diff --git "a/data_all_eng_slimpj/shuffled/split2/finalzzrznn" "b/data_all_eng_slimpj/shuffled/split2/finalzzrznn"
new file mode 100644--- /dev/null
+++ "b/data_all_eng_slimpj/shuffled/split2/finalzzrznn"
@@ -0,0 +1,5 @@
+{"text":"\nNom de Plume\n\nA (Secret) History of Pseudonyms\n\nCarmela Ciuraru\n\nDedication\n\nFor Sarah, everything\n\n(and for Oscar)\nEpigraphs\n\nWorld is crazier and more of it than we think.\n\nIncorrigibly plural. I peel and portion\n\nA tangerine and spit the pips and feel\n\nThe drunkenness of things being various.\n\n\u2014LOUIS MACNEICE, \"Snow\"\n\nOn whom, then, my God, am I the onlooker? How many am I? Who is me? What then is this gap between myself and me?\n\n\u2014FERNANDO PESSOA\n\n\"Must a name mean something?\" Alice asked doubtfully.\n\n\"Of course it must,\" Humpty Dumpty said, with a short laugh. \"My name means the shape I am and a good handsome shape it is, too. With a name like yours, you might be any shape.\"\n\n\u2014LEWIS CARROLL, Alice's Adventures in Wonderland\n\nThe self is like a bug. Every time you smack it, it moves to another place.\n\n\u2014PAT STEIR\nContents\n\nCover\n\nTitle Page\n\nDedication\n\nEpigraphs\n\nIntroduction\n\nChapter 1\n\nAnne, Charlotte, and Emily Bront\u00eb & Acton, Currer, and Ellis Bell (1816\u20131855)\n\n\"Once there were five sisters. . . .\"\n\nChapter 2\n\nGeorge Sand & Aurore Dupin (1804\u20131876)\n\n\"It began with an ankle-length gray military coat, matching trousers, a cravat, and a waistcoat. . . .\"\n\nChapter 3\n\nGeorge Eliot & Marian Evans (1819\u20131880)\n\n\"Charles Dickens was suspicious. . . .\"\n\nChapter 4\n\nLewis Carroll & Charles Dodgson (1832\u20131898)\n\n\"A show of hands if you've never heard of Alice in Wonderland. . . .\"\n\nChapter 5\n\nMark Twain & Samuel Clemens (1835\u20131910)\n\n\"How the protean Samuel Clemens became the world's most famous literary alias will never be known for sure. . . .\"\n\nChapter 6\n\nO. Henry & William Sydney Porter (1862\u20131910)\n\n\"If you are now reading or have recently read a short story by O. Henry, you are most likely a middle-school student. . . .\"\n\nChapter 7\n\nFernando Pessoa & His Heteronyms (1888\u20131935)\n\n\"You will never get to the bottom of Fernando Pessoa. . . .\"\n\nChapter 8\n\nGeorge Orwell & Eric Blair (1903\u20131950)\n\n\"Had Eric Arthur Blair been a working-class bloke from Birmingham instead of an Old Etonian . . .\"\n\nChapter 9\n\nIsak Dinesen & Karen Blixen (1885\u20131962)\n\n\"She was descended from Danish royalty, but her childhood was filled with the traditional privileges of an aristocratic upbringing. . . .\"\n\nChapter 10\n\nSylvia Plath & Victoria Lucas (1932\u20131963)\n\n\"She was a good girl who loved her mother. . . .\"\n\nChapter 11\n\nHenry Green & Henry Yorke (1905\u20131973)\n\n\"He's the best writer you've never heard of. . . .\"\n\nChapter 12\n\nRomain Gary & \u00c9mile Ajar (1914\u20131980)\n\n\"He was a war hero, a Ping-Pong champion, a film director, a diplomat, and an author who wrote the best-selling French novel of the twentieth century. . . .\"\n\nChapter 13\n\nJames Tiptree, Jr. & Alice Sheldon (1915\u20131987)\n\n\"On May 19, 1987, a seventy-one-year-old woman and her eighty-four-year-old husband were found lying in bed together, hand in hand, dead of gunshot wounds. . . .\"\n\nChapter 14\n\nGeorges Simenon & Christian Brulls et al. (1903\u20131989)\n\n\"He claimed to have had sex with ten thousand women. . . .\"\n\nChapter 15\n\nPatricia Highsmith & Claire Morgan (1921\u20131995)\n\n\"She was one of the most wretched people you could ever meet, with mood shifts that swung as wildly as the stock market. . . .\"\n\nChapter 16\n\nPauline R\u00e9age & Dominique Aury (1907\u20131998)\n\n\"Not many authors can boast of having written a best-selling pornographic novel. . . .\"\n\nAcknowledgments\n\nTime Line\n\nBibliography\n\nEpigraph\n\nAbout the Author\n\nCredits\n\nCopyright\n\nAbout the Publisher\nIntroduction\n\nAt its most basic level, a pseudonym is a prank. Yet the motives that lead writers to assume an alias are infinitely complex, sometimes mysterious even to them. Names are loaded, full of pitfalls and possibilities, and can prove obstacles to writing. Virginia Woolf, who never adopted a nom de plume herself, once expressed the fundamental and maddening condition of authorship: \"Never to be yourself and yet always\u2014that is the problem.\" She was describing the predicament of the personal essayist, but identity can seem crippling to any writer. A change of name, much like a change of scenery, provides a chance to start again.\n\nTo a certain extent, all writing involves impersonation\u2014the act of summoning an authorial \"I\" to create the speaker of a poem or the characters in a novel. For the audacious poet Walt Whitman, it was possible to explore other voices simply as himself. He embraced his multitudes. (\"Do I contradict myself? \/ Very well then, I contradict myself.\") But some writers are unable to engage in such alchemy, or don't want to, without relying on an alter ego. If the authorial persona is a construct, never wholly authentic (no matter how autobiographical the material), then the pseudonymous writer takes this notion to yet another level, inventing a construct of a construct. \"[T]he cultivation of a pseudonym might be interpreted as not so very different from the cultivation in vivo of the narrative voice that sustains any work of words, making it unique and inimitable,\" wrote Joyce Carol Oates in a 1987 New York Times essay. \"Choosing a pseudonym by which to identify the completed product simply takes the mysterious process a step or two further, officially erasing the author's (social) identity and supplanting it with the (pseudonymous) identity.\" Elide your own name, and imaginative beckoning can truly begin. As the French journalist and writer Fran\u00e7ois Nourissier once noted (in a piece entitled \"Faut-il \u00e9crire masqu\u00e9?\"), a nom de plume provides a space in which \"obstacles fall away, and one's reserve dissipates.\"\n\nThe merging of an author and an alter ego is an unpredictable thing. It can become a marriage, like a faithful and sturdy partnership, or it can prove a swift, intoxicating affair. A clandestine literary self can be tried on temporarily, to produce a single work, then dropped like a robe; or the guise might exist as something to be guarded at all costs. The attraction is obvious and undeniable. Entering another body (figuratively, ecstatically) is almost an erotic impulse. Historically, many writers have been lonely outsiders, which is why inhabiting another self offers an intimacy that seems otherwise unobtainable. In the absence of real-life companionship, the pseudonymous entity can serve as confidant, keeper of secrets, and protective shield.\n\nThe term \"alter ego\" is taken from Latin, meaning \"other I.\" This suggests the writer is not so much wearing a mask as becoming another person entirely. Have the two selves met? Maybe not, and it's probably better that way. Sometimes there's no reason to explore how or why the other half lives. Knowing that it does is enough.\n\nIn his influential 1974 book The Inner Game of Tennis, author Timothy Gallwey applied the notion of doubleness to the tennis player, describing how each self hinders or enhances performance. With almost no technical advice, he provides a prescriptive guide to mastery. He focuses on what he describes as two arenas of engagement: Self 1 and Self 2. When his book was first published, Gallwey's ideas were so radical that thousands of readers wrote to express their gratitude, saying that they'd successfully applied his principles to pursuits other than tennis, including writing.\n\nGallwey, who majored in English literature at Harvard University, portrays Self 1 as \"the talker, critic, controlling voice,\" and notes its \"persistence and inventiveness in finding opportunities to get in the way.\" Self 1 berates you, calls you an incorrigible failure. But the nonjudgmental Self 2 represents liberation in its purest form. As Gallwey writes, Self 2 is \"much more than a doer. It is capable of a range of feelings that are the most uniquely human aspect of life. These feelings can be explored in sports, the arts . . . and countless other activities. Self 2 is like an acorn that, when first discovered, seems quite small yet turns out to have the uncanny ability not only to become a magnificent tree but, if it has the right conditions, can generate an entire forest.\" In the context of authorship, the freeing of an alternate identity (Self 2) can reveal not just a forest but new worlds, boundless and transgressive, thrilling beyond one's wildest dreams.\n\nA pseudonym may give a writer the necessary distance to speak honestly, but it can just as easily provide a license to lie. Anything is possible. It allows a writer to produce a work of \"serious\" literature, or one that is simply a guilty pleasure. It can inspire unprecedented bursts of creativity and prove an antidote to boredom. For that rare bird known as the commercially successful author, there is typically less at stake in toying with a pen name. If the book produced by an ephemeral self fails, it will be viewed as a silly misstep. All is forgiven when an author retires a pen name and returns to giving critics and fans exactly what they want: the familiar. Lesson learned, let's move on. If you're writing the equivalent of high-fructose corn syrup, perhaps it's unwise to serve up organic spelt, even under a different brand name.\n\nFor best-selling authors like Nora Roberts (a truncated version of her actual name, Eleanor Robertson)\u2014who has written more than two hundred novels, including under the pen name J. D. Robb\u2014having a transparent or \"open\" pseudonym is a savvy marketing strategy, a way to keep up her busy production line and show off her versatility. Roberts had initially resisted writing as someone else, but her agent had talked her into it by explaining, \"There's Diet Pepsi, there's regular Pepsi, and there's Caffeine-Free Pepsi.\" It's all about brand extension.\n\nA new work by Stephen King, whose books have sold more than 500 million copies worldwide, is a reassuring promise of success to his publisher. It's also critic-proof. Yet in the late 1970s, feeling hemmed in by his phenomenally prolific output, King introduced the pen name Richard Bachman. As he later said, it was easy to add someone to his interior staff:\n\nThe name Richard Bachman actually came from when they called me and said we're ready to go to press with this novel, what name shall we put on it? And I hadn't really thought about that. Well, I had, but the original name\u2014Gus Pillsbury\u2014had gotten out on the grapevine and I really didn't like it that much anyway, so they said they needed it right away and there was a novel by Richard Stark on my desk, so I used the name Richard, and that's kind of funny because Richard Stark is in itself a pen name for Donald Westlake, and what was playing on the record player was \"You Ain't Seen Nothin' Yet\" by Bachman Turner Overdrive, so I put the two of them together and came up with Richard Bachman.\n\nKing's practical measure to avoid saturating the market (and avoid openly competing with himself for sales) was a success. But in 1985, a bookstore clerk in Washington, D.C., did some detective work and exposed King's secret. The author subsequently issued a press release announcing Bachman's death from \"cancer of the pseudonym.\" King dedicated his 1989 novel The Dark Half (about a pen name that assumes a sinister life of its own) to \"the late Richard Bachman.\"\n\nProminent writers such as Robert Ludlum, Joyce Carol Oates, Anthony Burgess, Anne Rice, Michael Crichton, John Banville, Ruth Rendell, and Julian Barnes are also known to have indulged in pseudonymous publication. The Nobel laureate Doris Lessing, who tested out a nom de plume in the early 1980s, learned that she was better off sticking with her own identity. One of her aims had been a respite from the public's perception of her work; she sought to upend preconceptions of what it meant to read a \"Doris Lessing novel.\"\n\nShe also had something to prove. Lessing wanted to see how her books would be received if no one knew they were by the author of The Golden Notebook (a novel that had sold nearly a million copies), as well as more than twenty other books. \"I wanted to highlight that whole dreadful process in book publishing that 'nothing succeeds like success,'\" she said later in an interview. \"If the books had come out in my name, they would have sold a lot of copies and reviewers would have said, 'Oh, Doris Lessing, how wonderful.'\"\n\nThat's debatable, but on another level, Lessing had revenge in mind: the ruse was a way to strike back at critics who she felt had \"hated\" her then-recent Canopus novels, a five-volume science-fiction series of which she was extremely proud. (She considered the series her most important work.) So Lessing became \"Jane Somers\" and wrote the novel The Diary of a Good Neighbour, which her longtime UK publisher, Jonathan Cape, rejected, insisting that it was not commercially viable. The novel traced the friendship between two women: a middle-aged magazine editor and an octogenarian. After Lessing found a publisher, Michael Joseph, the book was released in the UK in 1983. (The coy jacket copy indicated, falsely, that Somers was the pen name of \"a well-known English woman journalist.\") It sold only a few thousand copies, and the American edition fared poorly, too.\n\nWas its failure due to people's fixation on famous authors, or was it a bad book? Lessing blamed the former. Was her test nothing but an egotistical publicity stunt? A critic from the Washington Post, Jonathan Yardley, seemed to think so. He argued that it was not at all the \"success syndrome\" that had troubled Lessing, but rather that \"reviewers refused to be seduced by her name on the 'Canopus' novels and picked them to pieces.\"\n\nRegardless, Lessing followed up a year later with a Somers sequel, If the Old Could, and soon after its publication she confessed that she had written both books. \"The reviews were more or less what I expected,\" she said of her experiment. \"It was interesting to be a beginning writer again because I found how patronizing reviewers can be.\"\n\nOf course, authorial charlatanism isn't always provoked by malice, fear, guilt, or any other dark motive. The best-selling author Tom Huff, who died in 1990, was a Texan who published gothic novels, but he rechristened himself Jennifer Wilde to venture convincingly into bodice-ripping historical romance. He did so with the 1976 novel Love's Tender Fury, and although he had used other female pseudonyms, none earned him the kind of success he experienced as Wilde.\n\nTerry Harknett, the prolific author of nearly two hundred books, wrote westerns\u2014as in gun slinging and tobacco chewing\u2014using rancher-sounding names like George G. Gilman. Harknett once described himself as a frustrated suspense writer: \"For fifteen long years, I wrote mystery novels that were published twice yearly\u2014and sank without trace at the same rate.\" In a rather unlikely way, he had stumbled into the genre of westerns, and his Gilman novels went on to sell millions of copies. Not bad for a British man from Essex with a decidedly unmasculine name.\n\nSometimes, however, literary fakery crosses the line from being a harmless alias, employed for the author's private, benign purpose. It is perceived as mendacity, as an appalling betrayal of trust. The consequences of this exploitation can tarnish the poseur's reputation irrevocably. And when not only does the supposed background of an author prove fraudulent, but the material presented as autobiographical is itself a lie, the backlash is especially dreadful.\n\nIn early 2008, a writer named Margaret B. Jones published Love and Consequences: A Memoir of Hope and Survival. This was a harrowing story of the author's experiences as a foster child and a Bloods gang member in South Central Los Angeles. She recalled one of the crucial lessons she had learned in her former life: \"Trust no one. Even your own momma will sell you out for the right price or if she gets scared enough.\"\n\nWriting in the New York Times, in a review accompanied by the headline \"However Mean the Streets, Have an Exit Strategy,\" the critic Michiko Kakutani called the book \"humane and deeply affecting\" and praised the author for writing \"with a novelist's eye for the psychological detail and an anthropologist's eye for social rituals and routines.\"\n\nThe book was a fabrication, and \"Margaret B. Jones\" did not exist. (The author's duplicity was exposed by her own sister.) \"Jones,\" it turned out, was the persona of Margaret Seltzer, a thirty-three-year-old white woman living with her daughter in a four-bedroom 1940s bungalow in Eugene, Oregon. Seltzer had grown up with her biological parents in affluent Sherman Oaks, California, and had attended a private Episcopal day school. She did not have a black foster mother whom she called \"Big Mom,\" nor foster siblings named Terrell, Taye, Nishia, and NeeCee. She was neither a Blood nor a Crip. And she had not, at fourteen years old, received a gun as a birthday gift.\n\nRiverhead Books, the publisher of Love and Consequences, promptly canceled the author's publicity tour, recalled copies of the book, and offered refunds to those who had purchased it. For her part, Seltzer claimed that her intentions had been honorable. \"I thought it was my opportunity to put a voice to people who people don't listen to,\" she said in an interview. \"I was in a position where at one point people said you should speak for us, because nobody else is going to let us in to talk. Maybe it's an ego thing\u2014I don't know. I just felt that there was good that I could do and there was no other way that someone would listen to it.\" Seltzer had written much of the book at a Starbucks in Los Angeles.\n\nThe morbidly shy young writer JT LeRoy, a teenage drifter and recovering drug addict from West Virginia, courted (mostly by phone, mail, and fax) the sympathetic attention of Hollywood celebrities such as Winona Ryder and Drew Barrymore, and prominent authors including Mary Karr and Dennis Cooper. Another fan of his work, Madonna, once sent LeRoy some books on kabbalah as a gift. No one actually met him.\n\nHe maintained an enigmatic allure, and it wasn't long before rumors circulated that there was no JT LeRoy. (Chlo\u00eb Sevigny said that he was definitely real because \"he's left several messages on my answering machine.\") When the writer Mary Gaitskill wanted to meet him in person, the \"real\" LeRoy\u2014Laura Albert, a former phone-sex operator from Brooklyn\u2014paid a nineteen-year-old boy she'd met on the street (\"You want to make fifty bucks, no sex?\") to meet Gaitskill quickly at a San Francisco caf\u00e9, \"get freaked out,\" and leave. Later, other \"stunt doubles\"\u2014always wearing sunglasses and a blond wig\u2014were hired to embody LeRoy for public appearances.\n\nFollowing publication of the cult favorites Sarah and The Heart Is Deceitful Above All Things, LeRoy was praised as a wunderkind and his work described as a \"revelation.\" Although both books were works of fiction, LeRoy's marketability (and his many celebrity friendships) depended on his image as a wounded kid with a hardscrabble background. The director Gus Van Sant spoke to LeRoy by phone for hours every day, and gave him an associate-producer credit on the 2003 film Elephant. Dave Eggers edited (and wrote the foreword to) LeRoy's 2005 novella, Harold's End, which appeared first in McSweeney's. Eggers wrote that LeRoy's books would prove to be \"among the most influential American books in the last ten years.\"\n\nSeveral months later, a journalist revealed LeRoy's true identity, and the fallout was immediate and severe. A company that had optioned the film rights to Sarah successfully sued Albert for fraud. Still, in the wake of the ignominious scandal, the middle-aged author was unapologetic: \"I went through a minefield,\" she said, \"and I put on camouflage in order to tell the truth.\" Albert felt victimized by the media and insisted that she could not have written LeRoy's works under her own name. She denied that she had perpetrated a hoax. \"It really felt like he was another human being,\" she told the Paris Review in a 2006 interview. \"He'd tell the story and I was the secretary who would take it down and say, OK, thank you, now I'm going to try to turn it into craft. But while I wouldn't sit there and think of myself as JT, as long as I was writing I didn't have to be Laura either.\"\n\nWhat's in a name? Everything. Nothing. Some writers find that crafting prose under the name they were born with is too restrictive. It can seem oddly false, or perhaps not grand enough to accompany their literary peregrinations. A name carries so much baggage; it can seem tired and dull. Too ethnic. Too stultifying. Too old. Too young. In such instances, an author may be unable to proceed if he is, say, Samuel Clemens, but feels capable of achieving impressive feats if he is Mark Twain. Imagination blooms. Assume an alias, and the depths of the mind can be plumbed at last, without fear of retribution, mockery, or\u2014worst of all\u2014irrelevance. The erasure of a primary name can reveal what appears to be a truer, better, more authentic self. Or it can attain the opposite, by allowing a writer to take flight from a self that is \"true\" yet shameful or despised.\n\nA nom de plume can also provide a divine sense of control. No writer can determine the fate of a book\u2014how the poems or novels are interpreted, whether they are loved or grossly misunderstood. By assuming a pen name, though, an author can claim territory, seize possession of a work before the reader or critic inevitably distorts it. In this way, the author gets the last laugh: despise my book as much as you like; you don't even know who wrote it. However petty, such trickery yields infinite pleasure. Obfuscation is fun!\n\n\"Every writer\u2014after a certain point, when one's labors have resulted in a body of work\u2014experiences himself or herself as both Dr. Frankenstein and the monster,\" Susan Sontag once lamented. Authorial identity can become a trap that causes creative fatigue or even halts literary output altogether. As many writers know firsthand, the literary world is tough: one minute you're the toast of the town; the next minute you're just toast. The desire to emancipate oneself from the shackles of familiarity and start anew, under an altogether different name, makes perfect sense. In fact, why not more pseudonyms?\n\nIn the nineteenth century, the curious phenomenon of pseudonymity reached its height, and as early as the mid-sixteenth century, it was customary for a work to be published without any author's name. It is interesting that the decline of pseudonyms in the twentieth century coincided with the rise of television and film. As people gained more access to the lives of others, it became harder to maintain privacy\u2014and perhaps less desirable. In today's culture, no information seems too personal to be shared (or appropriated). Reality television has increased our hunger to \"know\" celebrities, and even authors are not immune to the pressures of self-promotion and self-revelation; we are in an era in which, as the biographer Nigel Hamilton has written, \"individual human identity has become the focus of so much discussion.\" This is not entirely new, but with the explosion of digital technology, things seem to have spiraled out of control. Fans clamor to interact, online and in person, with their favorite writers, who in turn are expected to blog, sign autographs, and happily pose for photographs at publicity events. Along with their books, authors themselves are sold as products. Even though the practice of pseudonymity is still going strong, it has lost the allure it once had, and for the most part it is applied perfunctorily in genres such as crime fiction or erotica. Today, using a pen name is less often a creative or playful endeavor than a commercial one. Reticence is not what it used to be.\n\nFor each of the authors in this book, hiding behind a nom de plume was essential. However varied their literary styles and their reasons for going undercover, all of them longed to escape the burdens of selfhood\u2014whether permanently or for a brief period in their lives. To publish their work, many risked their reputations, their means of subsistence, and even the relationships they held most dear. Three of the authors committed suicide (Sylvia Plath, Romain Gary, and Alice Sheldon); others had contemplated killing themselves or attempted it; at least one author (Alice Sheldon) was bipolar; and several\u2014including the Bront\u00eb sisters, George Eliot, Isak Dinesen, and George Orwell\u2014suffered from chronic health issues. Many succumbed to strange compulsions, addictions, and self-destructive habits. Almost all were lonely, and few were adept at friendship, marriage, or parenthood. One was a convicted criminal. A number of them, including Henry Green, Georges Simenon, and Patricia Highsmith, were alcoholics. Some achieved literary success in their twenties, while others were late bloomers who found recognition in midlife. But the Portuguese writer Fernando Pessoa, who channeled more than seventy different identities, lived in obscurity and never achieved acclaim. At the time of his death, he left behind more than thirty thousand fragments of his unpublished writings in a trunk. For Romain Gary, the best-selling French author of the twentieth century, pseudonymity became a cage, much like fame.\n\nMost of these authors had endured childhoods with domineering, neglectful, or cruel parents. They suffered profound trauma early on, such as the death of a parent (in the case of Dinesen's father, by hanging himself) or of one or more siblings. Mark Twain outlived his spouse and all but one of his children; Georges Simenon's daughter killed herself. For these troubled authors whose lives seemed to bring impediments without surcease, an alter ego served as a kind of buffer, protecting them (at least up to a point) from the painful aspects of their lives.\n\nThis book is a selective chronicle of pseudonymity over a hundred-year period, beginning in the mid-nineteenth century and ending in the mid-twentieth century. To explore this peculiar tradition is to tap into, among other themes, the complex psychological machinery of authorial identity; the perils of literary fame; the struggles of the artist within a society generally hostile to such a vocation; courage and faith; and the nature of creativity itself. In certain respects, delving into pseudonymity is a frustrating endeavor. No pithy or singular conclusions can be made. It's a puzzle. By definition, this is a history riddled with lacunae: there are thousands of recorded noms de plume, but many more that we will never know.\n\nIn reflecting on the tumultuous lives of the authors in this book, it's hard not to consider the literary deprivation we might have suffered had they not found the protective cover they needed to write. But that would mean contemplating a world without, say, Jane Eyre, Middlemarch, or Alice in Wonderland. Instead, let us celebrate the sense of liberation, however short-lived, that these writers found through pseudonymity. In carving out their secret identities, they went to astonishing lengths. Each of these authors possessed extraordinary determination and resilience.\n\nHere are their stories.\nThey were dead by the age of forty\n\nChapter 1\n\nAnne, Charlotte, and Emily Bront\u00eb & ACTON, CURRER, AND ELLIS BELL\n\nOnce there were five sisters. In 1825, Maria and Elizabeth Bront\u00eb, the two eldest, died of tuberculosis. That left Charlotte (born in 1816), Emily (born in 1818), and Anne (born in 1820), as well as a brother, Branwell, born in 1817. Their mother, Maria Branwell Bront\u00eb, died of cancer a year after Anne's birth. Their Irish minister father, Patrick, would outlive them all, dying in 1861 at the age of eighty-four.\n\nThe Bront\u00eb children grew up in a manufacturing village at the edge of the Pennine moors in West Yorkshire, England, and would spend, almost without exception, their entire lives at their father's parsonage at Haworth. The plain, two-story early Georgian building where they once lived is now a museum. Eventually, Haworth would be known as Bront\u00eb country. It might have been known as Brunty country, had their father not changed his family surname while studying at Oxford. (\"Bront\u00eb\" means \"thunder\" in Greek.)\n\nLiving with their father and an aunt, Elizabeth, who helped raise them (and whom they did not love), the children lacked playmates but had one another. Precocious and bookish, they retreated into their own private world. They roamed the moors, and, as Charlotte later wrote, Emily especially loved doing so. \"They were far more to her than a mere spectacle; they were what she lived in and by, as much as the wild birds, their tenants, or the heather, their produce. . . . She found in the bleak solitude many and dear delights; and not the least and best loved was\u2014liberty.\"\n\nThe children kept dogs, cats, and birds as pets, made drawings, and invented stories, creating elaborate fantasy worlds in which they could lose themselves. Lonely in the absence of their mother, the children developed rich sagas of imaginary cities and kingdoms. Their grand creation was \"Great Glass Town Confederacy,\" presided over by the \"Four Genii,\" named Tallii, Brannii, Emmii, and Annii. They conceived histories of Glass Town and even composed Glass Town songs. Later came the kingdoms of Angria, invented by Charlotte and Branwell, and Gondal, as dreamed up by Emily and Anne. There were kings, queens, pirates, heroes, romances, armies, schools, and struggles between good and evil. These apparently silly children's games gave rise to a flurry of literary activity, proving to be exercises in developing their craft. By their late teens, the Bront\u00ebs had a command of plot, characterization, and pacing.\n\nAnother significant detail from their childhood was the rather unorthodox pedagogical method their father applied with them: the children would put on masks, and Patrick would question them intensively, one by one, about various subjects to test their knowledge. He believed that by wearing masks the children would feel unself-conscious and learn to speak with confidence and candor.\n\nWhen Branwell created the Young Men's Magazine at the age of twelve, the siblings (most of all Charlotte) contributed essays, plays, and illustrations. Like Charlotte, Branwell was ambitious about his writing and desired a readership beyond the family. He believed he was destined for greatness. At twenty, he wrote a sycophantic letter about his literary efforts to William Wordsworth, enclosing samples of his own work, but the poet never replied. (Wordsworth reported to others that he was \"disgusted\" by Branwell's letter.)\n\nAt twenty-one, Charlotte also took the bold step of writing to a famous author, the poet laureate Robert Southey, asking for his opinion of her work. She shyly confessed to him that she longed \"to be forever known\" as a poet. Southey was a poor choice for a potential mentor; cranky, elderly, and in poor health, he had no interest in a young woman's literary aspirations. (She wrote to him using her own name.) Three months later, he replied by acknowledging her obvious talent and then putting her in her place. He issued a stern admonition that young poets hoping to get published \"ought to be prepared for disappointment,\" and that, above all, \"Literature cannot be the business of a woman's life, and it ought not to be.\" Surely he did not expect or even want a response to his missive, but he got one anyway: a letter from Charlotte that was almost comical in its expression of meek obedience. \"In the evenings, I do confess, I do think,\" she wrote, \"but I never trouble any one else with my thoughts. I carefully avoid any appearance of preoccupation and eccentricity, which might lead those I live amongst to suspect the nature of my pursuits. . . . Sometimes when I'm teaching or sewing I would rather be reading or writing; but I try to deny myself.\" She closed her letter by thanking him again \"with sincere gratitude\" for essentially crushing her dreams. If her misguided literary ambition should arise again, Charlotte told him, she would simply reread his letter \"and suppress it.\"\n\nThe vast trove of Bront\u00eb juvenilia is larger than all their published works put together. Most of the material was recorded in nearly microscopic handwriting, on tiny folded sheets of paper\u2014some only 2 inches by 1\u00bd inches. These were stitched and bundled together, complete with title pages and back covers made from scraps of wrapping paper and bags of sugar. For her part, Charlotte was already documenting her own literary accomplishments\u2014all twenty-two volumes\u2014with a detailed record titled \"Catalogue of My Books, With the Period of Their Completion Up to August 3, 1830,\" when she was just fourteen years old. Three years later she wrote a novella, The Green Dwarf, under the name \"Wellesley.\"\n\nThe sisters wrote constantly, but had it not been for Charlotte, their efforts might have remained private. She dreamed of making writing her vocation and was unafraid to pursue it. Her foray into publishing was inspired not by her own work, however, but by Emily's.\n\nCharlotte later described how she came across one of her sister's small notebooks and, although this was a violation of privacy, read what Emily had written: \"One day, in the autumn of 1845, I accidentally lighted on a MS volume of verse in my sister Emily's handwriting. Of course, I was not surprised, knowing that she could and did write verse: I looked it over, and something more than surprise seized me\u2014a deep conviction that these were not common effusions nor at all like the poetry women generally write. . . . To my ear, they had also a peculiar music\u2014wild, melancholy and elevating.\" Emily was furious when she found out what Charlotte had done. It was only after breaking down her sister's resistance that Charlotte \"at last wrung out a reluctant consent to have the 'rhymes' as they were contemptuously termed, published.\"\n\nLeft to her own devices, Emily probably would have kept her work private, much like another nineteenth-century Emily\u2014Dickinson, the \"belle of Amherst\"\u2014with whom she had a certain temperamental kinship. (Bront\u00eb's poem \"Last Lines\" would be read at Dickinson's funeral in 1886.)\n\nUnlike Anne or Charlotte, Emily was by nature reclusive and always the least inclined to speak. She felt no need to reach the world beyond Haworth. As Charlotte later explained, her sister tended toward seclusion, and \"except to go to church, or take a walk on the hills, she rarely crossed the threshold of home.\"\n\nAnne, too, Charlotte noted, had \"a constitutional reserve and taciturnity,\" but she was also ambitious. Finally, at Charlotte's urging, the sisters decided to publish, under assumed (and gender indeterminate) names, a volume of poems by all three of them: twenty-one poems by Emily, nineteen by Charlotte, and twenty-one by Anne. Branwell was excluded from this endeavor. His life\u2014and his tremendous artistic potential\u2014would be curtailed by alcoholism, opium addiction, and the often reckless behavior that embarrassed his family. He understood his predicament but felt helpless to fix it. \"I have lain during nine long weeks utterly shattered in body and broken down in mind,\" he wrote during one of his typical bad stretches. Branwell was too much of a mess to be let in on his sisters' secret identities; they had to shut him out. He was a loudmouth drunk who would, they were sure, inevitably spill the news of their pseudonyms.\n\n\"My hopes ebb low indeed about Branwell,\" Charlotte wrote of the brother she had once idolized. \"I sometimes fear he will never be fit for much.\" He was dead at thirty-one.\n\nCharlotte took the initiative with regard to publication by sending query letters to publishers, but she had trouble even getting a response. Presenting herself as an \"agent\" writing on behalf of the authors, she sent a letter to the firm Aylott & Jones in January 1846:\n\nGentlemen\u2014May I request to be informed whether you would undertake the publication of a Collection of short poems in I vol. oct.\n\nIf you object to publishing the work at your own risk, would you undertake it on the Author's account\u2014I am gentlemen,\n\nYour obdt. Hmble. Servt.\n\nC. Bront\u00eb\n\nThey agreed to accept the book for publication, provided it was at the authors' own expense. Charlotte had very specific ideas about how the book should be presented: \"I should like it to be printed in 1 octavo volume of the same quality of paper and size of type as Moxon's last edition of Wordsworth,\" she wrote. \"The poems will occupy\u2014I should think from 200 to 250 pages.\" She also expressed herself emphatically on the printing: \"clear type\u2014not too small\u2014and good paper.\"\n\nHaving reached an agreement, Charlotte sent the manuscript (as \"C. Bront\u00eb Esq\") to Aylott & Jones. \"You will perceive that the Poems are the work of three persons\u2014relatives\u2014their separate pieces are distinguished by their separate signatures,\" she explained.\n\nWhen Poems by Currer, Ellis, and Acton Bell came out in the summer of 1846, the savvy Charlotte oversaw advertising and promotion. She had directed the design, and now she suggested how the book should be released to the public and which publications ought to review it. She was gratified by the positive critical reception that Poems received. \"It is long since we have enjoyed a volume of such genuine poetry as this,\" one reviewer wrote, expressing curiosity regarding \"the triumvirate\" and wondering whether the Bells might be pseudonymous authors. Another contemplated the possibility that the trio might be \"one master spirit . . . that has been pleased to project itself into three imaginary poets.\" Charlotte was more than happy to feed public curiosity: writing a letter to one magazine editor (under her pseudonym), she thanked him for his very kind review and referred to \"my brothers, Ellis and Acton.\"\n\nFour years later, in the posthumous editions of Wuthering Heights and Agnes Grey, Charlotte would explain fully the motive behind their pseudonyms:\n\nAverse to personal publicity, we veiled our own names under those of Currer, Ellis, and Acton Bell; the ambiguous choice being dictated by a sort of conscientious scruple at assuming Christian names positively masculine, while we did not like to declare ourselves women, because\u2014without at that time suspecting that our mode of writing and thinking was not what is called \"feminine\"\u2014we had a vague impression that authoresses are liable to be looked on with prejudice; we had noticed how critics sometimes use for their chastisement the weapon of personality, and for their reward, a flattery, which is not true praise.\n\nDespite the positive reviews of the book, it was a failure financially. Only two copies were sold. (The initial print run was around a thousand.) Charlotte was not the least bit discouraged. \"The mere effort to succeed had given a wonderful zest to existence,\" she wrote. \"It must be pursued.\"\n\nA year later, seeing that nothing had come of their poetic debut, Charlotte, tenacious as ever, sent copies of the slim green volume to various celebrated authors, including Tennyson, Wordsworth, and De Quincey, with an imploring letter to each:\n\nSir,\n\nMy relatives, Ellis and Acton Bell, and myself, heedless of the repeated warnings of various respectable publishers, have committed the rash act of printing a volume of poems.\n\nThe consequences predicted have, of course, overtaken us; our book is found to be a drug; no man needs or heeds it. In the space of a year our publisher has disposed but of two copies, and by what painful efforts he succeeded in getting rid of these two, himself only knows.\n\nBefore transferring the edition to the trunk-makers, we have decided on distributing as presents a few copies of what we cannot sell\u2014We beg to offer you one in acknowledgement of the pleasure and profit we have often and long derived from your works.\n\nI am, sir, yours very respectfully,\n\nCurrer Bell.\n\nUndeterred by the Bells' lackluster debut, Charlotte wrote a follow-up letter to Aylott & Jones, advising them that \"C. E. & A. Bell are now preparing for the Press a work of fiction\u2014consisting of three distinct and unconnected tales which may be published together as a work of 3 vols. of ordinary novel-size, or separately as single vols\u2014as shall be deemed most advisable.\" And she brashly advised them to respond soon, as other publishers might be interested as well. They declined the solicitation.\n\nWhat they foolishly turned down, of course, were novels that would become part of the canon of English literature: Anne was writing Agnes Grey. Emily had begun Wuthering Heights (whose ferocity of emotion Charlotte found rather off-putting). And Charlotte had collected all the material she needed for her novel Jane Eyre, having worked, quite miserably, as a governess\u2014but the novel she'd written first was The Professor, with its male narrator, Charles Grimsworth, who teaches at a girls' school in Brussels. The story, which she'd completed in June 1846, was based on her own formative time at a Brussels girls' school, where she fell in love (unrequited) with her headmaster before homesickness set in and she returned, deeply depressed, to the refuge of Haworth.\n\nThough she tried submitting their works for consideration elsewhere, she had no luck. Finally, Wuthering Heights and Agnes Grey were accepted by a minor publisher, Thomas Cautley Newby, but he didn't want The Professor. Charlotte sent it to other publishers, and it was repeatedly rejected. In fact, she would not see the novel published in her lifetime. It came out in 1857, two years after her death.\n\nAmazingly, the year 1847 would bring publication for all three sisters, almost at once. Charlotte completed Jane Eyre, which she'd written in small square books. As she wrote, she suffered from an almost unbearably painful toothache and gum disease that would linger for years. (By 1851, Charlotte had very few teeth left.) But she persevered, and Jane Eyre was accepted with enthusiasm by the obscure publishing house Smith, Elder and Company in London.\n\nIt wouldn't remain unknown for long; in the latter half of the century, Smith, Elder became known as the distinguished publisher of Elizabeth Gaskell, Matthew Arnold, George Eliot, Thackeray, Browning, and Ruskin. The firm's eventual success could be traced to having taken a chance on an unknown writer named Currer Bell.\n\nCharlotte submitted the manuscript to her publisher in August 1847, with a note indicating casually that \"[i]t is better in future to address Mr Currer Bell, under cover to Miss Bront\u00eb, Haworth, Bradford, Yorkshire, as there is a risk of letters otherwise directed not reaching me at present.\" Later, George Smith, the head of the firm, recalled his suspicions about Currer Bell: \"For my own part I never had much doubt on the subject of the writer's sex; but then I had the advantage over the general public of having the handwriting of the author before me.\"\n\nPublished just six weeks later on October 16, Jane Eyre, with its declarative opening line\u2014\"There was no possibility of taking a walk that day\"\u2014proved shocking to many Victorians, and even an assault against decorum. Yet it was immediately recognized as a masterpiece, and could count among its admirers Queen Victoria, who read it aloud to her \"dear Albert.\" Thackeray, who'd received an early review copy, wrote to Charlotte's publisher:\n\nI wish you had not sent me Jane Eyre. It interested me so much that I have lost (or won if you like) a whole day in reading it. . . . Who the author can be I can't guess, if a woman she knows her language better than most ladies do, or has had a \"classical\" education. . . . Some of the love passages made me cry. . . . I don't know why I tell you this but that I have been exceedingly moved and pleased by Jane Eyre. It is a woman's writing, but whose?\n\nElizabeth Barrett Browning thought it a fine novel (and superior to the subsequent Shirley and Villette) but wrote to a friend, \"I certainly don't think that the qualities, half savage and half freethinking, expressed in Jane Eyre are likely to suit a model governess or schoolmistress.\" Although she found these \"qualities\" repugnant and expressed her disapproval, she was excited by the mystery of the authorship\u2014particularly the scandalous gossip that \"Currer Bell\" was actually a young governess. Another critic declared that the novel was \"[w]orth fifty Trollopes and Martineaus rolled into one counterpane, with fifty Dickenses and Bulwers to keep them company,\" but added that the author of Jane Eyre was \"rather a brazen Miss.\"\n\nCompared with her sisters' novels, Charlotte's debut achieved by far the greatest commercial and critical success. Sales exceeded all expectations, and within six months Jane Eyre went into a third printing. Charlotte\u2014or, rather, her nom de plume\u2014became the most celebrated author in England. Deepening the mystery was the book's curious title page: \"Jane Eyre: An Autobiography. Edited by Currer Bell.\" It had been George Smith's idea to add the provocative subtitle. The novel was very autobiographical indeed\u2014for Charlotte, that is. Some critics believed that Bell was a woman, but to others it seemed obvious that the novel was simply too good to have been written by a female author. \"It is no woman's writing,\" wrote one reviewer confidently. \"Although ladies have written histories, and travels, and warlike novels, to say nothing of books upon the different arts and sciences, no woman could have penned the 'Autobiography of Jane Eyre.' It is all that one of the other sex might invent, and much more.\" The critic George Henry Lewes wrote that the novel was perhaps not autobiographical \"in the naked facts and circumstances,\" but it certainly appeared to be \"in the actual suffering and experience.\"\n\nSome speculated that perhaps Acton and Currer Bell were the same person. A baffled critic surmised that the author's identity was divided, \"if we are not misinformed, with a brother and sister. The work bears the marks of more than one mind and more than one sex.\" One writer argued that the novel's \"mistakes\" about \"preparing game and dessert dishes\" proved beyond a doubt that the author was a man, because no female author would have been so clueless. But another claimed that \"only a woman or an upholsterer\" could have written the section about sewing on brass rings. Yet another reviewer was convinced that the name was a pseudonym, perhaps an anagram, and that the book was definitely by a woman from the north of England. \"Who, indeed, but a woman could have ventured, with the smallest prospect of success, to fill three octavo volumes with the history of a woman's heart?\"\n\nAs Elizabeth Gaskell wrote in her biography of Charlotte, following the publication of Jane Eyre Charlotte's life became \"divided into two parallel currents,\" that of Bell and Bront\u00eb, and \"there were separate duties belonging to each character\u2014not opposing each other; not impossible, but difficult to be reconciled.\" Gaskell noted ruefully that when a man becomes an author, \"it is probably merely a change of employment to him,\" but for a woman to take on the same role, especially in secret, the burdens seem too great to overcome. \"[N]o other can take up the quiet, regular duties of the daughter, the wife, or the mother,\" Gaskell wrote. Sequestered at the parsonage, where the most exciting part of her day was the postman's call, Charlotte was somewhat protected from the pressures of her fame\u2014but not entirely.\n\nLiterary London was buzzing about Currer Bell. Most agreed that whoever the author was, he or she had extraordinary talent. \"This is not merely a work of great promise,\" one critic said, \"it is one of absolute performance. It is one of the most powerful domestic romances which has been published for many years.\" There came an inevitable backlash\u2014among other things, the novel was said to be coarse and immoral\u2014but those reviews were drowned out by the praise. (Some critics wanted it both ways: The Economist declared the novel a triumph if written by a man, \"odious\" if written by a woman.)\n\nCharlotte could not resist sharing a copy of the book (along with some laudatory reviews) with her gruff father, who had no idea that she'd been published. All of Patrick's support, interest, and hope for the future had been lost with his son. But he read the novel one afternoon, summoned his daughters to tea, declared the book \"a better one than I expected,\" and did not mention it again for the next few years.\n\nAlthough Charlotte found refuge in her anonymity, her happiness about the novel's triumphant reception was tempered by the drubbing that Emily took for Wuthering Heights. Agnes Grey (like poor Anne) did not stir a strong reaction in anyone. Their novels were published together in December 1847, just as Charlotte was preparing for the second edition of Jane Eyre. Unfortunately, Emily and Anne found their publisher to have done a shamefully shoddy job; their books were riddled with mortifying mistakes of spelling and punctuation that they'd corrected on proof sheets, and new errors had been introduced. Most of the reviews of Wuthering Heights were unkind. Although critics recognized the power of Ellis Bell's writing, one reviewer deemed the characters \"grotesque, so entirely without art, that they strike us as proceeding from a mind of limited experience.\" And readers were warned that they would be \"disgusted, almost sickened by details of cruelty, inhumanity and the most diabolical hate and vengeance\" in Wuthering Heights. Emily, always reclusive, did not speak of her pain at reading the negative reviews; nor did she admit how hurtful it was to see Charlotte's work bask in adulation at the same time. But after her death it was discovered that tucked inside her desk, Emily had saved the clippings of the reviews comparing her novel unfavorably with Jane Eyre.\n\nMeanwhile, Charlotte clutched the protective umbrella of Currer Bell as the storm of publicity raged around her. In a letter to her editor, she wondered \"what author would be without the advantage of being able to walk invisible?\"\n\nFor the third edition of Jane Eyre, she wrote a brief author's note \"to explain that my claim to the title of novelist rests on this one work alone. If, therefore, the authorship of other works of fiction has been attributed to me, an honour is awarded where it is not merited; and consequently, denied where it is justly due. This explanation will serve to rectify mistakes which may already have been made, and to prevent future errors.\" Dated April 13, 1848, it was signed \"Currer Bell.\" She'd written it as an irked response to Emily's and Anne's disreputable publisher, who had led readers to believe that one \"Mr. Bell\" was responsible for the works by all three sisters. The Bell brothers were thus accused of \"trickery.\" This misrepresentation had brought trouble for Charlotte on a number of levels, including a need to assure her own publisher, George Smith, that his author was not working for a competitor behind his back.\n\nThat year, Jane Eyre was sold in the United States, also to great acclaim, and the New York publisher Harper & Brothers had eagerly submitted a high bid to acquire the rights to Currer Bell's next novel.\n\nAt home, people were clamoring to know who the elusive Bell was. Charlotte could not contain her secret much longer; nevertheless, she wrote to her publisher insisting that the author's identity remain protected at all costs. \" 'Currer Bell' only I am and will be to the Public; if accident or design should deprive me of that name,\" she wrote, \"I should deem it a misfortune\u2014a very great one. Mental tranquility would then be gone; it would be a task to write, a task which I doubt whether I could continue.\"\n\nIn July 1848, Charlotte made a dramatic decision: without giving notice, she traveled to London to introduce herself\u2014her real self\u2014to Smith and to her editor, W. S. Williams. Deeply grateful for everything the firm had done for her, she felt obliged to be forthright and to prove that one author was not responsible for the novels of all three. Originally she'd planned to surprise Smith at his office accompanied by both Anne and Emily, but Emily refused to go. She was upset about the turn of events and viewed the confession as a betrayal. Charlotte felt terribly guilty. Following her visit to the office she wrote to Williams, asking him to pretend that their meeting had never happened, at least as far as Emily was concerned.\n\n\"Permit me to caution you not to speak of my sisters when you write to me,\" Charlotte advised. \"I mean, do not use the word in the plural. Ellis Bell will not endure to be alluded to under any other appellation than the nom de plume. I committed a grand error in betraying his identity to you and Mr. Smith. It was inadvertent\u2014the words 'we are three sisters' escaped me before I was aware. I regretted the avowal the moment I had made it; I regret it bitterly now, for I find it is against every feeling and intention of Ellis Bell.\" Even after her sisters died, she maintained \"Currer Bell\" as her authorial identity.\n\nApart from Emily's agitation about the trip, it had been wonderful in every way. Charlotte and Anne had stayed in Paternoster Row, in the shadow of St. Paul's Cathedral, at the Chapter Coffee House, which had once been a meeting place for luminaries such as Dr. Johnson\u2014\"the resort of all the booksellers and publishers; and where the literary hacks, the critics, and even the wits, used to go in search of ideas or employment,\" as Elizabeth Gaskell would describe it in her biography of Charlotte.\n\nThe sisters' arrival at the publisher's office was priceless: when Charlotte showed up, along with Anne, Smith was confused by the sudden appearance of two \"rather quaintly dressed little ladies, pale-faced and anxious-looking.\" (He wasn't joking about the \"little\" part\u2014at five feet three, Emily was the tallest of the sisters; Charlotte was a mere four feet nine.) He was also annoyed because the two strangers\u2014women, at that\u2014had shown up uninvited on a busy workday demanding to see him. They declined to give their names. \"One of them came forward and presented me with a letter\u2014addressed in my own handwriting to 'Currer Bell, Esq.,'\" he recalled. \"I noticed that the letter had been opened, and said with some sharpness: 'Where did you get this from?' 'From the post office,' was the reply. 'It was addressed to me. We have both come that you might have ocular proof that there are at least two of us.'\"\n\nHowever much Smith had suspected Currer Bell to be a woman, at first he could not put two and two together in the presence of Charlotte Bront\u00eb. Utterly stunned, he looked at the letter and at his author and back again at the letter. It took him a few moments to recover from his shock; Charlotte tried to suppress a laugh. As the truth dawned on Smith, he received them graciously\u2014insisting that the sisters extend their London visit and entertaining them with trips to the opera, art museums, and more. Charlotte cautioned him that although they had disclosed the truth about their identities, the revelation should go no further: \"To all the rest of the world we must remain 'gentlemen' as heretofore.\"\n\nBecause Smith could not tell anyone who his companions really were, his family and friends were perplexed as to why he had brought \"a couple of odd-looking countrywomen,\" as Charlotte wryly recalled, to dine with them one evening. They were introduced as \"the Misses Brown.\" What the urbane young Londoner was doing socializing with \"these insignificant spinsters\" was anyone's guess, but in typical British fashion, no one spoke of it. Charlotte and Anne were amused at the awkwardness and dazzled by the grandeur of Smith's family residence.\n\nHe later described Anne as \"a gentle, quiet, rather subdued person, by no means pretty, yet of a pleasing appearance.\" Though he was fascinated by Charlotte and awestruck by her intellect, his appraisal of her appearance confirmed there was no danger of falling in love with his unmasked author (though her feelings for him were far more complex). For one thing, he took note of her missing teeth and her ruddy complexion. Also, \"Her head seemed too large for her body. . . . There was but little feminine charm about her; and of this fact she was herself uneasily and perpetually conscious.\" Charlotte once lamented her \"almost repulsive\" plainness to her dear friend Elizabeth Gaskell, but understood that her power lay elsewhere. \"Though I knew I looked a poor creature,\" she wrote, \"and in many respects actually was so, nature had given me a voice that could make itself heard, if lifted in excitement or deepened by emotion.\"\n\nShe returned home from her London trip tired but giddy at having unburdened herself. The future seemed full of promise.\n\nInstead, the next year of her life would bring extraordinary suffering. The dissolute lost soul, Branwell, died in September of tuberculosis. His sisters never told him about the novels they'd published. In a letter to W. S. Williams a month after Branwell's death, Charlotte admitted, \"I do not weep from a sense of bereavement\u2014there is no prop withdrawn, no consolation torn away, no dear companion lost\u2014but for the wreck of talent, the ruin of promise, the untimely, dreary extinction of what might have been a burning and a shining light.\"\n\nThe worst was still to come. Emily caught a severe cold at Branwell's funeral and had difficulty breathing. Her health deteriorated steadily from then on, and she did not leave the house again. She developed consumption but refused medical treatment, and her behavior became increasingly erratic; she would not rest or eat and bristled at familial displays of sympathy. (Charlotte described witnessing her sister's abrupt decline as causing \"pain no words can render.\") Just thirty years old, Emily died on December 19, 1848, at two o'clock in the afternoon. Three days later a memorial service was held, and her beloved bulldog, Keeper, accompanied the family to the church. (After her death, he had howled outside her door.) Emily was buried in the vault of the same church where her mother and brother now lay. Her coffin was only seventeen inches wide.\n\n\"For my part I am free to walk on the moors,\" Charlotte wrote later, \"but when I go out there alone\u2014everything reminds me of the times when others were with me and then the moors seem a wilderness, featureless, solitary, saddening\u2014My sister Emily had a particular love for them, and there is not a knoll of heather, not a branch of fern, not a young bilberry leaf, not a fluttering lark or linnet but reminds me of her.\" Charlotte did not think she could go on as a writer: \"Worse than useless did it seem to attempt to write what there no longer lived an 'Ellis Bell' to read,\" she informed her publisher.\n\nBecause Anne had shared a bedroom with Emily, it was not entirely shocking that in January 1849 Anne was diagnosed with tuberculosis. She had managed to publish another novel, The Tenant of Wildfell Hall, the year before, but it would be her last. As if she'd had a presentiment of her death, in the sharply worded preface to the novel's second edition she boldly defended the need for authorial privacy. The essay reads almost as a manifesto:\n\nRespecting the author's identity, I would have it to be distinctly understood that Acton Bell is neither Currer nor Ellis Bell, and therefore let not his faults be attributed to them. As to whether the name be real or fictitious, it cannot greatly signify to those who know him only by his works. As little, I should think, can it matter whether the writer so designated is a man, or a woman, as one or two of my critics profess to have discovered. I take the imputation in good part, as a compliment to the just delineation of my female characters; and though I am bound to attribute much of the severity of my censors to this suspicion, I make no effort to refute it, because, in my own mind, I am satisfied that if a book is a good one, it is so whatever the sex of the author may be. All novels are, or should be, written for both men and women to read, and I am at a loss to conceive how a man should permit himself to write anything that would be really disgraceful to a woman, or why a woman should be censured for writing anything that would be proper and becoming for a man.\n\nJuly 22nd, 1848.\n\nAnne died on the afternoon of May 28, 1849, at the age of twenty-nine. A lifelong friend of Charlotte later recalled the last words Anne had uttered to her sister: \"Take courage, Charlotte.\"\n\n\"When my thoughts turn to Anne,\" Charlotte said of her sister, \"they always see her as a patient, persecuted stranger,\u2014more lonely, less gifted with the power of making friends even than I am.\" She wrote a poem in Anne's memory that began, \"There's little joy in life for me, \/ And little terror in the grave; \/ I've lived the parting hour to see \/ Of one I would have died to save.\"\n\nIn life, Anne had been overshadowed by her sisters (and her legacy remains so), yet her preface is a deeply captivating personal document, remarkable for its forcefulness of expression and eloquence. Her argument is also impossible to refute.\n\nAs the only survivor of her siblings, Charlotte was inconsolable. \"Why life is so blank, brief and bitter I do not know,\" she wrote. Her faith sustained her: \"God has upheld me. From my heart I thank Him.\" She proceeded with her next novel, Shirley, which she completed in August 1849. \"[T]hough I earnestly wish to preserve my incognito,\" she wrote to her editor, \"I live under no slavish fear of discovery\u2014I am ashamed of nothing I have written\u2014not a line.\" Still, she thanked him for preserving her secret.\n\nThat Shirley is considered her weakest novel can be forgiven, considering the circumstances under which it was written. Regardless, it had been a balm for the author, who admitted to her editor that in the aftermath of enormous losses, work was her favorite companion: \"[H]ereafter I look for no great earthly comfort except what congenial occupation can give.\"\n\nIt was published in October to mostly respectable reviews, and Charlotte said that she would have to be a \"conceited ape\" to be dissatisfied with them. But the best thing to come of the book's publication was a warm letter from Elizabeth Gaskell. In response, Charlotte explained, \"Currer Bell will avow to Mrs. Gaskell that her chief reason for maintaining an incognito is the fear that if she relinquished it, strength and courage would leave her, and she should ever after shrink from writing the plain truth.\" Aside from keeping up the nom de plume, the sentiments expressed in Charlotte's letter were completely honest.\n\nGaskell was delighted at having extracted some small bit of biographical information from the mysterious author. She excitedly wrote to a friend: \"Currer Bell (aha! What will you give me for a secret?) She's a she\u2014that I will tell you.\"\n\nIn 1850, Charlotte's social circle began to widen, and she met Mrs. Gaskell in person during a visit to the Lake District. \"She is a woman of the most genuine talent,\" Charlotte said, \"of cheerful, pleasing and cordial manners and\u2014I believe\u2014of a kind and good heart.\" They became close, and Gaskell's loving and sympathetic (if flawed) biography, The Life of Charlotte Bront\u00eb (published in 1857), is still considered one of the great works of Victorian literature. Gaskell's book was significant for being the first full-length biography of a woman novelist written by another woman. The legend, long upheld by scholars and readers alike, of Charlotte as the saintly sister\u2014dutiful, modest, almost mouselike, and above reproach\u2014can be traced to Gaskell, who created it.\n\nAfter the deaths of her sisters, Charlotte made regular visits to London, where she had the privilege of meeting writers she admired, including Thackeray. She attended lectures, saw plays, and visited museums. She even sat for a portrait by the popular artist George Richmond\u2014a gift from George Smith to Charlotte's father that now resides in London's National Portrait Gallery, along with Branwell's iconic painting of Emily, Anne, and Charlotte, circa 1835, with his own image inexplicably blurred out of the portrait.\n\nEven as she extended herself beyond Haworth, Charlotte remained discreet about her alter ego. She railed against \"vulgar notoriety,\" yet speculation was rampant. She was even openly confronted, though she tried to brush such incidents aside. One evening, at a dinner party at Thackeray's home, the author called Charlotte \"Currer Bell\" in front of the other guests. She was not amused. \"I believe there are books being published by a person named Currer Bell,\" she said curtly, \"but the person you address is Miss Bront\u00eb\u2014and I see no connection between the two.\" (Thackeray had himself used various noms de plume in his early works, including Michael Angelo Titmarsh, George Savage Fitz-Boodle, and Charles James Yellowplush.)\n\nCharlotte was also on the defensive with George Lewes, who had initially praised her work, offering advice and encouragement, but who began lecturing \"Bell\" sternly in his letters and then maligning the author in reviews. She entered reluctantly into what became a rather contentious correspondence. It seems bizarre that the man who would become George Eliot's most passionate supporter just a few years later would engage in reductive criticism on grounds of gender, but he did. \"I wish you did not think me a woman,\" she wrote to him in 1849. \"I wish all reviewers believed 'Currer Bell' to be a man; they would be more just to him. You will, I know, keep measuring me by some standard of what you deem becoming to my sex; where I am not what you consider graceful, you will condemn me.\" She went on: \"I cannot when I write think always of myself\u2014and of what is elegant and charming in femininity\u2014it is not on those terms or with such ideas I ever took pen in hand; and if it is only on such terms my writing will be tolerated\u2014I shall pass away from the public and trouble it no more. Out of obscurity I came\u2014to obscurity I can easily return.\"\n\nLewes ignored her response, reviewing Shirley in the Edinburgh Review and finding fault with the work based on the author's gender. (The headlines of the article's first two pages read, \"Mental Equality of the Sexes?\" and \"Female Literature.\") Charlotte was outraged and hurt by what she viewed as his cruelty toward her, and at having her fiction judged by a double standard. The note she subsequently addressed to \"G. H. Lewes, Esq.\" was damning and brief: \"I can be on guard against my enemies, but God deliver me from my friends.\" It was signed \"Currer Bell.\" (About a year later, after Charlotte met him in person, she said, \"I cannot hate him.\")\n\nAt home as well, her secret had begun to unravel. Her father had started telling neighbors who his daughter was. Excited fans made pilgrimages to the village, hoping to come upon the genius in person. And on February 28, 1850, a local newspaper announced, in a burst of pride, that Charlotte Bront\u00eb, the reverend's daughter, was \"the authoress of Jane Eyre and Shirley, two of the most popular novels of the day, which have appeared under the name of 'Currer Bell.'\" The charade was officially over.\n\nIn 1851, thirty-five-year-old Charlotte received the third marriage proposal of her life and the third she would decline. (When the latest suitor approached her to propose, Charlotte admitted, \"my veins ran ice.\") Caring for her aging father, and suffering from health problems of her own, including a liver infection, she was lonely\u2014but she didn't want a husband.\n\nDiscouraging her further was the news that despite all her success, Smith, Elder still declined to publish The Professor. The firm suggested that she instead begin work on a new novel, and she did\u2014often in a state of despair. Two years later, Villette was published. The title page read, \"VILLETTE. BY CURRER BELL, AUTHOR OF 'JANE EYRE,' 'SHIRLEY,' ETC.\" Feeling burned after having her pseudonymous cover unmasked, Charlotte longed to become invisible again. She had asked George Smith if he might consider publishing Villette under yet another pen name: \"I should be much thankful for the sheltering shadow of an incognito,\" she implored. But \"Currer Bell\" was now an enviable brand in Victorian society; \"he\" was a towering figure whose name on a book almost guaranteed sales. The publisher reluctantly denied her request.\n\nVillette, which Virginia Woolf would later deem to be Bront\u00eb's \"finest novel,\" drew on Charlotte's own breakdowns and was her most overt exploration to date of the struggle between a woman's will and the constraints of society. Even though it made demands on the reader and lacked a happy ending, it proved a great success. George Eliot, then still known as Mary Ann Evans, read it three times. \"I am only just returned to a sense of real wonder about me, for I have been reading Villette, a still more wonderful book than Jane Eyre,\" she wrote to a friend. \"There is something almost preternatural about its power.\" She would later praise Charlotte to George Lewes, who had met Charlotte and saw her as a plain \"old maid.\" Eliot, however, recognized the beauty of Charlotte's inner life: \"What passion, what fire in her!\" she said. \"Quite as much as in George Sand, only the clothing is less voluptuous.\" Charlotte happened to have great respect for Sand, whom she considered \"sagacious and profound\"; this favorable view was in contrast to her opinion of Jane Austen's work, which she found uninteresting, with its \"ladies and gentlemen, in their elegant but confined houses.\"\n\nIn 1853, Charlotte was just two years from her death. She'd begun writing yet another novel, but abandoned it after reluctantly marrying her father's curate, Arthur Bell Nicholls, who had pursued her for years. (The sisters' pseudonymous surname was taken from his middle name.) She consented to marry Nicholls in June 1854, only a short time after George Smith had married. (That event was quite painful for Charlotte to digest.) She married Nicholls accepting that there was only companionship, not passion, between them. At least she would no longer be alone. \"Doubtless then it is the best for me,\" she wrote to a friend. Soon after marrying, she offered a sober assessment of her new role: \"It is a solemn and strange and perilous thing for a woman to become a wife. . . . My time is not my own now.\"\n\nIn the early hours of March 31, 1855, Charlotte died at the age of thirty-eight. She is believed to have been pregnant at the time.\n\nThe defiant opening stanza of Emily Bront\u00eb's most famous poem conveys the inspiring resilience and fierce spirit of Emily, Anne, and Charlotte:\n\nNo coward soul is mine,\n\nNo trembler in the world's storm-troubled sphere:\n\nI see Heaven's glories shine,\n\nAnd faith shines equal, arming me from Fear.\n\nThe Bront\u00eb sisters had aggressively offended, challenged, and violated Victorian morals with their revolutionary works, which were profoundly disturbing for their era. In concealing their true identities, the Bront\u00ebs could speak the truth without facing judgment. By overturning rigid societal notions of a distinctly \"male\" or \"female\" imagination, they raised provocative questions about the nature of creativity.\n\n\"If men could see us as we really are,\" Charlotte once wrote, \"they would be amazed.\"\nShe was a bisexual, cigar-smoking cross-dresser\n\nChapter 2\n\nGeorge Sand & AURORE DUPIN\n\nIt began with an ankle-length gray military coat, matching trousers, a cravat, and a waistcoat. Clothes may make the man, as Mark Twain famously noted, but in this instance, they made the woman the man.\n\nShe was born Amandine-Aurore-Lucile Dupin in Paris in the summer of 1804, shortly before Napoleon became emperor of France. She was known as Aurore. \"My father was playing the violin and my mother wore a pretty pink dress,\" she would report of her birth in her epic, two-volume memoir, Histoire de ma vie. \"It took but a minute.\" Her parents had married secretly weeks before, making their daughter legitimate. In later years Aurore claimed to have walked at ten months, and to have been an adept reader by age four. \"My looks gave promise of great beauty, a promise unkept,\" she recalled, with no trace of regret. \"This was perhaps my own fault, because at the age when beauty blossoms, I was already spending my nights reading and writing.\"\n\nEven in childhood Aurore was unconventional, finding delight and power in her own precociousness. She was an explorer, constantly testing boundaries in her behavior and pushing back at authority figures. Though her mother valued beauty above all, Aurore deplored the notion of \"living under a bell jar so as to avoid being weather-beaten, chapped, or faded before your time.\" She shunned hats and gloves and lessons in becoming a proper young lady. Gestures of reticence and grace were of no interest. She didn't rebel for the sake of rebellion, but \"I could not be coerced.\" She daydreamed endlessly, befriended boys and girls alike, and cultivated a certain wildness of intellect and character. Already she displayed hints of the adult she would become, magnanimous and brave.\n\nAurore's identity evolved largely in opposition to her mother's character. Yet in one regard, her mother unwittingly exerted a profound influence. A former stage actress and prostitute, Antoinette Sophie-Victoire Delaborde Dupin had a lifelong tendency toward melodrama and instability\u2014thus teaching Aurore that selves could be cycled through and discarded at will. Aurore's mother was raised in poverty, the daughter of a bird seller; her husband was descended from a family of aristocrats. (Both had illegitimate children from previous relationships\u2014he had a son, Hippolyte; she had a daughter, Caroline.) The class schism was a source of tension in their relationship, and would become a recurring theme in Aurore's fiction. Antoinette's mother-in-law came to accept, and even adore, Aurore but could not endorse her son's marriage, which she deemed \"disproportionate,\" and she considered his wife contemptible. The best state the two women would ever settle into was a kind of benign antipathy.\n\nOver the years the mercurial Antoinette preferred to be called Victoire, and then, after her marriage, Sophie. At times she had only a tenuous grip on reality, and this condition worsened as she aged. She was impulsive and manipulative. Roles and selves were interchanged to suit her circumstances. \"When she was in good spirits,\" Aurore recalled of her mother, \"she was truly charming, and it was impossible not to be swept up in her buoyant gaiety and vivid witticisms. Unfortunately, it would never last an entire day; lightning would strike from some remote corner of heaven.\"\n\nAt least one thing kept the mother-daughter relationship close: the art of storytelling. As a girl, Aurore delighted in hearing her mother read stories and sing lullabies to her. She loved the sounds of words, developed a rich imagination, and became a compulsive storyteller herself. \"I used to compose out loud interminably long tales which my mother used to call my novels,\" she later recalled.\n\nShe had a knack for embellishment, or perhaps a dubious memory (though she once dismissed forgetfulness as unintelligence or inattention). In her autobiography, Aurore shared what she claimed was the first memory of her life, an incident that occurred when she was two years old and that she recalled in remarkable detail:\n\nA servant let me fall out of her arms onto the corner of the fireplace; I was frightened, and I hurt my forehead. All the commotion, the shock to the nervous system opened me to self-awareness, and I saw clearly\u2014I still see\u2014the reddish marble of the mantelpiece, my blood running, the distraught face of my nursemaid. I distinctly remember the doctor who came, the leeches which were put behind my ears, my mother's anxiety, and the servant dismissed for drunkenness.\n\nWas this recollection accurate in every, or even any, aspect? Did the incident happen at all? No matter. Aurore cast her younger self at the center of a drama vividly told. She was both subject and object. She also claimed to remember perfectly the apartment her family lived in a year later, on Rue Grange-Batali\u00e8re, and she said that from then on, \"my memories are precise and nearly without interruption.\" It's an astonishing claim, regardless of how attentive young Aurore must have been to the world around her.\n\nSince her mother was often unavailable physically and emotionally, Aurore spent countless hours in solitude. She craved touch. When she wasn't telling stories to her rapt listener, a pet rabbit, she was beginning to discover the thrills of playing with different personae. The first time she called out into the empty flat and heard her own voice call back the same words, she recalled thinking, \"I was double and somewhere nearby was another 'me' whom I could not see but who always saw me since it always answered me.\" She didn't realize that it was only the echo of her own voice calling back until her mother later told her this, but the idea of \"doubleness\" had been planted, and it delighted her. She gave the voice a name and would call out, \"Echo, are you there? Do you hear me? Hello, echo!\"\n\nAlthough Sophie's husband, Maurice Dupin, a military officer, was rarely present, his letters home showed how much he loved his family: \"How dear is our Aurore!\" he wrote in September 1805, two days before the Battle of Austerlitz. \"How impatient you make me to come back and take both of you into my arms! . . . Tell me about your love, our child. Know that you'd destroy my life if you should cease to love me. Know that you're my wife, that I adore you, that I love life only because of you, and that I've dedicated my life to you.\" Elsewhere he implored, \"May you always feel gloomy in my absence. Yes, beloved wife, that is how I love you. Let no one see you, think only of taking care of our daughter, and I'll be happy as I can be far from you.\"\n\nSophie seemed to take his words to heart. In her husband's absence, the Dupin house, with only mother and daughter, was lonely and listless. (The illegitimate children mostly lived elsewhere, though Hippolyte would become quite close to his half sister Aurore.) Maurice's periodic returns invigorated their lives, at least temporarily. \"[My parents] found themselves happy only in their little household,\" Aurore would later recall. \"Everywhere else they suffocated from melancholy yawning, and they left me with this legacy of secret savagery, which has always made society intolerable for me and 'the home' a necessity.\"\n\nIn the autumn of 1808, Maurice was thrown from his horse, Leopardo, and instantly killed. He was thirty. Just eight days earlier, he and Sophie had suffered the devastating loss of their infant son, Louis. After Maurice died, Aurore saw her mother crying one day and shyly approached her. \"But when my daddy is through being dead,\" she said, \"he'll come back to see you, won't he?\"\n\nShe recalled that the house was \"plunged into melancholy.\" Sophie's fragile, shifting self may have been a means of resilience against the hardships she suffered. But whatever the cause, her condition worsened after the deaths of Maurice and Louis, and her perpetual instability provided a template for Aurore's own ideas about identity: \"It seems to me that we change from day to day and that after some years we are a new being,\" she reflected late in life. This notion was liberating\u2014Aurore was a fearless risk-taker, rushing headlong into new experiences\u2014but it also had a grievous effect, leaving her with a lifelong pining for love and intimacy that, occasional salves aside, would never be filled.\n\nFollowing the losses of her brother and father, Aurore's love of daydreaming and storytelling became obsessive; imagination was no longer merely a retreat from boredom and solitude but a life raft, a need. \"She's not trying to be difficult; it's her nature,\" her mother would explain. \"You may be sure that she's always meditating on something. She used to chatter when she daydreamed.\" Aurore never relinquished her belief in the virtues of clinging to the imagination: \"To cut short the fantasy life of a child is to go against the very laws of nature,\" she wrote.\n\nHer grandmother was increasingly troubled by her peculiar behavior, and in 1817 Madame Dupin decided to rectify it by sending the thirteen-year-old to a convent, Couvent des Anglaises, in Paris. Aurore later recounted her grandmother's harsh assessment of her at the time: \"You have inherited an excellent intelligence from your father and grandparents, but you do all in your power to appear an idiot. You could be attractive, but you take pride in looking unkempt. . . . You have no bearing, no grace, no tact. Your mind is becoming as deformed as your body. Sometimes you hardly reply when spoken to, and you assume the air of a bold animal that scorns human contact. . . . It is time to change all this.\"\n\nAt the convent, Aurore alternated between subversive behavior (\"Let me say in passing that the great fault of monastic education is the attempt to exaggerate chastity,\" she later wrote), and austere withdrawal. Despite the excessive instruction, she admitted, \"I still slouched, moved too abruptly, walked too naturally, and could not bear the thought of gloves or deep curtsies.\" Aurore said that when her grandmother would scold her for these \"vices,\" \"it took great self-control for me to hide the annoyance and irritation these eternal little critiques caused me. I would so like to have pleased her! I was never able to.\" Aurore could not (and had no wish to) shake off her propensity for daydreaming\u2014\"my mind, sluggish and wrapped up in itself, was still that of a child.\" She wrote poems at the convent, and even completed two novels; the second was \"a pastoral one, which I judged worse than my first and with which I lit the stove one winter's day.\"\n\nDuring this period, her grandmother exerted tremendous control over Aurore's education and development. (She accomplished this by threatening to disinherit Aurore.) Dying, Madame Dupin was concerned, as ever, about what she saw as the toxic effects of Sophie's influence on her granddaughter. She was determined to instill in Aurore moral and intellectual development; a socially acceptable degree of independence; a permanent distrust of Sophie and her family; and to remove Aurore from a \"lower-class environment\" into established society\u2014which also meant finding a proper husband.\n\nAs a result, Sophie gave up custody of Aurore for a time. \"It seemed as though she was ready to accept for herself a future in which I was no longer an essential party,\" Aurore wrote of her mother. Resigned to Madame Dupin's authority and dominance, Sophie would not engage in a contest of wills with her mother-in-law. The distance between mother and daughter was painful. \"My mother seemed to have abandoned me to my silent and miserable struggle,\" Aurore recalled. \"I was desolate over her apparent abandonment of me after the passion she had showered on me in my childhood.\" Maternal nurturing was provided instead by a nun, Sister Alicia, the first woman for whom Aurore had powerful feelings of love. In later years, with other women, Aurore would find physical intimacy, great passion, and much torment.\n\nShe left the convent in 1820, at age sixteen; within two years, she was married. It was the result of an extensive contractual agreement, following lengthy negotiations and financial haggling between the families. Her husband, Casimir Dudevant, the son of a baron, was a handsome twenty-seven-year-old sublieutenant in the French army. Sophie never warmed to him, explaining her dislike by saying that his nose didn't please her. But Aurore found him an agreeable and reasonable companion, if not an ideal romantic suitor: \"He never spoke to me of love, he admitted to being little disposed to sudden passion, or enthusiasm, and in any case, was incapable of expressing these sentiments in a seductive manner,\" she recalled.\n\nNonetheless, by the spring of 1823 her first child was on the way: a son, Maurice, named for her father. And she already had stirrings of doubt about her marriage. Self-abnegation did not suit Aurore, who recognized it as an essential but unpleasant aspect of her union. \"In marrying, one of the two must renounce himself or herself completely,\" she noted. \"All that remains to be asked, then, is whether it should be for the husband or the wife to recast his or her being according to the mould of the other.\" On a family trip to the Pyrenees on her twenty-first birthday, she wrote in her diary: \"I have to get used to smiling though my soul feels dead.\"\n\nDespite periods of depression, she delighted in motherhood and gave birth to a daughter, Solange, in the fall of 1828. Aurore was dissatisfied with her marriage intellectually, emotionally, and sexually. It was a functioning partnership, nothing more. She was slowly recasting herself, but hardly as the dutiful wife\u2014though she had genuinely tried: \"I made enormous efforts to see things through my husband's eyes and think and do as he wished,\" she later wrote. \"But the minute I had come to agree with him, I would fall into dreadful sadness, because I no longer felt in agreement with my own instincts.\"\n\nThe headstrong young woman was keenly interested in exploring other, more flamboyant and expansive roles. Just a year before she married, Aurore had made her first public appearance in a male disguise. She'd been riding her horse, Colette, one day, dressed in equestrian clothes, and was mistaken for a man. In a nearby village, she'd been addressed as \"monsieur\" by a woman, who had blushed and narrowed her eyes in \"his\" presence. Aurore was thrilled about her cross-dressing experiment and delighted by her own power. The illicit erotic charge wasn't bad, either.\n\nAlthough she was certainly adjusting to the mold of another, her new form did not belong to the dull Casimir but to George Sand, her literary persona, who would become France's best-selling writer and would be among its most prolific authors. In considering what Sand accomplished, and the inspiring way she went about it, a dictum from the inimitable artist Louise Bourgeois comes to mind: \"A woman has no place in the art world unless she proves over and over again she won't be eliminated.\"\n\nBoth Aurore and Casimir had casual affairs during their marriage, but at twenty-six, Aurore met a young Parisian who would play an important role in her life. When they fell in love, Jules Sandeau was nineteen and, like her, a writer. The first syllable of his surname (Sand) would become the surname of her pseudonym. Though their love affair didn't last long, Sandeau proved enormously influential and helped her find her path toward a wholly independent life. \"Inspiration can pass through the soul just as easily in the midst of an orgy as in the silence of the woods,\" she wrote in her autobiography, \"but when it is a question of giving form to your thoughts, whether you are secluded in your study or performing on the planks of a stage, you must be in total possession of yourself.\" By 1830, she was well on her way.\n\nThe following year, she decided to assert her will rather forcefully. She told Casimir that she would live in Paris for half the year with Solange, returning in the other months to care for Maurice. Yet she went through many periods of replicating her own mother's treatment of her\u2014abandoning both children for long (and damaging) stretches to caretakers and tutors, in the single-minded pursuit of her own desires and ambitions. She wrestled with this but did not always remedy the situation to her children's liking.\n\nAurore's loneliness in Paris, at first, was \"profound and complete.\" She felt useless. There was no doubt in her mind that literature alone \"offered me the most chance of success as a profession.\" The few people she confided in about it were skeptical that writing and monetary concerns could successfully coexist\u2014at least for a woman.\n\nShe dabbled in other, more pragmatic attempts at work. Feeling despair over not being able to help the poor in any meaningful way, she became \"a bit of a pharmacist,\" preparing ointments and syrups for her clients gratis. She tried translation work, but because she was meticulous and conscientious with the words of others, it took too long. In attempting pencil and watercolor portraits done at sittings, she said, \"I caught the likenesses very well, my little heads were not drawn badly, but the m\u00e9tier lacked distinction.\" She tried sewing, and was quick at it, but it didn't bring in much money and she couldn't see well enough close up. In another profitless venture, she sold tea chests and cigar boxes she'd varnished and painted with ornamental birds and flowers. \"For four years, I went along groping, or slaving at nothing worthwhile, in order to discover within me any capability whatsoever,\" she recalled. \"In spite of myself, I felt that I was an artist, without ever having dreamed I could be one.\"\n\nJules Sandeau would play an integral role in her becoming a \"public\" writer, as she had already written prolifically in private. He was part of a bohemian circle that Aurore eagerly joined, one that provided stimulating political, artistic, and intellectual discourse. These were people she felt an affinity with (as she most certainly did not with her husband), and they would become her close friends. It was an exciting time, and she took full advantage, throwing herself passionately into the affair with Sandeau.\n\nThe tricky issue of financial independence lingered. In the winter of 1831, Aurore reluctantly arranged an interview, through an acquaintance, with the publisher of Le Figaro, Henri de Latouche. She cringed at the thought of newspaper work, but recognized it as a useful entry point to literary endeavors. Also, she appreciated Latouche's intensity and fervent antibourgeois sensibility. He offered Aurore a job as columnist\u2014making her the only woman on the staff and paying her seven francs per column. She was more than willing to prove herself. \"I don't believe in all the sorrows that people predict for me in the literary career on which I'm trying to embark,\" she wrote in a letter to a friend. But when she called on an author to seek advice about the Parisian publishing world, the meeting was a disaster: \"I shall be very brief, and I shall tell you frankly\u2014a woman shouldn't write,\" he said before showing her the door. She recalled in her autobiography that because she left quietly, \"prone more to laughter than anger,\" he ended his harangue on the inferiority of women with \"a Napoleonic stroke that was intended to crush me: 'Take my word for it,' he said gravely, as I was opening the outer door to his sanctum, 'don't make books, make babies!'\"\n\nNever mind: Aurore was more determined than ever. As she once wrote, in another context, \"I was not a coward, and I could not have been if I tried.\"\n\nShe continued to immerse herself in her social circle, and she and Sandeau collaborated on their writing. They received enthusiastic support from Balzac, who would drop by Aurore's flat from time to time. She later described him fondly as \"childlike and great; always envious of trifles and never jealous of true glory; sincere to the point of modesty, proud to the point of braggadocio; trusting himself and others; very generous, very kind, and very crazy.\" Other notable men she called her friends included Baudelaire, Turgenev, Dostoevsky, Zola, Henry James, and Dumas. (John Ruskin, William Thackeray, and Thomas Carlyle, however, disliked her work intensely.) Later, Flaubert became a lifelong friend and confidant. Their letters were beautiful and mutually consoling. \"There you are feeling sad and lonely, you say, and here I am feeling the same way,\" Flaubert wrote to her in 1866. \"Where do they come from, do you think, these black moods that engulf us like this? They rise like a tide, you feel as if you are drowning and you have to escape somehow. What I do is lie, floating, letting it all wash over me.\" In 1876, a few months before she died, Flaubert wrote: \"[Y]ou've never done me anything but good and I love you most tenderly.\"\n\nAt the end of the summer of 1831, Aurore and Jules began work on the bawdy Rose et Blanche, a planned five-volume novel for which they'd secured a publishing contract, and which they'd signed with the joint pseudonym \"J. Sand.\" (Latouche, who had become a devoted mentor to Aurore, invented the name.) But Aurore ended up doing the bulk of the writing.\n\nThe novel was released to mixed reviews, yet it had moderate success and gave Aurore the confidence to publish entirely on her own. The following year, she published Indiana\u2014a semi-autobiographical novel, and an unapologetic denunciation of marriage that she expected \"to please very few people.\" Instead, it won international acclaim and became a best seller. An envious Victor Hugo (her rival for the status of France's best-selling author) called it \"the finest novel of manners that has been published in French for twenty years.\" The author of this lauded novel was \"George Sand,\" a name that would not only endure as her nom de plume but serve as her identity for the rest of her life. After completing Indiana, \"I was baptized,\" she explained. \"The [name] I was given, I earned myself, after the event, by my own toil. . . . I do not think anyone has anything to reproach me for.\"\n\nShe was amused by the number of reviewers who spoke enthusiastically of \"Mr. G. Sand,\" but insisted that a woman must have had a hand in refining some of the novel's more emotional aspects. They were stumped because \"the style and discrimination were too virile to be anything but a man's.\"\n\nIn 1832, her romantic relationship with Sandeau collapsed, and just as she was beginning to achieve professional success, she felt increasingly isolated. But in January 1833, she met Marie Dorval, a famous stage actress in her mid-thirties whose presence toppled and intoxicated Sand, and who would become\u2014as she later described it\u2014the one true love of her life. Both women were married (and had other lovers) at the time, but Sand legally separated from her husband in 1835. She pursued Dorval\u2014initially, in the name of \"friendship\": \"For my part I feel I love you with a heart brought back to life and rejuvenated by you,\" Sand wrote to her early on. \"If it is a dream, like everything else I have wished for in life, do not steal it from me too quickly. It does me so much good.\" Meanwhile, Dorval's lover at the time, Alfred de Vigny, gave a detailed assessment of his rival: \"Her hair is dark and curly and falls freely over her collar, rather like one of Raphael's angels,\" he wrote of Sand. \"She has large black eyes, shaped like those of mystics whom one sees in paintings, or in those magnificent Italian portraits. Her face is severe and gives little away, the lower half is unattractive, the mouth ill-shaped. She has no grace of bearing, and her speech is coarse. In her manner of dress, her language, her tone of voice and the audacity of her conversation, she is like a man.\" Vigny had good cause to be concerned.\n\nSand played a male role in public because doing so offered her a much broader range of experience, and she loved freedom. Elizabeth Barrett Browning affectionately called her \"thou large-brained woman and large-hearted man.\" She wrote a sonnet, \"To George Sand: A Recognition,\" in 1844:\n\nTrue genius, but true woman! dost deny\n\nThy woman's nature with a manly scorn,\n\nAnd break away the gauds and armlets worn\n\nBy weaker women in captivity?\n\nAh, vain denial! that revolted cry\n\nIs sobbed in by a woman's voice forlorn!\u2014\n\nThy woman's hair, my sister, all unshorn\n\nFloats back disheveled strength in agony,\n\nDisproving thy man's name: and while before\n\nThe world thou burnest in a poet-fire,\n\nWe see thy woman-heart beat evermore\n\nThrough the large flame. Beat purer, heart, and higher,\n\nTill God unsex thee on the heavenly shore\n\nWhere unincarnate spirits purely aspire!\n\nSand was a cigar-chomping rebel who had brazen affairs as she wished, and with whomever she desired. She could practically roll a cigarette with her eyes closed, and she loved to smoke a hookah. She reveled in her own mischief. In one of her novels, Sand boldly suggested that monogamous marriage was an abnormal, unnatural state that deprived men and women of experiencing true sexual pleasure. Her significant lovers included Alfred de Musset, Franz Liszt, and Fr\u00e9d\u00e9ric Chopin, who reported to his family, \"Something about her repels me.\" Her decade-long relationship with Chopin ended badly in 1847, when Sand suspected that he had fallen in love with her daughter.\n\nEven after it became an open secret in literary circles (and a source of malicious gossip) that Aurore Dupin was the notorious George Sand, she continued her transgressive style of dress and behavior, simply because she enjoyed it. She loved the idea of being in disguise. With her trousers, vest, military coat, hat, and tie, \"I was the perfect little first-year student,\" she recalled in her autobiography. \"My clothing made me fearless.\" And walking in her solid, sturdy boots was far preferable to the fussy discomfort of women's shoes: \"With those little iron heels, I felt secure on the sidewalks. I flew from one end of Paris to the other.\" In her male attire, she was a voyeur, seeing without being seen. \"No one knew me, no one looked at me, no one gave me a second thought; I was an atom lost in the immense crowd.\"\n\nAt theaters, she sat in the pit, where only men were permitted, and she always pulled off the ruse with ease\u2014\"the absence of coquettishness in costume and facial expression warded off any suspicion,\" she explained. \"I was too poorly dressed and looked too simple\u2014my usual vacant, verging on dumb, look\u2014to attract or compel attention. . . . There is a way of stealing about, everywhere, without turning a head, and of speaking in a low and muted pitch which does not resound like a flute in the ears of those who may hear you. Furthermore, to avoid being noticed as a man, you must already have not been noticed as a woman.\"\n\nIn her autobiography, Sand recalled that one of her friends, who was privy to her sartorial secret, began calling her \"monsieur\" in public. But just as he would get used to addressing her this way, she would appear the following day dressed as a woman, and he couldn't keep up with the relentless change of costume. Confused by her various corrections, he took to addressing her only as \"monsieur\" from then on.\n\nThere was a less amusing aspect to dabbling in androgyny: having to deal with the fallout from her marriage. Casimir meticulously kept a log of his (soon to be former) wife's crimes and misdemeanors\u2014among them, \"She writes novels.\" Even worse, \"Mme D. affecting the manners of a young man, smoking, swearing, dressed as a man and having lost all the feminine graces, has no understanding of money.\" Once tolerant and blithe about their marital arrangement, which allowed her to veer off on an independent path, Casimir came to detest the liberty she'd achieved and was disgusted by her \"bohemian\" lifestyle. She had to enter a nasty and protracted legal battle to end the marriage, and in the end had to divide her fortune with him.\n\nNo matter how messy her personal life became at any given time, she held steady with her writing, producing a staggering number of novels, plays, essays, and other works. She also painted, and she was an astonishingly prolific letter writer; her published correspondence includes more than fifteen thousand letters. Yet she also happily engaged in so-called women's work\u2014making jam, doing needlework, and immersing herself in her beloved garden. Although she would periodically take stock of \"the irregularity of my essentially feminine constitution,\" she was never shaken by what she viewed as the mutability of the self. Given the choice between conforming to prevailing customs and doing as she wished, she simply alternated between the two. It was not always easy, yet she was constitutionally incapable of remaining in a fixed state:\n\nI was not a woman completely like those whom some moralists censure and mock; I had in my soul an enthusiasm for the beautiful, a thirst for the true; and yet I was a woman like others\u2014dependent, nervous, prey to my imagination, childishly susceptible to the emotionalism and anxieties of motherhood. But did these traits have to relegate me to secondary standing in artistic and family life? That being society's rule, it was still within my power to submit patiently or cheerfully.\n\nAs Sand's biographer Belinda Jack noted, \"[H]er modernity lies less in her feminism or her socialism, and more in her acceptance of loose, even freewheeling ideas about the self. . . . She had strong intuitions about the subconscious and the need to be aware of our inner unthinking, but acutely responsive, selves.\"\n\nTo Sand, this was a natural, normal idea. It was far ahead of her time; she worked tirelessly so that others might embrace it. In her autobiography, Sand expressed a desire to achieve societal acceptance not for herself only, but for other women. \"I was going along nourishing a dream of male virtue to which women could aspire,\" she wrote, \"and was constantly examining my soul with a na\u00efve curiosity to find out whether it had the power of such aspirations, and whether uprightness, unselfishness, discretion, perseverance in work\u2014all the strengths, in short, that man attributes exclusively to himself\u2014were actually unavailable to a heart which accepted the concept of them so ardently. . . . I wondered why Montaigne would not have liked and respected me as much as a brother.\"\n\nNo less than George Eliot's future partner, the critic George Henry Lewes, declared in 1842 that Sand was the most remarkable writer of the century. Dostoevsky considered her \"one of the most brilliant, the most indomitable, and the most perfect champions.\"\n\nThe last years of her life were often filled with sadness, as by then many of her friends and former lovers were dead. But she was one of the most influential and famous women in France, and possessed remarkable serenity after all that she'd endured. Unfortunately, her reputation did not hold up well after her death. Her prodigious output was eclipsed by the shocking, scandalous details of her life. Compared with her contemporaries, she is hardly read today. \"The world will know and understand me someday,\" Sand once wrote. \"But if that day does not arrive, it does not greatly matter. I shall have opened the way for other women.\" In that regard, she succeeded beyond measure.\n\n\"What a brave man she was,\" Turgenev recalled of Sand, \"and what a good woman.\"\n\nHer old friend Flaubert, a notorious misanthrope and recluse, outlived her by four years. Of her funeral in 1876, he said: \"I cried like an ass.\"\nShe had a big nose and the face of a withered cabbage\n\nChapter 3\n\nGeorge Eliot & MARIAN EVANS\n\nCharles Dickens was suspicious. \"I have observed what seem to me to be such womanly touches, in those moving fictions, that the assurance on the title-page is insufficient to satisfy me, even now,\" he wrote to George Eliot in January 1858. The candid letter was written a year after the publication of Scenes of Clerical Life, a collection of three stories first serialized, anonymously, in Blackwood's Edinburgh Magazine. Dickens praised their \"exquisite truth and delicacy\" but was convinced that the writer was a woman. Elizabeth Gaskell, however, insisted that the author was a man named Joseph Liggins of Nuneaton. The Saturday Review, meanwhile, harbored its own suspicions, noting that George Eliot was rumored to be \"an assumed name, screening that of some studious clergyman . . . who is the father of a family, of High Church tendencies, and exceedingly fond of children, Greek dramatists and dogs.\"\n\nNot quite: George Eliot was a thirty-eight-year-old woman named Mary Anne (Marian) Evans, a politically progressive atheist raised in a stern, religious household, unmarried, childless, and living openly with a married man. She was a formidable intellectual who had begun educating herself after her mother's death in 1836 and would publish seven astonishing novels in her lifetime, including The Mill on the Floss, Middlemarch, and Daniel Deronda. How Evans became one of the great Victorian novelists is the story of an eccentric young woman from the Midlands region of England who broke just about every taboo of her time. \"She was never content with what was safely known and could be taken for granted,\" one critic wrote of her extraordinarily restless life.\n\nBorn on November 22, 1819, in Warwickshire, she was her parents' third child, following the birth of a daughter and a son. (Her father, Robert, also had two children from a previous marriage; his first wife died.) The birth of a second daughter was terribly disappointing. Sons were valued and valuable; girls, until married off, were a financial drain and nothing but a burden on the family. Mary Anne was no great prize. Twin boys arrived fourteen months later, but they died soon after birth, and Mary Anne's mother, Christiana, never recovered from the loss. She made no effort to hide that fact from her daughter.\n\nMary Anne eventually dropped the \"e\" from \"Anne\" and later changed her name to Marian, but at the end of her life, she reverted to \"Mary Ann.\" (That's why, in biographies, you'll find her first name spelled with confusing variation: what to call her?) Since she lived with a mother who never doted on her, her childhood was marked by isolation and sadness. Luckily, her father was kinder, and gave her a copy of her very first book: The Linnet's Life. But whatever bond she shared with him, it was never enough to replace the maternal affection she was denied.\n\nUnkempt, frequently melancholy, and extremely sensitive, she was an unsightly irritant to Christiana, who may have blamed her own poor health and depression on having given birth to Mary Anne. The Evanses' youngest child was obstinate, fearful, and given to emotional outbursts. At the age of five, in 1824, she was sent to a boarding school. A few years later, her parents would move her to another boarding school, where Mary Anne became close to a teacher named Maria Lewis. Even for the Victorian era, five was quite young to be shipped away for one's education, though she did come home on weekends. A timid and socially awkward student, Mary Anne would eventually find academic success and earn the admiration of her peers, but her insecurity lingered and she was always harshly critical of her own achievements.\n\nAt seven, Mary Anne began reading Sir Walter Scott's Waverley. This event marked the first hint of her future vocation: when the book was returned to a neighbor before she'd had a chance to finish reading it, she was terribly upset. She did the next best thing by writing out an ending herself.\n\nWhen she was twelve, Mary Anne attended a girls' school in the Midlands run by evangelical sisters. She excelled there, impressing her teachers with her mastery of every subject, especially literature. She received a novel in the mail from her beloved former teacher Maria Lewis, and sent a thank-you letter back, describing the sustaining role that books had played in her life. \"When I was quite a little child I could not be satisfied with the things around me,\" she wrote. \"I was constantly living in a world of my own creation, and was quite contented to have no companions that I might be left to my own musings and imagine scenes in which I was chief actress.\" It was Lewis, in 1839, who encouraged Mary Anne to submit her work for publication. The poem, her print debut, was signed \"M.A.E.\" and appeared in the Christian Observer. It began:\n\nAs o'er the fields by evening's light I stray,\n\nI hear a still small whisper\u2014come away;\n\nThou must to this bright, lovely world soon say\n\nFarewell!\n\nThe effects of her feeling of estrangement from those around her\u2014and dealing with her mother's death, when she was seventeen years old\u2014would lead her to be perpetually in search of mother figures and to form fierce attachments to the people she loved\u2014including her brother Isaac. (Their close yet complex bond informed the sibling relationship of Maggie and Tom Tulliver in The Mill on the Floss.) She was desperate for intimacy, a longing that never left her. \"Before I had your kind letter,\" she wrote to a friend in 1842, \"one of the ravens that hovered over me in my Saul-like visitations was the idea that you did not love me well enough to bestow any time on me more than what I had already robbed you of, but that same letter was a David's harp that quite charmed away this naughty imagination.\" (By this time, too, she had begun spelling her name Mary Ann.)\n\nAfter her mother died, she became more withdrawn. While caring for her widowed sixty-three-year-old father, she dutifully\u2014though not happily\u2014took over running the household, and felt like little more than a maid. But she used the seclusion to further her education. In what spare time she had, she read (and reread) widely: history, literature, poetry, philosophy, science, and music (she became an accomplished pianist), and studied Latin, Greek, German, French, and Italian. With her capacity for deeply felt emotion, she could not ignore the fact that daily life was constricting and pallid. Still, her emotional deprivation was offset by the riches of learning, of cultivating a powerful and capacious intellect. The hunger of the heart was sublimated into the hunger of the mind.\n\nAlways a thoughtful, contemplative girl, Mary Ann grew increasingly analytical and developed a keen interest in ideas concerning morality, modesty, and character. She was also intrigued by the conflict between individual will and the stifling demands of convention. \"Certainly those determining acts of her life were not ideally beautiful,\" Evans, as Eliot, would write in Middlemarch, widely regarded as her greatest work. \"They were the mixed result of young and noble impulse struggling amidst the conditions of an imperfect social state, in which great feelings will often take the aspect of error, and great faith the aspect of illusion. For there is no creature whose inward being is so strong that it is not greatly determined by what lies outside it.\" As a novelist, Eliot would prove to be an astute social observer, a historian, and a philosopher. Yet she also captured the despair of insatiable yearning, a condition she understood all too well.\n\nThe fervent desire to love and be loved, which had driven her back upon herself throughout her childhood, stayed constant even after it had been fulfilled. Despite her reputation as an author whose novels reflected her vast intellect, she was very much invested in matters of the heart.\n\nThe English poet William Ernest Henley, best known for his 1875 poem \"Invictus,\" once dismissed Eliot as \"George Sand plus Science minus Sex.\" Yet the heart, if not sex, was more present in Eliot's work than is generally recognized. In Middlemarch she wrote (in the voice of her heroine, Dorothea Brooke) that \"surely the only true knowledge of our fellow-man is that which enables us to feel with him\u2014which gives us a fine ear for the heart-pulses that are beating under the mere clothes of circumstance and opinion. Our subtlest analysis of schools and sects must miss the essential truth, unless it be lit up by the love that sees in all forms of human thought and work, the life and death struggles of separate human beings.\"\n\nEven in her personal correspondence, such matters weighed on her mind, as in a letter to Lady Mary Elizabeth Ponsonby, the wife of Queen Victoria's Private Secretary, with whom Mary Ann corresponded until Ponsonby's death: \"Consider what the human mind en masse would have been if there had been no such combination of elements in it as has produced poets. All the philosophers and savants would not have sufficed to supply that deficiency. And how can the life of nations be understood without the inward life of poetry\u2014that is, of emotion blending with thought?\"\n\nThat Mary Ann had such a propensity stemmed from the extreme loneliness of her growing-up years. \"I have of late felt a depression that has disordered my mind's eye and made me alive to what is certainly a fact (though my imagination when I am in health is an adept at concealing it), that I am alone in the world,\" she wrote to a friend at the age of twenty-one. \"I do not mean to be so sinful as to say that I have not friends most unreservedly kind and tender, and disposed to form a far too favourable estimate of me, but I mean that I have no one who enters into my pleasures or my griefs, no one with whom I can pour out my soul, no one with the same yearnings, the same temptations, the same delights as myself.\" Four years later, in another letter, she reflected on her years of suffering: \"Childhood is only the beautiful and happy time in contemplation and retrospect: to the child it is full of deep sorrows, the meaning of which is unknown.\" She was absolutely convinced that \"the bliss of reciprocated affection\" was something she would never know.\n\nMary Ann had been marked early on as an ugly duckling, a characterization that would take on an even crueler edge for her as an adult. Someone once told her that she was, in fact, too ugly to love. Henry James called her the \"great horse-faced bluestocking.\" And her publisher, upon learning her identity, described her to his wife as \"a most intelligent pleasant woman, with a face like a man.\" Many went so far as to regard her as Medusa-like\u2014not merely plain but hideous. She had a large head, a big nose, and unflattering physical proportions. She dressed badly. And she was rather humorless, a trait that added severity and heaviness to her face. She was the first to acknowledge her ungainly appearance, once describing herself as \"a withered cabbage in a flower garden.\" Still, she had kind eyes, and Henry James wrote of this \"magnificently ugly, deliciously hideous\" woman that \"in this vast ugliness resides a most powerful beauty which, in a very few minutes, steals forth and charms the mind, so that you end, as I ended, in falling in love with her.\" Even her obituary in the Times, though praising her as \"a great and noble woman,\" could not refrain from mentioning her \"irradiated features that were too strongly marked for feminine beauty.\"\n\nShe had been raised in an intolerant family, which rejected those who didn't readily fit in. Aside from her \"ugly\" appearance, she held, from an early age, provocative views that distanced her from her family, particularly her father. Although she'd read theology texts passionately and had gone through a lengthy period of religious fervor, she became disenchanted. Eventually, her love of science and her passion for rational thought took over; a love of Wordsworth began to steer her into Romanticism and away from God. Moreover, when she and her father moved to Coventry, in 1841, she happily came into contact with agnostics, atheists, and freethinking intellectuals. She became especially close to her neighbors Cara and Charles Bray, both of whom openly enjoyed affairs outside their marriage.\n\nSoon afterward, Mary Ann renounced her faith and stopped going to church. Rather than give up her newfound principles, she told her outraged father that she would leave home and make her own way in the world. He made no effort to stop her. She eventually returned to care for him, and even attended church again, but their last years together, until his death in 1849, were difficult. \"My life is a perpetual nightmare,\" she confided to a friend, \"and always haunted by something to be done, which I have never the time, or rather the energy, to do.\" While serving as her father's nurse, she did read aloud to him a recently published novel, Jane Eyre, by a writer called Currer Bell. And in his final months, she managed the frivolous task of translating Spinoza's Tractatus Theologico-Politicus.\n\nThis period was yet another that led her to ruminate on notions of obligation versus independence, fulfilling duty versus chasing desires. In Romola, her historical novel set in fifteenth-century Florence, she would explore the question of where \"the duty of obedience ends and the duty of resistance begins.\"\n\nHow Mary Anne, Mary Ann, or Marian Evans\u2014full of secret ambition but lonely, prim, and lacking confidence\u2014transformed herself into George Eliot is a remarkable story. She often felt that she'd been given the mind of a man but not his opportunities. At thirty-one, she was numb, still grieving after her father's death, revealing in a letter that \"the only ardent hope I have for my future life is to have given to me some woman's duty\u2014some possibility of devoting myself where I may see a daily result of pure calm blessedness in the life of another.\" Around this time, she became Marian, another in a line of appellation shifts. And somewhere, \"George Eliot\" was patiently waiting to meet her.\n\nCharles and Cara Bray had introduced Marian to Ralph Waldo Emerson, who commented on her \"calm and serious soul.\" Her provincial world had expanded considerably. And with the support of Charles, she began writing book reviews (anonymously) for the newspaper he owned. She was still about a decade away from publishing her first novel. But one friend was wise enough to observe of Marian at thirty-two that \"[l]arge angels take a long time unfolding their wings; but when they do, soar out of sight. Miss Evans either has no wings or, which I think is the case, they are coming, budding.\" She was right.\n\nCharles also took a great interest in Marian's head\u2014or, to be more specific, her skull. As a keen believer in phrenology, Bray introduced Marian to one of its leading proponents in London. \"Miss Evans' head is a very large one,\" the expert astutely concluded. He added in his assessment that \"the Intellect greatly predominates\" (true), and that \"in the Feelings, the Animal and Moral regions are about equal; the moral being quite sufficient to keep the animal in order.\" That sounded about right, too. Most promising of all, he said, \"She was not fitted to stand alone.\"\n\nCompanionship would come later. For now, Marian was writing and editing for London's Westminster Review. However, there was one small snag in her newfound work. As Eliot's biographer Brenda Maddox has noted in her lively account, \"A female editor was as unheard of as a female surgeon; to be known to have one would have done no service to the review.\" While, in her own way, Marian was becoming entrenched in London's intellectual circles (the rare woman to have done so), she had to keep quiet about it. She was there, she was known socially, but her name could not be attached to the work she produced. Still, for the first time in her life, she experienced a real sense of popularity and demand for her presence. Young women she encountered, dazzled by her supple mind, developed crushes on her.\n\nConsidering her privileged position, Marian was more than happy to comply with the discretion demanded of her, and was even helpful in suggesting how to manage the situation. She told her boss that it might be best if \"you are regarded as the responsible person, but that you employ an Editor in whose literary and general ability you confide.\"\n\nThis rush of good fortune was cold comfort, however. She still lacked a husband, and she wanted one. But meeting a man named George Lewes would prove transformative. She could never legally marry him, but their relationship would become the most significant of her life. He was a prolific author, two years older than she, and they'd gotten to know each other better through a friend. She didn't know much about Lewes's personal life, but her first impression of him was that he talked too much. Soon she admitted, \"He has quite won my liking, in spite of myself.\" She found out that he was unhappily married, the father of four sons, and that he had a well-earned reputation for promiscuity. It was public knowledge that his wife, Agnes, had been having an affair with a friend of his, too. Lewes was even \"uglier\" than Marian, with a pockmarked face, an unkempt mustache, and unfashionable clothing, all of which she found off-putting. Even his friends called him \"Ape\" and declared him the ugliest man in London. (Charlotte Bront\u00eb, however, once remarked that she saw something of her sister Emily in him.) Henry James found him \"personally repulsive.\"\n\nLewes was cosmopolitan and Evans was provincial; his family, with its background in theater, was as flamboyant as hers was listless and austere. But by March 1853, she was already telling a friend that she found Lewes \"genial and amusing,\" and that he had \"won my liking, in spite of myself.\"\n\nWithin a year, they were living together\u2014and she started calling herself Marian Evans Lewes. Though he was still married to Agnes, Evans was able to confide to a friend, \"I begin this year more happily than I have done most years of my life.\" Divorce was out of the question for Lewes, but both he and Marian, despite their trepidation about whisperings of their supposed immorality, charged forward in their relationship\u2014living together \"in sin\" and hoping that her reputation in particular would not suffer irrevocably. They were prepared to lose friends to preserve their love, and did. \"I have counted the cost of the step I have taken and am prepared to bear, without irritation or bitterness, renunciation of all my friends,\" she wrote. \"I am not mistaken in the person to whom I have attached myself.\"\n\nBoth she and Lewes had already experienced their own forms of social persecution and were familiar with its toll. Yet they lost family, too: Marian had waited a few years to reveal her relationship to her siblings, and when she did, her brother Isaac (whom she adored) cut her off and encouraged his sisters to ostracize her. Defiant, she referred to Lewes as \"my husband.\"\n\n\"We are leading no life of self-indulgence,\" she wrote, \"except indeed that, being happy in each other, we find everything easy.\" Further, she insisted that she wasn't prepared to settle into someone else's notion of a virtuous life. She could be only herself. \"Women who are satisfied with light and easily broken ties do not act as I have done,\" she wrote. \"They obtain what they desire and are still invited to dinner.\" She paid the penalty without complaint or regret.\n\nIt is fair to say that without this passionate, supportive partnership, which would last until Lewes's death in 1878, George Eliot would not have been born. Lewes offered Evans a kind of love she had never known, unquestioning and absolute. (Despite rumors of his infidelity, there is no known evidence.) He wasn't an entirely enlightened man\u2014after all, he had once claimed condescendingly that even the best women writers were \"second only to the first-rate men of their day\"\u2014but he did heartily encourage her to write a novel. Journalistic work provided money but little satisfaction. \"It is worth while for you to try the experiment,\" he urged her\u2014and finally, in 1856, she confided in her journal: \"I am anxious to begin my fiction writing.\"\n\nShe embarked on this phase of her writing career by sending stories to John Blackwood, editor of Blackwood's Edinburgh Magazine; the first was published in January 1857 under the name \"George Eliot.\" She didn't send the pieces directly to Blackwood\u2014she submitted them via Lewes, who was already a regular contributor to the journal, as an added buffer. A month later, she wrote to Blackwood's brother and colleague, William: \"Whatever may be the success of my stories, I shall be resolute in preserving my incognito, having observed that a nom de plume secures all the advantages without the disagreeables of reputation.\"\n\nEliot was, of course, not the first woman to adopt a male pseudonym: the Bront\u00ebs had done it, and so had the French writer George Sand, who was much admired by Eliot. But she felt that her controversial subject matter\u2014depicting the lives of clergymen in her own native county of Warwickshire, and invoking autobiographical ideas about religion, faith, and unrequited love\u2014demanded secrecy. Not only that, but her social position was shaky enough because of her unconventional living situation. She was already infamous.\n\nIt turned out to be a good thing that she'd kept her identity hidden, as Blackwood wrote to Lewes (in a letter whose subtext was none too subtle): \"I am glad to hear that your friend is, as I supposed, a clergyman. Such a subject is best in clerical hands.\"\n\nIn 1858, Scenes of Clerical Life, which contained the stories serialized in Blackwood's magazine, was published in two volumes, under the name George Eliot. She was now a real author, and asked her publisher to send review copies to contemporaries she admired, including Dickens, Ruskin, and Tennyson.\n\nThe first part of her new pen name was inspired by her devoted partner (and was also the name of her uncle); the surname \"Eliot\" was chosen simply because she thought it was a \"good mouth-filling, easily pronounced word.\" \"Under what name could she have published her fiction?\" wrote a critic in 1999, referring to her various names. \"It is clear that neither 'Evans' nor 'Lewes' would have done. Her invented title became the only fixed point in a shifting world of reference.\"\n\nWhen John Blackwood showed up at Lewes's flat one day, hoping to meet the esteemed Mr. Eliot in person, the couple broke the news to him in a rather playful way. \"Do you wish to see him?\" Lewes asked. He and Marian left the room, then walked right back in\u2014and Blackwood was introduced to the man (woman) himself.\n\nHe was more than gracious about it, and happy to keep their secret safe. In 1859, with the publication of Adam Bede (a masterful depiction of rural domestic life, whose title character was based on her father), Marian kept her gender and name private\u2014though not for long. For one thing, too much of the story was recognizable, with identifiable characters; her brother Isaac read it and said that no one but his sister could possibly be the author. But the greater issue, as had been true for Charlotte Bront\u00eb with Jane Eyre, was the book's success: Queen Victoria was a fan. Dickens raved, \"I cannot praise it enough,\" even though Adam Bede had outsold A Tale of Two Cities. Alexandre Dumas called it \"the masterpiece of the century.\" And the Times declared that the mysterious author ranked \"at once among masters of the art.\" Critics loved Adam Bede, and so did the public\u2014a rare feat. The novel was a huge best seller. People wanted to know who George Eliot was, and false \"authors\" came forward to claim the glory. One man from Warwickshire insisted that he had written Adam Bede and Scenes of Clerical Life, and that he'd been cheated out of royalty payments.\n\nMarian's efforts to hide her identity were increasingly in vain. It did not escape the notice of the Leweses' friends that their purchase of a large house, filled with new furniture and staffed by servants, happened to coincide with the launch of George Eliot. One friend wrote to Marian saying that she would \"go to the stake\" if Marian was not George Eliot. She received a warm, open, but stern reply from the author: \"Keep the secret solemnly till I give you leave to tell it, and give way to no impulses of triumphant affection.\" Lewes added to the letter that \"you mustn't call her Marian Evans again; that individual is extinct, rolled up, quashed, absorbed in the Lewesian magnificence!\" From those who did realize the truth, the author pleaded for discretion. \"Talking about my books,\" she explained, \"has the same malign effect on me as talking of my feelings or my religion.\"\n\nWhen The Mill on the Floss came out in 1860, she was by then one of the most acclaimed authors of her day, and it became well known that George Eliot was a woman living with a married man. (Why she clung to her pseudonym even after her true identity was revealed is unclear.) People loved her books but judged her as immoral for her unorthodox relationship. Lewes's wife was cast as the victim in this drama, and Marian Evans as the predator. Never mind that Agnes had given birth to not one but another four sons outside her marriage. Although Lewes had forgiven her, he had ceased to think of her as his wife. He went on with his life in a discreet and dignified manner\u2014and did not embarrass Agnes as she had embarrassed him. He continued to support his family financially, yet his loyalty to Marian was unwavering. And she did not live with him until she knew that he would never again live under the same roof with Agnes.\n\nIn response to the flurry of scandal, \"George Eliot\" took full ownership of her new self, replying to letters addressed to \"Miss Evans\" with a chilly correction, informing one friend, \"I request that any one who has a regard for me will cease to speak of me by my maiden name.\" Marian Evans represented a lonely, ugly country girl whom the author no longer knew and now deemed \"extinct.\" George Eliot, her \"real\" self, was famous and influential (however immoral). She produced Silas Marner in 1861, and Romola two years later. Set in Renaissance Florence, Romola was a poorly received departure from her earlier works. She was not dissuaded by disappointment, and kept writing: Felix Holt the Radical came out in 1866\u2014and four years later came her masterpiece, Middlemarch. (Emily Dickinson wrote to a cousin: \"What do I think of Middlemarch? What do I think of glory?\")\n\nBy 1876, when Eliot published Daniel Deronda, another breathtaking accomplishment (notable for its sympathetic portrait of Jews), she was forgiven. She was the pride of her country and was proclaimed the greatest living English novelist. Her work, finally, spoke for itself, and a judgmental public had listened and fallen silent. She was adored and admired, a literary giant\u2014and a very wealthy woman. Whereas she and Lewes had once been exiles in London society, now they were celebrated, visited by Emerson, Turgenev, and other eminent intellectuals. A handsome American banker, John Cross, whom they affectionately called \"dear nephew,\" managed their business affairs. All was well.\n\nBut on November 30, 1878, Lewes was dead by evening. Eliot had reported months earlier to a friend that Lewes was \"racked with cramps from suppressed gout and feeling his inward economy all wrong.\" The sixty-one-year-old had succumbed to cancer, though he had never received the diagnosis.\n\nThey'd been together for more than two decades, and although Eliot was melancholic by nature, these had been the best years of her life. In a sense, Lewes had made everything possible. And when Eliot had received a manuscript of Adam Bede, bound in red leather, from her publisher, she had inscribed it to Lewes: \"To my dear husband, George Henry Lewes, I give this M.S. of a work which would never have been written but for the happiness which his love has conferred on my life. Marian Lewes, March 23, 1859.\"\n\nIn her grief-stricken stupor, she felt unable to attend his funeral. Each new day without him represented \"a new acquaintance with grief.\" Her old friend Turgenev sent a letter of condolence assuring her that the whole of \"learned Europe\" mourned with her. When she responded to such letters, she signed herself \"Your loving but half dead Marian.\" She was severely depressed and weighed just over a hundred pounds. She found a sense of purpose by establishing a \u00a35,000 grant in Lewes's name at Cambridge University, and by devoting her waking hours to editing his final work. Eliot never wrote another novel. She would be dead within two years.\n\nHer fans demanded her attention more than ever; it seemed that her fame had grown after her loss, which she found deeply unsettling. Requests for photographs of the famous George Eliot were politely declined, as the author explained that she treasured her privacy and did not wish to be stared at in public. One particularly aggressive autograph hunter was finally silenced with a form letter, a reply that the author had dictated: \"Mrs. Lewes (George Eliot), whom he has mistakenly addressed as Miss Marian Evans, has no photograph of herself and systematically abstains from giving her autograph.\"\n\nOne might expect that at this late stage of life\u2014she was sixty\u2014her knack for courting scandal would have been a distant memory. But she provoked rebuke once again, in May 1880, by marrying John Cross, who was twenty years her junior. He'd proposed to her three times before she accepted. Now she would have the legal marriage she'd always longed for; in this regard, she was rather old-fashioned, and had suffered from being unable to legitimize an otherwise blissful longtime union. At last, she could marry, if not the love of her life, a man she loved.\n\nFor their honeymoon, John and Marian traveled to Venice, where a strange mishap occurred. One morning, suffering from a depressive episode, Cross jumped from the balcony of their suite at the Hotel Europa (where luminaries such as Proust and Verdi had stayed) and landed in the Grand Canal. He was perhaps embarrassed, but physically unharmed. Venetian newspapers reported the incident, and the local police recorded it as a suicide attempt. Eliot alerted John's brother by telegram, and he joined them for the rest of their honeymoon. They blamed the heat for John's bizarre leap, and the trio traveled on to Munich.\n\nUpon their return home, the couple attended a dinner party in their honor, after which a guest wrote a petty and unkind missive to her sister: \"George Eliot, old as she is, and ugly, really looked very sweet and winning in spite of both. She was dressed in a short soft satin walking dress with a lace wrap half shading the body, a costume most artistically designed to show her slenderness, yet hiding the squareness of age.\" She added that there was not a single person in the room (including Eliot's husband) \"whose mother she might not have been. . . . She adores her husband, and it seemed to me it hurt her a little to have him talk so much to me. It made her, in her pain, slightly irritated and snappish. . . . He may forget the twenty years difference between them, but she never can.\"\n\nEvans changed her name yet again, to Mary Anne Cross, but the marriage lasted less than a year. She died unexpectedly on December 22, 1880, at sixty-one\u2014the same age at which Lewes died. Only a few days earlier, she and John had attended a concert and seen a performance of Agamemnon. In what is believed to be her final utterance, she complained of \"great pain in the left side.\" Then she was gone.\n\nLeft to tend to the legacy of this towering figure, Cross had his own minor identity crisis; he was referred to as \"George Eliot's widow.\" He only bolstered his image as \"Mr. Eliot\" when he published a biography of his late wife in 1885. It would be a stretch to assume that his marriage to Eliot had been consummated, but he is said to have truly loved and revered her. \"I am left alone in this new House we meant to be so happy in,\" he wrote to a friend. He never remarried.\n\nEven in death, Eliot paid a steep price for her unconventional life: in her will, she asked to be buried at Westminster Abbey, but the request was denied. She was dismissed as \"a person whose life and opinions were in notorious antagonism to Christian practice in regard to marriage, and Christian theory in regard to dogma.\" (This was certainly true.) Further, the church noted that despite the author's wish for a funeral in the Abbey, \"[o]ne cannot eat one's cake and have it too. Those who elect to be free in thought and deed must not hanker after the rewards, if they are to be so called, which the world offers to those who put up with its fetters.\"\n\nIt was not until the centenary of her death that she would receive a memorial stone in Poet's Corner. (She was in good company in that regard: Lord Byron, whose life was shockingly scandalous, died in 1824 and wasn't given a stone until 1969.) The eminent scholar Gordon Haight had the honor of delivering the speech for the unveiling of her stone at Westminster Abbey on June 21, 1980, five years before his own death. \"The novels of George Eliot provide the most varied and truthful picture we have of English religious life in the nineteenth century,\" he said. Whereas the novel had often previously served as a trivial pastime, he noted, Eliot elevated it into \"a compelling moral force.\"\n\nAfter Eliot died, Henry James paid her a glorious tribute: \"What is remarkable, extraordinary\u2014and the process remains inscrutable and mysterious\u2014is that this quiet, anxious, sedentary, serious, invalidical English lady, without animal spirits, without adventures, without extravagance, assumption, or bravado, should have made us believe that nothing in the world was alien to her; should have produced such rich, deep, masterly pictures of the multifold life of man.\"\n\nToday we take for granted how much Eliot sacrificed to become one of the greatest authors in the history of Western literature. She is simply George Eliot, literary master, staid historical figure, required college reading, admired by generations of authors. Her iconic Victorian visage now adorns posters, calendars, coffee mugs, stationery. But this pioneer could never forget the toll of her fame.\n\nReflecting on her story, it is tempting to interpret one of the concluding passages of The Mill on the Floss as the author's weary assessment of her own life:\n\nNature repairs her ravages, but not all. The uptorn trees are not rooted again; the parted hills are left scarred; if there is a new growth, the trees are not the same as the old, and the hills underneath their green vesture bear the marks of the past rending. To the eyes that have dwelt on the past, there is no thorough repair.\nHe was obsessive-compulsive and collected books about fairies\n\nChapter 4\n\nLewis Carroll & CHARLES DODGSON\n\nA show of hands if you've never heard of Alice in Wonderland. That's what I thought. You'd have to have fallen down a rabbit hole to be unfamiliar with Lewis Carroll's 1865 masterpiece, which in the past hundred years has been adapted for television and film numerous times, including three silent films, a British musical, a pornographic movie, an animated Disney version, a Japanese anime TV series (Fushigi no Kuni no Alice), and in 2010, a 3-D blockbuster directed by Tim Burton. It has been turned into graphic novels, plays, and operas, and it was even appropriated as the title of an execrable album by Jewel (Goodbye Alice in Wonderland). It has been translated into 125 languages, including Yiddish, Swahili, and Pitjantjatjara, an Aboriginal language of Australia. It has influenced James Joyce and Jefferson Airplane. There have been Alice theme parks, mugs, teapots, soap dishes, chess sets, T-shirts, and tea towels. Aside from Shakespeare, and the Bible, it's the most widely translated and quoted book of all time. Following the first edition illustrated by John Tenniel, subsequent versions have been accompanied by drawings from artists such as Arthur Rackham, Mervyn Peake, Ralph Steadman, and Salvador Dal\u00ed. Many woefully misguided authors have attempted sequels to Alice. Parodies have been published\u2014some brilliant, some without merit. Vladimir Nabokov translated a Russian edition when he was just twenty-four years old. And through all its iterations, Alice has never been out of print.\n\nThis classic story, perhaps the most-read children's book in the world, has also been banned on at least a few occasions. In the early twentieth century, a high school in New Hampshire censored Alice in Wonderland owing to its \"expletives, references to masturbation and sexual fantasies, and derogatory characterizations of teachers and of religious ceremonies.\" (Fair enough.) And in 1931, China deemed it forbidden material because \"animals should not use human language.\"\n\nThe author of Alice's Adventures in Wonderland (its original title) was Lewis Carroll, but that name was a hiding place. The eccentric Reverend Charles Lutwidge Dodgson, a shy, eminent Oxford mathematician and lecturer, had created the nom de plume as a means of shelter from which he could let his imagination run wild. He wanted his \"day job\" to remain undisturbed and private. Reflecting his obsession with wordplay since childhood, the pseudonym was a clever transposition of his real name: \"Lewis\" was the anglicized form of Ludovicus (Latin for \"Lutwidge\"), and \"Carroll\" was an Irish surname similar to the Latin Carolus, from which the name \"Charles\" is derived.\n\nHe was so mortified by publicity that he refused to acknowledge his alter ego. Whenever he was a guest in someone's home, if the name \"Lewis Carroll\" arose in conversation, he would leave. Autograph hunters were turned away without exception.\n\nHer Majesty Queen Victoria loved Alice and its sequel, Through the Looking-Glass, so much that she wrote a letter to Lewis Carroll, asking if he would send her the rest of his books. Unable to decline a request from the Queen, the humble author obliged as best he could, sending her numerous volumes\u2014all by Charles Dodgson, and all mathematical texts, including the popular beach read Condensation of Determinants, Being a New and Brief Method for Computing Their Arithmetical Values.\n\nUntil the end of his life, this reticent polymath maintained a strict divide between himself and the fanciful Lewis Carroll. \"For 30 years I have managed to keep the 2 personalities distinct,\" he boasted in a letter written three years before his death, \"and to avoid all communication, in propria persona, with the outer world, about my books.\"\n\nFastidious in everything he did\u2014today, we might apply the clinical term \"obsessive-compulsive,\" a mental disorder\u2014Dodgson went so far as to conceal his own handwriting. When he had to handle official correspondence for Lewis Carroll, he'd ask someone to copy out his response so that no one would have a sample of his writing. In 1883, he wrote to the divinity school at Oxford, begging the staff never to release anything he had handwritten. \"It is a thing I often have to do\u2014people seeming to assume that everybody likes notoriety,\" he explained, \"and scarcely believing me when I say I dislike it particularly. My constant aim is to remain, personally, unknown to the world.\"\n\nMore than a hundred years after his death, it is still hard to believe that the same man who wrote whimsical, exuberant classics of Victorian literature also produced arcane texts such as Notes on the First Two Books of Euclid, Designed for Candidates for Responsions; and An Elementary Treatise on Determinants with Their Application to Simultaneous Linear Equations and Algebraical Geometry. No wonder he needed a pen name.\n\nTo recount, even broadly, the achievements of Dodgson's life is the equivalent of tracing the lives of ten extraordinary men. The sheer vastness and absurd variety of his accomplishments, beyond his literary success, is exhausting to contemplate. He defies comprehension; only speculation is possible.\n\nHis beginnings were unremarkable. Charles Lutwidge Dodgson was born in Daresbury, Cheshire, on January 27, 1832, the third child (and first son) of eleven children\u2014there would be five more sisters and three brothers. His father, Charles Dodgson, and mother, Frances Jean Lutwidge, were first cousins. This genetic intermix might be to blame for the severe stammer that afflicted their son throughout his life, as well as most of his siblings. In childhood, he suffered a high fever making him permanently deaf in his right ear.\n\nThe young Dodgson had a fantastic imagination. He devised elaborate games with lists of rules, performed magic tricks to entertain his family, and created a puppet theater, writing plays and handling the troupe of marionettes himself. He wrote stories and poems (including acrostics), drew sketches, and wrote, edited, and illustrated magazines for his family. This was a common activity for many Victorians; what was unusual was for a young boy to lead the creative efforts and make all the booklets almost entirely alone. He was educated at home in his early years and proved a precocious reader, supposedly tackling The Pilgrim's Progress at age seven. Frances, who doted on her son, kept a record of his endeavors\u2014\"Religious Reading: Private,\" \"Religious Reading with Mama,\" and \"Daily Reading: Useful\u2014Private.\" In an 1898 biography of Dodgson, his nephew Stuart Dodgson Collingwood wrote that \"the boy invented the strangest diversions for himself . . . [and] numbered certain snails and toads among his intimate friends. . . . [He] lived in that charming 'Wonderland' which he afterwards described so vividly; but for all that he was a thorough boy, and loved to climb the trees and to scramble about in the marl-pits.\" It was an idyllic childhood.\n\nHe may have been a \"thorough boy,\" but he was wary of other boys. \"I am fond of children (except boys),\" he famously wrote, and admitted once that \"little girls I can now and then get along with . . . but with little boys I'm out of my element altogether.\" His negative sentiment might be traced to his time at Richmond Grammar School, where he was sent at the age of twelve. In a letter home, he recounted his unhappy initiation: \"The boys have played two tricks upon me which were these\u2014they first proposed to play at 'King of the cobblers' and asked me if I would be king, to which I agreed, then they made me sit down . . . and immediately began kicking me and knocking on all sides.\" He was bitten by another student, too. Although this was a rude awakening from his early years at the parsonage at Daresbury, where his father was a vicar (eventually rising to archdeacon), the boy quickly adapted to his new life. The headmaster sent an enthusiastic report to Dodgson's father, saying that \"he possesses, along with other and excellent natural endowments, a very uncommon share of genius.\"\n\nHe switched schools after a year and a half, but left feeling confident, intellectually superior to his peers, and, toughened by experience, unafraid to challenge would-be bullies. For the next four years he attended the public school Rugby; founded in 1567, it was one of Britain's most prestigious boarding schools (and the source of the sport). At the time Dodgson enrolled, Rugby was considered the best public school in England. Here, however, the hazing and cruelty proved far more brutal than had been the case at Richmond.\n\nIn 1849, he returned home, where he would stay before heading off to Christ Church, Oxford, two years later, following in his father's footsteps. His university education got off to a bittersweet start, as his mother died at forty-seven, just two days after he'd arrived at Oxford. He had been especially close to her, and this loss marked the definitive end of his childhood.\n\nAt twenty-one, he wrote a poem called \"Solitude\" that revealed, despite his love of jokes, puzzles, and riddles, a pensive side, glum and highly sensitive. It also revealed what would become a lifelong craving for a return to innocence, manifested in his preference for close friendships with children rather than adults. The final stanzas of the poem read:\n\nFor what to man the gift of breath,\n\nIf sorrow be his lot below;\n\nIf all the day that ends in death\n\nBe dark with clouds of woe?\n\nShall the poor transport of an hour\n\nRepay long years of sore distress\u2014\n\nThe fragrance of a lonely flower\n\nMake glad the wilderness?\n\nYe golden hours of Life's young spring,\n\nOf innocence, of love and truth!\n\nBright, beyond all imagining,\n\nThou fairy-dream of youth!\n\nI'd give all wealth that years have piled,\n\nThe slow result of Life's decay,\n\nTo be once more a little child\n\nFor one bright summer-day.\n\nEven as he earned a degree in mathematics at Christ Church, Dodgson wrote poems and stories on the side. In 1855, he submitted \"Solitude\" for publication in a literary journal called The Train. This is the earliest recorded appearance of \"Lewis Carroll.\" Two years earlier, he'd placed a poem and a short story in another literary journal, signing both with the alias \"B. B.\"\n\nHe was tall (six feet), slim, and handsome, yet he often showed considerable discomfort in social situations. Then there was the matter of his being ordained, which further isolated him from a wider community. However, as the bishop of Oxford recalled years after Dodgson's death, he did not pursue his religious studies as far as he might have done. \"He was ordained,\" the bishop wrote, \"but he never proceeded to priest's orders. Why he stopped at the Diaconate I do not know, but I think his stammer in speech may have had something to do with it. He was rather sensitive about this and it made him shy of taking clerical duty in church.\" Although Dodgson was not prepared to devote himself wholly to parochial life, as his father had, the bishop wrote, \"No one who knew him could doubt that he took his position as an ordained man seriously, or that his religion was a great reality to him, controlling his thoughts and actions in a variety of ways.\" It may have not only controlled but crippled him. Dodgson never married or had children; many scholars have asserted that the relationship with his young muse, Alice Liddell, was the single great romance of his life.\n\nDodgson found a home at Christ Church, partly by winning a distinguished \"studentship\" honor, given to only the best undergraduates. This appointment offered him lodgings and a small stipend for the rest of his life, along with permanent affiliation with Christ Church\u2014and access to its astounding resources with no obligation to teach or publish academic papers. There was a catch, of course; he could keep the fellowship as long as he never violated its restrictions. As a fellow, he was required to remain celibate and unmarried, and to progress to holy orders as an ordained priest. It's unclear why he never got further than deacon; instead, he appealed to the dean, Reverend Henry George Liddell, for permission not to advance. For reasons also unknown, Liddell allowed him to retain his position as Christ Church fellow, even though this was a violation of the rules and unprecedented. Despite deciding against entering the priesthood, Dodgson was by all accounts devout and pious, obsessed throughout his life by notions of sin and guilt. He was extremely conservative in his political and personal beliefs. This is yet another reason why he seems inscrutable, and so unlikely as the creator of Lewis Carroll.\n\nThe transition from stellar undergraduate to undergraduate tutor was not enough for him. As one writer commented of Dodgson's living quarters at the college, \"the very intensity of his tidiness indicates what forces were pent up within this environment.\" To the extent that he could, he satisfied his creative yearnings by slyly infusing his mathematical lessons with puzzles and riddles. One former student recalled, \"I always hated mathematics at school, but when I went up to Oxford I learnt from Mr. Dodgson to look upon my mathematics as the most delightful of all my studies.\" But something larger and more urgent stirred in his blood, and could not stay pent up for long.\n\nAt the age of twenty-three, Dodgson now had a secure position as a scholar and lecturer, and a regular income. His life changed profoundly when he was introduced to Dean Liddell's children.\n\nAmong his many skills and hobbies, Dodgson took an early interest in photography when it was still a wondrous new invention. Like Mark Twain, Dodgson was a gadget freak\u2014whatever the nineteenth-century equivalents of iPhones and iPods, he couldn't wait to try the next big thing. The camera was no exception, and with his eye for composition, his artistic sensibility, and his desire to tinker with new toys, Dodgson loved taking pictures. It was a cumbersome process, the very opposite of today's point-and-shoot, but he enjoyed it all, including the preparation of the plates. He constantly sought out subjects for his photographs, especially children.\n\nAs Liddell, a photography enthusiast himself, became better acquainted with Dodgson, he invited the young man to take pictures of his family. Dodgson began spending time with Liddell's little girls, Lorina (known as Ina), Alice, and Edith, and their brother, Harry, taking them on picnics and boating trips.\n\nThe \"golden afternoon\" of July 4, 1862, would prove transformative for them all. Years later, Alice Liddell recalled the day: \"The sun was so hot we landed in meadows down the river, deserting the boat to take refuge in the only bit of shade to be found, which was under a newly made hayrick. Here from all three of us, my sisters and myself, came the old petition, 'Tell us a story,' and Mr. Dodgson (that is Lewis Carroll) began it.\" He made it up as he went along.\n\nIn the presence of children, particularly the Liddells, there was no awkwardness: Dodgson was at his most charming. That July afternoon, as he later remembered it, \"in a desperate attempt to strike out some new line of fairy-lore, I had sent my heroine straight down a rabbit-hole, to begin with, without the least idea what was to happen afterwards.\" Ten-year-old Alice begged him to write down for her the story that he'd told. He sat up the entire night, working on a draft, and eventually made it into a green leather booklet called Alice's Adventures Under Ground, which he illustrated himself and gave to her as a Christmas gift.\n\nDodgson shared his story with a select few, including his friend Henry Kingsley, a novelist, who urged him to consider publishing it. He expanded and revised the manuscript, commissioned John Tenniel (already celebrated for his political cartoons for Punch) to do the illustrations, and submitted it to Macmillan, which agreed to publish Alice's Adventures in Wonderland by Lewis Carroll. The first chapter, \"Down the Rabbit Hole,\" began:\n\nAlice was beginning to get very tired of sitting by her sister on the bank, and of having nothing to do; once or twice she had peeped into the book her sister was reading, but it had no pictures or conversations in it, 'and what is the use of a book,' thought Alice, 'without pictures or conversations?'\n\nSo she was considering in her own mind (as well as she could, for the hot day made her feel very sleepy and stupid) whether the pleasure of making a daisy chain would be worth the trouble of getting up and picking the daisies, when suddenly a white rabbit with pink eyes ran close by her.\n\nThe book was initially released in 1865, but only fifty copies of a planned edition of two thousand were issued. Publication ceased when an unhappy Tenniel insisted on suppressing it because of imperfections in the printing process, which had affected his illustrations. Those who had purchased early copies were asked to return them to the publisher, and Macmillan donated the rejected books to children's hospitals.\n\nAfter the necessary corrections were made and a new printer was found, Alice was published, in 1866, in an edition of four thousand that Dodgson proudly declared to be a \"perfect piece of artistic printing.\" (Only twenty-three copies of the withdrawn 1865 version are known to survive, and in 1998 an anonymous buyer paid $1.54 million at auction for one of those precious books.)\n\nAlice was an instant success and sold out right away. Dodgson was thrilled at the reviews proclaiming his book \"a glorious artistic treasure.\" Like Charlotte Bront\u00eb, Dodgson requested that his publisher send him clippings of every review that came out, and he kept records of them in his diary.\n\nThe sequel, Through the Looking-Glass, and What Alice Found There, which made no reference to its predecessor, was published in time for Christmas 1871, with a first printing of nine thousand copies bound in gilt-stamped red cloth. Today, the original manuscript is in the British Museum.\n\nA section called \"The Wasp in a Wig\" had been omitted from the second book at Tenniel's suggestion, partly because he didn't think it could be drawn. He dismissed it with no small amount of condescension, informing the author that \"the 'wasp' chapter doesn't interest me in the least . . . a wasp in a wig is altogether beyond the appliances of art.\" Tenniel was apparently something of a diva\u2014he'd initially refused to sign on as the illustrator for Looking-Glass, and only after more than two years of nudging was Dodgson able to persuade him to say yes. Tenniel agreed, but noted that he would draw the pictures only if he could find the time.\n\nAlthough Looking-Glass was not as universally praised as Alice had been, it was a best seller. Immediately after the first printing sold out, Macmillan went back to press for six thousand more. It's no wonder that the critical reaction to the book, while favorable, was not entirely rapturous. The sequel, though brilliant, was more of an acquired taste than its predecessor (those coded chess moves!), if no less enchanting.\n\nAs the novelist Zadie Smith commented in her introduction to the 2001 Bloomsbury edition, Looking-Glass is \"a more tenebrous animal than its sister, both in style and quality of its fame. When I came to pick it up once more after an absence of years, I found I couldn't quite remember it other than as the repository where missing stories you thought were in Wonderland turn out to be\u2014like a second, darker, larder.\"\n\nLooking-Glass introduced what many consider to be the greatest piece of nonsense verse ever written, \"Jabberwocky.\" It ranked in the top ten in a poll of Britain's favorite children's poems, along with Edward Lear's \"The Owl and the Pussycat\" and T. S. Eliot's \"Macavity the Mystery Cat.\" \"Jabberwocky\" begins:\n\n'Twas brillig, and the slithy toves\n\nDid gyre and gimble in the wabe;\n\nAll mimsy were the borogoves,\n\nAnd the mome raths outgrabe.\n\n\"Beware the Jabberwock, my son!\n\nThe jaws that bite, the claws that catch!\n\nBeware the Jubjub bird, and shun\n\nThe frumious Bandersnatch!\"\n\nCarroll went on to write other books, including, in 1876, the mock-heroic nonsense poem The Hunting of the Snark (141 rhymed four-line stanzas) and important texts on mathematics and logic, but the Alice books remained his crowning achievements. His writing career had reached its apogee. As Robert McCrum wrote in the Guardian on the occasion of Tim Burton's \"charmless mash-up\" of a movie adaptation, the Alice books \"continue to exert an indestructible spell: teasing, phantasmagorical, narcotic, existential and profoundly English.\"\n\nThat he knew fame (not to mention great wealth) in his lifetime was a decidedly mixed blessing for C. L. Dodgson, as he was often known. Managing it filled him with terrible anxiety. On the rare occasions when he admitted that Dodgson and Carroll were the same man, he was either speaking openly with friends or corresponding with children and encouraging their letters. Otherwise, he said once, \"I use the name of 'Lewis Carroll' in order to avoid all personal publicity.\" Over and over, he lamented the unrelenting pressure to become a \"public figure,\" since he'd chosen a pseudonym precisely to protect himself from the burdens of celebrity. Dodgson hated the idea of strangers knowing anything about his personal life or what he looked like. Even those close to him could not resist feeding the myth of his enigmatic nature. He was \"not exactly an ordinary human being of flesh and blood,\" one friend reported, but rather \"some delicate, ethereal spirit, enveloped for the moment in a semblance of common humanity.\"\n\nTo that end, when he received letters for \"Lewis Carroll,\" he marked most of them \"Return to sender.\" Requests for photographs, even from relatives, were routinely denied. (He gave out photographs of himself only to children, usually young girls.) He begged friends to keep his real name private. When a bookshop catalog cited him as the author of Through the Looking-Glass, he wrote a letter demanding that Charles Dodgson's name no longer be printed \"in connection with any books except what he has put his name to.\"\n\nDesperate to keep his pseudonym private, he implored the Bodleian Library at Oxford to delete all cross-references between his names. The request was refused. Even though his identity as Carroll was an open secret, he was distressed by his inability to control its distribution.\n\nHe achieved a minor triumph when an editor contacted him for the Dictionary of the Anonymous and Pseudonymous Literature of Great Britain. \"I use a name, not my own, for writing under, for the one sole object, of avoiding personal publicity,\" he wrote, \"that I may be able to come and go, unnoticed, to all public places.\" He added that \"it would be a real unhappiness to me to feel myself liable to be noticed, or pointed out, by strangers.\" And he begged for respect in not \"breaking through a disguise which it is my most earnest wish to maintain.\"\n\nThe Dictionary editor, surprisingly, agreed to omit his name\u2014and so the book was published in 1882 with a glaring omission: this very famous pseudonym was nowhere to be found.\n\nDodgson remained a vigilant sentry of his privacy. In 1890, exasperated by the barrage of mail he received, he printed a circular to be enclosed with all replies to letters addressed to \"Lewis Carroll.\" The statement declared that Mr. Dodgson \"neither claims nor acknowledges any connection with any pseudonym, or with any book that is not published under his own name.\" (He might as well have added, \"So please bugger off.\")\n\nThere's a passage from Alice in Wonderland that invites interpretation as a commentary on the double-edged sword of fame, with its demands, expectations, and vicissitudes\u2014and as an expression of Dodgson's ambivalence toward his legacy:\n\n\"It was much pleasanter at home,\" thought poor Alice, \"when one wasn't always growing larger and smaller, and being ordered about by mice and rabbits. I almost wish I hadn't gone down that rabbit-hole\u2014and yet\u2014and yet\u2014it's rather curious, you know, this sort of life! I do wonder what can have happened to me! When I used to read fairy-tales, I fancied that kind of thing never happened, and now here I am in the middle of one! There ought to be a book written about me, that there ought!\"\n\n(The meaning is heightened, too, if you buy into the notion that the heroine is a stand-in for the author.) By all accounts, what Dodgson desired most was the power of invisibility. Though he was a fanatic about photography and loved taking pictures of people, he treasured his own privacy, and struggled to reconcile this requisite to his well-being with the fame he'd achieved. \"I don't want to be known by sight!\" he once said in despair.\n\nHe may have been paranoid about fame, but he was pragmatic. In 1879, he wrote to his publisher: \"I cannot of course help there being many people who know the connection between my real name and my 'alias,' but the fewer there are who are able to connect my face with the name 'Lewis Carroll' the happier for me.\" After all, he'd never intended to make Alice in Wonderland public; it had been created as a gift for Alice Liddell, and only at the urging of friends had he considered publishing it. He had hardly conceived it as a commercial product. Another reason for his strict separation of church (Dodgson) and state (Carroll) was purely professional: he wanted his mathematical books to be regarded seriously, and feared that if scholars connected him with Carroll, those works would be dismissed.\n\nAlthough Dodgson could accept that at a certain point his real name was not exactly a secret, there was the matter of preserving his privacy. He bristled at what he considered even the slightest invasion of his personal life\u2014such as being accosted in public to receive compliments about his work. It was exhausting. (\"But it's no use now,\" says Alice after falling down the rabbit hole, \"to pretend to be two people! Why, there's hardly enough of me left to make one respectable person!\")\n\nHis encounters with eager fans left him uncomfortable, and he confessed that among the things he hated most were \"having a tooth drawn\" and listening to a stranger talk about his books. The notion of being gracious to admiring fans was lost on him. In 1891, he reported to a friend an anecdote he'd read about a pompous author who greeted someone with the line, \"Have you read my book?\" It left him mortified. \"If ever I ask such a question of a stranger,\" he wrote, \"it will be due to 'temporary insanity!'\"\n\nEven though some biographical accounts of Dodgson portray him as a cloistered academic, he wasn't that, exactly. Between the age of twenty-nine and his death at sixty-five, he wrote a staggering number of letters\u2014nearly a hundred thousand in all\u2014proving that although he was shy with his public, he was not a recluse. That most of his letters were addressed to children shows his frequent unease in the world of adults; for the children he adored, he kept records of their birthdays and sent them letters with jokes, puns, puzzles, acrostics, and drawings. He toyed with inventive forms for his correspondence, including looking-glass letters that the recipient had to hold up to a mirror to read; rebus letters to be decoded; pinwheel-shaped letters; and delightful letters composed in such tiny script, on paper the size of a postage stamp, that a magnifying glass was needed to read them.\n\nBecause Carroll was a writer of the highest achievement who was also widely popular, it's understandable that his interactions with adults were sometimes marked by wariness and formality, rather than the broad affection he showed children. For them, he was whatever they wanted him to be. There is no evidence, in his diaries or elsewhere, of any long-term romantic relationships with women, which surely contributed to suggestions (however indirectly, even in his lifetime) of pedophilia. There was his affinity for taking photos of nude girls, of course. However, in the Victorian era, child nudes were not an uncommon artistic subject; it was perhaps Dodgson's excessive ardor for little girls, and his compulsive pursuit of their friendship, that called his behavior into question. (One might regard him as a Victorian-era Michael Jackson, but that is a topic for another time.)\n\nConsidering Dodgson's conscientious temperament, his openheartedness, and his religious fervor, it seems likely that his sexual urges, however inappropriate, remained repressed and were never acted on. Still, adding to the intrigue are four volumes missing from Dodgson's diaries of various periods dating from 1853 to 1863. (Nine volumes in all have survived.) What happened? Did he have something to hide? Was he chaste or deviant? Did a relative remove the diary pages after his death, to protect the family's reputation or the author's own? A record by the Lewis Carroll Society on Dodgson's \"Journal 8\" (from the period of his burgeoning acquaintance with Alice Liddell and her siblings) comments:\n\nA noticeable feature of this journal is the use Dodgson makes of these pages for recording prayers and supplications to help him lead a better life. Although prayers occurred in earlier volumes, the frequency and earnestness began to take on greater proportions in this journal. There has been much speculation about the reasons and purposes of the prayers. Reading them in the context of his unfolding life, there is no clear and obvious reason which can account for them. They do show that he experienced moments of great self-doubt and guilt. Some prayers indicate that feelings of slothfulness and lack of attention to his duties as mathematical lecturer gave rise to regret. However, there are some prayers which are more personal and poignant. One gets a deep sense of Dodgson's inability to come to terms with the troubles in his mind, and a feeling that he was unable to control these feelings which caused him such anguish and concern, whatever the cause may have been.\n\nIn 1863, a falling-out occurred with the Liddell family, and even though the mysterious rift was mended, the relationship was sporadic from then on. Was Dodgson a man with ignominious secrets? Did this partly explain his extreme need to protect himself from the scrutiny surrounding \"Lewis Carroll\"? Was the shame he carried regarding his fixation on prepubescent girls the reason he never progressed from deacon to priest? Why were his diaries filled with angst-ridden contemplations of guilt, temptation, and self-rebuke?\n\nTaken together, these questions are no more answerable today than they have been over the past century, but Alice's refrain throughout her journey in Wonderland\u2014\"Who in the world am I?\"\u2014resonates further when one considers the author's complex history. For unknown reasons, Dodgson often felt tormented by his own thoughts: he once wrote of having been \"haunted by some worrying subject of thought, which no effort of will is able to banish.\"\n\nThe man who loved puzzles was himself a deep mystery. Photography, which had once captivated him, was abandoned in the summer of 1880. Over more than two decades, he'd become an excellent photographer and had even considered earning a living with his hobby. He'd taken thousands of pictures, yet that year stopped the activity that had given him so much pleasure. As far as anyone knows, Dodgson never took another photograph for the rest of his life. One reason may be the unpleasant rumors that circulated about his penchant for photographing nude children. He was quite aware of how it might be perceived, telling one mother that her children's \"innocent unconsciousness is very beautiful, and gives one a feeling of reverence,\" and expressing remorse if he had overstepped any bounds with them. Though he discussed plans for future portraits, they were dropped. Another, more mundane, explanation for this may have been his extreme disdain for the latest, more advanced photographic processes, which he regarded as inferior.\n\nIn no way did this curious mathematician add up. He was a distinguished member of society, though he wore his hair longer than was considered proper for a Victorian gentleman. His letters alone were often works of genius, as exhilarating and imaginative as his Alice books, yet many who knew Dodgson found him stodgy and dull. \"He held himself stiffly,\" a relative recalled, \"one shoulder slightly higher than the other; in his almost overemphasized erectness there was an old-fashioned seriousness, an air of punctiliousness.\" Even Alice Liddell remembered him as having \"carried himself upright, as if he had swallowed a poker.\" In the company of adults, if he knew someone well and felt at ease, he appeared handsome, charming, funny, and confident; yet one colleague called him \"peculiar and paradoxical, and the topics on which he loved to dwell were such as would bore many persons.\"\n\nDodgson was quite odd. He wrote most of his books, including Alice, while standing up. (He calculated that he could work standing at his desk for up to ten hours.) His contradictions, eccentricities, and obsessive routines were truly astonishing\u2014apart from standing to write, he would map out entire journey routes well in advance, determining the precise time required to complete each leg of the trip. He also tallied the amount of money he would need at each stop, for each potential activity, and planned accordingly. Even his tea-brewing was a fanatical ritual: it must be steeped for exactly ten minutes, not a second more or less, or he would consider it undrinkable. And as it brewed, he would walk up and down his sitting room, swinging the teapot gently back and forth\u2014always for precisely ten minutes. When he entertained dinner guests, he prepared a seating chart and kept records of their dining preferences for future events. (\"By keeping the cards,\" he wrote, \"one gets materials for making up other dinner-parties, by observing what people harmonise well together.\") As a mathematician, he was fascinated by theories of randomness, but in life he was indefatigably controlling.\n\nHe loved taking long walks\u2014sometimes for twenty miles\u2014as an aid to problem solving, composing verse, and reflection. During his treks, he liked to time himself, record his average speed, then compare the numbers with those from previous walks. It makes sense that his mathematician's mind would have found satisfaction in this self-tracking; it's harder to understand why he took extensive notes on the condition of his feet after each walk.\n\nSteeped as he was in logic and science, Dodgson believed that Tuesday was his lucky day and forty-two his lucky number. He was a charter member of the Society for Psychical Research and the Ghost Society. He collected books about fairies and the occult.\n\nCharles Dodgson would have been a fascinating subject of study had he done nothing but produce the Alice books. One could spend years dissecting them and attempting to \"know\" the man whose phenomenal imagination made them possible. For those accomplishments alone, his name\u2014or, rather, Lewis Carroll's\u2014would have been embedded in the popular psyche for generations to come. But he spent his lifetime bursting with acts of invention, none of which adds up to a cohesive whole.\n\nWhere to begin? Dodgson can be credited with the idea of printing the title of a book on the spine of its dust jacket, which he conceived for The Hunting of the Snark. (That innovation proved fairly influential, to say the least.) He also developed, in his late forties, a system to correct flawed voting procedures that resulted in unjust outcomes; elements of his \"Parliamentary and Proportional Representation\" theory are still used in elections today. He also applied it to lawn tennis tournaments in which superior players were unfairly eliminated in early rounds, and in 1883 he published the treatise Lawn Tennis Tournaments: The True Method of Assigning Prizes. (This from the guy who wrote \"Jabberwocky\"?) Then there was his role as Common Room Curator at Christ Church, which was not curatorial in any artistic sense; Dodgson spent an inordinate amount of time organizing the wine cellars, creating accounting systems, conducting audits, and doing other tedious but important administrative tasks.\n\nBut wait, there's more: Dodgson lobbied for government support to relocate to Australia or the Cape the residents of Tristan da Cunha, an archipelago off the coast of South Africa considered the most remote inhabited locale in the world. (He'd adopted this as a political cause after his youngest brother, Edwin, had served as an Anglican missionary there in the 1880s.) He wrote both \"serious\" poems and comic verse. He spoke out on the benefits of vaccinations. He invented a portable chessboard. He was a passionate theatergoer and had corresponded with W. S. Gilbert and Arthur Sullivan\u2014as in Gilbert and Sullivan\u2014about a collaboration to set his poems to music and produce a musical Alice. (It never happened.) He created sketches to improve a kind of three-wheeled cycle known as the velociman, making it easier to steer. Shunning celebrity himself, he enjoyed meeting famous people of his time such as Trollope, Tennyson, and Ruskin. Portmanteau words such as \"chortle\" and \"galumph\" originated with Dodgson. He invented a new kind of postal money order, double-sided adhesive strips, a method for right-margin justification on a typewriter, an Alice in Wonderland postage-stamp case, a variation on conventional backgammon, a mnemonic system known as Memoria Technica for recalling dates and events, a writing tablet called a Nyctograph that could be used for taking notes in darkness (take that, iPad!), brainteasers, and word games, including an early version of what endures today as Scrabble.\n\nDespite all that he accomplished in his life, he was always modest. Dodgson wanted his work, regardless of context, to stand alone. When a friend once inquired about what The Hunting of the Snark \"means,\" he replied in a letter, \"I'm very much afraid I didn't mean anything but nonsense. Still, you know, words mean more than we mean to express when we use them; so a whole book ought to mean a great deal more than the writer means. So, whatever good meanings are in the book, I'm glad to accept as the meaning of the book.\" (Humpty Dumpty was just as cagey in explaining the meaning of his utterances. As he told Alice, \"When I use a word . . . it means just what I choose it to mean\u2014neither more nor less.\")\n\nIn his final years, Dodgson continued lecturing, sketching, writing letters, juggling work projects, and making time for the friends he cherished. He suffered increasingly from bronchial trouble, and he died on January 14, 1898, two weeks before his sixty-sixth birthday. He left instructions that his funeral be \"simple and inexpensive, avoiding all things which are merely done for show,\" and that there be \"no expensive monument. I should prefer a small plain head-stone.\"\n\nAs someone who had always drawn gossip with what might gently be called an unconventional lifestyle, Dodgson did little to dispel the rumors that swirled around him. Despite ugly whispers about his relations with children, he gave widely to charities that advocated on their behalf\u2014and kept all his donations private. He supported more than two dozen child-welfare organizations, and was so generous in giving away money that he incurred debt; a bank manager had to set limits on his overdrafts. As Morton N. Cohen noted in his excellent 1995 biography, Lewis Carroll, Dodgson never judged himself based on the opinions of others:\n\nCharles recognized earlier than one might suppose that his inner springs differed from most men's, that his heart beat to a different drum, that in order to be true to himself he would be compelled to lead a life that was not only outside the norm but would come under particular scrutiny and raise suspicions, one not generally condoned and subject to severe reprimand, sneers, lampoons, and ridicule. Be that as it may, he determined to follow his own star in spite of raised eyebrows and possible social censure. \"Let them talk\" was his answer; his own conscience would be his only judge.\n\n\"People want Carroll to be some sort of mad hatter,\" the chairman of the UK Lewis Carroll Society said in a 2010 interview. \"They find it difficult that somebody who could write something as crazy as Alice in Wonderland could still be a jolly decent chap.\"\nHe was a profligate spender who smoked forty cigars a day\n\nChapter 5\n\nMark Twain & SAMUEL CLEMENS\n\nHow the protean Samuel Clemens became the world's most famous literary alias will never be known for sure. Sly and droll, never one to shy away from the making of his own myth, Clemens claimed that his pen name derived from the years he spent working on riverboats, where water at a depth of two fathoms, or twelve feet, was considered safe for the boat to pass over. This distance was measured on a sounding line, a length of rope with lead on the end. The crew would call out, \"Mark twain!\" (meaning the mark on the line was at two fathoms) to indicate clear passage.\n\nLook up the archaic word \"twain\" in the Oxford English Dictionary and you will find an interesting entry. The adjective is defined as \"[o]ne more than one, two; forming a pair, twin.\" \"Consisting of two parts or elements; double, twofold.\" \"Separate, apart; estranged, at variance.\" As in the eighteenth-century hymnal by the priest and Oxford tutor John Keble: \"Five loaves had he, \/ And fishes twain.\" Or, from the Shakespearean sonnet, \"We two must be twain, \/ Although our undivided loves are one.\" The noun is defined as \"[t]wo persons or things identified contextually.\" In a nautical context, \"[t]wo fathoms. Esp. in mark twain, the two fathom mark on a sounding-line.\"\n\nClemens liked to explain that his appellation had been swiped from a man named Captain Isaiah Sellers\u2014a well-known steamboat man and sometime river correspondent for New Orleans newspapers. In Life on the Mississippi, Twain wrote that the captain \"was not of literary turn or capacity, but he used to jot down brief paragraphs of plain practical information about the river, and sign them 'MARK TWAIN,' and give them to the New Orleans Picayune. They related to the stage and condition of the river, and were accurate and valuable. . . . At the time that the telegraph brought the news of his death, I was on the Pacific coast. I was a fresh new journalist, and needed a nom de guerre; so I confiscated the ancient mariner's discarded one, and have done my best to make it remain what it was in his hands\u2014a sign and symbol and warrant that whatever is found in its company may be gambled on as being the petrified truth; how I have succeeded, it would not be modest in me to say.\"\n\nThere are other stories and legends as to how \"Twain\" came to be. Perhaps to varying degrees all versions are true, perhaps none. Some have ascribed to Clemens a Jekyll-and-Hyde nature; some have remarked on pseudonymity as a conventional choice for Victorian humorists, especially those tilted sharply toward satire. Perhaps both are true. One thing is beyond dispute: Twain is the best-known author in America's history, and his work is taught in every high school and college. With his pitch-perfect ear for the American vernacular, he is unrivaled (or, at least, secure among the all-time greats). \"I am not an American,\" he wrote in his notebook in 1897. \"I am the American.\"\n\nAdopting a pseudonym was for Clemens an exercise in playfulness, in fooling the public simply because he could. \"Some people lie when they tell the truth,\" Clemens once said in an interview. \"I tell the truth lying.\" (The poet, philosopher, and critic George Santayana once described truth as \"a jewel which should not be painted over; but it may be set to advantage and shown in a good light\"\u2014an aphorism that Twain would surely have endorsed.) And because he was someone who occasionally made enemies with his writing, having the pseudonymous cloak gave him a small measure of protective cover.\n\nThe jocular master of obfuscation was savvy about his own brand, eventually registering his alias as a trademark. He was his own best publicist and marketing director. He even incorporated himself under his nom de plume, so he officially became Mark Twain, Inc. He also trademarked the slogan on a box of \"Mark Twain\"\u2013branded cigars that read \"MARK TWAIN: KNOWN BY EVERYONE\u2014LIKED BY ALL.\" Although this pen name was the one that stuck, it was not his first: he'd previously experimented with other names, including \"W. Epaminondas Adrastrus Blab,\" \"Rambler,\" \"Josh,\" \"Sergeant Fathom,\" and \"Thomas Jefferson Snodgrass.\"\n\nAmong his contemporaries, the use of pseudonyms was not only common practice but considered a fashionable accessory. Humorists in particular adopted pen names: Charles Farrar Browne was a famous writer and lecturer who signed his writing as \"Artemus Ward\"; he was greatly admired by Abraham Lincoln and known for his delightfully awful puns. Other popular humorists included David Ross Locke, who wrote as \"Petroleum V. Nasby\"; and Robert Newell, whose pen name \"Orpheus C. Kerr\" was a pun on \"office seeker.\"\n\n\"Mark Twain\" was born in 1863, but Samuel Langhorne Clemens was born on November 30, 1835, in Florida, Missouri, in a two-room rented cabin. The red-haired infant's arrival was two months premature and he narrowly survived his birth. He spent his first four years frail and bedridden. Even his mother later admitted, \"I could see no promise in him.\"\n\nHe was one of seven children, three of whom would die young, and was raised in the nearby riverside town of Hannibal. \"If you are born in my state, you pronounce it Missourah,\" he once said. \"If you are not born in my state, you pronounce it Missouree. But if you are born in my state, and you have to live your entire life in my state, you pronounce it misery.\"\n\nHe adored his mother, Jane, and avoided his stern, aloof father, John, whom he could not remember ever having laughed. (John died of pneumonia when Clemens was eleven years old.) He was a high-strung child, a sleepwalker, and he suffered from nightmares. Yet he was as exuberant, magnetic, and funny as his father was austere. It's no wonder that Clemens struggled to escape the provincial, restrictive milieu of his boyhood and went in search of a more expansive world. Despite being a poor student, he displayed an early knack for language and mimicry and a great love of storytelling. If he found little enchantment in his own house, he cultivated it endlessly through his fertile imagination. At sixteen, he was already working for a small newspaper, the Hannibal Western Union, and writing humorous sketches. And the first in a lifelong series of get-rich-quick schemes hooked him at the age of eighteen\u2014he had a quixotic plan to sail the Amazon, where he would make a vast fortune in \"a vegetable product of miraculous powers\" that he'd read about. It was said to be \"so nourishing and so strengthening that the natives of the mountains of the Madeira region would tramp up hill and down all day on a pinch of the powdered coca and require no other substance.\" This claim about being able to tramp up and down hills all day was undoubtedly true. \"Coca\" is better known today as cocaine.\n\nClemens was detoured from his grand plan, however, and went to work as an apprentice steamboat pilot, eventually getting his license. It was a job he loved. The stint ended with the advent of the Civil War, and it had been marked by tragedy. Clemens had convinced his younger brother Henry to join him in steamboat work, and Henry died in 1858 when the steamboat he was working on exploded. Clemens never forgave himself for his brother's death. Adding to his horror and guilt, he'd had a dream, not long before Henry died, in which he saw his brother lying in a metal casket.\n\nIn the summer of 1861, at his brother Orion's insistence, Clemens had headed west by stagecoach, hoping to strike it rich in Nevada's silver rush. He failed as a prospector. A job as a mill laborer didn't work out, either. One day Clemens asked his boss for a raise, naming the figure of $400,000 a month in his request. He was promptly fired.\n\nHis next move brought better luck, if modest pay: he was hired as a reporter, at a salary of twenty-five dollars a week, by the Territorial Enterprise, a newspaper in Virginia City, Nevada. Well liked and clearly talented, Clemens soon upped his wages to six dollars a day. \"Everybody knows me,\" he boasted in a letter to his mother, \"& I fare like a prince wherever I go, be it on this side of the mountains or the other. And I am proud to say I am the most conceited ass in the Territory.\"\n\n\"Mark Twain\" made his debut on February 3, 1863, launched in an Enterprise column with the line, \"I feel very much as if I had just awakened out of a long sleep.\" It was signed, \"Yours, dreamily, Mark Twain.\" Two years later, while living in San Francisco, Twain became an official success: his short story \"Jim Smiley and His Jumping Frog\" was published in the Saturday Press in New York. It was reprinted all over the country (later retitled as \"The Celebrated Jumping Frog of Calaveras County\") and won him nationwide acclaim. \"The foremost among the merry gentlemen of the California press, as far as we have been able to judge,\" wrote one New York critic, \"is one who signs himself 'Mark Twain.' He is, we believe, quite a young man, and has not written a great deal. Perhaps, if he will husband his resources and not kill with overwork the mental goose that has given us these golden eggs, he may one day rank among the brightest of our wits.\"\n\nIt was obvious from the start, even in his slightest pieces, that down to his marrow Twain was a writer: \"The difference between the almost right word and the right word is really a large matter\u2014it's the difference between the lightning bug and the lightning,\" he once noted.\n\nAs Clemens's career as a newspaper reporter took off, he used his Twain pseudonym irregularly, but eventually it supplanted his real name. He slipped into Twain as if into an elegant new pair of shoes. Some of his friends began calling him \"Mark,\" and his letters home were signed that way, too. As Twain's biographer Ron Powers has noted, even early correspondence displayed the young man's knack for embellishment: \"His indifference to the boundary between fact and fantasy became a hallmark of his literature, and later, of his consciousness.\" At the age of twenty-eight, the transformation was complete: Clemens was a buried man. The sobriquet stuck, and everything published subsequently would appear under this alter ego. \"Mark Twain\" gave Clemens a kind of solid self-confidence he had never known as a boy. At one point he even joked that an \"independent Double\" was going around causing the kind of mischief that Sam Clemens wouldn't dream of attempting: \"It gets intoxicated\u2014I do not. It steals horses\u2014I do not. It imposes on theatre managers\u2014I never do. It lies\u2014I never do.\" He was a restless lover of reinvention, and his new name allowed him to step into a role that he had conjured, and that he alone controlled.\n\nWhen his book The Innocents Abroad was published in 1869, it was an instant hit. But his 1867 story collection, The Celebrated Jumping Frog of Calaveras County, and Other Sketches, had been a huge flop, and Twain said he hoped that every remaining copy would be burned. Even so, his public lectures had already made him a much-adored entertainer, with packed houses, and audiences hanging on his every word and rewarding his droll performances with roaring applause and standing ovations. (Powers has described Twain as \"the nation's first rock star.\") He charmed everyone he met. For the most part, he was able to repress his darker side and the grudges he held against those perceived to be his enemies. Yet he was gripped by bouts of depression and suicidal impulses, and often craved public validation as a means of steadying himself again. Periods of idleness threatened his equilibrium. Even in good times, though, he could be unpredictable, acting like a petulant prima donna: yelling at hotel employees in cities he visited; canceling lectures at the last minute; smashing a window shutter with his fists over a scheduling glitch; angrily throwing his shirts out a window.\n\nHe became a husband at thirty-five, marrying Olivia Langdon, the daughter of a wealthy merchant from Elmira, New York. Her skeptical father asked his future son-in-law for references, one of whom reported, \"I would rather bury a daughter of mine than have her marry such a fellow.\" However difficult Clemens could be (which was very), and however frequent his absences from home, he and his wife were utterly devoted to each other until her death left him a widower.\n\nThe couple met on New Year's Eve 1867, through her family, and spent the evening in Manhattan attending a reading by Charles Dickens. Their courtship lasted seventeen months. Marrying into money left Clemens conflicted\u2014after all, he had humble beginnings and claimed to hate the rich. (He would mock the nation's culture of materialism and greed in 1874's The Gilded Age.) Yet Mark Twain had boundless ambition and extravagant tastes. It seems fair to assume that even as he commissioned the ostentatious Gothic Revival mansion in Hartford, Connecticut, where he would settle with Olivia (known as Livy), some part of him must have burned with self-loathing. Louis Comfort Tiffany and Company designed part of the interior, which included custom stained-glass windows, polished marble floors, ornate brasswork, a carved oak Venetian bed, a mantelpiece from a Scottish castle, a billiards room, and modern conveniences such as central heating and flush toilets. In all, there were nineteen rooms and seven bathrooms. Although Twain would experience his greatest literary success while in that residence, it was also where he would experience ravaging losses. (Eventually, beset by financial ruin, he would be forced to sell the house.) In Buffalo, New York, he and Livy had already suffered the death of their first child, who died of diphtheria at eighteen months old. In Hartford, where the couple would spend the next twenty years, they raised three girls\u2014Susy, Clara, and Jean\u2014who venerated and feared their mercurial father. Harriet Beecher Stowe was the family's next-door neighbor, though she lived in a much more modest brick house.\n\nTwain's admirers included Charles Darwin, Joseph Conrad, William Faulkner, T. S. Eliot, Eugene O'Neill, Oliver Wendell Holmes, and George Bernard Shaw. Twain would rarely admit to reading other writers, but he liked Shaw, whom he praised as \"quite destitute of affectation.\" Shaw wrote a letter to Twain in 1907, mentioning that he'd met William Morris, an \"incurable Huckfinomaniac.\" He addressed the letter to \"My dear Mark Twain\u2014not to say Dr Clemens (though I have always regarded Clemens as mere raw material\u2014might have been your brother or your uncle).\" A year later, Thomas Edison remarked, \"An American loves his family. If he has any love left over for some other person, he generally selects Mark Twain.\" Nietzsche recommended The Adventures of Tom Sawyer to friends.\n\nAt the height of his fame, Twain was bombarded by fan mail, including manuscripts from aspiring writers who wanted his opinion of their work and assistance with publication. Letters poured in from around the world, some addressed simply to \"Mark Twain, Hartford, Connecticut.\" Some asked for money. He filed away many letters under the heading, \"From an ass.\" He wrote to his mother, \"I have a badgered, harassed feeling, a good part of the time.\" Yet he was paradoxical as ever: even though he often checked into hotels incognito, using a variety of aliases including \"S. L. Samuel\" and \"C. L. Samuel,\" he was always thrilled to be recognized. Sometimes he would actually strut up and down busy streets in Manhattan, just as church services were ending and crowds were pouring out, so that he could bask in the sight of heads excitedly turning toward the great celebrity in their midst.\n\nOnce Twain was asked why the fame of many other humorists had been so ephemeral. \"Because they were merely humorists,\" he replied. \"Humorists of the 'mere' sort cannot survive. Humor is only a fragrance, a decoration. Often it is merely an odd trick of speech or of spelling . . . and presently the fashion passes and the fame along with it.\" Restless and ambitious all his life, Twain knew that to secure his legacy, his output had to transcend \"mere\" comic sketches and journalism. His reputation would ultimately rest on two masterpieces: The Adventures of Tom Sawyer, published in 1876 when the author was forty-one; and The Adventures of Huckleberry Finn, published nearly a decade later. Ernest Hemingway claimed that all of American literature was derived from from the latter novel, calling it \"the best book we've ever had. There was nothing before. There's been nothing as good since.\" The playwright Arthur Miller once said of Twain in an interview, \"He wrote as though there had been no literature before him.\"\n\nTwain, a popular writer, was also one hell of a trickster. As the scholar John Seelye notes of Tom Sawyer in his introduction to the Penguin edition of Huckleberry Finn, Tom is \"a prankster from the start,\" not unlike the author himself, who adored practical jokes. \"Where Huck Finn seems to be a projection of something mysterious deeply hidden in Mark Twain's psyche, Tom Sawyer is clearly an active agent of the author,\" Seelye writes.\n\nSwindler, con man, histrionic showman: Tom represents, at least on the surface, the essential Twain. Huck goes deeper; he evinces both halves of the author's troubled psyche (Clemens\/Twain), with all its contradictions, anxieties, and follies. But as Twain grew older, his private, Clemensesque qualities floated disruptively to the surface, threatening the impish, rambunctious public man he had become. The blithe, witty charmer was far more mercurial than his admiring public ever knew, and struggled (often painfully) to manage the two worlds and selves he inhabited. When he was drunk, however, his carefully constructed mask came undone. As one friend observed, \"He was always afraid of dying in the poorhouse. The burden of his woe was that he would grow old and lose the power of interesting an audience, and become unable to write, and then what would become of him?\" The more Clemens drank, the worse it got; there was no Twainian joviality or playful wit to accompany his alcohol consumption. Instead, his friend said, he would \"grow more and more gloomy and blue until he fairly wept at the misery of his own future.\"\n\nIn April 1894, the world's most famous author declared bankruptcy. The wealth he'd amassed could not match his debts, and he'd had to embark upon a grueling round-the-world tour to repay creditors and become solvent again. Like his late father, Clemens had an almost manic relationship to money and had invested his considerable earnings dreadfully. He'd backed failed gadgets and fraudulent schemes, founded a money-losing publishing company, and patented a few unsuccessful inventions of his own, at great expense. Among them was an adjustable elastic waist strap for men that could be buttoned onto the back of a pair of trousers to keep them from falling down.\n\nFoolishly, even though he was among the first Americans to have a telephone at home, he had declined to invest in Alexander Graham Bell's invention. He wasn't convinced that the telephone had much of a future. Twain himself acknowledged his gift for squandering his fortune. \"Now here is a queer fact,\" he wrote, \"I am one of the wealthiest grandees in America\u2014one of the Vanderbilt gang, in fact\u2014and yet if you asked me to lend you a couple of dollars I should have to ask you to take my note instead.\" Even he must have appreciated the perverse irony of having succumbed to the Gilded Age\u2014a lifestyle that he so despised.\n\nHaving brought his family to the brink of ruin, Clemens would endure greater tragedies in subsequent years. He lost several friends and relatives. His daughter Susy died in 1896; Livy died of heart failure in 1904, at the age of fifty-eight; and his daughter Jean died in 1909.\n\nThese catastrophic events left him lonely, bitter, brokenhearted, vindictive, and paranoid. Sam Clemens depended on Mark Twain to keep going, but the gentle, irreverent humor in his work gave way to a more cynical, dyspeptic edge. (He took to calling his famous white uniform his \"don'tcareadam suits,\" and boasted that they made him the most conspicuous man alive.) Although he'd always abhorred critics, he had previously displayed tolerance toward what he regarded as a necessary evil. \"I believe that the trade of critic, in literature, music, and the drama, is the most degraded of all trades, and that it has no real value,\" he wrote. \"However, let it go. It is the will of God that we must have critics, and missionaries, and Congressmen, and humorists, and we must bear the burden.\" Now, however, he was inclined to be far more bilious. If it's true that Clemens and Twain were polar opposites within the same deeply divided man, then it seems there was little actual Twain left in him at the end.\n\nHis insecurity often overwhelmed him, and his corrosive obsessions\u2014success, wealth, fame\u2014revealed a volatility that baffled even him. The \"periodical and sudden changes of mood in me,\" he once wrote, \"from deep melancholy to half-insane tempests and cyclones of humor, are among the curiosities of my life.\" He loved playing billiards, which provided yet another excuse for his explosive temper to manifest itself. \"When his game was going badly,\" Albert Bigelow Paine wrote in his 1912 Twain biography, \"his language sometimes became violent and he was likely to become critical of his opponent. Then reaction would set in, and remorse.\"\n\nToday, Mark Twain is still viewed as the mythic \"Colonel Sanders without the chicken, the avuncular man who told stories,\" as Ron Powers has described him. \"He's been scrubbed and sanitized.\" Yet a more comprehensive version of Twain emerged in 2010 with the publication of the first installment of his rambling three-volume autobiography. It presents Twain raw and uncensored; he instructed that his unedited recollections be withheld from the public for one hundred years after his death. (As ever, what a brilliant marketing ploy.) He dictated most of the 500,000-word manuscript to a stenographer during the four years before he died, then postponed its publication for a century to preserve his genial reputation and legacy. The strategy worked. Among towering American literary figures, Twain remains essentially unknowable. As one contemporary journalist aptly put it, he's \"still a mystery, a riddle wrapped in an enigma shrouded in a white suit.\"\n\nThe biographer Justin Kaplan\u2014whose 1966 account of Twain won the Pulitzer Prize and the National Book Award\u2014has spoken of the author's dark moods, which are more fully revealed in the new Autobiography. The private Twain evinced a side filled with \"rage and resentment . . . where he wants to get even, to settle scores with people whom he really despises. He loved invective,\" Kaplan noted in an interview. For instance, after having stayed in 1904 with his family in Florence, Italy (where Livy would die), Twain unleashed his fury against the rather unaccommodating countess who owned the villa they'd rented. He characterized her as \"excitable, malicious, malignant, vengeful, unforgiving, selfish, stingy, avaricious, coarse, vulgar, profane, obscene, a furious blusterer on the outside and at heart a coward.\" A lawyer and fellow investor who betrayed him was attacked as having \"the pride of a tramp, the courage of a rabbit, the moral sense of a wax figure, the sex of a tapeworm.\" And Twain's secretary and household manager, Isabel Van Kleek Lyon, with whom he had a close, tempestuous relationship for the last several years of his life, was in the end an object of obsessive condemnation. In a letter to his daughter Clara, Twain fumed that Isabel was \"a liar, a forger, a thief, a hypocrite, a drunkard, a sneak, a humbug, a traitor, a conspirator, a filthy-minded & salacious slut pining for seduction & always getting disappointed, poor child.\"\n\nIn the years before his death in Redding, Connecticut, on April 21, 1910, Twain was at his most miserable, full of malice and sadness and vitriol. His health was terrible, too, no doubt owing to his having smoked forty cigars a day for most of his life. Toward the end, he spent much of his time in bed.\n\nFacing his own mortality, he hoped for reconciliation. \"I think we never become really and genuinely our entire and honest selves until we are dead,\" he once wrote, \"and not then until we have been dead years and years. People ought to start dead and then they would be honest so much earlier.\" Not long before drifting off to sleep for the last time, he mumbled something about \"dual personalities.\" He died in his carved oak bed, with his daughter Clara at his side. Two days later, a letter appeared in the New York Times.\n\nTo the Editor:\n\nI wish to draw your attention to a peculiar coincidence.\n\nMark Twain, born Nov. 30, 1835.\n\nLast perihelion of Halley's comet, Nov. 10, 1835.\n\nMark Twain died, April 21, 1910.\n\nPerihelion of Halley's comet, April 20, 1910.\n\nIt so appears that the lifetime of the great humorist was nearly identical (the difference being exactly fifteen days) with the last long \"year\" of the great comet.\n\nR. FRIDERICI.\n\nWestchester, N. Y., April 22, 1910\n\nMark Twain would have loved that coincidence. In fact, he had once predicted it himself: \"The Almighty has said, no doubt: 'Now here are these two unaccountable freaks; they came in together, they must go out together.'\" The comet was visible from Earth when he died, the final triumph of an inimitable showman.\nHe was Federal Prisoner 30664\n\nChapter 6\n\nO. Henry & WILLIAM SYDNEY PORTER\n\nIf you are now reading or have recently read a short story by O. Henry, you are most likely a middle-school student. He was the greatest short story writer of his generation, but O. Henry\u2014who died at forty-seven with twenty-three cents in his pocket\u2014isn't read much these days, except as homework.\n\nHis stories are known for their irony, aphorisms, plot twists, and moral lessons, and the surprise endings he called \"snappers.\" They were formulaic, but the formula worked. \"[H]e never told his story in the first paragraph but invariably began with patter and palaver; like a conjurer at a fair, it was the art of the anecdote that hooked the public,\" wrote the critic Francis Hackett. \"He planned, first of all, to make his theme straight and clear, as a preacher does who gives the text. Then he established his people with bold, brilliant strokes, like a great cartoonist. But the barb was always a surprise, adroitly prepared, craftily planted, and to catch him at it is an exercise for a detective.\"\n\nWilliam Sydney Porter was born in Greensboro, North Carolina, on September 11, 1862. His middle name was originally spelled \"Sidney,\" but he changed it; later in life he would drop \"William\" and be known as Sydney Porter.\n\nBy the time he was three years old, his mother was dead of tuberculosis. Along with his father, Dr. Algernon Sidney Porter, William moved into a boardinghouse run by his grandmother. Algernon\u2014a heavy drinker, just as William would become\u2014was also an aspiring inventor with plans for a flying machine and a horseless carriage driven by steam.\n\nThe year 1865 brought the end of the Civil War and the assassination of President Lincoln. William began attending a one-room schoolhouse run by his aunt, who served as a surrogate mother and whom he later credited with inspiring his love of art and literature. As a boy he had a talent for drawing, thanks to his aunt's attentive instruction; and he devoured Dumas, Hugo, Thackeray, Dickens, Wilkie Collins, and others. \"I did more reading between my thirteenth and my nineteenth years than I have done in all the years since, and my taste was much better then,\" he once told a reporter.\n\nAlthough he loved learning, college was for the rich, which meant that for him it was out of the question. At fifteen, William was sent to work in his uncle's pharmacy, and at nineteen he became a licensed pharmacist. \"The grind in the drugstore was an agony to me,\" he later admitted. Had he not received an invitation in 1882 to join a family friend in Texas, doing ranch work, William Porter might have lived and died a pharmacist rather than become the prolific writer O. Henry.\n\nLa Salle County, Texas, was not destined to be his last stop, but it was at least an escape from his tedious life at home. He was always reading poetry, especially Tennyson, and while herding sheep, he carried around a copy of Webster's Unabridged Dictionary. He wrote stories, too, but after reading them aloud to a family friend, he'd rip them up and throw them away.\n\nNext he made his way to Austin, where, supposedly, he first used his future pen name: he had a habit of calling \"Oh, Henry!\" to a girlfriend's cat, said to respond only to that greeting. (True or not, the phrase has no connection to the candy bar of that name, launched in 1924.) He signed his girlfriend's autograph album as \"O. Henry,\" and composed a poem, \"A Soliloquy by the Cat,\" using this name. When he proposed marriage to his girlfriend, she rejected him; she came from a wealthy family, and he was a nobody with a dead-end job. Although he lost the girl, he'd found his pen name. Or so one version of the story goes; there are many. Porter was a good liar who enjoyed spinning fabrications about himself.\n\nHe had a series of drab jobs, finally working as a draftsman at the Texas Land Office, where he earned a hundred dollars a month. He wasn't thrilled by the work, but had no trouble finding friends. He played cards, charmed rapt listeners with his storytelling, and joined local singing and theater groups. He became a popular local figure and was known for always being impeccably dressed.\n\nIn 1888, following a speedy courtship, Porter eloped with seventeen-year-old Athol Estes. They had a son who died the day he was born. A year later, the couple had a daughter, Margaret. Porter, feeling settled and happy, was ready to pursue his true ambition: writing. After sending a journalism piece to the Detroit Free Press, he received an encouraging reply: \"Am sorry it is not longer,\" the editor wrote. \"Check will be sent in a few days. Can you not send more matter\u2014a good big installment every week?\" Porter began selling freelance articles, mostly humor pieces, to newspapers and journals around the country.\n\nIn 1891, he took a job as a teller at the First National Bank of Austin, a position that seemed ideal at first\u2014it was mindless, and would allow him to write in the evenings\u2014but would later turn out to have damaging and long-lasting consequences. After working at the bank for three years, he resigned when an audit revealed shortages in his till. Though he was charged with embezzlement, the case was dismissed for lack of evidence. Porter decided to focus on his writing, and he turned entrepreneurial, buying a used printing press and publishing an eight-page weekly satirical magazine called The Rolling Stone, for which he served as writer, illustrator, typesetter, and printer. \"It rolled on for about a year,\" he said later, \"and then showed unmistakable signs of getting mossy.\" He shut it down but had no regrets; the experience had boosted his confidence. His family moved to Houston, where he worked as a reporter, cartoonist, and columnist for the Houston Post, a job he loved.\n\nUnfortunately, his falling-out with the Austin bank came back to haunt him just six months later. The embezzlement case had been reopened by federal auditors, and he was arrested. Although he insisted that bank executives regularly \"borrowed\" money without keeping records of their transactions (and that they rarely repaid what they'd withdrawn), he had no proof. Whether Porter was a fall guy or a criminal, no one will ever know, but he couldn't face the thought of imprisonment. After being released on a $2,000 bond posted by his wealthy father-in-law, Porter hopped on a night train to New Orleans, and, a few weeks later, boarded a freighter bound for Honduras. It was a frightening experience at the time, but would prove excellent fodder for fiction. (Life as a South American fugitive was chronicled in his 1904 debut story collection, Cabbages and Kings.) When asked once why he did not read more fiction written by others, he replied, \"It is all tame, as compared with the romance of my own life.\"\n\nPorter regretted his evasion of justice, but he argued until the end of his life that he was an innocent man who had no choice other than to flee. \"I am like [Conrad's] Lord Jim,\" he told a friend, \"because we both made one fateful mistake at the supreme crisis of our lives, a mistake from which we could not recover.\" Honduras was a smart choice\u2014it had no extradition treaty with the United States\u2014and he had some vague plan for his wife and daughter to join him in exile. It never happened. When Porter found out that Athol was dying of tuberculosis, he rushed back home.\n\nA year later, after a three-day trial in Austin, Porter\u2014now a grieving widower with a ten-year-old daughter\u2014pleaded not guilty. He was convicted of embezzlement and sentenced to five years in a Columbus, Ohio, penitentiary. \"I care not so much for the opinion of the general public,\" he wrote in a letter to his mother-in-law, \"but I would have a few of my friends still believe that there is some good in me.\"\n\nBecoming Federal Prisoner 30664 would launch his writing career and complete his transformation into O. Henry. Despite a painful separation from Margaret, with whom he was close, prison was the ultimate writing colony. The three years he spent there proved to be his MFA program, his refuge from the demands of the outside world.\n\nHe wrote stories during his night shifts in the prison infirmary, a plum job he had obtained because of his background as a licensed pharmacist. After saving the life of a warden who'd overdosed on arsenic, Porter gained additional privileges with minimal supervision, including sleeping at the infirmary and being able to roam the grounds more freely than other prisoners. Still, the inhumane conditions were difficult to witness, and the experience of being in prison left him shattered. Even after his early release for \"good behavior,\" he was never quite the same. Imprisonment left him ashamed, ended relationships, exacerbated his mercurial temper, and turned a gregarious, easygoing man into a solitary hard drinker (often consuming two quarts of whiskey a day)\u2014a habit that would kill him in the end.\n\nBut in prison, Porter was disciplined and productive in his writing, making the best of grim circumstances. A guard recalled his routine: \"After most of his work was finished and we had eaten our midnight supper, he would begin to write. . . . He seemed oblivious to the world of sleeping convicts about him, hearing not even the occasional sigh or groan from the beds which were stretched before him in the hospital ward, or the tramp of the passing guards. After he had written for perhaps two hours he would rise, make a round of the hospital, and then come back to his work again.\"\n\nHe was already a published author; his first short story, \"Miracle of Lava Canyon,\" appeared the year his wife died. He didn't use a pseudonym, exactly, but he did sign the story as the eminent-sounding \"W. S. Porter.\" For other stories, he'd toyed with various pen names: Sydney Porter, James L. Bliss, T. B. Dowd, Howard Clark, S. H. Peters, and Olivier Henry. Even in his personal correspondence, he sampled all sorts of names, signing letters as Panhandle Pete, S. P., Hiram Q. Smith, and so on. Later, working with the young editor Witter Bynner (who would become a poet and scholar), Porter almost never called him by his actual name. Instead, he addressed Bynner affectionately as Honored Sir, Doubleyou B, Mr. Man, Pal, My Dear Person, Willie, Witt, B. Binny, and Mr. Bitterwinter, among other appellations.\n\nFrom prison, Porter published more than a dozen stories, signing them \"O. Henry,\" the name with which he became the most widely read author of his time. He kept a small notebook in which he recorded the names of his stories and where they had been submitted. The first story he published as O. Henry was \"Whistling Dick's Christmas Stocking,\" which appeared in McClure's Magazine in 1899. Because he used an intermediary in New Orleans to submit his stories to editors, no one knew they were written by a convicted felon. His friend would place each story in a different envelope and then mail them from his own address.\n\nIn 1901, Porter was a free man. He'd made sure that Margaret had no idea where her father had been during his absence; she knew only that he was away on \"business.\" He'd written letters to her regularly from prison:\n\nJuly 8, 1898. MY DEAR MARGARET: You don't know how glad I was to get your nice little letter to-day. I am so sorry I couldn't come to tell you good-bye when I left Austin. You know I would have done so if I could have. Well, I think it's a shame some men folks have to go away from home to work and stay away so long don't you? But I tell you what's a fact. When I come home next time I'm going to stay there. . . . Now, Margaret, don't you worry any about me, for I'm well and fat as a pig and I'll have to be away from home a while yet and while I'm away you can just run up to Nashville and see the folks there. And not long after you come back home I'll be ready to come. And I won't ever have to leave again. . . . Look out pretty soon for another letter from me. I think about you every day and wonder what you are doing. Well, I will see you again before very long. Your loving PAPA.\n\nPorter was a changed man. He'd cut off several friendships rather than reveal the fact of his imprisonment. He had no wish to explain himself, and he hoped that no one would ever learn how he'd spent the past thirty-nine months of his life. He was determined to keep his secret and start anew.\n\nThe first step toward reinvention was no surprise: he shut down the name William Sydney Porter. Having adopted O. Henry in prison (and with no one able to trace it to an actual person), he made the transition easily. As William Porter, he was merely a journalist; as O. Henry, he was an author.\n\nIn 1902 he moved to New York City. The geographic change brought him closer to the center of the publishing industry and provided distance from his former self. In New York, where he had no friends or acquaintances, he was more prolific than ever, writing and publishing hundreds of stories. His popularity soared.\n\nFrom 1903 to 1907, Porter lived in Manhattan's Gramercy Park neighborhood, which had been created in 1831 by the developer Samuel Ruggles. The area was just as Ruggles had envisioned it: \"a bastion of civility and serenity.\" Over the years, Gramercy Park became known for its literary figures\u2014among them, Henry James, William Dean Howells, Herman Melville, Mark Twain, and the impoverished Stephen Crane, who lived with three aspiring artists in a tiny studio apartment. Melville, a customs inspector by day, was a resident for nearly thirty years, suffering there through the tepid reception of each of his novels, including Moby-Dick. Yet Porter\u2014or O. Henry\u2014is perhaps the author most closely identified with the neighborhood. He lived at 55 Irving Place in a first-floor brownstone apartment, and for the first time in his life, he was financially comfortable, having been given a contract by the New York World to write a weekly story, at the rate of a hundred dollars each.\n\nDespite the financial incentive, he often missed deadlines\u2014perhaps owing to his drunkenness. His editor refused to pay him until they arranged a compromise. For the first half of the story he delivered, he'd receive an advance; after submitting the other half, he'd be paid the remainder of his fee. Critics have noted that some of the beginnings and endings of O. Henry's stories seem disconnected, almost like Mad Libs. His quirky payment system might have had something to do with that.\n\nLater, as his fame grew, various stories were released about the origins of his pen name. Porter told the New York Times that he came across the surname \"Henry\" in the society pages of a New Orleans newspaper, and that he wanted something short for a first name. A friend suggested using a plain initial. \"O is about the easiest letter written,\" Porter decided, \"and O it is.\"\n\nThere was yet another version. After having dabbled in a number of pseudonyms, Porter took his name from Orrin Henry, a guard at the Columbus prison. Some said that the pseudonym came from the French pharmacist Etienne-Ossian Henry. Others said that the author had used \"O. Henry\" as an expletive so often that someone suggested it as his pen name.\n\nThe scholar Guy Davenport had his own rather dubious theory about the name, arguing that it was an assemblage from the first two letters of \"Ohio\" and the second and last two of \"penitentiary.\"\n\nSo, take your pick.\n\nIn 1904, Porter got a shock when he was asked to meet with an editor at the Critic, a monthly literary magazine. The editor said, \"You are O. Henry, are you not?\" Caught off-guard, Porter didn't deny it, but he did claim that there was no real mystery about writing under a different name. He hoped that a mundane story would defuse any desire by the editor to publish an expos\u00e9, and to dig into his past. He spoke as if confiding in the editor, saying that he was simply shy and averse to publicity, and that his lack of confidence had led him to use pen name. He then changed the subject, and hoped that the matter would go no further.\n\nBut a few weeks later, he picked up the new issue of the Critic and saw that the editor had proceeded with his scoop anyway. The article noted that the public was delighted by \"certain fantastic and ingenious tales\" bearing \"the strange device O. Henry as a signature.\" It went on: \"No one seemed to know the author's real name, and immediately vague and weird rumors began to be afloat and the nom de guerre was soon invested with as much curiosity as surrounds an author after his decease.\" Fortunately for Porter, the editor had simply published what he'd been told\u2014so now it would be known, at least to some, that Porter was O. Henry, but no one had connected him back to the bank teller who'd been arrested and convicted. \"[L]ike most mysteries, when it was probed there was no mystery,\" the article said of the unmasking. \"O. Henry's real name is Mr. Sydney Porter, a gentleman from Texas, who, having seen a great deal of the world with the naked eye, happened to find himself in New York.\" Porter's real secrets remained safe. Still, he fretted over how the Critic had found the story in the first place, who had tipped off the editor, and how the magazine had gotten hold of an old photograph of him to accompany the story. Luckily, the fact of Porter's pseudonym did not spread to the rest of the country right away. He could relax for a while, though he lived in fear that at any time he'd be found out and ruined. He decided that even if some people knew that he was O. Henry, he would at least minimize how much information was known about William Porter.\n\nAfter the publication of O. Henry's well-received Cabbages and Kings came The Four Million, in 1906, spreading his fame even further. The book included what would become his most celebrated story, \"The Gift of the Magi,\" with its famous opening:\n\nOne dollar and eighty-seven cents. That was all. And sixty cents of it was in pennies. Pennies saved one and two at a time by bulldozing the grocer and the vegetable man and the butcher until one's cheeks burned with the silent imputation of parsimony that such close dealing implied. Three times Della counted it. One dollar and eighty-seven cents. And the next day would be Christmas.\n\nThere was clearly nothing to do but flop down on the shabby little couch and howl. So Della did it. Which instigates the moral reflection that life is made up of sobs, sniffles, and smiles, with sniffles predominating.\n\nThe much-anthologized story is required reading for most students, but the story behind it is not well known. The night that the piece was due, his editor, in desperation, sent an illustrator out to track down O. Henry and extract it in person. When the illustrator arrived at the writer's apartment, he found that O. Henry had not even started. Supposedly, O. Henry then handed him a roughly drawn sketch and said, \"Just draw a picture of a poorly furnished room. . . . On the bed, a man and a girl are sitting side by side. They are talking about Christmas. The man has a watch fob in his hand. . . . The girl's principal feature is the long beautiful hair that is hanging down her back. That's all I can think of now, but the story is coming.\" Then he finished a few hours later.\n\nAs usual, the details of anything to do with William Porter are sketchy at best. According to another story about \"The Gift of the Magi,\" O. Henry wrote the entire story in a booth at Pete's Tavern, near Gramercy Park\u2014a bar established in 1864 whose tagline is \"The Tavern That O. Henry Made Famous.\" He is said to have gone to Pete's every morning. When he was in the midst of writing, though, he would order a bottle of Scotch to be delivered to him.\n\nGilman Hall was the magazine editor who'd given Porter his first writing contract, and they became friends. \"I was sure that he had a past,\" he once recalled, \"though he did not tell me of it and I did not inquire into it. It was not till after his death that I learned of the years spent in Columbus. I used to notice, however, that whenever we entered a restaurant or other public place together he would glance quickly around him as if expecting an attack.\"\n\nPorter did a fine job of keeping the most painful parts of his past a secret. In a wide-ranging interview he gave to the New York Times in the spring of 1909, the reporter George MacAdam commented that \"so far as the public is concerned, all he will do is to materialize between the covers of magazine and book . . . while he himself remains invisible behind the pen name.\"\n\nNoting that \"for the past six or seven years O. Henry has been one of the most popular short-story writers in America,\" MacAdam mentioned that even though \"he has kept himself under a bushel,\" his real name was now well known, having \"leaked from a hundred and one different sources.\"\n\nThe Times was clearly proud of having obtained unprecedented access to its elusive subject. MacAdam showed a dash of smugness in pointing out, \"Many are the interviewers who have sought him, but he has turned a deaf ear to their siren song.\"\n\nNow Porter was talking, but he wasn't necessarily telling the truth. \"Let me see: I was born in 1867,\" he told the reporter. (He wasn't.) Taking out a pencil and a scrap of paper to calculate his age, he added, \"That makes me 42, almost 43 years old, but put down 42.\"\n\nHe was asked what he had done after The Rolling Stone had ceased publication.\n\n\"A friend of mine who had a little money . . . suggested that I join him on a trip to Central America,\" he said, \"whither he was going with the intention of going into the fruit business.\" (Or, more accurately, whither Porter was going to avoid being sent to prison.) After that, instead of mentioning where he'd actually spent the next three years, he said that he moved to New Orleans and \"took up literary work in earnest.\" If by \"New Orleans,\" he meant \"Columbus, Ohio,\" then yes, he was telling the truth. There was no mention of his years in Austin, his years in prison, or even his marriage and daughter.\n\nHis few straightforward responses in the interview came when he was asked to talk about his writing. On his advice to young writers: \"I'll give you the whole secret to short story writing,\" he said. \"Rule 1: Write stories that please yourself. There is no Rule 2.\" And on the virtues of his work, he said, \"People say I know New York well. Just change Twenty-Third Street in one of my New York stories to Main Street, rub out the Flatiron Building, and put in the Town Hall and the story will fit just as truly in any upstate town. At least, I hope this can be said of my stories. So long as a story is true to human nature all you need do is change the local color to make it fit in any town.\"\n\nA woman who knew Porter socially in New York once spoke of how difficult it was to engage him in conversation, except superficially, because \"he protected himself from the crude and rude touch of the world in a triple-plated armor of mirth and formality.\" He bristled at personal questions (though he didn't mind reminiscing about his early years in North Carolina), and felt most at ease in the role of raconteur. \"His wit was urbane, sophisticated, individual; entirely free from tricks and the desire to secure effects,\" the woman recalled. \"It was never mordant nor corrosive; it did not eat or fester; it struck clean and swift and sure as a stroke of lightning.\"\n\nIt must have flattered him when, in his early days in New York, as his fame was growing and people began to speculate about his true identity, at least one impostor emerged. Gilman Hall recalled that only a few editors knew who O. Henry was and where he lived. An editor from a competing magazine boasted to Hall one day that he'd just learned that \"the real O. Henry\" was a college undergraduate who'd \"admitted\" that he was the author. Hearing this, Hall laughed and informed the editor that the real \"real O. Henry\" had in fact just left his office. When Hall related the amusing anecdote to Porter, he replied that so long as the paychecks were sent to the right man, he didn't care how many other aspiring O. Henrys there were.\n\nHaving established himself as an important writer was all the more reason to guard his privacy\u2014particularly any unsavory aspects of his past that didn't conform to his image as a man of letters. His rise to prominence was remarkable: one critic argued that O. Henry \"took the place of Kipling as a literary master,\" and said that on \"the shelf of my prized American classics\" were Poe, Thoreau, Whitman, Crane, Sarah Orne Jewett, W. D. Howells\u2014and O. Henry.\n\nAnother critic insisted that O. Henry should be considered a source of national pride: \"More than any author who ever wrote in the United States, O. Henry is an American writer. And the time is coming, let us hope, when the whole English-speaking world will recognize in him one of the great masters of modern literature.\"\n\nPorter's personal life, too, had finally brought a measure of happiness\u2014if short-lived, yet again. In 1905, after reading one of O. Henry's short stories, a childhood friend from Greensboro, Sara Lindsey Coleman, wrote a letter to the author inviting him to visit her. She'd gained her own impressive reputation as a short story writer, albeit locally, in North Carolina. Her family was prominent, as her father had served as a colonel in the Confederate army. She was witty and gracious, and Porter corresponded with her for a while before inviting her to come visit him in New York. (A diehard southerner, she admitted to him that she loathed the city.) Upon seeing her again, on his forty-fifth birthday, Porter fell in love and proposed. He confessed the entire (true) story of what he said was his wrongful imprisonment, and his journey to becoming a writer. They were married on November 27, 1907, in Asheville, and Gilman Hall served as best man. But within two years, owing mostly to Porter's alcoholism, the marriage deteriorated. They never divorced, however. His wife lived until the age of ninety-one; she died in North Carolina in 1959. She outlived even her husband's daughter: Margaret died in California at the age of thirty-seven.\n\nBut 1907 was a good year for the author: he was married and at the height of his fame. The third O. Henry story collection, The Heart of the West, was published, as well as a fourth, The Trimmed Lamp. He repeated the same feat for the next few years, issuing two story collections annually\u2014but these were his final years. (He would die at the age of forty-seven.) Porter had begun to resent his success and admitted that he felt constrained by it. Everyone by now knew what an \"O. Henry story\" was, and even he had tired of his predictable story structure. He boasted that he would write a novel, but he never did.\n\nAlthough his fame was accompanied by a very comfortable income, Porter was perpetually in debt. He used his earnings to buy Scotch, wine, and beer; tipped waiters at restaurants in amounts that matched the check for his meal; gave money freely to panhandlers; and generously treated his friends. He was compulsive in his giving, always ending up flat broke himself. Some of his debt, apparently, could be traced to silencing blackmailers. One woman from Austin was prepared to reveal to the press that he was a convicted embezzler. For her silence she requested a thousand dollars, an astronomical sum at the time, and he caved in to her demands. Perhaps fearing that she could be arrested for blackmail, she left Porter alone and never approached the media with her story.\n\nDespite the agony he had suffered over his past and the memories that haunted him, he received adulation from the public. Fans wrote to him asking for autographs, inscribed books, and photographs (which he usually declined to provide).\n\nBy 1909, his wife was living in North Carolina with her mother while Porter remained in New York. When he saw a doctor that summer, he was told that he had an enlarged heart, bad kidneys, and a severely compromised liver. During periods of relative recovery, he smoked and drank heavily, in denial that he was killing himself, and was more deeply in debt than ever.\n\nOn June 3, 1910, his kidneys failed. He called for help, then passed out. When he arrived by taxi (at his insistence) at New York Polyclinic Hospital on East Thirty-fourth Street, he wanted to protect his privacy. He requested permission to register under an assumed name, and as if casually checking into a hotel, he signed in as \"Will S. Parker.\" Following an emergency operation, his condition stabilized, and his wife began to make her way up to New York by train from North Carolina. She arrived too late to see him alive again.\n\nAt around midnight on June 5, Porter told a hospital nurse: \"Turn up the lights. I don't want to go home in the dark.\" He was dead before seven o'clock in the morning.\n\nHis career had been brief\u2014just under a decade\u2014but in that time he'd won international acclaim and his work was translated into a dozen languages. Two years after he died, Doubleday published a deluxe, limited edition of his collected stories, which included an original manuscript page with each copy. Only twelve were printed. Priced at $125, they sold out right away.\n\nIn the obituary that ran in the New York Times, Porter was called \"one of the best short story writers in America.\" The article also noted that a year before his death, \"O. Henry did something he was not in the habit of doing. He gave to the New York Times a story of his life, and it was the real story and not the invented narrative that went the rounds.\" (He died having fooled the Times.)\n\nThe \"real\" story came out only in 1916, in the first biography of O. Henry, which fully exposed the imprisonment of William Porter and the launching of O. Henry's writing career. Additional volumes of O. Henry's short stories were released posthumously, and continued to sell millions of copies. In 1918, the O. Henry Memorial Award Prizes were established, given each year to the best short stories published in the United States and Canada, and intended to \"strengthen the art of the short story and to stimulate younger authors.\" Doubleday published the first collection of prizewinning stories in 1919. Today, Porter is best known for this award, rather than his own work, but at the time it proved that his name, above all others, was synonymous with the short story.\n\nO. Henry was buried in Riverside Cemetery in Asheville, North Carolina. In honor of those famous first six words of \"The Gift of the Magi,\" visitors have made a tradition of leaving $1.87 at his grave\u2014money he would no doubt have spent if he could.\nHe died a virgin\n\nChapter 7\n\nFernando Pessoa & HIS HETERONYMS\n\nYou will never get to the bottom of Fernando Pessoa. There are too many of him.\n\n\"After looking for him in the poems, we look for him in the prose,\" wrote the scholar and translator Edwin Honig. Yet we find him nowhere. This was, after all, a poet whose maxim was, \"To pretend is to know oneself.\" Cyril Connolly noted that Pessoa \"hived off separate personalities like swarms of bees.\" He pretended relentlessly, employing more than seventy personae in his self-searching circus. They were not so much disguises as extensions and iterations of himself. \"How idyllic life would be,\" he once wrote, \"if it were lived by another person.\" When he looked in the mirror, he saw a crowd.\n\nFor some authors, the task of writing is a descent into the self. Pessoa ventured in the opposite direction, using his heteronyms as a means of departure and claiming that within his mini-populace, he was the least \"real\" and compelling of the bunch. The others were constellations swirling around him. In the context of psychoanalysis, a split identity is seen as a wound that needs healing. But in Pessoa's mind(s), there was nothing disorienting about it. \"I've divided all my humanness among the various authors whom I've served as literary executor,\" he explained. \"I subsist as a kind of medium of myself, but I'm less real than the others, less substantial, less personal, and easily influenced by them all.\"\n\nAlthough the basic facts of his life are now known, attempting to create a \"biography\" of Pessoa is a slippery task indeed. \"There never was a good biography of a good novelist,\" F. Scott Fitzgerald wrote in his journals. \"There couldn't be. He is too many people, if he's any good.\"\n\nFernando Ant\u00f3nio Nogueira Pessoa was very, very good.\n\nSome things about him can be said for sure. He was born on June 13, 1888, in Lisbon, Portugal, and spent his first seven years there. His surname, ironically, means \"person\" in Portuguese. He was five when his father, the music critic Joaquim de Seabra Pessoa, died of tuberculosis. Six months later, Fernando's infant brother, Jorge, died. His paternal grandmother suffered from episodes of insanity and was in and out of mental hospitals for the last twelve years of her life. After his father died, his mother, Maria Madalena Nogueria Pessoa, remarried, and the family moved to South Africa, where the boy's stepfather, Jo\u00e3o Miguel Rosa, served as the Portuguese consul of Durban, a British-governed town. By that time, the precocious Pessoa could read and write, thanks partly to his cultured, nurturing mother. He produced what is believed to be his first poem in the summer of 1895, when he was seven years old, in response to learning that the family would be moving to South Africa. The poem was called \"To My Dear Mother\":\n\nHere I am in Portugal,\n\nIn the lands where I was born.\n\nHowever much I love them,\n\nI love you even more.\n\nHe attended a primary school run by French and Irish nuns and became fluent in French and English. Later, at Durban High School, he was a brilliant student. He won awards and shunned sports. A former classmate, Clifford Geerdts, recalled a boy who was morbid, as well as \"meek and inoffensive and inclined to avoid association with his schoolfellows.\"\n\nPessoa gained three younger half siblings from his mother's second marriage: Henriqueta (with whom he was closest), Lu\u00eds, and Jo\u00e3o. He read and loved Keats, Shelley, Shakespeare, Dickens, Poe, and Byron. He began using false names to write: Charles Robert Anon, also known as C. R. Anon, and Alexander Search, for whom he printed calling cards. (Search once wrote a short story called \"A Very Original Dinner,\" in which the guests feast on human flesh.) Then there was Jean Seul, who wrote only in French. The shy boy created poems and stories, and even \"edited\" fake newspapers\u2014not unlike an early-twentieth-century version of The Onion\u2014with news, spoofs, editorials, riddles, and poems, all written by a staff of \"journalists\" who'd sprung from his imagination and whose biographies he'd made up. Later, in recalling his childhood, Pessoa wrote that \"[a]ny nostalgia I feel is literary. I remember my childhood with tears, but they're rhythmic tears, in which prose is already being formed.\" Nothing really mattered to him apart from his writing. Real life was beside the point. \"I've always belonged to what isn't where I am and to what I could never be,\" he once wrote, conceding his fixation on dreaming and escape. \"All I asked of life is that it go on by without my feeling it.\"\n\nIn 1905, at the age of seventeen, Pessoa returned to Lisbon to attend university. (He would never again leave the city.) Though he dropped out after two years, he got a fine education on his own by sequestering himself in the National Library to read literature, history, religion, and philosophy. He began writing short stories, some of them under the name \"David Merrick,\" as well as poems and essays, occasionally in Portuguese but more often in French and English.\n\nPessoa, who had very poor vision and wore glasses, lived with relatives or in rented rooms, chain-smoking, reading, writing, and earning a modest salary as a translator for companies that conducted business abroad. Later he worked as a bookkeeper. He had few friends. \"Since childhood I had the tendency to create around me a fictitious world, surrounding myself with friends and acquaintances that never existed,\" he wrote later. (As a boy, he'd invented the Chevalier de Pas, a faithful \"playmate\" who sent letters to him.) In 1910, the twenty-two-year-old admitted that \"[t]he whole constitution of my spirit is one of hesitancy and doubt. Nothing is or can be positive to me; all things oscillate round me, and I with them, an uncertainty unto myself.\" That his identity seemed so unstable was both distressing and consoling: \"Am I happy or sad?\" he asked in one poem. \"My sadness consists in not knowing much about myself. But then my happiness consists in that too.\"\n\nHis heteronyms, too, were filled with contradictions. \"In each of us there is a differingness and a manyness and a profusion of ourselves,\" wrote one of his mental offspring. This notion of endless expansiveness offered tremendous freedom. \"I suffer the delicacy of my feelings with disdainful attention,\" Pessoa explained, \"but the essential thing about my life, as about my soul, is never to be a protagonist. I've no idea of myself, not even one that consists of a nonidea of myself. I am a nomadic wanderer through my consciousness.\" Put it like that, and you can't help but envy him.\n\nIt is crucial to make the distinction that Pessoa's \"others\" were heteronyms rather than pseudonyms. He insisted that they were separate from him. \"I'm the empty stage where various actors act out various plays,\" he once wrote. In Pessoa country, unification was not possible or even desired. He was a breeder of beings, and always in pursuit of another. \"I break my soul into pieces,\" he wrote, \"and into different persons.\" He explained:\n\nA pseudonymic work is, except for the name with which it is signed, the work of an author writing as himself; a heteronymic work is by an author writing outside his own personality: it is the work of a complete individuality made up by him, just as the utterances of some character in a drama of his would be.\n\nAlthough Pessoa was timid and introspective and lived accordingly, he was no hermit. Nor did he attempt to hide his heteronyms\u2014he was quite transparent about the fact of their existence. Unlike many pseudonymous authors, Pessoa was not secretive but the opposite: utterly guileless, psychologically honest, earnest rather than serving up ironic posturing. His heteronymic conceit didn't spring from a desire to fool anyone or attract attention. This was a private matter.\n\nIn his writings, Pessoa went so far as to explain the genesis of his heteronyms; he understood that readers would be curious. Suggesting that the identities derived from \"an aspect of hysteria that exists within me,\" he diagnosed himself as either \"simply a hysteric\" or a \"neurasthenic hysteric,\" but leaned toward the latter. Also, he noted, \"The self-division of the I is a common phenomenon in cases of masturbation.\"\n\nHe claimed that the various people he had \"procreated\" often sent him greetings, and that he could hear and see them, even if no one else could. (\"Imaginary figures have more depth and truth than real ones,\" he once wrote.) Was this the result of talent or sickness? He stopped short of calling himself crazy. Throughout his life Pessoa grappled with the possibility of his insanity\u2014an anxiety undoubtedly fueled by his grandmother's illness\u2014but he was never able to draw conclusions about himself one way or the other. Perhaps he recognized that what mattered was being sound enough to produce his work. That he was so obsessively drawn to Shakespeare's Hamlet was more telling than he may have realized.\n\nHe argued that just as a novelist becomes annoyed when readers assume that a character's feelings and experiences are mere stand-ins for the author's own, so too should people accept that Pessoa's heteronyms were utterly separate from him. If the heteronyms occasionally happened to express his ideas, so be it; but this was not by calculation on his part, only chance. Although he acknowledged the strangeness of all this, he felt it was not for him to judge whether the heteronyms actually did or did not exist. Besides, he noted, he wasn't even sure which one, Hamlet or Shakespeare, was more real\u2014or \"real in truth.\" (He added that he had no proof that Lisbon existed, either.) Further, he said that he agreed with some of the theories expressed by his heteronyms but disagreed with others. All their work was dictated to him, yet they weren't seeking his advice or consent. He was not artist but amanuensis, nothing more.\n\nPessoa kept tight control over his social interactions, meeting acquaintances in coffeehouses and restaurants. One scholar noted that people who knew Pessoa described him as cordial, if inscrutable: \"He could be a delightful man, full of charm and good humor, a humor that was very British, though with none of the traditional grossness in it. But this role was also that of a heteronym, which saved him from intimacy with anyone while allowing him to take a modest part in the normal feast of daily life.\" A man who knew Pessoa in later years recalled, \"Never, when I bade him goodbye, did I dare to turn back and look at him; I was afraid I would see him vanish, dissolved in air.\"\n\nThere is no evidence that Pessoa yearned for more than his \"modest part\" in daily life, or that, in any case, he was willing to exert much effort. He once wrote that he wanted to be loved, but never to love: \"Passivity pleased me. I was only content with activity just enough to stimulate me, not to let myself be forgotten.\"\n\nHe was a lifelong outsider, but in 1910 he founded the magazine A \u00c1guia, and eventually he became part of the nascent Portuguese avant-garde, a group of intellectuals in Lisbon who founded a journal, Orpheu, introducing modernist literature to the country. Initially, it was ridiculed, but soon the publication won respect, and the criticism that appeared in Orpheu became highly influential. Only a few issues were released before it folded\u2014but within this group of intellectuals, Pessoa found a strong sense of kinship. He went on to work with other literary journals (both as editor and writer), publish chapbooks, issue a political manifesto called O Interregno, and start a press called Olisipo, which failed. For a London editor, he translated into English three hundred Portuguese proverbs. The years leading up to 1920 were most productive for this young bohemian.\n\nLiterary activity constituted his \"real\" life, but Pessoa paid the bills with his dreary day job, working as a clerk. (He had this dull occupation in common with fellow toiling authors Herman Melville, Franz Kafka, and Constantine Cavafy.)\n\nHe wrote and wrote\u2014in the daytime when he could, or else at night, and usually while standing up. On March 18, 1914, he had a kind of breakthrough: \"I wrote some thirty-odd poems, one after another, in a sort of ecstasy, the nature of which I am unable to define,\" he recalled. \"It was the triumphant day of my life. . . . What followed was the appearance of someone in me to whom I immediately gave the name Alberto Caeiro. Forgive the absurdity of the sentence: In me there appeared my master.\"\n\nCaeiro, the first of Pessoa's major heteronyms, had been \"born\" in 1889, lived with an elderly aunt in the country, and would die in 1915. He had \"no profession or any sort of education,\" was of medium height, pale, with blue eyes, and died consumptive. Once, Caeiro spoke in an \"interview\" of his humble accomplishments: \"I don't pretend to be anything more than the greatest poet in the world,\" he said. \"I noticed the Universe. The Greeks, with all their visual acuity, didn't do as much.\" He was joined by another heteronym, \u00c1lvaro de Campos, born in Tavira on October 15, 1890 (\"at 1:30 pm\"). Campos was a bisexual, unemployed naval engineer who'd studied in Glasgow and was now living in Lisbon. He was tall, Pessoa noted\u2014\"1.75 meters tall, two centimeters taller than I\"\u2014and \"slender with a slight tendency to stoop.\" He was \"fair and swarthy, a vaguely Jewish-Portuguese type, hair therefore smooth and normally parted on the side, monocled.\" And he was a dandy who smoked opium and drank absinthe. In him, Pessoa invested \"all the emotion that I allow neither in myself nor in my living.\" Ricardo Reis was a classicist and physician born in 1887 (\"not that I remember the day and the month, though I have them somewhere,\" Pessoa wrote) and living in Brazil. Pessoa explained that Reis \"is a Latinist by virtue of school training and a semi-Hellenist by virtue of his own efforts.\"\n\nThen there was the \"semi-heteronym\" Bernardo Soares, an assistant bookkeeper living in downtown Lisbon who \"seems always to be tired or sleepy.\" He was the closest to Pessoa's own voice, experience, and sensibility, and therefore the closest identity to a pseudonym. These men formed Pessoa's \"dramatic ensemble,\" and Campos even claimed that Pessoa did not exist.\n\nBecause he never had children of his own, Pessoa was father to his heteronyms, and they were quite a handful. There was the suicidal Baron of Teive, who produced just one manuscript, The Education of the Stoic, having allegedly destroyed everything else he had written. Raphael Baldaya was an astrologer. Maria Jos\u00e9 was a nineteen-year-old hunchback consumptive suffering madly from unrequited love. And Thomas Crosse was an ardent advocate of Alberto Caeiro's work. Yes, Pessoa's heteronyms actually critiqued\u2014sometimes savagely, sometimes kindly\u2014one another's writings. They also collaborated on projects (Crosse worked with his brother, I. I. Crosse) and translated one another's work. These diverse personae\u2014or, Pessoae, you might say\u2014wrote thousands and thousands of pages, and most of those texts were left behind as fragments to be transcribed and translated after Pessoa's death. It's a vast archive, much of it untouched even to this day.\n\nAside from Pessoa's almost spiritual devotion to his work, his life in Lisbon was uneventful and his routine predictable. He was a strange and lonely man. He smoked eighty cigarettes daily and drank a lot. He hated having his photograph taken. He never arrived on time for an appointment, always showing up too early or too late. He had terrible posture. He was very interested in the occult. He dressed formally, with a bow tie and homburg hat. Obsessed with horoscopes, he considered making his living as an astrologer. He produced horoscopes for himself, his acquaintances, and even his heteronyms. He lost some of the few friends he had to suicide.\n\nHe is known to have had only one significant love affair\u2014with a young woman named Of\u00e9lia de Queir\u00f3s. (She eventually married, and died in 1991.) When they met, the aptly named Of\u00e9lia was nineteen and working as a secretary at the same firm where the thirty-one-year-old Pessoa worked. He declared his love for her one day with lines taken from Hamlet, and then kissed her, she recalled, \"like a madman.\"\n\nAfter the failure of the relationship, Pessoa decided that love was a false notion, anyway. \"It's our own concept\u2014our own selves\u2014that we love,\" he wrote, arguing that \"the repression of love sheds much more light on its nature than does the actual experience of it.\" Yet Of\u00e9lia claimed that Pessoa was entirely to blame for their breakup. \"Little by little, he withdrew until we stopped seeing each other altogether,\" she recalled. \"And this was done without any concrete reason whatsoever. He did not appear or write for several days because, as he said, there was something wrong with his head and he wanted to go to the insane asylum.\" He had written her more than fifty letters\u2014some affectionate, drunk with love, others bitter and accusatory: \"Why can't you be frank with me?\" he demanded in March 1920. \"Why must you torment a man who never did any harm to you (or to anybody else) and whose sad and solitary life is already a heavy enough burden to bear, without someone adding to it by giving him false hopes and declaring feigned affections? What do you get out of it besides the dubious pleasure of making fun of me?\"\n\nElsewhere, he expressed moments of insecurity and alienation: \"I'm all alone\u2014I really am. . . . I'm going crazy from this sense of isolation and have no one to soothe me, just by being near, as I try to go to sleep.\" Yet he was just as quick to assume control and withdraw. \"By the way,\" he wrote a few weeks later, \"although I'm writing you, I'm not thinking about you. I'm thinking about how much I miss the days when I used to hunt pigeons.\" Pessoa also had Alvaro de Campos (\"Naval Engineer\") write to Of\u00e9lia on his behalf, explaining that his friend's \"mental state prevents him from communicating anything, even to a split pea.\"\n\nSome scholars contend that Pessoa was a latent homosexual who sublimated his sexual impulses.\n\nUltimately, the author remains, like his work, \"vastly unfinished, hopelessly unstructured, and practically unknown,\" as the Pessoa scholar and translator Richard Zenith has written. It is no accident that one volume of verse Zenith translated is titled Pessoa & Co. The Portuguese writer formed a Corporation of One, of which he was CEO and every employee from the top of the ladder to the bottom rung. Pessoa's dozens of constructed alternate selves, Zenith noted, \"were instruments of exorcism and redemption. They were born to save him from this life that he felt ill-equipped to live, or that offended his aesthetic and moral sensibilities, or that simply bored him.\" Although alter egos had become fashionable accessories for European writers in the early twentieth century, no one took the device as far as Pessoa did\u2014and certainly no one has done so since. As the scholar Jorge de Sena said in 1977, at the first international symposium on Pessoa's work (held at Brown University), Pessoa was hardly the first to eradicate any trace of autobiography from his writing. Yet de Sena noted that even though the alter egos of modernists such as Gide, Joyce, and Eliot produced masterpieces, they never went to the extremes that Pessoa did. He annihilated himself in the name of artistic creation. \"Unceasingly I feel that I was an other, that I felt other, that I thought other,\" Pessoa wrote. \"I am a spectator of myself. . . . I created myself, crevasse and echo, by thinking. I multiplied myself, by introspection. . . . I am other even in my way of being.\"\n\n\"Poets don't have biographies,\" Octavio Paz wrote in his introduction to A Centenary Pessoa. \"Their work is their biography.\" Who could make a stronger claim to this than Pessoa? \"I am, in large measure, the selfsame prose I write,\" he confessed. \"I unroll myself in periods and paragraphs, I make myself punctuation marks. . . . I've made myself into the character of a book, a life one reads.\"\n\nGeorge Steiner called Pessoa \"one of the evident giants in modern literature.\" John Hollander declared that if Pessoa had never existed, Jorge Luis Borges would have had to invent him. C. K. Williams praised Pessoa's \"amazing audacities, his brilliance and his shyness.\" Harold Bloom included Pessoa on a list of twenty-six writers he considered essential to the Western canon, including Dante, Shakespeare, and Proust, and argued that Pessoa was not a madman but a reborn Walt Whitman, \"who gives separate names to 'my self,' 'the real me,' or 'me myself,' and 'my soul,' and writes wonderful books of poetry for all of them.\"\n\nPessoa was the loving ringmaster, director, and traffic cop of his literary crew. He tended to each of their biographies with meticulous specificity, and attentively varied their styles, idioms, techniques, genres, ideologies, and interests. He killed some off and let others live. Whereas the work of poets is typically fed by outside stimuli, Pessoa's creativity seems to have fed off itself\u2014like one of the contemporary artist Dana Schutz's famous \"Self-Eater\" paintings. One persona stirred another and another, and perhaps that apparently arbitrary transmission of energy explains why so much of the work by Pessoa & Co. took shape in unfinished fragments. The ideas born of this collective were too much for one man to set down on paper. \"My character of mind is such that I hate the beginnings and the ends of things, for they are definite points,\" he explained.\n\nWhat was Pessoa aiming for with his menagerie? What drove him to it? Because \"true\" biographical information about him is so limited, it is difficult to say. All we have are his written accounts of his motives and the speculations of others. It seems that Pessoa was in pursuit of self-abdication. He wanted to escape both body and mind. \"Pessoa sought to expel not only his sexual desires,\" Zenith wrote, \"but his friendly affections, his religious tendencies, his aggressive feelings, his humanitarian urges, his longing for adventure, his dreams, and his regrets.\" Anyone attempting to define Pessoa reductively as a cluster of pathologies should think again. As Zenith noted, \"Psychoanalysis is too poor a science to explain the case of Pessoa, who seems to have been simply, mysteriously, possessed by a demon\u2014that of detachment.\"\n\nIn a 1977 interview, Edwin Honig echoed the notion of Pessoa's essential unknowability: \"Being both complex and simple, he is always hovering over some piece of mysterious ground, like moonscapes with mile-deep craters\u2014terribly attractive but also very forbidding.\" It's understandable that Pessoa has been compared to T. S. Eliot and Wallace Stevens, both masters of the elusive. \"Reading [Pessoa's] best poems,\" Honig said, \"you never know if you're plumbing the depths or if you're dangling there above without even touching ground. There's always that paradox in his secret, something unanswerable. Though he invites you to share it, he resists your advance the moment you accept the invitation.\" (This was not unlike his personal life. In work and in his social dealings, he always preferred a bit of distance.)\n\nBy taking leave of himself, becoming invisible to the extent that he could, he was free to roam in contradiction, paradox, and complexity without being labeled as this or that kind of writer. He could hold up mirrors, play with them, and then smash them to bits. As Borges wrote in his \"Ultra Manifesto,\" the true artist does not reflect himself, but razes himself and creates from there. \"Two aesthetics exist: the passive aesthetic of mirrors and the active aesthetic of prisms,\" he wrote. \"Guided by the former, art turns into a copy of the environment's objectivity or the individual's psychic history. Guided by the latter, art is redeemed, makes the world into its instrument, and forges\u2014beyond spatial and temporal prisons\u2014a personal vision.\"\n\nIn private life, Pessoa was a demure and awkward man. But his \"personal vision\" as a writer was startling and brave, anything but ordinary.\n\nMuch more than mere pseudonyms, Pessoa's heteronyms were so wildly different from one another that they allowed him to explore his imagination endlessly, without paying any price. Well, up to a point: that very messiness, the refusal to be defined as just one man, explains why he is not more widely known today. (Pessoa once described his oeuvre as \"a drama divided into people instead of into acts.\")\n\nCertainly to literary types he is a significant figure (the blessing of Harold Bloom is no small thing), but his books are not easily found. It's true that more of his work has been translated into English over the past decade, but Pessoa hardly helped the matter of his legacy: he left behind a trunk full of journalism, cultural criticism, philosophy, plays, poems, political essays, and horoscopes, much of the work illegible and unfinished. The trunk was discovered, after his death, in his rented room in Lisbon.\n\nThe material\u2014nearly thirty thousand manuscript pages\u2014is daunting for even the most intrepid scholar to sift through. Some have begun, then abandoned, their Pessoa projects. The task of deciphering, organizing, and translating his work is still in progress, and perhaps will never be finished. Pessoa wrote haphazardly in different languages, on loose scraps of paper, in journals and notebooks, on the backs of envelopes, and on the official stationery of the firms for which he worked. As Richard Zenith has written, the work stands \"like variously sized building blocks\u2014some rough, others exquisitely fashioned\u2014of an impossible but marvelous monument.\" Pessoa didn't care for cohesiveness in any area of his life. Yet the quality of much of these thousands of texts, however fragmented or arbitrary, is generally exceptional; these are much more than the ramblings of a crazy person.\n\nIn his lifetime, he wasn't quite the Emily Dickinson of Lisbon\u2014except for having apparently died a virgin. Mostly he kept to himself, to be sure, but he also published hundreds of poems, journalistic pieces, and essays. He became a respected intellectual figure, if not quite a celebrity, yet his literary genius was not widely recognized until after he died. In his home country he is now considered the greatest Portuguese poet since Lu\u00eds de Cam\u00f5es, the sixteenth-century author of the epic Os Lus\u00edadas (which Pessoa is said not to have cared much about). He is also regarded as one of the greatest modernists in any language and is one of the most fascinating figures in the history of literature.\n\nOn November 29, 1935, the forty-seven-year-old Pessoa suffered from abdominal pain and developed a high fever. He was taken to the Hospital de S\u00e3o Lu\u00eds in Lisbon, where he wrote, in English, his last words: \"I know not what tomorrow will bring.\" The next day he died from cirrhosis of the liver.\n\nA statue of Pessoa now stands near one of the coffeehouses he used to frequent. At the time of his death, those who knew his work understood that the country had lost an important man. \"Fernando Pessoa is dead,\" a young doctor (later to become a distinguished literary figure) named Miguel Torga wrote in his journal. \"As soon as I heard the news in the paper, I closed my surgery and plunged into the mountains. There, with the pines and the rocks, I wept for the death of the greatest poet of our times, whom Portugal watched pass by in his coffin, on his way to immortality, without even asking who he was.\"\n\nIn the opening lines of what is perhaps his best-known poem, \"The Tobacco Shop,\" Pessoa declares:\n\nI'm nothing.\n\nI'll always be nothing.\n\nI can't want to be something.\n\nBut I have in me all the dreams of the world.\n\nHe was someone who felt like \"nothing\" to such an extent that he strove for self-expulsion, yet like Whitman, he contained everything that he needed, desiring nothing from the universe beyond his imagination. His statement presents the speaker as both meek and grandiose: I have nothing, I am nothing, but don't you wish you had what I have? Don't you wish to be what I am? Pessoa's self-abnegation is the source of his power and vitality. In his free-floating way, he implicates us, his readers, in the telling and interpretation of his story. As he wrote in his posthumously published masterpiece, The Book of Disquiet:\n\nI am the suburb of a non-existent town, the prolix commentary on a book never written. I am nobody, nobody. I am a character in a novel which remains to be written, and I float, aerial, scattered without ever having been, among the dreams of a creature who did not know how to finish me off.\n\nPessoa has been dead for decades. We haven't even begun to finish him off.\nHe slept with prostitutes, hated bad smells, and dressed like a tramp\n\nChapter 8\n\nGeorge Orwell & ERIC BLAIR\n\nHad Eric Arthur Blair been a working-class bloke from Birmingham instead of an Old Etonian, George Orwell might never have existed. By the age of six, Blair aspired to become a writer, and as a young man he knew that he wanted to explore the lowest stratum of society in his work. Given his genteel family background, this kind of subject matter might have been problematic. If he wanted to write, he would have to conceal himself.\n\nBlair was born on June 25, 1903, in Motihari, a village in colonial India near the Nepalese border. His parents were stationed there while his father, Richard, held a minor post with the Indian Civil Service. They were not wealthy\u2014Eric would later describe his family as \"lower-upper-middle-class\"\u2014but both his parents came from prominent families in decline. Richard was descended from West Indian slave owners (his great-grandfather was rich and had married the daughter of an earl), and was instilled with a strong sense of public service; Blair's mother, Ida, grew up in Burma, the daughter of a French timber merchant who himself came from a distinguished family of artisans.\n\nIda was working at a boys' school in India when she met Richard, who was thirty-nine, unmarried, and in a dead-end job that paid poorly. He was eighteen years older than Ida. They married in 1897, and she gave birth to a daughter, Marjorie, the following year. (Another daughter, Avril, was born five years after Eric, in 1908.) Without being affluent, they enjoyed the usual perks of colonial life, including servants and access to a whites-only club. Soon after Eric was born, Ida took the children back to England; she wanted them to enjoy a comfortable middle-class existence (and education) in Oxfordshire. In the town where they settled, which dated to the fourteenth century, Ida found an active social life, something she'd missed terribly.\n\nGrowing up, Blair was keenly aware of his family history and of the divisions of caste and class systems. He would later reject organized religion and declare himself an atheist, but he had a strong sense of moral duty (even when he didn't live up to his own code, which was often). He was a stubborn, sensitive, and studious boy who loved reading Dickens, Swift, Defoe, and especially Kipling, whom he called a \"household god\" and whose work would greatly influence his own. His mother recorded his first word, uttered when he was eighteen months old: \"beastly.\"\n\nIn temperament, Blair was more like his soft-spoken, introverted father than his outgoing, chatty mother, and he found unbearable the frivolous tea parties he was forced to attend. \"As a child I was taught to say 'Thank you for having me' after a party, and it seemed to me such an awful phrase,\" he recalled later. Ida loved being part of a well-to-do social set, playing croquet, shopping, going to theater and music events in London, attending a local regatta, and watching tennis at Wimbledon. Yet she was an attentive, loving mother, and Blair is said to have inherited his vicious wit from her. \"I barely saw my father before I was eight,\" he recalled in the opening of his 1946 essay, \"Why I Write.\" \"For this and other reasons I was somewhat lonely, and I soon developed disagreeable mannerisms which made me unpopular throughout my schooldays.\"\n\nComing of age in Edwardian England, when \"the sheer vulgar fatness of wealth\" was everywhere, and \"without any kind of aristocratic elegance to redeem it,\" Blair assumed the stance of a critical outsider. \"[T]he social status of nearly everyone in England could be determined from his appearance, even at two hundred yards' distance.\" Social order was not an abstract notion; it was present in his everyday life, and he was made to understand its significance both at home and at school. \"I was forbidden to play with the plumber's children; they were 'common' and I was told to keep away from them,\" he wrote in his 1937 book The Road to Wigan Pier. \"This was snobbish, if you like, but it was also necessary, for middle-class people cannot afford to let their children grow up with vulgar accents. So, very early, the working class ceased to be a race of friendly and wonderful beings and became a race of enemies.\"\n\nIt was partly because of his chronic ill health that Blair was highly attuned to disparities in social conditions. (He had defective bronchial tubes and a lesion in one lung, which was not diagnosed until later in his life.) He knew what it was to feel helpless, to feel apart from one's own community, to be judged as weak and inferior. He was not yet two years old when he endured a bout of bronchitis, the first of many (along with influenza) to recur throughout his life. A decade after failing an army medical exam in 1940, Blair would be dead of tuberculosis at the age of forty-six. His entire life was spent with a sense of urgency regarding his work, with the constant knowledge that he was running out of time. \"Until I was about thirty I always planned my life not only on the assumption that any major undertaking was bound to fail, but that I could only expect to live a few years longer,\" he once wrote.\n\nPerhaps because he was confined to bed so often as a child, in enforced solitude, he developed a rich imagination. He believed in ghosts and was enchanted by ghost stories. He also believed that his dreams had symbolic meaning and were sometimes prescient. And he was highly superstitious, a believer in black magic. When his father died in 1939, he placed pennies on Richard's eyes and threw the pennies into the sea.\n\nIn fact, years later, Blair is said to have thought that assuming a pseudonym meant no one could use his real name against him for evil purposes. The notion of peeling off identities appealed to him, anyway. \"I had the lonely child's habit of making up stories and holding conversations with imaginary persons,\" he later recalled, \"and I think from the very start my literary ambitions were mixed up with the feeling of being isolated and undervalued.\"\n\nBlair produced his first poem at the age of four, dictated to his mother. Seven years later, in 1914, he published an exuberantly patriotic poem in a local newspaper, with the opening stanza:\n\nOh! Give me the strength of the lion,\n\nThe wisdom of Reynard the fox,\n\nAnd then I'll hurl troops at the Germans,\n\nAnd give them the hardest knocks.\n\nHe also produced what he later described as \"bad and unfinished 'nature poems' in the Georgian style,\" a rhyming play, and short fiction\u2014most of which he regarded as embarrassing. But he recognized his facility with language and his love for it. \"From a very early age, perhaps the age of five or six, I knew that when I grew up I should be a writer,\" he later recalled. \"Between the ages of about seventeen and twenty-four I tried to abandon this idea, but I did so with the consciousness that I was outraging my true nature and that sooner or later I should have to settle down and write books.\"\n\nEven in childhood he was cultivating \"the making up of a continuous 'story' about myself, a sort of diary existing only in the mind. . . . As a very small child I used to imagine that I was, say, Robin Hood, and picture myself as the hero of thrilling adventures, but quite soon my 'story' ceased to be narcissistic in a crude way and became more and more a mere description of what I was doing and the things I saw.\" (That storytelling self would later be manifested as George Orwell.) He always felt a need to describe things, events, and people, and his early stories were, if nothing else, impressive in their descriptive quality.\n\nIn 1911, a fateful event occurred: Ida decided to send her son away to St. Cyprian's, a fashionable preparatory school in Sussex for boys aged eight to thirteen. The five years he spent there traumatized him and filled him with contempt, yet the school's \"values\" did shape his socialist views\u2014and proved formative in the making of George Orwell, whom V. S. Pritchett called \"the conscience of his generation.\"\n\nHe set down an account of his sufferings at the \"expensive and snobbish school\" in the ironically titled, fifteen-thousand-word essay, \"Such, Such Were the Joys,\" which took him years to write. (It was not published in the UK until 1968, after the widow of the cruel headmaster died.) Soon after his arrival at St. Cyprian's, he recalled, \"I began wetting my bed. I was now aged eight, so that this was a reversion to a habit which I must have grown out of at least four years earlier.\" The guilt and self-mortification he'd acquired from a Catholic school education was exacerbated by his time at St. Cyprian's. \"[I]t was looked on as a disgusting crime which the child committed on purpose and for which the proper cure was a beating,\" he wrote. \"Night after night I prayed, with a fervour never previously attained in my prayers, 'Please God, do not let me wet my bed! Oh, please God, do not let me wet my bed!' but it made remarkably little difference.\"\n\nFor such an elite institution, \"the standard of comfort was in every way far lower than in my own home,\" Blair recalled bitterly, \"or, indeed, than it would have been in a prosperous working-class home.\" He found that there was never enough food, and what was available tasted awful\u2014including porridge containing unidentifiable black lumps. (He resorted to stealing stale bread from the pantry in the middle of the night.) The boys were allowed a hot bath only once a week, and the towels were damp, with a foul smell. \"Whoever writes of his childhood must beware exaggeration and self-pity,\" Blair admitted. \"But I should be falsifying my own memories if I did not record that they are largely memories of disgust.\"\n\nHe was surrounded by boys boasting about \"my father's yacht,\" \"my pony,\" \"my pater's touring car,\" and the like. \"How much a year has your pater got?\" \"What part of London do you live in?\" \"Is that Knightsbridge or Kensington?\" \"Have you got a butler?\" and \"How many bathrooms has your house got?\" were the kinds of interrogations intrinsic to the school's culture. The boys were constantly keeping score and ranking themselves socially above or below their peers; Blair was always below. He was well aware that aside from money or a title, he lacked every other virtue that might bolster his standing\u2014athleticism, good looks, confidence, and charm.\n\nOne of his few friends at St. Cyprian's was the future literary critic Cyril Connolly, who later recalled Blair's appearance as grotesque: \"Tall, pale, with his flaccid cheeks, large spatulate fingers, and supercilious voice, he was one of those boys who seem born old.\"\n\nBlair was bullied as well as beaten at school, and he found no comfort in his holidays at home. His father, now fifty-five, had retired with a modest pension and returned to the family. Ida, having been left alone to raise three children, was chilly and remote to Richard. He was a stranger to Eric and did nothing to cultivate closeness between them. Eric felt no love for his father, and was mortified by Richard's habit of removing his false teeth and setting them on the table at mealtime. \"Most of the good memories of my childhood, and up to the age of about twenty, are in some way connected with animals,\" he wrote in \"Such, Such Were the Joys.\"\n\nEven then, as a morose and timid boy, haunted by \"a sense of desolate loneliness and helplessness,\" as he once recalled of his younger self, Blair aspired to greatness. He knew he would become an author someday\u2014and not just any, but a famous one. He announced that his writing name would be the distinguished-sounding \"E. A. Blair,\" rather than \"Eric Blair,\" which he deemed too plain.\n\nHis education at St. Cyprian's prepared him for a spot at the Mount Olympus of English public schools, Eton, where he was awarded a scholarship in 1916. (His parents could never have afforded the full boarding and tuition fees.) But Blair had already been worn down by his unhappy experience at St. Cyprian's, and he hated Eton. He felt more miserable than ever, and even more alone. One of his classmates once described him in even more unflattering terms than did Cyril Connolly (who also attended Eton with Blair), as having had \"a large, rather fat face, with big jowls, a bit like a hamster.\" Another said that Blair was \"pretty awful\" and \"a bit of a bastard.\" One boy Blair particularly disliked was Philip Yorke\u2014the oldest brother of Henry Yorke, who would assume the authorial name Henry Green.\n\nAt Eton, Blair cranked out stories and plays in his notebooks, all of which he signed \"Eric the FAMOUS AUTHOR.\" He savored a few aspects of his time there, including having been taught French by Aldous Huxley. And his reading experiences were extraordinary: Zola, Maupassant, Flaubert, Twain, and Milton were among his favorites. He also took up smoking and cultivated a rebellious streak, which won him the admiration of his peers.\n\nAt St. Cyprian's, Blair had at least soared academically, but at Eton his grades were poor. His tutor found him lazy and impudent. Upon graduation, for reasons he never explained, Blair took a commission with the Indian Imperial Police in Burma, where he spent five monotonous years feeling exiled. (Among his few pleasures were frequent visits to Burmese brothels, which, as a sexual late bloomer, he found addictive.) Perhaps his decision to enter government service was an easy way to deal with his confusion about what to do next, and to figure out what kind of man he should become. The experience would buy him time. Cambridge and Oxford\u2014the two universities of destiny for Eton's finest\u2014held no interest for him, and in any case he was considered by Eton to be \"unsuitable\" for either. That was upper-class code for \"an embarrassment.\"\n\nIn 1927, Blair returned to England a heavy smoker, gaunt (having suffered his usual bronchial problems, along with dengue fever), and, as one of his parents' neighbors noted, someone who \"looks as though he never washes.\" His classmates from Eton had already started to publish and even achieve renown. Although Blair would eventually exorcise the bad memories from his time in Burma, which represented wasted years (and lost innocence), in his 1934 novel Burmese Days, for the time being he was still six years away from his publishing debut, Down and Out in Paris and London.\n\nHe rented a cheap room in Notting Hill for a while to fashion himself into a \"FAMOUS AUTHOR,\" but it wasn't until he set off for Paris that things seemed to click into place. Like so many other literary expatriates, Blair felt that in Paris life would truly begin. He arrived in 1928 in search of culture, education, writing material, and undoubtedly romance. (Brothels were legal at the time, so sex could be obtained one way or another.) The city had a buzz that dour London seemed to lack. Henry Miller was there, as were Joyce, Hemingway, and Fitzgerald, among other famous writers.\n\nBlair soon managed to complete his first novel, but when it was rejected for publication he burned it. At that time his heart was still set on fiction\u2014he had no intention of becoming a celebrated essayist, even though he was deeply political (while refusing to join any one party) and interested in provocative reportage. He wasn't sure how he intended to use the sketches he wrote about the beggars and tramps he encountered on the city's streets, but \"common people\"\u2014the kind he'd been raised to ignore, like a good and proper snob\u2014interested him most. The self-declared socialist was drawn to down-and-out types much more than to writers or artists, and least of all to anyone with the odor of affluence.\n\nWith political unrest brewing in Europe, Blair eased up on his single-minded focus on fiction; he needed money. Even though he also wrote poetry, he realized that no earnings would come of it. He started writing for a left-wing weekly publication and other newspapers, with an eye toward stories with sociological and political issues\u2014in particular the implications of censorship (exploring ideas that would incubate and later shape his dystopian masterpiece Nineteen Eighty-four), and the homeless. He started signing these pieces \"E. A. Blair.\"\n\nUnfortunately, he was hardly getting by in Paris; he would learn firsthand what it felt like to be impoverished. His experiences there felt desultory. He was reduced to fishing (without success) in the Seine, rationing his food supply, and even pawning some of his possessions. \"I underwent poverty and the sense of failure,\" he recalled of his time in Paris. \"This increased my natural hatred of authority and made me for the first time fully aware of the existence of the working classes.\"\n\nAfter doing menial work and finding it wretched, he was pleased that a publication in London had accepted one of the essays he'd submitted. He decided that moving back to London would not signify failure but offer greater potential for becoming a professional writer. In December 1929 he left Paris and returned to his parents' house. \"England is a very good country when you are not poor,\" he wrote a few years later. Still, it was better to struggle in his own country than in France.\n\nIn no way embarrassed by having to work as a babysitter and take occasional odd jobs, Blair (who looked like a bum) started writing a nonfiction book about beggars and outcasts, based on his own experiences, which would evolve into Down and Out in Paris and London. He also began publishing criticism. It didn't earn him much money, but he established himself as a respected reviewer, or at least the beginnings of one.\n\nThough slowly finding his way toward his vocation, Blair didn't fit neatly into any single category: he came from a snobbish family that was not wealthy; he'd been given the most prestigious public school education a student could hope for\u2014yet unlike many of his contemporaries, who had already achieved fame and wealth, he had little to show for it. He disowned Eton but wore it as a badge of honor. He spoke in a posh accent but dressed in ill-fitting, rumpled clothing. And having immersed himself in Shakespeare, Chaucer, Twain, Poe, Ibsen, Dickens, and Thackeray, among others, he was well read and intellectual, but he had rejected a university education. Although he was bitter about not having gone to Oxford or Cambridge, it was also a point of pride that he had not. He was austere, but he enjoyed comfort. He was stridently political and deplored politics. He was unlucky in love and perpetually unable to sustain relationships with women. (Prostitutes, however, he did fine with.) He appeared to love women and despise them; even some of his friends described him as a misogynist. He sought out tramps and beggars, yet he was an intellectual snob and ill at ease in the presence of those who did not share his interests. He relished immersing himself in vagrant life, but had a pathological aversion to bad smells and dirt, and was oblivious to the foul stench of his own smoking habit. He was happy only when writing, but no matter how hard he worked, he couldn't earn a living doing it. He was frustrated by his frequent illnesses, which kept him from writing, yet he did not take responsibility for his health\u2014he smoked heavily even while coughing up blood. All the intriguing contradictions of Eric Blair would find their way into the work of George Orwell. Blair might be judged by others as mentally unstable, paranoid, troubled, sadistic, and aberrant\u2014but George Orwell? He was a noble and brilliant author.\n\nRegardless of his quirks, and there were many, it was almost unnerving to see how little Blair cared about others' opinions of him. Still, he kept his writing ambition largely private. Slowly, Blair was developing a pioneering, novelistic style that blended reportage and memoir. His work was investigative yet highly personal, driven by a sense of moral outrage at social injustices. (The genre might be called Proletarian Lit\u2014not exactly sexy stuff.) He had also taken to hanging out with vagrants in London and sometimes dressing like a tramp, sleeping in Trafalgar Square covered in newspapers. \"He didn't look in the least like a poor man,\" a friend recalled of Blair decades later. \"God knows he was poor, but the formidable look didn't go with the rags.\"\n\nNor did the rags go with the name Eric Arthur Blair. It was time to invent George Orwell. Blair had always been secretive in every respect; adopting a pseudonym would allow him to release the various facets of his personality. Doing so was not without some degree of shame: he wrote in A Clergyman's Daughter (in which a character uses a pseudonym) that \"[i]t seemed a queer thing to have to do, to use a false name; dishonest\u2014criminal, almost.\"\n\nAs Eric Blair, he accepted a teaching job at a boys' school\u2014hardly a posh one\u2014which would make the twenty-nine-year-old seem somewhat respectable in the eyes of his parents. (Even though he had no university degree, his Eton schooling was impressive enough to win him the job.) He was bored by the work, and described the school as \"foul.\"\n\nThat summer, he received the best news he'd heard in a long time: he'd found a publisher for Down and Out in Paris and London. Under a different title, the manuscript had been rejected by Jonathan Cape, and also by T. S. Eliot at Faber and Faber\u2014Eliot's elitist sensibility did not exactly savor tales of the malodorous downtrodden. In refusing the book, he wrote to Blair that it was \"too loosely constructed.\"\n\nThe final version of the manuscript was a semiautobiographical story narrated by an anonymous, penniless English writer\u2014or, rather, it was a collection of essays about Blair's own experiences, recounted in fictionalized form. Most of the events in the book had occurred, but some fabrications were thrown in. It was startling for its up-close exploration of street people and others left behind by society. It was also a shocking expos\u00e9 of harsh, filthy, inhumane conditions in the restaurant kitchens of Paris, where Blair had toiled as a lowly dishwasher.\n\n\"I can point to one or two things I have definitely learned by being hard up,\" the narrator reflects in the book's final paragraph. \"I shall never again think that all tramps are drunken scoundrels, nor expect a beggar to be grateful when I give him a penny, nor be surprised if men out of work lack energy, nor subscribe to the Salvation Army, nor pawn my clothes, nor refuse a handbill, nor enjoy a meal at a smart restaurant. That is a beginning.\"\n\nThe publisher Victor Gollancz had accepted the work and paid Blair an advance of forty pounds. After some discussion about the title and potential libel issues, there was one significant matter to settle: the name of the author. Blair had informed his agent that he wished to use a pseudonym. \"If by any chance you do get it accepted,\" he wrote, \"will you please see that it is published pseudonymously, as I am not proud of it.\" (Perhaps he was ashamed by the rejections he'd received, and certain that his execrable book was doomed to failure.) Then there was the matter of his family: he did not want to embarrass them with sordid (if thinly disguised) tales of his adventures. He also wrote to Gollancz that \"if the book has any kind of success I can always use the same pseudonym again.\" The editor suggested simply signing the book with the letter \"X,\" but Blair wished to find a suitable name, perhaps thinking about his future career. He had trouble settling on a nom de plume, so he sent Gollancz four suggestions: H. Lewis Allways, P. S. Burton, Kenneth Miles\u2014and George Orwell, which was his favorite.\n\nHis anxiety about concealing his authorship from his parents may have been genuine, but he didn't try very hard. Portions of the book had already appeared in literary periodicals under his own name; he confessed to his sister Avril that he was publishing his first book using a pseudonym; and he allowed his mother to read the book. Still, the pen name at least shielded the family from public scrutiny. It seems that another compelling reason for using an alter ego was the fact of his background. How credible was it for an Eton graduate to go undercover by living on the margins of society, rejecting respectability, and plunging himself into the lives of outcasts? It could also be perceived as highly offensive that such a genteel young man would \"slum it\" for the sake of creating a literary masterpiece. For him, vagrancy was a choice: if his situation became too dire, he could always borrow money from his mother; and he could find a place to sleep whenever he wished. He was certainly in a bad way, yet he could afford to be a part-time tramp; it was a role to play more than anything else.\n\nWriting about poverty demanded authorial authenticity, and that meant erasing all traces of Eric Arthur Blair. He had to \"pass\" as a man living on the margins, and Eric Blair was not that man. Changing his name was also appealing because he claimed to detest his birth name. Perhaps it had to do with the strained relationship he'd had with his father.\n\nDown and Out in Paris and London by George Orwell was published in January 1933, with an initial print run of 1,500 copies. The author was relieved to have some validation of his efforts. \"Isn't it a grand feeling when you see your thoughts taking shape at last in a solid lump?\" he wrote to a friend.\n\nThere are a few reasons why Blair had settled on \"George Orwell\" as his literary persona. Some have speculated that the first name came from his admiration for the late-nineteenth-century writer George Gissing, who influenced his work. The surname seemed to have derived from the River Orwell in Suffolk, which Blair is said to have loved\u2014Defoe had written of it\u2014or from the village of Orwell in Bedfordshire, which Blair had once passed through.\n\nIn any case, it seemed perfect, and \"George Orwell\" became the most famous English pseudonym of the twentieth century. As well, thanks to his novel Nineteen Eighty-four, the adjective \"Orwellian\" became part of the lexicon. (It has a much better ring than, say, Milesian or Allwaysian, had he settled on his other choices.)\n\nAnthony Powell once asked his friend if he'd ever considered adopting \"George Orwell\" as his legal name. \"Well, I have,\" he told Powell, \"but then, of course, I'd have to write under another name if I did.\" Why he felt such a profound need to separate himself in private life from his \"writing self\" is a mystery. But duality is present throughout his work: in A Clergyman's Daughter, for instance, and elsewhere Orwell's characters lead double lives and harbor hidden selves. \"He was as secretive about his private life as any man I ever knew,\" a friend recalled of him.\n\nThe book was well received in England and, upon its international publication, by critics abroad. \"George Orwell is but trembling on the age of 30 this year, but he appears to have had about as much experience so far as the seamy side of life is concerned as a man of 50,\" wrote a reviewer in the New York Times in 1933, adding that Orwell's chilling account \"is apt to put an American with a ticklish stomach off filets mignon in the higher-priced hotel restaurants for ever. It is Mr. Orwell's argument bolstered by numerous horrible examples, that the more you pay for food in Paris, the less clean it is.\"\n\nWith the modest success of Down and Out, Blair's metamorphosis into George Orwell was complete. He'd received fan letters addressed to Orwell, and had, for the first time, even signed a book review as Orwell. The persona endured. His family, friends, publisher, and agent knew him as Eric Blair, but to the public he was firmly established as George Orwell. He'd accepted that neither Eric Arthur Blair nor even \"E. A. Blair\" had ever found success as a writer, and that only Orwell would be taken seriously. Eric Blair was a loser.\n\nHis books came in rapid succession: Burmese Days (1934), A Clergyman's Daughter (1935), Keep the Aspidistra Flying (1936), The Road to Wigan Pier (1937), Homage to Catalonia (1938), and in the last few years of his life, Animal Farm (1945) and Nineteen Eighty-four, published a year before his death. Even when he was highly productive, his usual reaction was to be dismissive of his output. \"I have never been able to get away from this neurotic feeling that I was wasting time,\" he wrote in his diary. \"As soon as a book is finished, I begin, actually from the next day, worrying because the next one is not begun, & am haunted with the fear that there never will be a next one.\" Nevertheless, despite having exasperated so many people with his polemics, he had by then endeared himself, more or less, to the literary establishment. \"He writes in a lucid conversational style which wakens one up suddenly like cold water dashed in the face,\" V. S. Pritchett wrote of Orwell's work.\n\nAlthough he instructed later that A Clergyman's Daughter and Keep the Aspidistra Flying not be republished once they had fallen out of print\u2014he dismissed them as \"silly potboilers\"\u2014they were crucial building blocks in what would prove a highly successful and even lucrative career.\n\nA friend once commented on Orwell's obsessive writing process. He walked in one day to find Orwell sitting at a table with books by W. Somerset Maugham and Jonathan Swift, reading passages from both, closing them, then copying out sections from memory. \"I'm trying to find a style which eliminates the adjective,\" Orwell explained. It was not unusual for Orwell to write for ten hours a day, to rewrite entire book drafts three times, or to revise individual passages five or ten times, until he was satisfied.\n\nHis fussiness also extended to his personal life. In an entry written in 1940 for an American directory of authors, he revealed, \"I dislike big towns, noise, motor cars, the radio, tinned food, central heating and 'modern' furniture.\" His list of approved things included English beer, French red wine, Indian tea, strong tobacco, vegetable gardening, and comfortable chairs. He added: \"My health is wretched, but it has never prevented me from doing anything that I wanted to. . . . I ought perhaps to mention that though this account that I have given of myself is true, George Orwell is not my real name.\"\n\nIn 1941, a critic (and former Eton classmate) named Christopher Hollis wrote a withering review of Orwell's book The Lion and the Unicorn, attacking the author as a coward: \"Many things interest me about Mr. Orwell,\" he wrote, \"and not the least among them the question why he prefers to confront the world with that peculiar name rather than with the very respectable one under which I have had the honour of knowing him for the last quarter of a century.\" This must have come as a shock to those acquaintances who knew Orwell only as Orwell. By that time, he was signing his work correspondence \"George Orwell,\" and sometimes signing personal letters \"E. A. B. (George Orwell).\" Most of his old friends still called him Eric. Despite the confusion, he refused to have his name legally changed. He may have taken some pleasure in being able to flit at will between one self and the other, as suited the occasion.\n\nThe 1930s had been a kind of golden age for the author, apart from his occasional hospitalizations and periods of convalescence. He established himself as a famous writer and he found love, or at any rate an acceptable version of it. After meeting an Oxford graduate named Eileen O'Shaughnessy at a party, he decided that she was \"the type of girl I'd like to marry.\" In 1936, they did, but like so many other things in his life, the marriage would prove ephemeral. In 1944, they adopted an infant son, whom they named Richard Horatio, but Eileen died a year later during an emergency operation. Because Orwell had been unfaithful to her, his grief was mingled with guilt. \"It wasn't an ideal marriage,\" he admitted to his housekeeper. \"I don't think I treated her very well.\" Her absence left him lonely and depressed, and with his recurrent bouts of flu and bronchitis, reminded him that he was probably running out of time himself.\n\nHe was eager to find another wife, and at the age of forty-six his wish was fulfilled. On October 14, 1949, the Associated Press issued a brief announcement: \"George Orwell, novelist, married yesterday Miss Sonia Brownell, an editor, in University College Hospital, where the author, who is suffering from tuberculosis, is confined.\" Orwell remarked on their travel plans. \"I don't know when I shall be allowed to get up,\" he said, \"but if I am able to move, we shall go abroad for the worst part of the winter, probably for January and February.\"\n\nHe was dead by the end of January.\n\nV. S. Pritchett paid tribute to Orwell, calling him \"sharp as a sniper\" and praising him as \"a writer of extraordinary honesty, if reckless in attack; to the day he died, nearly three weeks ago, he had never committed an act of political hypocrisy or casuistry.\" Six decades after his death, Orwell was named by fifty Penguin authors as the publisher's most popular author ever. Animal Farm and Nineteen Eighty-four are still required reading in schools. He influenced scores of writers, including Kingsley Amis, Norman Mailer, and Anthony Burgess. And he is considered one of the twentieth century's finest essayists. \"If you want to learn how to write non-fiction, Orwell is your man,\" wrote Jeremy Paxman in the Daily Telegraph in 2009. \"The impeccable style is one thing. But if I had to sum up what makes Orwell's essays so remarkable is that they always surprise you.\"\n\nAfter Orwell's death, his friends remembered him fondly while acknowledging that he was often difficult. One spoke of him, aptly, as having been \"easier to love than to like.\" Stephen Spender offered a more generous assessment: although he found Orwell disingenuous in earnestly aligning himself with the working class, Spender recognized his essential decency and the purity of his motives. \"Even his phoniness was perfectly acceptable,\" he recalled. \"Orwell had something about him like a character in a Charlie Chaplin movie, if not like Charlie Chaplin himself. He was a person who was always playing a role, but with great pathos and great sincerity.\"\nShe weighed seventy pounds when she died\n\nChapter 9\n\nIsak Dinesen & KAREN BLIXEN\n\nShe may not have been descended from Danish royalty, but her childhood was filled with the traditional privileges of an aristocratic upbringing. Karen Cristenze Dinesen was born on April 17, 1885, and over the course of her life would be known alternately as Tanne, Tanya, and Tania by her family and close friends. \"Tanne\" was a nickname that originated from her youthful mispronunciation of her own name (and was one she was said to dislike), but it stuck nonetheless. She grew up on her family's estate in Rungstedlund, on the Danish coast midway between Copenhagen and Elsinore. Her father bought the house, a former inn, in 1879, and Dinesen would spend her final years there in relative seclusion.\n\nKaren's great-grandfather on the side of her mother, Ingeborg, was a ship baron and one of the wealthiest men in Copenhagen. Her father, Wilhelm, came from a family of major landowners. After the Franco-Prussian War, he traveled to America, where he spent time among Indians who gave him the name \"Boganis,\" meaning hazelnut. He later published a book, Letters from the Hunt, using Boganis as a pseudonym, which was Karen's first encounter with a nom de plume. She was the second of five children, raised in a puritanical household, and easily her father's favorite; they had a close, confiding relationship that seemed to exist in its own private, obsessive realm, outside the rest of the family. But in 1895, a month before Karen's tenth birthday, Wilhelm hanged himself at the age of fifty. She never forgave him for abandoning her. She was left, as she would later say, with an abiding terror \"of putting one's life into, and abandoning one's soul to something that one might come to lose again.\"\n\nEven at a young age, Karen knew the depths of sadness. She struggled to find a secure place within her competitive family and used her rich fantasy life as a frequent means of escape. She wrote thoughtful, world-weary plays, essays, stories, and poems and kept a diary. Her mother was strict, forbidding the children to enter certain rooms of the house without permission and refusing to intervene in sibling squabbles. \"Whoever is angry must absent himself from the public spaces, and from the stairs and corridors, so long as the anger lasts,\" she decreed. (Crying aroused neither Ingeborg's sympathy nor any gestures of maternal comfort.)\n\nAt fourteen, Karen fell in love with Shakespeare, marveling at the epic scale of his romances and tragedies. She read widely: Stendahl, Chekhov, Voltaire, Conrad, Turgenev, Hans Christian Andersen, and poets such as Percy Bysshe Shelley. She was passionate about art and at eighteen was accepted at the Royal Academy in Copenhagen. The following year, in 1904, Karen began to write what she called \"Likely Stories,\" which revealed even then her predilection for the gothic and fantastical. Her favorite poet was Heinrich Heine, and she often recited these lines from his Buch der Lieder: \"You haughty heart, you wanted it like this! \/ You wanted to be happy, infinitely, \/ Or infinitely wretched, haughty heart, \/ And now you are wretched.\"\n\nAt the Royal Academy, she met someone who would become her first reader (apart from her family) and literary mentor: Mario Krohn, a young intellectual whose father was a museum curator. She spent a lot of time with Krohn, though her affection for him seems to have been largely platonic. (He died of tuberculosis in 1922.) With his encouragement, she sent her stories to the editor of Denmark's most prestigious literary journal, who responded to one piece by calling it \"too broad and a little too artistically contrived, and the whole tone too hearty and simpleminded. It is also too long.\" Yet he recognized her talent and decided to accept one of the stories, \"The Hermits.\" She would publish two more stories in the journal, all under the pseudonym \"Osceola.\" This she'd borrowed from an unlikely source: her father's German shepherd. It was a name that Wilhelm had borrowed from a leader of the Seminole Indians in Florida. Osceola had led his tribe's resistance when the American government tried to remove the Seminoles from their land, and he died in prison a few months after being captured.\n\nBefore giving birth to Isak Dinesen, Karen Dinesen would have to meet the man who would become her husband: her Swedish second cousin, Baron Bror von Blixen-Finecke, a distant cousin of King Christian of Denmark. (Bror had a twin brother, Hans, whom Karen was in love with, but he was not interested in her.) Bror, who pursued her assiduously until she relented, was handsome and gregarious. Karen did not find him compelling or even intelligent. Her family was not impressed, either. Bror was inept with money, a fact that would have disastrous consequences for the couple later on. Still, their early years together were fairly happy, and when Bror's uncle suggested, \"Go to Kenya, you two,\" they did.\n\nIn 1913, the Blixens set off for what was then British East Africa, setting up a 4,500-acre coffee plantation called the Swedo-African Coffee Company, twelve miles from Nairobi. (Eventually they would own 6,000 acres.) Bror was giddy at the financial potential of the business. Never mind that he knew nothing about growing coffee. Nor did he consider fluctuating coffee prices or realize that locusts, droughts, acidic soil, and the elevation of the land made it inhospitable for his ambitious endeavor, and destined it to fail. \"The land was in itself a little too high for coffee,\" Karen recalled with typical understatement in Out of Africa, \"and it was hard work to keep it going; we were never rich on the farm. But a coffee-plantation is a thing that gets hold of you and does not let go, and there is always something to do on it: you are generally just a little behind with your work. . . . Coffee-growing is a long job.\" But in Africa she had found a spiritual home, and she described the thought of ever leaving as \"Armageddon. After that\u2014nothing.\"\n\nIn addition to her own elegiac memoir\u2014with its famous opening line (\"I had a farm in Africa, at the foot of the Ngong Hills\") and notable omissions (not much mention of her husband, and only a platonic rendering of her lover, the hunter Denys Finch Hatton)\u2014many accounts of her years in Africa have been written elsewhere: the decline of the plantation; the pileup of financial debt; her contracting of syphilis (from which she never recovered) from her philandering husband in the first year they were married; the breakdown of her marriage; her relationship with Finch Hatton; and so on. Those years, by turns enchanting and filled with frightening adversity, had a profound impact. \"When I was a young girl,\" she recalled later, \"it was very far from my thoughts to go to Africa, nor did I dream then that an African farm should be the place in which I should be perfectly happy. That goes to prove that God has a greater and finer power of imagination than we have.\"\n\nEven after her marriage ended, she continued to manage the farm on her own. At the time, this was certainly an odd way for a woman of her class to live, but it was a testament to her attachment to the land and its people. \"Here at long last one was in a position not to give a damn for all conventions, here was a new kind of freedom which until then one had only found in dreams,\" she said.\n\nPartly because of the collapse of the coffee market, she was eventually forced to sell the plantation and return to Denmark. The decision broke her heart. \"I was driven out of my house by the fear of losing it,\" she wrote. \"When in the end, the day came on which I was going away, I learned the strange learning that things can happen which we ourselves cannot possibly imagine, either beforehand, or at the time when they are taking place, or afterwards when we look back on them.\"\n\nThe painful losses she endured\u2014of her farm, of her beloved horses and dogs, of Finch Hatton (who died in a plane crash in May 1931)\u2014drove her back to Denmark and to writing.\n\n\"I really began writing before I went to Africa,\" she told the Paris Review in an interview six years before her death. \"But I never once wanted to be a writer.\" (That single-minded devotion to process\u2014the ardor for writing itself, rather than the vanity of having written\u2014is, of course, the mark of a true writer.) She had done some writing in Africa as well; two of the stories in Gothic Tales, believed to be \"The Dreamers\" and \"The Old Chevalier,\" were written there. She wrote them while trying to distract herself from the distressing problems of the farm. \"One of my friends said about me that I think all sorrows can be borne if you put them into a story or tell a story about them, and perhaps this is not entirely untrue,\" she said years later. In Africa she refined her skills as a storyteller: \"I had the perfect audience,\" she told the Paris Review. \"White people can no longer listen to a tale recited. They fidget or become drowsy. But the natives have an ear still. I told stories constantly to them, all kinds. And all kinds of nonsense. I'd say, 'Once there was a man who had an elephant with two heads' . . . and at once they were eager to hear more. 'Oh? Yes, but Memsahib, how did he find it, and how did he manage to feed it?' or whatever. They loved such invention.\" In a 1957 interview with the New York Times, she insisted, \"I am not a novelist, really not even a writer; I am a storyteller.\" Some would dispute that assessment. When Hemingway was awarded the Nobel Prize in 1954, he said that he would have been \"happy\u2014happier\u2014today\" if it had gone instead to \"that beautiful writer, Isak Dinesen.\"\n\nWhen she returned to Denmark from Africa, in 1931, she was a lost soul. Even her identity was in tatters; she'd lost her title as baroness after Bror remarried in 1929 and found a new Baroness Blixen. Though Karen was often accused, perhaps unfairly, of being a vain snob, she did cherish her title and was angry and indignant when she was stripped of it. People continued to call her the Baroness in later years, which surely pleased her. But at that point, she didn't know what to call herself. And at forty-six years old, she was destitute, forced to move back to her mother's home. (In a 1986 essay, John Updike, an admirer of Dinesen's work, described her return at that time as \"ignominious,\" noting that she was received into the household as \"a prodigal daughter, a middle-aged adolescent.\") In a letter that year, Karen told her brother Thomas, \"I have wondered whether I could learn to cook in Paris for a year or two, and then perhaps get a post in a restaurant or a hotel.\" She also suggested that she could take care of \"mad people.\" Fortunately, she set her ambitions elsewhere, confiding to Thomas, \"I have begun to do what we brothers and sisters do when we don't know what else to resort to\u2014I have started to write a book.\" She decided to write in English because it had been her primary language in British East Africa, and she was comfortable with it; and because she believed that potentially, an English-language book would reach a larger audience and be more profitable. (She was right.)\n\nThe book she was writing became Seven Gothic Tales. She said later that she used the word tale after Shakespeare, or \"in the na\u00efve view of a child or primitive who sees a story as neither tragic nor comic but marvelous.\"\n\nOf course, Karen needed to support herself as she wrote, so she asked Thomas to finance her for two years, promising that by the end of this time she would become independent. Writing from her family's estate, Karen felt the presence of her father once again. After all, he had gone to America and lived with the Plains Indians, then returned to Denmark to write his books. \"So you see, it was natural for me, his daughter, to go off to Africa and live with the natives and after return home to write about it,\" she explained to the Paris Review.\n\nBiographical events intersected in other ways. Not long after the farm was sold to a Nairobi real estate developer, Karen had attempted suicide by slashing her wrists. And shortly before her father's suicide, Wilhelm was told by a doctor that he had a disease \"which could only conclude in a dark, helpless future.\" (It was most likely syphilis.) \"My father's destiny has, curiously enough, to a great extent, been repeated in my own,\" she said.\n\nThree years after the scaffolding of her life collapsed beneath her, Karen would become a published author in England and the United States. She would evolve into Isak Dinesen. The transformation was a struggle, however. (\"No one came into literature more bloody than I,\" she once said.) In those days\u2014as is true in today's publishing climate\u2014a short story collection, especially by an unknown author, was not a desirable commodity. Publishing is a profit-driven business, like any other, and story collections aren't known for being lucrative. Karen's manuscript was rejected by at least two publishers, including the London house of Faber and Faber, which was then a relatively new (founded in 1929) but prestigious firm.\n\nKaren had been preparing the material that would form Seven Gothic Tales, on and off, for a decade. She later described her process as beginning with a \"tingle, a kind of feeling of the story I will write. Then come the characters, and they take over, they make the story.\" She said that she began only with the \"flavor\" of a tale, and that her characters led her toward their fates\u2014\"I simply permit them their liberty,\" she explained. (In his 1976 Paris Review interview, John Cheever\u2014speaking not of Dinesen specifically, but of the notion that fictional characters take on identities of their own\u2014dismissed the romantic idea of the author as a passive creative vessel. \"The legend that characters run away from their authors\u2014taking up drugs, having sex operations, and becoming president\u2014implies that the writer is a fool with no knowledge or mastery of his craft,\" he said.)\n\nIn any case, Karen would devote herself fully to writing, well aware of the radical nature of her task. Two decades later, she admitted in a speech that if she had been a man, \"it would be out of the question for me to fall in love with a woman writer.\" Working away in her father's old office\u2014which was also the same room where, in the late eighteenth century, Denmark's greatest lyrical poet of the era, Johannes Ewald, was said to have written\u2014she sat at the Corona typewriter she'd brought home from Kenya, allowing few intrusions into her time and space. She was openly resentful of social interruptions, whether from family or friends; as a result, some visitors who came to the house were put off by her foul mood and deemed her behavior selfish (not an adjective one might have ascribed to a male writer). Her seclusion provided a kind of freedom, psychological if not physical: the permission for her imagination to roam at will, exploring and reaching beyond the bounds of self, mining the material of both her dreaming and her waking life.\n\nShe later explained that her decision to publish under a pseudonym was not unlike how her father \"hid behind the pseudonym Boganis. . . . [It was to] express himself freely, give his imagination a free rein. He didn't want people to ask, 'Do you really mean that?' Or, 'Have you, yourself, experienced that?'\" She decided to use her maiden name, Dinesen, and chose the first name Isak, meaning \"laughter\" in Hebrew. (In the Old Testament, Isaac was born to Sarah when she was quite old; his birth seemed almost like a miraculous prank by God.) The name reflected Karen's comic spirit and her love of humor, particularly irony. It was an element in her work, even in the \"tragic\" stories, that was never given its proper due by most critics. Also, as an author who \"gave birth\" to her first book at the age of forty-nine, she was a late bloomer herself; the name was apt.\n\nWhen she completed her manuscript in 1933, she took it first to London, where a family friend arranged a luncheon for her. Karen was introduced to an American-born publisher, Constant Huntington of the British firm Putnam's. She charmed him and asked if he would be willing to read her work. When she mentioned that it was a collection of short stories, he threw up his hands and refused even to look at it. \"A book of short stories by an unknown writer? No hope!\" he said. She returned home angry and despondent, almost ready to give up. But one of her mottoes in life was: \"Often in difficulties, never afraid.\"\n\nShe made use of a contact from her brother Thomas\u2014an American author, Dorothy Canfield Fisher, who lived in Vermont. She mailed her the manuscript, hoping something might come of it. Fisher immediately recognized its value and passed it along to her neighbor, the publisher Robert Haas, who accepted it, taking a big chance on an unknown European writer. (When Random House bought the firm owned by Haas and his partner, Harrison Smith, the company acquired not only Isak Dinesen but other prominent authors, including Jean de Brunhoff and William Faulkner.) Haas considered the deal a labor of love, and imposed two conditions before publication: that the book include a foreword by Fisher, a distinguished figure whose name might generate some good publicity; and that he not pay an advance until at least a few thousand copies had been sold. Dinesen agreed to his terms, but wrote to Fisher to express concern regarding authorship. \"I don't want the book to come out under my own name,\" she wrote, \"and at the same time I don't want people to know that it is myself who has written it, even though that is not a serious problem in America!\u2014I'm going to have to find a name to publish it under.\"\n\nWhen Seven Gothic Tales was published in January 1934, it was a critical and commercial success. (Dinesen's first check from Haas, for $8,000, arrived that Christmas.) Upon seeing that the book had met with such great acclaim, Constant Huntington wrote a letter to Haas, praising the book, pleading for the author's address, and\u2014oh, the audacity\u2014insisting that he (and Putnam's) publish the British edition. Dinesen was amused. \"He had met me as Baroness Blixen,\" she recalled later, \"while Mr. Haas and I had never seen one another. Huntington never connected me with Isak Dinesen.\" Putnam's released the work in England in September 1934.\n\nThe persona \"Isak Dinesen\" made the author a figure of mystery in the literary world. Rumors swirled about the true identity of this \"slender, pale, large-eyed, middle-aged Danish woman,\" as a critic would later describe her. They said she was really a man, or \"Isak\" was a woman, or argued that the author was a collaboration between a brother and sister. He or she was a recluse. A nun. Actually French, not Danish. And so on.\n\nIn \"The Dreamers\" (the penultimate story in Seven Gothic Tales), the character Pellegrina Leoni, an alter ego of sorts for Dinesen, implores another character to lose himself: \"Be many people,\" she says. \"Give up this game of being one. . . . You must, from now, be more than one, many people, as many as you can think of. I feel, Marcus\u2014I am sure\u2014that all people in the world ought to be, each of them, more than one, and they would all, yes, all of them, be more easy at heart.\" This reflected the author's desire to escape her own self. For her, the willful expanse of identity was a path that led away from suffering, from the daily sorrows that trapped her.\n\nA passage in the story \"The Old Chevalier\" seemed to express Dinesen's need, after so much loss, to overturn the circumstances of her life and start anew: \"Reality had met me, such a short time ago, in such an ugly shape, that I had no wish to come into contact with it again,\" she wrote. \"Somewhere in me a dark fear was still crouching, and I took refuge within the fantastic like a distressed child in his book of fairy tales. I did not want to look ahead, and not at all to look back.\"\n\nThemes of truth and deceit are everywhere in Dinesen's fiction. In \"The Deluge at Noderney,\" the opening story of Gothic Tales, a cardinal explains the virtues (and power) of masquerade: \"The witty woman, Madame, chooses for her carnival costume one which ingeniously reveals something in her spirit or heart which the conventions of her everyday life conceal; and when she puts on the hideous long-nosed Venetian mask, she tells us, not only that she has a classic nose behind it, but that she has much more, and may well be adored for things other than her mere beauty. So speaketh the Arbiter of the masquerade: 'By thy mask I shall know thee.'\"\n\nIsak Dinesen's lauded debut was a Book-of-the-Month Club selection, with a print run of fifty thousand copies\u2014an astonishing number at the time. The BOMC newsletter ran an announcement, along with a simple notice: \"No clue is available as to the pseudonymic author.\" On March 3, 1934, the New York Times posted the selection in its \"Book Notes\" column: \"Seven Gothic Tales, by a European writer who uses the pen name of Isak Dinesen, is to be the Book-of-the-Month Club choice for April. Smith & Haas will publish it, with an enthusiastic introduction by Dorothy Canfield.\" Five weeks later, John Chamberlain, a columnist for the newspaper's \"Books of the Times,\" wrote that he was unimpressed by the selection: \"[W]e found it impossible to get interested in Isak Dinesen's Seven Gothic Tales. . . . We are willing to grant the eerie light in the book, and the slanting beauty of phrase, but the predicaments of the characters leave us cold. If you prick Mr. Dinesen's people, they do not bleed.\"\n\nRegardless, it was a hit, and Dorothy Canfield Fisher's introduction to Seven Gothic Tales encouraged a sense of intrigue about the author's identity. She proved a great advocate for the book, writing, \"I am so much under its spell (it feels exactly like a spell),\" and also letting the reader know that the material did not fit easily into any familiar genre or literary movement. \"The person who has set his teeth into a kind of fruit new to him,\" Fisher wrote, \"is usually as eager as he is unable to tell you how it tastes. It is not enough for him to be munching away on it with relish. No, he must twist his tongue trying to get its strange new flavor into words, which never yet had any power to capture colors or tastes.\"\n\nDevour the book, she urged, but claimed she could offer no insight into who had written it: \"I can't even tell you the first fact about it which everybody wants to know about a book\u2014who is the author.\" Fisher continued, cryptically: \"In this case, all that we are told is that the author is a Continental European, writing in English although that is not native to his pen, who wishes his-or-her identity not to be known, although between us be it said, it is safe from the setting of the tales to guess that he is not a Sicilian.\"\n\nBut Isak Dinesen was perhaps the shortest-lived pseudonym in literary history. The book had created such a stir that the Danish press immediately set out to learn the author's real identity, and, following a tip that \"he\" was in fact a \"Danish lady,\" reporters from the newspaper Politiken found her. At the end of April, Smith and Haas announced formally that Isak Dinesen was Baroness Blixen of Rungstedlund. A week later, the competition began among Danish publishers to acquire translation rights to her book. She decided to undertake the job of translation herself, a practice she would follow from then on\u2014writing most of her stories first in English, then in Danish. But these were never direct translations; she would rewrite as well, even changing the endings to create original stories for a different audience.\n\nSeven Gothic Tales (or Syv Fantastiske Fortaellinger) was published in Denmark in September 1935, when Dinesen was fifty (the same age at which her father committed suicide). The critical reception was decidedly harsh. Her work was dismissed as too artificial, too perverse, too shallow, too elitist, and too foreign. One young reviewer criticized the book on many counts, noting that \"[t]he erotic life which unfolds in the tales is of the most peculiar kind.\" In the end, he wrote, \"There is nothing . . . behind [the author's] veil, once it is lifted.\"\n\nSome critics were annoyed by Dinesen's decision to write first in English\u2014an apparent breach of etiquette\u2014and by the fact that her breakthrough had occurred in the United States rather than her homeland. To avoid offending them again, subsequent books were issued simultaneously in Danish and English\u2014or first in Danish. Also, she reserved her pseudonym only for books that came out in North America; in Denmark she reverted to Karen Blixen\u2014perhaps in an attempt to prove her \"authenticity\" and appeal to national pride. Still, she never felt that she achieved enough popularity in Denmark, certainly not compared with the adulation she received abroad. In the United States, she had an impressive roster of admirers. Truman Capote yearned for a movie adaptation of \"The Dreamers,\" with Greta Garbo in the lead role. Ralph Ellison, Pearl Buck, and Marianne Moore loved her work. Orson Welles said that he considered Dinesen superior to Shakespeare. William Maxwell praised Dinesen as \"the most original, the most perceptive, and perhaps the best living prose writer.\" Eudora Welty called her \"a great lady, an inspired teller of her own tales, a traveler, possessed of a learned and seraphic mind.\" Carson McCullers was also a fan. \"When I was ill or out of sorts with the world,\" she said, \"I would turn to Out of Africa, which never failed to comfort and support me.\" In 1957, Dinesen was made an honorary member of the American Academy of Arts and Letters; other member inductees that year included John Dos Passos, Flannery O'Connor, Mary McCarthy, and W. S. Merwin. Meanwhile, at home, Dinesen confided to a friend, \"Lately, I have had the feeling in Denmark of being under suspicion, almost as if I were on parole.\"\n\nWhen Out of Africa was published in 1937, it, too, was a Book-of-the-Month Club selection. With its lovely, straightforward prose, not the least bit baroque or decadent, and with no questionable subject matter, the memoir elicited a positive critical response in her homeland. Grounded in the story of a land and its people, it was \"realistic\" rather than fantastical. In Denmark she called the book Den afrikanske Farm (The African Farm); for the American edition, she'd chosen the title Ex Africa, but Robert Haas persuaded her to use Out of Africa instead. Dinesen insisted that it be published on the same day in the United States, Scandinavia, and England, rather than releasing first to Danish readers and then elsewhere\u2014a request her publisher resisted because of the logistics. \"America took me in when I could not even make the publishers in Europe have a look at my book,\" she explained, \"and the American reading public received me with such generosity and open-mindedness as I shall never forget. I was delighted with the reviews of the American critics. I feel the deepest gratitude toward you all.\" She worried (however irrationally) that delaying American publication might convey the impression that she had lost interest in her fans there or no longer valued them. Despite the case she'd made, her request was denied, thus preserving a schism in her literary identity that could not be made whole: living as one persona abroad and another at home.\n\nOut of Africa was praised by Time magazine as \"a restrained, formalized book, which has little in common with her first book.\" She captured the African landscape, its people and animals \"with the eye of a painter and a novelist.\" The New York Times called the book \"rare and lovely,\" and praised its \"penetration, restraint, simplicity and precision which, together, mark the highly civilized mind, and that compassion, courage and dignity which mark civilization, in the best sense, in the human heart.\"\n\nIt must have annoyed Dinesen that a book by her former husband came out at the same time\u2014also a memoir of Africa, published by Knopf. Time magazine was scathing in its review: \"By comparison with his former wife's volume, 50-year-old Baron von Blixen-Finecke's African Hunter is little more than a handbook for big-game hunters. . . . Baron Blixen-Finecke does not care much for natives. Now married to an adventurous, pretty, 29-year-old Englishwoman, he remembers his first wife (Isak Dinesen) for one incident, when she flew unarmed at two lions that had attacked an ox, lashed them into the jungle with a stock whip.\" (Bror would marry a third time and die in a car crash in Sweden in 1946.)\n\nOn May 10, 1943, Dinesen's third book, Winter's Tales, was published in the United States. (It had come out in Denmark a year earlier.) This, too, was sold to the Book-of-the-Month Club and was a huge success. Despite having been unmasked seven years earlier, Dinesen still had a seductive aura of intrigue, one that cast her as imperious and remote. William Maxwell noted that although \"Isak Dinesen\" was \"now generally known to be the pen-name of a Danish woman . . . the Baroness herself is still something of a mystery. The facts concerning Baroness Blixen supplied by her publishers are definite enough; there just aren't many of them.\" And when the New York Times columnist Orville Prescott reported the publication of Winter's Tales, his piece, with its dramatic opening, read more as if he were writing about a witch than an author: \"In Denmark lives a baroness, a strange and grandly gifted woman who by some odd chance has strayed into the twentieth century from distant regions beyond time and space. . . . A serene and frosty genius, she is an artist of pr\u00e9cieux and impeccable talent who scorns the conventional, the direct and the clearly understandable. A writer, she forsook her native Danish tongue and has written her books in an English of such coldly glittering beauty she has hardly a living rival as a literary stylist. Her books are signed Isak Dinesen.\" Prescott proclaimed the arrival of another book from this enchantress as \"rather like a nightingale singing in a boiler factory, like a phoenix materializing in Union Square on May Day.\" He may as well have been referring to the author herself when he said that Winter's Tales was \"aloof and separate from every world that ever was.\"\n\nWinter's Tales\u2014which was Dinesen's own favorite of her books\u2014had a rather unlikely path to publication. This collection came out of Denmark in the midst of World War II, by secret diplomatic mails, to America. First Dinesen had traveled with the manuscript to Stockholm, where she visited the American embassy with an odd request: would someone there be willing to carry the manuscript on one of the planes headed for the United States? She was told that only political or other official papers could be transported. Then she went to the British embassy to make the same request. After she provided a few references in high places (including Winston Churchill), the favor was granted and the manuscript was sent to America on her behalf. Along with her stories, Dinesen had enclosed a note to her publisher, indicating that she was unable to communicate further: \"I can sign no contract and read no proofs,\" she wrote. \"I leave the fate of my book in your hands.\"\n\nShe would have no idea how things turned out until the war ended. \"I suddenly received dozens of charming letters from American soldiers and sailors all over the world,\" she said later. \"The book had been put into Armed Forces Editions\u2014little paper books to fit a soldier's pocket. I was very touched. They gave me two copies of it; I gave one to the King of Denmark and he was pleased to see that, after all, some voice had spoken from his silent country during that dark time.\"\n\nThe book was critically well received, though without making the same splash as Seven Gothic Tales. \"Many people, I feel sure, will read all eleven Winter's Tales as I did\u2014as fast as possible in order to have as soon as possible the pleasure of reading them for the second, the third, and, inevitably, the fourth time,\" William Maxwell wrote. Still, Dinesen was feeling bored, restless, and frustrated\u2014partly because of the monotony of daily life brought on by the war\u2014and suffered through periods of poor health, due to the syphilis she'd contracted years earlier. She was convinced that she would never produce a novel, but held out hope that she might.\n\nA Frenchman named Pierre Andr\u00e9zel would do it for her.\n\nHere was yet another persona for Dinesen, at the age of fifty-nine, during the German occupation of Denmark. She had created Andr\u00e9zel out of boredom, because she felt caged in as herself and wished to toy with a new disguise. The novel, The Angelic Avengers, was (as its title suggests) a thriller. Years later, Dinesen would laugh it off as \"my illegitimate child.\" She had done it, she insisted, simply to amuse herself. She asked her Danish publisher in Copenhagen for an advance, and for a stenographer to whom she could dictate the novel. Unsure of the story before she began, she wrote by improvising, dictating a little each day. \"It was very baffling to the poor stenographer,\" she said. It was also problematic: she would begin a session by announcing that a certain character would enter a room, only to be reminded by the stenographer, \"Oh dear, he can't! He died yesterday in Chapter Seventeen.\"\n\nWhen the book came out, Dinesen denied that she had anything to do with it (or with Andr\u00e9zel), despite a surge of rumors fueled by her own publisher. She said that even if she were the author, she would never admit it. When a friend wrote to say that he'd read the novel and found it \"a profound joke,\" she replied that she knew who the author was, but refused to reveal his identity until others discovered it for themselves (just as she knew, inevitably, they would).\n\nThe novel, which some readers interpreted as an allegory of the fall of Nazism, was published as Ways of Retribution in Copenhagen, in 1944. Although Dinesen refused to claim authorship, it wasn't long before she was unmasked, again by the pesky press. Dinesen was upset that journalists would not respect her desire to go incognito, a privilege lost to her long ago. The book became a best seller in Denmark (it was reviled by critics) and was published in the United States a few years later. The Book-of-the-Month Club chose it as half of the dual selection for January 1946 (along with Mr. Blandings Builds His Dream House, by Eric Hodgins). In its announcement, the BOMC remarked that The Angelic Avengers\u2014\"a fascinating story of mystery, adventure, and pure young love\"\u2014was written in wartime, and that \"Pierre Andr\u00e9zel\" was surely a pseudonym, but: \"Of whom? There were shrewd guesses, but nobody ever really knew. The author, whoever it is, continues to guard this anonymity. All that has been divulged by him (or her) is a plainly fictitious autobiographical note sent to the American publisher.\"\n\nDespite the author's contention that her latest novel was a bit of an embarrassment, something she had written to have \"a little fun,\" The Angelic Avengers marked another grand success, selling ninety thousand copies in America. One reviewer wrote that Dinesen was dealing with \"somewhat coarser material than in the best of her tales, but dealing with it in such a way that this novel will certainly widen the circle of her readers.\"\n\nAfter the publication of Winter's Tales and The Angelic Avengers, Dinesen didn't publish again for more than a decade. In the final years of her life came Last Tales (1958), which she dictated to her assistant and said was written \"with a leg and a half in the grave\"; two years later came Shadows on the Grass and Anecdotes of Destiny. Her health had steadily worsened, owing to the syphilis. There were periods in which she would rally, but once her decline had begun she was never quite the same. A frail, gaunt figure, weighing less than eighty pounds, Karen was in and out of the hospital. In the morning she took amphetamines, stimulants that caused her to talk compulsively, in an odd, almost trancelike state. At night she swallowed barbiturates to fall asleep. Because of the wasting away of her spine, she was sometimes unable to stand or walk. In those last years she led a fairly isolated life, and in periods of illness she was especially ill tempered, sarcastic, depressed, and paranoid. Her moods, she admitted, were \"coal black.\"\n\nDinesen was well aware that she could be as difficult as she was charming. \"As long as I live it will be bothersome for you to have to deal with me,\" she once told a dear friend. She was a leading contender for the Nobel Prize until her death but never won, a fact that proved an ongoing disappointment. Yet by the time she died, in 1962, she was an international celebrity and her books had been published in twelve languages. When Sydney Pollack's Academy Award\u2013winning film adaptation of Out of Africa was released in 1985, a new audience was drawn to Dinesen's work, and there was a resurgence of interest in the author as well. To the end, whether inhabiting Tanne, Karen, Isak, or any of her other selves, she believed absolutely that it was her right to assume a pseudonym, and that readers were obliged to respect it. Although her aliases had been promptly uncovered, a friend once wrote of his unknowable, inscrutable friend that \"Karen Blixen as a person was always pseudonymous in varying degrees, [and] that she always wanted to be suspected behind her texts but under no circumstances caught.\"\n\nIn her final months, she grew weaker still, her weight down to seventy pounds. She subsisted on glasses of vegetable and fruit juice, oysters, and biscuits\u2014the few things she could keep down. She could no longer stand without losing her balance, and admitted in a letter to a friend that a doctor had said \"that I have all the symptoms of a concentration camp prisoner, one of them being that my legs swell so that they look like thick poles and feel like cannon balls. This last thing is terribly unbecoming and for some reason very vulgar. Altogether I look like the most horrid old witch, a real Memento Mori.\" On September 7, 1962, she spent the evening listening to Brahms. That night she fell into a coma and died in her narrow wooden bed. She was buried on the family property under a beech tree.\n\nFive years before her death, an interviewer asked Dinesen whether she had led a happy life. \"Yes, and with all my heart,\" she replied. \"At times I have been so happy that it has struck me as overwhelming, almost as supernatural.\" She was asked what, exactly, had made her so happy. \"In a way I believe that the only true, sure happiness one can talk about here is the pure joy of living, a sort of triumph simply because one exists.\"\nShe found sexual satisfaction in picking her nose\n\nChapter 10\n\nSylvia Plath & VICTORIA LUCAS\n\nShe was a good girl who loved her mother. That, at least, was the benign impression Sylvia Plath gave the outside world\u2014a smiling fa\u00e7ade of conformity; feminine, pure of heart; accommodating, polite, bright-eyed, and pretty. She admired her mother, Aurelia, and was desperate for her approval. There were no secrets between them. Aurelia was nurturing and boundlessly devoted; Sylvia was her dutiful, adoring daughter. Such was the seamless porcelain exterior of their relationship, and both players were invested in protecting it. Meanwhile, writing in her journals, Plath recorded the brutal truth. \"I lay in my bed when I thought my mind was going blank forever and thought what a luxury it would be to kill her, to strangle her skinny veined throat which could never be big enough to protect me from the world,\" she wrote on December 12, 1958, following a session with her therapist. \"But I was too nice for murder. I tried to murder myself: to keep from being an embarrassment to the ones I loved and from living myself in a mindless hell.\" She resigned herself to the ineluctable role she'd been cast in: \"I could pass her on the street and not say a word, she depresses me so. But she is my mother.\" Sylvia was adept at dealing with Aurelia. Before speaking to her, it was as if Sylvia had trained herself to neatly tuck in her fury and put it to bed, permitting it to stir again only in her mother's absence.\n\nPlath's biography is familiar to just about every English literature major, reader of contemporary poetry, and suicidal teenager. She was toxic because she was so seductive, and seductive because she was so toxic. Her fame is immeasurable. Even many nonliterary types know that Sylvia Plath was the mercurial poet who gassed herself in an oven.\n\nShe was born at 2:10 p.m. on October 27, 1932, in a suburb of Boston, Massachusetts. Her father, Otto Plath, was a biology professor at Boston University, a well-regarded etymologist, and twenty-one years older than his wife. \"At the end of my first year of marriage,\" Aurelia later wrote, \"I realized that if I wanted a peaceful home\u2014and I did\u2014I would simply have to become more submissive, although it was not my nature to be so.\"\n\nBy the time she was three years old, Sylvia proved quite brilliant. Once, while her mother was baking in the kitchen, she played alone on the living room floor. She was unusually quiet. Otto went to check on her, and, as Aurelia recalled, both parents were stunned to see what their daughter had done. Using a set of mosaic tiles she'd received as a gift, she reproduced \"unmistakably the simplified outline of the Taj Mahal, the picture of which was woven into a mat in our bathroom.\"\n\nWhen Sylvia was eight years old, her father died of an embolism brought on by complications of diabetes. We know how well she came to terms with that loss; those who don't should read her notorious poem \"Daddy,\" which says it all.\n\nShe had a younger brother, Warren, born in 1935, with whom she felt competitive for her parents' affection, especially her mother's. Sylvia was always driven to be the best, and often was. The siblings' relationship did not become markedly closer until she attended Smith College (on a scholarship) and Warren was at Phillips Exeter Academy and then Harvard. Years later, their mother described the family with her typical fondness for nostalgia (steeped in denial). This false portrait presented a family close and uncomplicated in its affections: \"We three loved walking by the sea, in the woods, huddling close by the fire and talking, talking, talking\u2014or sharing a companionable silence,\" she said.\n\nPlath always knew that she stood apart from others. Because she was viewed as \"dangerously brainy,\" she felt it was in her interest to mask her sharp intellect and turbulent emotions. Not only did she embody the role of a perfect, straight-A student, but she was determined to become popular. She also pursued the approval of adults, both at school and at home. Other students might merely work hard, but she burned with determination. Before her first short story appeared in Seventeen (in the August 1950 issue), Plath had submitted forty-five pieces to the magazine. At eighteen, she berated herself in her journal: \"What is my life for and what am I going to do with it? I don't know and I'm afraid. I can never read all the books I want; I can never be all the people I want and live all the lives I want.\"\n\nOne of her early poems, written when she was in tenth grade, was called \"I Thought That I Could Not Be Hurt.\" A teacher who read it expressed amazement that \"one so young could have experienced anything so devastating.\" In this instance, the source of suffering was her unwitting grandmother, who had accidentally smudged one of Plath's pastel drawings. The final stanza read,\n\n(How frail the human heart must be\u2014\n\na mirrored pool of thought. So deep\n\nand tremulous an instrument\n\nof glass that it can either sing,\n\nor weep.)\n\nSuch intensity of feeling would never leave her, despite her efforts to conceal and tame it. Aurelia added to this unbearable pressure by making Sylvia feel responsible for the well-being of both mother and daughter. Yet Aurelia might also be credited for Sylvia's supreme sense of confidence, her innate belief that she was \"special\" and destined for greatness. \"The worst enemy to creativity is self-doubt,\" Plath once wrote, but when it did creep in, she pounded it like a Whac-A-Mole until her achieving self could surface once again. Then all was right with the world. And she was at least able to find consolation in what she once described as the \"minute joys\" in life: she admitted in her journals that she loved the \"illicit sensuous delight\" she felt when picking her nose. \"God what a sexual satisfaction!\" she wrote.\n\nEarly on, Plath was a baffling mix: highly empathetic but also self-obsessed. She absorbed everything and everyone around her. By the age of seventeen, she was investigating the bounds of the self and how to manage her troubled psyche. \"Sometimes I try to put myself in another's place, and I am frightened when I find I am succeeding,\" she wrote in her diary in 1949. \"How awful to be anyone but I. I have a terrible egotism. I love my flesh, my face, my limbs with an overwhelming devotion. . . . I want, I think, to be omniscient. . . . I think I would like to call myself 'The girl who wanted to be God.' Yet if I were not in this body, where would I be? . . . But, oh, I cry out against it. I am I\u2014I am powerful, but to what extent? I am I.\"\n\nThe struggle between selves would torment her for her entire life\u2014in poems such as \"An Appearance,\" \"Tulips,\" and \"In Plaster,\" among others\u2014and it served as a frequent subject of her journals. At Smith, she wrote a long paper on the theme of double personality in Dostoevsky's novels. Even when she was relatively happy, or at least emotionally stable, her inner turmoil never abated. It must have been exhausting. Often her fixation on duality and falseness reached a crisis pitch. \"Look at that ugly dead mask here and do not forget it,\" she wrote in a lacerating note to herself in a 1953 diary, referring to a recent photograph. \"It is a chalk mask with dead dry poison behind it, like the death angel. It is what I was this fall, and what I never want to be again.\" That year, she attempted to kill herself by overdosing on sleeping pills.\n\nDespite her recurring depressions, treatments with electroshock therapy, and flirtations with suicide, she was not entirely obsessed with death. As much as she was preoccupied with it, she was also seeking to end her ego self, with its oppressive, needy demands that were impossible to fulfill. Perhaps it wasn't her whole life she wanted to stop, but a \"shameful\" part of herself. Over and over she expressed frustration at not measuring up to other poets and for feeling stalled in her work. \"I, sitting here as if brainless wanting both a baby and a career,\" she wrote in her journal in 1959. \"What inner decision, what inner murder or prison-break must I commit if I want to speak from my true deep voice in writing . . . and not feel this jam up of feeling behind a glass-dam fancy-fa\u00e7ade of numb dumb wordage.\"\n\nWhat she seems to have craved most, in fact, was a chance at rebirth, at resurrection. Even though she was sometimes able to produce (or recover) what she deemed an \"authentic\" self, the success did not prove sustainable. Plath's obsession with split selves\u2014the pretty, superficial good girl who does everything easily and well, versus the raging, violent demon lurking within\u2014left her perpetually confused: Which one was real? Which one should be shed? Which one should she kill off? In the end, the demon won.\n\nIn 1961 Plath won the Eugene F. Saxton Memorial Fellowship, a writing grant of $8,000. She had by then graduated summa cum laude from Smith, published poems and stories, and won a Fulbright scholarship to Cambridge University. There she met the dashing British poet Ted Hughes, whom she married on June 16, 1956. (That date is Bloomsday.) After winning the Saxton, she was especially excited because she had applied previously, for poetry, but had been rejected. This time, she'd gone for it with a different project in mind. Although she had in fact completed her first novel, The Bell Jar, and even signed a contract for the manuscript with the British publisher Heinemann, this award would give her time to make revisions before the book's publication and provide monthly living expenses as well. Money was extremely tight. That fall, Plath wrote one of her usual cheery letters home to her mother, assuring her that all was well. (Many of Plath's missives to Aurelia opened with the effusive \"Dearest-Mother-whom-I-love-better-than-anybody.\")\n\nShe mentioned that the New Yorker had just accepted her poem \"Blackberrying\" and shared the news about the Saxton. \"Well, I applied for a grant for prose this time and got the amount I asked for,\" she wrote. \"They pay in quarterly installments as parts of a project are completed, so I should get my first lot in a week or two!\" She continued: \"Life in town has been more and more fun.\" The letter began and ended with her standard loving greeting (\"Dear Mother\") and sign-off (\"x x x Sivvy\").\n\nLess than two weeks later, she sent Aurelia another chatty letter, referring again to the grant but neglecting to explain what, exactly, her writing project was about. \"I finished a batch of stuff this last year, tied it up in four parcels and have it ready to report on bit by bit as required,\" she reported vaguely. \"Thus I don't need to write a word if I don't feel like it. Of course, the grant is supposed to help you do writing and is not for writing you've done, but I will do what I can and feel like doing, while my conscience is perfectly free in knowing my assignments are done.\"\n\nWhat \"Sivvy\" failed to mention was that the \"batch of stuff\" was an autobiographical novel that would have killed her mother, or at least broken her heart. The narrator's voice, as in Plath's poetry, was icy and lucid. It was about the \"crackup\" of a well-behaved young woman named Esther Greenwood, described in the flap copy of the 1971 Harper & Row hardcover edition as \"brilliant, beautiful, enormously talented, successful\u2014but slowly going under, and maybe for the last time.\" The story was also, to put it mildly, an exploration of Esther's strained relationship with her mother, and how her repressed anguish leads to madness. There was only one way this devastating novel could be published by a \"good girl\" such as Plath, and that was to hide behind a pseudonym. She chose \"Victoria Lucas\": \"Victoria\" was a favorite cousin of Ted Hughes; \"Lucas\" was the name of Hughes's good friend Lucas Myers. Heinemann published The Bell Jar in London on January 14, 1963. Twenty-eight days later she killed herself.\n\nPlath lived long enough to read the reviews of her novel, and they didn't please her. The reception in Britain was tepid and condescending. \"There are criticisms of America that the neurotic can make as well as anyone, perhaps better, and Miss Lucas makes them brilliantly,\" Laurence Lerner wrote in the Listener. A critic in the Times Literary Supplement wrote that \"if [Lucas] can learn to shape as well as she imagines, she may write an extremely good book.\" Worse, Plath had hoped for publication in the United States, too, but that didn't seem forthcoming. Just after Christmas, she'd received a jarring letter of rejection from Alfred A. Knopf in New York, which had published her poetry book The Colossus the year before. A second rejection came from Harper & Row (\"The experience remains a private one,\" the editor wrote of the narrative, which seemed more a \"case history\" than a novel.) In the letter from Knopf, the editor expressed her regret: \"We didn't feel that you had managed to use your materials successfully in a novelistic way. . . . Up to the point of her breakdown the attitude of your young girl had seemed a perfectly normal combination of brashness and disgust with the world, but I was not at all prepared as a reader to accept the extent of her illness.\" The same could be said of Plath. No one\u2014not even those closest to her, who were well acquainted with her despair\u2014could fully comprehend its sheer velocity, its manic and unstoppable force.\n\nThe 1989 Plath biography Bitter Fame, by the poet and critic Anne Stevenson, opens with an apt epigraph from Dostoevsky's The Devils:\n\nThere was a tremendous power in the burning look of her dark eyes; she came \"conquering and to conquer.\" She seemed proud and occasionally even arrogant; I don't know if she ever succeeded in being kind, but I do know that she badly wanted to and that she went through agonies to force herself to be a little kind. There were, of course, many fine impulses and a most commendable initiative in her nature; but everything in her seemed to be perpetually seeking its equilibrium and not finding it; everything was in chaos, in a state of agitation and restlessness. Perhaps the demands she made upon herself were too severe and she was unable to find in herself the necessary strength to satisfy them.\n\nPlath, volatile to say the least, was once described by the poet W. S. Merwin as \"a cat suspended over water.\" Like many others who knew her, he found her a \"determined, insistent, obsessive person who snapped if things did not go her way, and flew into sudden rages.\"\n\nThe manuscript of The Bell Jar is another interesting manifestation of Plath's fragmented selves. Here she had produced the most shocking work of her young life, filled with harrowing insights into her own psyche\u2014yet she typed these words on dainty pink Smith College memo paper. \"Got a queer and most overpowering urge today to write, or typewrite, my whole novel on the pink, stiff, lovely-textured Smith memorandum pads of 100 sheets each,\" Plath wrote in her journal on March 3, 1958, while she was back at Smith, working as an instructor in the English department. She proudly noted that she'd helped herself to plenty of school stationery: \"Bought a rose bulb for the bedroom light today & have already robbed enough notebooks from the supply closet for one & 1\/2 drafts of a 350 page novel.\"\n\nWell before the novel came out, the phrase \"bell jar\" had popped up in Plath's writings. At Smith, she described feeling overwhelmed by her own mind, by the demands made on her, socially and otherwise\u2014and admitted that she found things especially hard without a prescribed routine to follow. She could never give herself a break:\n\nWorking, living, dancing, dreaming, talking, kissing\u2014singing, laughing, learning. The responsibility, the awful responsibility . . . is rather overwhelming when there is nothing, no one, to insert an exact routine into the large unfenced acres of time\u2014which is so easy to let drift by in soporific idling and luxurious relaxing. It is like lifting a bell jar off a securely clockworklike functioning community, and seeing all the little busy people stop, gasp, blow up, and float in the inrush (or rather outrush) of the rarefied scheduled atmosphere\u2014poor little frightened people, flailing impotent arms in the aimless air. . . . What to do? Where to turn?\n\nElsewhere, she wrote that \"it's quite amazing how I've gone around for most of my life as in the rarefied atmosphere under a bell jar.\" And a 1959 journal entry recorded feelings of frustration and gloom: \"The day is an accusation. Pure and clear and ready to be the day of creation, snow white on all the roof tops and the sun on it and the sky a high clear blue bell jar.\"\n\nThe few years leading up to the publication of The Bell Jar had brimmed with creativity. That period provided an argosy of material, but it may have ultimately contributed to her death. In the spring of 1959, she was writing the searing poems of The Colossus. Those took a toll, and the book's themes would inform her novel as well. She also had an appendectomy, and in 1960 she gave birth to a daughter, Frieda. After suffering a miscarriage, she became pregnant again, and in January 1962 gave birth to her son, Nicholas. (He would have his own lifelong battle with depression; he died in 2009 by hanging himself.) By the fall, her marriage had fallen apart: Hughes left her for another woman. This time, recovery was not possible.\n\nThe following January, as an overwhelmed, exhausted, and isolated young mother, Plath numbly witnessed her novel's debut. She was living in a dreary London flat (where W. B. Yeats had once lived) at 23 Fitzroy Road in St. Pancras. There was no telephone and electricity was intermittent. That winter in England, following a \"bone cold\" autumn, was bleak, snowy, and icy, one of the worst on record. Plath and her children had the flu, and she was terribly anxious about money.\n\nAt night, she could not fall asleep without medication. She was waking at four o'clock each morning to crank out the poems of Ariel. (Several drafts had already been handwritten on the reverse of the \"lovely-textured\" pink Smith College stationery on which she had typed her Bell Jar manuscript.) On February 11, 1963, as her children lay sleeping, she sealed off the door to their bedroom with wet towels and opened their window wide. She left them milk and bread. Then she put her head inside that infamous gas oven and ended it all.\n\nAs she'd immersed herself in the early stages of her novel, Plath had been understandably secretive with her mother about writing The Bell Jar, but she had openly shared her fiction-writing ambition with a friend: \"I have been wanting to do this for ten years but had a terrible block about Writing a Novel. Then suddenly . . . the dykes broke and I stayed awake all night seized by fearsome excitement, saw how it should be done, started the next day & go every morning to my borrowed study as to an office & belt out more of it.\" This was a real breakthrough, considering Plath constantly berated herself for not having accomplished enough. \"Prose writing has become a phobia to me: my mind shuts & I clench,\" she wrote in her diary in 1957. \"I can't, or won't, come clear with a plot.\" Her self-flagellation is present throughout her journals. \"Why can't I throw myself into writing?\" she wrote. \"Because I am afraid of failure before I begin.\"\n\nIn an entry dated December 12, 1958, Plath wondered, \"Why don't I write a novel?\" Following that question, she'd gone back a mere three years later and giddily amended the entry: \"I have! August 22, 1961: THE BELL JAR.\"\n\nTo describe the writing of it as cathartic is an understatement. Plath called The Bell Jar \"an autobiographical apprentice work which I had to write in order to free myself from the past.\" She had found a safe alter ego in Esther Greenwood\u2014rendered even more secure by the mask of Victoria Lucas\u2014through which the author could exorcise, among other things, her electroshock therapy, mental breakdowns, repressed sexual desires, and hatred of her mother. In one scene, Esther expresses revulsion at watching her mother awaken: \"My mother turned from a foggy log into a slumbering, middle-aged woman, her mouth slightly open and a snore raveling from her throat. The piggish noise irritated me, and for a while it seemed to me that the only way to stop it would be to take the column of skin and sinew from which it rose and twist it to silence between my hands.\" With such cruelly drawn characters, Plath could malign anyone who'd ever caused her pain or failed to give her what she craved\u2014and her mother above all would be punished.\n\nYears after Plath's death, Aurelia refused to accept her daughter's dark feelings toward her, attributing the lapse to mental anguish. \"My mother was always my best friend and I'd hoped that my daughter would be too,\" she said. \"She became ashamed of our friendship during her breakdown. I don't want to accuse anybody. I don't want to blame anybody, but . . . somebody had to be the scapegoat.\"\n\nThe American edition of The Bell Jar wasn't published until April 1971, and it would remain on the New York Times best-seller list for six months (fueled no doubt by the author's posthumous fame). Finally, the novel also achieved critical acclaim. When the paperback was issued a year later, three editions sold out within a month.\n\nAurelia had done her best to stop publication. The year before, she implored Plath's editor at Harper & Row to reconsider. \"I realize that no explanation of the why of personal suffering that this publication here will create in the lives of several people nor any appeal on any other grounds is going to stop this, so I shall waste neither my time nor yours in pointing out the inevitable repercussions,\" she wrote. Nearly every character in The Bell Jar, she claimed, \"represents someone\u2014often in caricature\u2014whom Sylvia loved; each person had given freely of time, thought, affection, and, in one case, financial help during those agonizing six months of breakdown in 1953 [the year in which the novel is set] . . . as this book stands by itself, it represents the basest ingratitude.\"\n\nTo Aurelia, the novel also gave the world a gross distortion of her daughter's supposedly true self, undone by mental illness. \"Sylvia never wanted it to be published here,\" she told a New York Times reporter in a 1979 interview, which took place in the white frame suburban house where Sylvia and Warren grew up. \"She'd had two babies and an appendectomy and needed money. 'I have to write a best-seller,' she told me. 'I want to write a potboiler. What would you suggest for a subject that wouldn't fail?' I suggested a child-parent conflict. I little knew what shape it would take.\" As usual, Aurelia made everything all about her, and she came across as self-absorbed and self-pitying. She spoke repeatedly of her vulnerability and painted herself as a victim. \"When The Bell Jar came out in 1971, it became a very hard time for me,\" she said. \"It was accepted as an autobiography, which it wasn't. Sylvia manipulated it very skillfully. She invented, fused, imagined. She made an artistic whole that read as truth itself.\"\n\nJust a few days before Plath died, she had written optimistic letters about her future\u2014including horseback riding again, an activity she loved. When she committed suicide, the first rumor in the United States was that she had died of pneumonia, a rather sunnier cause of death that Aurelia Plath, always in willful denial of monstrous truths, wanted to believe. (The official cause was deliberate carbon monoxide poisoning.) Ted Hughes handled the grim task of identifying his wife's body and confirming her name, age, occupation, and address. Their children did not attend the funeral. Plath's tombstone inscription read, EVEN AMIDST FIERCE FLAMES THE GOLDEN LOTUS CAN BE PLANTED.\n\nSixteen years after her daughter's death, Aurelia continued to reckon with the grotesque portrait of herself that was presented in The Bell Jar. \"Can you imagine what it is like to relive it over and over and over again?\" she said of her daughter's crippling legacy. \"It is only because I've been compelled to. It is because I have the name Plath. Anytime I meet anyone, the same thing happens. It happens to my daughter-in-law, their two girls, my son, of course. I was on Nantucket recently having a joyous time with a dear friend. She introduced me at a party and the other woman said: 'Oh . . . you are, aren't you?' I just can't escape it. The warm greeting until the name strikes them and they think of The Bell Jar, and of Mrs. Greenwood, the uncaring mother. 'Oh so you are Mrs. Greenwood,' they say.\"\n\nAurelia once wrote to the scholar and poet Judith Kroll, who had published the first full-scale critical study of Plath's poetry, Chapters in a Mythology. It was evident in her letter that Aurelia's trauma would never heal. (She died at the age of eighty-seven in 1994.) \"[Sylvia] made use of everything and often transmuted gold into lead,\" she explained. \"These emotions in another person would dissipate with time, but with Sylvia they were written at the moment of intensity to become ineradicable as an epitaph engraved on a tombstone. . . . She has posthumous fame\u2014at what price to her children, to those of us who loved her so dearly and whom she has trapped into her past. The love remains\u2014and the hurt. There is no escape for us.\"\nHe was a stinky drunkard with brown teeth and dirty hair\n\nChapter 11\n\nHenry Green & HENRY YORKE\n\nHe's the best writer you've never heard of. If you have read any or all of Henry Green's nine novels, you know that you're in on a too-well-kept secret. You probably wish that everyone with fine literary taste (such as yours) could experience the intense pleasure of reading him for the first time. That's no easy feat, since most of his books are out of print and the few that aren't are nearly impossible to find. Quiz a bunch of people who consider themselves well read, and a surprising number will admit that they have never read Green's books and are not even familiar with his name. Depending on your temperament, this response will leave you feeling disappointed or smug.\n\nHenry Green was the nom de plume of Henry Vincent Yorke, an Englishman born on October 29, 1905, in Gloucestershire. He grew up in a fourteenth-century manor, called Forthampton Court, on a 2,500-acre estate. Henry came from fancy stock: his handsome, athletic father, Vincent, had attended Eton and Cambridge and was a former archaeologist and explorer turned businessman; his mother, Maud, was the daughter of a baron who owned one of the grandest houses in England, a man who was among the richest British aristocrats of his era. One of her uncles was prime minister. Her great-grandfather was an earl and a well-known patron of the arts, one of the first supporters of J. M. W. Turner. Although Maud was an affectionate mother to her three sons\u2014Henry and his older brothers Philip, who would die at sixteen of lymphatic leukemia, and Gerald\u2014she preferred spending time with her beloved dogs and horses and indulging in her great love, reading. She was born with a curvature of the spine, yet had been quite athletic in her youth\u2014shooting pheasants, hunting avidly, and breeding racehorses. (She'd continued riding horses well into the sixth month of her pregnancy with Henry; he later insisted this had undermined his health and been the source of his neurotic temperament.) Maud, who spoke in a clipped military diction, was a witty, intelligent woman who loved to gossip. She was an eccentric character, said to have instructed her gardener to bowl turnips down a grass slope so that she could shoot at them. Maud almost always wore black or navy blue, and because she was a chain-smoker (of Turkish cigarettes), she was left in old age with only one brown-stained tooth. She refused to wear dentures.\n\nIn childhood, the Yorke boys were left largely in the care of nannies and servants who taught them proper manners and reined them in when necessary. Their father was an aloof presence in their lives. \"We were well brought up and saw our parents twice a day,\" Henry later wrote, \"that is to say my father worked in London and we only saw him at weekends.\" When he was there, Vincent was taciturn to the point of hostility. Unlike Maud, he lacked a sense of humor, and he envied her social ease. Most often, Vincent behaved toward his family like an irascible bully. Affection played no part in his emotional repertoire.\n\nPhilip and Gerald appeared somewhat more in the mold of their father\u2014brash, confident, excellent athletes and hunters\u2014but their younger brother was timid, awkward, lonely, and plump. He had no knack for academics or sports. (Henry described gym class as \"harrowing.\") He took his family's wealth for granted, yet he also felt estranged from it, identifying with servants, butlers, and working-class men far more than with his fellow aristocrats, whose company he found boring. Opulence was lost on him, which partly explained why Vincent found Henry such an awful disappointment.\n\nNicknamed \"Goosy\" at home, Henry was educated at Eton and Oxford but was not able to match the impressive academic records of his brothers. His time at Eton was unremarkable. Philip in particular had been a star there, and after his death in 1917, Henry felt even more inadequate. \"I needed praise badly,\" he wrote later, \"and if I had had it might be even less of a person now, but from the lack of it at that time found everything pointless, so blind that no effort at work or play ever seemed worth while.\"\n\nBut he and some friends did form a Society of Arts, a creative outlet that gave him a sense of belonging, and he began writing short stories. \"This point is a watershed, after this there was no turning back,\" he later wrote. \"I determined to be a writer . . . and a nom de plume was chosen, of all names Henry Michaels.\" He published a few pieces in College Days, the school literary magazine, an accomplishment his parents regarded with suspicion and disdain. One of his stories, \"Bees,\" which appeared in 1923, follows a clergyman from \"a slum parish in Liverpool\" who suffers from malaise in a \"sleepy, unenthusiastic\" village. Though brief and spare, it is well written and reveals a certain psychological acuity; one wouldn't necessarily guess that the story had sprung from the mind of an eighteen-year-old:\n\nAll day long he thought of how he was to stand the blow of his daughter's death, and, although it was eighteen months since she had died, he was still composing answers in his mind to the letters of condolence that never came. In the busy buzz of his bees he detected the sympathy he could not discover in the world outside. His wife, whom he always regarded as a drone, could do nothing with him. He was sure that every man's hand was against him. He detected an insult in the butcher boy's whistling as he delivered the meat. So he turned to his bees, who always sympathized, and were so practical, and who were not useless like his family.\n\nWithout telling Henry, his mother sent his stories to a friend, the Scottish writer John Buchan (most famous for The Thirty-nine Steps, adapted into a film directed by Alfred Hitchcock). Buchan offered encouraging words about Henry's work. \"Whatever your boy's stories are, they are not a waste of time,\" he wrote. \"They are curious stories, rather like the kind of thing that Hans Christian Andersen wrote in his youth. They show great powers of observation, great sensitivity to scenery, and the nuances of temperament, and a strangely mature sense of the irony of life.\" He added, however, that one ought to use writing as a hobby rather than a profession, and that Henry \"seems to have literary gifts of a high order, but he wants the discipline of more normal subjects. It would be exceedingly good for him to try his hand at concrete objective narrative for a change.\" Concreteness, as readers would later discover, was perhaps the quality most absent in Henry's writing\u2014and perhaps most abhorred by him.\n\nHenry became close to one of his Eton classmates, Anthony Powell, who would become an author as well. The boys shared their enjoyment of storytelling and even began (but abandoned) collaborating on a novel. Years later, Powell recalled his friend as \"always interested in words, repeating unfamiliar ones (e.g., hirsute) over to himself, laughing at them, discussing them.\" Another close friend, Robert Byron, admired Henry's peculiarity and shared his irreverent humor. Byron later said of him, \"He can talk like no other person I've ever met.\"\n\nIn 1924, Henry began writing a draft of what would become his impressionistic first novel, Blindness, published when he was just twenty-one. The original typescript was signed \"Henry Browne.\" Eventually he would settle on the bland pseudonym \"Henry Green\"\u2014never publishing a book under his real name. That choice may have had to do with his aristocratic upbringing, which frowned upon such a self-centered vocation. His friend and classmate Harold Acton did not approve of his pseudonym. \"There are Greens of so many shades writing novels that one wishes he had selected another colour,\" he said. (Henry \"Green\" later befriended the novelist Graham Greene, whose full name was Henry Graham Greene.) Henry was forever caught between the desire for revelation, for confession, and the reticence expected of someone of his class. He wanted to remain enigmatic and private while at the same time fully exploring human emotion and experience. Anonymity seemed to offer a comfortable compromise. \"Names distract, nicknames are too easy, and if leaving both out . . . makes a book look blind then that to my mind is no disadvantage.\" Perhaps he meant that in making a book \"blind,\" cloaking it cleverly enough, more distance would be placed between reader and writer; keeping the reader slightly \"in the dark\" was not a bad idea. The author himself could not be examined too closely\u2014only his work. Res ipsa loquitur: the thing speaks for itself.\n\nAt Oxford, Henry's tutor was C. S. Lewis, who had no respect for Henry's literary interests, especially for his appreciation of \"experimental\" writers. Henry regarded Lewis as \"rude and incompetent.\" Studying was not Henry's priority, anyway. He estimated that he put in no more than six hours of academic work a week. Most nights he was drunk. He played billiards, stayed up late, and slept until around noon. His first meal of the day, accompanied by a brandy and soda, was always fried sole and sausages because \"I thought that by not varying my food I was giving my stomach less to do.\" He had another routine: going to the cinema every afternoon, sometimes twice a day, and returning to his room to write; he admitted that \"it became the last foothold to write just one more page a day, the last line of defence because I was miserable in fits and starts and felt insane.\"\n\nHenry's debut novel, which he characterized three decades later as \"mostly autobiographical,\" follows the callow sixteen-year-old John Haye (\"It sounds an awful thing to write, but I seldom meet anyone who interests me more than myself,\" he admits). Like the author, John enjoys reading Carlyle, Turgenev, and Dostoevsky. On his way home from a repressive boarding school called Noat (a thinly disguised Eton, which was \"Note\" in earlier drafts), John is blinded. The account of his accident is clinical: a boy throws a rock at a train; the window smashes; John, sitting behind it, loses his sight. His stepmother (who is obviously Maud) spends her time hunting and horseback riding. She tries to marry him off to any girl of the \"right\" social class, so she won't have to spend the rest of her life caring for him. The story is told from multiple points of view, including those of a young girl and a drunken clergyman. Henry dedicated the novel to his mother.\n\nA review appeared in the New York Times on November 14, 1926, shortly after the book was released in the United States: \"It is reported to be the first novel of a very young man. In spite of certain defects of workmanship, of prolonged episodes, meandering dialogue and confusion of method, it does convey a sense of character under stress. It is a creditable performance.\"\n\nBlindness doesn't have much of a plot\u2014Green, like modernists such as Woolf and Joyce, was far more interested in the interior life, memory, emotion, language, and metaphor than in creating tidy, linear, plot-driven stories. \"I write for about six people (including myself) whom I respect and for no one else,\" he once said. In Green's work, there was no authorial guidance as to how a reader should \"feel\" about any character. Ambivalence reigned. Henry was already an eccentric and sophisticated thinker. He understood that less is more, and that sometimes, nothing is even more.\n\nThe gaps and flaws in Blindness, including its too-abrupt ending, could be attributed to the immaturity of the author (of which he was well aware), yet even in later novels he favored an oblique approach that did not fill in many narrative blanks. Of his nine novels, not one is like another. Motifs change from book to book. Likable characters are not considered crucial. There are no feel-good endings. And no character learns a moral lesson or is transformed by experience. Life simply goes on. In a sense, you might say that Green was the Jerry Seinfeld of his day. Calling to mind the comedian's approach to humor, with his \"show about nothing,\" Green believed that \"the novel should be concerned with the everyday mishaps of ordinary life,\" as he told an interviewer in 1950. (John Updike once proclaimed Green \"a saint of the mundane.\")\n\nWith its knotty diction and odd syntax, his fiction, he knew, was not for everyone. Fortunately, Henry found a sympathetic ear in Nevill Coghill, an Irish don at Oxford who became a close friend; Coghill believed absolutely in the young writer's talent and proved a steady source of support and advice. In a 1925 letter, written while on holiday, Coghill was filled with regret as he reported that his brother and a cousin had picked up Henry's manuscript and were not so taken with it: \"Alas they think it difficult, depressing, ungrammatical (!!!) carelessly written!!! This so infuriates me that I shout at them, telling them it is a work of undying genius and that they are too crapulous to understand it. To which they reply 'Ah, but I like a good story.' Poor Henry. I am so sorry. But I am sure that your way of writing is a very good way and is right for you.\"\n\nAt the time, Henry may not have been aware of Emily Dickinson's \"Tell all the truth but tell it slant\u2014 \/ Success in Circuit lies,\" but he certainly practiced it in his writing. He was keenly interested in playing with different stylistic techniques, and in applying Chekhov's notion of significant irrelevance, in which details were teased out through indirect means. \"Irrelevancy means so much,\" he wrote to Coghill, \"it shows you what a person is & how he thinks, & conveys atmosphere in a way that is inconceivable if you have not seen Tchekov's Cherry Orchard.\" He finished his novel on May 30, 1925, a few months before his twentieth birthday, noting the precise time and date of completion on the last page of the manuscript. He promptly (and rather boldly) sent it to Chatto & Windus, the distinguished London publishing house of authors such as Wilkie Collins and Samuel Beckett, and the first English translation of Marcel Proust's novel \u00c0 la recherche du temps perdu. (Henry was a great admirer of Proust.) The editor who received Henry's novel was not impressed. \"I do not make much of this MS., which depressed me at the start (by the boringness of the schoolboy mind) and went on depressing me (by the boringness of everything) to the end,\" he wrote in a memo. \"Nevertheless,\" he added, \"the author should not be lightly condemned, because he evidently is very fluent, and his talent may develop.\"\n\nHe mailed the manuscript back to the author with a standard letter of rejection. Henry was furious and told Coghill that Chatto was a \"despicable firm.\" His friend suggested that he send his work to the publisher J. M. Dent, who accepted it after requesting some revisions. An editor there asked Henry, \"How did you ever come to write anything so good?\"\n\nDent was known for creating the Everyman's Library\u2014handsome limited editions of classic literature, offered at an affordable price (one shilling). Because Henry was still legally a minor when his book was accepted, his father had to sign his publishing contract. Gerald Yorke later described their parents' response to the novel as \"not quite horror but complete misunderstanding and great doubt.\" They did nothing to assuage Henry's anxiety about disappointing or upsetting them. One of Henry's aunts interpreted Blindness as a cry \"for sympathy which he doesn't find at home.\"\n\nAt Oxford in the fall of 1926, Henry was very lonely. Many of his friends had graduated and moved on, and he fell into a depression. He wanted out. \"Everyone is rich and vapid or poor and vapid & one & all talk about Oxford day & night,\" he complained in a letter to his mother. He wanted to go to work \"in a factory with my wet podgy hands\" for the Birmingham branch of H. Pontifex and Sons, his family's coppersmithing company. Vincent Yorke had several enterprises, including positions in banking, insurance, and railways\u2014all secured for him by his father, John Reginald Yorke, who had purchased Pontifex for him as well. The company, which was then failing, had once made plate engravings for William Blake. Vincent proved a savvy businessman, making Pontifex profitable again by moving the factory to a cheaper site (Birmingham instead of London) and by expanding the manufacturing business into bathroom plumbing and brewery equipment.\n\nHenry decided that he was finished with Oxford's academic pressures. He resented being forced to spend his time studying, he said, \"when I have my own work always running in my mind.\" In December, shortly after the publication of Blindness, he dropped out of Oxford without earning a degree. His friend Evelyn Waugh, at work on his own first novel at the time, was enthusiastic about Henry's literary debut: \"It is extraordinary to me that anyone of our generation could have written so fine a book.\"\n\nHenry was eager to trade his stuffy university environment for the factory floor, partly because he sensed that the rhythms and sounds of proletarian idioms could provide material for his next novel, and partly because he longed to experience what he called \"the deep, the real satisfaction\" of manual labor. (His interest in the working class was not unlike that of his contemporary, and fellow pseudonymous writer, George Orwell.)\n\nIn January 1927, Henry reported at the Midlands iron foundry, where he would quickly build up his muscles moving heavy machinery for eight hours a day, earning twenty shillings a week. Although coworkers assumed that he'd been assigned the job as some kind of shameful punishment, it was at Henry's insistence that he started as an apprentice, working his way up from the bottom and living in a Victorian boardinghouse.\n\nThe setting was as unrefined as he'd hoped for, and he toiled away at Pontifex quite happily for the next two years. \"I had been an idler who had at last found something to occupy his mind and hands,\" he later wrote. A hundred and fifty people worked at the factory that he would one day control. There was no lack of colorful characters to keep him entertained. Henry proudly reported to his mother that he had met a man who \"bites the heads off mice to kill them when the trap hasn't.\" Despite working harder physically than he ever had, he still found time to indulge in one of his favorite hobbies\u2014his addiction to the cinema\u2014and to write a few hours each day. \"Going home it would be dark again and I would be tired,\" he later recalled. \"But after no more than thirty minutes in a chair I was ready for hard work again.\" His moviegoing and his writing were not unrelated; Henry aptly described the draft of what would become his second novel, Living, as \"a kind of very disconnected film.\" The novel, tentatively titled \"Works,\" was set in a Birmingham iron foundry and captured the monotony of the workers' lives. Written with extraordinary sensitivity and empathy, without a trace of condescension or sentimentality, it was a reflection of the author's lifelong affection for the working class, and of his ambivalence about his own pedigree. He once told his mother bluntly that she should accept the fact that \"by nature I am not the sort of person who dresses for dinner every night, in fact I am not what is generally known as a gentleman.\"\n\nHe led two lives. By day, he was Henry Yorke, laborer and aspiring businessman; in his private writing time, he was Henry Green. He preferred that the two personae would never meet. For one thing, as he later explained, \"I write books but I am not proud of this any more than anyone is of their nails growing.\" And for another, as he explained in a 1958 Paris Review interview with a close friend, the American novelist Terry Southern, \"I didn't want my business associates to know I wrote novels.\" The role of artist seemed pretentious and ill fitting. While contemporaries such as George Orwell (Eric Blair) were engaged in polemical writings and political activism, Green was quietly crafting his strange fiction. He shied away from publicity, avoided being photographed in public, and had deliberately chosen a pseudonym that was unremarkable and did not call attention to itself. He explained that \"if you are trying to write something which has a life of its own, which is alive, of course the author must keep completely out of the picture.\" As Sebastian Yorke (Henry's son) later noted, Henry's own father regarded his son's books \"with silent contempt because they did not make money,\" which only reinforced the notion that Henry Yorke ought to remain as invisible as he could.\n\nIn 1949, upon the U.S. publication of Green's fifth novel, an American critic asked:\n\nWho is Henry Green? Well, there's an elaborately built-up mystery about that, though you could probably soon find out in England. Particularly if you could inspect British income-tax records. He is [according to his publisher] a fellow with a passion for anonymity, a Birmingham manufacturer, an Etonian, an Oxonian, possibly a Bolognian, too, no less. . . . It may be that he is really Graham Greene. It may be that he is Ivy Compton-Burnett's great-grandfather.\n\nBy 1958, most of his colleagues at Pontifex, at least, were well aware that Henry Yorke had an alter ego called Henry Green, and he admitted in his interview with Southern that the revelation had affected his relationships with them. \"Yes, yes, oh yes\u2014why, some years ago a group at our Birmingham works put in a penny each and bought a copy of a book of mine, Living,\" he said. \"And as I was going round the iron foundry one day, a loam molder said to me, 'I read your book, Henry.' 'And did you like it?\" I asked, rightly apprehensive. He replied, 'I didn't think much of it, Henry.'\"\n\nIt was no wonder: aside from the brilliant music of common speech, which Green captured beautifully (\"I got you fixed in me mind's eye tucking away lamb with mint sauce\"), his prose style was scrambled and demanding. There were sentences with loose grammar, absent nouns, cryptic references, and articles dropped at will: \"Hundreds went along road outside, men and girls.\" \"Range made kitchen hotter.\" \"Baby howled till mother lifted him from bed to breast and sighed most parts asleep in darkness.\"\n\nAlthough Green's admirers placed him alongside authors such as Woolf, Joyce, Lawrence, Kafka, and Sterne, he claimed no influences himself. \"As far as I'm consciously aware,\" he said, \"I forget everything I read at once, including my own stuff.\" He explained, too, that \"Joyce and Kafka have said the last word on each of the two forms they developed. There's no one to follow them. They're like cats which have licked the plate clean. You've got to dream up another dish if you're to be a writer.\"\n\nBy the time Green was twenty-four, he'd already written two boldly experimental novels. Yet in the memoir\u2014if you can call it that\u2014that he published in 1940, Pack My Bag, he hardly mentions the publication of either book, or his third novel, Party Going, which came out the year before; or any sense of pride in his accomplishments. He names almost none of the people in his life, not even his wife and young son; he does not reveal that the main schools he attended were Eton and Oxford; and in no way does he describe the effect that Philip's death had on him. He entirely skips the decade of his life prior to 1938 (his story stops when he is twenty-four). And he does not even bother to explain why Henry Yorke had become Henry Green. The book's subtitle, \"A Self-Portrait,\" seems a kind of joke. It is a work of great originality, but one in which the author, as usual, omits the most basic details and presents the rest mostly through a blurry viewfinder. (Kingsley Amis said of the book that it seemed \"the author was drunk whilst writing it.\") There is a willful perversity in the way Green hoards and obfuscates information. Evelyn Waugh wrote to him at the time that \"it was a book no-one else could have written and it makes me feel I know [you] far less well than I did before which, in a way, I take to be its purpose.\"\n\nDespite its baffling omissions, Pack My Bag was deeply important to the author. Written when Green was thirty-three, his \"interim autobiography\" was the result of his terror that he would die in the impending war. He published his book hastily because \"we who may not have time to write anything else must do what we now can.\"\n\nThe memoir finished as enigmatically as it began, and abruptly, too\u2014though on a somewhat tender note, alluding at the very end to his epistolary courtship of the woman he would marry: \"It was not hunting when it was no fun, not having to go shooting, it was not having to be polite to masters who were fools, it was to lose convictions, at a blow it was life itself at last in loneliness certainly at first, but, in that long exchange of letters then beginning and for the ten years now we have not had to write because we are man and wife, there was love.\"\n\nGreen delivered his tersely titled novels in efficient succession: Party Going (1939), Caught (1943), Loving (1945), Back (1946), Concluding (1948), Nothing (1950), and Doting (1952). Then came silence, a literary purgatory that lasted until Green's death in 1973.\n\nIn 1929, the year that Henry Green published Living, Henry Yorke married an upper-class Englishwoman, Mary Adelaide Biddulph, known as Dig. Waugh affectionately called the couple \"Mr H. Yorke the lavatory king and his pretty wife.\" Henry, along with his parents, had decided that once he'd completed his latest novel, he would move to London and assume a new role at the family firm: managing director. He knew that the structure of an office job would keep him stable, yet he was also \"violently depressed\" at the time. \"My fucking novel is so absolutely mediocre,\" he told Anthony Powell. His editor, too, had commented that Green's elliptical prose style was \"difficult, & a trifle affected.\" When it was released, the book was neglected critically, perhaps owing to the crowded, exceedingly impressive publishing field that year: Ernest Hemingway, Italo Svevo, Rebecca West, V. S. Pritchett, and Robert Graves all brought out new works. But Green's prominent literary friends helped boost his spirits, providing a welcome antidote to reviews such as one from the Times Literary Supplement, which asserted that the author \"does not seem to care in the least whether the reader is thrilled, bored, delighted, or irritated.\" Waugh considered Living a masterpiece and compared Green's dazzling technical feats to those of T. S. Eliot. He also emphasized that the author's radically ambitious aims made it \"necessary to take language one step further than its grammatical limitations allow. The more I read it the more I appreciate the structural necessity of all the features which at first disconcerted me.\" Regardless of its originality, the novel failed to sell many copies.\n\nThe decade-long interval between the publication of Green's second and third novels was the result of frustrations and distractions, bouts of depression and paranoia; the demands of business, of upper-class society, of fatherhood (his son was born in 1934); and the onset of the war. As one critic later wrote of Green, the neglect of his literary legacy came about partly because he \"lived several lives not sequentially but in parallel.\" He never fully inhabited one identity or the other. The ambivalence evident in his work was also reflected in his personal attitudes. As Anthony Powell noted of his friend, \"[I]f one side of Yorke found the silver spoon a handicap to respiration, another accepted it as understandably welcome; and coming to terms with opposed inner feelings about his family circumstances, his writing, his business, his social life, was something he never quite managed to achieve to his own satisfaction.\"\n\nDuring the London Blitz, Green volunteered for the Auxiliary Fire Service, an experience that would provide material for Caught. His preceding novel, Party Going, comically followed a group of aimless young rich people stranded in London's Victoria Station during a heavy fog. Not much happens, and the characters aren't particularly likable. (Seinfeld again comes to mind.) The novel's startling, bizarre opening line set the tone for the disorientation that lay ahead: \"Fog was so dense, bird that had been disturbed went flat into a balustrade and slowly fell, dead, at her feet.\"\n\nAfter the war, Henry assumed his position as managing director at Pontifex, and the novels he continued to write were greatly admired by W. H. Auden, Christopher Isherwood, Elizabeth Bowen, Roald Dahl, and other prominent literary figures. He also kept busy as a serial philanderer who was as cruel as he was charming. \"Hurting\u2014that should be the title of your next novel,\" one of his girlfriends suggested bitterly.\n\nThat Henry Green published nothing after 1952 is explained by the sad decline of Henry Yorke. His lifelong despair started to overtake him and never loosened its grip, eventually leaving him adrift even from himself. Although he was mostly deaf (a condition that worsened during the war), he refused to wear a hearing aid, which isolated him still further from others. He drank and drank. Half his days were spent in pubs, and sometimes he'd return to a pub after dinner and stay until closing time. \"To the regulars he was simply Henry who always sat at the same table wearing his raincoat and hat with a glass of gin and water beside him,\" Sebastian recalled of his father.\n\nHenry had not lost sight of the mission of writing fiction, but he could no longer fulfill it. \"Prose is not to be read aloud but to oneself alone at night,\" he believed, \"and it is not quick as poetry but rather a gathering web of insinuations which go further than names however shared can ever go. Prose should be a long intimacy between strangers with no direct appeal to what both may have known. It should slowly appeal to feelings unexpressed, it should in the end draw tears out of stone.\"\n\nHe once claimed that he could \"only get myself right by writing.\" He insisted that writing alone had given him happiness, and that he relied upon it to stay sane. Yet, for some reason, he could no longer gain access to the part of himself that yielded such pleasure.\n\nHenry continued to oversee Pontifex, which had experienced a brief postwar boom, but the company too began to decline. With his stubborn inattention to detail, his pessimism and pathological indecisiveness, and his increasingly erratic behavior, he proved a poor chairman, and the company suffered. In 1958, it was discovered during a board meeting that Henry's water glass contained neat gin. He was forced to retire a year later.\n\nAfter 1960, the man whom Terry Southern had called a \"writer's writer's writer\" rarely left his house. He dictated the beginning of an intended sequel of sorts to his memoir, called \"Pack My Bag Repacked,\" a project that, like many others, he soon abandoned. In one draft, he refers to himself in the third person: \"Green lives with his wife in Belgravia. He has now become a hermit. . . . Green can write novels, but his present difficulty is to know quite how to do it.\" He spent much of his time watching TV, especially sports. He often wandered around the house in a shabby state, littered with cigarette ash and wearing mittens because he said that his hands were always cold. Sebastian recalled that his father's hearing grew steadily worse. He once phoned home and asked to speak to \"Mummy,\" to which Henry replied, \"So sorry, I have absolutely no money.\"\n\nIn 1962, a BBC interviewer asked Green, \"Are you going to write any more books?\" He replied wearily: \"No\u2014never\u2014never. . . . It's too exhausting, I can't do it.\" He'd lost his drive and was convinced that no one wanted him to find it again. \"I'm absolutely finished as far as the public's concerned,\" he said. \"I mean, I'm out, I don't sell books any more, and the critics despair of me. No, I don't exist any more.\"\n\nIn spite of his black moods, he wasn't entirely gone. He loved reading books (but hated talking about them), and consumed about eight a week\u2014always novels, no poetry or nonfiction. Contemporary British and American fiction appealed to him; he had catholic tastes and read widely, but he refused to read Georges Simenon or C. P. Snow. Like Simenon, Henry idolized Faulkner, and meeting him in 1950 was one of the highlights of Henry's life. He told an interviewer that he wanted Faulkner, more than anyone else, to read his books. (It isn't known whether Faulkner did.)\n\nOver the years, Henry alienated many of his friends with his drunken, maudlin, self-destructive behavior, and they stopped calling on him. He'd become a charmless embarrassment. One friend recalled observing his rare presence at a dinner party, \"talking away as if driven by a demon, looking very much the worse for wear.\" Another compared him to F. Scott Fitzgerald: \"He drank because he couldn't write and he couldn't write because he drank.\" This was perhaps the most succinct diagnosis of Henry's predicament. And in a letter to Nancy Mitford, Evelyn Waugh, who could be quite vicious about Henry's diminished state, described him with sheer disgust: \"He looked GHASTLY. Very long black dirty hair, one brown tooth, pallid puffy face, trembling hands, stone deaf, smoking continuously throughout meals, picking up books in the middle of conversation & falling into maniac giggles, drinking a lot of raw spirits, hating the country & everything good. . . . I really think Henry will be locked up soon.\" In 1968, after much coaxing, Sebastian convinced his father to accompany him to an event at London's Albert Hall, to which Henry came unshaven and wearing bedroom slippers.\n\nWhat had become of the promise of Henry Green, a writer who, as the author and translator Tim Parks put it, \"must be the most highly praised, certainly the most accomplished, of twentieth-century novelists not to have made it into the canon, not to be regularly taught in universities, not to be considered 'required reading'\"? One critic astutely described Green as having shown \"more subtlety and virtuosity than any other novelist of his generation in England. And yet Green's very mastery of his medium has kept him from the recognition he deserves.\" Eudora Welty, who'd met Green once and adored him, lavished praise on his underappreciated work in a 1961 essay: \"The intelligence, the blazing gifts of imagery, dialogue, construction, and form, the power to feel both what can and what never can be said, give Henry Green's work an intensity greater, this reader believes, than that of any other writer of imaginative fiction today.\" And the critic James Wood has written that after D. H. Lawrence and Virginia Woolf, Green was the greatest English modern novelist.\n\nThe fact remained that despite occasional success (Loving had appeared briefly on best-seller lists in the United States), no single book of his had sold more than ten thousand copies in England. His novels had slipped in and out of print even in his lifetime. Of course, Green had sabotaged himself through bizarre financial and marketing decisions: declining offers for paperback sales because of paranoia about income tax debt, and refusing to provide photographs or biographical information to his publishers. Though he reluctantly agreed to come to New York to help launch Loving, he registered at his hotel under another pseudonym, H. V. Yonge, whose initials at least matched those of his real name. But he didn't mask his hatred of publicity and made it as difficult as possible for admirers to meet him. Considering how disinclined he seemed to achieve a wider readership, it is no surprise that his publishers were unable to earn a profit from his work. They promoted him as best they could under rather challenging circumstances, undoubtedly out of absolute belief in his prodigious talent. As with everything else to do with him, nothing was straightforward or even rational.\n\nAfter he'd stopped publishing, Green succumbed to sporadic bouts of inconsolable weeping, telling anyone who would listen that he suffered from a lack of recognition and believed he was a failure. Once, while visiting his brother and sister-in-law, he cried as he complained, \"I've never won any of the good prizes.\" If Green was misunderstood or neglected, he seemed oblivious that any of this was his own doing.\n\nOf the relatively few people who were aware of Green's novels, perhaps too many shared the view of the New York Times critic who found his work baffling to the point of irritation, and who dismissed Green as another case of Emperor's New Clothes. Green was blasted by the critic for writing \"peculiar, artificially mannered novels of limited appeal which are extravagantly overpraised by a few critics whose pride it is to admire books which lesser mortals don't appreciate.\" This naysayer, however, later revised his opinion, admitting, \"I didn't like green olives the first few times, either. Maybe Mr. Green is an acquired taste.\"\n\nHenry Yorke died at the age of sixty-eight on December 13, 1973, from bronchial pneumonia, after being bedridden for quite some time. Henry Green had been dead for years. \"He was a very very complicated and tricky person,\" Anthony Powell recalled of his old friend. \"And although we knew each other so well, of all the people I've ever known I really never got to the bottom of him.\"\nHe could fool some of the people all of the time\n\nChapter 12\n\nRomain Gary & \u00c9MILE AJAR\n\nHe was a war hero, a Ping-Pong champion, a film director, a diplomat, and an author who wrote the best-selling French novel of the twentieth century. Being famous made him tired. He wanted to be someone else, but one invented persona was not enough.\n\nRoman Kacew was born on May 8, 1914, in Vilna, Russia, and raised by a Jewish single mother, Nina Owczinski, a former stage actress. By the age of thirteen, he believed that he was destined to become a great writer. At this age, too, he took up smoking, a habit encouraged by his mother. (She would smoke three or four cigarettes when she woke each morning, and happily shared her Gauloises with her son.) Nina was a devout Francophile, and emigrated with her son in 1928 to Nice, where she instructed Roman to change his name so he could become famous. \"You must choose a pseudonym,\" she said. \"A great French writer who is going to astonish the world can't possibly have a Russian name.\"\n\nHe began experimenting with pen names\u2014spending hours each day hunched over his exercise book and testing out \"noble-sounding\" noms de plume in red ink. He toyed with \"Hubert de la Vall\u00e9e\" and \"Romain de Roncevaux,\" among many others. \"The obvious trouble with pen names,\" he discovered, \"even with the most inspired and impressive ones, was that they somehow failed to convey truly the full extent of one's literary genius.\"\n\nSuch healthy self-regard was inspired in no small part by his mother, who considered her son the center of the universe. Anyone who failed to recognize that fact was an idiot. Neighbors who were annoyed by Nina's constant proclamations of his glory were denounced as \"dirty little bourgeois bedbugs.\" If her son scored poor marks at school, it was everyone else's fault. In his fictionalized 1960 memoir Promise at Dawn, he recalled an exchange that occurred one day when his mother asked how things were going at school:\n\n\"I got another zero in math.\"\n\nMy mother thought this over for a moment.\n\n\"Your teachers don't understand you,\" she said firmly.\n\nI was inclined to agree. The persistence with which my teachers kept giving me zeros in science subjects seemed to indicate some truly crass ignorance on their part.\n\n\"They'll be sorry one day,\" my mother assured me. \"The time will come when your name will be inscribed in letters of gold on the wall of their wretched school. I'll go and tell them so tomorrow.\"\n\nRoman found his mother's relentless adoration both awe-inspiring and paralyzing. She encouraged him to become a \"giant of French literature\" only after suffering disappointment that he was not a violin prodigy, a budding Jascha Heifetz. She'd pointed out that if he were to become a famous violinist, \"our real name, Kacew, or even better, my stage name, Borisovski, would be excellent.\" But it was not to be. After Nina bought Roman a secondhand violin when he was seven years old, she'd signed him up for private lessons, but the instructor dismissed him after three weeks. \"A great dream had left us,\" Roman recalled.\n\nHe and his mother led an itinerant life, dependent mostly on the latest way she'd devised to reinvent herself. He would grow up in Russia and Poland and on the French Riviera. \"My mother was always waiting for the intrusion of the magical and marvelous into her life,\" he wrote, \"for some deus ex machina that would suddenly come to her rescue, confound the doubters and the mockers, take the side of the dreamer and see to it that justice was done.\" She earned a living making hats, running a hotel, selling furs and antiques, and other occupations, but perhaps her most memorable venture was in \"second-hand teeth\"\u2014buying teeth containing gold or platinum, then reselling them at a highly marked-up price.\n\nThough Nina eventually achieved financial security, her dream of being a famous actress never left her. This larger-than-life woman was the consummate stage mother, pushing her son to succeed as the artist she would never become. Roman was instilled with ambition, fear, frustration, and dread, but he was always determined to please her, to lay the world at her feet. In search of the vocation that would bring them acclaim and fortune, they exhausted various possibilities\u2014such as painting, acting, and singing\u2014before settling on literature, which he later noted \"has always been the last refuge, in this world, for those who do not know where to lay their dreaming heads.\" Not only did Nina expect her son to become an artist of renown, but she dreamed that someday he would become an ambassador of France and wear bespoke suits made in London. (Both came to be true.)\n\nAs mother and son plotted his future, Roman applied himself to crafting the pseudonym that would inspire literary masterpieces to flow like water. It was, he later recalled, no easy task to discover a name \"grand enough to compensate for my own feeling of insecurity and helplessness at the idea of everything my mother expected from me.\" Despite Roman's intensive brainstorming, nothing seemed right\u2014and both he and Nina were chagrined that names such as \"Shakespeare\" and \"Goethe\" were already spoken for. \"We were both getting terribly impatient to know, at last, under what name we were to become famous,\" he recalled. Fifteen years later, when he heard the name \"Charles de Gaulle\" for the first time, he felt it would have been the perfect pseudonym. None of the names he came up with satisfied him or his mother: \"Alexandre Natal,\" \"Armand de la Torre,\" \"Romain de Mysore\"\u2014these just weren't good enough.\n\nNina suffocated her son in a more significant way: she believed that no other woman should ever have him. He belonged to her alone. Worse, she refused to give herself to another man, and took great offense at the suggestion that she ought to try. Any attempts by the teenage Roman to explore his sexual appetite with beautiful young women were invariably crushed by his mother. \"I am not saying that mothers should be prevented from loving their young,\" he wrote later. \"I am only saying that they should have someone else to love as well. If my mother had had a husband or a lover I would not have spent my days dying of thirst beside so many fountains.\"\n\nThe grievous effect of such vast and forceful love was that for the rest of his life, Roman would seek, in vain, to recapture it. His craving for companionship was best fulfilled by a succession of devoted friends\u2014notably, Mortimer, Nicholas, Humphrey, and Gaucho, all of them cats\u2014and a dog named Gaston.\n\nAs he and his mother steeped themselves in French culture, Roman became \"Romain.\" Whenever Nina was harassed in France as an outsider, mocked as a \"dirty foreigner,\" she would retort by coolly informing the moronic offender that her son \"is an officer of the French Air Force and he tells you merde!\" Romain noted his mother's inability to distinguish between \"is\" and \"will be.\" (Her ardent idealism and willful denial were part of her charm.)\n\nMeanwhile, Romain felt desperate to somehow make his mark. He discovered that despite his failure at sports such as swimming, running, and tennis, he had a real knack for Ping-Pong. One of the engraved medals he later won at a tournament sat on Nina's bedside table until the day she died.\n\nIn his late teens, he also grew more serious about writing. \"Attacked by reality on every front, forced back on every side and constantly coming up against my own limitations, I developed the habit of seeking refuge in an imaginary world where, by proxy, through the medium of invented characters, I could find a life in which there was meaning, justice and compassion,\" he recalled. But because Romain had inherited that marvelous flair for self-mythologizing, he could not simply sit down and write. Under the watchful eye of his mother, as ever, he took a Method-acting approach to his craft. That was the only way he could become, as he hoped, \"the youngest Tolstoy of all time,\" and thus reward his mother for all the sacrifices she had made on his behalf.\n\nRomain flung himself headlong into his task. His dramatic first step was to assemble a pile of three thousand sheets of paper, which he estimated to be equivalent to the manuscript of War and Peace. Then, he recalled,\n\nMy mother gave me a dressing gown of ample proportions, modeled on the one which had already made a great literary reputation for Balzac. Five times a day she opened the door, set a plate of food on the table and tiptoed out again. I was, just then, using Fran\u00e7ois Mermonts as a pen name. Since, however, my works were regularly returned to me by the publishers, we decided that it was a bad choice, and substituted for it, on my next effort, that of Lucien Bulard.\n\nStill no luck. But in 1933, at the age of nineteen, Romain finally won a respite from his mother's overwhelming expectations. He enrolled, as a practical consideration, in law school at the University of Aix-en-Provence. He described the experience of bidding good-bye to Nina as \"heartrending.\" It was a healthy and much-needed separation. He spent his free time lingering at caf\u00e9s, and managed to write a novel. He promptly sent the manuscript to various publishers, and one responded by including a report from an acquaintance\u2014a well-known psychoanalyst to whom he had shown the novel. She indicated in her report that the author of this demented book suffered from a castration complex, a fecal complex, necrophilic tendencies, and other pathologies. In any case, the manuscript was politely declined. Undaunted, Romain took pride in being told that he had a fecal complex, which he felt marked him indisputably as a tormented soul\u2014and a genuine artist. He completed law school in Paris, neglecting his studies to spend several hours a day on his writing. Eventually, he published a few of his stories, and was thrilled to learn that one had even been translated and published in the United States. His early work, signed as \"Romain Kacew,\" was marked by a maturity and economy of prose that was impressive in someone not yet twenty-two years old. In 1935, he became a naturalized French citizen.\n\nDuring the Nazi occupation, Romain was admitted to what was regarded as the oldest and \"most glorious\" bomber squadron, the Lorraine, serving under de Gaulle in the Free French forces. Around this time he had begun using the surname \"Gari,\" anglicized as \"Gary.\" For some reason he never explained this rather crucial fact in Promise at Dawn. The name was not a random choice; it was yet another tribute to his mother, who had used \"Gari\" as one of her stage names.\n\nDuring World War II he had various postings throughout Europe and North Africa. Despite the physical and emotional battering he suffered, he was nonetheless chastised from afar by his ever-looming mother, who insisted that he ought to keep up with his writing. He knew that it would be futile to defend his inactivity by reminding her that there was a war on. So the obedient son set himself to work.\n\nNoting that it was hard to unleash his creative genius \"on a ship's deck or in a tiny cabin shared by two others,\" he persisted, attempting to cobble together stories that might turn into a coherent whole. Part of what would become his first novel, \u00c9ducation europ\u00e9enne (A European Education), was written on a steamer ship that carried him into battle. The latter half was composed at night, in a shared corrugated-iron hut. Every night, Romain\u2014wearing his flying jacket and fur-lined boots\u2014would write until three or four in the morning, \"with numbed fingers, my breath rising in visible vapor in the freezing air.\" He completed the novel in 1943, in Surrey, England.\n\nLater, Romain recalled his harrowing wartime experiences, the aftermath of which left him in a state of alienation unlike any he had ever known\u2014and one he would never quite shake. \"After four years of fighting with a squadron of which only five members are still alive, emptiness has become for me a densely populated place,\" he recalled. \"All the new friendships I have attempted since the war have made me only more conscious of that absence which dwells beside me.\" He also shared an insight about himself that would acquire an eerie and profoundly tragic meaning after he died. \"A fool I shall always be, when it is a matter of . . . smiling in the face of nothingness,\" he admitted. \"There is no despair in me and my idiocy is of the kind that death itself cannot defeat.\"\n\nHis combat service transformed him in many ways. He had survived dangerous missions, typhoid fever, and a plane crash that killed everyone aboard but himself. He was decorated with some of France's highest honors, including the Cross of the Liberation, the Legion of Honor, and the Croix de Guerre. Upon his return to Paris, he married Lesley Blanch, a British writer and former features editor at Vogue. He also entered the diplomatic corps, serving first with the French embassy in Bulgaria, then Moscow, then Switzerland. Later, he became first secretary of the French delegation to the United Nations, as well as the French Consul General in Los Angeles. The postwar years were an exciting time and would launch the amazing Romain Gary in earnest. (In 1951, it became his legal name.) He would satisfy his mother's great expectations after all.\n\nOnly one essential source of happiness was missing: his mother. Romain had returned home from the war to learn that she had been dead for more than three years. How was that possible? He had received a steady flow of letters from her all along. That's because just a few days before she died, Nina had written more than two hundred short, undated letters to her son and sent them to a friend in Switzerland, with instructions to forward them to Romain at regular intervals. And so, as far as he'd known during combat, his mother had been there for him, sending constant words of love and support. The last letter he'd received ended, \"Be tough, be strong. Mama.\"\n\nIn 1945, the year after he was married, A European Education appeared in print to great acclaim and won the Prix des Critiques. The author and journalist Joseph Kessel raved, \"In the last ten years, ever since we heard the names of Malraux and Saint-Exup\u00e9ry, there has not been a novel in French fed by a talent as deep, new, and brilliant as this one.\" Raymond Queneau declared Gary's debut a triumph, with \"such a particular and original tone.\" Jean-Paul Sartre considered it possibly the finest novel about the Resistance. Gary received an admiring letter from Albert Camus. And in reviewing the American edition, published in 1960, the New York Times noted, \"He can forge a great conception with all the incandescence of a romance novelist\u2014then give it final definition by tempering it in sad irony.\"\n\nThis new toast of the literary world, thirty-one years old in 1945, was on his way to becoming what he'd always wanted: rich and famous, and one of France's most prominent authors. Ultimately, his success would kill him.\n\nThe year 1956 brought the publication of his fifth novel, the 443-page Les racines du ciel (The Roots of Heaven), along with France's premier literary award, the Prix Goncourt. (The eleven-member jury included Maurice Blanchot and Jean Paulhan.) In truth, the prize was a mixed blessing. For Gary, winning meant a surge in sales, of the kind that only the imprimatur of Oprah Winfrey's book club can inspire today. Along with that, however, came the need to address the demands of promotion and celebrity, while also managing a confused identity and increased self-doubt. He and other winners over the years found themselves derailed by what they had most coveted. One winner of the prize referred to the \"GP\" as his \"General Paralysis.\" Another, Jean Carri\u00e8re (who won in 1972), expressed a similar sentiment. \"After having believed that one was writing for a couple hundred or thousand readers, one finds oneself in front of an arena packed with spectators who gasp every time they spot a sign of failure\u2014or the renewal of the artist's exploit,\" he wrote. \"It is enough to paralyze your pen and call into doubt the slightest word traced by your hand.\" Following his award, Carri\u00e8re expressed a sense of resigned duty toward the writing that had once been his passion, saying he felt as though his identity had been hijacked and \"a puppet was bearing it in my place.\" His disillusion worsened over time, to the point, he said, that \"names strike me as fraudulent.\" Still, Carri\u00e8re would survive his plunge into depression and feelings of profound alienation. Romain Gary would not.\n\nBetween his celebrated debut and his fifth novel, a strange thing had happened: the adored Romain Gary had been neglected by his public. The novels published after A European Education were not well received, and Gary found his career stalled. He sent a despairing letter to his publisher. \"I know full well that the public has forgotten me,\" he complained. \"I will have passed like a dream. It's horrible. Sometimes when I look back and see my brilliant beginning and what I am today, a knot forms in my throat.\"\n\nThe Roots of Heaven marked a triumphant comeback. The author whom everyone had once celebrated was again relevant. Gary was no fool; he knew that he had to capitalize on his resurgence. He committed to hundreds of media appearances, and in interviews, he enjoyed inventing amusing and outlandish anecdotes about himself (including a story about his seduction of Clark Gable's girlfriend in a London bar). He treated the \"truth\" behind his authorial persona like a piece of taffy, something to be stretched and pulled.\n\nOne writer noted that \"[Gary's] legend as a charmer is not overblown.\" It worked in his private life, too. In 1959, he met the Iowa-born film actress Jean Seberg. She was twenty-one; he was forty-five. Nine months later, Seberg divorced her husband, Fran\u00e7ois Moreuil. Gary divorced Lesley Blanch in 1961. (She lived to the age of 102, dying shortly before her birthday in 2007.)\n\nIn 1962, Gary and Seberg had a son, Alexandre Diego (known as Diego), and they subsequently married, but Gary would lie about the order in which these events occurred, transposing them so that marriage came first, and even falsifying his son's birth certificate.\n\nFollowing the publication of his memoir, S. ou l'esp\u00e9rance de vie, in 2009, Diego recalled his father in an interview. \"Even when he was present,\" he told Paris Match, \"my father was not there. Obsessed by his work, he greeted me, but he was elsewhere.\" Today, Diego maintains his father's literary estate and tends to his legacy.\n\nSeberg and Gary were a glamorous couple whose social whirl included dining with the Kennedys and spending time with famous actors. But the marriage collapsed in 1970. Its failure could be attributed in part to an affair Seberg is said to have had with Clint Eastwood, and another with a college student (while she and Gary were separated). That relationship resulted in a daughter. Seberg and Gary were divorced by the end of the year, yet they remained extremely close. They jointly filed a lawsuit against Newsweek, which, along with other publications, had alleged that the father of Seberg's daughter was a Black Panther, a cousin of Malcolm X. The stress from this gossip led Seberg to attempt suicide and to give birth prematurely. The baby died two days later.\n\nThroughout the 1960s, Gary published a number of books, but he also acquired new credentials as a director and screenwriter. (Both of the films he directed starred Seberg.) Others adapted Gary's books for the screen as well, and these productions involved some big names. John Huston directed The Roots of Heaven, starring Errol Flynn, Trevor Howard, and Orson Welles. The film The Man Who Understood Women starred Henry Fonda and Leslie Caron. Peter Ustinov directed Paul Newman, Sophia Loren, and David Niven in Lady L. Charlotte Rampling appeared in an adaptation of The Ski Bum.\n\nGary's amazing feat of self-invention now seemed complete. This Russian Jew turned Frenchman was a war hero, a diplomat, a renowned and widely translated author, and a film director, and for eight years he had been the husband of a young and beautiful Hollywood actress. He owned residences in Paris, Majorca, and Switzerland, and on the French Riviera. He was fluent in Russian, Polish, French, and English, and knew some German, Bulgarian, Arabic, and Hebrew. He was a legend of his own making, and against all odds, he had pulled it off. Even though his reputation as a writer had waned somewhat in the 1960s, he still seemed to lead a rather enviable life. His story should end there, it seems, but instead it starts anew. This is where things get really interesting, and deeply sad.\n\nWith all the gaps in biographical information\u2014and all the misinformation\u2014concerning Romain Gary, it is difficult to assemble a comprehensive narrative of his entire life, though biographers in recent years have tried. One fact, however, is well established: at a certain point, Roman Kacew no longer wished to be Romain Gary. Feeling as though he'd been typecast, he reached an impasse. So he became someone else.\n\nIn January 1974, the French publisher \u00c9ditions Gallimard received a manuscript called La solitude du python \u00e0 Paris. It arrived in an envelope that appeared to have been sent from Brazil, by a French businessman on behalf of his friend. Eventually the publisher passed on it, but sent it along to Mercure de France, a division of Gallimard. The novel, later called Gros-C\u00e2lin (also known as Cuddles) was published that year. It told the story of a lonely IBM employee who lived in a Paris apartment with his pet python. The author was \u00c9mile Ajar. It was an immediate best seller.\n\nOnly a select group knew that Ajar was Gary: his typist, his son, Seberg, his attorneys in Geneva and New York, and a longtime friend. They carefully protected his secret. Once, when he was young, Diego watched a show on television in which a critic mercilessly trashed the work of Romain Gary. She then exclaimed, \"Ah! Ajar\u2014now there's a talent of a quite different order!\" The boy glanced toward his father and slyly winked at him.\n\nIt's unclear how Gary arrived at his nom de plume, but some speculate that \"\u00c9mile\" was derived from the bastard child of Gauguin, whom Gary had fictionalized in a novel. \"Ajar\" is Russian for \"glowing embers\" and was also the acronym for a Jewish veterans group.\n\nWhen Gros-C\u00e2lin was short-listed for the Renaudot Prize, Gary found himself in an ethical quandary. The prize was intended for the first novel by a new, undiscovered talent. Not wanting to deprive a young writer of a significant prize, Ajar withdrew his work from consideration. This honorable act merely fed the flame of public interest, and Gary quickly enlisted a cousin, Paul Pavlowitch, to play the \"real-life\" role of Ajar. Now people could put a face to the mysterious author (or so they thought). \"Ajar\" had his photograph taken and even gave interviews. \"It was a new birth,\" Gary admitted later. \"I was renewing myself. Everything was being given me one more time.\"\n\nBefore the birth of Ajar, Gary had already begun planning a second act. Initially he'd considered a kind of performance art ruse, in which an old friend named Sacha Kardo-Sesso\u00ebf would sign his name to detective fiction that Gary had written. His friend declined, as did another, so the role-playing idea was tossed. Instead, Gary produced a trial run for Ajar, under a different guise. In the spring of 1974, a spy novel called Les t\u00eates de St\u00e9phanie, by Shatan Bogat, was published. This unknown author was praised by critics for writing \"with the stroke of a master.\" The press release featured a detailed (and peculiar) biography: \"Thirty-nine years old, son of a Turkish immigrant, Shatan Bogat was born in Oregon. He directs a fishing and shipping business in the Indian Ocean and Persian Gulf. The black market arms trade inspired one of his novels. He won the Dakkan Prize in 1970 for his coverage of international gold and weapons traffickers.\" The prize did not actually exist, nor any earlier novels, but no one had bothered to verify the information.\n\nThe critics loved Bogat. One reviewer said that the author's style was \"100% American, both explosive and relaxed, but with an appreciation of the Persian Gulf's local color that is not from the eye of a tourist.\"\n\nUnfortunately, sales were sluggish. The publisher, Robert Gallimard, decided to out the author in a radio interview, hoping the news might provide a much-needed sales boost. He revealed that \"Bogat\" was actually Romain Gary. In a later edition of the novel, Gary explained his use of a pseudonym in that instance: \"I did it because I sometimes feel the need to change identities, to break free of myself, if only for the duration of a book.\"\n\nIf Gallimard had not exposed the hoax, would anyone have discovered the author's identity? Perhaps not. Journalists can be a lazy bunch.\n\nIn any case, now Gary was ready to become Ajar.\n\nHis alter ego was an Algerian immigrant, born in 1940, and a former medical student who, after performing an illegal abortion, had fled to Brazil, where he now lived. Some critics were suspicious about Ajar's identity, wondering whether an eminent figure such as Raymond Queneau or Jean Paulhan might have taken a pseudonym. Yet as the scholar Ralph Schoolcraft notes in his fascinating 2002 study, Romain Gary: The Man Who Sold His Shadow (the first major examination in English of Gary's life and work), the author had left plenty of clues that Ajar was a mask. Anyone who poked around enough would have found evidence linking Ajar's work to Gary's own novels. (Gary later admitted that Ajar's books \"often contained the same sentences, the same turns of phrase, the same human beings.\") Yet no one picked up on the trail of crumbs. With Ajar, Gary was trying to shed the influence of the literary establishment of which he was now a familiar member. \"I was an author who was classified, catalogued, taken for granted,\" he later complained. Ajar opened the door to experimentation and novelty, and to another new start for his career. Critics would have to approach the work from a fresh perspective because Ajar was an unknown quantity, free of baggage.\n\nFollowing the success of Gros-C\u00e2lin, in 1975 Ajar published a second novel, La vie devant soi (also known as Madame Rosa). A reviewer in Le Monde proclaimed it \"a Les Mis\u00e9rables for the twentieth century.\" The novel explored the relationship between an orphaned Arab boy named Momo and Madame Rosa, a heavyset sixty-eight-year-old Auschwitz survivor who was once a \"lady of the night.\" (A film adaptation was released in 1968.) With this work, Ajar's reputation was assured. The first printing of fifty thousand copies sold out quickly and the book became a best seller. The author could count Marlene Dietrich among his fans. Today, the novel remains the top-selling French novel of the twentieth century, with more than a million copies sold.\n\nAlthough some suspected that Ajar was a pseudonym, no one associated it with Gary. In news accounts, Gary's name had been mentioned, but simply as another example of a pseudonymous author. Some were convinced that Ajar was a Lebanese terrorist; others believed that the eccentric author was an American; still others said that the work was the product of a clandestine collective. And once, Gary met a woman who claimed to have had an affair with Ajar. \"He was a terrific fucker,\" she said.\n\nEventually, this mystery would prove to be the most scandalous event in the French literary world since the publication of Pauline R\u00e9age's Histoire d'O. One half-joking theory was that the savvy culprit behind Ajar was R\u00e9age's illegitimate son.\n\nAll was mere fun and games until La vie devant soi won the 1975 Goncourt. Because Paul Pavlowitch had done such a fine job selling himself as Ajar, the jury members had all they needed to see that the author was real, that they were not being played for fools. Satisfied that Ajar really did exist, they awarded the deserving young author his prize.\n\nThis event was no happy accident. Gary had worked tirelessly behind the scenes, managing his accommodating cousin like a puppet. Pavlowitch eventually gave in-person interviews, but first he had to trick Ajar's own publishing house, Mercure de France, into believing, beyond any doubt, that Ajar was flesh and blood. It was an absolutely brilliant scheme. As Ralph Schoolcraft recounts in his book:\n\nGary then prepared a couple meetings, plotting out Pavlowitch's role in minute detail. The impersonation would be something of a high-wire act, for Pavlowitch had to improvise his demeanor and remarks within the boundaries of Gary's prearranged script. Pavlowitch began by sending Mercure de France a blurry photograph of himself for promotional use (the photo, taken years earlier in Guadeloupe, had the advantage of showing him prior to the growth of the bushy, long hair and extravagant moustache that he was sporting in 1975).\n\nPavlowitch-as-Ajar even signed the publishing contracts and collected a check in person. He went so far as to enlist his wife, Annie, to play the role of Ajar's girlfriend. When the head of his publishing house wanted to spend more time with Ajar, a weekend together in Copenhagen was arranged, which went off without a hitch. During that weekend, Pavlowitch autographed a stack of \"his\" books as a favor to the publisher. He dutifully personalized his inscriptions, just as she requested, addressing them to members of various prize juries, including the Goncourt.\n\nAs Gary himself would explain later, the politics behind the Goncourt were rather heated, and authors had to make nice to become literary darlings. It was a highly rarefied and incestuous world. \"I am not the only person to have spoken of the 'literary terror,' of the coteries, of the cliques with their claques, of cronyism, of 'you scratch my back and I'll scratch yours,' of debts repaid and accounts settled,\" he wrote. \"Outside Paris there is no trace of that pathetic little will to power.\" The back-scratching was exhausting and humiliating, and after a while Gary had come to detest his critics and the phoniness of his milieu: \"I developed a profound disgust of publishing anything.\"\n\nPleased with the success of the encounters he'd concocted for Ajar, Gary upped the demands on his cousin, who complied with each new directive. Personal information was given to the press, but not too much; and with his unkempt appearance and slouchy demeanor, Pavlowitch had no trouble passing as a bohemian writer in exile. His performance wasn't always flawless (he occasionally got minor details wrong), but the public was so eager to embrace \"Ajar\" that discrepancies went unnoticed. After a while, he and Gary could simply sit back and enjoy the fruits of their labors. Journalists did all the rest. \"As soon as it became public,\" Pavlowitch later revealed, \"it no longer depended on us.\" When a reporter once suggested to Gary the similarities between his and Ajar's work, Gary replied that he was flattered, and that perhaps Ajar was guilty of plagiarism.\n\nFor some factions in the literary world, the selection of Ajar for the Goncourt was highly controversial, and the usual protests took an especially ugly turn that year. There were bomb threats. Gary, growing nervous, attempted to heed the advice of one of his lawyers, who'd urged him to have \"Ajar\" decline the prize as a magnanimous gesture. Recusing himself, it turned out, was not Gary's choice to make. The Goncourt jury issued a terse, huffy, unambiguous statement, announcing that \"the Academy votes for a book, not a candidate. The Goncourt Prize cannot be accepted or refused any more than birth or death. Mr. Ajar remains the laureate.\" And that was that.\n\nThe problem? An author can be awarded the Goncourt only once. Romain Gary had already won. That he could (secretly) win again gave his ego a significant boost and confirmed that, at sixty-one, he was still an important cultural figure\u2014even if under the cloak of someone else. He'd shown that his talent was still intact. To throw people off the scent, Gary provided a friendly but neutral comment in support of Ajar. \"I liked Gross-C\u00e2lin,\" he said, \"but I haven't read Madame Rosa yet. I don't think the author will stay in hiding much longer.\"\n\nHe was right. Events took another bizarre twist, though, when more than one reporter tracked \"Ajar\" to Pavlowitch's home, and even uncovered Pavlowitch's relation to Gary. But instead of recognizing that Pavlowitch was a proxy for Gary, who was the real man behind Ajar, the press assumed that the bold Pavlowitch had acted alone\u2014and that the has-been Gary must have envied his relative's turn in the spotlight. Rather than attempt to seize control of this narrative, Gary and his cousin embraced it. Pavlowitch took the hit, crafting a story about how he'd adopted the Ajar pseudonym to launch his own career independently, so as not to exploit Gary's celebrity. This story made Pavlowitch a sympathetic figure and drew attention away from Gary. Meanwhile, Gary cheered on his cousin from the sidelines, joking to the media that there was no way he could have found the time to write Pavlowitch's books as well as his own\u2014and encouraging the literary world to accept the talented Pavlowitch into its fold. He responded angrily to a journalist who persisted in suggesting that Gary himself, not Pavlowitch, was Ajar. \"Your maneuver consists of cutting the balls off a newcomer by attributing his work to me,\" he said, \"all the while protecting yourself with a 'maybe.' Even by Parisian standards, this is truly low.\"\n\nGary was beginning to come undone, increasingly unable to deal with the pressure of keeping up his fabricated self. Determined to put a definitive end to lingering guessing games, he sat down to write. In a state of almost manic fury\u2014just two weeks after the (false) revelation that Pavlowitch was Ajar\u2014in his \"Geneva hideout,\" Gary finished another manuscript.\n\nEntitled Pseudo and published in December 1976, the book purported to be a complete, uncensored confession of the entire Ajar affair. It sold modestly. Written as a novel, it was nonetheless meant to be interpreted as autobiography. The narrator was a madman telling his story from a psychiatric ward, but many of the events and motivations he described were true. (They were, however, told in a highly distorted form, and ascribed to the wrong person.) To the world, it seemed that the story of Ajar had at last been unraveled by the man himself. This should be the end of the story, but it isn't. Not quite.\n\nThere was one glitch: Pseudo was presented as the confession of Paul Pavlowitch, not Romain Gary. This is a confusing twist, but Gary had largely told the truth about his own story, providing many accurate details\u2014he had simply attached the wrong name to it. Some of the issues he \"revealed\" as belonging to Pavlowitch\/Ajar were invented, but others were actually his own. (There's mention of a doctor telling the author that he masturbates too much, a colorful anecdote that may or may not have been true.) In this way, Gary was able to seek redemption and at the same time deny his identity. He'd told a story that was at once fictional and true. He even inserted himself into the novel as a character called Uncle Bogey. The Princeton University scholar David Bellos, who translated Pseudo for the 2010 American edition (as Hocus Bogus), called the book \"one of the most alarmingly effective mystifications in all literature. . . . Almost every sentence of the book is a double take.\"\n\nIn Pseudo, Ajar-as-Pavlowitch describes being pressured to adopt a pseudonym:\n\nPublish! It'll be good for you. Use a nom de plume. And don't worry! Nobody will guess you could do it. If it's any good, they'll say it's got art and technique and that it can't have been done by a beginner. That it's the work of a real pro. They'll leave you alone. They'll say you're just a straw man or a ghost. Or a whore.\n\nBecause Gary was unable or unwilling to speak as himself, he hid yet again, like a coward, behind his cousin. (He phoned Pavlowitch to tell him what he'd done only after it was completed.) At his cousin's expense, Gary had cleared his own name for good, and presto! Mystery solved. Ajar was Pavlowitch, who was a lunatic.\n\nBut this time Pavlowitch was not a willing accomplice. He'd loved his cousin dearly\u2014and felt grateful that Gary had paid for his education at Harvard\u2014but now his devotion reached a breaking point. He felt used and discarded. The neurotic, paranoid, delusional \"narrator\" of Pseudo had been presented under Pavlowitch's actual name, and this was unforgivable. He worried that his reputation might be harmed beyond repair, and he had his wife to consider as well. To Pavlowitch, this book seemed an aggressive and repugnant act. The rift between the men did not heal, as Gary showed little remorse toward his cousin, no gratitude for all that Pavlowitch had done on his behalf, and no real grasp of the perilous implications of Pseudo.\n\nNor does the story end there. Gary wrote yet another confession, but this one he gave only to friends, with the assumption that it should be released posthumously. (It was.) Titled Vie et mort d'\u00c9mile Ajar (The Life and Death of \u00c9mile Ajar), the piece explained his motivations and frustrations. \"The truth is that I was profoundly affected by the oldest protean temptation of man: that of multiplicity,\" he wrote. \"A craving for life in all its forms and possibilities, which every flavor tasted merely deepened. . . . As I was simultaneously publishing other novels under the name of Romain Gary, the duality was perfect.\"\n\nHe signed off the piece, dated March 21, 1979: \"I've had a lot of fun. Good-bye, and thank you.\"\n\nUpon the release of this text, the French literary establishment was outraged. They perceived Gary's doubleness as mockery directed at them, an attack on the very institutions that had crowned him, and they retaliated as they saw fit. Indeed, Gary's posthumous reputation would suffer as a result. Most of his books are out of print in the United States, some have never been translated into English, and those that are available are not easy to find. (In France, however, his books have never gone out of print.)\n\nA few months after the author had written the true confession of his Ajar pseudonym, he was shattered by devastating news: Jean Seberg was dead. On August 30, 1979, having gone missing for eleven days, she was found on the backseat of her car. Her death at age forty was an apparent suicide\u2014a verdict some still consider questionable\u2014caused by an overdose of alcohol and barbiturates. Gary was inconsolable.\n\nOn December 2, 1980, in the Paris apartment where he lived alone, he shot himself in the head. He left behind a suicide note in ninety-six words.\n\nFor the press\u2014\n\nNothing to do with Jean Seberg. Devotees of the broken heart are requested to look elsewhere. Obviously it could be blamed on a nervous depression. But if so it would be one which I've had since I became a man and which enabled me to succeed in my literary work. But why, then? Perhaps you should look for the answer in the title of my autobiographical book The Night Will Be Calm and in the last words of my last novel: \"There's no better way to say it, I have expressed myself completely.\"\nShe was bipolar and sexually confused\n\nChapter 13\n\nJames Tiptree, Jr. & ALICE SHELDON\n\nOn May 19, 1987, a seventy-one-year-old woman and her eighty-four-year-old husband were found lying in bed together, hand in hand, dead of gunshot wounds, at their home in McLean, Virginia.\n\nJust before midnight, the woman had phoned a family attorney to warn him that she planned to kill her husband and herself. She calmly asked that he notify the police. When the officers arrived at the house, they found the couple alive, concluded that the situation was under control, and left. Two hours later, the woman phoned the lawyer to tell him that she had killed her husband. Again she asked him to summon the police. Then she called her husband's son and said that she had shot his father. Although she claimed that she and her husband had agreed in advance upon a suicide pact, she had waited until he fell asleep to kill him. At about 3:30 in the morning, she shot herself in the head.\n\nThis event marked the tragic and dramatic end to the lives of Huntington Sheldon and his wife, Alice Bradley Sheldon. It was sick and scandalous, like something out of a gothic novel. In fact, Alice had been a wildly imaginative writer, intensely driven, producing science fiction for more than a decade using a male pseudonym. She kept her alter ego a secret even from those closest to her. (\"At last I have what every child wants, a real secret life . . . nobody else's damn secret but MINE,\" she wrote in her diary in 1970.) Assuming this guise gave her the confidence to write and allowed her to become the \"son\" she believed her father had always wished he'd had. It also freed her to explore another deeply buried self\u2014one that harbored a shameful yet undeniable sexual desire for women.\n\nAside from becoming famous\u2014and considered among the most important science-fiction authors of the twentieth century, along with writers such as Philip K. Dick\u2014Alice Sheldon led many extraordinary lives. She was an exceptional painter, a brilliant storyteller, and passionately interested in science; she had eloped at age nineteen, become pregnant, and had an abortion in her first year of marriage; divorced, enlisted in the army, and worked for the CIA; she had become a poultry farmer; and she had earned an undergraduate degree at age forty-three, followed by a Ph.D. in experimental psychology. Literary success came later still.\n\nBorn in Chicago in 1915, Alice Hastings Bradley (later known as Alli) was the only child of charismatic, wealthy, glamorous, and eccentric parents. Her formidable mother, Mary, was a prolific travel and fiction author and a popular lecturer; her attorney father, Herbert, was also an explorer and hunter who led expeditions into unmapped regions of central Africa. Those trips into the Congo provided Mary with material for two children's books. Yet Mary did not just accompany her husband on African hunting expeditions; she carried her own rifle and killed lions and tigers herself, proudly bringing back the skins as souvenirs. The Bradleys, both Republicans, were often featured in the society and gossip pages of local newspapers. They had a large circle of friends; loved to give parties; and employed nannies, a chauffeur, and a cook. Alice was pampered and spoiled, but she was also lonely and never felt comfortable in her affluent surroundings. \"I was unpopular,\" she once complained, \"except with dull adults.\"\n\nWhen Alice was four, her mother gave birth to another daughter, Rosemary, who lived for only a day. Mary never recovered from her grief. \"She didn't provide a model for me,\" Alice wrote later of her mother. \"She provided an impossibility.\" The barrier was, among other things, vocational, but above all, Mary's idealization of femininity left Alice anguished, her identity a blur. As Alice struggled throughout her life to achieve a sense of wholeness, to feel at peace with her gender, Mary projected confidence, accomplishment, and uncomplicated, effortless sensuality.\n\nAlice's artistic inclinations were encouraged\u2014but only so far. Mary's needs and her desire for attention (especially from male admirers) always came first. \"She had emotion enough for 10,\" Alli wrote of her mother, \"but I got it all, and was always\u2014perhaps wrongly\u2014aware that had the others existed she wouldn't have cared much for me.\" Later she would recall that having a mother who seemed to do everything well was \"bad for a daughter because you identify with her. And without meaning to, you compete. And to be in competition with Mary was devastation, because anything I could do she could do ten times as well.\" Alice's father was cool and distant, a welcome contrast to her mother's dependent, possessive behavior.\n\nMary's emotional neediness was too much pressure for a child to bear. Alice felt compelled to be a compliant \"good girl,\" managing her own anxieties, anger, and unhappiness so as not to upset her mother. Although Alice suspected that \"everybody wants to wipe the world out a couple of times a day,\" she kept such notions to herself. Decades later she admitted that she'd lived with \"a silent inner terror\" of not succeeding enough to warrant her parents' praise. \"[A]ll my early life was lanced with that fear; if I wasn't somehow Somebody, it would represent such a failure I'd have to kill myself to keep my parents from knowing how I'd betrayed their hopes.\"\n\nShe also had to suppress her intense dislike of her own name, which carried \"joyless connotations of 'Alice, eat your spinach.' 'Alice, go to bed.'\" As she discovered early on, there was power, and a thrilling sense of escape, in naming yourself, in reclaiming your identity. She fled the unpleasantness of daily life through books. Alice was an avid reader and especially loved Kipling. Later she insisted that everything she knew about writing stories and plotting \"came from Kipling, and will probably end there.\" Eventually (following the unmasking of her pseudonym), Alice gave an interview in which she pointedly quoted the end of his poem \"The Appeal\":\n\nAnd for the little, little span\n\nThe dead are borne in mind,\n\nSeek not to question other than\n\nThe books I leave behind.\n\nAs a child, Alice also enjoyed reading science fiction, including H. P. Lovecraft and a pulp magazine called Weird Tales. It was exactly the kind of literary material that her mother would have found vulgar and unseemly, and this made her love it even more.\n\nAlice was sent to boarding school in Switzerland, where she did her best to fit in, but she was socially awkward, moody, and a poor student. She made her first suicide attempt there, cutting herself with razor blades. Lonely and struggling with what would be a lifelong battle with depression\u2014fifty years later, Alice was diagnosed as bipolar\u2014she was desperate to return home. But her father wrote to her that \"it would not be fair to your school, nor to us, nor to you to come home in the middle of the year.\" Convinced that the challenge of an academic experience abroad would build character, Herbert urged her not to give up. In any case, he gave her no choice. \"I'll trust you to be a good sport and see it through like a little lady,\" he wrote in another letter. Mary, who deplored candid displays of emotion as much as her husband did, wrote to Alice cheerily, \"You are taking life the right way, darling, if you keep jolly and keep going\u2014that's all any of us can do.\" Eventually Alice attended a small boarding school in New York, and even though she felt happier there, the headmistress observed astutely that \"[t]he task of adjusting herself to her contemporaries is not an easy one.\"\n\nAs her sexuality developed, Alice felt ambivalent toward other girls. In some ways she preferred the company of boys, who seemed much more straightforward emotionally. Her relations with them were flirty, easy, and fun. In the presence of girls, Alice often felt annoyed by their frivolous, superficial behavior and their hierarchical approach to friendship, yet she felt strongly attracted to them as well. Girls turned her on. They excited her in ways she found deeply unsettling, but she did not pursue her feelings beyond a few fumbling encounters. The passion she felt was unrequited, anyway, and remained so: her desire for women would never be fulfilled (at least, not as far as anyone knows; it has never been confirmed that Alice had any affairs with women). Sexual love provoked frustration and torment, but nothing more. The only coming out Alice experienced was as a debutante, in 1934, when she was nineteen years old.\n\nThat was the year she met a wealthy twenty-one-year-old Princeton student, Bill Davey, who was a guest at her debutante party. Alice was still an undergraduate at Sarah Lawrence College. They eloped almost immediately\u2014the wedding was front-page news in the Chicago Tribune\u2014and Mary coldly informed Alice that she had broken her father's heart. The marriage lasted just six years. \"He was beautiful, he was charming, he was a poet, he had references from the deans at Princeton,\" Alice would recall years later, \"but they forgot to mention that he was an alcoholic and supporting half the whores in Trenton.\"\n\nInitially, getting married seemed to promise Alice a chance to liberate herself from her parents, and from her sense of inertia and sexual confusion. It would prove that she was a \"normal\" heterosexual woman fulfilling what was expected of her. But this marriage was hell. Bill was as mercurial as his wife, who also drank too much. Both of them slept with other people. Their fights were often physically violent, and their reconciliations were short-lived. The sex was mutually unsatisfying. (Alice described it in her journal as \"a mechanical farce.\") She was uncomfortable with her own body, and quite miserable having sex with a man. \"Oh god pity me I am born damned they say it is ego in me I know it is man all I want is man's life,\" she wrote in a notebook five years before her death, \"my damned oh my damned body how can I escape it. . . . I am going crazy, thank god for liquor.\" It was not surprising that Alice's gender dysphoria would lead her to inhabit a male self so that she could feel in control as an author. Even after her second marriage, to Huntington Sheldon, her struggles with sex and sexuality continued. \"I am (was) notoriously fucked up about sex,\" she once admitted in a letter to a friend.\n\nMeanwhile, during her tempestuous first marriage Alice was beginning to find her way as an artist, though as a painter, not as a writer. She started to show her work and was included in a group exhibition at the Art Institute of Chicago. For the next five years, she toiled away at her paintings while struggling through her moribund marriage. \"Happy is the person who has never loved another,\" she wrote in her journal in 1941. Alice became convinced that she was constitutionally incapable of intimacy, and realized that she had to end her marriage. That summer, she left Bill. He promptly filed for divorce and remarried within a few months.\n\nThe collapse of this relationship ended her ambitions as a painter, too. Although she knew that she was talented, Alice felt certain that her true vocation lay elsewhere. She offered a harsh self-assessment of her potential as a visual artist: \"I was a good grade B, no more, only with a quickness at new tricks which made ignorant souls call me an A.\" She decided to invest her intellect and energy in writing instead. Her parents helped her get a job as an art critic for the recently launched Chicago Sun, where she earned sixty dollars a week. She didn't especially like journalism, but she knew she had to start somewhere.\n\nIn 1942, when the controversial Women's Army Auxiliary Corps (later the Women's Army Corps) was created by Congress, Alice decided to join. She wanted to serve her country, move toward a different kind of career, and feel useful and accomplished. She also wanted to put more distance between herself and her failed marriage, and to cultivate more structure and discipline in her life. The pretty twenty-seven-year-old arrived at the recruiting office in \"three-inch heels and my little chartreuse crepe-de-chine designer thing by Claire somebody, and my pale fox fur jacket.\" When she showed up for basic training in Des Moines, Iowa, she marveled at the sight of women \"seen for the first time at ease, unselfconscious, swaggering or thoughtful, sizing everything up openly, businesslike, all personalities all unbending and unafraid.\" At the time, it was the most exciting experience of her life to be surrounded by twelve thousand women. \"What a range!\" she later marveled.\n\nEventually she went to the Pentagon, where she did intelligence work during World War II, and spent the next few years having affairs with men. She seemed to have resigned herself to the fact that her romantic future, however imperfect, inauthentic, or unsatisfying, would be with a man. And she spent her spare time writing fiction. Her efforts, filled with autobiographical elements, fell flat. (\"'Ouch' simply is not a story,\" she wrote years later, in a letter to a friend.) It would take the authority and secrecy of a male pseudonym, and the genre of science fiction, to transform her pain, anguish, and desire into compelling material.\n\nStationed in London in 1945, Alice met the man with whom she would spend the rest of her life: Huntington \"Ting\" Sheldon, a forty-two-year-old army colonel who had been a Wall Street banker. He fell in love with her\u2014hard. Ting came from the \"right\" social class; like Alice, he'd found a sense of purpose during wartime and had used military service to escape the confines of his past. He was born in Greenwich, Connecticut, to a family that had earned a fortune in banking and lost it in the Great Depression. He attended boarding school at Eton and university at Yale. Though Ting was a calm, steady, dependable presence, he was not without baggage\u2014he'd already been married and divorced twice and had three children.\n\nBut now Ting wanted to marry her, and as a thirty-year-old woman, Alice felt she was in no position to refuse. She wanted children; she wanted to feel cared for and secure. For the most part, Ting proved a supportive, easygoing partner who gave her space when she needed it. He also put up with her mood swings. But there were problems. Like her first husband, Ting drank a lot. He was emotionally distant and did not share her love of reading. And their sex life was terrible. A year later, Alice's literary agent, Harold Ober, submitted a short story she'd written to the New Yorker, and it appeared in the magazine on November 16, 1946. \"The Lucky Ones\" was the first and last piece she would publish under her own name. Nor would she ever submit another story to the New Yorker. She had been unhappy with the intensive editing process, complaining that \"it was astounding how they edited me into New Yorkerese,\" and she found the magazine as a whole too polished and genteel.\n\nDespite her family pedigree, Alice was anything but polished and genteel. Among the multitudes within her were, she said, \"a female wolf who howls, and a gross-bodied workman who moves things and sweats, and a thin rat-jawed person who is afraid and snaps . . . [and] a disastrous comedian who every so often comes roaring out of the wings and collapses the show. Now it seems clear that while one might get one or two of these characters to write for a living, most of them won't go along, and the comedian's opinion is unprintable.\"\n\nHer impressive publishing accomplishment at such a young age notwithstanding, the next several years hardly indicated that Alice was on her way to becoming a famous writer\u2014one who, in the words of Isaac Asimov decades later, \"has produced works of the first magnitude and has won the wild adulation of innumerable readers.\" In fact she seemed about as far away as possible from a literary life. She felt lost. Depressed by her lack of sexual chemistry with Ting and her ambivalence toward their marriage, she tried to leave him at one point. \"What shall I do?\" she wrote in a letter to her husband, announcing her departure. \"Lie and deceive, put on a bold face and knock the bottom out of everything? Drift in this void and try to work? I cannot hold the beast that is me in check much longer.\" But she didn't leave, and apparently never even gave him the letter. (Eventually, the couple agreed on an open marriage.) Alice abandoned her attempts at journalism, having experienced little success at selling pieces as a freelancer. Ting, too, was adrift in his work, unable to secure a new job on Wall Street.\n\nAfter seeing an ad in the New York Times offering a chicken hatchery for sale in New Jersey\u2014with promises of high income and working only half the year\u2014the Sheldons impulsively decided to buy the business, which they ran for nearly five years. The work was hard and the routine dull, but at first the rigid structure of their days was good for Alice. When she realized, however, that she could not conceive a child, she was devastated and began spending the little free time she had writing both poetry and prose, including the beginning of a mystery novel and some science fiction. In knowing that she would never become a mother, she felt betrayed by her own body. She decided to confront this issue in an essay, asserting that a woman's body was an \"unpredictable, volcanic, treacherous, merry, rather overpowering thing to live with.\" She likened her body to \"a large and only partly tamed animal, day and night the damn thing is being itself, with its own semi-inscrutable operations.\"\n\nThe characteristics of her gender\u2014punishing and restrictive, yet wildly untamable\u2014left her feeling repeatedly \"derailed\" in life, and she described being a woman as an almost debilitating condition, and certainly a steep disadvantage. She argued (rather reductively, even for the era) that if she had been born male, she might have been more aggressive and could have become \"a rather prosy young engineer or research scientist,\" married with children. \"Instead of which, I was born a girl,\" she wrote, \"and my life has been quite different. . . . I have had about four different and disparate careers. I have been married twice. I have seriously upset a great many of the people who came close to me. . . . I have been called brilliant, beautiful, neurotic, suicidal, restless, amoral, anarchic, dangerous, diffuse, weak, strong, perverse, and just plain nuts.\" It seemed to Alice that the impossible fact of living as a woman was enough to make anyone despondent or crazy. She devised no solutions to her profound quandary, but she did extol the virtues of \"a great deal more homosexual activity on the part of women.\" Rather than adhering to binary notions of gender, Alice proposed five: men, women, children, mothers, and \"human beings.\" Unsure of where or how she fit into her own odd schema, she concluded wearily that it was perhaps best \"in most of the waking hours of a non-pregnant woman to consider her a kind of man.\"\n\nIn 1952, when Alice was thirty-seven, she and Ting turned to their former military and government contacts, sold the (woefully unprofitable) hatchery business, and moved to Washington, D.C., to work as analysts for the CIA. Ting worked in high-level intelligence positions for the next seventeen years, whereas Alice's career was low-ranking, much to her frustration, and lasted only a few years. As a woman in a male-dominated agency, she stood no chance of having a powerful or well-paying job, but as someone who placed a high value on secrecy and privacy, she felt entirely comfortable in an environment that promoted covertness as policy. \"I always had a feeling there were big things going on in her life that she would share with nobody,\" one friend recalled of Alice. \"She could have been living three or four lives at once.\"\n\nAlice was by nature flirtatious, but at the CIA she remained sartorially gender-neutral, a look she found appealing. \"Boyish clothes look younger, or healthier,\" she noted in an unfinished memoir in 1957, \"because they contrast a woman's features with a man's, rather than with a girl's. In a clean white shirt I still look like a perverse young boy, and this is about my best effect, from the standpoint of attraction.\"\n\nIn 1955, Alice was in her third year at the CIA, on the verge of turning forty and deeply unhappy. Her mood swings were even more pronounced. She became addicted to prescription pills and at times felt suicidally depressed. Like many women, she felt conflicted about the gap between society's demands on women and her own desires. Which should she reject, and at what cost? She felt that as a writer, she had nothing important to say\u2014or at least nothing that would be heard. She was expected to be a devoted wife and a faithful, hardworking CIA operative; with whatever energy was left, she could attempt to write. The most obvious effects of such strong pressure were her increased hostility toward Ting and her general sense of inertia. \"O, how I want to be loved, me myself\u2014\" she confided in a letter that year, \"\u2014and how I fear it\u2014and what bliss it might be\u2014brrr!\u2014and how easy to shelve this whole thing.\" That summer, she quit her job, left Ting, rented an apartment, and \"really destroyed all traces of my former personality.\"\n\nSelf-creation and reinvention are deeply and quintessentially American notions (e.g., The Great Gatsby), and their appeal was not lost on Alice. A full decade before she would assume her pseudonym and launch her literary career, she felt the lure of inhabiting another identity. Alice was trapped in her nondescript life, and simply wanted to be someone else. For a start, she wanted the freedom of divorce and solitude. \"I figure I have enough sub-personalities so I can build one up to where it is quite companionable,\" she wrote. She was convinced that \"[t]here is no way I can be peacefully happy in this society and in this skin. I am committed to Uneasy Street.\"\n\nAfter a year of aimless soul-searching, she returned to her husband, having resolved nothing. \"So ensued a period of more milling (I'm a slow type) including some dabblings in academe,\" she later recalled.\n\nBecause she was unable to commit herself wholly to one enterprise, she accepted Ting as an essential and permanent part of her life and struggled to find fulfillment elsewhere. Her inability to give herself full time to writing was partly due to her profound ambivalence toward the task itself. She questioned its value and believed that writing was \"an act of aggression.\" It was a betrayal, selfish, an act of exploitation. As Joan Didion famously noted, \"Writers are always selling somebody out.\" Janet Malcolm, too, has described even journalism as \"morally indefensible,\" and has characterized the journalist as \"a kind of confidence man, preying on people's vanity, ignorance, or loneliness, gaining their trust and betraying them without remorse.\"\n\nAlice decided to go to back to school to complete her undergraduate degree, which she earned, summa cum laude, at the age of forty-three. A friend and mentor at the time advised her that \"the greatest favor you can do to others is being yourself as much as you can,\" but Alice was still grappling with what that meant. \"Being, I imagine, must be very simple,\" she wrote back. \"It is Becoming which is so messy and which I am all for.\"\n\nIn 1959, Ting and Alice moved to McLean, Virginia. Eight years later, Alice earned her doctoral degree in psychology at George Washington University. Her mother's physical health had severely declined\u2014perhaps freeing Alice creatively\u2014and she herself had survived another long period of depression. \"Too much motor for the chassis,\" she noted of her emotional and mental vulnerability. She was burned out in every sense and on the verge of physical collapse. Amphetamines, cigarettes, and coffee were her sustenance. Just as she was completing her dissertation (and perhaps realizing that academia was too confining for her ambitions, and too boring), Alice began writing fiction again. Sci-fi authors such as Samuel R. Delaney, Ursula K. Le Guin, and J. G. Ballard were gaining prominence. Soon she would take her place among them.\n\nMidlife is often said to be a period of reinvention, and Alice, with typical intensity, accomplished this to an extreme degree. At the age of fifty-two, she abandoned her role as a research scientist and scholar and began submitting her stories to science-fiction magazines. These were hardly highbrow literary publications, but she wasn't aiming for her work to appear in the New Yorker again. \"I have a modest view of my talent. I haven't the ear for rhythm or the feel for style to encourage me to compete in the serious mainstream,\" she later admitted. \"And I certainly haven't the stomach to write 'mainstream' schlock, like Jaws or Gone with the Wind. Science fiction suits me just right. SF is the literature of ideas, and I am, I think, an idea writer.\"\n\nStarting out, Alice wasn't fearless enough to submit her writings under her own name. Anonymity seemed best. \"I am a reclusive type, afraid of meeting people, except on paper,\" she once admitted. A fateful trip to the supermarket with her husband in 1967 provided inspiration. Spotting a jar of Wilkin & Sons marmalade, she was struck by the label: \"Tiptree,\" in a distinctive cursive print. (The name came from the English village near which Wilkin & Sons owned farmland and orchards.) For the impulsive Alice, it held the key to her new identity. \"James Tiptree,\" she said to Ting. \"Junior,\" he replied, without missing a beat. They laughed, but the name stuck and an author was born. Alice had intended to use a different pseudonym for each short story she submitted to magazines, but as it happened, Tiptree had such a rapid rise to success that she kept him.\n\nIn a biographical sketch written more than a decade later for Contemporary Authors, Alice offered a cursory description of her bold postdoctoral transformation: \"At this point a heart problem forced temporary retirement at semester's end. Meanwhile, some SF stories written as a hobby were all selling, to the author's immense surprise. As health returned, the temptation to write more won out. The author rationalized this activity as a claim for a broader concept of 'science' than rocketry and engineering, and the aim of showing SF readers that there are sciences other than physics, that bio-ethology or behavioral psychology, for instance, could be exploited to enrich the SF field.\"\n\nShe continued: \"But this writing had to be kept secret; the news that a new PhD with offbeat ideas was writing science fiction would have wakened prejudice enough to imperil any grant and destroy my credibility. . . . Luckily, the challenge of writing exerted its spell; retirement from university work became permanent without any great traumas, and the author found herself with a new line of effort ready-made for somewhat erratic health. . . . The first SF stories were naturally not expected to sell, so a pseudonym was selected at random (from a jam pot).\"\n\nBy the time of that entry, her secret identity had been exposed for three years and \"James Tiptree, Jr.\" was already buried.\n\nShe had enjoyed enviable success as Tiptree, however. On some level, the experience must have been bittersweet: only after she inhabited the role of a male author did she achieve fame. As herself, just another woman writer, no one had paid much attention. \"I have this childish fascination with brute power,\" she admitted in an essay, written in her post-Tiptree years. \"And since I have none, I am nothing.\" As only herself, Alice felt oppressed by a sense of powerlessness and believed that her \"authentic\" self lay elsewhere. \" 'I' am not a writer,\" she wrote in her diary. \" 'I' am what is left over from J.T. Jr., a mindless human female who 'lives' from day to day.\" Interestingly, in the late 1960s, women writers in increasing numbers had taken up science fiction and fantasy, and although this was a male-dominated field by any measure, it was not impossible for a woman to become successful in the genre. Alice did not see herself among them, but there were women she admired who did just that, such as Le Guin, a contented housewife and mother living a \"conventional\" life in Portland, Oregon; and Joanna Russ, an outspoken feminist best known for her award-winning, stylistically inventive novel The Female Man. Russ was also a lesbian, and this was not without its complications for her writing career\u2014yet somehow it gave her permission to work freely, beyond the standard definitions of gender. Neither of those models would prove a comfortable fit for Alice, but Le Guin and Russ became two of Tiptree's favorite correspondents.\n\n\"Becoming\" a man had seduced Alice partly because she believed it gave her access to power\u2014and the possibilities that accompany power. \"Alli Sheldon has no such choice,\" she lamented, and imagined what life might have been like if she had been born a boy. No matter how accomplished she was as a woman\u2014and she was extraordinarily so, however hard on herself she was\u2014Alice never felt relief from what she viewed as the constraints of her gender. \"Always draining us is the reality of our inescapable commitment,\" she wrote, arguing that it is only women who \"feel always the tug toward empathy, toward caring, cherishing, building-up\u2014the dull interminable mission of creating, nourishing, protecting, civilizing\u2014maintaining the very race. At bottom is always the bitter knowledge that all else is boys' play\u2014and that this boys' play rules the world.\" Of course, for a woman of her generation, the prospect of being a writer hardly carried the same stigmas and constraints as it had for nineteenth-century iconoclasts such as the Bront\u00ebs, George Sand, and George Eliot. But the \"giants\" of literature were men. And to become a major sci-fi writer, a woman within a cloistered subculture, she might as well have been living in the previous century. To her, this realm truly seemed an impenetrable boys' club. Things aren't nearly so dire now, but it remains a male-dominated field (less so as a result of her pioneering efforts).\n\nAlice was airing her concerns about gender imbalance in an era of so-called second-wave feminism, but rather than accept herself as a passive victim, she never stopped pushing back. She never gave up. \"Maybe all one can do is to say the hell with it,\" she wrote. \"But\u2014life is to use. Only, how? How? How? How?\"\n\nWriting under the cloak of Tiptree, she soon achieved success. She described once how \"this letter from Cond\u00e9 Nast (who the hell was Cond\u00e9 Nast?) turns up in a carton. Being a compulsive, I opened it. Check.\" Her story \"Birth of a Salesman\" had been sold to the sci-fi magazine Analog; the story \"Fault\" was bought by another editor for twenty-five dollars. Within a few weeks, a third sold. Letters and checks were addressed to \"James Tiptree, Jr.,\" so her alter ego began to seem like a real person, separate from her. He became a card-carrying member of the Science Fiction and Fantasy Writers of America (SFWA). He even had a nickname\u2014he insisted on being called \"Tip.\" And he enjoyed flirting in his correspondence with women. (Tip complimented one editor's assistant by calling her a \"superdoll.\") Some women developed crushes on him in return. One editor invited Tip to his wedding; of course, Tip had no choice but to decline. He mentored aspiring sci-fi writers\u2014by mail, of course\u2014and wrote fan letters to fellow authors he admired or envied, including Italo Calvino, Anthony Burgess, and Philip K. Dick. He was generous with praise for his fellow writers. \"Who do I admire in SF?\" he once wrote. \"You and you and you as far as eye and memory reach, sir and madam. Some for this, some for that. All different. But more than that\u2014I love the SF world. And I don't love easy.\"\n\nWhen editors asked to meet Tiptree in person, they were given lame excuses. (One editor tried to call him, only to find that Tiptree was not listed in the phone book.) In retrospect, it seems incredible that the ruse was so easy to pull off. But it worked, so Alice simply kept going. Even though she regarded her early sci-fi stories as \"mechanical and banal,\" they were selling, and the act of writing proved a pleasant diversion from the episodes of crushing depression that came on without warning. Yet her two selves were at odds: the charming Tiptree longed to connect, to find acceptance and kinship, to establish a sense of community in the sci-fi realm. He was witty, generous, and kind, a great raconteur, and always supportive of the endeavors and ambitions of his peers. But Alice was forced to act as his vigilant sentry, rejecting intimacy, withholding information, keeping outsiders at arm's length to protect her colossal secret. This internal clash between concealment and revelation was confusing and often painful to bear. \"I've lived so deep under masks, my interior was built to satisfy me alone,\" she wrote in a letter five years before her death. \"I have lived 60 years almost totally alone, mentally, and quite content to have it so. I'm fond of a hundred people who no more know 'me' than they know the landscape of Antarctica.\"\n\nAlthough other science-fiction writers were secretive, they rarely hid from both editors and readers. Tiptree was especially reclusive and protective of his privacy, which only encouraged rumors about his motivation. Once, a group of curious fans, attending a local sci-fi convention, staked out Tiptree's P.O. box in McLean. Luckily Alice was in Canada at the time.\n\nReaders wondered whether Tiptree was very young, Native American, secretly gay, or working undercover for the CIA. One editor wrote, \"It has been suggested that Tiptree is female, a theory that I find absurd, for there is to me something ineluctably masculine about Tiptree's writing. I don't think the novels of Jane Austen could have been written by a man nor the stories of Ernest Hemingway by a woman, and in the same way I believe the author of the James Tiptree stories is male.\"\n\nAnyone who attempted to extract biographical information met with resistance, aside from learning broad, generic facts. To Le Guin, with whom Tiptree's epistolary friendship endured even after his pseudonymous cover was blown, he once described himself as \"an old battered Airedale, one-eyed and droop-eared, whose scarred paws have travelled a lifetime of lava plains.\" That's about as descriptive as he got about his appearance. He did, however, once venture so far as to send a \"baby picture\" to one of his correspondents\u2014actually a photograph of Alice Sheldon at age one, in which she might easily have been mistaken for a boy.\n\nIn her extensive correspondence, Alice carelessly offered many of her own life experiences as Tip's, rather than making them up entirely\u2014a misstep that would lead to the downfall of her alias. For instance, Tip told people that he was born in the \"Chicago area,\" had traveled around colonial India and Africa as a child, joined the army, and had \"some dabblings in academe.\" When Alice's elderly mother was ill, Tip described the burden of caretaking as his own. \"At the moment I'm in and around the Chicago area, partly attending to family matters in the shape of an aged and ornery mother,\" he revealed in correspondence.\n\nPerhaps Alice's inability or unwillingness to create an entirely fictional background\u2014familial or professional\u2014for Tiptree indicated that on some level she hoped someone would discover her secret and that she would be made whole\u2014freed from the burdens of duality. But for a while, no one did. And because Tiptree had no voice or body for others to know, people gave free rein to their fantasies about him. One person imagined him as Ichabod Crane\u2013like. Another believed him to be exceptionally handsome. One fanzine publisher wrote to Tiptree with his take on the author's physicality: \"You like wild shirts and ties. You smoke a pipe. You type fast and grin a lot.\"\n\nWhenever pressed about personal matters, Tiptree either ignored the queries or pushed back. \"Does a writer ever stop telling you who he is?\" he wrote in an interview conducted by mail with the editor Jeffrey Smith, arguing that an author's work should speak for itself and that it told readers everything they needed to know. \"[M]aybe I believe . . . that the story is the realest part of the storyteller. Who cares about the color of Coleridge's socks? (Answer, Mrs. C.) Of course, I enjoy reading a writer's autobiography\u2014or rather, some writers! A few. By far the most of them make me nervous, like watching a stoned friend driving a crowded expressway. For Chrissakes, stop!\" He also insisted that \"my mundane life is so uninteresting that it would discredit my stories.\" (Well, that was not exactly true.) Tiptree did reveal that \"part of my secretiveness is nothing more than childish glee.\"\n\nHe suggested that one way to inhabit authorial identity was to use the \"self as an experience laboratory, no sacred wall around the sealed black box of Me.\" In other words, it was merely a play space. Regardless, he believed (or claimed to believe) that an author's \"real\" self \"leaks at every sentence,\" so that attempts to shield biographical details from the public were futile, anyway. Justifying his motives a dozen different ways, Tiptree remained defiant. \"You know as well as I do we all go around in disguise,\" he wrote, describing each person as a \"roomful\" of human beings. Beneath our everyday decorum, he argued, were layers of ugly and messy emotions, including terror, rage, obsession, love, and shame. \"So who the fuck cares whether the mask is one or two millimeters thick?\"\n\nTip did a good job most of the time at maintaining his own mask\u2014a little macho posturing here, a little raunchy joking there\u2014but it wasn't always a flawless performance. \"Do you know, there's a good deal about you that seems to me more like women I know than like men I know in the way you handle your feelings?\" Joanna Russ wrote to him.\n\nHe kept people intrigued by his brilliant talent as well as his demand for absolute privacy. Some claimed that his reticence was a put-on, a \"publicity trick,\" as Alice later wrote. Curiosity about him continued to grow along with his reputation, perhaps because he defied categorizing in every sense. As Jeffrey Smith (who would become Tiptree's literary executor) noted, \"What I was most interested in was the fact that in 1970, when there was a virtual war declared between the Old Wave and the New Wave in science fiction, Tiptree was being claimed by both camps.\"\n\n\"It's futile to ask as new a writer as me where he's tending or what his style might become,\" Tiptree wrote in response to a question from Smith. \"Does a kid whose voice is changing know what's going to come out of his mouth?\"\n\nFor whatever reason, Tiptree was rather expansive in his correspondence with Smith, and developed an unusual closeness and trust with him. Smith, who was respectful without being sycophantic, seems to have impressed Tiptree with astute interpretations of his work. \"I'm beginning to feel like this was my last will and personal Time Capsule and it contains more on Tiptree than anybody including me will ever likely see or want to again,\" Tiptree confided to Smith in the final letter of their interview by mail, which went back and forth from December 3, 1970, until the end of January 1971.\n\nThe following year, having inhabited Tiptree for half a decade, Alice Sheldon began to feel constrained by writing as a man. She wanted to express her \"feminine\" voice, yet she wasn't willing to unmask herself entirely. She did the next best thing: Alice introduced Raccoona Sheldon, another alter ego. What a perfect name: raccoons, after all, are mask-wearing bandits, stealthy and clever.\n\nIt was actually Tiptree who announced the arrival of Raccoona\u2014an old friend of his from Wisconsin\u2014to Smith, mentioning that she was a gifted writer. Alice took just as much care of her female pseudonym as she had taken of the enigmatic Tiptree, buying Raccoona her own Olivetti typewriter\u2014this one with a black ribbon to distinguish it from Tiptree's blue ribbon. Raccoona was given a distinct handwriting and signature, and a mailbox in her name at the post office.\n\nYet just as Alice would slip up in covering Tiptree's tracks, here, too, she was somewhat sloppy. For one thing, she'd given Raccoona her own surname. And supposedly Raccoona had also been published in the New Yorker. She was a talented illustrator, had dabbled in academia, and she'd had an abortion. She described herself as a former East Coast resident and a retired schoolteacher, but insisted that \"really the less said the better\" when it came to talking about her personal life. She had in common with Tiptree the requisite elusiveness, existing entirely on the page. (Raccoona's stories didn't pack the same wallop as Tiptree's work, however; he was by far the better writer.) Tiptree emphasized her shroud of mystery to Smith, warning him that his friend \"is even more recessive than me and hard to talk to.\" When another editor accepted one of Raccoona's stories for publication, he was puzzled to receive no reply from the author. Eventually, she wrote apologetically to explain that her mother \"had a heart attack down South.\" Soon afterward, an \"embarrassed\" Tiptree dashed off a letter about his friend as well: \"I can't imagine what happened to Sheldon (Raccoona), unless she's been abducted by aliens . . . [It's possible] some of her multitudinous parasitic family has her tied up.\"\n\nRaccoona had mixed success in getting her stories published, and better luck only when her pal Tiptree wrote cover letters of recommendation on her behalf. It is amusing to note that Raccoona felt exasperated and jealous that Tiptree\u2014a man, of course\u2014was getting his stories published by the same editors who were rejecting her work\u2014with Alice being the dutiful midwife to them both. By this point, Alice wished, in a sense, that Tiptree were dead, but killing him off wasn't an option. The strain of maintaining relationships solely by mail was getting to her. \"They're real,\" she wrote privately of the friendships she'd cultivated in the science-fiction world, \"yet unreal insofar as they're carried on under an assumed name and gender. A lot of genuine relation comes through, but it's tainted to an unknown degree by falsity. Here I seem to have contrived another odd trap for myself.\" Why she set these traps is impossible to say, but surely the destructive messages lingering from her childhood had a lot to do with submerging herself in other selves.\n\nIn 1974, at least one of Tiptree's friends rightly sensed something amiss in his letters, though it was hard to know how to respond. \"Is my friend whom I know and do not know troubled beyond all touch or reassurance?\" wrote a worried Le Guin. \"Is he in trouble? Is there nothing his friends whom he knows and does not know can do, or say, or be? Nothing that would help?\"\n\nAlice's ambivalence toward her male alter ego had started to affect her ability to play the role. She was tired and lonely. She asked Ting to lock her prescription pills in the medicine cabinet because she feared she would overdose. And at her lowest depths, she fantasized about killing Ting and then herself, but she wasn't yet able to go through with it.\n\nTiptree strained her nerves more than ever. He seemed pointless, this man named after a jar of jam at the supermarket. The following year, Alice described herself as descending into a \"black pit\" and admitted, \"I personally am dying.\" As if to force Tiptree to fade away, Raccoona pointedly downplayed her relationship with him in a letter to Smith: \"There seems to be some confusion about me and Tip Tiptree,\" she wrote. \"Several people have written me as though I were an authority on him. I did know him when we were in the local 4th and 5th grades together, but I have not seen him in person for a couple of years.\" She continued: \"We correspond in fits and starts. I take care of his mail when he comes through here to see his mother.\"\n\nIn November 1976, Tiptree sent Smith a letter as intimate and confessional as a diary entry: \"Mother died last week,\" he wrote, \"leaving me with a new dark strange place in the heart, and flashes of a lively, beautiful, intelligent, adventurous red haired young woman whom I had once known.\" The subsequent biographical details about Tiptree's mother, unfortunately, were too specific\u2014they included where Tip's parents had lived for sixty-four years (\"Father built the building and they took the whole top and made the first roof garden in Chicago\")\u2014and too similar to the newspaper obituaries of Alice's mother. It was already well known that Tiptree's mother (like Mary) had been an African explorer, hardly a typical biographical detail. Tip seemed to recognize that he'd spilled too much personal information. He ended by saying, \"Well, this is a weird letter.\" It was. Yet he mailed it anyway.\n\nThe author had (inadvertently? deliberately?) laid out all the clues that would link Tiptree to Alice. It's no wonder: she was exhausted, anxious, and in very bad shape, despite Ting's efforts at managing her moods. She was hooked on prescription pills, including Percodan, Dexedrine, Valium, and Demerol. As for Tiptree, he'd become like one of the distorted figures in Francis Bacon's paintings\u2014tortured and grotesque. The charade had run its course.\n\nThe outpouring of fact and emotion in Tiptree's letter was not lost on Smith. Nonetheless, he felt highly protective of the dear friend he'd never met or even spoken with on the phone. He didn't want Tip's cover blown, and didn't want to pry, but he couldn't resist investigating whether Tip's revealing missive was indeed a \"road map to a newspaper obituary,\" as he recalled later. His research didn't take long: the first Chicago newspaper he found at the library, a copy of the Tribune, led him to the death notice of ninety-four-year-old Mary Hastings Bradley, who was survived by one child, a daughter. The obituary, aside from a minor element or two, matched the details of Tiptree's letter. How to reconcile \"Uncle Tip\" with the posh Alice Hastings Bradley Sheldon?\n\nIn her 1980 biographical sketch for Contemporary Authors, in a section she titled \"The Pseudonym That Got Away,\" Alice wrote that when \"the author's mother died after a long illness . . . Tiptree\u2014who wrote only the truth in all letters\u2014had imparted so many of the details of Mary Bradley's unusual life that when her obituary was read by certain sharp-eyed young friends, James Tiptree, Jr., was blown for good\u2014leaving an elderly lady in McLean, VA, as his only astral contact.\"\n\nTo ease the aftermath of Alice's broken secret, Smith opened up to Tiptree first. In a gently honest letter, ever respectful of his friend's privacy, he wrote that he was not making \"a demand for information,\" but warned, \"I am going to be getting questions, and whatever you choose to disclose or withhold from me, please pass along the Party Line that I'm supposed to tell others.\"\n\nHe received a response\u2014not from Tiptree, but from Alice Sheldon, who introduced herself. She asked that Smith keep her secret for a bit longer. He agreed. \"How great,\" Alice wrote, but she was relieved beyond measure that the consuming role was no more. Her reply was casual: \"Yeah. Alice Sheldon. Five ft 8, 61 yrs, remains of a good-looking girl vaguely visible, grins a lot in a depressed way, very active in spurts. Also,\" she added, \"Raccoona.\"\n\nTo the very end, however, Alice insisted that there was no such thing as \"male\" or \"female\" writing. Instead, she believed there were only separate and varying styles of bad writing, and whether a weak voice belonged to a man or a woman was beside the point. She allowed that men perhaps had the edge when it came to black humor, and women had a knack for \"heart-wringing,\" which was an odd statement of gender stereotyping by someone whose writing career had defied such notions.\n\nIt is intriguing that as Tip, she displayed a certain swagger, while Raccoona was more diffident and a less compelling writer. Alice admitted in an essay (\"A Woman Writing Science Fiction,\" written six months before her suicide) that she didn't feel proud of using a male pseudonym to get ahead. She happened to choose a man's name as a lark, and stuck with it only because it worked so seamlessly. Frankly, she kept exploiting it because of the superior treatment she received as a man: her work was taken seriously, she was well regarded by the women with whom she corresponded as their \"understanding\" and empathetic male friend, and she occupied a place of power and influence among her peers\u2014allowing her to challenge editors to publish more women writers. Alice said that she was ashamed of using a male guise to earn her place, while other women writers had languished or succeeded entirely on their own terms. \"I had taken the easy path,\" she admitted.\n\nAs she began to make amends for her ruse, the responses she received were almost entirely supportive. \"Dear Jim or Tip or Alice or Allie,\" one friend addressed her in a letter, reassuring her, \"You are still the same person and I am still the same person and here we are.\" Ursula Le Guin was similarly kind: \"And it is absolutely a delight, a joy, for some reason, to be truly absolutely flatfootedly surprised\u2014it's like a Christmas present!\" Joanna Russ, upon learning that Tip was a woman, didn't suppress her delight at the news, admitting that she liked \"old women,\" expressing hope that they could meet \"in the flesh\"\u2014they never would\u2014and telling Alice bluntly that she should consider herself \"well and truly propositioned. I was in love with you when you were 'James Tiptree Jr.' and have been able to transfer the infatuation to Allie Sheldon.\" Eventually, Alice declared that she was a lesbian in a letter to Russ, but she took things no further with Russ or any other woman.\n\nAlice was crushed to find that some of the male writers she'd considered true friends\u2014those who had ostensibly admired her work as Tiptree\u2014turned their backs on her. (\"Oh, how well we know and love that pretentiously amiable tone, beneath which hides the furtive nastiness!\" she wrote.) She was heartbroken that some men were suddenly patronizing and condescending toward her, or that they abandoned her altogether. \"If that is how I would have been received from the start,\" she wrote, \"my hat is off to those brave women writing as women.\"\n\nIn her \"Woman Writing Science Fiction\" essay, she couldn't resist a dig at her erstwhile \"friends.\" Noting that some of the male writers who'd been \"a touch snotty\" to her were perfectly nice to other women writers, she went straight to the core of the problem: \"People dislike being fooled, and, quite innocently, I did fool them for ten years. Moreover, it seems to be very important, especially to men, to know the sex of the person they are dealing with. What's the use of being Number One in a field of two\u2014i.e., male\u2014if people can't tell the difference? I had not only fooled them, I had robbed them of relative status.\" Apparently, they felt emasculated, something they didn't find funny or even forgivable.\n\nAfter the initial dizzying rush of revealing her true identity, Alice became severely depressed again. (Rightly so: being exposed meant that a part of her was now dead.) It was something like the shattering remorse that sets in after a breakup. Alice had gotten rid of this troublesome character, and now she wanted him back. Like an ex-lover, Alice could remember only the good that Tip had brought into her life; he had made her a celebrated science-fiction author and given her a supportive community, the likes of which she had never known. Without him, she felt crazy and unable to write.\n\nIn her journals, Alice detailed her sense of deprivation. The language she used was like that of someone wanting a sex change: \"I do not 'match' my exterior.\" She wrote of feeling as if she inhabited her body like an alien and even yearned explicitly to become a man. Within her, too, remained a fervent desire to someday love a woman erotically as a woman: not to resort to sublimation, as she always had done, but to satisfy raw urges. This pull was profoundly disorienting, and the sudden limbo\u2014for both her professional and her personal identity\u2014intensified her self-hatred.\n\n\"Some inner gate is shut,\" she wrote. The revelation was terrifying. She was left with nowhere to go, no way out. As Tiptree, she'd immersed herself in his unbridled imagination; as Alice B. Sheldon, she noted ruefully that she had no discernible prose style other than \"Enclosed please find payment.\" She was convinced that no one wanted to know her simply as Alice, and she called herself a \"poor substitute\" for Tip. Although she toyed with the idea of another pseudonym, Sylvester Mule, nothing came of it.\n\nIn an interview for Contemporary Authors (which would accompany her biographical sketch), Alice expressed her attitude toward separating a writer's work and life, and the damage that results when the latter overshadows the former. She felt this problem was especially acute in science fiction\u2014a genre \"that carries some sense of wonder\"\u2014and said that when \"the camera suddenly pans and picks up the writer himself, he's slouched in a haze of smoke over his typewriter, and it's all come out of his little head. . . . Magic gone.\" She insisted that most writers were obnoxious or dull (never mind that she was neither), and spoke of fa\u00e7ades not in her writing persona, but in daily life. The interview offered plenty of fascinating material. Alice revealed that since she suffered from paralyzing shyness, \"Tiptree's elusiveness was no pose.\" She said that even though she was capable of chatting with people at the grocery store, she had to put on a kind of polite veneer to do it, and \"what no one sees is the cost of the fa\u00e7ade.\" (They would after she killed herself.) She spoke of having done two interviews with \"pleasant strangers\" the previous week, for which she \"couldn't help impersonating Miss Vitality\" (yet another reference to impersonation), but that the moment those interviews had ended, \"I collapsed for the rest of the day in a dark room with a cold rag on my head.\" She wasn't exaggerating. No one but Ting knew the toll that social interaction exacted from her. This was why he often asked friends to keep their visits short or, better yet, not to come at all.\n\nHer contradictory feelings about the loss of Tiptree were unrelenting and painful. In a passage from the original transcript of her Contemporary Authors interview (which she decided to omit in the final version), Alice said that in regard to Tiptree, she would do nothing differently if she had to do it over. Yet she was still shaken by his absence:\n\nI think that Tiptree's death was long overdue. I had considered taking him out and drowning him in the Caribbean, but I knew I couldn't get away with that. It's a little frightening to find oneself almost being possessed by this personality that one isn't or that only one part of one is. It was an extraordinary experience. He had a life of his own. He would do things and he would not do other things, and I didn't have much control over him.\n\nAs Alice felt increasingly dejected after having been outed, she talked openly about wanting to die, telling friends that if Ting's health continued to deteriorate she had no intention of outliving him. She also said that if life got too bad, she'd kill them both. She started seeing a psychiatrist and was taking several antidepressants, but nothing seemed to help. She complained that \"so far nobody will give me what I deepest crave, a lead-nose .38 bullet in the parietal lobe. I dream about oblivion the way other people dream of good sex.\" She would also describe herself with an eerie metaphor to an interviewer in 1982: \"I'm a loaded gun, an achingly loaded gun wholly unable to get a shot at those who are my enemies.\" (Years after her death, one of Alice's editors remembered her as having been \"notable for her jocular and ironic determination to survive in spite of her admitted desire to die.\")\n\nAlice didn't actively attempt suicide, but she took terrible care of herself. She had \"accidents\" that caused injuries, health issues (including open-heart surgery), and for a while she lived on nothing but vanilla custard with frozen raspberries. Although she continued to correspond with some of Tip's friends and kept up with people by telephone, her interactions were undeniably awkward. She knew that and withdrew even further. After starting to write fiction again under her own name, she never achieved Tiptree's magic or even came close. She knew that, too. Maybe her enormous talent would have eventually returned, but Alice didn't live long enough to find out.\n\nToward the end, Ting had a stroke and was partially blind and deaf; Alice's most serious illness was mental. Her suffering had become intolerable. She'd written a suicide pact for them years ago, but at eighty-four years old, despite his frail health, Ting still wanted to live. Alice had been heading toward oblivion for so long that it was impossible to trace the starting point of her fateful decline. She'd anticipated her premature death, hungrily waited for it.\n\nOn May 18, 1987, Alice sent a brief note to Ursula Le Guin, along with a magazine article she thought her friend would find amusing. She signed off, as usual, \"Tip\/Alli.\" There was no hint of the gruesome scene to come in the middle of the night: Ting fell asleep; Alice shot him in the head. Then she wrapped her own head in a towel, held Ting's hand, and shot herself. Proving this event had been a long time coming, she left behind a suicide note dated September 13, 1979. Their bodies were donated to George Washington University's medical school.\n\n\"She had enormous critical success and was very highly thought of by intellectuals,\" Alice's literary agent, Virginia Kidd, told the New York Times after her death. \"But she never made the numbers.\"\nHis mother didn't love him but he was in love with himself\n\nChapter 14\n\nGeorges Simenon & CHRISTIAN BRULLS ET AL.\n\nHe claimed to have had sex with ten thousand women, so it is surprising to learn that communication posed a problem. Clearly, he was able to fulfill his needs. But the challenges of verbal intercourse obsessed him throughout his life, as he revealed in an interview with the Paris Review in 1955. The Belgian author Georges Simenon was asked about the most significant issues he'd dealt with in his fiction, and which themes he expected to contend with in the future. He replied:\n\nOne of them, for example, which will probably haunt me more than any other, is the problem of communication. I mean communication between two people. The fact that we are I don't know how many millions of people, yet communication, complete communication, is completely impossible between two of those people, is to me one of the biggest tragic themes in the world. When I was a young boy I was afraid of it. I would almost scream because of it. It gave me such a sensation of solitude, of loneliness. That is a theme I have taken I don't know how many times. But I know it will come again. Certainly it will come again.\n\nFor someone so acutely aware of the efforts and failures of everyday speech, Simenon seemed to embody a phenomenal will to express himself to the world. How else to explain his voluminous literary output\u2014hundreds of novels, translated into nearly fifty languages? Many of his novels were best sellers; he sold more than 500 million books worldwide. Preposterously prolific, he was capable of producing eighty pages of prose a day, six books a year; somehow he found time to publish more than a thousand articles and short stories as well. He makes Joyce Carol Oates look like Harper Lee.\n\nSimenon, who died in 1989 at the age of eighty-six, was often more famous for his louche ways than for his work. He brought it on himself. Simenon was \"larger than life,\" known for his hubris, self-infatuation, and a capacity for excess that reached astonishing proportions. He never had an agent, choosing instead to oversee all his own publishing contracts, which he did very shrewdly. His kindness and magnanimity, when he cared to display them, were stupendous in equal measure. He began using pseudonyms at age sixteen and published more than two hundred novels using more than two dozen noms des plume. Nearly two hundred other novels were written under his own name, and twenty-one volumes of memoirs. He was itinerant, moving house dozens of times in his life, including a decade-long stretch in the United States, when he lived in Arizona, California, Florida, and Connecticut. He owned a gold watch that a reporter described as \"the size and shape of a brioche.\" He was an international celebrity and the subject of countless flattering magazine and newspaper profiles. \"He Writes a Book in 33 Hours,\" proclaimed one typically hyperbolic headline. \"World's Most Prolific Novelist\" was another.\n\nMost of the anecdotes he told about his life were false\u2014they were fantasies he spun to amuse himself and impress (or confuse) others. He was a legend in his own mind. This was a man as intoxicated by himself as others are by fine wine. But he liked wine, too\u2014also, champagne, whiskey, and beer, even while he wrote. On the advice of his doctor, he restricted himself to two bottles of red Bordeaux daily. (He did go through periods of renouncing alcohol for Coca-Cola.) One friend recalled a common sight: Simenon throwing up a bottle's worth of cognac in the garden, \"two fingers down his throat, after he finished a chapter.\"\n\nHe told an interviewer that he had become \"hungry for all women\" at age thirteen. That was apparent in the vast number of his sexual conquests\u2014ten thousand was perhaps a conservative estimate\u2014most of whom were paid. (He was more often a customer than someone's lover.) Allegedly, Simenon liked to make love several times a day, which would put his stamina right up there with that of Warren Beatty, Wilt Chamberlain, and other reputedly record-breaking sex fiends. He once said that he suffered physical pain at the thought of so many women in the world with whom he would never get to have sex. \"I would have liked to have known all females,\" he said. Simenon married and divorced twice\u2014the first time, at age twenty; the second time, a day after the dissolution of his first marriage\u2014and was an incorrigible philanderer. He was never boring.\n\nGeorges Joseph Christian Simenon was born in Li\u00e8ge, Belgium, on Friday, February 13, 1903. Even his birth involved an act of deceit: his superstitious mother insisted that the date be recorded, falsely, as February 12. When his grandmother saw him for the first time, she is said to have exclaimed to her daughter-in-law, \"My God, Henriette, what an ugly baby!\"\n\nAlthough Georges worshipped his father, D\u00e9sir\u00e9, an insurance clerk, he regarded his domineering, high-strung mother with contempt, and in his later writings, he savaged her mercilessly. Their relationship wasn't helped by her obvious and unabashed preference for his younger brother, Christian, and her blatant disdain for Georges. She adored Christian and always referred to him as \"my son\"; Georges, however, was \"le fils de D\u00e9sir\u00e9.\" Henriette exacerbated Georges's resentment of his younger brother and his bitterness toward the mother he perceived as rejecting him. He acted out in a number of ways, which had the effect not of gaining Henriette's sympathy, as he desperately wished, but of provoking her ire; she found him annoying and peculiar. His parents' marriage was unhappy, too. D\u00e9sir\u00e9 died at age forty-four of a heart attack in 1921, when Georges was eighteen years old. Just as his mother's withholding behavior would mark him for life\u2014and surely influence his dysfunctional relationships with women, as well as his writing\u2014so would the loss of his father. \"The most important day in a man's life is the day of his father's death,\" he wrote some thirty-five years later. When Henriette remarried in 1929, Georges considered it an act of treachery. Even more galling was that she kept the name Simenon; he was quite famous by that time and resented her exploitation of his celebrity.\n\nAs a child, Georges excelled at school to show his mother that he was no failure, that he was worthy of her love. She was oblivious. He supposedly learned to read at age five, and as a student at a local Catholic school, he was industrious, conscientious, and exceptionally gifted. At age eight, he won a student prize for French composition, earning the praise his mother denied him. By the age of thirteen, the precocious boy was signing his homework using the pseudonym \"Georges Sim,\" just for fun.\n\nYet by 1918, he'd shed his \"good boy\" persona, and his grades suffered as a result. \"I rebelled more or less against the taboos that imprisoned me and also against the mediocrity that surrounded me,\" he told a reporter for Paris Match in 1967.\n\nThanks to his brilliance, he got away with a lot. He mocked authority figures, skipped school, rejected any thought of entering the priesthood\u2014the vocation his mother had pressured him toward\u2014and, finally, dropped out of school. \"I wanted to get laid, and the Church told me I'd be damned for it,\" he once said. \"So I left.\" Had he stuck with it, he might have been expelled. Georges didn't care. He felt he could no longer continue being a mindless slave to any institution, least of all school or religion, and for the rest of his life he would devote himself wholly to two compulsions: sex and writing (not necessarily in that order).\n\nLike nearly every other biographical detail about Simenon, there are multiple versions of the story of how, as a teenager, he landed a newspaper job. Any or all of them may be apocryphal. But it seems that he walked into the offices of the Gazette de Li\u00e8ge and talked his way into a position as a reporter, earning forty-five francs a month to start. His debut was an article about the city's first horse fair since the Armistice, and he managed to impress his editors. Although he'd had no burning ambition to become a journalist, he was getting plenty of practice writing. He loved it. Even better, the deadline-driven, high-pressure environment turned him into a writer who could crank out copy quickly, a habit that would help him become the famous author of hundreds of novels.\n\nAt the Gazette, he resurrected the pseudonym he'd used at school, \"Georges Sim,\" whose byline first appeared in print on January 24, 1919. He happily took on the reporting assignments that no one else at the newspaper wanted, and proved himself a quick study, ambitious, full of energy, and enthusiastic about each new assignment. Soon his editor gave him the crime beat, furnishing him with a paid education that would later serve his detective fiction. In addition to the access he gained to police and criminal matters, he learned a great deal about forensic science.\n\nWithin a few months Georges was also given his own daily column, \"Hors du Poulailler\" (\"From Outside the Hen Coop\"). He signed it with the pseudonym \"M. Le Coq\" (\"Mr. Rooster\"). Whereas Sim was a straight news reporter, Le Coq's tone was funny, cavalier, and snarky. Writing about a criminal trial in 1921, Le Coq described the gathering of journalists in the courtroom: \"They form a small, closed circle which lives very much at its ease. There they sit, sharpening their pencils, munching chocolate, swapping jokes, until suddenly the trial takes an interesting direction and they start to scribble furiously. . . . They frequently break off between sentences to swig from bottles which they have brought into court, right under the judge's nose.\" And at the ripe old age of eighteen, Georges defined a journalist as \"a man who can stay awake at political meetings\" and \"a man who writes a column or two on a subject he knows absolutely nothing about.\"\n\nHe found the world of journalism fascinating, every aspect of it, and perhaps some part of him knew even then that his experiences would prove useful for his fiction writing. As his newspaper articles garnered more attention (a fact that thrilled him), his confidence grew. He knew that he was a real writer and that his ambition and talent extended beyond journalism. He proved it by writing his first book, Au pont des Arches, subtitled \"A short humorous novel of Li\u00e8geois mores.\" The author was Georges Sim. He followed this a few months later with a second novel, which he later admitted had been written while he was quite drunk. Within a year he cowrote a third novel with a friend\u2014a parody of a detective novel.\n\nIn December 1922, Georges resigned from his newspaper job. He was engaged to be married to a painter, R\u00e9gine Renchon; he decided that he disliked her name and rechristened her \"Tigy,\" which stuck. She was no great beauty, but she was strong-willed and intellectual and three years older than he\u2014and the first woman he'd been attracted to who was not a prostitute. They moved to Paris, as Georges knew he must leave Belgium to truly achieve success. Later, Simenon would confess that when he married Tigy he was in love with her sister, but the marriage got off to a promising start anyway. They had a son, whom they named Marc.\n\nThe Simenons felt at home in Paris, where Georges began to submit stories to literary journals and magazines. In 1923, he sent his work to the fiction editor of the daily newspaper Le Matin, who happened to be Colette, already famous for her novel Ch\u00e9ri. She rejected his work again and again, but one day, she encouraged him by saying that he was close to being published, just not quite there. And she offered some unforgettable advice: \"You're too literary. No literature! Get rid of all the literature, and you've got it.\" He finally did; he was published in Le Matin, and felt eternally grateful to Colette for transforming his approach to writing. (He went on to become a regular contributor.) His less-is-more style limited the use of adverbs and adjectives and favored short, clear sentences and brief paragraphs:\n\nThere is not a single light on Quai de l'Aiguillon. Everything is closed. Everyone is asleep. Only the three windows of the Admiral Hotel, on the square where it meets the quay, are still lighted.\n\nOver the next several years, Simenon obsessively honed his craft, trying out different themes and developing his voice. He churned out an absurd number of novels and more than a thousand short stories\u2014all pseudonymously, all pulp fiction\u2014with astonishing economy and efficiency. He would watch movies at night, sleep for a few hours, drink wine, and write and write. Any fear of being \"too literary\" was gone. These short novels were messy, even incoherent, but they were still good stories\u2014lowbrow page-turners intended for popular consumption. (He was thinking in \"chick lit\" terms long before that genre ever existed, describing his early works as \"novels for secretaries.\") They were not works of art, but he had no illusions about that.\n\nIn an interview Simenon gave to the New Yorker in 1945, he described the rigorous routine of his early career: \"Every day was like a prizefight,\" he said. \"My schedule was two hours of work, typing at high speed, followed by an hour of rest or physical exercise. Often my wife would give me a rubdown. Then I would return for another two hours of writing. When evening came, I was depleted.\"\n\nHe admitted that by 1924 he was engaged in \"the careful manufacture of semi-luxurious literary products. I became successful. I had a yellow Chrysler Imperial sedan and a chauffeur who delivered my manuscripts to the publishers and collected my checks. Also, I had a servant to fill my pipes for me. Every morning she would place forty filled pipes on my desk, enough to last me for two hours. I did not have to stop to fill my pipes myself and lose valuable time. After a while I worked more slowly, spending as much as two weeks on a single book.\" Another luxury he enjoyed was what he claimed to be the first private bar in Paris, in his own apartment. He later recalled that after one of his frequent raucous parties, with friends passed out on the floor, \"dawn would find me stepping over the cadavers and making my way to the typewriter.\"\n\nSimenon often boasted about the ease with which he produced books. If he was not ashamed of what he'd written, why had he chosen to write them using multiple pseudonyms? The roster included his old friend Georges Sim; Christian Brulls, a combination of his younger brother's name and his mother's maiden surname; Georges-Martin Georges; Gom Gut; Jean du Perry; Georges d'Isly; Bobette; Plick et Plock; Jacques Dersonne; Germain d'Antibes; and Poum et Zette.\n\nHe may have been a pulp fiction factory, but he didn't necessarily want everyone to know. (As in The Wizard of Oz, the idea was to \"pay no attention to the man behind the curtain.\") Perhaps using so many names allowed him to skip from crime novels to steamy romance novels to adventure novels, and so on, employing as many clich\u00e9s and hackneyed plots as he wished, freely and often hilariously, with no fear of criticism to slow him down. He could write eighty pages a day without breaking a sweat. Because he was in disguise, nothing (nor any dismissive critic) could stop him from exploring his imagination in whatever form or direction he wished. And even here, in what would not unreasonably be called dreck, there were seeds of the glorious Simenon novels to come\u2014including the acclaimed Maigret detective series, which made him one of the best-selling writers in the world\u2014and hints of the author whom Andr\u00e9 Gide called \"the greatest French novelist of our times.\"\n\nIf his writing life was orderly and productive, his personal life was a mess. In 1925, he and Tigy vacationed in Normandy, where he met Henriette Liberge, a local fisherman's daughter whom the Simenons hired as their maid. Just as he'd renamed his wife, Georges started calling Henriette \"Boule.\" His wealth grew along with his writing output, and although he remained as disciplined as ever in his work\u2014Boule woke him at four o'clock each morning with a cup of coffee, and he immediately went to work at his typewriter\u2014his libido was about to wreak havoc.\n\nBoule became Simenon's mistress. But that same year, he saw a nineteen-year-old African-American singer and dancer, Josephine Baker, perform in the show La revue n\u00e8gre. He fell in love. Baker was the toast of Paris, and Simenon was but one of her many lovers and admirers. He was so preoccupied with her that in 1927, his typically manic productivity nearly ceased. His wife seems to have had no inkling of his affair with Baker, even though it consumed his attention. (He and Baker remained lifelong friends.) The following year, he was able to break away from his obsession, at least enough to resume almost his usual output\u2014forty-four novels in 1928. A sense of frustration was beginning to set in; he wanted something more than journalism and pulp novels written under pseudonyms. He had plenty of money now, enough to buy a boat, and then an even larger boat that he had custom-built. Still, he was dissatisfied, maybe because his greatest creation, Inspector Jules Maigret, had yet to be born.\n\nAlways self-mythologizing, Simenon claimed that Maigret came to him a fully formed character one day as he sat in a caf\u00e9. \"I began to picture the powerful, impassive bulk of a gentleman I thought would make a passable inspector,\" he told an interviewer decades later. \"I added various accessories as the day wore on: a pipe, a bowler hat, a thick overcoat with a velvet collar.\" Maigret made his first appearance in 1929's Une ombre dans la nuit (A Shadow in the Night), written under the pseudonym Georges-Martin Georges. In this novel, Maigret is a doctor, and he has only a minor role. It is interesting that Simenon gave the early Maigret a medical profession, as the author frequently mentioned that he might have become a doctor if his writing career had failed.\n\nSimenon published other pulp novels (under different names) that year, some of which featured police inspectors who were essentially composites of the author himself.\n\nThe Maigret character was fleshed out over the course of four novels. It was almost as if Simenon was getting to know his signature character, experimenting with his creation before committing an entire novel to the hard-drinking, pipe-smoking detective. Simenon was starting to realize that he could produce higher-quality fiction, but the slow emergence of Maigret was caused by stubborn resistance from publishers, who weren't sold on the character. They viewed Simenon as a reliable cash cow\u2014and if it ain't broke, why fix it? They didn't want to tamper with a successful formula and had little regard for the author's wish to take his career in a different direction. Nor did they see any need for him to publish under his own name, which he was keen to do. It wasn't enough to be a lucrative and prolific author. Simenon yearned to be admired\u2014and moreover, to take credit for his work.\n\nEven several editors he worked with didn't know his real name. In fact, some believed that \"Georges Simenon\" was Georges Sim's pseudonym. Frustrated by the confusion for which he was responsible, Simenon announced dramatically to a journalist that his days of alter egos were about to end: \"From now on I'm going back to my real name, and I'll sign my books as Georges Simenon.\"\n\nHe was taking a huge risk by exposing his true name and attempting a more ambitious, nuanced writing style\u2014placing greater emphasis on character development and shedding the hackneyed plots of his pulp novels. His Maigret series would tweak the detective genre so that the answer to \"Whodunit?\" was not always wholly resolved, and the unorthodox detective could be counted on for his eccentric, highly unscientific investigative methods and empathy toward criminals. There were no obvious heroes or villains.\n\nEver fond of excess, Simenon decided that he needed a proper party to introduce his new, improved, more literary self. For someone who had worked pseudonymously for so long, he knew how to win publicity when he needed it. \"It's not enough to have talent,\" he told a friend. \"You have to make it known.\" He was hardly shy. In February 1931, he hosted a decadent society ball in his own honor at a Montparnasse nightclub. The savvy Simenon even hired a company to film his guests as they arrived, just like a Hollywood red-carpet premiere. He invited the most glamorous people in Paris\u2014a mix of high-society types, celebrities, journalists, and artists\u2014ensuring that it would be a much-talked-about event. Nearly a thousand people came. The party lasted all night and, like most other things Simenon attempted, it was a smashing success. Although some critics dismissed him as a publicity whore, he now had all the validation he needed to write under his own name. (He did continue publishing other novels under his nom de plume Christian Brulls for the next few years, but then he retired his alter egos.)\n\nWriting as himself did not slow his output; Simenon could easily complete a book a month, or even every few weeks. A New York Times piece once noted that Simenon was a man who \"can write a good novel in the time it takes a fallible human to turn out a passable book review.\" And a Life magazine article by Henry Grunwald pointed out that \"Simenon turns out a book in about the time the average writer needs to draft a single chapter.\"\n\n\"I write fast, because I haven't the brains to write slow,\" Simenon once said.\n\nFor him, writing provided an equilibrium that kept a darker side under control. He couldn't stand being between books. He took long walks, sometimes for hours on end, as ideas percolated in his mind.\n\nHis second wife, Denyse, described the difficulty of living with him during the gestation of each new work: \"Normally a happy person, full of vitality and strength, [he] would suddenly look and act strange, become short-tempered and even morose,\" she said. \"I used to think that I had done something to hurt him. The answer usually came three or four days later, when he would announce to me, 'I am going to start a new book!\"\n\nSimenon did not seek approval from his fellow writers, which was lucky, since he had offended so many by behaving like a pompous ass in his interviews. After all, he was only twenty-nine years old in 1932, and he displayed an arrogance that people felt he had not earned. He boasted about never creating outlines for his manuscripts but simply sitting down at his typewriter and essentially allowing the entire story and all its characters to unfold before him. His muse, it seemed, never took a vacation day or called in sick. Further, he didn't hesitate to reveal that all his novels were written \"in one take,\" with no revisions and \"no touchups or modifications.\"\n\nOne journalist recounted an irritating interview with Simenon. \"I wish I could be anonymous again, walk around unrecognized,\" he told her, rather disingenuously. \"It's terrible, you know, not to be able to go into a bar or restaurant without people elbowing each other and whispering, 'Look! It's Georges Simenon!' They read my books all over the world, you know.\" He also insisted that he had no taste for the great wealth he'd worked so hard to accumulate, even suggesting that he found money tedious. \"If I spend half a million francs a year,\" he said, \"it's only because I have to see the world. I have to know how it feels to lose a fortune in Monte Carlo, or to own a yacht and have a chauffeur. But as soon as I've amassed the material I need, it'll be over with, and I'll go back to a quiet, peaceful, life.\" Never mind that Simenon enjoyed Savile Row suits, custom-made silk shirts, and expensive wines.\n\nHis self-regard was insufferable. \"Provide me with a typewriter and this very instant I would be able to get started on a new book,\" he once boasted, displaying an ego the size of a small nation. \"I am fortunate in that I can write anywhere and under any conditions. I do not need to wait for inspiration. I am always inspired.\"\n\nSimenon argued that he had written his pseudonymous pulp novels to make enough money for writing more \"serious\" books. Yet he didn't want to limit his literary efforts to an elite readership. He said that his goal was \"to write a novel capable of capturing the interest of all audiences.\" Yet he admitted, \"This is not as easy as it sounds: not to repulse the learned while remaining comprehensible to simple folk.\"\n\nBy 1933, Simenon had written nineteen Maigret novels. He felt that he had entered what he called his \"literary period,\" but he was not satisfied with his status. \"When I am 40 I will publish my first real novel,\" he announced in 1937, at the age of thirty-four, \"and by the time I am 45 I will have won the Nobel Prize.\"\n\nIt is amazing that Simenon found time for writing at all: because Tigy supposedly had little need for sex, he cheated on her several times a week, with Boule and other women. Sometimes he was unfaithful several times a day. Most years, he was able to maintain the frenzied pace of his writing; when his life was consumed with additional distractions, his average output was still four novels a year (more than some writers produce in a lifetime).\n\nLong after Simenon resolved to publish books openly as himself, the intensity of his writing process caused him to inhabit other selves, in a manner of speaking. Although he was no longer using other names, he adopted the mannerisms, facial expressions, and gaits of his characters, and used sense memory (such as smells, colors, and sounds) to create settings. \"[While writing my novels] I shall not be myself,\" he once said. \"Of course, I will eat with my family, but I will not be Simenon but someone else.\"\n\nEntering into a trancelike state, diving into his subconscious\u2014these were necessary triggers for the act of creation. He was not inventing stories from his imagination, or from an intellectual place. Essentially, he still had to become someone else to write\u2014if not by using a pseudonym, then by allowing a character's \"self\" to take shape fully, without the author's control or intervention. \"I'm not an intelligent man and I don't have an analytical mind,\" he told a reporter in 1971. \"My books are therefore written by intuition alone. . . . The intuition just comes\u2014on condition that I am, in a sense, completely empty.\"\n\nHe would achieve a neutral mind-set in which his subconscious took over, temporarily abandoning Georges Simenon to discover characters that were waiting to rise to the surface. \"I actually live the part of my characters,\" he said. \"It's no longer I who write, but they.\" At one point in the process, the author would pose a question to yield more information, as he revealed in a 1955 interview: \"Given this man, where he is, his profession, his family, what can happen which will push him to his limit?\"\n\nThe first procedure he used to \"empty\" himself before writing was cleaning his desk, a perfunctory but necessary ritual. \"It's the character who commands, not me,\" he said. His method may have been pretentious (or invented for the sake of a good anecdote), but he claimed that it was the only way his books could be written.\n\n\"All the day I am one of my characters,\" he once said. \"I feel what he feels. The other characters are always seen by him. So it is in this character's skin I have to be. And it's almost unbearable after five or six days. That is one of the reasons my novels are so short; after eleven days I can't\u2014it's impossible. I have to\u2014it's physical. I am too tired.\"\n\nBy 1945, Simenon was still married to his first wife, but the marriage wouldn't last. (Still, he managed to stay close to Tigy for the rest of his life.) Within weeks of moving his family to the United States, he began an affair with a twenty-five-year-old French-Canadian woman, Denyse Ouimet. His fixation on name changing continued, as he promptly changed the spelling of hers to \"Denise.\" Because her former lover's name was Georges, he wanted to be renamed as well, and asked her to call him Jo.\n\nFour years later, she was pregnant with the first of their three children: Jean, Pierre, and Marie-Jo. He divorced Tigy in 1950 and immediately married Denyse. They lived for a time in California, where he met and became friends with Charlie Chaplin.\n\nIn the same random fashion in which he did most things, Simenon moved his family to Lakeville, Connecticut, where he bought an eighteenth-century home on fifty acres. He woke at six each morning and went to work in a soundproofed office, the curtains drawn. Denyse would prepare everything for him before he sat down at his IBM typewriter. He placed a \"Do Not Disturb\" sign\u2014stolen from New York's Plaza Hotel\u2014on the doorknob. His favorite pipes were filled and ready to be smoked, and his stacks of paper, maps, and dictionaries were by his side, as well as the telephone directories from all over the world that he used for naming his characters. In moments of solitary contemplation, he toyed with a monogrammed solid gold ball that Denyse had ordered from Cartier. His dozens of pencils were pre-sharpened daily, and he would switch on a hot plate to keep coffee brewing. He always began by drafting, on the back of a manila envelope, a list of his characters, their addresses and phone numbers, their ages, and other basic information\u2014including places to which they might travel, and possible medical ailments. If his writing \"spell\" was ever broken by some interruption from the outside world, he immediately shut down and discarded whatever he had written until that point. (Interruptions were rare.)\n\nSupposedly he wore the same outfit while writing each novel. For a normal writer, that might seem eccentric, but for Simenon, who could produce a book in a matter of days or a week, wearing the same clothes for the duration wasn't so odd. And he weighed himself before and after completing each new book, so as to measure how much sweat the project had cost him.\n\nSimenon submerged himself completely while writing at his feverish pace, refusing to see anyone or speak on the phone. It was the only way he could work. There is a well-known story (perhaps a joke?) that goes like this:\n\nAlfred Hitchcock once called to speak with the author. Simenon's secretary apologized, explaining that her boss couldn't come to the telephone because he had just started writing a new novel. \"That's all right,\" Hitchcock replied. \"I'll wait.\"\n\nBetween books, Simenon was fully engaged with the people around him. \"I'm a bit like a sponge,\" he once said. \"When I'm not writing I absorb life like water. When I write I squeeze the sponge a little\u2014and out it comes, not water but ink.\"\n\nHe would produce twenty-six novels during his five years in Lakeville.\n\nIt was there, in 1955, that a reporter from the Paris Review came to interview Simenon. The subject was described as \"cheerful, efficient, hospitable, controlled,\" which seemed to be Simenon's manner at all times, unless he was in bed with a woman. In the interview, Simenon provided insight into his revision process, which was brutally efficient and, he claimed, never involved changing the plot in any way. Asked what kinds of cuts he made to his work, he replied, \"Adjectives, adverbs, and every word which is there just for the sentence. You know, you have a beautiful sentence\u2014cut it.\"\n\nHe was just as unsentimental about word choice in general. \"[M]ost of the time I use concrete words,\" he said. \"I try to avoid abstract words, or poetical words, you know, like 'crepuscule,' for example. It is very nice, but it gives nothing.\"\n\nSimenon never had any interest in participating in the \"literary life,\" or even reading the work of his contemporaries. His own masters were dead. \"I should tell any young man who wanted to follow in my footsteps to read the novels of Dickens, Stevenson, Dostoevsky, Balzac, and Daniel Defoe,\" Simenon once told an English journalist. \"Then\u2014forget them. He must stop reading and start living. He mustn't be like Zola, who cross-examined a carpenter in his workshop about the tricks of his trade, then sat down to hammer out a book on the life of a carpenter.\"\n\nYet he did admire a few of his contemporaries, including John Steinbeck, Erskine Caldwell, and especially William Faulkner. He once said in an interview that he wished he could have been Faulkner, because \"he was able to contain the whole of humanity in a small county in the south of the United States.\" Faulkner was also greatly admired by Simenon's contemporary Henry Green. They may have had this in common, but Green couldn't stand Simenon's work.\n\nSimenon regarded Ian Fleming's James Bond novels as insipid, but Fleming was a huge fan of Simenon. So was T. S. Eliot. The film directors Federico Fellini and Fran\u00e7ois Truffaut were admirers, too.\n\nHe met Dashiell Hammett and James Thurber, and formed friendships with Thornton Wilder and Henry Miller, both of whom he corresponded with. \"For us Americans who have just discovered you in translation,\" Miller wrote to him in 1954, \"it is like a new star rising on the horizon.\" And in his longtime friendship with Andr\u00e9 Gide, Simenon opened up about aspects of his personal life that he shared with no one else. But when it came to his writing he was like a magician; he knew better than to reveal too much about how his tricks worked. So when Gide, always awestruck by his friend's extraordinary output, once pressed him in a letter about his creative process, Simenon replied, \"It's a form of self-deception, nothing more.\"\n\nSimenon was flattered by Gide's attention, but he admitted later that he found Gide's work unreadable.\n\nOne fan of Simenon (whom he never met) was the British author John Cowper Powys, who described Simenon as \"my new favorite writer\" and considered him superior to Arthur Conan Doyle. \"I never thought I'd live to see the day that I'd be reading detective stories,\" Powys wrote to a friend, \"but the detective element of Simenon's books is their weakest aspect, generally rather unconvincing. All the rest\u2014atmosphere, composition, narration, and characters\u2014is wonderful, at least for me. It's been years since I've come upon an author who has so pleased me, with so many books, all equally charming.\"\n\nThe novelist, critic, and Flaubert scholar Francis Steegmuller (who was married to the author Shirley Hazzard) was an occasional user of pseudonyms in his own novels. Steegmuller wrote that when Simenon was at his best, he\n\nis an all-round master craftsman\u2014ironic, disciplined, highly intelligent, with fine descriptive power. His themes are timeless in their preoccupation with the interrelation of evil, guilt and good; contemporary in their fidelity to the modern context and Gallic in precision, logic and a certain emanation of pain or disquiet. His fluency is of course astonishing. His life is itself a work by Simenon.\n\nSimenon might have acted nonchalant about how others perceived him, but he soaked up the glory. To his credit, he harbored a degree of humility that lingered from childhood. \"I like plain people,\" he explained in a 1953 interview with Look magazine, \"people who are not all the time thinking about the impression they make and taking notes on themselves. The best thing is for the writer to know the garbage collector.\"\n\nEven after conquering the world with his Maigret novels, he could not succeed in shaking his mother's critical attitude. When Henriette was well into her seventies, her disapproval had not diminished. \"Why don't you ever write a book about nice people and good Catholics,\" she said to her son, \"instead of all these criminals?\"\n\nIn 1961 Simenon's career was still going strong. His work had been (lucratively) adapted for television and film, and he was deep into another affair, with Teresa Sburelin, the family's Italian housekeeper, who was twenty-three years younger than he. Years later, Denyse offered her opinion of her husband's incorrigible ways. \"We made love three times a day every day, before breakfast, after an afternoon nap, and before going to sleep,\" she said. \"Sometimes I wondered whether he didn't think of me as a prostitute. . . . He had contempt for women, but I'm the only one he respected while still showing that contempt. You want to know why he felt the need to cheat on me when he was getting what he needed at home? Definitely to reassure himself. He overdid everything: speaking, writing, publishing, and making love. This was a reflection of his temperament.\"\n\nSupposedly, one afternoon Simenon enjoyed a marathon session of sex with four women in a row while Denyse packed their suitcases in the adjoining room.\n\nHe once said that he viewed sex as \"the only possible form of communication with women.\" Because he had no memory of tenderness from his mother\u2014he claimed that she had never even held him in her lap\u2014his attitude is not surprising. He spent a lifetime trying to move beyond that early abandonment.\n\n\"I have no sexual vices,\" he told Fellini, \"just a need to communicate.\"\n\nAlthough the author's name was worth a fortune\u2014he was a one-man celebrity brand\u2014his personal life fell apart in middle age. For a man who never met a brothel he didn't like, his sexual fervor remained strong as ever, but it started to take its toll. He went through bouts of depression, and even he recognized that his life was in disarray.\n\nHis malaise did not go unnoticed by a French journalist who visited him in 1963, surprised by how grouchy and anxious his interview subject seemed. (Simenon had abruptly moved his family to Switzerland, having enjoyed his time in America but unable to resist his nomadic impulse.) Recalling the interview later, the journalist said he had not come away with a favorable impression. He took a jab at Simenon, describing him as \"an industrialist of literature. He produced, and he sold what he produced.\" Even more damning were his observations of the author's paranoia:\n\nSimenon dreaded a world war or some other catastrophe; hence the enormous laundry and operating room at his home, driven by a generator ready to go at a moment's notice. The house was replete with microphones, supposedly installed so that Simenon would know if one of his children was calling or crying, but I think he also used them to eavesdrop on what others besides his children might be saying about him. And finally, he detested wood, in which any number of undesirable beasts might find shelter. The furniture was of glass, leather, and metal. A curious impression: I listened to Georges Simenon for hours but never really got to know him.\n\nWhether Simenon's decline had to do with his desperately unhappy marriage is unclear, but it is likely. There were violent incidents between him and Denyse. \"He's afraid of her,\" one of his editors said of the couple's relationship. \"She's mad.\"\n\nPerhaps in denial about how bad things were, Simenon made the reckless decision to custom-build a grand home in Epalinges. In the beginning, Simenon claimed that the house was so immense he did not know the exact number of rooms. It was a fortress designed to accommodate his large staff of servants, nannies, and secretaries, and his paintings by Matisse and Picasso. It provided ample space to park his fleet of luxury cars\u2014including a Mercedes, a Jaguar, and a Bentley. Dollar signs were built into the front gates of the grounds. The house had (depending on the source) either eleven or twenty-one telephones; a vast library of his own works, translated into several languages; a service elevator installed specifically to deliver Simenon's meals; and a pool, among other extravagances. Charlie Chaplin and his family were frequent guests.\n\nUnfathomably rich and famous, Simenon became jaded about his career. In 1969, despite being the world's best-selling author, he had grown tired of his beloved detective. \"When I first began Maigret I was 26 and he was 45,\" he told a reporter that year. \"I was his son, he was my father. Now I am 66 and he is only 52, and he is my son and I am his father.\"\n\nA few years later, having written more than eighty Maigret novels, and thousands of pages in multiple genres under various names, he published the final volume and announced that he would never again write fiction in any genre, under any name. His children and friends refused to believe him\u2014they were convinced that he had another surprise in store\u2014but this time he did not. People magazine ran a profile of him accompanied by the headline, \"After 500 Novels and 10,000 Women, Georges Simenon Has Earned His Retirement.\"\n\nUnable to let go of storytelling entirely, he spent years dictating twenty-one volumes of his memoirs into a tape recorder. He addressed the public's immense curiosity about his prolific writing career: \"People will speak of a gift. Why not a malediction?\" He'd once said that whenever he went to his doctor while suffering from a mysterious illness, his doctor would offer the same prescription: \"Write a book.\" He always did, and noticed that he felt better instantly. Writing was his affliction and his cure. \"I'm happy when I've finished,\" he told a reporter five years before his death. \"But during the time I'm writing, it's something awful.\"\n\nAt seventy, he'd endured years of trauma and heartbreak: the collapse of his marriage to the manic-depressive, alcoholic Denyse; the death of his mother, which left him with complicated feelings of grief and anger. And in 1978, his daughter, Marie-Jo, committed suicide at age twenty-five in her apartment in Paris. She shot herself in the chest with a pistol, and a heartbroken Simenon could not recover from the loss. Less than two months earlier, Denyse had published a spiteful, extensively detailed account of their marriage, Un oiseau pour le chat (A Bird for the Cat). Simenon never forgave her for this betrayal, and refused to say her name aloud.\n\nIn his professional life, too, strains became apparent. He felt that he still hadn't received the acclaim he deserved, even though he'd flooded the world with hundreds of millions of copies of his books. The self-described \"imbecile of genius\" could not overcome his spite at being passed over for the Nobel Prize. It had been bad enough when his friend Gide won in 1947, but Simenon felt even more bruised when Albert Camus became the Nobel laureate in 1957. (There had been international rumors that perhaps Simenon would win that year.) The choice of Camus made him furious. \"Can you believe that asshole got it and not me?\" he had complained to Denyse.\n\nHaving abandoned fiction, he also gave up the house at Epalinges. He and Teresa, now his companion, moved first into a high-rise apartment, then into a small, cramped house in Lausanne. (After his death, his ashes would be scattered under an old cedar tree in their garden.) He placed most of his possessions in storage. He changed the \"Occupation\" line of his passport from \"homme de lettres\" to \"sans profession.\" He took a daily nap after lunch. It was a simple life. He and Teresa were devoted to each other.\n\nThe profile in People magazine described a blissful couple: \"[L]ife with Teresa appears serene. They are inseparable. They take a daily promenade together and eat their meals on a precise schedule.\"\n\nAs his health declined and he was confined to a wheelchair, he was philosophical about dying: \"I don't fear death, but I fear causing trouble by my death to those who survive me. I would like to die as discreetly as possible.\"\n\nPerhaps because Simenon had so effortlessly inhabited his many pseudonyms and had experienced such huge success, even writing as himself, he was never unduly preoccupied with how others regarded him. \"I have a very, very strong will about my writing,\" he once said, \"and I will go my way. For instance, all the critics for twenty years have said the same thing: 'It is time for Simenon to give us a big novel, a novel with twenty or thirty characters.' They do not understand. I will never write a big novel. My big novel is the mosaic of all my small novels. You understand?\"\n\nHe had always drawn attention because of his gargantuan appetites, including his sexual escapades, yet in private he was an ordinary man who followed a rigid routine\u2014as cited in a 1969 New York Times profile, just a few years before his final novel was published: \"Mr. Simenon lives by order and discipline. Not only does he rise at 6 on the dot, but he also goes to bed at the first stroke of 10, whether he is in the middle of the sentence or watching a drama on one of his seven TV sets. He falls asleep immediately.\"\n\nHe had the luxury of adhering, without interference, to the simple routine he had designed\u2014never having to do a single thing, for work or pleasure, that he did not schedule himself. And on September 4, 1989, he didn't feel like waking again. With nothing left to say, the great Simenon died serenely in his slumber at 3:30 in the morning.\n\nHe could not have written a better ending.\nShe kept snails as pets\n\nChapter 15\n\nPatricia Highsmith & CLAIRE MORGAN\n\nShe was one of the most wretched people you could ever meet, with mood shifts that swung as wildly as the stock market. Patricia Highsmith was born eleven years before Sylvia Plath, and the two women had a similar temperament. Like Plath, Highsmith possessed a legendary cruel streak and harbored feelings of murderous rage that were directed at family members, lovers, and innocent bystanders alike. One friend said that although she appreciated Highsmith's startlingly direct manner, unaccompanied by tact, she did not care for \"the ranting and raving, the nastiness, the hatred which would overflow.\" When a biographer of Highsmith was asked why she'd become interested in her subject, she replied, \"I have always been interested in women who go too far\u2014and Highsmith went further than anyone.\"\n\nThat point is hard to dispute. Highsmith was a heavy smoker (Gauloises), an alcoholic, and sexually promiscuous. She had affairs with both men and women\u2014almost all of these relationships were intense and unhappy\u2014and she compulsively recorded her sexual encounters. She revised her work by retyping her manuscripts in their entirety \"two-and-a-half times\" on a manual typewriter. She was living proof that not all women have a maternal instinct. She was secretive, misanthropic, gruff, cheap, rude, and generally mean. She had wanderlust. She collected maps. She had an eating disorder and described food as her \"b\u00eate noire.\" She felt disgusted by feminists. She was openly and relentlessly anti-Semitic, and felt that the Holocaust didn't go far enough. She wrote hateful letters, critical of Israel, to politicians and newspapers, using more than forty pseudonyms (including \"Phyllis Cutler\" and \"Edgar S. Sallich\") and disguised signatures. She saved, in her edition of the Holy Bible, an old article with the headline \"Archaeologist Finds the Tomb of Caiphus, the Jewish High Priest Who Handed Jesus Christ Over to the Jews.\" She said that she refused to sell Israel the rights to publish any of her books, and when the ham sandwiches she liked were no longer served in first class on airline flights, she blamed \"the yids\" for it. Yet she had Jewish lovers and friends. She had huge hands. She loved cats and owned many books about cats. She was a racist who believed that if black men didn't have sex many times a month, they became ill. She simultaneously cursed her fame and courted it. She was a compulsive liar. She had a febrile imagination and boasted that she had ideas \"as often as rats have orgasms.\" One of her editors described her as being like a \"child of 10 or 11.\" On her left wrist, she had a tattoo of her initials in Greek letters. She enjoyed watching violent scenes in movies, but shielded her eyes during sex scenes, which repelled her. She always wanted to play the harpsichord. She did play the recorder. She kept snails as pets because she enjoyed watching them copulate, liked their indeterminate gender and self-sufficiency, and said they provided a sense of tranquillity\u2014this from someone almost incapable of relaxation. Her fondness for snails was such that she kept three hundred of them in her garden in Suffolk and insisted on traveling with them. When she moved to France in 1967, she smuggled snails into the country by hiding them under her breasts\u2014and she made several trips back and forth to smuggle them all. Her favorite snails were named Hortense and Edgar. Her favorite flower was the carnation. She liked her Scotch neat. She had bad teeth. She was lonely and anxious, ambidextrous, and physically clumsy. She was sensitive to noise and despised it. She was obsessed by routine and repetition in all areas of her life. She believed that her phone was being wiretapped by people who wanted to steal her money. She liked to read the dictionary every evening before dinner. She was known to start drinking screwdrivers at seven o'clock in the morning. She made furniture. She felt that her best quality was perseverance. She was a gifted visual artist and admired the work of Francis Bacon because \"he sees mankind throwing up into a toilet.\" She was tall, dark, and handsome. She slept with many women named Virginia. She was paranoid and controlling. She contemplated suicide, but rejected the act as too selfish.\n\nPatricia Highsmith was born in Fort Worth, Texas, in 1921, and grew up in New York City. She never felt at home in the United States and left permanently for Europe in 1963. Expatriate life suited her well. \"My most persistent obsession\u2014that America is fatally . . . off the mark of the true reality, that the Europeans have it precisely,\" she wrote in her notebook at age twenty-seven. Her childhood could hardly be described as happy; she despised her equally vicious mother, Mary. Highsmith said that she \"learned to live with a grievous and murderous hatred early on.\"\n\nAfter falling out with Mary in 1974, Highsmith did not see her for the last seventeen years of Mary's life. (It rankled her that her mother lived to the age of ninety-five.) Among what she considered countless slights and misdeeds, Highsmith deeply resented Mary's refusal to accept responsibility for her daughter's character, \"or to put it bluntly queerness.\" When she was fourteen years old her mother asked, \"Are you a les? You are beginning to make noises like one.\" This belittling remark served to alienate Highsmith further from everyone around her.\n\nWhen she was nearly sixty years old, Highsmith was asked by a reporter why she did not love her mother. \"First, because she made my childhood a little hell,\" she said. \"Second, because she herself never loved anyone, neither my father, my stepfather, nor me.\" One of Highsmith's former lovers once commented that Mary was \"high-strung, jealous, and possessive,\" and that mother and daughter \"enjoyed a certain folie \u00e0 deux.\" Although Highsmith dedicated a few books to her mother, she said that she did it only to impress the woman who found fault with everything she did.\n\nIn her diary, Highsmith described herself as feeling \"like a glacier or like stone\" until the age of thirty, but that sense of remove would never leave her. She had a lifelong aversion to being touched, and she bristled when someone shook her hand. (Many acquaintances learned never to do this with her.) Highsmith was perpetually anxious about maintaining boundaries with people. She viewed living with a romantic partner as \"catastrophic.\" Being alone was her preferred state: \"My imagination functions better when I don't have to speak to people,\" she said.\n\nShe was well aware that her taut, self-protective carapace had been caused partly by her upbringing and that it was \"certainly tied up with the fact I had to conceal the most important emotional drives of myself completely.\" Those yearnings were directed toward other women, a fact that drew baffled contempt from her mother.\n\nHighsmith's parents divorced a few days before she was born, and five months before the birth, Mary had tried to abort the fetus by ingesting turpentine. \"Highsmith\" was actually the name of Patricia's stepfather, who the girl believed was her biological father until she was ten years old. (Her initial surname, Plangman, belonged to her father, but she never used it.) When she learned the truth about her stepfather, she wasn't terribly shocked, because she'd suspected for a while that he wasn't her real father. Still, the revelation added another confounding element to her already fragmented sense of identity. The experience of shifting and shedding selves would prove a recurring theme in her work. It was a conundrum she was never able to solve and one that never ceased to fascinate her.\n\nAs a child, Highsmith was reticent, hypersensitive, and self-conscious; she had difficulty forming attachments. By age six, she was aware of an inchoate longing for other girls, which she tried to suppress. An itinerant childhood added to her struggle with (and ambivalence toward) making new friends. But she was a sophisticated and voracious reader, which provided solace. She immersed herself in Dostoevsky, Kafka, Poe, Woolf, and Proust, among others.\n\nWhen she was just eight years old, she discovered The Human Mind, the first book by the influential American psychiatrist Karl Menninger. \"He writes about pyromaniacs, kleptomaniacs, schizos and so on; their case histories, whether they're cured or not,\" she later recalled. \"I found this very interesting, and it was only much later that I realized that it had had such an effect on my imagination, because I started writing these weirdo stories when I was fifteen or sixteen.\" The opening sentence of the first story she wrote was, \"He prepared to go to sleep, removed his shoes and set them parallel, toe outward, beside his bed.\" (Even when she was a teenager, her obsessive-compulsive tendencies were set. These were efforts at control\u2014a coping mechanism in response to the tumult of her early years.)\n\nShe was a lifelong diarist and a relentless maker of charts, sketches, and lists that included ratings of lovers by character trait and category. At her death, she left behind about eight thousand pages from her diaries and \"cahiers,\" as she called her notebooks. (The diaries were for chronicling personal experiences; the \"cahiers\" recorded ideas for stories, poems, and other creative endeavors.) These writings were searching, anguished, and intimate. \"Every move I make on earth is in some way for women,\" she wrote. \"I adore them! I need them as I need music, as I need drawings.\"\n\nShe struggled with the gap between who she was and who she longed to become: \"What and why am I? There is an ever more acute difference . . . between my inner self which I know is the real me, and various faces of the outside world.\" Her identity seemed in perpetual flux, and it was quite a lot to manage. \"Dostoevsky is criticized for ambivalence, for illogic, contradictions\u2014worst of all, ambivalences in his philosophy,\" she once wrote in her diary. \"But there are always two. Perhaps this wonderful, magical, creative, public & private number is the mystic secret of the universe. One can love two people, the sexes are within all of us, emotions directly contrary do exist side by side. This is the way I see the world too.\"\n\nOn December 31, 1947, she wrote a private \"New Year's Toast\": \"[T]o all the devils, lusts, passions, greeds, envys, loves, hates, strange desires, enemies ghostly and real, the army of memories, with which I do battle\u2014may they never give me peace.\" Her own happiness, whatever that meant, was not relevant. Nor did anyone else's well-being matter to her, and in that sense she was a bit like the sociopathic characters in her stories.\n\nIn 1942, Highsmith graduated from Barnard College. Thus began a series of failed job interviews with various magazines. This was (and remains) a common entry-level field for literary college graduates in Manhattan. But no one would have her. Time, Fortune, Good Housekeeping, and Mademoiselle were among the publications that turned her down. Her interview with Vogue was comically disastrous, even though she did have a flair for clothing and usually displayed a distinctive, androgynous style. She was also meticulous about ironing, a domestic task she'd mastered at a young age and found satisfying. Yet for some reason, Highsmith showed up for her much-coveted interview looking like a mess. She appeared at the offices of the world's most glamorous and prestigious fashion magazine \"with a stained and wrinkled blouse, bad hair, and, in the formal 1940s, a head unadorned by a hat,\" as her biographer Joan Schenkar noted. She appeared to have rolled out of bed and gone straight to her interview. In her diary, Highsmith was angry about the rejection (which was clearly her fault). \"Well, I did wash my hair just before going in,\" she wrote. \"There'll come a time when I shall be bigger than Vogue and I can thank my lucky star I escaped their corruptive influences.\" Unlikely as it was, she would prove to be right.\n\nAfter Barnard, she had a secret life: writing comic strips (story lines and dialogue) for at least seven years. Later, as Schenkar discovered, Highsmith attempted to remove, without explanation, all traces of this extensive work from her archives. Still, she seemed oddly suited to writing comics if you consider that she specialized in superheroes with alter egos\u2014secret lives and clandestine identities that shifted from day to night. One of her few pleasures in life was fiercely guarding secrets about herself, down to the most banal details.\n\nIn 1950, she would publish her first novel, Strangers on a Train. It promptly launched her career. The story\u2014which follows two men, Guy and Bruno, who meet on a train and form a murder pact, as well as a twisted, homoerotic bond\u2014had been rejected by six publishers. Yet upon publication it was an immediate success, and Alfred Hitchcock adapted it into a well-received film. (Highsmith was unhappy that the director had paid only about $7,000 to secure the rights. She never got over it.) The process of getting the script written proved challenging; writers such as Dashiell Hammett and John Steinbeck turned down the project. Raymond Chandler wrote an early draft but was fired by Hitchcock. That was probably for the best, as Chandler admitted that he had struggled with the material. \"It's darn near impossible to write, because consider what you have to put over: a perfectly decent young man (Guy) agrees to murder a man he doesn't know, has never seen, in order to keep a maniac from giving himself away and from tormenting the nice young man,\" Chandler wrote. \"We are flirting with the ludicrous. If it is not written and played exactly right, it will be absurd.\"\n\nOther film adaptations of Highsmith's work over the years included Ren\u00e9 Cl\u00e9ment's Purple Noon and Anthony Minghella's The Talented Mr. Ripley. In the 1980s, a smart, talented young film director named Kathryn Bigelow, who would go on to direct the Academy Award\u2013winning film The Hurt Locker, wrote a script on spec for a Highsmith novel she loved. The project never went anywhere, but Highsmith liked Bigelow very much.\n\nTruman Capote was responsible for helping the author complete her draft of Strangers on a Train. In the summer of 1948, thanks to his endorsement, Highsmith was awarded a residency at Yaddo, the prestigious writers' and artists' colony in upstate New York. Also there that summer were Chester Himes and Flannery O'Connor. Highsmith finally got the space and time she needed to finish the manuscript, despite her two-day hangovers. She was thrilled: \"If I cannot give birth in the supreme hospital of Yaddo, where can I ever?\" Fifty years later, in a rare magnanimous gesture, Highsmith would show her gratitude to Yaddo by naming it the sole beneficiary of her estate, along with a $3 million bequest.\n\nShe recalled being instantly taken with the spritelike Capote, if not his writing, and particularly appreciated his openness about being gay. He was entirely unacquainted with the hang-ups that froze Highsmith and left her struggling with her sexuality. Once he told her that at the age of fourteen, he came out to his parents with a simple, jubilant declaration: \"Everybody is interested in girls, only I, T.C., am interested in boys!\"\n\nHighsmith's second novel, as far as anyone knew at the time, was The Blunderer, in 1954; it would be followed a year later by The Talented Mr. Ripley, the book that would ensure her reputation and fame. With that accomplishment she established herself as a master of crime fiction\u2014even though she disliked being typecast in a particular genre\u2014and a creator of psychologically complex characters who, beneath their mannered fa\u00e7ades, were misfits, deviants, and sometimes psychopaths. The British novelist Graham Greene, a great fan of Highsmith's work, described her as a \"writer who has created a world of her own\u2014a world claustrophobic and irrational which we enter each time with a sense of personal danger. Nothing is certain when we have crossed this frontier.\" It was a world that often reflected her interior state and her own disturbing obsessions. Perhaps most troubling of all, Susannah Clapp wrote in a 1999 piece in the New Yorker, was that \"her narratives suggest a seamlessness between bumbling normality and horrific acts. You never hear the gears shift when the terrible moment arrives.\"\n\nIn truth, Highsmith had published her second novel two years before The Blunderer\u2014yet it was not a work she wished to claim credit for. This one was a secret.\n\nThe Price of Salt came out in 1952 under the name of Claire Morgan, who did not exist. Although Highsmith would never again use a pseudonym for any of her novels or stories, this radical narrative demanded a furtive identity. \"Oh god,\" she said, \"how this story emerges from my own bones!\" Homoeroticism was pervasive in her fiction, but always obliquely and within the context of troubled, amoral characters. In a scene from The Talented Mr. Ripley, relations between Tom Ripley and the object of his fixation, Dickie Greenleaf, begin to take an ugly turn when Dickie walks in on Tom dressed in his clothes:\n\n\"Marge and I are fine,\" Dickie snapped in a way that shut Tom out from them. \"Another thing I want to say, but clearly,\" he said, looking at Tom, \"I'm not queer. I don't know if you have the idea that I am or not.\"\n\n\"Queer?\" Tom smiled faintly. \"I never thought you were queer.\"\n\nDickie started to say something else, and didn't. He straightened up, the ribs showing in his dark chest. \"Well, Marge thinks you are.\"\n\n\"Why?\" Tom felt the blood go out of his face. He kicked off Dickie's second shoe feebly, and set the pair in the closet. \"Why should she? What've I ever done?\" He felt faint. Nobody had ever said it outright to him, not in this way.\n\n\"It's just the way you act,\" Dickie said in a growling tone, and went out of the door.\n\nThe Price of Salt, however, depicted consensual (and satisfying) romantic love between two women. It was Highsmith's most autobiographical novel, and it laid bare the emotional drives she had worked hard to keep hidden for so long. Moreover, it was the first gay or lesbian novel with a happy ending. This was not pulp fiction. No one went insane, committed suicide, or was murdered. No one \"converted\" to heterosexuality or found God. This was a breakthrough for the era in which it was written, and surprisingly, the novel was well received by critics. The paperback edition, issued by Bantam a year later, sold more than a million copies. Grateful letters trickled in for years afterward, from both men and women, addressed to Claire Morgan in care of her publishing house. \"We don't all commit suicide and lots of us are doing fine,\" wrote one fan.\n\nIf it was true, as Highsmith wrote in her diary in 1942, that \"[a]ll my life's work will be an undedicated monument to a woman,\" then The Price of Salt was the culmination of that ambition. No wonder it demanded concealment.\n\nThe idea for the novel had arisen from a single but transformative moment. In December 1948, in need of cash and feeling depressed, she took a temporary job during the pre-Christmas rush in the toy department of Bloomingdale's in Manhattan. Though she was hired for a month, she lasted only two and half weeks there. She'd gotten the job partly to pay for her psychoanalytic treatment, which she'd begun in a halfhearted effort to \"cure\" herself of the homosexual urges that alternately tormented her and left her in a manic state of bliss. \"When you're in love it's a state of madness,\" she said.\n\nOne morning, a few days after Highsmith started the job, a beautiful blond woman in a mink coat walked into the toy department, purchased a doll for her daughter, then left the store. Highsmith never saw her again. Yet that brief transaction captivated Highsmith, who had a habit of projecting her fantasies and yearnings onto unsuspecting women she barely knew. \"She could be called the balladeer of stalking,\" Susannah Clapp noted of Highsmith in her New Yorker piece. \"The fixation of one person on another\u2014oscillating between attraction and antagonism\u2014figures prominently in almost every Highsmith tale.\"\n\nTo Highsmith, the woman she'd met \"seemed to give off light.\" And though it had been a routine encounter in which no flirtation had occurred, she was left feeling \"odd and swimmy in the head, near to fainting, yet at the same time uplifted, as if I had seen a vision.\" That night, she went home to the apartment where she lived alone and wrote eight pages in longhand, a broad version of the novel's plot. \"It flowed from my pen as if from nowhere\u2014beginning, middle and end,\" she recalled. \"It took me about two hours, perhaps less.\" Then she fell ill with chicken pox.\n\nBecause this bewitching customer had paid by credit card and asked for the purchase to be sent to her home, Highsmith had the woman's name and address: Mrs. E. R. Senn of Ridgewood, New Jersey. In Highsmith's imaginative retelling, Senn was cast as the seductive older woman, Carol, and Highsmith as the na\u00efve nineteen-year-old shopgirl, Therese. The department store was fictionalized as Frankenberg's. When Carol invites Therese out for lunch, the young protagonist, despite having a boyfriend, feels the first stirrings of love. \"An indefinite longing, that she had been only vaguely conscious of at times before, became now a recognizable wish,\" Highsmith wrote. \"It was so absurd, so embarrassing a desire, that Therese thrust it from her mind.\" Some passages in the novel were taken verbatim from the author's own notebooks and diaries. Although the initial writing of the novel came easily to her, the revision stage brought out dark emotions. As the publication date grew closer, Highsmith suddenly crashed, hitting one of the lowest points of her life. She became self-destructive to a terrifying extent, going on drinking binges and feeling more miserable than ever. At the very moment she should have been celebrating a work that she felt proud of, she experienced an agonizing case of writer's remorse. She wanted to withdraw the novel from publication: it was so deeply personal that she feared it would destroy her, both personally and professionally. The use of an invented name was only a mild anodyne for her anxiety. Mostly, she felt sick with worry and shame: \"These days are on the brink again. The least thing depresses me to the point of suicide.\"\n\nIn fact, suicide was the fate of Mrs. E. R. Senn\u2014a grim twist worthy of a Highsmith tale. Married to a rich businessman, the beautiful woman who had aroused Highsmith's ardor was an alcoholic who had been in and out of psychiatric hospitals. She had absolutely no idea that she'd inspired a lesbian love story. In the fall of 1951, Kathleen Wiggins Senn killed herself by carbon monoxide poisoning in the garage of her lavish home in Bergen County.\n\nIt wasn't until the 1990 British edition of The Price of Salt was released that Highsmith explained in an afterword why she'd decided to publish under a pseudonym. Both her publisher and agent seemed determined to have her keep writing the same books over and over, confining her to so-called crime fiction. After the publication of Strangers on a Train, she'd been tagged instantly as a certain kind of writer, even though in her mind it was \"simply a novel with an interesting story.\" (The reductive business of branding and marketing is unchanged even today.) She found this rather frustrating, and in objecting to being labeled she had her share of supporters.\n\n\"Patricia Highsmith is often called a mystery or crime writer,\" a newspaper critic noted, \"which is a bit like calling Picasso a draftsman.\" To Gore Vidal, who shared her expatriate anti-American views, she was simply one of the greatest modernist writers. And the playwright David Hare admired her work because \"behind it lies the claim that, once you set your mind to it, any one human being can destroy any other.\"\n\nHighsmith knew that her literary genius transcended any single genre, and she detested any kind of categorization. She considered herself a neglected master. \"If I were to write a novel about a lesbian relationship,\" she wrote in the afterword, \"would I then be labeled a lesbian book-writer? That was a possibility, even though I might never be inspired to write another such book in my life. So I decided to offer the book under another name.\" She also must have wanted to protect her reputation and nascent career, although she never admitted this outright. (Nor did she wish to upset her eighty-four-year-old grandmother, Willie Mae.)\n\nAfter all, she noted, those were the days when homosexuals were widely viewed as perverts, when \"gay bars were a dark door somewhere in Manhattan, where people wanting to go to a certain bar got off the subway a station before or after the convenient one, lest they be suspected of being homosexual.\"\n\nThe pseudonym gave her the safety she craved. Because the story of her two characters ends on a sweet, hopeful note, this novel seemed an exercise in wish fulfillment for the author. In her own life, Highsmith almost always experienced thwarted love, painfully brief relationships, and bitter rejections.\n\nIn 1959, she began an on-and-off relationship with the author Marijane Meaker, who would also publish under pseudonyms, including Vin Packer, Ann Aldrich, and most famously M. E. Kerr. They met at L's, a lesbian bar in Greenwich Village, and in her memoir Meaker later recalled Highsmith wearing a trench coat, drinking gin neat, and looking like \"a combination of Prince Valiant and Rudolph [sic] Nureyev.\" She admired Highsmith's resistance to societal attitudes toward homosexuality: \"I don't care for acceptance,\" Highsmith told Meaker, who was in her early thirties at the time and foolishly believed that she had found her life partner. (The relationship would last two years.) They lived together for a time, but Meaker later confessed that if they hadn't had \"such good horizontal rapport,\" the affair would have ended much sooner.\n\nAfter breaking up, they stayed in touch\u2014which meant that Meaker had to deal with Highsmith's narcissism by mail instead. \"Did I tell you that Bloomsbury liked my latest Ripley so much they gave me an advance that in American money comes to about $115,000?\" Highsmith wrote in one letter. \"I never got that much for a book. You know, in the U.S. no one really recognizes me, but in Europe I'm often recognized and treated like a celebrity.\" In other letters she railed against Jews, adding in one postscript that \"USA could save 11 million per day if they would cut the dough to Israel.\"\n\nThat wasn't all. Immediately after their breakup, Highsmith wrote a novel called The Cry of the Owl, in which an \"unsuccessful artist\" was a thinly disguised version of Meaker. The character was viciously knifed to death for several pages at the end.\n\nBy 1983, many people suspected that Highsmith had been the author of The Price of Salt; although she refused to address the truth in any way, it had become a poorly kept secret. When contacted by Barnard's alumni magazine for an article about her, and asked directly whether she had written the novel, Highsmith replied that the less said about the subject, the better, and left it at that. In the same year, Naiad Press bought the rights to reissue the book, but Highsmith declined to publish it under her own name. Naiad tried to tempt her by offering a $5,000 advance for publishing with full disclosure, or $2,000 for publishing under a pseudonym. She refused to take the bait.\n\nFor the 1990 UK edition, Highsmith finally came around to coming out. She had stubbornly resisted even then, but she did consent to putting her real name to the work. The novel was also released with a new title, Carol. Although Highsmith had no wish to analyze her decision for the press or the public, the book spoke for itself. Writing it had been an act of courage, even if the author wanted no part in acknowledging that fact. Today, it remains one of her best works, a novel worth reading and revisiting.\n\nHighsmith spent her last thirteen years alone in a two-hundred-year-old farmhouse in southern Switzerland. She died of cancer in 1995, at the age of seventy-four, and her body was cremated. Of the author's final weeks, a neighbor recalled, \"There was a tranquillity about her. She seemed to be quite peaceful, and as lucid as could be.\"\nShe liked whips and chains\n\nChapter 16\n\nPauline R\u00e9age & DOMINIQUE AURY\n\nNot many authors can boast of having written a best-selling pornographic novel, much less one regarded as an erotica classic\u2014but Pauline R\u00e9age could. Make that Dominique Aury. No: Anne Desclos.\n\nAll three were the same woman, but for years the real name behind the incendiary work was among the best-kept secrets in the literary world. Forty years after the publication of the French novel Histoire d'O, the full truth was finally made public. Even then, some still considered it the most shocking book ever written. When the book came out, its purported author was \"Pauline R\u00e9age,\" widely believed to be a pseudonym. Although shocking for its graphic depictions of sadomasochism, the novel was admired for its reticent, even austere literary style. It went on to achieve worldwide success, selling millions of copies, and has never been out of print. This was no cheap potboiler. There was nothing clumsy, sloppy, or crude about it. Histoire d'O was awarded the distinguished Prix des Deux Magots, was adapted for film, and was translated into more than twenty languages.\n\nDesclos (or, rather, Aury, as she became known in her early thirties) was obsessed with her married lover, Jean Paulhan. She wrote the book to entice him, claim him, and keep him\u2014and she wrote it exclusively for him. It was the ultimate love letter.\n\nWhips and chains and masks! Oh, my. When Histoire d'O appeared in France in the summer of 1954, it was so scandalous that obscenity charges (later dropped) were brought against its mysterious author. Even in the mid-twentieth century, in a European country decidedly less prudish than the United States, the book struck like a meteor. That the writer had evidently used a pen name provoked endless gossip in Parisian society. Speculation about the author's identity became a favorite sport among the literati: was the author prominent, obscure, male, female, perverted, crazy? The authorial voice was too direct, too cool, to be that of a woman, some argued; others insisted that no man could have offered such a nuanced exploration of a woman's psyche. One thing was certain: the person who wrote this novel had no shame.\n\nStory of O, the title of the English edition, is an account of a French fashion photographer, known only as O, who descends into debasement, torment, humiliation, violence, and bondage, all in the name of devotion to her lover, Ren\u00e9. Over the course of the novel she is blindfolded, chained, flogged, pierced, branded, and more. As the story opens, O is a passive figure who does precisely what she's told:\n\nHer lover one day takes O for a walk in a section of the city where they never go\u2014the Montsouris Park, the Monceau Park. After they have taken a stroll in the park and have sat together side by side on the edge of a lawn, they notice, at one corner of the park, at an intersection where there are never any taxis, a car which, because of its meter, resembles a taxi.\n\n\"Get in,\" he says.\n\nShe gets in.\n\nThe book is like an erotic version of those childhood tales in which a character steps accidentally into an alternate reality and is induced into a hallucinatory state. (Paulhan once insisted that \"fairy tales are erotic novels for children.\") Think of Alice falling down the rabbit hole, or the magic wardrobe leading to Narnia. That was Story of O, albeit with a much darker vision. By the novel's eleventh page, O has been abandoned by her lover at a ch\u00e2teau outside Paris. Alone, she is subdued, quietly following instructions without resistance. She undresses and is fitted with a locked collar and bracelets and a long red cape. Blindfolded, she cries out as a stranger's hand \"penetrated her in both places at once.\" Thus begins her odyssey as a sexual slave to the mostly anonymous men and women who have their way with her. \"O thought she recognized one of the men from his voice,\" R\u00e9age writes, \"one of those who had forced her the previous evening, the one who had asked that her rear be made more easily accessible.\" Willing to do anything with anyone, she reveals an existential longing for release. Aury once observed that \"O is looking for deliverance, to thrust off this mortal coil, as Shakespeare says.\"\n\nYears after the book was published, Aury offered insight into her protagonist's apparent fa\u00e7ade of passive acceptance. \"I think that submissiveness can [be] and is a formidable weapon, which women will use as long as it isn't taken from them,\" she said. \"Is O used by Ren\u00e9 and Sir Stephen, or does she in fact use them, and . . . all those irons and chains and obligatory debauchery, to fulfill her own dream\u2014that is, her own destruction and death? And, in some surreptitious way, isn't she in charge of them? Doesn't she bend them to her will?\"\n\nThe novel also featured scenes of women seducing women. Those encounters seemed genuine rather than forced, contrary to accusations that the author had written such scenes to satisfy the \"male gaze.\" Aury considered herself bisexual and admitted her preference for the female body. Describing her first real-life exposure to male anatomy, she said, \"I found that stiffly saluting member, of which he was so proud, rather frightening, and to tell the truth I found his pride slightly comical. I thought that that must be embarrassing for him, and thought how much more pleasant it was to be a girl. That, by the way, is an opinion I still hold today.\"\n\nThroughout the story, O readily offers herself. She responds to pain and suffering with acceptance or gratitude. The narrative culminates in an all-night party in which she is led along on a dog leash, naked, wearing an owl mask. After she has had a depilatory, to please her master, a chain is attached to rings inserted into her labia. (Her journey seemed to confirm the French writer Georges Bataille's dictum: \"Man goes constantly in fear of himself. His erotic urges terrify him.\") O's response to such terror is absolute surrender, allowing her experiences to lead her into a realm of no pathology, analysis, or consequence. As just about every self-help book advises, opening yourself to the unknown can feel very good. It can transform you. Then again, it can also make you insane.\n\nDepending on your erotic wishes and habits, Story of O will disturb you, frighten you, make you angry, make you upset, confuse you, disgust you, or turn you on. Maybe everything at once. Decades after its publication, the novel has not lost its shock value. In 2009, a commentary in the Guardian following a Radio 4 program, \"The Story of O\u2014The Vice Fran\u00e7aise,\" explained that the late-night timing of the program was apt, because the material was \"strong stuff\" and might have made people queasy. One listener had remarked, on the air, that hearing excerpts from the book provoked \"a rush of blood to the non-thinking parts.\"\n\nAs the author once revealed, the character O actually began as Odile, the name of a close friend who'd once been deeply in love with Albert Camus. \"She knew all about the name and was enchanted,\" Aury said. \"But after a few pages I decided that I couldn't do all those things to poor Odile, so I just kept the first letter.\" Contrary to speculation over the years by feminists, academics, psychoanalysts, and general readers obsessed with the book, the name O, she said, \"has nothing to do with erotic symbolism or the shape of the female sex.\"\n\nHowever depraved her novel seemed, Aury had set out to create a profoundly personal work of art, not cheap porn. (\"That Pauline R\u00e9age is a more dangerous writer than the Marquis de Sade follows from the fact that art is more persuasive than propaganda,\" declared an essayist in the New York Review of Books.) Aury was making something new, working with conventions as no one had attempted in quite the same way. \"Debauchery conceived of as a kind of ascetic experience is not new, either for men or for women,\" she explained, \"but until Story of O no woman to my knowledge had said it.\"\n\nAury seemed an unlikely candidate to produce a book showcasing violent penetration. From childhood she'd been a serious reader, immersing herself in Boccaccio, Shakespeare, Baudelaire, and the Bible. She once boasted of a period in which she'd read and reread the whole of Proust each year for five years. It seemed inconceivable that a woman with such a drab exterior could explore a sexual compulsion that drove her protagonist toward oblivion. Also distinguishing the novel from what one critic called \"volumes sold under the counter\" were its intricate ideas about human behavior\u2014that \"we are all jailers, and all in prison, in that there is always someone within us whom we enchain, whom we imprison, whom we silence,\" as she later explained. Story of O is about power, the pleasure of having it, and finally the pleasure of letting it go. For her part, the author admitted her comfort with the notion of obedience, at least in certain contexts. \"I think I have a repressed bent for the military,\" she said. \"I like discipline without question, specific schedules and duties.\"\n\nPaulhan, the impetus for Aury's cri de coeur, was one of France's leading intellectuals and the publisher of the preeminent literary journal Nouvelle Revue Fran\u00e7aise. His affair with Aury lasted thirty years, until his death in 1968. Throughout their relationship, Paulhan remained married to his second wife, Germaine, who had Parkinson's disease. She was well aware of her husband's philandering, which he expected her to tolerate without protest. And Aury was not his only mistress. After his death, his daughter-in-law remembered him as \"quite the ladies' man.\" (It's interesting that Aury used precisely the same phrase in recalling her own father.)\n\nWhen she met Paulhan, Aury was in her early thirties and he was in his fifties. (She was born in 1907; he was born in 1884.) She'd been married briefly and had a son, Philippe. Her father, an acquaintance of Paulhan, had introduced them. At the time, she was hoping to publish a collection of sixteenth- and seventeenth-century French religious poetry, and Paulhan was an editor at the distinguished publishing house Gallimard. She did not describe their meeting as love at first sight. \"It was slow, but it went very\u2014efficiently,\" she said, recalling her initial impression of him as handsome, charming, and funny. They bonded through shared intellectual passion; during the Nazi occupation of France, while doing work for the Resistance, they became lovers. \"Dominique Aury was fascinated by intelligence,\" a friend recalled. \"The intelligence of Paulhan was obvious. And for her it became a kind of obsession.\"\n\nUntil her fateful meeting with Paulhan, Aury hadn't yet found the love of her life, and her sexual history was hardly remarkable. \"By my makeup and temperament I wasn't really prey to physical desires,\" she once said. \"Everything happened in my head.\" That would explain the electricity between her and Paulhan, which would exert a hold on her for the rest of her life. Although she could talk extensively about sex, her personal life was fairly tame. She did once joke, however, that she'd considered prostitution as a potential vocation: \"I told myself that had to be absolutely terrific: to be constantly wanted, and to get paid besides, how could you go wrong?\" she said. \"And what happens? At the first opportunity, what do I do but turn into a stupid prude!\" Yet she had also wondered what it might be like to become a nun\u2014drawn to it, no doubt, by the stern uniform.\n\nOf course, Aury was destined not for prostitution but to live, work, and breathe intellectual society. She toyed with her identity well before Histoire d'O was published. At some point during the war, while working as a journalist and translator, she discarded her original name, Anne Desclos, erasing it entirely from her professional and personal life. Almost no one knew that Aury was not actually her own name; she kept that fact a secret. She had chosen \"Dominique\" for its gender neutrality, and \"Aury\" was derived from her mother's maiden name, \"Auricoste.\"\n\nAlthough it's true that Story of O was inspired by Paulhan's offhand remark to Aury that no woman could ever write a \"truly\" erotic novel, a more compelling motive was her fear, however irrational, that their relationship might end. \"I wasn't young, I wasn't pretty, it was necessary to find other weapons,\" she later revealed. \"The physical side wasn't enough. The weapons, alas, were in the head.\" She plunged into the task: writing through the night, in pencil, in school exercise books, while lying in bed, and she produced\u2014three months later\u2014her intimate masterpiece. The first sixty pages, she said, flowed \"automatically\" and appeared in the book exactly as they had come to her.\n\nThe novel was written as a challenge to Paulhan's dare (or assignment, if you want to call it that). \"I wrote it alone, for him, to interest him, to please him, to occupy him,\" she told the documentary filmmaker Pola Rapaport shortly before her death. Aury never intended the novel to be made public, but Paulhan insisted on it. For her, the manuscript was simply a long letter that had to be written. She hoped this gift would ensure the permanence of their relationship. \"You're always looking for ways to make it go on,\" she said. \"The story of Scheherazade, more or less.\"\n\nThe content of the novel was graphic, but the author's prose was highly controlled, disciplined, and spare. Her \"voice\" was at odds with the erotic material, making it hard to dismiss as pornography. For Paulhan, the book was \"the most ardent love letter that any man has ever received.\" He did not abandon her.\n\nThe author said later that Story of O, written when she was forty-seven, was based on her own fantasies. She was influenced, too, by her lover's admiration for the Marquis de Sade. Later she described her feverish writing process as \"writing the way you speak in the dark to the person you love when you've held back the words of love for too long and they flow at last . . . without hesitation, without stopping, rewriting, discarding . . . the way one breathes, the way one dreams.\"\n\nPaulhan was awestruck. When he excitedly asked if he could find a publisher for her work, she agreed on the condition that her authorship remain hidden, known only to a select few. She gave herself the pen name \"Pauline R\u00e9age\": \"Pauline\" after Pauline (Bonaparte) Borghese, elder sister of Napoleon, who was famous for her sensual, decadent pursuits; as well as Pauline Roland, the late-nineteenth-century French women's rights activist. Despite the apparent blur between \"Pauline\" and \"Paulhan,\" Aury said later that her appellation had nothing to do with him. (Some insisted, wrongly, that she chose the name because it sounded like the French for \"Reacting to Paulhan.\") As for \"R\u00e9age,\" she'd supposedly stumbled upon it in a real estate registry.\n\nPeople assumed that aspects of Story of O were highly autobiographical, yet Aury wasn't so sure. Some twenty years after the book came out, she admitted that her own joys and sorrows had informed it, but she had no idea just how much, and did not care to analyze anything. \"Story of O is a fairy tale for another world,\" she said, \"a world where some part of me lived for a long time, a world that no longer exists except between the covers of a book.\"\n\nShe characterized \"Pauline R\u00e9age\" in vague terms as well\u2014someone who \"is not me entirely and yet in some obscure way is: when I move from one me to the other the fragments scatter, then come back together again in a pattern that I'm sure is ever-changing. I find it harder and harder to tell them apart anymore, or at least not with sufficient clarity.\" Like many pseudonymous authors, Aury saw identity as unstable and felt perfectly at ease inhabiting a self that refused to remain a fixed star.\n\nShe knew that finding a publisher for her novel (whether or not she took a pen name) would not be easy. It was Paulhan who demanded that the book reach the public, and he fought for it. In this instance, however, his prestige within the literary world carried no clout. Gallimard promptly refused the work, not wanting to deal with the inevitable (and expensive) hassle of a court case. \"We can't publish books like this,\" Gaston Gallimard told her. This was especially disappointing because Aury had worked for him. A few years before her death, Aury said that she had never forgiven Gallimard's rejection of her novel, since he'd already published Jean Genet, whose work was \"much nastier.\"\n\nPaulhan persuaded Jean-Jacques Pauvert\u2014an ambitious twenty-seven-year-old publisher who'd issued Sade's complete works, and who was already a veteran of obscenity trials\u2014to accept Story of O. \"It's marvelous, it'll spark a revolution,\" Pauvert said to Paulhan after reading it overnight. \"So when do we sign the contract?\"\n\nIn 1954, Pauvert published a gorgeously designed first edition of two thousand copies. It had a laudatory preface by Paulhan, \"Happiness in Slavery,\" in which he argued that women in their truest nature crave domination; that O is empowered by confessing her desire; and that, in truth, slaves love their masters, would suffer in their absence, and have no wish to achieve independence. Indeed, as one reviewer noted, the more O is brutalized, the more \"perfectly feminine\" she becomes. This is one of the elements that makes the novel more disturbing than arousing.\n\nPaulhan conceded that there was \"no dearth of abominations in Story of O. But it sometimes seems to me that it is an idea, or a complex of ideas, an opinion rather than a young woman we see being subjected to these tortures.\"\n\nThe book was a sensation, but hardly a blockbuster. Although it was a topic of titillating gossip among the cognoscenti, a year after publication, the initial printing had not sold out. Aury was not hopeful about the book's prospects; she believed it was doomed to be relegated to the \"reserved\" section of libraries, if it was ordered at all.\n\nIts status as a best seller was achieved slowly, as the mystique around it continued to build and as other international editions were issued. Initially, because many French booksellers assumed that the novel had been banned, they tended to conceal it under the counter\u2014thus ensuring that sales would be poor. \"Everyone talked about it in private,\" the author recalled, \"but the press acted as though the book had never been published.\"\n\nWhatever attention Histoire d'O did receive focused on the author's identity, not on the text itself as something worthy of consideration and analysis. Susan Sontag was the first major writer to recognize the novel's merit and to defend it as a significant literary work.\n\nIn her 1969 essay \"The Pornographic Imagination,\" Sontag insisted that Story of O could be correctly defined as \"authentic\" literature. She compared the ratio of first-rate pornography to trashy books within the genre to \"another somewhat shady sub-genre with a few first-rate books to its credit, science fiction.\" She also maintained that like science fiction, pornography was aimed at \"disorientation, at psychic dislocation.\"\n\nIf so, that aim is far more interesting than what most generic \"mainstream\" novels set out to do. No one could describe O as predictable or sentimental. Its vision was dark and unrelenting; everything about it was extreme. Sontag also compared sexual obsession (as expressed by R\u00e9age) with religious obsession: two sides of the same coin. \"Religion is probably, after sex, the second oldest resource which human beings have available to them for blowing their minds,\" she wrote. In her disciplined effort toward transcendence, O is not unlike a zealot giving herself to God. O's devotion to the task at hand takes the form of what might be described as spiritual fervor. She loses herself entirely\u2014and, after all, the loss of self is a goal of prayer.\n\nIf O is willing to sustain her devotion all the way through to her own destruction, so be it. She wants to be \"possessed, utterly possessed, to the point of death,\" to the point that her body and mind are no longer her responsibility. \"What does a Christian seek but to lose himself in God,\" Aury, a devout atheist, once said. \"To be killed by someone you love strikes me as the epitome of ecstasy.\"\n\nSontag's essay was notable for refusing to conflate all porn as bad or to dismiss it all as \"dirty books.\" It was a thoughtful, rational piece on the aesthetic virtues of pornography at its best. In arguing that some so-called pornographic books were legitimate works of art, she acknowledged that staking such a claim was a daunting task: \"Pornography is a malady to be diagnosed and an occasion for judgment. It's something to be for or against . . . quite a bit like being for or against legalized abortion or federal aid to parochial schools.\"\n\nHer case for the literary value of Story of O was compelling and highly specific: \"Though the novel is clearly obscene by the usual standards,\" she wrote, \"and more effective than many in arousing a reader sexually, sexual arousal doesn't appear to be the sole function of the situations portrayed. The narrative does have a definite beginning, middle, and end. The elegance of the writing hardly gives the impression that its author considered language a bothersome necessity. Further, the characters do possess emotions of a very intense kind, although obsessional and indeed wholly asocial ones; characters do have motives, though they are not psychiatrically or socially 'normal' motives.\" All R\u00e9age did was bring into the open the kinds of impulses many people harbor in their bedrooms, alone, late at night. And, from Sontag's perspective, Story of O was not really pornography but \"meta-pornography, a brilliant parody.\"\n\nWho would suspect that Dominique Aury was Pauline R\u00e9age? In midlife, Aury was a respected figure: an influential editor, a writer, and a jury member for various literary prizes. She'd earned the L\u00e9gion d'Honneur; she had translated into French works by authors such as T. S. Eliot, Evelyn Waugh, F. Scott Fitzgerald, and Virginia Woolf; and she had been the only woman to serve on Gallimard's esteemed reading committee. Her demure appearance gave no hint of owl masks or dog collars. She was polite, refined, elegant, shy. She could not be described as beautiful. A friend remembered Aury as \"very self-effacing,\" and as having worn \"soft, muted colors which really matched her personality.\" She dressed quite plainly and wore almost no makeup. At least on the surface, nothing about her was subversive. (She said that dressing in a kind of basic uniform made life simpler.) If anything, Aury seemed conservative, even severe\u2014and to look at her, you might assume that her sexual fantasies would be as stimulating as staring at a dusty library shelf.\n\nThe glaring incongruity between her work and her personal life was not lost on Aury. That was why the pen name was so crucial. She insisted that \"it would have been wrong to mix what was for so long a time secret with something that was always banal and devoid of interest.\" Aury never felt a need to justify the distinction to anyone; it was what she wanted, and it was nobody's business. She was not \"living a lie,\" because Dominique Aury was not \"Pauline R\u00e9age,\" who had produced the scandalous work. \"For a long time I've lived two parallel lives,\" Aury explained. \"I have meticulously kept those two lives quite separate, so separate in fact that the invisible wall between them seems to me normal and natural.\"\n\nUpon the publication of Story of O, guessing games were rampant about the author's identity. Contenders included Raymond Queneau, Andr\u00e9 Malraux, and the most unlikely of all, George Plimpton, founding editor of the Paris Review. \"It wasn't me,\" Plimpton told a reporter in the early 1990s, \"but it's a rumor I prefer not to scotch.\" Paulhan, too, was a possibility, suspected at the very least of knowing the enigmatic author's identity. As Sontag noted, the theory that Paulhan was the author seemed credible partly because of his introductory essay for the novel. It called to mind the mask of Georges Bataille, who, having written his Madame Edwarda under the pseudonym \"Pierre Angelique,\" also contributed its preface under his own name.\n\nFor a while, Paulhan was under intense scrutiny, and his longtime, already volatile friendship with the writer Fran\u00e7ois Mauriac was threatened by the novel's publication. Mauriac, a devoted if somewhat conflicted Catholic, acknowledged that he hadn't read O, but nonetheless publicly attacked the book. He was convinced that Paulhan had written it, and Paulhan responded by lashing out, accusing Mauriac of being the real author.\n\nSome readers believed that Paulhan had heavily edited the kinky text. He denied doing so, and Aury, too, insisted later that he hadn't altered so much as a comma. She said the extent of his editing consisted of omitting a single adjective: \"sacrificiel.\" Pauvert, who'd known Aury for more than a decade (and knew of her nom de plume), had no doubt that the novel was Aury's alone. \"I recognized her style immediately when I first saw the manuscript,\" he said. \"She is a great writer and absolutely uncopyable. Paulhan said that he could not write like that\u2014that his own style was quite different, very dry, ironic, and he could not change it.\"\n\nA hastily released English translation came within weeks, issued by Maurice Girodas of the Olympia Press. This edition \"horrified\" Aury; she found it \"vulgar\" and said that \"it cheapens the character of the book.\" (She did approve, however, of the translation published by Barney Rosset's Grove Press in 1965.) Fan mail and hate mail poured in. Such a fuss was made that Pauvert and Girodas were interrogated by French police after the novel won the Deux Magots prize. Both men refused to reveal the whereabouts of Pauline R\u00e9age, and despite an investigation, no legal action came of it.\n\nAs a prime suspect in the making of this scandalous text, Paulhan paid a price. When he was nominated for membership in the elite Acad\u00e9mie Fran\u00e7aise\u2014which consists of forty members known as \"immortals\"\u2014the opponents of his candidacy are said to have placed a copy of Histoire d'O on every Academy member's chair in protest. (He was elected anyway.) He was also forced to provide a deposition to the vice squad in 1955 as it held hearings to determine whether legal action should be taken against the book. Of course he lied in his testimony. He declared that \"Mme. Pauline R\u00e9age (a pseudonym) paid me a visit in my office . . . and submitted to me a thick manuscript.\" There was some truth in Paulhan's deposition\u2014his feelings about the manuscript and why he had championed it. He revealed that he was struck by the book's literary quality \"and, if I may say, in the context of an absolutely scabrous subject, by its restraint and modesty.\" He said nothing about being in love with the author, but he was completely honest in recalling his first response to it. \"I had in my hands a work that was very important in both its content and its style,\" he said, \"a work that derived much more from the mystical than from the erotic and that might well be for our own time what Letters to a Portuguese Nun or Les Liaisons Dangereuses were for theirs.\"\n\nHe concluded his statement by reiterating that R\u00e9age did not wish to reveal her true identity, and that he intended to protect her desire for privacy. \"Nonetheless,\" he added, \"since I do see her fairly regularly, I shall inform her of the statement I have just made, and in case she should change her mind I shall ask that she get in touch with you.\"\n\nAury had her own dealings with the police: They showed up at her house one day to interrogate her about the book, and she feigned ignorance. Inexplicably, they chose not to pursue the matter\u2014a courtesy for which she was grateful. But she did feel terribly guilty that the vice squad had focused so intently on her lover and her publisher.\n\nStill, she suffered her share of awkward encounters, snubs, insults, scorn, and ignorant and rude remarks. Because of her anonymity, people felt free to express their opinions about the book. If anyone asked her directly whether she was R\u00e9age, she'd simply reply, \"That is a question to which I never respond.\" (It was a clever response, neither an admission nor a denial.) She was startled to read a characterization of her book as \"violently and willfully immoral\"; such commentary served to confirm that the pseudonym had been the best way to go. Aury was not na\u00efve, and understood that the novel was quite racy, but she didn't think it was offensive. To her, one's sense of morality was assaulted daily by reading the newspaper. \"Concentration camps offend decency,\" she said, \"as does the atomic bomb, and torture; in fact, life itself offends public morals every minute of the day, in my opinion, and not specifically through the various and sundry methods of making love.\"\n\nOnce, at a dinner party, she was amused to hear a friend confidently announce that people who wrote books such as Histoire d'O were very sick. Another time, in the presence of her mother, a family friend abruptly turned to Aury and said he believed that she'd written Histoire d'O. She panicked, but said nothing. There was an awkward silence. Then her mother said, \"She never mentioned it to us.\" After their guest left, Aury's mother offered her more tea and never spoke of it again. \"My freedom lay in silence, as my mother's lay in hers,\" Aury later recalled. \"Hers was the refusal to know; mine, the refusal to say.\"\n\nAlthough her father had an extensive erotica collection and had spoken frankly to her early on about sex, Aury's mother was another matter. \"She didn't like men,\" Aury said. \"She didn't like women, either. She hated flesh.\"\n\nSome of the vitriol directed against Aury's book was quite shocking. People described it as trash. (How many had actually read the book?) The author was accused of being antifeminist and of dishonoring all women. Never mind that no man who had written pornography was ever blamed for debasing his gender. Aury received plenty of nasty letters addressed to her alter ego: one writer called her a \"damned bitch\" who catered to the lowest common denominator for money. Another cursed the womb that bore her. Perhaps one of the most perplexing letters was from a man who told her that although the fantasy S&M world she wrote about did exist, it was only between men and boys. He claimed that it was much easier to dominate young boys than women.\n\nAs for Aury's son, Philippe, who was in his twenties when O appeared, he told a journalist after his mother's death that he'd had no clue what she had been up to. \"I didn't know she was the author,\" he said. \"She never told me, really. I only found out in 1974, when there was talk of making a film and people came round to discuss it.\" The film, made in 1975, is universally acknowledged to be dreadful. However, he added, \"It is a very good book.\"\n\nJean-Jacques Pauvert once told an amusing anecdote about being on holiday with his wife in 1957 or 1958 and overhearing a conversation at a restaurant. A group of people were seated at a table behind them\u2014\"well dressed, in their late forties or fifties, probably notables of the town, quite cultivated people, talking about books.\" Suddenly, Pauvert recalled:\n\nOne of the men said, \"You must understand that since Paulette wrote Histoire d'O she has had a very difficult time\u2014isn't that right, Paulette?\" His wife, a good-looking woman, about forty-five years old, wearing a fine pearl necklace, replied, \"Yes, you know, it's been terrible for me. If I had only known what it would turn into, what with my husband's position. . . . It's absolutely terrible.\" This seemed to be going on all over France. There were literally hundreds of people claiming to be the author of O.\n\nEach of the three introductory notes in the novel expresses bafflement as to the author's real identity. The translator Sabine d'Estr\u00e9e\u2014and more on that pseudonym later\u2014pointed out that Pauline R\u00e9age was \"a name completely unknown in French literary circles, where everyone knows everyone.\" Aside from corresponding with the author about the translation, d'Estr\u00e9e admitted, \"I have never met Pauline R\u00e9age.\" The shock of the book itself paled in comparison with the public's curiosity about the name of the person who wrote it. \"Until her identity was bared,\" d'Estr\u00e9e wrote, \"people found it difficult to assume a reasonable stance vis-\u00e0-vis the work; if Pauline R\u00e9age was the pseudonym of some eminent writer, they would feel compelled to react one way; if she were a complete unknown, another; and if indeed she were a literary hack merely seeking notoriety, then still another.\"\n\nThe quality of R\u00e9age's prose made it clear that the last alternative was highly unlikely, if not impossible. \"To this day,\" the translator wrote, \"no one knows who Pauline R\u00e9age is.\"\n\nFor his part, Paulhan offered no clues. \"Who is Pauline R\u00e9age?\" he wrote in his preface, describing the novel as \"one of those books which marks the reader, which leaves him not quite, or not at all, the same as he was before he read it.\" He proclaimed it a \"brilliant feat\" from beginning to end, one that read more like someone's private letter than a diary. \"But to whom is the letter addressed?\" he asked, disingenuously. \"Whom is the speech trying to convince? Whom can we ask? I don't even know who you are.\"\n\nNothing about the novel was straightforward, as the New York Review of Books noted in 1966 about the Grove Press edition: \"[O]ne is struck by an atmosphere of prestidigitation, of double and triple meanings that suggest an elaborate literary joke or riddle which extends even to the question of O's authorship. Pauline R\u00e9age, except as author of the present book and of the preface to another, seems not otherwise to exist: None of her admirers claims to have met her, she has not been seen in Parisian literary circles, and it has been said that she is actually a committee of literary farceurs, sworn to guard their separate identities, like the pseudonymous authors of a revolutionary manifesto.\"\n\nThe NYRB had a mixed response to the novel but conceded that it was too coolly executed not to be taken seriously: \"If it is not a joke then it is madness, though not without brilliance and not without pathos.\" The reviewer seemed convinced that it was the work of Paulhan, noting (wrongly) that Paulhan's preface was in a style \"not unlike that of the novel itself.\" But this reviewer added, more tenably, that \"Pauline (Paulhan?) R\u00e9age, whoever she, he, or they may be, is surely perverse and may indeed be mad, but she or he is no fool and is as far as can be from vulgarity.\"\n\nThe Columbia University professor Albert Goldman, reviewing Story of O for the New York Times, was effusive in his praise, calling it \"a rare instance of pornography sublimed to purest art\" and describing its \"evidently pseudonymous author\" as \"a more dangerous writer than the Marquis de Sade.\" Rather than issuing propaganda or a \"call to arms,\" R\u00e9age, with her simple, direct style, aims, he argued, \"to clarify, to make real to the reader those dark and repulsive practices and emotions that his better self rejects as improbable or evil.\" Yet the critic Eliot Fremont-Smith, also writing in the Times, described the book in more ambivalent terms as \"revolting, haunting, somewhat erotic, rather more emetic, ludicrous, boring, unbelievable and quite unsettling.\" He added that it was of \"undeniable artistic interest.\"\n\nIn any case, Pauline R\u00e9age stayed silent, and Dominique Aury continued her respectable life as a cultural \u00e9minence grise. For years there were rumors, hints, and speculation connecting the two, and at some point the connection had become an open secret in literary circles\u2014yet her privacy was respected.\n\nPaulhan's daughter-in-law, Jacqueline, later claimed that she had learned the truth only at Paulhan's funeral in 1968. \"There was a very big bouquet of flowers with no name attached,\" she told a journalist. \"I was standing next to Dominique Aury, whom of course I knew well, and I remarked, 'I suppose they must be from Pauline R\u00e9age.' Dominique turned to me and said, 'Mais, Jacqueline, Pauline R\u00e9age, c'est moi.'\"\n\nDecades later, Aury offered a full and public confession. Her lover had been dead a long time. Her parents were dead. She felt she was reaching the end of her own life. There was nothing to lose, nothing at stake now.\n\nThe August 1, 1994, issue of the New Yorker printed an excerpt from a forthcoming book by the British writer John de St. Jorre, The Good Ship Venus (Venus Bound in the United States), about the infamous novels published by the Olympia Press\u2014including Vladimir Nabokov's Lolita, William S. Burroughs's Naked Lunch, and Pauline R\u00e9age's Story of O. When the author interviewed Aury for his book, he was treated to \"a double surprise\": he learned definitively that she was R\u00e9age; and he learned that the name Dominique Aury \"was itself a disguise.\" Although she asked that he not publish her actual name, the now elderly lady was otherwise ready to confess at last.\n\nSt. Jorre landed a fascinating interview with Aury, whom he described as a \"calm, clearheaded woman who answered my questions easily and with dry humor.\" She dismissed the scandal that had erupted over her novel all those years ago as \"much ado about nothing.\"\n\nIn 1975, Aury had given a long, wide-ranging interview to R\u00e9gine Deforges, an author whom Aury admired. She was interviewed as \"Pauline R\u00e9age,\" and she provided honest answers about her life and work, and her philosophical views on art, sex, war, feminism, and so on, without disclosing her true name or getting too specific in her personal anecdotes. She could open up while remaining anonymous. The interview was published in book-length form as Confessions of O, first by Pauvert in France, and then, four years later, by Viking Press in the United States. (No photo of R\u00e9age appeared in the book.) The jacket copy noted, \"In these pages one senses clearly a presence, a person, where once there had been only a pseudonym. The face may still be shrouded in mystery, but now, at last, the voice is clear, authoritative, and of a rare intelligence.\"\n\nAury never intended to give another interview after that one, so the New Yorker's profile was a coup for the reporter. It was the first time Aury admitted in public that she had written Story of O.\n\nAlthough she had led a quiet, comfortable life in the years following the publication of O, she did not entirely relinquish Pauline R\u00e9age. In 1969, she'd published a sequel of sorts, Retour \u00e0 Roissy, which included the first novel's original (unpublished) final chapter, and a third-person account (titled \"Une Fille Amoureuse,\" or \"A Girl in Love\") about the genesis of O, signed by R\u00e9age. She'd worked on it as Paulhan lay dying in a hospital room in a Paris suburb. Aury slept in his room each night for four months, until his death at eighty-three in October 1968. Later, she recalled Paulhan's extraordinary passion for life. \"Existence filled him with wonder,\" she said. \"Both the admirable and the horrible aspects of existence, equally so. The atrocious fascinated him. The enchanting enchanted him.\" One friend of Aury said that after Paulhan died, \"She pulled back from the world and lost her short-term memory.\"\n\nIt's clear from St. Jorre's profile in the New Yorker that this \"small, neat, handsome woman with gray hair and gray-blue eyes\" never recovered from the loss of Paulhan and led a fairly solitary life afterward. \"Their relationship underscored the centrality of love to life,\" St. Jorre wrote, \"the creative and destructive forces that passion can unleash, and the ease with which a human heart can be broken.\" He concluded the piece by observing that Aury had no regrets \"as her days and nights gather speed, taking her toward what she calls 'a great silence.'\" She died in 1998, at the age of ninety.\n\nAside from its major revelation, the article delved into a subplot of the saga: the pseudonymous translator of the English edition of Story of O. There was no evidence of deception, aside from the translator's suspiciously florid name, \"Sabine d'Estr\u00e9e.\" Yet no one seemed to know the mysterious woman. The Grove edition included no biographical note on her, and she mentioned in her \"Translator's Note\" that she'd never met R\u00e9age but had been \"in indirect communication (via the French publisher, Jean-Jacques Pauvert) and received the author's comments.\" Aury, in her interview with St. Jorre, told him that she had no memory of any contact with d'Estr\u00e9e, nor any idea who she (or he) might be. St. Jorre had a theory, however: the New York editor, translator, and publisher Richard Seaver.\n\nIn the early 1950s, Seaver had lived in Paris as a Fulbright scholar studying at the Sorbonne. He cofounded a literary journal that published early pieces by Jean Genet and Eug\u00e8ne Ionesco in English. He was an early champion of the then-unknown Irish playwright Samuel Beckett, and had been instrumental in arranging a book deal for Beckett with Barney Rosset (who hired Seaver). Eventually, Seaver worked his way up to editor in chief at Grove, where he was celebrated for advocating challenging and censored books. He stayed at Grove for twelve years before moving to Viking and then to Holt, Rinehart; along with his French wife, Jeannette, he founded Arcade Publishing in 1988. Jeannette's middle name is Sabine.\n\nSt. Jorre's attempt to extract information from Seaver himself went nowhere. Seaver insisted that he'd been sworn to secrecy about d'Estr\u00e9e's identity but told St. Jorre that he would seek permission from d'Estr\u00e9e\u2014whom he called a \"very shy, secretive person\"\u2014and get back to him. He never did.\n\nSo the journalist did his own research, carefully going through the Grove Press correspondence archive at Syracuse University Library's Special Collections Department. He found it curious that there were variant spellings of \"d'Estr\u00e9e,\" and that one letter purporting to be from d'Estr\u00e9e herself requested that all payments be addressed to an attorney in Manhattan, Seymour Litvinoff. After St. Jorre tracked down the lawyer, Litvinoff said that had represented both Seaver and d'Estr\u00e9e, but \"I cannot say who Sabine is. I don't know who she is.\"\n\nSt. Jorre also discovered that d'Estr\u00e9e had continued to translate French erotica\u2014at least four other books\u2014in collaboration with Seaver, who kept \"hiring\" her even after changing publishing jobs. She did translation work for no one else. Seaver was long believed to be d'Estr\u00e9e, but he kept quiet about it. The mystery was solved in January 2009, when he died of a heart attack at the age of eighty-two. His wife finally confessed. \"He wanted people to guess,\" Jeannette told a reporter. \"But yes, he did it.\"\n\nSeaver's stint as d'Estr\u00e9e has been largely forgotten, but the novel still resonates around the world, affecting readers in ways that are deeply personal. \"Ever since I remember,\" an anonymous American woman admitted on an online message board, \"I have always used some form of power exchange fantasy in masturbation. I had no words for it, no framework, and O was the first book to provide that.\"\n\nStory of O also influenced writers of erotica for decades after its release, though it set a standard that few, if any, could meet. The person who could perhaps claim the closest literary kinship with R\u00e9age is the contemporary author and art critic Catherine Millet, whose \"autobiography,\" Sexual Life of Catherine M., was published in 2002. Edmund White went so far as to call it \"the most explicit book about sex ever written by a woman.\" The book detailed the author's early experiences with masturbation and her abiding fondness for orgies (in which she began to dabble at the age of eighteen), sex in public places, and so on. She said that of her countless lovers, mostly men, she would be able to recognize at best only fifty faces or names. The book was cast as a memoir, but J. G. Ballard wondered if it was \"the most original novel of the year.\"\n\nAlthough Millet's sexual proclivities hardly mirrored those of R\u00e9age\u2014Millet did admit to enjoying having her nipples pinched, along with more aggressive forms of sex\u2014their profiles were strikingly alike. Both women published confessional, shockingly graphic books in midlife. Millet is French, and by day she, too, is a bourgeois intellectual who appears respectable enough. Her book, like O, was well written and even literary. As Jenny Diski noted in the London Review of Books, Millet \"anatomises her sexual experiences and responses as a Cubist might the visual field.\" That Millet's project was both intellectual and sexual (and possibly even spiritual) calls R\u00e9age to mind yet again. \"[Millet] takes her radical philosophy from Bataille, and admires Pauline R\u00e9age's \u00fcber-underling O for her perpetual readiness for sex, her propensity for being sodomised and her reclusiveness,\" Diski wrote. Millet, like R\u00e9age, feels no guilt about her sexual life, and similarly writes about sex \"as plainly as if she were a housewife describing her domestic round.\"\n\nHad R\u00e9age not published Story of O, perhaps Millet could not have published Sexual Life, at least not under her own name. Aury had endured stigma and shame and had emerged a success. That legacy gave Millet license to tell her story. And it explains why Jane Juska, for example, could celebrate, in A Round Heeled Woman, the pleasures of promiscuous geriatric sex via the NYRB classified ads. It also freed a young woman, Melissa Panarello (known as \"Melissa P.\"), to publish an erotic autobiographical novel in 2004. Called One Hundred Strokes of the Brush Before Bed, it chronicled, in diary form, group sex, S&M, and other experiences. This book was an immediate best seller in the author's native Italy, and was hailed as \"a Story of O for our times.\"\n\nO's enduring significance was evident on the fiftieth anniversary of its debut, when the French government proudly announced that Histoire d'O would be included on a list of \"national triumphs\" to be celebrated that year.\n\nTwo years later, in 2006, R\u00e9age's works were part of an auction at Christie's in Paris, featuring the \"Biblioth\u00e8que Erotique\" of G\u00e9rard Nordmann, a businessman in Geneva who had assembled a library of almost two thousand erotic manuscripts and rare books. An edition of O from its limited first run of six hundred copies was cited as \"First edition of the most important erotic novel of the postwar period.\" Another lot by R\u00e9age was described as \"[the] complete holograph working manuscript in pencil and ballpoint of what is arguably the finest erotic novel (1954) of the post-war period and its sequel (1969), which describe unconditional love as total sexual submission carried to its ultimate consequence.\" The manuscripts of Histoire d'O and Retour \u00e0 Roissy sold for $127,000.\n\nIn assessing the life and work of Dominique Aury, it is striking how brave she was to risk everything for the man she loved. The pseudonym could have been exposed early on, destroying her reputation and wrecking friendships. \"If you care enough about something, you have to pay the price,\" Aury once said. Hers was a life without compromise, highly moral, and one lived without regret.\n\nAfter Aury's death in a suburb south of Paris, a longtime friend declared it unremarkable that the author had hoarded a nom de plume for so long. \"Everyone is double, or triple, or quadruple,\" she said. \"Every character has its hidden sides. One doesn't reveal one's secrets to all.\"\nAcknowledgments\n\nAbove all, I cannot thank enough the amazing Tina Bennett. Without her, there would be no book, or it would exist only in my head. I'm endlessly grateful for her wisdom, kindness, patience, and enthusiasm, and for always laughing at my jokes. I feel very lucky.\n\nThanks to Gillian Blake for acquiring this book, then nudging and nurturing it along.\n\nTo all at HarperCollins\u2014the wonderful Rakesh Satyal, Katie Salisbury, David Koral, Heather Drucker, and Leah Wasielewski\u2014thank you for everything.\n\nAlso at Janklow & Nesbit, I'm grateful to Stephanie Koven and to Svetlana Katz (a real-life superhero). Dorothy Irwin provided expertise at crucial moments. And I did nothing to earn such generosity from Amy Grace Loyd, but there it was, anyway. The phenomenal Nicholas Latimer is busy enough with his many responsibilities at Knopf, yet he gave freely of his time, his brilliant advice, and more. Whenever I tell people that I know Gretchen Koss, they usually respond by saying, \"She's the best.\" It's true, and having her help has been a great gift.\n\nLove and gratitude also to Alice Quinn and Laurie Kerr; Michelle Williams; Devon Hodges, Eric Swanson, Tristan Swanson, and Cecily Swanson; the Rosabals; my dear friends at 37 Montgomery Place; and above all, Sarah (and her parents, Rosalind and Colin) and Oscar and Freddy Fitzharding.\nTime Line\n\nGeorge Sand born | 1804\n\n|   \n---|---|---\n\nCharlotte Bront\u00eb born | 1816\n\n|\n\nEmily Bront\u00eb born | 1818\n\n|\n\nGeorge Eliot born | 1819\n\n|\n\nAnne Bront\u00eb born | 1820\n\n|\n\nLewis Carroll born | 1832\n\n|\n\nMark Twain born | 1835\n\n|   \n|\n\n1848 | Emily Bront\u00eb dies\n\n|\n\n1849 | Anne Bront\u00eb dies\n\n|\n\n1855 | Charlotte Bront\u00eb dies\n\nO. Henry born | 1862\n\n|   \n|\n\n1876 | George Sand dies\n\n|\n\n1880 | George Eliot dies\n\nIsak Dinesen born | 1885\n\n|\n\nFernando Pessoa born | 1888\n\n|   \n|\n\n1898 | Lewis Carroll dies\n\nGeorges Simenon born | 1903\n\n|\n\nGeorge Orwell born | 1903\n\n|\n\nHenry Green born | 1905\n\n|\n\nPauline R\u00e9age born | 1907\n\n|   \n|\n\n1910 | Mark Twain dies\n\n|\n\n1916 | O. Henry dies\n\nRomain Gary born | 1914\n\n|\n\nAlice Sheldon born | 1915\n\n|\n\nPatricia Highsmith born | 1921\n\n|\n\nSylvia Plath born | 1932\n\n|   \n|\n\n1935 | Fernando Pessoa dies\n\n|\n\n1950 | George Orwell dies\n\n|\n\n1962 | Isak Dinesen dies\n\n|\n\n1963 | Sylvia Plath dies\n\n|\n\n1973 | Henry Green dies\n\n|\n\n1980 | Romain Gary dies\n\n|\n\n1987 | Alice Sheldon dies\n\n|\n\n1989 | Georges Simenon dies\n\n|\n\n1995 | Patricia Highsmith dies\n\n|\n\n1998 | Pauline R\u00e9age dies\nBibliography\n\nThe sources below were invaluable to me in researching and writing this book. Dates refer to the editions used, rather than the date of first publication. My supplemental research sources\u2014hundreds of archival magazine, journal, and newspaper articles\u2014are far too extensive to be cited in full here. Any source errors or omissions are wholly unintentional and will be corrected in future editions.\n\nAnne, Charlotte, and Emily Bront\u00eb & Acton, Currer, and Ellis Bell\n\nBarker, Juliet. The Bront\u00ebs: A Life in Letters. New York: Viking, 1997.\n\nBentley, Phyllis. The Bront\u00ebs. London: Thames and Hudson, 1997.\n\nBront\u00eb, Charlotte. Jane Eyre. New York: Penguin, 2003.\n\n\u2014\u2014\u2014. Villette. New York: Penguin, 2004.\n\nBront\u00eb, Emily Jane. The Complete Poems. New York: Penguin, 1992.\n\nFraser, Rebecca. Charlotte Bront\u00eb: A Writer's Life. New York: Pegasus, 2008.\n\nGaskell, Elizabeth. The Life of Charlotte Bront\u00eb. New York: Penguin, 1997.\n\nGordon, Lyndall. Charlotte Bront\u00eb: A Passionate Life. New York: Norton, 1994.\n\nMiller, Lucasta. The Bront\u00eb Myth. New York: Knopf, 2003.\n\nGeorge Sand & Aurore Dupin\n\nJack, Belinda. George Sand: A Woman's Life Writ Large. New York: Vintage, 2001.\n\nMaurois, Andr\u00e9. L\u00e9lia: The Life of George Sand, trans. Gerard Hopkins. New York: Harper, 1954.\n\nSand, George. Lucrezia Floriani. Chicago: Academy Chicago, 1985.\n\n\u2014\u2014\u2014. Marianne. New York: Carroll and Graf, 1998.\n\n\u2014\u2014\u2014. Story of My Life: The Autobiography of George Sand, ed. Thelma Jurgrau. Albany: State University of New York Press, 1991.\n\nGeorge Eliot & Marian Evans\n\nBooth, Alison. Greatness Engendered: George Eliot and Virginia Woolf. Ithaca, NY: Cornell University Press, 1992.\n\nEliot, George. Brother Jacob. London: Virago, 1989.\n\n\u2014\u2014\u2014. Daniel Deronda. New York: Penguin, 1986.\n\n\u2014\u2014\u2014. The Lifted Veil. London: Virago, 1985.\n\n\u2014\u2014\u2014. Middlemarch. New York: Modern Library, 2000.\n\n\u2014\u2014\u2014. The Mill on the Floss. New York: Penguin, 1985.\n\n\u2014\u2014\u2014. Scenes of Clerical Life. New York: Penguin, 1998.\n\n\u2014\u2014\u2014. Silas Marner. New York: Penguin, 1996.\n\nHanson, Lawrence, and Elisabeth Hanson. Marian Evans and George Eliot: A Biography. New York: Oxford University Press, 1952.\n\nHughes, Kathryn. George Eliot: The Last Victorian. New York: Farrar, Straus and Giroux, 1999.\n\nMaddox, Brenda. George Eliot in Love. New York: Palgrave Macmillan, 2009.\n\nLewis Carroll & Charles Dodgson\n\nBassett, Lisa. Very Truly Yours, Charles L. Dodgson, Alias Lewis Carroll. New York: Lothrop, Lee and Shepard, 1987.\n\nCarroll, Lewis. Alice's Adventures in Wonderland. New York: Bloomsbury, 2001.\n\n\u2014\u2014\u2014. Alice's Adventures in Wonderland and Through the Looking Glass. New York: Puffin, 1962.\n\nCohen, Morton N. Lewis Carroll: A Biography. New York: Vintage, 1995.\n\nWilson, Robin. Lewis Carroll in Numberland: His Fantastical Mathematical Logical Life. New York: Norton, 2008.\n\nWoolf, Jenny. The Mystery of Lewis Carroll: Discovering the Whimsical, Thoughtful, and Sometimes Lonely Man Who Created Alice in Wonderland. New York: St. Martin's, 2010.\n\nMark Twain & Samuel Clemens\n\nFishkin, Shelley Fisher, ed. The Mark Twain Anthology: Great Writers on His Life and Works. New York: Library of America, 2010.\n\nGrant, Douglas. Mark Twain. New York: Grove, 1962.\n\nKaplan, Justin. Mr. Clemens and Mark Twain. New York: Simon and Schuster, 1966.\n\nNeider, Charles, ed. Life As I Find It: A Treasury of Mark Twain Rarities. New York: Cooper Square, 2000.\n\nPowers, Ron. Mark Twain: A Life. New York: Free Press, 2005.\n\nQuirk, Thomas, ed. The Portable Mark Twain. New York: Penguin, 2004.\n\nTrombley, Laura. Mark Twain's Other Woman: The Hidden Story of His Final Years. New York: Knopf, 2010.\n\nTwain, Mark. Adventures of Huckleberry Finn. Norton Critical Edition, ed. Sculley Bradley et al. New York: Norton, 1977.\n\n\u2014\u2014\u2014. The Adventures of Tom Sawyer. New York, Penguin, 1986.\n\nWard, Geoffrey C., and Dayton Duncan. Mark Twain: An Illustrated Biography. New York: Knopf, 2001.\n\nO. Henry & William Sydney Porter\n\nHenry, O. The Complete Works. New York: Doubleday, 1928.\n\nLangford, Gerald. Alias O. Henry. New York: Macmillan, 1957.\n\nO'Connor, Richard. O. Henry: The Legendary Life of William S. Porter. New York: Doubleday, 1970.\n\nSmith, C. Alphonso. O. Henry, Biography. New York: Doubleday, Page, 1916.\n\nStuart, David. O. Henry: A Biography of William Sydney Porter. Chelsea, MI: Scarborough House, 1990.\n\nFernando Pessoa & His Heteronyms\n\nBorges, Jorge Luis. On Writing. New York: Penguin, 2010.\n\nMonteiro, George, ed. The Man Who Never Was: Essays on Fernando Pessoa. Providence, RI: G\u00e1vea-Brown, 1981.\n\n\u2014\u2014\u2014, ed. The Presence of Pessoa: English, American, and Southern African Literary Responses. Lexington: University Press of Kentucky, 1998.\n\nPessoa, Fernando. Always Astonished: Selected Prose, ed. and trans. Edwin Honig. San Francisco: City Lights, 1988.\n\n\u2014\u2014\u2014. The Book of Disquiet: A Selection, trans. Iain Watson. London: Quartet, 1991.\n\n\u2014\u2014\u2014. The Book of Disquiet, ed. and trans. Richard Zenith. New York: Penguin, 2003.\n\n\u2014\u2014\u2014. The Education of the Stoic: The Only Manuscript of the Baron of Teive, ed. and trans. Richard Zenith. Cambridge: Exact Change, 2005.\n\n\u2014\u2014\u2014. Fernando Pessoa: A Centenary Pessoa, ed. Eug\u00e9nio Lisboa and L. C. Taylor. Carcanet, 2003.\n\n\u2014\u2014\u2014. Fernando Pessoa & Co.: Selected Poems, ed. and trans. Richard Zenith. New York: Grove, 1998.\n\n\u2014\u2014\u2014. A Little Larger Than the Entire Universe: Selected Poems, ed. and trans. Richard Zenith. New York: Penguin, 2006.\n\n\u2014\u2014\u2014. The Selected Prose of Fernando Pessoa, ed. and trans. Richard Zenith. New York: Grove, 2001.\n\nGeorge Orwell & Eric Blair\n\nBowker, Gordon. George Orwell. London: Abacus, 2003.\n\nCrick, Bernard. George Orwell: A Life. New York: Penguin, 1982.\n\nCrick, Bernard, and Audrey Coppard. Orwell Remembered. New York: Facts on File, 1984.\n\nDavison, Peter. George Orwell: A Literary Life. London: Macmillan, 1998.\n\nOrwell, George. Animal Farm. New York: Plume, 2003.\n\n\u2014\u2014\u2014. Down and Out in Paris and London. New York: Harcourt, 1961.\n\n\u2014\u2014\u2014. Essays. New York: Everyman's Library, 2002.\n\n\u2014\u2014\u2014. Facing Unpleasant Facts: Narrative Essays, ed. George Packer. New York: Mariner, 2009.\n\n\u2014\u2014\u2014. Why I Write. New York: Penguin, 2004.\n\nIsak Dinesen & Karen Blixen\n\nDinesen, Isak. Anecdotes of Destiny and Ehrengard. New York: Vintage, 1993.\n\n\u2014\u2014\u2014. Last Tales. New York: Vintage, 1991.\n\n\u2014\u2014\u2014. Letters from Africa 1914\u20131931, ed. Frans Lasson. Chicago: University of Chicago Press, 1984.\n\n\u2014\u2014\u2014. Out of Africa and Shadows on the Grass. New York: Vintage, 1989.\n\n\u2014\u2014\u2014. Seven Gothic Tales. New York: Modern Library, 1934.\n\n\u2014\u2014\u2014. Winter's Tales. New York: Vintage, 1993.\n\nHenriksen, Aage. Isak Dinesen\/Karen Blixen: The Work and the Life. New York: St. Martin's, 1988.\n\nThurman, Judith. Isak Dinesen: The Life of a Storyteller. New York: St. Martin's, 1982.\n\nSylvia Plath & Victoria Lucas\n\nKendall, Tim. Sylvia Plath: A Critical Study. London: Faber and Faber, 2001.\n\nPlath, Sylvia. The Bell Jar. New York: Harper and Row, 1971.\n\n\u2014\u2014\u2014. The Bell Jar. New York: HarperPerennial, 1999.\n\n\u2014\u2014\u2014. Letters Home by Sylvia Plath: Correspondence 1950\u20131963, ed. Aurelia Schober Plath. New York: Harper and Row, 1975.\n\n\u2014\u2014\u2014. The Unabridged Journals of Sylvia Plath, ed. Karen V. Kukil. New York: Anchor, 2000.\n\nRose, Jacqueline. The Haunting of Sylvia Plath. Cambridge, MA: Harvard University Press, 1991.\n\nStevenson, Anne. Bitter Fame: A Life of Sylvia Plath. London: Viking, 1989.\n\nWagner, Erica. Ariel's Gift. New York: Norton, 2001.\n\nHenry Green & Henry Yorke\n\nGreen, Henry. Loving; Living; Party Going. New York: Penguin, 1993.\n\n\u2014\u2014\u2014. Nothing. New York: Dalkey Archive, 2000.\n\n\u2014\u2014\u2014. Nothing; Doting; Blindness. New York: Penguin, 1993.\n\n\u2014\u2014\u2014. Pack My Bag: A Self-Portrait. New York: New Directions, 2004.\n\n\u2014\u2014\u2014. Party Going. London: Vintage, 2000.\n\n\u2014\u2014\u2014. Surviving: The Uncollected Writings of Henry Green, ed. Matthew Yorke. New York: Viking, 1993.\n\nRussell, John. Henry Green: Nine Novels and an Unpacked Bag. New Brunswick, NJ: Rutgers University Press, 1960.\n\nTreglown, Jeremy. Romancing: The Life and Work of Henry Green. New York: Random House, 2001.\n\nRomain Gary & \u00c9mile Ajar\n\nBellos, David. Romain Gary: A Tall Story. London: Harvill Secker, 2010.\n\nGary, Romain (\u00c9mile Ajar). Hocus Bogus, trans. David Bellos. New Haven and London: Yale University Press, 2010.\n\nGary, Romain. The Life Before Us (Madame Rosa), trans. Ralph Manheim. New York: New Directions, 1977, 1978.\n\n\u2014\u2014\u2014. Promise at Dawn: A Memoir, trans. John Markham Beach. New York: New Directions, 1961.\n\n\u2014\u2014\u2014. White Dog. Chicago: University of Chicago Press, 1970.\n\nSchoolcraft, Ralph. Romain Gary: The Man Who Sold His Shadow. Philadelphia: University of Pennsylvania Press, 2002.\n\nJames Tiptree, Jr. & Alice Sheldon\n\nPhillips, Julie. James Tiptree, Jr.: The Double Life of Alice B. Sheldon. New York: St. Martin's, 2006.\n\nTiptree, James, Jr., Meet Me at Infinity. New York: Tom Doherty Associates, 2000.\n\nGeorges Simenon & Christian Brulls et al.\n\nAssouline, Pierre. Simenon: A Biography. New York: Knopf, 1997.\n\nCarter, David. The Pocket Essential Georges Simenon. Chicago: Trafalgar Square, 2003.\n\nMarnham, Patrick. The Man Who Wasn't Maigret: A Portrait of Georges Simenon. Harvest, 1992.\n\nSimenon, Georges. Pedigree. New York: New York Review Books, 2010.\n\nPatricia Highsmith & Claire Morgan\n\nHighsmith, Patricia. Carol. London: Bloomsbury, 1990.\n\n\u2014\u2014\u2014. The Selected Stories of Patricia Highsmith. New York: Norton, 2001.\n\n\u2014\u2014\u2014. Strangers on a Train. New York: Norton, 2001.\n\n\u2014\u2014\u2014. The Talented Mr. Ripley, Ripley Under Ground, Ripley's Game. New York: Knopf\/Everyman's Library, 1999.\n\nMeaker, Marijane. Highsmith: A Romance of the 1950s. San Francisco: Cleis, 2003.\n\nSchenkar, Joan. The Talented Miss Highsmith: The Secret Life and Serious Art of Patricia Highsmith. New York: St. Martin's, 2009.\n\nWilson, Andrew. Beautiful Shadow: A Life of Patricia Highsmith. London: Bloomsbury, 2003.\n\nPauline R\u00e9age & Dominique Aury\n\nDeforges, R\u00e9gine. Confessions of O: Conversations with Pauline R\u00e9age. New York: Viking, 1979.\n\nR\u00e9age, Pauline. Return to the Ch\u00e2teau. New York: Ballantine, 1995.\n\n\u2014\u2014\u2014. Story of O. New York: Ballantine, 1973.\n\nSontag, Susan. Styles of Radical Will. New York: Picador, 2002.\n\nBond, Jenny, and Chris Sheedy. Who the Hell Is Pansy O'Hara?: The Fascinating Stories Behind 50 of the World's Best-loved Books. New York: Penguin, 2008.\n\nGross, John, ed. The New Oxford Book of Literary Anecdotes. New York: Oxford University Press, 2006.\n\nHalpern, Daniel, ed. Who's Writing This? New York: Ecco, 1995.\n\nHamilton, Nigel. Biography: A Brief History. Cambridge, MA: Harvard University Press, 2007.\n\nJones, Emma. The Literary Companion. London: Think, 2004.\n\nMar\u00edas, Javier. Written Lives, trans. Margaret Jull Costa. New York: New Directions, 2007.\n\nMotion, Andrew, ed. Interrupted Lives. London: National Portrait Gallery Publications, 2004.\n\nMullan, John. Anonymity. Princeton, NJ: Princeton University Press, 2007.\n\nProse, Francine. The Lives of the Muses: Nine Women and the Artists They Inspired. New York: HarperCollins, 2002.\n\nRose, Phyllis. Parallel Lives: Five Victorian Marriages. New York: Vintage, 1983.\n\nSchnakenberg, Robert. Secret Lives of Great Authors. Philadelphia: Quirk, 2008.\n\nShowalter, Elaine. A Literature of Their Own: British Women Novelists from Bront\u00eb to Lessing. Princeton, NJ: Princeton University Press, 1999.\n\nWoolf, Virginia. A Room of One's Own. New York: Harcourt, 1989.\nEpigraph\n\nWho is it that can tell me who I am?\n\n\u2014KING LEAR\nAbout the Author\n\nCARMELA CIURARU is not a pseudonym. Her anthologies include First Loves: Poets Introduce the Essential Poems That Captivated and Inspired Them and Solitude Poems. She is a graduate of Columbia University's School of Journalism and a member of the National Book Critics Circle. She has written for the Los Angeles Times, the San Francisco Chronicle, the Boston Globe, the Wall Street Journal, Newsday, Elle Decor, ARTNews, O, The Oprah Magazine, and other publications. She lives in Brooklyn.\n\nwww.carmelaciuraru.com\n\nVisit www.AuthorTracker.com for exclusive information on your favorite HarperCollins authors.\nCredits\n\nCover design by Jarrod Taylor\nCopyright\n\nNOM DE PLUME. Copyright \u00a9 2011 by Carmela Ciuraru. All rights reserved under International and Pan-American Copyright Conventions. By payment of the required fees, you have been granted the nonexclusive, nontransferable right to access and read the text of this e-book on-screen. No part of this text may be reproduced, transmitted, downloaded, decompiled, reverse-engineered, or stored in or introduced into any information storage and retrieval system, in any form or by any means, whether electronic or mechanical, now known or hereinafter invented, without the express written permission of HarperCollins e-books.\n\nEPub Edition \u00a9 JUNE 2011 ISBN: 9780062109569\n\nFIRST EDITION\n\nLibrary of Congress Cataloging-in-Publication Data\n\nCiuraru, Carmela.\n\nNom de plume: a (secret) history of pseudonyms \/ Carmela Ciuraru.\n\np. cm.\n\nIncludes bibliographical references.\n\nISBN: 978-0-06-173526-4\n\n1. Anonyms and pseudonyms\u2014History. I. Title.\n\nZ1041.C55 2011\n\n929.4\u2014dc22\n\n2010053603\n\n10 9 8 7 6 5 4 3 2 1\nAbout the Publisher\n\nAustralia\n\nHarperCollins Publishers (Australia) Pty. Ltd.\n\n25 Ryde Road (P.O. Box 321)\n\nPymble, NSW 2073, Australia\n\n<http:\/\/www.harpercollins.com.au\/ebooks>\n\nCanada\n\nHarperCollins Canada\n\n2 Bloor Street East - 20th Floor\n\nToronto, ON, M4W, 1A8, Canada\n\n<http:\/\/www.harpercollins.ca>\n\nNew Zealand\n\nHarperCollins Publishers (New Zealand) Limited\n\nP.O. Box 1\n\nAuckland, New Zealand\n\n<http:\/\/www.harpercollins.co.nz>\n\nUnited Kingdom\n\nHarperCollins Publishers Ltd.\n\n77-85 Fulham Palace Road\n\nLondon, W6 8JB, UK\n\n<http:\/\/www.harpercollins.co.uk>\n\nUnited States\n\nHarperCollins Publishers Inc.\n\n10 East 53rd Street\n\nNew York, NY 10022\n\n<http:\/\/www.harpercollins.com>\n","meta":{"redpajama_set_name":"RedPajamaBook"}}
+{"text":" \n**ON**\n\n**SEXUALITY**\n\n**AND**\n\n**POWER**\n\nBetween Men \u223c Between Women\n**On**\n\n**Sexuality**\n\n**and**\n\n**Power**\n\nALAN SINFIELD\n\nColumbia University Press New York\n\n**COLUMBIA UNIVERSITY PRESS**\n\n_Publishers Since 1893_\n\nNew York Chichester, West Sussex\n\ncup.columbia.edu\n\nCopyright \u00a9 2004 Columbia University Press\n\nAll rights reserved\n\nE-ISBN 978-0-231-50866-7\n\nLibrary of Congress Cataloging-in-Publication Data\n\nSinfield, Alan.\n\nOn sexuality and power \/ Alan Sinfield.\n\np. cm\u2014(Between men\u2014between women)\n\nIncludes bibliographical references and index.\n\nISBN 0\u2013231\u201313408\u20138 (cloth : alk. paper)\n\nISBN 0\u2013231\u201313409\u20136 (pbk. : alk. paper)\n\n1. Homosexuality, Male\u2014Psychological aspects.\n\n2. Power (Social sciences) 3. Sex (Psychology)\n\n4. Gays in literature. 5. Homosexuality and literature.\n\nI Title. II. Series.\n\nHQ76.S545 2004\n\n306.76\u20326\u2014dc22\n\n2004052773\n\nA Columbia University Press E-book.\n\nCUP would be pleased to hear about your reading experience with this e-book at cup-ebook@columbia.edu.\n\n_Designed by Lisa Hamm_\n**BETWEEN MEN \u223c BETWEEN WOMEN**\n\n_Lesbian, Gay, and Bisexual Studies_\n\nTerry Castle and Larry Gross, Editors\n\nAdvisory Board of Editors\n\nClaudia Card\n\nJohn D'Emilio\n\nEsther Newton\n\nAnne Peplau\n\nEugene Rice\n\nKendall Thomas\n\nJeffrey Weeks\n\nBetween Men \u223c Between Women is a forum for current lesbian and gay scholarship in the humanities and social sciences. The series includes both books that rest within specific traditional disciplines and are substantially about gay men, bisexuals, or lesbians and books that are interdisciplinary in ways that reveal new insights into gay, bisexual, or lesbian experience, transform traditional disciplinary methods in consequence of the perspectives that experience provides, or begin to establish lesbian and gay studies as a freestanding inquiry. Established to contribute to an increased understanding of lesbians, bisexuals, and gay men, the series also aims to provide through that understanding a wider comprehension of culture in general.\n**CONTENTS**\n\n_Acknowledgments_\n\n**1**\n\nIntroduction\n\n**2**\n\nTaxonomies\n\n**3**\n\nFantasy\n\n**4**\n\nPower\n\n**5**\n\nGender\n\n**6**\n\nAge\n\n**7**\n\nClass\n\n**8**\n\nRace\n\n**9**\n\nFiction\n\n_Notes_\n\n_Index_\n**ACKNOWLEDGMENTS**\n\nI am grateful for books, comments, and conversations to Peter Burton, Alex Evans, John Fletcher, Gowan Hewlett, Ann Rosalind Jones, Vicky Lebeau, Linda Logie, David Marriott, Vincent Quinn, Nick Rees-Roberts, David Rogers, Lynne Segal, Mark Sinfield, Peter Stallybrass, Chris West; students on the Sexual Dissidence and Cultural Change program at the University of Sussex. They are not responsible for the ideas.\n\nParts of this book have appeared in other forms in _Critical Inquiry_ and _GLQ (Gay and Lesbian Quarterly)_ ; Richard Phillips, Diane Watt, and David Shuttleton, eds., _De-centering Sexualities_ (London: Routledge, 2000); and David Alderson and Linda Anderson, eds., _Territories of Desire in Queer Culture_ (Manchester: Manchester University Press, 2000).\n\nGregory Woods's poem, \"Under where Catullus toyed,\" is quoted with permission from the Carcanet Press. John Minton's painting \"Painter and Model\" is used on the cover with permission from the Royal College of Art and the Russell-Cotes Art Gallery and Museum, Bournemouth, England.\n**1**\n\n**INTRODUCTION**\n\nReginald Shepherd is, he says, the person no one wants to know about: a black (African American) gay man with an unappeasable attraction to white men. \"I am in love with the image and idea of white manhood, which is everything I am not and want to be.\" Why is this such a fearful condition? Because of the historic oppression of blacks by whites. Shepherd is under no illusion about the role of power in his attraction: \"I think many gay men worship the power that oppresses them; I think too that all sexual relations in our society are about power over another or the submission to the power of another.\"\n\nShepherd's sexual desire is hinged on to racial difference, the most fraught political, economic, social, and cultural issue in the United States and Britain, and in many other countries. At this juncture, unavoidably, the psychic meets the social, fantasy meets history, desire meets politics. No wonder our societies find the subject hot to handle. We experience a marked unease about all hierarchical liaisons\u2014not only of race but also of age, gender, and class. In metropolitan contexts today, it is often said, gay people favor egalitarian relations. Shepherd himself seeks commonality in everything but racial difference: his dream lover is \"some beautiful cultured blond named Troy with whom I'd have everything in common, everything but that\" (56).\n\nYet power differentials are remarkably persistent, in gay fantasies and in the stories about gayness that circulate. I discern three reasons for this. One is that, while we may like to think of fantasy as free-ranging, in fact it often shows astonishing fixity. Shepherd's desire is by no means comfortable, but it appears ineluctable. \"So I hate him and desire him, fearing him and myself, too often despising both. So I continue to want him\" (56). Second, our desires are not ours alone; they are embedded in the power structures that organize our social being. Shepherd would like to step out of his personal history, but that would be \"to step out of the history of the nation.\" It would be to say: \"This is not America.\" The entire social system is tending to maintain his fantasy. And, anyway, who would he then be? \"The catch is that 'myself' is a product of that history both general and very particular.... What and who would I be without the burden of the past?\" (57; my elision).\n\nA third reason for the persistence of hierarchy in relationships, I will argue, is that _it is sexy_. Racial difference is not just a major influence on or component in Shepherd's sexuality, it is intricately implicated with what turns him on. \"I am in love with the image and idea of white manhood, which is everything I am not and want to be, and if I cannot be that at least I can have that, if only for the night, if only for the week or the month\" (54\u201355). For Shepherd, whiteness is sexy, despite or because of its bond with oppression.\n\nIt would be absurdly presumptuous to suppose that a (white) commentator might begin to unravel Shepherd's intimate experience. However, he has placed it, boldly, in the public domain, and I can record that it is there. A project of this book is to reexamine the implication of sexuality and power. I propose a framework for this in chapter 4, and continue with chapters that explore binary relations founded in gender, age, class, and race. These formations, as Jonathan Dollimore says, quoting Jacques Derrida, are \"violent hierarchies\": \"one of the two terms forcefully governs the other.\" They structure our societies; they constitute our psyches also.\n\nThe late and still-lamented Allon White, a keen Freudian, and I were talking about the unconscious. I was attempting to substantiate a materialist position which I associated with Jean-Paul Sartre, namely that human behavior may be explained without resort to the idea of a reservoir of unconscious repressed experience which could only be reached through the ministrations of the analyst. In order that traumatic experiences may be repressed into the unconscious, they have first to reach the threshold of the conscious, which is where they are censored as unacceptable. The images which Freud dredged up in his hysterical patients were not unavailable: rather, the patients did not want to admit to them. So there is no reason to posit an unconscious as such. \"That's right,\" cried Allon, puckishly. \"Some people have no unconscious. And you're one of them.\" I felt that at some not very subtle level I had lost the debate. Even a materialist needs a theory of psychic life. Chapters 2 and  of this book make a start on sketching what the terms for that might be.\n\nMy approach is broadly _cultural materialist_. It seeks to raise questions about cultural production, addressing typically the relations between dominant and subordinate cultures; the scope for dissidence; how far male dominance might be able to accommodate feminism and dissident sexualities; how culture is negotiated through institutions, including those which govern the definition and circulation of art and literature; how subcultural groups constitute themselves in and through culture. Cultural materialism is a kind of Marxism, and does not pretend to political neutrality. It looks for the transformation of an exploitative social order.\n\nCultural materialists believe that historic forces and the power structures that they sustain determine the direction, not just of our societies, but ultimately of our selfhood. Yet there is little agreement on how to describe the links between the psyche and the social order. Raymond Williams, who announced the main themes of cultural materialism in the 1970s and 1980s, placed his understanding of personal attachments in his novels, rather than his theory. Louis Althusser seeks to explain how the human subject is persuaded to recognize him- or herself as hailed (interpellated) by the norms of society, while experiencing him- or herself as a free individual. This sounds like the beginnings of a theory of subjectivity and psychic life. However, Althusser is ready to cede the individual psyche to Freud and his followers. He accepts that the unconscious and its effects are the specific object of psychoanalysis. \"History, 'sociology' or anthropology have no business here.\" Nevertheless, questions remain, for Althusser, about how historical origins and socioeconomic conditions affect psychoanalytical theory and technique.\n\nThe most ambitious theory has sought to combine Freudian and Marxian principles. Herbert Marcuse in _Eros and Civilization_ (1955) sought to reverse what was usually taken as a central Freudian principle\u2014that repression of natural instincts is a necessary condition for the development of civilization. Marcuse argued that while some repression is inevitable, modern societies experience _surplus repression_ , that is, repression in the service of domination. Further, what is specially repressed is sexuality, and especially insofar as it is restricted to genital organs and reproductive purposes. Marcuse wants us to retrieve the _polymorphous perversity_ that we experienced as infants: this, he holds, will threaten the overthrow of the order of procreation and the institutions that guarantee it. This was a theory for the exuberant and optimistic 1960s. Writing a little later, in 1976, Michel Foucault is more aware of the \"polymorphous techniques of power.\" A Marcusean \"great Refusal\" of the system supposes power to be unitary and solely oppressive. Rather, the opposite has come about: \"there is a plurality of resistances.\"\n\nAnyway, rather than repressing sex, Foucault points out, we have talked of almost nothing else, since Victorian times. This \"veritable discursive explosion\" (96) has been the cause and effect of diverse taxonomies of sexual and gender dissidence, especially in the traditions of anthropology and psychoanalysis; I mean to draw upon both. Anthropologists have observed structural correlations in diverse societies: same-sex passions, they find, are organized through age difference (commonly, an older partner initiating a younger partner) and gender difference (one partner takes the role of a man, the other of a woman). I have added race and class to age and gender.\n\nFor feminists, and lesbian and gay activists, psychoanalytic texts are fraught with problems. My emphasis on age, class, and race, as well as gender, has to resist a Freudian propensity to suppose that proper analysis of any sexual relation is likely to disclose a version of the ultimate drama through which the castrating father (allegedly) installs gender hierarchy. Consider Joan Riviere's much-cited essay, \"Womanliness as a Masquerade.\" An analysand who grew up in the American South reports a dream in which she is threatened by a black man, and resists him with the secret intention of attracting him. Riviere comments: \"The meaning was that she had killed father and mother and obtained everything for herself (alone in the house), [and] became terrified of their retribution.\" Although she mentions the American South, Riviere's interpretation involves no thought of the horrific violence that was enacted against black men accused of violating white women. One allegedly hostile man is pretty much like another, and it all comes back to your father.\n\nOther commentators have made a similar point. Michael Warner demands: \"Why is gender assumed to be our only access to alterity? It is not even the only line of sameness and difference that structures erotic images. Race, age, and class are capable of doing that as well.\" Lynne Segal traces a difficulty in articulating class to Lacanian psychoanalysis and points out that some feminists \"have addressed the significance of differences other than those of gender, noting that class, race, ethnicity and sexual orientation, for example, have no place in psychoanalytic formulations of subjectivity.\"\n\nMy aim is not to attack or defend psychoanalysis; there is little point, at this date, in doing either. A hundred years after the principles were enunciated, it is still difficult to imagine what a rival theory would look like. Getting beyond this point is going to require a paradigm shift, a wholesale reconception from a new stance. In the meantime, the omnipresence of Freud's ideas makes him the inevitable reference point. My approach here is pragmatic, perhaps appropriative. While setting aside theories which seem to me unfounded and unhelpful, I mean to make positive use of some Freudian concepts, including the distinction between desire-for and desire-to-be, the demarcation between gender identity and sexual object-choice, and the role of narcissism in structuring gay desire.\n\nIn this book elucidations and qualifications of theorists will be balanced and tested by more nuanced, complex, and perhaps vulnerable insights, quarried from works of fiction (novels, stories, and films) and autobiography. (I say little about theater because I have written about it elsewhere, albeit from other points of view.) Some of my references to fiction are quite brief, designed to support claims that such-and-such an attitude is commonly held. They are synoptic and symptomatic, borrowing purposefully from disparate sources in order to indicate that it is not just Hollywood writers, or just writers concerned with AIDS, or just male writers, who hold this-or-that position. Other discussions are larger, and are designed to show the complexity and subtlety of engagement that occur in certain writings, and, by inference, in human lives.\n\nI am not trying to write literary history, which seems to me impoverished in its defining assumption\u2014that literature is sufficiently self-contained to constitute, ultimately, a self-sustaining system. Nor am I interested in establishing literary value. I am captivated by some contemporary and recent fiction not because it tells us transcendent truths, but because it offers sophisticated narratives for exploration, reflection, and action. The literary is often seen as opposed to action, but I remain loyal to the cultural materialist idea that any writing of ambition might be more important than that. As Bertolt Brecht wrote of theater, \"it sets out society's experiences, past and present alike, in such a manner that the audience can 'appreciate' the feelings, insights and impulses which are distilled by the wisest, most active and most passionate among us from the events of the day or the century.\"\n\nReports of gay experience indicate repeatedly that fiction enlarges our sense of our potential. American novelist Jim Grimsley is typical:\n\nThe very first book I read in which gay men appeared was Mary Renault's _Fire from Heaven_ , in which Alexander and Hephaistion become lovers in the second half of the book. I was still in high school, probably 1970 or 1971. The fact that I could find some kind of affirmation that there were\u2014or had been\u2014other men like me was enthralling and I read the book over and over again. One of the ironies of that era was that the best books about gay men were all by women. Soon after that I would find John Rechy's _City of Night_ , which was extraordinary and powerful and which shook me deeply with its portrait of a gay man's narcissistic sexuality and indifference to commitment.\n\nI have argued elsewhere that the narratives which we revisit compulsively (in literary writing and many other forms) are those which in our cultures are unresolved: I call them _faultline stories_. When a part of our worldview threatens disruption by manifestly failing to cohere with the rest, then we reorganize and retell its story, trying to get it into shape\u2014back into the old shape if we are conservative-minded, or into a new shape if we are more adventurous. Faultline stories address the awkward, unresolved issues; they find their way, willy-nilly, into texts. There is nothing mysterious about this. Authors and readers want writing to be interesting, and these unresolved issues are the most promising for that. This is true in the culture at large, and in subcultural formations also. I deploy some of the techniques of poststructuralist literary criticism, but my aim is to explore how gay experiences have fed into books, films, plays, and cultural commentary, and how we, in turn, have read and pondered them, and reframed ourselves through them. In recognizing the plausibility of a story (Yes, I would act like that in those circumstances; or No, if he takes that line he's asking for trouble), we recognize ourselves, both individually and, implicitly at least, as part of a subcultural formation. We don't have to agree; the point is to have the conversation.\n\nTo be sure, I am exploiting different kinds of texts\u2014different genres, different media. A romance is likely to mobilize different impressions of gayness from a thriller. Yet each exerts a claim for plausibility within its own criteria, and readers learn to read accordingly. My main interest is in writing of the last thirty years, but contemporary writing is not necessarily the most influential. I discuss Walt Whitman, Oscar Wilde, Thomas Mann, Radclyffe Hall, E. M. Forster, and Jean Genet because they still bulk large in our collective imaginations. The aim is not to deduce facts from fiction, but to explore ideas and images; the kinds of representations that have been circulating. These texts display what people (gay and otherwise) have been saying about lesbians and gay men; or rather\u2014for writing must be seen as an intervention in a contested space\u2014what they have wanted readers to believe. In my concluding chapter I ponder whether there may be some use in the idea of a lesbian and gay canon.\n\nIt will be sensible to clarify a few other aspects of my approach. Much of what I have to say may easily be applied to heterosexuality. I have framed my thoughts and study in terms of gay subcultures because I know them better and am personally engaged there; because they offer a substantial and up-to-date body of work to build upon; and because there are plenty of books about sex from a straight point of view. It may be that a queer stance will be of wider relevance. Centers, after all, are defined by margins; dissident sexuality, being not the default position, is by definition always already problematized.\n\nI cannot envisage an effective study of gay men that will not be aware of what lies at the most immediate boundaries, those with lesbians and transgender people. I have drawn upon these neighboring discourses when it has seemed instructive to do so. I refer to both _gay_ and _queer_ , tending to use the former in modern and subcultural contexts, the latter in cases with more of a casual, provocative, and inclusive slant. _Sexual dissidence_ is at once vaguer, more purposeful, and more inclusive. _Same-sex_ aspires to neutrality. _Female_ and _male_ are always to be understood as the prevailing normative concepts, as are the notions of _masculinity_ and _femininity_ that conventionally accompany them.\n\nI have tried elsewhere to write about the interface between metropolitan and nonmetropolitan constructions of same-sex passion. The idea was not to find a universal model, still less to legislate for other peoples, but to gain a critical perspective on metropolitan assumptions, revealing them as a local and temporal creation\u2014one that tends to disavow, repudiate, or repress large areas of actual sexual opportunity. In this book my concern is mainly with gay subcultures that have arisen in the cities of North America and northwestern Europe, and to some extent with local versions of them as they have been distributed through other parts of the world by global interaction.\n\nIn practice, as it transpires, although this book was largely conceived and written in Britain, many of my texts are from the United States. As I have observed elsewhere, most of the metropolitan imagery of gayness is \"American\"; blue-jeans and T-shirts, short hair and mustaches, have been adopted into British gay subculture; so have candlelit vigils, quilting, buddying, and photo-obituaries. Because it is located in a more violent and fractured society, and because the AIDS emergency struck more precisely in more clearly demarcated communities, writing from the United States often seems to present the dilemmas of gayness with a special intensity. When I refer to \"our cultures\" I mean those of the United States and northwestern Europe.\n**2**\n\n**TAXONOMIES**\n\n**POSSIBLE AND PLAUSIBLE**\n\nIn a recent essay designed to kick-start gay history, David Halperin proposes a new categorization of same-sex relations. He distinguishes the _modern_ concept of homosexuality, and four _prehomosexual traditions_. These latter are found in specific European contexts\u2014from ancient Greece through early-modern Italy and France, and the molly houses of the eighteenth century, on into the emergence of the concept of \"sexual inversion\" in the writings of late-nineteenth-century sexologists. They are: (1) effeminacy, (2) pederasty or \"active\" sodomy, (3) friendship or male love, (4) passivity or inversion. The key factor linking these prehomosexual traditions, Halperin says, is the privileging of gender over sexual identity (defined by object-choice).\n\nIn \"modern\" homosexuality, which develops in the mid-twentieth century, it is the other way around. Both partners in a same-sex scenario are regarded as gay and neither need be positioned as feminine. Today, Halperin says, \"Homosexual relations cease to be compulsorily structured by a polarization of identities and roles (active\/passive, insertive\/receptive, masculine\/feminine, or man\/boy). Exclusive, lifelong, companionate, romantic, and mutual homosexual love becomes possible for both partners.\" However, traces of earlier patterns linger, and hence the confusions in some current ideas of gayness. Conversely, we may read modern egalitarian expectations back into earlier periods. Astutely, Halperin observes an inconsistency in Jamie O'Neill's acclaimed novel, _At Swim, Two Boys_ : while the Irish situation is presented with careful historical responsibility, the relationships between the gay boys are presented in the modern manner.\n\nHalperin's model is relatively local\u2014it charts particular social and historican contexts within Western European and classical tradition. He concedes that his \"categories are heuristic, tentative, and ad hoc,\" that his patterns \"do not reduce to a single coherent scheme.\" Bruce Smith's analysis of early-modern England and George Rousseau's account of Enlightenment Europe are in similar vein: they chart same-sex liaisons more or less in the terms in which they were regarded contemporaneously. In fact, their models may be regarded as local elaborations of the more general model advanced by anthropologists such as David Greenberg, Stephen Murray, and Gilbert Herdt. From a wide survey, they discover two main patterns of male same-sex relations: gender difference (one party is \"masculine,\" the other \"feminine\") and age difference. Like Halperin, they find that a new type\u2014egalitarian partnerships founded in similarity\u2014has been developing lately in metropolitan settings. More elaborate homosexual taxonomies, Jeffrey Weeks notes, have been generated by psychologists, such as Clifford Allen's twelve types (including the compulsive, the neurotic, and the alcoholic) and Richard Harvey's forty-six types (including the religious, the bodybuilder, and the ship's queen).\n\nAnthropological approaches have afforded a valuable way of reminding post-Stonewall gays that their way of doing things is not universal. I mean to take another route, trying to assemble a more abstract model\u2014one that will seek to map the range of _possible_ sex and gender positions, aspiring to coherence and even completeness. The _plausible_ positions, at any time and place, will be a particular selection and development from the possible positions. For instance, Halperin believes that while it was plausible in earlier centuries to have a close male friend, and to cross-dress, to combine the two in one person was almost inconceivable. However, it is always possible, and nowadays a drag queen may have close friends of all kinds.\n\nIt should not be supposed that the plausible modes delimit the totality of current behaviors. At present, for instance, there is no extant moral or legal ratification for the serial killer. Even psychologically, it is quite hard to envisage what gratification he or she might be obtaining. Nonetheless such people exist. Plausibility locates what can be said and done within normal parameters but, necessarily, it supposes the coexistence of varying degrees of implausibility, threatening always to infiltrate or shatter the normative. A more abstract taxonomy may remind us that the immediate evidence will not anticipate all possibilities.\n\nI do not intend this as an antihistorical, nor an ahistorical work. My aim is to explore frameworks within which historical differences may be better comprehended (especially, in the present study, the relations between the present and the recent past). The detecting of taxonomies is not a trivial matter. Eve Sedgwick suggests that it may contribute to \"the making and unmaking and _re_ making and redissolution of hundreds of old and new categorical imaginings concerning all the kinds it may take to make up a world.\"\n\n**THREE PROBLEMS WITH FREUD**\n\nThe other established taxonomies are in the Freudian tradition. Kenneth Lewes derives four \"explanations\" of male homosexuality from Freud's work; C. A. Tripp eight; Kaja Silverman three (one of them in two variants). As Silverman remarks, \"the various theories Freud proposes to account for the etiology of homosexuality are far from consistent, and it is often difficult to determine whether one is supposed to extend, supplement, or supersede those that came before\" (362). My intention is not to rehearse these theories, nor to attack or defend Freud, but to reconsider problems and opportunities for a materialist elucidation of dissident sexualities.\n\nFreud makes a number of gay-friendly utterances: most famously, that \"all human beings are capable of making a homosexual object-choice and have in fact made one in their unconscious.\" However, some of his ideas about homosexuality occur in the course of surprisingly flimsy essays (such as the one on Leonardo da Vinci). Lesbians and gay men have found three tendencies in Freudian theory particularly hard to deal with. One is the implication that gay men are in some fundamental way feminine, and lesbians correspondingly masculine. For many sexual dissidents, it should be said at once, this implication is not at all unsatisfactory. They are happy to think of themselves as embodying a significant element of the \"other\" sex. However, many are not. Gay men in particular have often appealed to David and Jonathan, the Theban Band, and Walt Whitman in order to establish that, so far from being effeminate, same-sex love may be quintessentially manly. Maurice, in E. M. Forster's novel, believed he had \"brought out the man in Alec, and now it was Alec's turn to bring out the hero in him.\" Notwithstanding, Silverman, expounding the Freudian corpus, warns that the gay man may have to accept that \"an identification with 'woman' constitutes the very basis of his identity, and\/or the position from which he desires.\"\n\nAs one might expect, many lesbians object to the idea that they might be \"really\" male. Sheila Jeffreys opposes bitterly the ascription of masculine roles to women. Teresa de Lauretis makes an elaborate case, broadly from within psychoanalysis, for freeing lesbians from the imputation that they are castrated and yearning for phallic maleness. The infant, she says, experiences a castrating loss (such as Freud posits), but it is the loss of the female\u2014of the mother\u2014not of the phallus. De Lauretis instances _The Well of Loneliness_ (1928), where Stephen's mother finds her daughter's body repulsive. This maternal withholding, which Stephen cannot address directly, is displaced onto a fetish: manliness. Her masculine bearing, therefore, signifies not phallic pretension but \"Stephen's desire for the (lost) female body.\" The difference from traditional Freudian versions is that Stephen's cultivation of male identity is linked only incidentally to the phallus. It is the adjustment that the sex\/gender system makes available: \"The popularity of _visible_ masculine signifiers as lesbian fetish in Western cultures is directly proportionate to the latter's enduringly hegemonic representation of lesbianism as phallic pretension of male identification\" (308).\n\nProbably Elizabeth Grosz is right to say that this attempt to render psychoanalysis lesbian-friendly still carries too much Freudian baggage. Nonetheless, de Lauretis' separation of gender from heterosexuality sets off reverberations that may be heard through the present study. My aim at this point is not to evaluate such theories but to note the firmness with which some lesbians repudiate Freudian arguments that would position them as second-class men. Molly in Rita Mae Brown's _Rubyfruit Jungle_ , encountering a butch\/femme bar, exclaims: \"That's the craziest dumbass thing I ever heard tell of. What's the point of being a lesbian if a woman is going to look and act like an _imitation man_? Hell, if I want a man, I'll get _the real thing_ , not one of these chippies.\" This is perhaps the kind of incident Judith Butler has in mind when she insists that homosexuality is not a _copy_ of heterosexuality. \"As a young person,\" she confesses, \"I suffered for a long time, and I suspect many people have, from being told, explicitly or implicitly, that what I 'am' is a copy, an imitation, a derivative example, a shadow of the real.\"\n\nA second problematic tendency in Freudian theory concerns \"arrested development.\" As Freud elaborates his idea of an Oedipus complex and adapts to it his experiences of homosexuality, the congenital claims of Havelock Ellis and the inversionists are largely replaced by a developmental model. In his famous letter to a mother about her son's homosexuality, Freud assured her that it was \"nothing to be ashamed of, no vice, no degradation.\" Nonetheless, he felt bound to say, it was \"produced by a certain arrest of sexual development.\" \"How does it come to be taken as self-evident that homo-erotics is really an arrested form of interest in oneself?\" Michael Warner demands.\n\nWhat is so revealing is the point at which the arrest is supposed to occur: it is at that moment in the Oedipal process at which the individual is caught in the \"wrong\" gender identity. In a note added to the _Three Essays_ in 1910, Freud declares:\n\nIn all the cases we have examined we have established the fact that the future [male] inverts, in the earliest years of their childhood, pass through a phase of very intense but short-lived fixation to a woman (usually their mother), and that, after leaving this behind, they identify themselves with a woman and take _themselves_ as their sexual object. That is to say, proceeding from a basis of narcissism, they look for a young man who resembles themselves and whom _they_ may love as their mother loved _them_. (56)\n\nIn other words, the homosexual behaves as his mother did (or as he wanted her to). Carole-Anne Tyler glosses: \"if a man desires another man, he must do so as a woman.\"\n\nFor women the Oedipal sequence is more complicated, but the outcome is similar. In his essay on \"Female Sexuality\" (1931), Freud sketches three lines of development for the girl as she \"acknowledges the fact of her castration.... Only if her development follows the third, very circuitous, path does she reach the final normal female attitude.\" Otherwise she may arrive at \"a general revulsion from sexuality,\" or she may \"cling with defiant self-assertiveness to her threatened masculinity,\" perhaps resulting in \"a manifest homosexual choice of object.\" In a footnote Freud grants cheerfully that \"men analysts with feminist views, as well as our women analysts, will disagree.\"\n\nA third problematic tendency, for many lesbians and gay men, is located around the term _narcissism_. The mythical Narcissus is a beautiful youth who refuses to be wooed, embracing only himself; gazing at his own reflection, he starves to death. Not the kind of guy you want as a role model.\n\nIn his essay \"On Narcissism: An Introduction\" (1914), Freud distinguishes _anaclitic_ (other directed) and _narcissistic_ love. According to the anaclitic type a man may love the woman who feeds him, and the man who protects him. According to the narcissistic type a man may love versions of himself\u2014what he is, what he was, what he would like to be, someone who was once part of himself. Freud straightaway mentions, but does not develop, \"the significance of narcissistic object-choice for homosexuality in men.\"\n\nOn Freud's own account, narcissistic love is far less limited than the name suggests. Like anaclitic love, it requires two people, and only in one variant are they supposed to be _the same_ ; otherwise there is a significant difference. In practice, a relationship with an individual who represents the person you have been, or might become, is likely to involve ceaseless negotiation. You are faced continually with both the distinctiveness of the other person (the extent to which s\/he does not embody your ideal self) and the contradictions and failures in your own yearning (your ideal self is not as likable, coherent, or attainable as you might wish to suppose). In fact \"anaclitic\" doesn't mean independent, but _attached_ ; specifically, \"leaning-on,\" Freud's editor explains, \"by analogy with the grammatical term 'enclitic,' used of particles which cannot be the first word in a sentence, but must be appended to, or must lean up against, a more important one.\" Narcissism, then, may operate in an anaclitic way. Freud does not offer them as opposed types: \"We have, however, not concluded that human beings are divided into two sharply differentiated groups.\" Nor, according to Freud, is narcissistic love distinctively gay. It characterizes many women, who, \"especially if they grow up with good looks, develop a certain self-contentment which compensates them for the social restrictions that are imposed on them in their choice of object.... Such women have the greatest fascination for men.\" Freud instances also the narcissism of parents and children.\n\nNotwithstanding, followers of Freud have tended to conclude that homosexual love is narcissistic and therefore at best immature, at worst pathological. It seems likely to involve age and status difference (a man loves what he was, what he would like to be), and hence to violate the modern egalitarian ethos. The common inference, as Kenneth Lewes puts it, is that homosexuality is \"not truly object-related, that it involves impoverished object relations and consequently operates through a primitive and defective superego, and that its mental organization is basically preoedipal.\" It is not surprising that lesbians and gay men have been uneasy at being labeled \"narcissistic.\" The only relation of difference that is validated is gender, and then only when a male and a female are involved.\n\nEven so, in my view Freud's four variants of narcissistic love do offer an intuitively relevant model for some kinds of lesbian and gay passion. Elements of hero worship and idealization, in or of a younger partner, abound in the histories we have created for ourselves, from Socrates and Sappho to Shakespeare, and on to Wilde and Forster. This is not just a male thing. Sarah Ponsonby was thirteen and Eleanor Butler nearly thirty when they made the commitment that was to become the Ladies of Llangollen. Stephen is thirty-one and Mary twenty-one when they fall in love in _The Well of Loneliness_. Audre Lorde in _Zami_ tells how she passed for thirty-five when she was actually twenty so as to take the protective role in her relationship with Muriel.\n\n**DESIRE-FOR AND DESIRE-TO-BE**\n\nIn pursuit of a more effective and more materialist taxonomy, I mean to resort to another Freudian construct\u2014one which is, I hope, less ideologically loaded than those I have discussed so far. In the essay _Group Psychology and the Analysis of the Ego_ (1921), Freud draws a distinction between _desire-for_ and _desire-to-be_. Typically, he says, the boy develops an anaclitic attachment to (desire-for) the mother. At the same time, also, he experiences an identification toward (desire-to-be) his father. These are \"two psychologically distinct ties: a straightforward sexual object-cathexis toward his mother and an identification with his father which takes him as his model.\" Propelled by \"the irresistible advance toward a unification of mental life, they come together at last; and the normal Oedipus complex originates from their confluence.\"\n\nFor Freud it is crucial, if your Oedipus complex is to work out properly, to get desire-to-be and desire-for the right way round:\n\nA little boy will exhibit a special interest in his father; he would like to grow like him and be like him, and take his place everywhere. We may say simply that he takes his father as his ideal. This behaviour has nothing to do with a passive or feminine attitude towards his father (and towards males in general); it is on the contrary typically masculine. It fits in very well with the Oedipus complex, for which it helps to prepare the way.\n\nHowever, it may not work out so conveniently. The father may be taken as the object of a feminine attitude, or the boy may develop an identification with his mother. The process, Freud admits, is precarious and hard to understand:\n\nIt is easy to state in a formula the distinction between an identification with the father and the choice of the father as an object. In the first case one's father is what one would like to _be_ , and in the second he is what one would like to _have_. The distinction, that is, depends upon whether the tie attaches to the subject or to the object of the ego. The former kind of tie is therefore already possible before any sexual object-choice has been made. It is much more difficult to give a clear metapsychological representation of the distinction. (135)\n\nWhat does seem clear is: (1) desire-to-be must be kept apart from desire-for, (2) this quarantine is unreliable, and (3) the consequence is sexual and gender dissidence. This insistence upon a precarious separation is Eve Sedgwick's theme in _Epistomology of the Closet_ where, she says, the idea was \"to demonstrate that modern, homophobic constructions of male heterosexuality have a conceptual dependence on a distinction between men's _identification_ (with men) and their _desire_ (for women), a distinction whose factitiousness is latent where not patent.\" Judith Butler also addresses the topic: \"The heterosexual logic that requires that identification and desire be mutually exclusive is one of the most reductive of heterosexism's psychological instruments: if one identifies _as_ a given gender, one must desire a different gender.\"\n\nSo how does passion cross the barrier between desire-to-be and desire-for? The answer: very easily! Wayne Koestenbaum observes: \"I spent much of my childhood trying to distinguish identification from desire, asking myself, 'Am I in love with Julie Andrews, or do I think I _am_ Julie Andrews?' I knew that to love Julie Andrews placed me, however vaguely, in heterosexuality's domain; but to identify with Julie Andrews, to want to be the star of _Star!_ , placed me under suspicion.\" John Fletcher declares: \"There can be no clear cut distinction between identification and desire, being and having, in the early stages\" of infant development; it is only a presumed Oedipal polarity that requires it. Noncompliance with this heteronormative demand is presented by Fletcher as a positive opportunity for gay relations:\n\nWhat is refused [in male homosexuality] is not masculinity or the phallic in itself, but the polarity at the heart of the Oedipal injunction: \"You cannot _be_ what you desire, you cannot _desire_ what you wish to be.\" The \"narcissism\" that characterizes certain gay male erotic scenarios, turning on images and terms of traditional masculinity and phallic positioning, often can be seen to have a reparative function, restoring an _alliance_ between being and having, identification and desire. (114; Fletcher's emphasis)\n\nLike de Lauretis in her argument for gender identity as fetish, Fletcher cleverly reorients the Oedipal calamity so that it becomes a gay advantage.\n\nNow, it is not necessary to tangle with psychoanalytic intricacies and purported explanations of gender and sexuality in order to find illumination in the distinction between desire-to-be and desire-for. The pattern has a formal aptness; it will admit, at least in the abstract, all the models I have discussed so far. It reorganizes, in relatively neutral terms, the gendered and narcissistic models of homosexuality which have troubled lesbians and gay men in psychoanalysis. It may be mapped onto, though not contained by, the schema posed so precisely by Sedgwick in _Epistemology of the Closet_ , of inversion (women's souls in men's bodies and vice versa) and gender separatism (same-sex bonding). It offers to connect up Halperin's four prehomosexual traditions. For instance, the effeminate man may be seen as cultivating desire-to-be of the feminine gender, while experiencing desire-for either women or men. Or, again, an \"active\" sodomite cultivates a desire-to-be male while experiencing desire-for a boy or man.\n\nFor the conventional model of heterosexuality you need a man who has desire-to-be male and desire-for a female, and a woman who has desire-to-be female and desire-for a male:\n\nA man has:\n\n| desire-to-be M| desire-for F  \n---|---|---\n\nA woman has:\n\n| desire-to-be F| desire-for M\n\nThe structure appears complementary at every point\u2014as it should do, for the terms are designed to ratify heteronormativity. You desire-to-be yourself (i.e., your own gender), which seems only right. You have desire-for-another, who is indeed other (another gender). Changing any of the positions disrupts the model. Such disruptions may be experienced, variously, as shameful weakness, moral dilemma, nervous strain, exhilarating kinkiness; some of them will produce gender identifications and object-choices which our cultures call homosexual.\n\nIt may be that the dichotomies I am invoking will strike you as a blunt instrument; so they do me. Their usefulness, it will emerge, resides largely in what we can learn from their inadequacies. My goal is not to fit the range of our relations into them, but to use them to disclose salient features of that range. For a start, the terms \"M\" and \"F\" must be problematized. They are to be understood as the prevailing normative concepts of male and female, together with the norms of masculinity and femininity that commonly accompany them. It is not my assumption that they are the positions that we have to occupy, but they are the positions we have to negotiate. In the initial, simple version of the model, it is supposed that gender identity will correspond to anatomical sex; however, this may not be so. In practice, little is uncontested in these matters. Some people referred to Margaret Thatcher as \"that bloody woman,\" others said she wasn't really a woman at all; interestingly, I don't remember anyone calling her a lesbian. Again: is it manly or cowardly for a man to assault his wife? Our cultures are not agreed on that.\n\n**MODEL (g)**\n\nTwo main dissident models initially appear. For the sake of simplicity, and because I feel more confident there, I am writing them as they apply to men. However, I believe they may admit lesbian experience as well; I indicate this from time to time, drawing upon instances and scholarship from lesbian traditions. In an attempt to evade preconceived historical and geographical notions, I call the two initial dissident models (c) and (g).\n\nThe former, model (c), is often associated with the ancient Greeks and endorsed by many gay men today. It shows males, without relinquishing their masculine gender identities, desiring other males:\n\n(c) A man has:\n\n| desire-to-be M| desire-for M  \n---|---|---\n\nThis model flies in the face of a (Freudian) inclination to impose a cross-sex pattern upon same-sex relations by distributing a same-sex couple as one pseudo-male and one pseudo-female. However, it may be authenticated (for Freudians) by Freud's recognition of desire-for the same gender in his account of narcissism. If the two men are more or less equal, they fit the modern egalitarian ethos. I return in a while to some of the complications in model (c).\n\nThe second dissident model, model (g), looks like this:\n\n(g) A man has:\n\n| desire-to-be F| desire-for M  \n---|---|---\n\nThis is the classic inversion model of the \"passive\" male homosexual: he wants to be female, and his desire, like that conventionally expected in a woman, is for a man. He may be said to have a woman's soul in a man's body, or a negative Oedipus complex. In the popular imagination, still, effeminacy is the badge of gayness. For instance, in Ned Cresswell's romance of small town to stardom via sexual intrigue, _A Hollywood Conscience_ , neither Brik nor Ryder has shown any sign of effeminacy. However, once Brik's gayness is recognized, camp becomes the inevitable marker of this knowledge. If Ryder is getting married he will want a matron of honor, Brik suggests. \"'Guess you'll have to wax your legs,'\" Ryder responds. The topic is not pursued by the boys; it is too risky, they pull back to their customary protocol. I trace some of the history of dissident gendering in chapter 5. Model (g) is about sexual dissidence organized around gender; we may call it the _gender_ model.\n\nMany gay men and lesbians, I have suggested, are uncomfortable with this model; it often appears in context with some element of disavowal or, at least, unease. In James Robert Baker's novel _Tim and Pete_ , Pete is a garage mechanic, performs heavy metal, and passes for straight. Tim is a film archivist. He likes Pete because he is not simply gendered\u2014not \"just a mechanic or a rock musician or a cute, butch guy.... It could be a long time before I met someone else with Pete's sensibility and humor.\" Meanwhile Tim risks being too effeminate, Pete accuses: \"'The day we went to Monte Carlo you looked like a fruit.'\" \"'Only in your mind. I was wearing a totally masculine, faded green tennis shirt and a sloppy straight guy's khaki shorts,'\" Tim retorts. \"'With your collar turned up like a queen,'\" Pete insists; two German guys made antigay remarks (73). However, neither of them is as feminine as Victor, who is British, lives with his mother, and likes Barry Manilow and Liza Minnelli.\n\nFurther, model (g) proves significantly inexact, in ways that indicate just how tangled and resistant to categorization gender is in our sex\/gender system. Consider: the obvious _partner_ for the man in model (g), who desires that he himself should be feminine and that his partner should be masculine, is:\n\nA man who has:\n\n| desire-to-be M| desire-for F  \n---|---|---\n\nOf course, this image is familiar: it represents the \"normal\" heterosexual man! Actually, that is not so strange. In the mid-twentieth century, men such as Quentin Crisp believed that effeminate homosexuals sought \"to win the love of a 'real' man.\" So the ultimate ideal partner was indeed a straight-identified man who desired the feminine. Unfortunately, desiring the feminine in the masculine form called the straightness of this man into question: \"A man who 'goes with' other men is not what they would call a real man.\" So the maneuver is bound to fail. In Latin cultures, however, this seems less of a problem: a masculine-identified man may manifest desire-for both women and effeminate men.\n\nIf, in one aspect, model (g) discloses a congruency with heterosexual desire, in another it fails to distinguish between the \"effeminate\" male homosexual and the transsexual anatomical male who feels that he really is female. Each manifests desire-to-be F, and may well experience desire-for M. This confusion is paradoxical, because one advantage of separating desire-to-be from desire-for is that it becomes easier to see the specificity of transgender. The ultimate distinction is that whereas the male homosexual priority is to get a man (desire-for), the transgender priority is to establish a dissident identity (desire-to-be). In Leslie Feinberg's _Stone Butch Blues_ , Jess has a dream which \"wasn't about being gay. It was about being a man or a woman.\" Judith Halberstam reports the case of Danny, a pre-operative female-to-male transsexual: Danny finds sexual satisfaction with men, but only so long as it is understood to be \"gay\" sex\u2014so long as s\/he is \"recognized\" as a man. Don Kulick has observed a comparable attitude among male-to-female _travestis_ in Brazil. They choose their macho boyfriends not for sexual fulfillment but because having such a man in the house reassures the _travesti_ that s\/he is female. In fact, transsexuals are not necessarily homosexual, and their model has to be written:\n\n(g) A man has:\n\n| desire-to-be F| desire-for M\/F  \n---|---|---\n\nThe transsexual also problematizes the initial term in the diagram, \"a man\" (or \"a woman\"), for s\/he may regard hirself as a male, a female, a mixture of the two, or neither. Notice that when we speak of the transsexual as a man who has desire-to-be a woman, that this is a loaded way of putting it\u2014a way that prioritizes anatomical gender. We might instead term this person a woman who has been born into the wrong body, prioritizing psychological gender.\n\nAs Jay Prosser observes, some transsexuals are refusing absolute gender categories. Kate Bornstein remarks: \"I identify as neither male nor female, and now that my lover is going through his gender change, it turns out I'm neither straight nor gay.\" This is not new. Theresa, Jess's lover in _Stone Butch Blues_ , calls herself a lesbian and urges Jess to join the women's movement\u2014\"'You're a woman!'\" she exclaims. But Jess denies this: \"'No I'm not,' I yelled back at her. 'I'm a he-she. That's different.'\" The transsexual is complicated also as a partner: is desire-for about hir masculinity, hir femininity, or both? Such relationships may be straight or gay, depending how you look at it.\n\nApplication of the categories \"M\" and \"F\" has been thrown into further disarray by recent academic and political attention to \"intersexuality\"\u2014approximately, what has been called hermaphroditism\u2014bearing anatomical indications of both genders. The Intersex Society of North America is objecting to the practice of surgeons who, coming upon infants whose sex appears mixed or indeterminate, intervene to construct what they regard as more satisfactory gender characteristics. The society urges that medically unnecessary surgery be deferred until the child can make an informed decision from within a supportive environment. Intersex people may feel themselves to be \"F,\" \"M,\" neither, or both. Clifford Geertz presents attitudes toward intersexuality in different cultures as an instance of the constructedness of common sense. The fact that transgender resists my diagrams is not surprising; it is an index of the difficulty our societies have in conceptualizing it.\n\nOne consequence of constructing a model that prompts a serious recognition of transgender is that the situation of the more typical lesbian or gay man comes more clearly into focus by comparison. Many such people experience a degree of dissident gender identification. However, they do not behave or regard themselves as thereby _not-male_ (for men), or _not-female_ (for women). Rather, they see themselves as embodying an element of the alternate gender. So camp men generally have more in common with other men than they do with women (women know this). Correspondingly, contributors to Sally Munt's collection _Butch\/Femme_ insist that while butches allude to masculinity, and even masquerade as men, their purpose is to pursue their particular ways of being women. Judith Butler writes:\n\nWithin lesbian contexts, the \"identification\" with masculinity that appears as butch identity is not a simple assimilation of lesbianism back into the terms of heterosexuality. As one lesbian femme explained, she likes her boys to be girls, meaning that \"being a girl\" contextualizes and resignifies \"masculinity\" in a butch identity. As a result, that masculinity, if that it can be called, is always brought into relief against a culturally intelligible \"female body.\"\n\nHalberstam cites Butler's argument, but reasserts that there are other positions: \"While some girls are content with boys who retain genetically female bodies, others desire the transgendered or cosmetically altered body.\"\n\nI am minded to conclude, from these complexities, that the difference between lesbian or gay and transgender variants of model (g) is one of degree. They all point toward a dissident gender identity, but they range, in a person of predominantly male anatomy, from sensitivity, through the screaming queen, to the person who seeks gender reassignment surgery. In the diagram\n\n(g) A man has:\n\n| desire-to-be F,  \n---|---\n\ntherefore, the terms comprise variable intensities in a continuum. If this argument feels wrong\u2014threatening to situate you close to something that you feel is not-you\u2014bear in mind that it is the proximate that demands most assiduous policing. Stephen Maddison, in _Fags, Hags, and Queer Sisters_ , suggests that we regard (some kinds of) homosexuality and transgender as \"alternative responses to similar conditions,\" adding: \"The two movements share a heritage.\" That is right: the diaries of Anne Lister and _The Well of Loneliness_ figure in both lesbian and transgender discourses (I develop this argument in chapter 5).\n\nWhat has made it difficult to unravel the diversity of the gender model is Freudian absolutism, which is committed to a distribution of psychic life between two poles: father\/mother, male\/female. For instance, in \"A Case of Homosexuality in a Woman\" (1920), Freud's analysand had \"entirely repudiated her wish for a child, her love of men, and the feminine role in general.\" The outcome was \"extreme\": \"She changed into a man and took her mother in place of her father as the object of her love.\" But the extremism is Freud's: not wanting a child, a male partner, or a feminine role does not turn a woman into a man! Again: finding at the end of puberty that the time has come for \"exchanging his mother for some other sexual object,\" a young man \"identifies himself with her; he transforms himself into her, and now looks about for objects which can replace his ego for him, and on which he can bestow such love and care as he has experienced from his mother.\" Such language (\"he transforms himself into her\") encourages the inference that the homosexual who develops a dissident gender identity _really has_ , at some level, changed gender.\n\nIn fact most gender-dissident individuals in the gender model, transsexuals apart, should be apprehended as aspiring, quite informally, to some kind of _mixing_ of gender identities, whereby a person may appropriate \"other\" gender attributes, without seeking to abandon his or her initially ascribed gender. I had thought to call this \"androgyny,\" meaning not a semimystical transcendence of gender and the body, but simply a strategic appropriation. However, Halberstam critiques Martha Vicinus' usage on this: \"The androgyne represents some version of gender mixing, but this rarely adds up to total ambiguity; when a woman is mistaken consistently for a man, I think it is safe to say that what marks her gender presentation is not androgyny but masculinity.\" So androgyny and female masculinity are distinct.\n\nThe wider point here is that a range of degrees and types of commitment may occur within the gender model, and only the most extreme should be understood as amounting to gender transformation. Desire-to-be is a relative matter, then, not an absolute difference. Perhaps the idea, as Maddison suggests, is not to be a woman but to disaffiliate from dominant heterosexual modes of manhood. For many or most gay men, desire-to-be in the gender model should be represented not by \"F\" but by \"RF,\" _relatively feminine_ :\n\nA man has:\n\n| desire-to-be RF| desire-for M  \n---|---|---\n\nWhile \"RF\" does embrace the carefully fashioned appropriation of the drag queen or king and the punk gender-bender, for the most part it entails the ordinary, day-to-day effects of an uneven gender identification. Indeed, it may have little to do with object-choice, and be compatible with a heterosexual desire-for. Desire-to-be \"RF\" is available for diverse disaffections of males from heavily masculine commitments: leisure-class men affecting a dandified style; schoolboys wanting to be aesthetes rather than athletes; artists, priests, and dons choosing to signal unworldliness; the present-day \"New Man.\" Halberstam points out that rural women may be considered masculine by urban standards, but merely practical in their own community.\n\nIn most cases I have written the dissident models as they locate individuals, but identities are interactive and the partner (long or short-term, actual or fantasized) is important because s\/he is the person who, above all, is expected to confirm one's own identity. In conventional heterosexual couples this may work contrastively: the man may feel more masculine in contrast with the woman, and vice versa. Plainly a lot of normative masculinity depends on that process; indeed, while it is conventionally supposed that a man has to be masculine in order to impress a woman, it may be the other way around\u2014the woman is called upon to ratify his masculinity. This is a point made by feminists. Today in metropolitan contexts the obvious partner for a man who has desire-to-be \"RF\" is _a man who has desire-for \"RF.\"_ In fact _both partners_ may experience desire-to-be \"RF\" _and_ desire-for \"RF.\" This does not mean that such couples will be symmetrical. The \"RF\" element may be distributed unevenly, producing complementarity at some points, conflict at others. Further, \"M,\" \"F,\" and \"RF\" are likely to be relative between the partners. In a butch\/femme couple, for instance, the partners may be gender-marked mainly as measured against each other, not in absolute terms of masculinity and femininity.\n\nThe relative nature of many gender-dissident identities points toward the diversity of innumerable actual lesbian and gay relationships. I think its occurrence is sufficiently pronounced to justify terming it the _relative-gender_ model, or model (rg). Insofar as it posits a weaker element of gender dissidence than has often been supposed, it begins to converge on model (c)\u2014to which I now return.\n\n**MODEL (c)**\n\nA man has:\n\n| desire-to-be M| desire-for M  \n---|---|---\n\nIn the most obvious version of (c), each man is attracted to someone who is very like himself. They may admire and inspire one another's masculinity, perhaps emulously, as between Shakespeare's Coriolanus and Aufidius, or in the gym. (When Aufidius calls Coriolanus \"boy,\" at the climax of the play, he ejects him from manly equivalence.) Somewhere around this point, many accounts of (c) and (g) seek to justify one and condemn the other. Adherents of (c) are accused of doubling phallic maleness, colluding with heteronormativity, and despising women; adherents of (g) are accused of being effeminate, colluding with heteronormativity, and (secretly) despising women. A taxonomist is not obliged to evaluate\u2014though in fact the terms I have been using are replete, inevitably, with premature evaluations; there is no neutral language.\n\nAs with model (g), for some men the positions in (c) might better be apprehended as \"RF\" or \"RM\" (relatively feminine or masculine). However, this does not affect the viability of the model. The main structural inadequacy in (c) is that diverse relations are incorporated together. If this model may be labeled narcissistic, that does not mean that a monochromatic sameness prevails. For very many gay men, it is crucial that the desired object in model (c) should be _male and different_ \u2014different, above all, in class, and\/or age, and\/or race. Let's take examples.\n\nAt the start of the film about David Hockney and his friends and lovers, _A Bigger Splash_ (Jack Hazan, 1974), we see a good-looking young man, hands behind his head, face composed. The camera swings between him and Hockney; they look pleased with themselves. Hockney speaks partly toward Joe, partly to camera:\n\nHow could I describe Joe? He's, erm, erm, he's tall; he's about my size [ _Hockney smirks and Joe giggles_ ], erm, he's handsome. He's got a complexion similar to mine. He's [ _they laugh_ ] witty, erm, he's, erm, sexy. And\u2014what else? He's artistic: I've decided you're artistic, Joe.\n\n\"When did you decide that?\" Joe asks\u2014\"When I said I liked your work?\" Joe suggests.\n\nHockney emphasizes the physical similarity of the two men; their complicit laughter allows the viewer to suppose that they are talking partly about genital equipment. However, everything else in the exchange speaks status difference. We don't need to have Hockney described because we know who he is and, anyway, the entire film is about him. He has the authority to decide who is to be called artistic; they have been to Paris with some of his pictures. Joe, we may gather, is stirred by Hockney's fame. At the end of the film we are in the same scene; Joe is talking about Hockney's paintings: \"They were so beautiful. And I said, 'My God, I know the painter that did that.' And I knew the person thought I was lying.\" Joe's admiration is flattering for Hockney who, though I deem nice-looking, is not a pinup boy; age is not mentioned, but Joe is clearly younger. The difference in status, I think for both men, is part of the attraction.\n\nClass difference was very common in mid-twentieth-century queer relationships, where it was often associated with (middle-class) effeminacy and (working-class) masculinity; I have written about this in various places. Some women cultivated it as well. Stephen in _The Well of Loneliness_ begins by falling in love with Collins, the maid. Ominously, in respect of her later sacrifice of Mary, Stephen prays to be allowed to take Collins's housemaid's knee, Christlike, upon herself; when praying doesn't work she tries kneeling for long periods. Disillusionment sets in when she comes upon Collins necking with the footman; she handles it by transferring her affections to her new pony, which she names \"Collins.\" As Prosser points out, Stephen is of a higher class than her lovers, Angela and Mary. Consider also Virginia Woolf's romance with aristocracy, as well as with Vita Sackville-West, for instance as displayed in _Orlando_. I am taking \"class\" approximately, as comprising hierarchies of wealth, status, and cultural sophistication, and their markers in attire, decor, and general lifestyle; it is in my view quite wrong to suppose that we have grown out of all that.\n\nClass hierarchy is disavowed and then acknowledged in _Tim and Pete_. Both men are into movies, but whereas Tim is an archival researcher and, as I have remarked, at risk of appearing effeminate because of his college-educated manner, Pete is an automobile mechanic, fronts an aggressively political band, and is passing as straight in his rough (lower-class) apartment building. Pete has misled Tim about his \"'middle-class background'\": \"the neighborhood was more lower-middle class or blue collar.\" Pete implied that he studied at Yale, when actually he'd only lived in New Haven with a history professor (class and age hierarchy there). Tim realizes that Pete had \"exaggerated so that we'd seem more equal.\" But does Tim really want Pete to be more equal? \"Once I'd sucked Pete's cock while he was only wearing his Yale T-shirt. Afterward he'd said, 'So do you like me better as a brilliant student or a dumb mechanic?'\" Tim had replied: \"'I like _you_ '\"\u2014the correct response for the modern, egalitarian gay man. But now he wonders whether it was true (37). When they saw the film of _Maurice_ they went home and acted joke variations on the roles (\"'Oh, Scudder. You're stretching me. My word'\"). Their parody effected a \"homoerotic catharsis... a genuine guilty pleasure\" (39; my elision). The endurance of class mobility as a theme is evident in _The Talented Mr. Ripley_ (Anthony Minghella, 1999) and _AKA_ (Duncan Roy, 2002), films in which desire-for and desire-to-be cross over to intriguing effect.\n\nIf model (c) sexualities may be regarded as narcissistic, and hence as involving what one was, would like to have been, or would like to be, then age difference is a likely component (it is a factor in _A Bigger Splash_ and _Tim and Pete_ ). In Felice Picano's novel _Like People in History_ , a young ACT UP activist is surprised to find Wally and the older narrator together: \"'I could never figure out why a great-looking guy like Wally would get involved in a transgen thing.'\" \"Read trans-generational,\" the narrator says to himself. \"Read I'm old enough to be his father but neither look it nor act like it. Read eternal Peter Pan.\" The age model flourishes notwithstanding. It figures in many of the most influential texts of our time\u2014 _The Immoralist_ (Andr\u00e9 Gide), _Death in Venice_ (Thomas Mann), _Maurice_ (E. M. Forster), _Funeral Rites_ (Jean Genet), _Hemlock and After_ (Angus Wilson), _Variation on a Theme_ (Terence Rattigan), _Sweet Bird of Youth_ (Tennessee Williams), _Entertaining Mr. Sloane_ (Joe Orton), _A Single Man_ (Christopher Isherwood), _The Swimming-Pool Library_ (Alan Hollinghurst), _Ready to Catch Him Should He Fall_ (Neil Bartlett), _Frisk_ (Dennis Cooper), _The Night Listener_ (Armistead Maupin). Its ordinariness is manifest in innumerable contact ads. In _The Beautiful Room Is Empty_ by Edmund White, the narrator has a substantial relationship with Lou, but it doesn't work out: Edmund is \"too big and educated to be the boy, and too much younger to be the man.\"\n\nDe Lauretis' founding of lesbianism in the loss of the mother seems likely to produce relationships characterized by age difference. She reports how the Milan Women's Bookshop Collective found disparities (social, educational, economic) among their reading group: some women \"were seen as authoritarian 'mothers' prevaricating over the preferences and interpretations of the others, who thus felt cast in the role of daughters.\" The collective decided not to outlaw this element of hierarchy, but to validate relations of _entrustment_ ,\n\nin which one woman gives her trust or entrusts herself symbolically to another woman, who thus becomes her guide, mentor, or point of reference.... Both women engage in the relationship\u2014and here is the novelty, and the most controversial aspect of this feminist theory of practice\u2014not in spite, but rather because of and in full recognition of the disparity that may exist between them in class or social position, age, level of education, professional status, income, etc. (21\u201322; my elision)\n\nAnd this, de Lauretis observes, is \"contrary to the egalitarian feminist belief that women's mutual trust is incompatible with unequal power\" (24). Working with disparities enables a proper recognition of the diversity of women.\n\nAn interesting corollary of the age version of model (c) is that it offers a way for lesbians and gay men to reproduce their kind: some girls and boys may pass through, serially, to the woman's or man's position, and so on from generation to generation. This is anticipated by Freud in his comments on the Greeks: \"As soon as the boy became a man he ceased to be a sexual object for men and himself, perhaps, became a lover of boys.\" Jean Genet is a notable individual who was involved in such a sequence. People hostile to homosexuality become especially distressed at the prospect that we might have our own, same-sex way of breeding.\n\nIn my discussion of the gender model (g), it was possible to recognize a range of desires by horizontal substitution\u2014complicating rather than changing the received terms of the model (\"M\" and \"F\"), and then adding \"RF\" and (rg). In model (c) vertical substitution is required. For instance:\n\n(c) A man has:\n\n| desire-to-be M| desire-for M  \n---|---|---  \n| who is old| who is old  \n| young| young  \n| old| young  \n| young| old\n\nThe asymmetry between the two models occurs because (g) is about gender, whereas in (c) gender tends to obscure other factors. I propose calling model (c) the _complementarity_ model, taking this to include both the sense of lack in narcissism and the potential affinity in difference; it may be particularized as _race complementary_ , _class complementary_ , and _age complementary_.\n\nAs with the gender model, the mapping of such elaborations discloses further complications. With the complementarity model there is no convenient starting assumption as to who does what. A boy in the age version, for instance, might be rough and \"active\" or docile (tractable, agreeable) and \"passive\"; indeed, he might be rough and \"passive\" or docile and \"active.\" Nor are these matters stable: a relationship may start in one vein and modulate into another. Delving further into the intimate potential of both fantasy and practice would produce more elaborate systems.\n\n**BISEXUALITY**\n\nA blatant disturber of neatly gendered models is bisexuality. Traditionally, lesbians and gay men have been suspicious of bisexuality, regarding it as a way of evading the stigma of gayness. This may have been partly true. It is plain, however, that very many people entertain, simultaneously or successively, divergent desires, both for and to-be. In _Gay and After_ I argue that in the 1970s and 1980s, to declare yourself gay or lesbian was such a strenuous project that to blur the effect by adding that sometimes you were a bit straight after all seemed just too complicated, and scarcely plausible. The notion that there are two distinct populations suited straights because it helped them to avoid contamination, and gays because it facilitated political and economic organization. However, since the mid-1990s, some young people are less daunted by such pressures. Meanwhile some noted lesbian and gay activists, who can hardly be accused of running scared, have been venturing beyond customary identities.\n\nTaxonomical thinking discloses an interesting symmetry: compare the position of the bisexual with that of the transsexual. Both are in-between\u2014irregular combinations of \"M\" and \"F.\" However, one is structured in desire-for, the other in desire-to-be.\n\nThe scope for entertaining, more or less together, formally incompatible desires is represented positively in Aiden Shaw's self-consciously contemporary novel, _Wasted_. At the beginning David and Joe are together, though David was with Flora before. David also loves his nephew, Ryan (age sixteen). Ryan is with Leila, but experiments in going to bed with David. Flora takes up with Don, who is straight but nonetheless drawn to Joe. David dies, and Joe connects with Flora as an artistic collaborator. Flora falls for Josie. Joe finds an immediate sexual rapport with Dylan, a student friend of Ryan. \"'Yeah. Meet the family,' said Flora.\" These are families of choice. Scarcely an eyelid is batted at these diverse developments. David is worried that his feelings for his young nephew are sexual (50); however, for Ryan it's cool: \"'You sound like an old queer from the Fifties.' He held out his arms to hug David. 'Why should it bother me?'\" (52). Flora doesn't hesitate with Josie: \"It felt so right in Josie's arms. Complete. Already it felt like the next phase in life\" (200); \"'What! Flora a dyke?'\" Ryan exclaims. \"'Fuck! That one's full of surprises'\" (218). Joe and Dylan feel obliged to defend the age difference between them (fifteen years), but it's no big deal. Among the younger generation more or less anything goes. Indeed, Ryan finds that in fantasy he can fill in the blank of David's former lover with diverse interchangeable images, including himself, like morphing in a pop video. In Ryan's drugged dream the characters of the book merge into each other: \"Leila was David. David was Leila, now both just one person. Sexual. Passionate. Loving. Breasts turned into arse cheeks, their dicks into syringes. Dylan also became a part of them. He merged with David and Leila. Ryan loved them, physically and mentally\" (257).\n\nHowever, free-ranging is not the only game in town. As a counterpart to the equable families of choice, _Wasted_ displays a preoccupation with involuntary sexual experience. David tells Ryan how he has been obsessed with him\u2014\"'hundreds of times I have jerked off about you... nine years of jerking off, and the different images I've had of you in my head during that time'\" (142; my elision). There is one rape in the novel, probably two. Also, as well as the usual array of clubbing drugs, the characters experiment with a sleeping tablet through which one person subjects himself to an oblivion in which anything can be done to him. \"'But he could have done anything.' 'Exactly. Sexy, huh?'\" (9). It is a mechanism of trust, but also of exploitation. More ominously, the closing episode of the book indicates that not everyone in the city is cool. The amiable, freewheeling milieu has no resources to match obsessional sexual violence.\n\nThe outcome is merely frustrating in a rueful instance offered by Sarah Schulman in her novel _Empathy_. Anna, a woman, and a man have been drinking and decide to play a game: \"each one would say their fantasy and the other two would fulfill it.\" The man goes first, and wants the two women to suck his dick, so they do. The woman wants Anna to be fucked by the man, so she is. However, when Anna wants the man to leave the room so she can make love to the woman, the woman says no.\n\nMy goal in this chapter has been to generate models of gender and sexual experience quite abstractly, so as to afford a possible frame within which the more local, empirical categories of Halperin, Smith, Rousseau, and anthropological scholars may be comprehended. I have derived desire-for and desire-to-be from Freudian writings, while trying to avoid the Freudian tendency to essentialize and absolutize gender. The dominant categories that emerge in my analysis are gender difference and gender complementarity. _Gender difference_ comprises heterosexuality, together with a range of intricate same-sex relations of identity and desire, in which a person with at least some biological male characteristics is apprehended, either by himself or his male partner, as feminine (and the other way about for women); gender difference turns out, often, to be _relative_. _Gender complementarity_ delineates relations where two men or two women have both desire-for the same gender and desire-to-be the same gender; while it may be addressed as a kind of narcissism, it emerges as preoccupied with hierarchies of class, age, and race, within ostensible sameness.\n\nArguably, I have myself colluded with heteronormativity by retaining versions of \"F\" and \"M\" as starting points in my models. Perhaps we should be taking more seriously a desire-to-be wealthy, taller, a doctor, or Barbra Streisand. Or this:\n\nA man has:\n\n| desire-to-be not-M| desire-for God  \n---|---|---\n\nIdentity might not be about gender, sexuality might not be genital. Many erotic practices are relatively diffuse\u2014involving pleasures of touch and smell. Some sadomasochists, fetishists, and pedophiles may be able to find satisfaction with either male or female partners. Concepts such as beauty, intelligence, sense of humor, and even virtue may be stimulating; they are not altogether in thrall to ideology. The point of my taxonomy, I have said, is not to limit identity or desire but, rather, to offer a base from which the specificity and multiplicity of the potential combinations and interactions may coherently emerge.\n\nEven with these provisos, I suspect that for many readers my brisk modeling feels too regulated, too standardized. It goes against the general postmodern-poststructuralist truism, that _any_ identity is, and should be, provisional, unstable, fuzzy around the edges, occupied only through processes of anxious iteration. Taxonomy refuses the ideology which asserts that we are all individuals, and that our sex lives belong to a private, personal, individual realm into which it is better not to inquire. At least (it may be averred) Freud produces an air of mystery. I have to say that I have never found individualism a very appealing or reassuring idea. As David Evans observes, capitalism invites us to see ourselves as \"unique individuals with needs, identities and lifestyles which we express through our purchase of appropriate commodities.\" In fact, advertisers and other cultural producers know how to corral us into niche markets where we can be conveniently targeted; individual choice is disturbingly congruent with the idea that the right designer label will enable us to complete our happiness. At the same time, it is only by combining that ordinary people gain any potential for political action\u2014for understanding, even. Insistence upon individuality amounts to a naive reluctance to acknowledge that oneself is actually quite like a lot of other people.\n\nThe presumption behind my models is that our behavior falls into patterns, and that they are not unconnected with those disclosed by surveys and focus groups. Sexuality is social. However, I do believe that those patterns are immensely complex. In _Stone Butch Blues_ Theresa protests when Jess decides to begin taking male hormones in order to pass as a man. \"'I'm a woman, Jess. I love you because you're a woman, too,'\" Theresa avers. \"'I just don't want to be some man's wife, even if that man's a woman.'\" This is not just a matter of object-choice. Theresa experiences her identity as interdependent with that of her partner: \"'I'm a femme, Jess. I want to be with a butch.'\" Otherwise her lesbian character is at stake: \"'If I'm not with a butch everyone just assumes I'm straight. It's like I'm passing too, against my will. I've worked hard to be discriminated against as a lesbian'\" (151).\n**3**\n\n**FANTASY**\n\n**DISSIDENT IDENTITIES**\n\nAfter decades of collaboration and occasional sexual experimentation together, Esther Newton and Shirley Walton realized that the reason they had never really got off was that, despite appearances, they both were tops. What is needed, they argue, is \"a more precise vocabulary to take us out of Victorian romanticism in sexual matters and toward a new understanding of women's sexual diversity and possibility.\" The categories Newton and Walton discover are:\n\nsexual preference (from which gender you usually select your partners)\n\nerotic identity (how you image yourself)\n\nerotic role (who you want to be in bed)\n\nand erotic acts (what you like to do in bed)\n\nWhat you prefer to do in bed cannot be inferred from whether you appear to be cultivating a masculine or a feminine image, or whether you are the older or the younger partner. My investigation of taxonomies entailed an admittedly schematic tendency; this chapter will involve a corrective assessment of the disorderly operations of fantasy, which challenge, solidify, and divert established identities and orthodox desires.\n\nYour fantasies may run quite counter to your self-presentation; they may be indelible, or fairly flexible; they may be conventional or, at least to others, radically inventive. There is, writes Vicky Lebeau, \"no limit to the reach of fantasy, its role in our attempts to contain the trauma, as well as the banality, of our lives.\" Fantasies are not, as I use the term, typically unconscious, though they may be. Leo Bersani declares their practical importance: \"What positions, what activities, what identifications excite us? What imagined object best helps the masturbatory process along? What do we prefer the other to be doing\u2014to us, for us, alone, with someone else?\"\n\nBy fantasy I mean the scenarios that we cultivate in our imaginings, typically of empowerment and humiliation (I seek to justify this emphasis on power in the next chapter). Fantasies are not necessarily sexual in form or origins; as I have indicated, a scenario of class or racial identification or domination is not necessarily to be _reduced to_ the sexual. Their most intense expression _may be_ sexual, however; that is where they enter the most vivid sites of pleasure and control. Fantasies are not necessarily solitary, secret, manipulated, or frustrated; the term includes attempted and successful realization in action, perhaps in collaboration with another. Getting someone to share your scenario is not only fun, it may help to make it plausible to you. Perhaps, as Aristophanes suggests in Plato's _Symposium_ , the desire to find one's lost other half is fundamental not just to love and desire, but to humanity. Fortunately, a consensual partner may be found for most practices. Conversely, Jean-Paul Sartre's vision of hell, in his play _Huis Clos_ ( _No Exit_ ; 1944), is three people trapped together and prevented by their incompatible psychic needs from confirming each other.\n\nRegrettably, that is not all. You can be at ease with yourself and with your partners, but if the social and political system is stigmatizing and criminalizing you, then you still have a problem. Consider the men in the British \"Operation Spanner\" case, who were found guilty of consensual S\/M practices about which they felt personally very happy.\n\nSophisticated analysis of these topics often begins at the intersection of psychoanalysis and film studies, with \"the gaze.\" A comparison may be broached between the way the subject locates him- or herself in the reading of a film (or other) narrative, and in a fantasy scenario. Laura Mulvey inaugurated much of this work by arguing that in Hollywood cinema male viewers are invited to identify with a male protagonist in looking at and desiring women as objects, while women are to identify with the female figures passively looked at. Such an analysis answers well to an intuitive sense of Hollywood as a monstrous dream machine for the industrialization of culture, dedicated to the preservation of conventional male and female roles. However, it seems to make women passive, not to say stupid, in their reading of film. Also, it makes the system appear more monolithic than is plausible.\n\nSubsequent work\u2014including by Mulvey herself\u2014has consisted of theorizing a way past the implications of the gaze, involving three main points. First, Hollywood does seem to offer more diverse possibilities. An instance that has delighted many gay men is the chorus, \"June Is Bustin' Out All Over,\" in Richard Rodgers and Oscar Hammerstein's _Carousel_ (Henry King, 1956). In the film's choreography, the girls dance with the fishermen and the sailors, but then the boys dance with each other, in pairs, and for each other as complementary groups; the girls watch the boys, the boys watch each other. Diverse spectatorial positions are available here for women and gay men. Unfortunately, _Carousel_ strives ultimately to contain such gender exuberance: the plot is resolutely heterosexual, and even comradely relations between men are shown as dishonest, violent, criminal, and fatal. Such a contradiction is not unusual in the Hollywood musical. But audiences do not have to respect closures; they may dwell imaginatively on the episodes that excite them. June may bust out.\n\nSecond, as Jean Laplanche and Jean-Bertrand Pontalis observe in a key formulation, fantasy \"is not the object of desire, but its setting.\" This means that the subject may locate him- or herself at more than one point in a scenario. A seduction fantasy, for instance, \"is a scenario with multiple entries, in which nothing shows whether the subject will be immediately located as _daughter_ ; it can as well be fixed as _father_ , or even in the term _seduces_.\" Desire-for alternates, overlaps, and tangles with desire-to-be. Freud is often credited with noticing this potential mobility of identification and desire in his essay \"'A Child Is Being Beaten.'\" There is a nice instance of it in James Robert Baker's novel, _Boy Wonder_ , where a leading theme is obsessional fantasy investments. \"As the film [ _Rebel Without a Cause_ ] reached its climax, and Sal Mineo died on the observatory steps, Shark wept. 'I felt as I _were_ Sal Mineo,' he said, 'but also Dean. In the end more Dean, the survivor, than Mineo, the martyr. But a _part_ of me died on those observatory steps.'\"\n\nChris Straayer considers how a Mulveyan viewing regime may be adapted to accommodate a lesbian spectator and her partner. In films such as _Entre Nous_ (Diane Kurys, 1983) and _Voyage en Douce_ (Michel Deville, 1979), the male in a triangular relationship with two women may be regarded as an \"intermediary\" for the feeling between the two women. Through hints such as the exchange of significant glances, a space for \"female bonding\" may be discovered. Again: Dorothy Allison notes that the clippings pinned above her desk include a young woman in a black lace dress and feathered hat, and a samurai woman sweeping her long sword. \"Some days I want to become one or the other of them. Some days I want to write the story of how they become lovers. Other days I can't stand to look at them at all.\"\n\nThird, any assumption that people want to identify with the nearest equivalent to their ostensible selves is unsatisfactory. On the contrary, fantasy is likely to be the place where we try out alternative identities and desires. Constance Penley has observed how women contributors to _Star Trek_ fanzines invest their libidinal energies in heroic, romantic, and sometimes sexy stories about Captain Kirk and Dr. Spock, rather than in women characters. Penley takes this as evidence \"that one can, no matter what one's gender, identify with either the man or the woman, or the entire scene itself, or the fictional place of the one who looks on to the scene.\" The motives of these women seem to be mixed. They are witty and self-parodic; fooling about, experimenting, conducting their own enterprising voyage into the unknown; they are also in earnest. They are claiming male freedoms in their imaginations, while refusing to announce themselves as feminists and rejecting the female body; they are happy to see men as erotically involved but reluctant to contemplate gayness. The main point, however, is perhaps that very many people are far more inventive and adaptable than has often been supposed.\n\nCross-gender identification is an obvious instance of unruly fantasy. In some aspects at least, it seems to be more disturbing to heteronormativity than dissident object-choice. The notorious version is the traditional gay male devotion to female stars such as Judy Garland and Maria Callas. The death of Garland is usually reckoned to be one of the direct stimuli for the 1969 Stonewall Riots. According to Richard Dyer, Garland \"could be seen as in some sense androgynous, as a gender in-between.\" Further, \"she sings of desire for men and of relationships with men going wrong. Male singers could not (still largely do not) sing of these things.\" Stephen Maddison posits two explanations. The gay man adopts the position of the woman\u2014perhaps, we might confess, elbowing her aside\u2014so as to commandeer her desire for the man. Also, a broadly \"feminine\" emotional stance is desired and the entire scenario is embraced. Lately many gay men have admired women such as Barbra Streisand and Madonna, who appear to have more control over their destinies.\n\nAt this point it seems appropriate to recall my argument in the previous chapter, that individuals who cultivate an element of gender dissidence, stopping short of a transsexual adjustment, do not want to be a different gender; they should be apprehended as aspiring, quite informally, to an amalgamation of gender attributes. Gay men who gain pleasure and strength from a vivid engagement with Streisand don't believe that they _are_ her. They are pirating aspects of the image for their own purposes. For gender is a negotiation, not a possession; there are innumerable reasons for trying to feel definite about it, but any such attainment is provisional. Fantasy should be understood, not as an absolute demand, nor as a unified core, but as a sequential, piecemeal, strategic adaptation. David Wojnarowicz remarks: \"Fantasized images are actually made up of millions of disjointed observations collected and collated into the forms and textures of thought.\"\n\nIf a mood of feminine emotional indulgence and sexual attraction has appeal for some men, the freedoms associated with masculinity have an obvious appeal for women. Lynne Segal describes how her path to a heterosexual and feminist identification passed through gay fiction: \"The lustfully desirous fantasies of my own youth were\u2014as they remain\u2014most easily aroused and fed by the words and images of male homosexual authors.\" Segal was drawn particularly to the black, and hence doubly forbidden, author James Baldwin. His gay characters afforded a more attractive route to desire for the male than many of the available images of women. Cora Kaplan describes a comparable youthful investment in Baldwin's writing. She appreciated\n\nthe lowered threshold he provided for fantasies that were not about the fixing of gender or sexual orientation but about their mobility and fluidity. Women could take up shifting and multiple fantasy positions within his fictional narratives: that possibility, itself wonderfully if terrifyingly liberating, allowed an identification not just with specific characters but with the scenarios of desire themselves.\n\nImages of gay men offered a way to gain a more flexible foothold among the extant sex and gender scenarios, evading a premature consolidation of fantasmatic desires and identifications within the limits of Cold War gender ideology.\n\nSome lesbians report a youthful, transitional reliance on male gay scenarios. They signaled, in the context of a relative sparsity of lesbian images, at least that not everyone is straight. For Cheryl Clarke, an African American, Baldwin figured the prospect of queer, black authorship. _Another Country_ , despite having nothing positive to say about lesbianism, \"made me imagine freedom from traditional monogamous heterosexuality and set me to thinking about the possibility of a 'variant' life.\" As a prominent dissident intellectual\u2014one who preferred not to live in the land of the free\u2014Baldwin represented broader prospects of alternative thought. Bia Lowe describes a double displacement, whereby she invested in actors who were admired for their virility while implicitly embodying an element of gayness. Lowe got from her mother the idea that actors such as Laurence Olivier and Rock Hudson\n\nwere men with enormous sex appeal and, now I realize, not without that certain je ne sais quoi [ _sic_ ]. Was I unknowingly drawn to gay men because of the model of my mother? Or because, as a budding Miss H, I was protected by them from the failure of heterosexual contact? Because gay men reminded me more of brothers than of fathers? Until I came out, I might as well have been a gay man, for male was the only gender I would spot in the \"pathology\" of same-sex love. I read _Giovanni's Room_ , saw _The Boys in the Band_. I eyed my mother's string of interior decorators. I listened for clues to my own stirrings in the swells and swirl of Tchaikovsky's music.\n\nAs well as race, class identifications may tangle with sexuality and gender. Sue-Ellen Case says she became queer through an adolescent identification with Arthur Rimbaud. Valerie Walkerdine remarks how, watching _Rocky II_ , she found that her identification was taking an unexpected path. She had not expected to enjoy its \"macho sexism,\" but she found herself identifying with the class feeling that informs it. \"The film brought me up against such memories of pain and struggle and class that it made me cry.... I too wanted Rocky to win. Indeed, I _was_ Rocky\u2014struggling, fighting, crying to get out.\" Class feeling overwhelmed gender principles\u2014or, rather, enabled a more complex experience of them.\n\nThere may be downsides to these irregular identifications. As always with appropriations, one gets more than one had bargained for. For Kaplan, to read Baldwin in the context of the limited ideology of femininity that prevailed in the late 1950s meant engaging \"not only in an empathetic, even desiring, identification with the figures of masculinity in his texts, but also (if only subliminally) in a repudiation of the feminine, if not exactly of women.\" Segal pursued her interest in gay men to the extent of becoming pregnant by and marrying one; it didn't work out.\n\n**SUBSTITUTIONS, CONFLATIONS, REVERSALS, LOOPS**\n\nAnthropologists have held that some societies organize same-sex passion around age, others around gender. In chapter 2, correspondingly, a complementarity model and a relative-gender model emerged. But these models are not always discrete. In Pai Hsien-yung's Taiwanese novel _Crystal Boys_ , the prostitutes are all boys and the punters are older men; the informing imagery, in this militaristic society, is of fathers and sons. This seems to be a same-sex community structured primarily around age. Nonetheless, the boys are called \"fairies,\" even the ones who might seem masculine: one who is \"husky as an ox\" is called \"Little Fairy,\" and one with \"a wonderful physique\"\u2014\"broad shoulders, and a muscular chest\"\u2014is called \"the Butch Queen.\"\n\nWhat seems to be happening here is a _conflation_ or _substitution_ of roles: since both boys and women figure subordination, they may be blurred together, or the one may stand for the other. It is beyond the scope of this book to track the range and intricacy of fantasy. In this section I mean to unravel some exemplary instances of fantasmatic maneuvering and explicit role-play, discovering a nexus of complications that seem especially prominent as ways of elaborating gay psychic experience. I distinguish substitutions, conflations, reversals, and loops.\n\nSometimes roles are substituted for (allegedly) tactical reasons. In single-sex institutions one male may fuck another without losing status, so long as he takes the \"active\" part and the other is regarded as a stand-in woman. In 1922, Alec Waugh invoked the substitutability of boys and women as an explanation for homosexuality in boarding schools:\n\nIn this environment there is nothing unnatural about the attraction exercised by a small boy over an elder one. A small boy is the nearest approach possible to the feminine ideal. Indeed a small boy at a Public School has many of the characteristics that a man would hope and expect to find in a woman. He is small, weak, and stands in need of protection.\n\nThat is a heteronormative way of putting it, of course; we might say that the attraction of women resides in their \"boyish\" characteristics. Boy-love as a substitute for girl-love is widely displayed in prison dramas, including _Little Ol' Boy_ by Albert Bein, _\"Now Barabbas...\"_ [ _sic_ ] by William Douglas Home, _Deathwatch_ by Jean Genet, and _Fortune and Men's Eyes_ by John Herbert.\n\nFreud's comment on same-sex passion among the Greeks finds that love of boys is really about love of women. \"What excited a man's love was not the _masculine_ character of a boy, but his physical resemblance to a woman as well as his feminine mental qualities\u2014his shyness, his modesty and his need for instruction and assistance.\" So gender hierarchy is maintained after all\u2014so long as you go along with the Victorian notion of what \"a woman\" is like. Foucault believes the opposite of the Greeks:\n\nit was the juvenile body with its peculiar charm that was regularly suggested as the \"right object\" of pleasure. And it would be a mistake to think that its traits were valued because of what they shared with feminine beauty. They were appreciated in themselves or in their juxtaposition with the signs and guarantees of a developing virility.\n\nCommenting on an earlier version of some of the ideas in the current study, David Halperin insists that not all systems conflate all subordinations, or in the same way. This is indeed my point; the conflations I observe are particular strategic adjustments, not instances of an essential process.\n\nIt is noticeable that substitution and conflation of roles are more commonly posited of subordinate figures. The masculine, together with the adult, the established, and the white, appears simply as itself, and claims the authority to reposition its others. Pai Hsien-yung's crystal boys, having run away from home, are lower class as well as young and feminine. Arthur, in Alan Hollinghurst's _Swimming-Pool Library_ , is black, and also younger and considerably poorer. Such conflations illustrate the malleability of fantasy, but also the ruthlessness of its appropriations, and its disregard for the stability of the subordinated person. If relative femininity and youth are regarded as metonymic, or perhaps even the same, then the conflation of roles may enable a more elaborate fantasmatic discourse; alternatively, it may lead to a confused identity.\n\nEdmund White reports: everyone on the New York scene was doing it: \"We were all obsessed with fantasies back then, which we kept exploring until they became absurd. One boy even said to me: 'I do father-son, sailor-slut, older brother\u2013younger brother, black rapist\u2013white secretary, trucker-hitchhiker, and a virgin couple on their wedding night.'\" You can slot into one or the other; it's all the same kind of thing; hierarchy is the point, as much as the particular terms in which it is framed. Perhaps it is a mistake to suppose an original menu of discrete dominations and subordinations, wherein everything was simply itself. In fact, an attraction of same-sex relations may reside in their potential to invoke, simultaneously, several social hierarchies in complicated combinations.\n\nThe fantasy identifications discussed so far seem to involve fairly simple substitutions; a person is able to cultivate feelings, typically of empowerment or submission, that would be hard to access through his or her regular identity. Of course, fantasy is not always so conveniently labeled or so comfortably experienced. Often in psychic life there is a tendency for roles to be _reversed_ \u2014such that one fantasizes oneself as the other. As Jacqueline Rose puts it, with case histories of Jewish Holocaust survivors in view, \"being a victim does not stop you from identifying with the aggressor; being an aggressor does not stop you from identifying with the victim.\" Role reversal was common in the cross-class liaison of the mid-twentieth century, where the bit of rough trade might be called upon to fuck his social superior.\n\nI remarked at the start of this book how Reginald Shepherd's desire for men is entangled with his experience of racial hierarchy. Shepherd both desires white men and has himself always wanted to be white. He asks, \"How much of wanting another man is the desire to be that man?\" The connection works quite literally: by being seen with a white lover, Shepherd becomes \"an honorary white man.\" He believes that sex is about dominance and submission: \"For a gay man both roles are simultaneously available.\" Gary Fisher, with similar issues in mind, ponders _Billy Budd_ : Melville might have been \"a bit more generous; he might have asked us to feel instead of to just watch, feel what it is to be victim and victimizer; white victim and then black victim; white victimizer and black victimizer; asked us to feel, to study and enjoy all the permutations, all the variations on a theme in this text.\"\n\nSuch reversals may be facilitated in gay relations. In the previous chapter I followed John Fletcher's argument about the breaking of the fragile barrier between identification and desire, to show that such confounding of the distinction between desire-to-be and desire-for is endemic in same-sex passion. Earl Jackson Jr. frames this factor in a revision of Mulveyan, cinematic terms: whereas the viewing pleasures of the heterosexual male may most easily entail identification with the man and objectification of the woman, the gay viewer \"regularly identifies with the figure he sexually objectifies. In other words, he experiences a coalescence of drives that are radically dichotomized in his heterosexual male counterpart.\" This is specially true in pornography. In my view Jackson may underestimate the perversity (to use the normative term) in much heterosexual passion; in horror films men may identify with the female victim. But we should take Jackson's point, that gay people may more readily cross the heavily policed line between identity and desire, making it relatively convenient to cultivate complex scenarios. In Robert Chesney's play _Jerker, or the Helping Hand_ , J. R. declares his engagement with both the prince and the princess: \"I was always more interested in _him_ than in the fairy tale princesses\u2014Snow White, Cinderella, whatever. _I_ identified with the Sleeping _Beauty_ : _I_ wanted that kiss.\" J. R. is committed to both characters: he wants to be both the prince who kisses and the princess who gets kissed\u2014and awakened into sexual response, such that the prince gets kissed in return.\n\nFinally, we may observe instances in which passion takes a _loop_ through one kind of identification or desire, in order to gain a role in relation to another. This was a resource for the young Edmund White, as he struggled with adolescent passion. In _A Boy's Own Story_ he falls for the gym teacher, Mr. Pouchet, and imagines himself to be Pouchet's girlfriend. White was prepared to be \"Julie or Helen or whoever else, just so long as I was in his mind somehow.\" His desire loops through the desire of the teacher and the person of the girlfriend.\n\nJonathan Dollimore describes a threesome in which a bisexual male (I call him \"the protagonist\") watches a man fucking with a woman:\n\nHis identifications here are multiple: he identifies with the man (he wants to be in his position, having sex with the woman) but he also wants to be her. And I mean _be_ her: he doesn't just want to be in her position and have the man fuck him as himself (though he wants that too); no, he wants to be fucked by the man with himself in the position of, which is to say, as, the woman.\n\nThe protagonist has desire-to-be the man, but this is for a purpose: \"he wants to be in [the man's] position, having sex with the woman.\" He wants to fuck the woman, and imagines doing it through the agency of the man. His masculine activity is routed through another. Elsewhere, we have found desire-for and desire-to-be to be autonomous, in the sense that nothing about the one can reliably be inferred from the other (drag artistes may be straight). In this case, one facilitates the other:\n\ndesire-to-be the man   desire-for the woman\n\nBoth these desires seem to secure the masculinity of the man. This, however, is not the protagonist's goal, Dollimore insists: \"he also wants to be her. And I mean _be_ her.\"\n\ndesire-for the man   desire-to-be the woman\n\nBisexuality is usually glossed, quite simply, as a static split: desire-for both genders. This is not an adequate account of the positionings of Dollimore's protagonist: he is performing an elaborate psychic loop through the possible permutations. He knows the pleasure of being fucked by a man, he adds, but in this scenario \"he also wants to be the woman; he wants to be fucked by the man in a way he imagines\u2014fantasizes\u2014only a woman can be.\" This way of putting it implies another variant:\n\ndesire-to-be the woman   desire-for the man\n\n\"Maybe he desires the man through her.\"\n\nIs this the goal, then? For the bisexual protagonist, Dollimore admits, \"the sexual attractiveness of the male is heightened by the fact that the latter is apparently desired by the woman; he excites the more because he is desired by her\" (529). Should we declare the protagonist in bad faith, then? His desire-for and desire-to-be take him in a loop through the woman, but his true goal is to share the identity and desire of the man (Dollimore is aware that the woman might be the most objectified figure in the scenario as he presents it; indeed, she might be in effect extinguished). However, I believe that would be a false inference. The care with which the sequence is elaborated indicates that the pleasure is in the entire process, not in any singular end product. Indeed, having reached the point of desiring the man, the protagonist may well go back to the beginning and, from the position of the man, desire the woman.\n\nGuy Willard's novel _Mirrors of Narcissus_ offers a complex sequence of the desires and identities that one young man might experience. We first see Guy looking at his reflection after working out, and enjoying the thought that women in the dorm opposite can see him through the window. His desire-to-be appears suitably masculine, and adequately depends on heterosexual validation. However, desire-to-be crosses into desire-for when he masturbates over a picture of a bodybuilder. He likes the thought of other men fancying his girlfriend, Christine. \"It was as if she were my doll and I was dressing her up to please the guys. And my pleasure in it was ignited by a process of reflection: the other boys' excitement excited me.\" So far so good, though the other boys seem rather prominent in the fantasy. Then identification turns to desire: \"I imagined that all the male attention she drew to her stuck to the surface of her skin, so that when I caressed her, I was caressing those male glances.... This was the only way I could get close to a boy.\" Guy contrives the positions of Christine and himself in lovemaking so that she appears like a boy: \"I lay on my back and she sat atop me straddling my thighs, the better to stroke my erection. From the way she was sitting it looked as if my upthrusting penis were hers, completing the illusion that she was a boy\" (26). The consequent orgasm may be credited to all three of them. Christine sees that Guy is turned on; she is open-minded, but he conceals the extent of his gay interest.\n\nThus far, Guy's desires may be represented like this:\n\ndesire-to-be M   desire-for F   desire-for M\n\nGuy manages to lever the woman out of the loop when he is invited to model as Narcissus by a gay artist, Peter. Guy is excited by the thought that Peter desires him. When the painting is finished, Guy finds himself aroused by the image of himself (mirroring again the Narcissus in the painting, of course).\n\nGuy falls for his straight roommate, Scott, who, in the manner we have seen elsewhere, is slightly feminine\u2014though not, of course, effeminate: \"The eyes were what held my attention. They were large and soulful, and hinted of artistic sensibilities. As if to confirm this, his skin was very fair, a shade too delicate for a boy, though it didn't make him effeminate in any way\" (63). Guy seeks to approach Scott by suggesting to Christine that they help him to lose his virginity. She refuses to do this; instead, she and Scott sleep together and discover their love for one another. For Guy, though only in imagination, this completes a loop: \"Through the channel of Christine's body, Scott and I were now one, linked by the most basic bonds vouchsafed to unrelated strangers. My skin, in nakedness, had touched Christine's, and her skin, in nakedness had touched his\" (169).\n\ndesire-for F   desire-for M\n\nThe implication in the novel is that Guy was really gay all along, and using loop strategies, exploitatively, to sort himself out. Thus any playful or adventurous potential in his fantasies is set aside. Christine's refusal to collaborate is wise, according to the narrative; Scott, despite his responsiveness to Guy's advances on one occasion, is \"Perfectly normal\" (187). The simplest models are adequate after all. _Mirrors of Narcissus_ finally offers a traditional view of dissident fantasy.\n\n**THE SUBJECT IN POSTSTRUCTURALISM**\n\nAs film theory has repudiated the deterministic notion of spectatorship found initially in Mulvey, it has sometimes imagined a free play of identities. Penley declares: \"An important emphasis has been placed on the subject's ability to assume, successively, all the available positions in the fantasmatic scenario.\" The mobility and intricacies of fantasy tend to undermine expectations of stability, thereby facilitating an elaborate range of libidinal investments. This account is too voluntaristic for Teresa de Lauretis. She objects to\n\nthe optimistically silly notion of an unbounded mobility of identities for the spectator-subject; that is to say, any spectator would be able to assume and shift between a variety of identificatory positions, would be able to pick and choose any or all of the subject-positions inscribed in the film regardless of gender or sexual difference, to say nothing of other kinds of difference.\n\nEven in these postmodern times, identity has to have some kind of structure, however provisional. I have shown readers investing in diverse aspects of scenarios, but that does not mean there are no constraints; indeed, movement within a scenario may help to keep it in place. This issue is the theme of this section.\n\nThe potential for mobility in psychic identifications has been a persistent motif in queer and poststructuralist thought (queer theory is best understood as a kind of poststructuralism). Judith Butler concludes _Gender Trouble_ with the prospect that we might evade the oppressions of difference by elaborating a multiplicity of fantasies and practices: \"Cultural configurations of sex and gender might then proliferate or, rather, their present proliferation might then become articulable within the discourses that establish intelligible cultural life, confounding the very binarism of sex, and exposing its fundamental unnaturalness.\" If we recognized innumerable sexualities, norms and stigma would collapse. Henning Bech reaches a compatible conclusion in his book _When Men Meet_. He argues that the admiration of gay men for masculinity has now become a harmless style choice:\n\nthe more the surfaces are detached and become autonomous, the more the roles are severed from nature, the more accessible they become for staging and pleasure, the more they can be treated _as_ surfaces, _as_ roles, _as_ images.... We can finally reach the point at which the dangerous in masculinity is maintained all the while it's suspended, the violence, the domination, the power display; it can stop when it isn't fun any more.\n\nThe more we experiment with masculinity, in other words, the less significance it has. Bech foresees the demise of the masculine\/feminine hierarchy.\n\nI am struck more by the repetition and fixity of fantasy, in the experience of very many people. I conceded in chapter 2 that my juggling of \"M\" and \"F\" might impress the reader as too standardized. It goes against the general postmodern\/poststructuralist truism that _any_ identity is, and should be, provisional, unstable. I do believe that psychic life is manufactured out of the typical building blocks of gender, age, class, race, and sexual orientation. These are the structures in which we live, and ongoing psychic life is an attempt to cope with the attendant triumphs and humiliations. I envisage our selfhoods as constructed through a kind of _bricolage_ \u2014the term proposed by Claude L\u00e9vi-Strauss to describe the development of cultures in anthropology. In John Clarke's account this means a piecemeal, appropriative process: \"the re-ordering and re-contextualisation of objects to communicate fresh meanings, within a total system of significances, which already includes prior and sedimented meanings attached to the objects used.\" As I have said, it is because the permutations are so numerous and so intricate that the outcome is experienced by many as implying the uniqueness of the individual, and, often, his or her ultimate freedom from the constraints of history and ideology.\n\nWhat is difficult to articulate, in the models I have been using, is the fourth dimension: time. In the formation of an individual subject, there will be moments of crystallization, in which a specific set of identifications and object-choices will become established, while others are repudiated. Fantasies attempt to manage those traumatic moments, often in tangled form; the individual subject, at any point of time, is the product of a sequence of pioneering and entrenched selves. Through these successive engagements, the subject is constituted.\n\nThe postmodern notion that one might manage better without some kind of working identity is intensely romantic. R. D. Laing, the 1960s theorist of damaged identity, remarks: \"It is difficult to imagine many who would choose unlimited freedom within a nexus of personal relations, if anything they did had no significance for anyone else. Would anyone choose freedom if nothing he did mattered to anyone?\" Would a boundless indeterminacy be sexy? Dollimore invokes the German film _Taxi zum Klo_ ( _Taxi to the Toilet_ : Frank Ripploh, 1981), where Frank passes a note through to the next cubicle asking, \"What are you into?\" The reply is, \"Everything. Anything.\" Frank walks out in disgust. No opposition, no substance, no turn-on. It is one of Foucault's key insights: power always entails\u2014is experienced only through\u2014resistance.\n\nIdentity, according to Laing, is neither essential, nor something you adopt and proclaim, like a political slogan. More fundamentally, it is \"that whereby one feels one is _the same_ , in this place, this time as at that time and at that place, past or future; it is that whereby one is identified. I have the impression that most people tend to come to feel that they are the same continuous beings through womb, to tomb. And that this 'identity,' the more it is phantasy, is the more intensely defended\" (86; Laing's emphasis). Because gay people may be out of touch with their birth families and closeted at work, they may appear to be unconstrained. A potential for anomie in gay culture is the theme of Andrew Holleran's _Dancer from the Dance_. \"We are free to do anything, live anywhere, it doesn't matter. We're completely free and that's the horror,\" Malone opines. \"'Perhaps you would like a Valium,\" Sutherland responds.\n\nThe extent to which one might be bound to an identity, and the consequences of abandoning it, are explored in Kevin Smith's film _Chasing Amy_ (1997). Placing his friendship and artistic collaboration with Banky in jeopardy, Holden falls in love with Alyssa, although he knows she is a lesbian. His feelings become unbearably intense, so he tells her of them. Alyssa's response is to climb out of the car and start hitchhiking. Has she no comment? Yes: \"Fuck you!\" It is unfair of Holden to unburden his soul to her, because by ignoring her declared lesbianism he is refusing to take her seriously: \"Do you remember for one fucking second who I am?\" \"People change,\" Holden replies. \"Oh, it's that simple. You fall in love with me and want a romantic relationship. Nothing changes for you.... I can't just get into a relationship with you without throwing my whole fucking world into upheaval\" (my elision). There's bound to be a period of adjustment, Holden replies. \"There's no period of adjustment, Holden, I am fucking gay. That's who I am, and you assume that I can just turn all that around because you've got a fucking crush!\" She follows him back to the car, however; they embrace heavily; next thing it's morning and they're sleeping on the couch together.\n\nAlyssa does seem able to abandon her declared identity after all. The outcome is notably uneven, however, as Holden falls into complacent assumptions. He is devastated to learn that Alyssa's adaptable identity includes a history of experimentation with boys at school. Also, he presumes that he can reengineer his relations with Banky, who is evidently jealous, by inviting him to explore his (alleged) latent homosexuality. In some circumstances some people may be able to change some parts of their identities in some directions, but they will still be carrying all kinds of debris, and indeed esteem, from their former selves, and the outcome may be uncomfortable.\n\nGenerally, erotic imagery proves amazingly stubborn, as people who have tried to change through psychotherapy and religious devotion know. In a memorable formulation, Lynne Segal presents intrepid fantasmatic adventuring as characteristic of psychic life: \"We insert ourselves, whatever our sex, at one and the same moment as both active and passive, powerful and powerless, giving and receiving: desire flows through binaries in all directions at once, all of the time.\" Yet I can also envisage a case for the opposite extreme: the fixation. The obsessional fetishist may be living more intensely than people who gain an easy, moderate pleasure, either from unconsidered custom, or from almost anything.\n\nJeanette Winterson's protagonist in _The PowerBook_ supposes that the idiom of the computer adds a new impetus to the idea of freedom to be who you will, if only for one night. This prospect is emblematic of our ability to rewrite the stories in which we figure: \"there is always a new beginning, a different end. I can change the story. I am the story.\" We assume that the screen we have open at the moment represents our lives, but there is another, less familiar window behind that, and yet another beyond that. \"We think of ourselves as close and finite, when we are multiple and infinite\" (103). However, _The PowerBook_ does not actually exemplify such freedom. Ali (as the narrator is most often called) engages in a sultry affair with a married woman, who is reluctant to leave her husband. This passion governs Ali's electronic explorations: \"That's why I trawl my screen like a beachcomber\u2014looking for you, looking for me, trying to see through the disguise. I guess I've been looking for us both all my life\" (64). This is hardly freedom, and hardly the sign of a new electronic age. It is a quest as purposeful and traditional as those pursued by the heroic knights of epic and romance\u2014whose stories are intercut with Ali's affair. The challenge to forsake everything, follow your heart, and live for the magical twosome is hardly a new narrative motif.\n\nThe invocation of freedom sits oddly with the air of obsession in _The PowerBook_. Indeed, links with other novels by Winterson suggest that she herself (like very many authors) is working out some compulsive stories of her own. The angry invocation of narrow childhood circumstances is reminiscent of _Oranges Are Not the Only Fruit_ , and her lover's flaming red hair and autocratic husband recall Louise in _Written on the Body_. Indeed, the latter novel and _The PowerBook_ suggest a dynamic, whereby the narrator is a masculinized figure (in _Written on the Body_ it is unclear whether s\/he is a woman or a man) who feels impelled to compete with the husband. The narrator in _The PowerBook_ identifies herself with male heroes, permitting little sense of herself as a woman; she doesn't allow that any particular pressures might attend a lesbian affair. In _The Passion_ , Henri venerates Napoleon, while Villanelle makes love to another woman in the guise of a man, and has to endure the sight through a window of the domestic affection of her lover and her husband. The patterns in Winterson's writing mark the extent to which we do not control our own stories.\n\nIt may be observed also that _The PowerBook_ affords an instance of the thesis I develop in the next chapter, concerning the effacement and ineluctability of power in our relationships. Ali insists that there is no legitimate overlap between power and love. However, the main narrative shows her seeking to control her lover by imposing her idea of how they should proceed. She complains specifically when she feels herself unable to exclude her lover's husband. \"The only power I have is the negative power of withdrawal.... A relationship where one person has no power or negative power, isn't a relationship, it's the bond between master and slave\" (187; my elision). Of course, this would explain Ali's male identification: men have power. While fantasy may prove transformative, it may also trap the subject in fruitless and perhaps dangerous compulsions.\n\n**THE FRONTIERS OF FANTASY**\n\nThe psychic investments discussed thus far in the present chapter are in fact less about freedom than the discovery of a flexible, but apparently suitable, identity. In other contexts, the scope of the fantasy scenario is a problem: it harbors rapists, stalkers, habitual familial abusers, serial killers. Not all roles can be legitimated, even within the superpermissive regime of Queer. There is a persisting problem with individuals who want to force their practices upon others. This should not surprise us. While the hierarchies of gender, age, class, and race often appear benign, and may afford opportunities for rewarding sexual adventures, it is evident that the social and political system, which sponsors such fantasies, can operate in intimidating and brutal ways when a serious threat is perceived. As these hierarchies are internalized by individuals and groups, often as competing psychic and social demands, they are bound to produce strenuous techniques of psychic management and vehement attempts to gain control of self and others. Violent mental disturbance, in other words, is what you would expect in societies like ours.\n\nMark Ravenhill's play _Shopping and Fucking_ represents the _m\u00e9nage \u00e0 trois_ of Robbie, Lulu, and Mark as unexceptional but foundering because of Mark's substance abuse. He decides, as part of his cure, to avoid dependency of any kind, but young Gary's story of abuse by his stepfather draws Mark in after all. Gary, however, doesn't want to be loved and protected, he wants to be owned and hurt. So why not gratify him? \"When someone's paying, someone wants something and they're paying, then you do it. Nothing right. Nothing wrong. It's a deal,\" Gary says. Is this right? For Robbie in particular, more or less anything goes. Like him, many of us are learning to acknowledge and accommodate a range of \"perverse\" practices that previously would have been thought embarrassing, if not disgusting. But does that mean we can have any experience that we can afford to pay for? When they operate a telephone chat line, Lulu is eventually sickened when a scenario comes too close to life. The question, then, is this: How far is fantasy liberating, how far constraining?\n\nThe protagonist in _Frisk_ by Dennis Cooper is drawn \"uncontrollably\" to a particular \"physical type.\" Ever since the age of thirteen, when he saw photos of an apparently murdered model, the desires of the narrator, Dennis, have been fixed on such a boy and such a scene (\"It looked as if someone had set off a bomb in his rectum\"; 27). The photos, he says, \"went on to completely direct or destroy my life\" (30). Five years later Dennis meets Henry, who claims to have been the model: the photos were fakes. But that doesn't dispel the fantasy.\n\nThe novel takes place at this interface between actuality and fantasy. Samson is Dennis's ideal type, so he maintains his fantasy scenario in his imagination during conventional lovemaking: \"In reality I was caressing him. In my head I'd be grabbing objects off the night table, crashing his skull, then mutilating his body, especially his ass, while he tried to dissuade me from murdering him in a brain-damaged voice\" (34). One night Dennis loses it with Samson and punches him repeatedly. Samson isn't upset: \"'I was _so_ out of it. And you were _so_ weird'\" (35; Cooper's emphases). However, Dennis is afraid, and for a few years avoids\n\nserious, ongoing relationships as a precaution. It wasn't that I didn't fantasize murdering hustlers. It's just that I tend to be too scared or shy the first few times I sleep with someone to do what I actually want. The worst that could, and did, happen was I'd get a little too rough. But the hustler would stop me, or I'd stop myself, before things became more than conventionally kinky, as far as he knew. (36)\n\nWhat is inhibiting Dennis from acting out his fantasies? He seems to have no trouble getting boys to go with him, especially when they are on drugs; he has money (his parents send it to him). He writes a story, in which Joe's wish to be hurt has placed him in the power of Gary, who fantasizes about murdering people. \"'But something usually stops me. I think it's beauty. But whatever it is, it's not there with you. I really want to kill you,'\" Gary says (63). He is not impressed by the conventional S\/M notion that the bottom is in charge. \"'Well, um, you shouldn't do it, because I don't want you to, and I'm half of this,'\" Joe protests. \"'If I don't do it,' Gary said, 'that'll be why. But it's the only reason, which is strange, because there should be others, right?'\" (64).\n\nWe live in a world, _Frisk_ is showing, in which it is not easy to supply better reasons. This is not a new dilemma (\"'I mean, I know there's no God'\"\u2014Dennis; 69). Compare Dorian Gray's exclamation when Lord Henry is executing his initial seduction: \"'Stop!' faltered Dorian Gray, 'Stop! you bewilder me. I don't know what to say. There is some answer to you, but I cannot find it.'\" Dennis's old friend Julian says he understands the appeal of murder, but is shocked by the idea of doing it. \"'I'm not being moralistic. I'm talking fairness, which is not a particularly bad rule to live by, as rules go'\" (112). This seems right, but it scarcely measures up to the intensity of Dennis's compulsion.\n\nUltimately he is inhibited, more simply, by the very extremity of the gap between ordinary life and his fantasy. Henry wants details of the photos in which he appeared, but they seem preposterous in real-life conversation: \"Spoken aloud, the descriptions seemed much more pretentious, ridiculous, amoral... something, than they'd ever been in the secret, uncritical world of my fantasies\" (30; Cooper's pause). This gap between what you can imagine and what you can actually, plausibly, say or do is neatly illustrated when Dennis fantasizes about sacrificing a boy called Finn on the top of an Aztec pyramid. \"Part of me wanted to kill and dismember him, which I probably could have done without getting arrested,\" Dennis reports; \"but most of me gave him a towel, then humored him until he left\" (38).\n\nDennis writes letters to his old friend Julian, describing how he has been killing boys in Amsterdam. The murders become more violent and disgusting (at least to me). Julian and his younger brother Kevin, who's always had a thing for Dennis, come to rescue him. Kevin lights on the idea of restaging the photos, and hence Dennis's original trauma. \"I'd wind up cured or exorcised or something\" (121). It appears to work; Julian goes home to his partner, while Kevin and Dennis stay together. But is Dennis cured? It all depends, as Claudia Card says of S\/M generally, on whether the cathartic or the addiction model is correct. The former means that occasional controlled indulgence may enable painful psychic material to be disposed of safely; the latter that enactment may produce a need for more intensity or more frequency. The closure of _Frisk_ depends on the cathartic model (Dennis's need to act out his desires diminishes), but elsewhere in the novel Dennis's obsessions feed on themselves and the addiction model appears to reign; certainly it seems to claim intense imaginative energy. Nor is mere indulgence in fantasy without eventual consequence. Dennis explains why he was unfazed by the first (pretended) murder: \"I guess I'd fantasized killing a boy for so long that all the truth did was fill in details. The feeling was already planned and decided for ten years at least\" (92). _Frisk_ does not flinch from the thought that fantasmatic desires may prove overwhelmingly addictive, no matter how repulsive they are.\n\nIn fact, although he has not actually murdered anyone, Dennis's motives in writing the letters were not therapeutic: \"I realized at some point that I couldn't and wouldn't kill anyone, no matter how persuasive the fantasy is.\" He was trying to attract an accomplice\u2014someone to \"come here, and give me the courage or amorality or whatever to actually kill somebody in league with them\" (123). Notice also the ambivalence in Dennis's key statement, that the snuff photos \"went on to completely direct or destroy my life\" (30). \"Destroy\" speaks Dennis's revulsion, but \"direct\" is more complacent.\n\nDennis's unrepentant streak nourishes and is nourished by a disturbing cross-current in the novel: the idea that the boys he approaches are neither uninterested nor unwilling. Henry is still eager to please, eager to be appreciated; Joe at the last moment appears to consent to his own murder; Pierre's partner predicts that a boy who has escaped from a \"kiddie porn ring\" will be terrified, but he blows a kiss to media reporters (86). One might argue, anyway, that consent is often no more than _internalized ideology_. For instance, when in the marriage service the partners say \"I will,\" this is perhaps because they are taking it for granted that matrimony is their natural destiny. It would be open to the sexual dissident to interrupt with an impediment, namely that marriage colludes with the wish of the state to control reproduction by fixing gendered and sexual roles. The bride and groom believe they are choosing freely, but they have been systematically conditioned. Where we consent, therefore, we may be most deluded. Not much can be done about this, but it undermines any straightforward reliance upon consent as an ethical and political principle.\n\nNonetheless, many readers may reflect that the ascription of readiness to the boys is all too convenient for Dennis. A novel is a kind of fantasy scenario, in which the characters may be arranged to suit an imaginative contrivance. The novel says, \"this is how people are,\" but readers may declare the outcome implausible or immoral\u2014merely (we may say) a fantasy. _Frisk_ actually draws attention to the contrivance of fiction. The narrative slips repeatedly between invention and (pretended) reportage, first and third person. Dennis presents himself continually as if he were in a film. \"I should include some reaction shots here,\" he says, meaning some indication of how he reacted to his own manic assault on Samson. \"But I doubt I had many. I felt numb, blank, so my face probably followed suit\" (34). He watches slasher movies avidly; his most substantial conversation is with Pierre, a porn star and hustler whom he has hired. He admits that writing down his fantasies \"'was and still is exciting in a pornographic way'\" (123). Such an explicit preference for fantasy over reality does not encourage the reader to trust Dennis's perceptions of other people. It is, however, a logical outcome to the indulgence in fantasy that some theorists are encouraging; _Frisk_ is about how you police fantasy when experimentation is offered as a good in itself.\n\nIn his next novel, _Try_ , Cooper presents similarly violent scenarios, but largely from the viewpoint of the teenager. Ziggy is used sexually by his adoptive fathers, Brice and Roger:\n\nZiggy's happy. It's drug-induced, no doubt. Still, for whatever reason, he suddenly knows, like, for sure, that a huge part of... sexual abuse, at least for him, is how he loves being a target for such intense feelings, especially from someone who knows him and isn't just stupidly thinking he's cute or whatever. That's why he hasn't killed Brice, or hired a hit man like other abused-type teens do.\n\nWe may grant that Ziggy's contentment is an interesting and important phenomenon, but it doesn't justify the exploitation. Indeed, the passage quoted acknowledges the extreme distress which \"abused-type teens\" may experience. Apart from the power (age and wealth) difference, the use of drugs negates any prospect of informed consent. Ziggy is talked into a threesome he doesn't really want; he is suddenly depressed and bursts into tears, but Roger is oblivious: \"'If you loved me...'\u2014Ziggy slugs\u2014'... you wouldn't _rim_ me while I'm _crying_.' This time he hits Roger's head so violently it's knocked loose. 'That's the _truth_ , you... _scum_!'\" (149; Cooper's pauses and emphases).\n\nThe quest for accomplices in Cooper's books discloses a frightening world of desperate, undernourished youngsters, lacking any evident parental or school guidance, oblivious to the risk of AIDS, taking without hesitation any drug they are offered, absurdly possessed by heavy metal bands, and making themselves available to far more powerful men in return for the most meager emotional consolations. Henry, for instance, has commodified himself to the point where he asks everyone he has sex with, \"'If you could change one thing about the way I was acting a minute ago, what would that be?'\" \"'You talk too much,' the guy said\" ( _Frisk_ , 8\u20139).\n\n_Frisk_ constitutes a limit case for any progressive, poststructuralist, or queer wish, that all fantasies might be exhilarating and all sexualities viable. If my contention (which I pursue in the next chapter) about the social and political constitution of desire is right, then it means, on the one hand, that we have to accept as inevitable and only realistic the lineaments of power relations in our sexualities. On the other hand, it means also that gross psychic deformations will appear, even as capitalism and patriarchy produce horrific exploitations. While in the first perspective it is vain to expect that the overwhelming run of our desire can be redirected, in the second perspective there will be perilous consequences to some fantasy scenarios, and it will be necessary to intervene. These consequences will not always be at the gruesome level displayed in Cooper's writings; there are other, meaner, and narrower kinds of fixation, which produce barely tolerable bullying, bigotry, and disconfirmation. _Frisk_ and _Try_ are valuable books because they take some readers at least into the world of abuse without abandoning them there, but also without harboring impractical prospects for reconciliation of aberrant desire.\n\nA later novel, _Guide_ , displays Dennis in a more realistic setting. Alongside murderous fantasies, the narrator exhibits a rather subdued, lovelorn stance. Contrary to the earlier novels, he doesn't get the boys he desires. Luke moves in to his apartment, to the alarm of Andy who has seen the novels: \"'Have you read them? They're all about serial murderers. And all the victims are boys. And all the boys look like you.'\" Luke doesn't feel threatened: \"'I think Dennis is more sort of someone who lives in his head,'\" he opines (170). Let's hope so.\n**4**\n\n**POWER**\n\nMen come to the brothel in Jean Genet's _The Balcony_ to act out scenarios of power. There is the Bishop and the Penitent, the General and his horse, the Ma\u00eetresse and the Beggar, and the Judge, the Executioner, and the Thief.\n\nThere are two kings of France with coronation ceremonies and different rituals, an admiral at the stern of his sinking destroyer, a dey of Algiers surrendering, a fireman putting out a fire, a goat attached to a stake, a housewife returning from market, a pickpocket, a robbed man who's bound and beaten up, a Saint Sebastian, a farmer in his barn... a missionary dying on the cross, and Christ in person.\n\nThere is no chief of police, as the actual incumbent ruefully observes. However, after he has put down the rebellion, the traditional authority figures wilt and men queue up to enter his scenario. The meaning of this fable is that, traditionally, the imagery of the chief of police is insufficiently charismatic; he is not recognized as part of the establishment, he is too functional. He has not figured in the fantasies of citizens. Now the fascist state has arrived, and the policeman bulks large in the psyches of citizens. His symbol is a man-sized phallus, his counterpart a slave, and his setting a mausoleum. Genet is showing that our sexual fantasies depend on the power arrangements in our societies. This chapter aims to appraise, in outline, the relations between sexual practice and fantasy, social organization, and hierarchies of gender, class, age, and race.\n\n**TWO BOYS TOGETHER CLINGING**\n\nAs a boy, Paul Monette sees his incipient queerness as a failure of manhood. He conceives an attachment to Elizabeth Taylor, he says in his autobiography _Becoming a Man_ : \"I'm not quite sure what I'd identified with, but it seemed to amount to a kind of _emotional_ drag\u2014trying on those steamy, gaudy feelings.\" He and two friends discover all this for themselves; \"If someone had told me I was exhibiting a sensibility, I probably would've frozen in horror, terrified my wrists were going limp.\" A teacher sees them camping around, and notes: \"'Paul spends too much time acting silly with his day student play-mates. It's not healthy. He's got a lot of growing up to do if he wants to be a man.'\"\n\nPaul is not happy with his relatively feminine stance. He \"hated the soft androgyny of [his] body, which somehow managed to be both scrawny and plump at once\" (70). When he gets fucked he hates himself \"for acceding to the _woman's_ role, when what I had been so desperate for was to prove I was a man\" (144; Monette's emphasis). Class difference is also at issue. Paul's hitherto virtuous life is turned into a more exciting path at the age of nine by his attachment to a lower-class boy, Kite. The \"turn-on\" was \"the twist of his dirty mouth, the punk veneer, the boot-camp father, like an urchin in _Oliver Twist_.\" The Lawrence who wrote of Lady Chatterley would have understood, and Forster too: \"this first fire in my loins was all about class. _Paul is perfect_ was slumming.\" Monette adds: \"Maybe Kite was a way of getting out from under the weight of gentility. I've always had a thing for men from unpaved places, not too polished, definitely not English\" (22\u201324; Monette's emphasis). Until his late twenties, Monette's attempts to do something with his unwanted gay sexuality involve class difference. \"The laughing man I was looking for was older than I and working-class, certainly no preppie\" (194). So there is age difference too, complicated by teaching in private schools, where he is seduced by the boys' flattering attentions and his own loneliness: \"I had become the thing the heteros secretly believe about everyone gay\u2014a predator, a recruiter, an indoctrinator of boys into acts of darkness. Sullying my mission as teacher and guide\" (197).\n\nBeing the seduced boy may be satisfying. Paul enjoys going around with Harold, who is older and wealthier:\n\nHe put on the carnival of events for my sake, treating me like a prince, and even as I raced about laughing on his arm I was thinking how it would be if this were a permanent thing. To be kept by Harold\u2014no more teaching meat-brain kids, no obligations except to be a poet. Wherever we went, running into Harold's friends, I'd see the flush of pride in his face as he showed me off. No, that's not right. I did all the showing off. (268)\n\nYet age difference takes Paul only so far: \"I needed the seventeen years' difference between us in order to put my trust in his sagacity and worldliness. But I also wanted a man my own age, to discover the world along with me\" (269). Paul is uncomfortable with these hierarchies of gender, class, and age. The resolution is his meeting with Roger: they remain together for seventeen years, until Roger's death.\n\nIn the continuation of Monette's autobiography, _Borrowed Time_ , the emphasis shifts to similarity, even sameness. Roger is a successful lawyer, of compatible professional stature and affluence. They are the same size, and hold shirts, underwear, and socks in common. Neither is relatively feminine; though marriage is invoked, roles are not differentiated. \"'But we're the same person,'\" Roger exclaims shortly before death, \"in a sort of bewildered delight. 'When did that happen?'\" The answer, according to a friend, is that Paul had anticipated the idea: \"'But that's what _you_ always used to say in Boston. Roger and you were just two names for the same person.'\"\n\nIn fact there are significant differences (this will be a recurring pattern). Roger is four years older\u2014thirty-two and twenty-eight when they meet. Paul is frenzied while Roger is calm; Paul is more dependent, Roger more stable and self-contained. \"I am the weather, Roger is the climate\" (65). Further: \"Over the years, relations between us had evolved to a place where he was the grown-up and I the child\" (194). Now Roger is sick and Paul cares for him \"like a mother\" (341). Notwithstanding, Paul insists that these differences are transcended. \"Between us we covered the night and the morning watch\"; \"Being as we were the same person, happily it all balanced out\"; \"How was it even physically possible to separate us now, with the two of us so interchangeably one?\" (29, 41, 315). Some people, Paul acknowledges, regard him as \"just a love junkie. What I experience as being known to the core, appetite and aspiration fused, some queers think of as confinement. Doomed to resemble a bourgeois marriage, straight-identified to boot.\" But Paul learned to love himself, \"because someone else finally loved me.\"\n\nThe more serious problem for the love junkie is living up to it all. In _Borrowed Time_ this is magnified by Roger's sickness with AIDS.\n\nI ran around the bed and clutched Roger's hand. \"We'll fight it, darling, we'll beat it, I promise. I won't let you die.\" The sentiments merged as they tumbled out. This is the liturgy of bonding. Mostly we clung together, as if time still had the decency to stop when we were entwined. After all, the whole world was right here in this room. (77)\n\nThe quality of love must ensure survival. We might notice the echo of John Donne's poem \"The Good-Morrow\": love \"makes one little room, an everywhere.\" \"Whatever dies, was not mix'd equally,\" the poem continues: Roger's life appears to depend on their equivalence and togetherness. Walt Whitman is there as well:\n\nWe two boys together clinging,\n\nOne the other never leaving.\n\nPoetry and gay tradition may help sustain them.\n\nNone of Monette's experiences is uncommon among gay men, though his awareness may be. They amount to a series of anxious negotiations of disjunctions and convergences, positioned around binary differences of gender, class, and age. Positively and negatively, as Monette presents them, these hierarchies constitute the available options; despite aspirations to transcend them in the name of equality, they structure the terms on which intense human interactions become available. Monette prefers to deny, or move beyond, hierarchy, and toward an idealized, egalitarian relationship; yet differentials are still apparent. My case is that hierarchies of gender, age, and class, and race also, are hard to expel from our personal lives because they constitute the principal hierarchies that structure our societies. Differences of masculine and feminine, old and young, upper and lower class, and white and black are not incidental or neutral alternatives. They flow through the power relations that we encounter daily in the world, and through our psyches also; we experience them, ultimately, as empowerment and abjection. Monette's negotiation of hierarchy and equality manifests a persistent strain in gay imaginings.\n\nThis proposition facilitates a materialist interpretation of gender and sexuality. Because these elements are so complex, their institutional apparatuses so contradictory, and the permutations so many and so intricate, we experience ourselves as unique individuals; probably that is what we are. Nonetheless, the hierarchies in fantasy and practice derive not from the individual psyche, but from the social relations that define our being. They are continuous with the stories that construct our psychic reality: social being determines consciousness. Egalitarian aspirations also are socially encoded; they are as exciting and difficult to sustain in personal relations as they are in the social order (I return to this in a while). But often, I will show, there is more hierarchy in the frame than is immediately admitted. Such an account of fantasy and power is cultural materialist on three counts: (1) it recognizes the priority of economic, social, and political structures in the constitution of consciousness; (2) it emphasizes the role of ideology; (3) it maintains an awareness of domination and exploitation.\n\nAs Foucault argues, power is to be envisaged as pervading the entire social order, in positive as well as negative aspects. It \"penetrates and controls everyday pleasure\u2014all this entailing effects that may be those of refusal, blockage, and invalidation, but also incitement and intensification: in short, the 'polymorphous techniques of power.'\" To say this is not to overlook the specific and massive apparatuses of government, law, business, and education in our societies. Power is, at once, both intimate and institutional. In _Discipline and Punish_ , Foucault seems to repudiate the idea that power relations are \"localized in the relations between the state and its citizens or on the frontier between classes.\" He denies that power may \"merely reproduce, at the level of individuals, bodies, gestures and behaviour, the general form of the law or government.\" This seems to position class, gender, race, and age as superficial modes, whereas power relations are more fundamental, reaching \"right down into the depths of society.\" Yet Foucault does grant \"continuity\" between the multiple modes of power\u2014not in any predictable analogy or homology, but through \"a specificity of mechanism and modality.\"\n\nIf, then, social structures may be said to inform what we experience as our individual sex\/gender formations, this is not to imagine some simple transmission of the world into the psyche. Teresa de Lauretis portrays the relations of representation, action, and fantasy as \"intimate... in the realm of the senses and in that of the law, in sexual practices as well as in the juridical-legislative domain.\" However, no easy transference should be supposed. We need to observe \"the different relations of production.\" As I remarked in the previous chapter, the erotic deployment of fantasies of power is inevitably tangled into substitutions, conflations, reversals, and loops.\n\nFantasies of dominance and subjection should be regarded as unsurprising transmutations of prevailing social relations of domination and subordination. Hierarchy is neither an aberration nor a misfortune in desire, but integral with it. Indeed, it may well be that power difference is the ground of the erotic; that it is sexy. That is the insight of Genet's _Balcony_.\n\n**THE EGALITARIAN IDEOLOGY**\n\nThe dominant metropolitan ideology suggests that the most suitable partner, gay or straight, will be of similar age, class, and race to oneself. Gender is the difference that is prized\u2014though only when it figures heterosexually. This ideology of similarity and equality informs the companionate marriage, as it has evolved from the 1920s' endorsement of reciprocity in sexual pleasure, through the 1950s' pram-pushing hubby, to the 1980s' \"New Man.\" Already in 1971, Geoffrey Gorer, reporting on a survey of attitudes toward sex and marriage, was remarking that twenty years previously the dominant model of marriage had been \"complementary,\" resting on a clear division of responsibilities. However, among younger people this was being displaced by a \"symmetrical\" model, which stresses \"comradeship, doing things together, and articulateness.\" The survey was conducted in England but, Gorer noted, key terms in the new model\u2014\"togetherness\" and \"communication\"\u2014were coming from the United States.\n\nAnthropologists and social historians have tracked this development among lesbians and gay men, looking for the emergence of egalitarian relations as a sign of progress. Romantic friendship, in the eighteenth and nineteenth centuries, seems to promise modernity and maturity, even though the equality may be more notional than actual, and the eroticism uncertain. Walt Whitman is celebrated for the dear love of comrades, though his own relationships seem to have been characterized by differences of class and age. The key to David Halperin's sense of modern homosexuality is the opportunity to transcend hierarchy. \"Homosexual relations cease to be compulsorily structured by a polarization of identities and roles (active\/passive, insertive\/receptive, masculine\/feminine, or man\/boy). Exclusive, lifelong, companionate, romantic, and mutual homosexual love becomes possible for both partners.\"\n\nA change in expectations along these lines has been confirmed lately among lesbians and gay men by Jeffrey Weeks, Brian Heaphy, and Catherine Donovan. They believe that there was once \"a prevalent stereotype about the inegalitarian nature of many homosexual sexual and emotional involvements, defined or fractured by generational, class, racial or domestic inequalities.\" But now, they find in interviews, \"The dominant ethos among lesbians, gay men, and bisexuals is of egalitarian relationships.\" Even so, \"The reality, inevitably, is more complex: non-heterosexuals strive to achieve equality in terms of intimacy, sexual relations and the division of labour in the household against all the inequalities that continue to structure our societies\" (109). Actually, around 60 percent of the respondents did not describe their relationships as \"equal\" (114). Whether they were finding any positive advantage in power differentials is hard to know, since Weeks, Heaphy, and Donovan evidently share the egalitarian ideology, and their interview questions take it for granted that valuable factors, such as \"communication, closeness, and intimacy,\" are scarcely to be found outside equal relationships (110).\n\nTo be sure, few people suppose that it is possible to have a totally egalitarian relationship. Notwithstanding, the dominant ideology says that power differentials are unfortunate and should be either avoided or overcome. Indeed, so strong is the ideology of equality that some S\/Mers are insisting that their routines are not just \"safe, sane, and consensual,\" but actually egalitarian. Lynda Hart and Joshua Dale note that in some quarters S\/M \"has become less a polarized expression of a master's power over a slave than a mutual exchange of power.\" Already in the 1970s, some practitioners began to refer to S\/M as \"sensuality and mutuality\"; by the early 1980s the \"mutualists,\" as Geoff Mains calls them, had become a prime element in the leather community.\n\nSimilarity and mutuality correlate with monogamy, respectability, and assimilation in _The Lost Language of Cranes_ by David Leavitt. The project of this novel is to sort out the good gays from the unfortunate approximations. The older generation finds it hard to benefit from recent developments in gay selfhood. Owen, who is married to Rose, is unable to talk to anyone about his yearning for gay sex; even at a pornographic cinema, which he visits regularly, he is too ashamed and frightened to speak to anyone or to follow up potential contacts. Two less prominent characters, Derek and Geoffrey, are old-style queens reminiscent of Oscar Wilde; they cultivate British accents, speak of men as \"girls,\" prepare a dinner in which all the food is blue, and include in their circle cultured Europeans who go to Tangiers where it is easy to buy young boys.\n\nAll this is regarded with a mixture of distaste and disbelief by the younger generation, represented centrally by Philip, the gay son of Owen and Rose. He had difficulty as an adolescent coming to terms with himself but, the narration suggests, he's doing it more or less right now. He postpones coming out to his parents until he believes he has achieved a gayness he can be proud of: \"I wanted to wait until I could show you that a homosexual life could be a good thing.\" This involves, above all, having a presentable partner: \"he had counted on Eliot's presence in their living room to justify all he had said to them, to justify his life\" (198). The alternative is cruising and porn movies, but Philip finds little satisfaction there; he meets partners socially, among friends at dinner parties. He has a favorite gay bar, but there is no back room. It is \"a friendly place, very social, a place where people go who really are comfortable with being gay, and know it's a lot more than a matter of who you sleep with\" (155).\n\nEliot, Philip's prized partner, proves unreliable. Probably he has been damaged by an overcasual upbringing, and spoiled by superior wealth and connections. However, the resolution, for Philip, is already to hand. Brad, an old school friend, is white, and of the same age, class, and educational attainment; they enjoy spending time together. In due course they find that sex is a natural part of that. When they first kiss, \"long and lovingly,\" it is \"spontaneous, without thought\" (311). So no sticky, sexually explicit, bathhouse, pickup scene is required; they appear to be a natural couple. They are innocent of gendered roles: there is nothing \"frilly or feminine\" about Brad (249), and nothing in Philip's appearance \"betrayed his homosexuality\" (33). Yet when he was at school\u2014although \"he hardly fit the stereotype of the sensitive, silent, 'different' boy who knows how to sew, is friends with the teacher and subject to colds\"\u2014the other boys \"routinely called him 'faggot' or 'fairy'\" (74). Plainly Philip was giving some kind of queer signal, but Leavitt cannot say what it was without admitting a demeaning hint of effeminacy. It is easy for the queer reader today to dismiss Philip and Brad as in thrall to a bourgeois, heteronormative lifestyle concept, but twenty years ago it was not easy for young people to accomplish such a thing, in the absence of role models, and even discussion.\n\nThe disqualification of hierarchy is confirmed in the stories of the other characters. Owen, prompted by Philip's coming out, does his best to catch up. He makes a suitable choice when he takes up with another married man of similar age and class. Philip's friend Jerene is African American; her adoptive parents\u2014black, middle-class Republicans\u2014reject her when she tells them she is a lesbian. She remains nonetheless nice, good, and wise. However, her new partner is perceived by Philip and Brad as rather a pain, in the manner of characters in Tennessee Williams' _The Glass Menagerie_ : \"If Laura's looks were Laura Wingfield\u2014fragile and transparent as a tiny glass animal\u2014her temperament was pure Amanda: loud and brash and indiscreet; full of hype and bombast; good-natured, loving, easy to hurt\" (251). So Laura is the dominating mother posing as the needful daughter; Jerene is subdued and silenced. These women are still involved in power games; they have not yet arrived at an adequately reciprocal partnership.\n\nSuch restagings of difference and similarity as manipulation and maturity are found in many other texts. In the film _An Early Frost_ (John Erman, 1985), Michael and Peter form a compatible couple, with just a touch of gender hierarchy: Michael is a lawyer, whereas Peter is artistic, sells collectibles, cooks, wears looser, noncorporate clothes. There are differences, then, but they appear not to signify. It is as if hierarchy is needed to make the relationship plausible, but nothing can be done with it for fear of compromising the image of the good gay. _Hollow Reed_ (Angela Pope, 1995) is similar. The issue in this film is Martyn's suspicion that his son is being assaulted by his ex-wife's partner. Meanwhile Martyn and his partner Tom constitute a \"good\" couple, sharing problems and being sympathetic and sexy for each other; they are a lot nicer than the heterosexuals in the film. There are differences: Martyn is a doctor, he wears a sport's jacket and a tie; Tom keeps a music shop, dresses in jeans, T-shirt, and denim jacket, appears younger and slighter; Martyn drives a car, Tom rides a bike. However, these do not affect the story.\n\nThe notoriously sanitized view of gay life in the film _Philadelphia_ (Jonathan Demme, 1993) includes a careful negotiation of sameness and difference. Andy and Miguel look about the same age and professional status, and have been together for more than nine years. Miguel appears comfortable with Andy's family; at ease and smart in the courtroom; at home he administers sophisticated medical treatment and is articulate; at a gay fancy dress ball (the opportunity for fantasy to burst forth), they dress the same\u2014as naval officers. The difference is that Miguel looks and sounds Spanish (and is played by the actor Antonio Banderas). However, at no point is this difference registered by anyone.\n\n_Jack_ , a novel by A. M. Homes, rewrites _Catcher in the Rye_. Holden Caulfield was harassed by nauseating perverts; Jack freaks out when he finds that the reason his father left home is that he's gay and lives with Bob. However, there is no need to worry because Bob is entirely presentable: he is a lawyer, Jack's dad is an accountant; they were acquainted socially before they got together, and they are sufficiently respectable to host a party to support a woman who is running for Congress. Meanwhile the apparently normal family of Jack's friend Max turns out to be far from ideal. My account makes it sound worthy, but this is a droll book. For a British instance see Anthony McDonald's romance, _Adam_. Adam (sixteen) falls intensely in love with a young Frenchman (twenty-two). Sylvain is a bit simple\u2014from inbred peasant stock, a child of the woodlands; he can't be introduced to middle-class family and friends. His devotion to Adam proves morbid and dangerous. After all, Adam finds that he has a more mature kind of sexual love with his long-standing school-friend\u2014same age, same class and education, same nationality.\n\nMichael Cunningham's novel _Flesh and Blood_ is one that does not take the superiority of sameness for granted. Will, at thirty-five, is \"tired of pretty boys\" from out of town. His new relationship is with the older Harry:\n\nhe'd be the beauty and Harry the one who paid cool, humorous tribute. Will loved and hated the idea. It surprised him. Here in this expensive but haphazardly furnished apartment, he was the one with the body and no cash. It wasn't where he'd expected to go.... It occurred to Will that he could be to Harry what he'd always wanted pretty men to be to him.\n\nThe sex is different: \"Ordinarily [Will] felt concealed by sex; he disappeared into the beauty of the other man. With Harry he was more visible. Sometimes he liked the sensation. Sometimes he thought he'd get up and leave\" (307). They come together initially on a friendly rather than a lustful basis, but they do meet in a gay bar and sex is central.\n\nAt the same time, Will and Harry are compatible in every other respect. No racial difference is remarked. Will teaches fifth grade (but went to Harvard); Harry is a doctor (but plays the saxophone). They are both clever; they talk all the time about everything. They both love _Anna Karenina_ and _Middlemarch_ , and have rebelled against oppressive fathers. They go to movies, eat in restaurants, drive to Provincetown. \"Will and he made no declarations; it just unfolded\" (307). Alongside Will's turning on to difference, then, Cunningham asserts a natural couple (in the manner of _The Lost Language of Cranes_ and _Borrowed Time_ ). Indeed, they become so compatible as to be interchangeable: Will \"lived as himself and he lived as the younger man who was loved by Harry and he lived, obscurely, as Harry, too\" (310). However, as in the complementarity (narcissistic) model of gay relations, these convergences are predicated on discrepancy; they secure sameness and difference at the same time.\n\n**IMAGE OF AN UNLIMITED EMBRACE**\n\nThe project of constituting gay respectability around the equal, if pressured, couple is, of course, contested. _The Lost Language of Cranes_ , _An Early Frost_ , and _Philadelphia_ are promoting one position in a cardinal, ongoing dispute. The contrary position values multiple and anonymous partners. Currently the dispute is often framed as one about \"gay marriage.\" I will argue that the two positions actually share a preference for sameness, and a persisting unease about hierarchy.\n\nPositive accounts of multiple and anonymous relations are not so easy to find as one might suppose. Notoriously, such gay classics as _Dancer from the Dance_ by Andrew Holleran and _Faggots_ by Larry Kramer represent gay subculture as promiscuous and hence necessarily frustrating and anguished. To be sure, pornography often promotes the idea of multiple and anonymous partners. However, it does so from a less prestigious sector of the gay cultural apparatus. Pornography is widely spoken of, by radicals as well as conservatives, as if it were an essential concept (often it is suggested that its images are distinctively objectified). Rather, we should ask what is the history and structure of such a categorization, and what interests it is tending to serve. It is not that this or that practice is bad and therefore pornographic, but that labeling a practice pornographic reflects a decision to regard it as bad. Pornography is not the opposite of worthwhile sexuality, but a way of asserting which sexualities are worthwhile and which are not. Because it is where we put illicit sexuality, pornography cannot confer legitimacy on its images. This outlaw status is reproduced in its irregular modes of circulation. Leavitt, on the other hand, can present the ideas that inform _The Lost Language of Cranes_ widely through authoritative media: Alfred A. Knopf, Penguin Books, and the BBC (who filmed the book for television).\n\nIn _The Farewell Symphony_ , Edmund White makes the important point that the pursuit of multiple and anonymous partners was not a new factor in the 1970s. Cruising was a gay tradition\u2014there was no break with the past. It was the same in England: bear in mind the extensive routines of men such as Tom Driberg, Michael Davidson, and Joe Orton. Nonetheless, White posits a distinctive post-Stonewall ethos: \"We saw gay men as a vanguard that society would inevitably follow. I thought that the couple would disappear and be replaced by new, polyvalent molecules of affection or Whitmanesque adhesiveness.\" \"Guys just sort of fell in with each other, buddies rubbing shoulders. We wanted sexual friends, loving comrades, multiple husbands in a whole polyandry of desire.\"\n\nSamuel R. Delany in _The Motion of Light in Water_ recalls the bar, tearoom (public toilet), and truck scenes of the 1950s, but a post-Stonewall orgy at the baths was qualitatively different:\n\nwhat _this_ experience said was that there was a population\u2014not of individual homosexuals, some of whom now and then encountered, or that those encounters could be human and fulfilling in their way\u2014not of hundreds, not of thousands, but rather of millions of gay men, and that history had, actively and already, created for us whole galleries of institutions, good and bad, to accommodate our sex.\n\nCasual sex might be a vehicle for noble, egalitarian aspirations. David Wojnarowicz finds peace, companionship, and empowerment in a one-off encounter: \"In loving him, I saw men encouraging each other to lay down their arms.... In loving him, I saw great houses being erected that would soon slide into the waiting and stirring seas. I saw him freeing me from the silences of the interior life.\" As Ben Gove observes, Wojnarowicz, like Genet, repositions romance by levering it away from its customary link with monogamy.\n\nTwo factors tend to complicate such visions. One is that hierarchy is not so easily expelled. Dennis Altman declares: \"The willingness to have sex immediately, promiscuously, with people about whom one knows nothing and from whom one demands only physical contact, can be seen as a sort of Whitmanesque democracy, a desire to know and trust other men in a type of brotherhood.\" However, as Leo Bersani points out, Altman admits that \"age and physical beauty set up their own hierarchies and barriers.\"\n\nHolleran insists on democracy in _Dancer from the Dance_. He sees in the disco \"a strange democracy whose only ticket of admission was physical beauty\u2014and not even that sometimes. All else was strictly classless.\" On the dance floor, he adds, even beauty might not matter: \"all of them mixed together on that square of blond wood and danced, without looking at anyone else, for one another.\" \"What a carnival of people.\" The abundance of anonymous contacts overwhelms the particularity of this or that partner. There may be plenty of difference, but it doesn't make any difference. Interestingly, effeminate boys are not excluded from Holleran's scene. For White too, \"Whitmanesque adhesiveness\" does not preclude gendered roles. He writes of polyandry (having more than one husband), and of a partner who referred to himself as a \"hubby,\" and of how he felt like a girl alongside him. These accounts are set in the mid-1970s. The development of the clone image (short hair, moustache, denim or leather) tended to produce a repudiation or an effacement of gendered roles.\n\nWhat is striking is that, insofar as they tend to erase hierarchies which reinsert themselves, these legitimations of casual sex converge, strangely, on _The Lost Language of Cranes_. Defenders of casual sex are evoking the very bars and cruising grounds that Philip eschewed, but they share nonetheless a suspicion of hierarchy. What have appeared to be the two poles of gay experience are united in their anxious and inconclusive treatment of power in relationships.\n\nNeil Bartlett presents the late 1970s moment of the clone as euphorically inclusive: \"It was a style that explicitly proposed a single culture. It offered to embrace everybody, to erase all differences in a generous, homogeneous, successful style. Commercially promoted on a mass scale, it seemed to absorb all the other, older styles.\" Not for nothing, Bartlett adds, was the biggest London discotheque opened under the name of \"Heaven.\" Even Kramer, despite his commitment to personal values in relationships and the generally caustic stance of _Faggots_ , finally concedes and celebrates, for once, the indifference of the scene on Fire Island:\n\nThe dancing's over for this night. Haven't we shared a night of nights! A night of fellowship. We have danced and partied and drugged and Meat Racked, and we have survived no sleep. Together. Together. Yes, we have braved and passaged all these rites together. Though we may not know each other's names nor will we necessarily speak when next we meet. The beach is filled with my friends.\n\nSharing, fellowship, togetherness, friends\u2014these terms are not far from Leavitt's, though they are adapted to Kramer's moment of self-abandonment, rather than a personal relationship.\n\nOf course, the status of casual sex has been transformed by HIV and AIDS. Oscar Moore's _A Matter of Life and Sex_ is one of many texts which accept a correlation between the delights of multiple and anonymous partners and the fate, as it appears in the novel, of infection and death. William M. Hoffman in his play and film _As Is_ makes Saul and Rich agree on the pleasures of \"promiscuous sex\"; they redefine it as \"nondirective, noncommitted, non-authoritarian\u2014\/ Free, wild, rampant\u2014.\" But now sex even with your lover is too dangerous to attempt. The most impressive antimonogamy texts attempt principled reassertions of multiple and anonymous partners in the face of AIDS. Thom Gunn in his poem \"The Missing,\" in _The Man with Night Sweats_ , reasserts a vision of unfettered sexual congress:\n\nContact of friend led to another friend,\n\nSupple entwinement through the living mass\n\nWhich for all that I knew might have no end,\n\nImage of an unlimited embrace.\n\nAgain, difference evaporates into immeasurable multiplicity.\n\nBersani is the theorist of anonymous sexual relations (antirelational relations). He is sharply critical of commentators who domesticate sex by reinventing it as _redemptive_ \u2014as \"less disturbing, less socially abrasive, less violent, more respectful of 'personhood' than it has been in a male-dominated, phallocentric culture.\" The broad indictment of male ascendancy, and of pornography in particular, by feminists such as Andrea Dworkin and Catherine A. MacKinnon, has in Bersani's view \"had the immensely desirable effect of publicizing, of lucidly laying out for us, the inestimable value of sex as\u2014at least in certain of its ineradicable aspects\u2014anticommunal, antiegalitarian, antinurturing, antiloving.\" Their expos\u00e9 of the untamable nature of sex affords a reason not for banning pornography, but for encouraging it.\n\nYet, as Bersani pushes these ideas along, a familiar theme emerges. In an essay on Genet (which I discuss elsewhere), \"an anti-monumental, antiredemptive aesthetic\" is discovered; it is endemic in homosexuality, Bersani believes. Within its orbit \"an _identity_ between the penetrator and the penetrated\" occurs; \"a fundamental sameness.\" Bersani has driven his repudiation of authentic personhood in sex to the point where it makes everyone the same\u2014identical in their anonymity and unrelatedness. Lately Bersani has theorized further his argument for the irrelevance of difference, through a reinterpretation of Aristophanes' fable of the divided creatures in Plato's _Symposium_. The missing portion, Bersani urges, is not the other, but a part of oneself. This is claimed for heterosexuals equally. What the lover lacks is not _the other_ , as followers of Jacques Lacan in particular have supposed, but \" _more of what he is_.\" This lack is based not on difference, therefore, \"but rather on the _extensibility of sameness_.\"\n\nAcross a wide spectrum of texts, and in diverse and ingenious ways, sameness is valorized and hierarchy is disallowed. Yet even, or especially, when the repudiation of hierarchy is associated with the highest aspirations, it is not easily banished. In later chapters I will explore the erotics of power in specific lived experience of gender, age, class, and race.\n\n**SEDUCTION AND IMPLANTATION**\n\nAn argument about sexuality and power has to engage eventually with the psychoanalytic tradition. What has often troubled cultural materialists and other social constructionists is Freud's intermittent reliance on what he presents as universal factors\u2014biological, phylogenetic (in the evolution of a race), primordial. In the wake of Darwin, such concepts tend to incorporate a murky, conservative view of human potential, in which sex is envisaged as designed for reproduction and gender roles are confined accordingly. For example, in Freud's \"Wolfman\" analysis we are told of a breakthrough at the point where the boy \"discovered the vagina and the biological significance of masculine and feminine. He understood now that active was the same as masculine, while passive was the same as feminine.\"\n\nAs I observe below, my argument that power difference informs sexuality consorts easily with sadomasochistic fantasies. But Freud, at least from time to time, prizes reproduction over perversion. He writes in _Three Essays on the Theory of Sexuality_ : \"The normal sexual aim is regarded as being the union of the genitals in the act known as copulation.\" Other practices, including fetishism, sadism, and masochism, are \"the perversions\"\u2014that is, \"sexual activities which either (a) _extend_ , in an anatomical sense, beyond the regions of the body that are designed for sexual union, or (b) _linger_ over the intermediate relations to the sexual object which should normally be traversed rapidly on the path towards the final sexual aim.\" The \"biological significance\" of an element of male aggressiveness may \"lie in the need for overcoming the resistance of the sexual object by means other than the process of wooing\"; it becomes sadistic\u2014perverse\u2014when it is \"independent and exaggerated\" (71). However, this approach will not easily account for male masochism. Perhaps it is an instance of a principle: \"Every active perversion is thus accompanied by its passive counterpart\" (81). So \"masochism is not the manifestation of a primary instinct, but originates from sadism which has been turned round upon the self.\"\n\nIn \"'A Child Is Being Beaten,'\" Freud holds that male masochists occupy a feminine position: \"they invariably transfer themselves into the part of a woman; that is to say, their masochistic attitude coincides with a _feminine_ one.\" This statement perhaps gestures toward the binary gendering that is associated with reproduction; Freud insists: \"It makes no difference if in a fanciful embellishment of the masochistic scene they keep up the fiction that a mischievous boy, or page, or apprentice is going to be punished.\" But then, \"confusingly enough,\" the chastiser also is a woman; Freud moves to another topic. At other points Freud connects masochism to \"the death drive,\" and \"a fixation in childhood.\" The last suggestion seems most plausible. To be sure, Freud offers other frameworks for thinking about masochism at other points in his writing. I have focused on his subsuming of power into reproduction because it occurs in major texts (the _Three Essays_ were reprinted twenty-one times in ten languages between 1910 and 1938), and because it works against my sense that power differentials are intrinsic to sundry sexual identities and interactions, and not dependent on heteronormative interpretation. The priority of reproduction is still asserted by reputable thinkers. Luce Irigaray, for example, declares: \"The human species is divided into _two genders_ which ensure its production and reproduction. To wish to get rid of sexual difference is to call for a genocide more radical than any form of destruction there has ever been in History.\" It is a mistake, therefore, \"to demand equality as women.\"\n\nMainly since the publication of Jean Laplanche's _New Foundations for Psychoanalysis_ , via a battery of conferences, translations, introductions, interpretations, and collections of essays, instigated and undertaken principally by John Fletcher, new theories have effected decisive revisions of Freud, and Lacan also, offering a new route beyond biologism. Freud, along with Darwin, Marx, and Nietzsche, has often been credited with a _decentering_ of man. Copernicus demonstrated that \"man\" is not at the center of the universe; these later thinkers have argued that he is not in charge of his own biology, history, and psychology. Laplanche declares this Freudian-Copernican revolution unfinished, and our dependence upon the other unacknowledged, so long as the infant is said to bring with him into the world fundamental drives:\n\nFor if the individual is henceforth governed, in classical psychoanalytic theory, by the unknown drives of the unconscious, this \"id\"\u2014however strange it is supposed to be\u2014is nonetheless not an alien. It is supposed to dwell _at the center_ of the individual, whom it governs in its own way, even if it has dethroned the ego. One sovereign in place of another, but well and truly installed in the keep of the castle.\n\nTo address this problem, Laplanche reviews the \"seduction theory,\" which Freud abandoned in 1897; this had understood neurotic symptoms as consequent upon the seduction or abuse which his patients reported from their childhood. Upon this abandonment, Freud posited the phases (oral\u2013anal\u2013phallic) of endogenous (growing from within) infantile sexuality, with its apparently biological sequencing.\n\nLaplanche believes that passivity and seduction do constitute the individual's originary moment. The mother (or other primary caregiver) _does seduce the child into erotic pleasure_ \u2014not at all, however, in any sinister way, but as part of the routine parental ministrations without which the child would die. The feeding and handling of the infant exposes him or her to the caregiver's fantasy life, inviting cooperation, through strategies of translation and repression. Fletcher explicates:\n\nWe are not talking here of abusive events. In Laplanche's sense seduction is ordinary. This leads him to talk of an _implantation_ of stimulating, arousing and traumatizing non-verbal signifiers with their unconscious, enigmatic significations: an implantation on the surface of the primitive body image or skin-ego of the infant. These are anchored or inscribed particularly in the erogenous zones as folds and openings in the body surface\u2014mouth, anus, genitals.\n\nThus is effected \"the primary mapping and zoning of the sexual body, indeed the very sexing of the body.\" Martin Stanton asks: \"Is there no input from the infant at all?\" Laplanche replies that feeding\u2014the initial demand\u2014is interactive, but the sexual message that accompanies it is \"a one-way action.\"\n\nFrom the beginning, one is active and the other is passive. But very quickly, the little human tries to turn this passivity into activity, that is, to make something of this message from the other. Still, there is this dissymmetry. This comes from the fact that the active one has more \"knowledge,\" more unconscious fantasies than the passive infant.\n\nThis way of explaining the development of individual psychic formations is notably compatible with my arguments about sexuality and power, in two main ways. First, the organization of fantasy in which the infant is involved is that of the mother (or other caregiver), and consequently already steeped in her tangle of scenarios. She, necessarily, has already her own personal take on the prevailing system of representation, fantasy, and unconscious desire, and this intrudes, also necessarily, upon the infant. Second, the infant's initial experience is of power imbalance. \"The primary situation that gives rise to the sexual drive in the human being,\" Fletcher writes, \"is one of a primary passivity and penetration by the other. It involves a breaking in that is characteristic of pain.\" For what is undeniable in Laplanche's \"situation of primal seduction\" is both\n\nthe wealth of its innate mechanisms of reciprocal communication between mother and child, and the profound _asymmetry_ between the adult with an already formed unconscious, the bearer of unconsciously determined enigmatic signifiers or messages, and the new-born infant assigned a gender on the basis of adult perceptions of its anatomy but [as yet] without sexual fantasies.\n\nIf, as I believe, Laplanche has evolved an important theory through which the power arrangements in our societies may be implanted in the infant, becoming the building blocks in the _bricolage_ of our psychic life, this should not be envisaged as an elementary mechanism of social conditioning, producing automata incapable of independent agency. The infant is subjected to extremely powerful inputs, but they are complex\u2014and it is the simplicity of a message that stifles agency, not its strength. The transmissions in Laplanche's theory are fraught with untranslated, resistant, and troubling remainders. Fletcher takes up Dominique Scarfone's argument that it is negation, repression, and enigma on the part of the adult and her messages that provoke the infant \"to translate, to reprise and rework the enigmatic and exciting messages, to substitute its own signifying sequences, fantasies, 'infantile sexual theories,' to interpret the blanks in the parental discourse, to sublimate by symbolising otherwise.\"\n\nFurther, the encounter may go wrong, implanting materials that resist adequate symbolization. There is a \"violent variant\" of implantation, according to Laplanche: _intromission_. This is a blocking process which \"short-circuits the differentiation of the agencies in the process of their formation, and puts into the interior an element resistant to all metabolism,\" creating the conditions for both the superego and psychosis. Perhaps it is the gaps and enigmas in the parental messages that create the space for the child to interpret, translate, and fantasize, whereas when parental fantasies are imposed, full on, translation is paralyzed\u2014which may be why sexual abuse in childhood is so destructive, impairing, and repetitive, from generation to generation.\n\nOf course, I am not equipped to evaluate theories of infant subjectivity. In any event, Laplanche's insistence on the primacy of the caregiver issues a reminder of the infant's experience of initial power disparity, while allowing space in which human beings might develop diverse and novel ways of relating.\n\n**ARGUMENTS AND VISIONS**\n\nIf hierarchies of sex and gender are both embedded in our psyches and a sexual turn-on, attempts to thwart them must be both futile and austerely abstemious. To be sure, power may be distributed irregularly between two people, such that A is powerful in respect of p, whereas B is powerful in respect of q. However, this does not mean that hierarchy evens out and becomes irrelevant. More likely, quite intricate and engrossing negotiations will be required to maintain the diverse claims; there will be more engagement with hierarchy, not less.\n\nThis is not to say that oppressions of gender, class, age, and race are either necessary or justified. We may campaign for radical transformations. Nor is it to say that it is acceptable to knock your partner around, or to manipulate him or her psychologically, or to use his or her relative poverty or general neediness to exert control. Nor is it to say that you should stay with a partner who abuses you in those ways. I write, I should say, as a relatively empowered person. I am white, male, salaried, and have published some books; I have some advantages of middle age (stability, security) and some of its disadvantages (likely to fall ill and die before long). It is perhaps easy for me to insist that power difference is sexy; my potential partner, if he is located on the more vulnerable side of those hierarchies, may be more exposed, psychologically, in his engagements.\n\nThe terms on which women might collude in hierarchies which derive from male-dominated ideologies have been effectively disputed in lesbian feminism. Sheila Jeffreys sets out from somewhere near the start of the present analysis: \"Since the concept of difference or 'male-female polarity' is the organising principle of the heteropatriarchy it is not surprising that it should so profoundly have shaped the consciousness even of many lesbians.\" However, Jeffreys contests any idea that lesbians should collaborate with it. She disagrees with women such as Amber Hollibaugh and Cherr\u00ede Moraga, who protest that feminist rejection of male dominance in heterosexuality has led to an ill-considered rejection of hierarchical relations between women. Hollibaugh opines: \"what feminism did, in its fear of heterosexual control of fantasy, was to say that there was almost no fantasy safe to have, where you weren't going to have to give up power or take it. There's no sexual fantasy I can think of that doesn't include some aspect of that.\" This is to say, Jeffreys replies,\n\nthat we cannot build a sexuality which is about equality and mutuality.... The refusal to see any kind of sex without dominance and submission as possible, rules out the feminist adventure in the total transformation of sexuality with the object of eliminating sexual violence and the objectification of women, almost before it's begun.\n\nMoraga responds with her personal experience. She grew up with \"the fantasy of capture, taking a woman\"; her \"identification was with the man, taking\": that is how her sexuality works.\n\nIn _Loving in the War Years_ , Moraga relates these arguments to class and race:\n\nWhat the white women's movement tried to convince me of is that lesbian sexuality was _naturally_ different than heterosexual sexuality. That the desire to penetrate and be penetrated, to fill and be filled, would vanish. That retaining such desires was \"reactionary,\" not \"politically correct,\" \"male identified.\" And somehow reaching sexual ecstasy with a woman lover would never involve any kind of power struggle.\n\nJudith Halberstam comments: \"Chicana lesbians cannot suddenly be expected to cast off these sex roles in favour of a lesbian feminist egalitarianism.\" Esther Newton and Shirley Walton are similarly skeptical. \"Do away with masculinity and femininity and the residuum is egalitarian sexuality: open, honest, caring, and non-oppressive\": that is the theory. However, \"Power and sexual desire are deeply, perhaps intrinsically connected in ways we do not fully understand and just can't abolish,\" Newton and Walton contend. \"It is true that men have more power than women in the sexual domain. But one cannot proceed directly from this fact to explain how sexuality works, any more than male domination of the art world, for example, explains aesthetic experience.\"\n\nIt will be evident where the present study stands in relation to these arguments. Our sexual imaginaries probably are informed by hierarchies that are ultimately oppressive, but we have to negotiate within, through, and beyond that insight. It cannot be realistic to suppose that we might simply, through good intentions, sidestep the hierarchies of capitalism and male dominance. They inform our daily interactions, the language through which we come to consciousness, our psychic formations. Islands of individual serenity are a strategic aspiration for therapy, but finally we must be talking about damage limitation. Socialism in a single psyche must be a chimera.\n\nWhat, then, of aspirations toward sex and gender liberation, and what prospects for the mutuality and harmony that have often been attached to sexual love in our cultures? Egalitarian impulses are not to be regarded as false, deluded, or partial. They too are produced within the system; where else could they come from? Often they are couched in terms afforded by the dominant, but this does not invalidate them. If the opportunities for containment of liberatory aspirations are large, so is the potential for idealism. For the cultural materialist, as well as critique, there is always the prospect of transformation. All the books discussed in this chapter witness to the vitality of reciprocity as an idea, even where they are marking the impediments that strew its path. Perhaps the most extreme instance of an unequal love relationship is between the commandant of a Nazi work camp and a boy inmate. Ursula Zilinsky makes this plausible in her novel _Middle Ground_. The commandant declares, even in this context, \"I don't believe love is possible except between equals.\"\n\nCollaborative impulses certainly do figure in our sexual imaginaries. They have been important in heroic and romantic friendship; in some forms of romantic love; in partnerships founded in complementarity; in the element of interchangeability in some S\/M relations. They have inspired our love poems. I have quoted Monette quoting Donne and Whitman. Philip Sidney's _Arcadia_ (1580) features cross-dressing and same-sex passion. The singer of this song is a young man who says he overheard it sung by a young maid, so the effect is of a male addressing a male:\n\nMy true love hath my heart, and I have his,\n\nBy just exchange, one for the other given.\n\nYet this sonnet continues in a surprisingly violent manner:\n\nHis heart his wound received from my sight:\n\nMy heart was wounded, with his wounded heart,\n\nFor as from me on him his hurt did light,\n\nSo still methought in me his hurt did smart:\n\nBoth equal hurt, in this change sought our bliss:\n\nMy true love hath my heart, and I have his.\n\nThe poem circles back to its initial declaration, reinstating reciprocity, but there's a lot of wounding and hurting in between, suggesting not just the convention of Cupid's arrows but also a social order where violent death might be the consequence of inappropriate loving.\n\nAlfred Tennyson in _In Memoriam_ (1850) sometimes celebrates the equality and reciprocity of the deceased Arthur and himself:\n\nI know that this was Life,\u2014the track\n\nWhereon with equal feet we fared;\n\nAnd then, as now, the day prepared\n\nThe daily burden for the back.\n\nNor could I weary, heart or limb,\n\nWhen mighty Love would cleave in twain\n\nThe lading of a single pain,\n\nAnd part it, giving half to him.\n\nHowever, the poet more often dwells upon Arthur's superiority and inaccessibility:\n\nI vex my heart with fancies dim:\n\nHe still outstript me in the race;\n\nIt was but unity of place\n\nThat made me dream I ranked with him.\n\nNow Alfred is left behind again, as the confident, privileged, and accomplished Arthur ascends into a more spacious life among the souls of the departed.\n\nW. H. Auden allows that sexual passion may prompt an ideal vision in \"Lay your sleeping head, my love\" (1937). To lovers, \"Soul and body have no bounds\";\n\nGrave the vision Venus sends\n\nOf supernatural sympathy,\n\nUniversal love and hope.\n\nYet for the boy with him he wishes that the \"mortal world\" will be enough, with all its mundane unevenness. In another poem, \"The More Loving One\" (1955), Auden remarks that, while the language of love celebrates mutuality, it is unusual for two people's loves to match precisely:\n\nIf equal affection cannot be,\n\nLet the more loving one be me. (282)\n\nEvocations of ideal harmony may not be entirely unalloyed, then; they may smack of wishful thinking, tactics, and ideology; they may not sit easily with actual power differentials. Yet they do not have to be unalloyed to be important.\n\nAlthough we tend to think of lesbians and gay men as subject to distinctively complex psychic, ethical, and political demands, the idea of egalitarian sex is far more problematic for heterosexuals. While same-sex partners may choose to engage with hierarchical imagery, the copresence of a man and a woman has to start from it (though other hierarchies of class, age, and race may undermine or counteract it, and perhaps put the woman in the dominant position). The most sustained quest for a viable feminist heterosexuality, from within current progressive thought, has been mounted by Lynne Segal, particularly in _Straight Sex_. She draws from Naomi Segal the five elements of pleasure that may be said to characterize women's sexual desire:\n\npurposeless playfulness;\n\na recovery of childhood feelings (or whatever consciousness can tolerate of their original polymorphous perversity);\n\na connection with nurturance;\n\ngames with power (especially the pleasure of feeling power over the powerful);\n\na narcissistic sense of completion through access to the body of another.\n\nWhile appreciating the elements of sharing, reassurance, and play in this sequence, we may notice also that it seems designed to accommodate and subdue hierarchy. The play has to be purposeless; only so much perversity must be recovered as consciousness can tolerate; the narcissism must not be merely self-regarding. \"The pleasure of feeling power\" is allowed only if you are generally the powerless one, and only by way of \"games.\"\n\nLynne Segal's goal is an egalitarian containment which will both license and control the pleasures and dangers of hierarchy. It is a fine balance, though. For once you venture beyond mutual reassurance and a bit of playful slap-and-tickle, you get back into dominance and subordination. And unfortunately, Segal admits, men seem less amenable to routing their sense of male power through play. In fact, when Segal wants to assert the scope and vigor of fantasy against the puritanical antipornography stance of Dworkin and MacKinnon, she appeals not to heterosexual theorists but to the raunchy stance of Hollibaugh's and Pat Califia's lesbian feminist S\/M genre (251). It is not easy to get companionate marriage and vigorous sexuality into the same frame.\n\nAn argument about dominance and subordination must finally confront the particular anxieties that cluster around the formalized role-play which we call \"S\/M,\" particularly when it entails the cultivation of scenarios from fascism and slavery. In my view it is a mistake to infer that power play necessarily involves distinctive gear, dedicated bars, and playrooms, and that in other contexts, conversely, the rest of us are egalitarian. Rather, S\/M is continuous with other, less specific, hierarchical formations, such as I have been discussing. Consider John Rechy's attempt at quarantining S\/M in _The Sexual Outlaw_ : \"I do cultivate a certain tough appearance because it attracts people sexually, and I do equate sex with power. But I know the difference between that and the most negative aspect within the gay world\u2014S & M.... Pain and humiliation have nothing to do with love.\" Rechy's distinction sounds less than secure.\n\nBersani discusses Foucault's attempt in an interview to distinguish the master-slave relation in S\/M from actual power structures. S\/M, Foucault said, is not \"a reproduction, within the erotic relationship, of the structure of power. It is an acting out of power structures by a strategic game that is able to give sexual pleasure or bodily pleasure.\" To be sure, there can be no simple \"reproduction\"; the S\/M representation will indeed be \"a strategic game.\" It will be oblique, displaced, allusive\u2014even, perhaps, parodic. However, Bersani asks, \"what is the game without the power structure that constitutes its strategies?\" Would erotic pleasure survive the evaporation of its ultimate allusion to a real-world model? Does the attraction of leather, say, stem from its texture and derivation from cattle, or would it not become less interesting if policemen and men with heavy manual jobs stopped wearing it? For a time, its residual associations would persist, but eventually it would become quaint.\n\nInsofar as he regards sexuality as necessarily, in the circumstances in which we currently live, about power, Bersani's argument is congruent with that offered in the present study. His refusal to accept a consolatory story about the innocence of S\/M also seems right. However, he derives this from Freudian principles and actively disputes Jeffrey Weeks's materialist suggestion that \"the erotic acts as a crossover point for a number of tensions whose origins are elsewhere: of class, gender and racial location, of intergenerational conflict, moral acceptability and medical definition.\"\n\nMy goal here is not to justify or dismiss S\/M, but to observe its continuities with fantasies of dominance and subordination that are inevitable in our kinds of societies, and also with the everyday routines of sexual and interpersonal power, as they present themselves through the prevailing structures. The task, then, is to find ways of making hierarchy, in our sexual and personal relations, productive of pleasure and the other rewards of intimacy, and productive also of insights into the psychic economies through which we handle the triumphs and humiliations that the system bestows\u2014while maintaining also the credibility and integrity of our political engagements. This is the project proposed by Claudia Card, who observes the influence of\n\nroles of dominance and subordinance that characterize not only authoritarian adult-child relationships within the family or authoritarian religious relationships but, more generally, the norms of a patriarchal, misogynist society that is also riddled with homophobia, racism, anti-Semitism, and other forms of oppression. On this understanding, sadomasochistic desires have roots not simply in individual psychologies but in society at large; they are not mysterious givens but social constructions.\n\nCard's answer is to approach the issue on different levels. Interpersonally, we have to work with the idea of informed consent, despite its awkwardnesses. Politically, we have to oppose oppressive societal structures.\n\n**INSURRECTION**\n\nJames Robert Baker's novel, _Tim and Pete_ , explores the overlap between interpersonal hierarchies and those which sustain the state. The eponymous heroes believe in love and monogamy, at least in the time of AIDS. Spontaneous sex can be rewarding, Tim allows, but not the 1970s bathhouses:\n\n\"it wasn't joyful. It seemed to bring out the worst in people. It wasn't sex as catharsis or redemption, but sex as a drug. A crazed, compulsive abuse of what sex can provide. Cynical and loveless. Sex as ego gratification, as ephemeral validation. Sex as a product, something you need to feel better about yourself for a while. A new shirt, a new car, a new fuck. Capitalism. I didn't like the music either.\"\n\n\"Sleaze\" is their term for cruising and occasional partners; however, they accuse each other of it continually, and it transpires that at least one of them has fucked at some time with almost everyone they meet.\n\nTheir quest for a renewal of their love takes them across Los Angeles, desolate and threatening after the riots of 1991, looking as if it were a sequence of film sets. \"In fact, it was so deserted it almost felt like a set. The battered 1920s storefronts, the deep dusk blue of the sky, even the burned-out ruins, had a weird _designed_ look, like a soundstage street for a Janet Jackson video. Or a 'gritty' background for a black-and-white Guess? ad\" (92; Baker's emphasis). They view their situation through a filter of film and fantasy, as they reprise recent gay history. They love to improvise pornographic scenarios out of famous Hollywood films and clean-cut surf movies. However, there is a persistent political edge to their invention. \"So we'd imagine, in the most extreme and lurid cinematic terms, the obliteration by gunfire of different right-wing people we disliked.... we could machine-gun George Will, for example, in Washington, and suck each other off in Mexico in the same sentence\" (179; my elision). Pete heads up a post-punk rock band, and does a song called \"What This Country Needs (Is a Baader-Meinhof Gang).\" Tim has reservations: \"I'd seen his point, that anything in art was permissible, that to depict something was not the same as to advocate it, let alone do it, and I felt that our fantasies, besides being fun, were a kind of harmless way of blowing off steam\" (179). But suppose some people experience such a song as a rallying cry?\n\nPete's young friend Joey is involved with Glenn, who is forty-two; they have what Pete considers \"'an extremely fucked-up dynamic,'\" a \"'possessive, obsessive dance-of-death thing'\"\u2014not formally S\/M, but emotionally; deriving from Joey's childhood abuse. Tim calls it \"'a dad-and-lad scene'\" (166). \"'I love Joey. We're both extremely fucked up and bad for each other, but I love him,'\" Glenn says (226). It transpires that Glenn is leading a group plotting to assassinate ex-president Reagan. Initially, Glenn's interest in political terror was \"on a kind of camp, ironic level\" (199); this is familiar enough to poststructural critics, who have imagined that a purposeful gender-bending might seriously discomfort the system. Already in their paintings Glenn and Joey have drawn upon Tim and Pete's inventions, trying to break out of the customary boundaries of art; these paintings \"were very much like some of the fantasies we used to spin\" (179). Glenn tells Pete: \"'You've shaken us out of our catatonic grief with your inflammatory call to violence'\" (221). Fantasies, and their expression, cannot be corralled into a safe world of art.\n\nIt may gradually dawn on the reader that there is a continuity rather than a contrast between Tim and Pete, as a couple, and Glenn and Joey. As I note in chapter 2, Tim and Pete present their relationship to themselves as free from hierarchies of class, age, and gender, but in practice these power differences afford them a guilty pleasure. A lesbian who gives them a lift advises that they might enjoy experimenting with dominance and submission (however, her vehemence destabilizes her own relationship). When Pete gives in and agrees to get back together with Tim, he signals this by singing \"I Wanna Be Your Dog\" (135\u201337). Having captured Tim and Pete, Glenn readily intuits the link between their current situation in a real political drama and the fantasies they have cultivated informally: \"'I'll bet my entire Colt pornography collection _you've_ fantasized being stripped and restrained by a gang of rowdy outlaws'\" (217). In fact Tim has admitted to this fantasy: \"'I thought about tying you up once. Like a western thing. But serial killers always tie up their victims. What if I snapped?'\" (93). Power is everywhere in the system\u2014in the nice guys as well as in the assassins and the president.\n\nThe most developed theory of countercultural art and the dramatic insurrectionary gesture derives from Guy Debord, Raoul Vaneigem, and the Situationist International group in the 1960s. It was influential in the May 1968 _\u00e9v\u00e9nementes_ in Paris, and the bombings of financial and governmental targets by the Angry Brigade in London in 1969\u20131971. The initial idea was to create the _situation_ that would disrupt the rhythms of everyday life and the spectacle constructed by the state; it was insurrection at the level of representation. Because the state depends upon the spectacle of the commodified image, the revolutionary event may operate by reorienting those images. These ideas informed the underground press and counterculture of the 1960s and 1970s, and the work of prominent British dramatists such as Howard Brenton and David Edgar. For if power was theatrical, theater might be powerful. Leading artistic and musical figures in the punk movement professed familiarity with situationist thought.\n\nThis is what Tim and Pete have been doing in their ad hoc appropriations of movies, Pete also in his music, and Glenn and Joey in their artwork. (It is interesting that people around the Angry Brigade, some of whom were Gay Liberation Front activists, called themselves \"The Wild Bunch\" and \"Butch Cassidy and the Sundance Kid.\") ACT UP and Queer Nation are in this line of political thought but, according to Pete, they \"don't go nearly far enough.... They should go into the Vatican, remove the fag art, and dynamite the place. Douse the pope with gasoline, set him on fire\" (143; my elision). Glenn and Joey have made their pictures as transgressive as they can. The next step, as for the Situationists, is a counter-theater of momentous violence, intended not (as in traditional revolutions) to seize the apparatus of the state, but to undermine its spectacular facade by the radiant symbolism of the intervention. The most plausible instance of a positive effect from political violence in a modern state is perhaps the Black Panthers, who changed the self-awareness of African American people. \"'I miss the Black Panthers,'\" Pete says. \"'At least they knew who the real enemy was.'\" \"'Which is why they got wiped out,'\" Tim replies (80).\n\n_Tim and Pete_ ends before the culminating act of violence which Glenn and Joey intend. Attention is switched to the fact that Pete's mother is there and would be killed. Of course, once she is factored in, the case for political violence is hard to sustain: nearly half the world is potentially somebody's mother. Finally Tim envisions the slaughter at the convention, translating it immediately into cinematic terms ( _The Godfather_ ), and then imagining that he's \"looking at a painting of Pete and me, as we were right now, the way Joey would have painted it\" (256). You can't quarantine fantasy from political action. The question, still, is whether they can work productively together.\n\nActually I think _Tim and Pete_ is about something else as well. One senses a less focused, more gestural sense of apocalyptic violence\u2014for instance, in the references to Charles Manson. A strain in U.S. culture\u2014both on the right and the left\u2014has always been ready for apocalypse now; the conjunction of AIDS and the end of the millennium made it unavoidable.\n\nWhat is at work is less a political analysis than a community that is anguished and desperate. _Tim and Pete_ is strewn with harrowing stories of official and personal disdain for people with AIDS. However, this is not just about the loss of lovers and friends and the threat of sickness and death: it is about social exclusion. There is one strategy running through Tim and Pete's numerous, impertinent appropriations of movie scenarios: at every point gay men are being inserted into recognized images of \"America.\" (Pedro Almod\u00f3var, conversely, is mentioned simply with respect; only American scenarios are appropriated.) Already these gay men have been rejected by their families. Now they are mourning \"America.\" The president, once elected, is supposed to stand for all citizens. But George Bush the elder is quoted as admitting in 1987 that there was still a \"giggle factor\" in the government's approach to AIDS (9). Gay Americans find that their condition has excluded them from the nation. Centrally \"American\" images can be occupied by them only in a parodic or violent way.\n\nThey are, in one of David Wojnarowicz's titles, in the shadow of the American Dream. A leitmotiv in Wojnarowicz's _Close to the Knives_ is that if only the president would pay attention to AIDS, we would begin to get somewhere with it. I find here a strange vein of surprise that gay rights have not been acknowledged; Wojnarowicz is shocked to read of a Supreme Court ruling that \"only people who are heterosexual or married or who have families can expect these constitutional rights.\" Lamenting the suicide of a friend, he asks: \"Man, why did you do it? Why didn't you wait for the possibilities to reveal themselves in this shit country, on this planet?\" (241). \"America,\" surely, will come good in the end. I have remarked elsewhere this wish to believe in America, noting its role in the thought of Randy Shilts, Larry Kramer, Bruce Bawer, and Andrew Holleran.\n\nAndrew Sullivan is so keen to believe in \"America\" that he claims that it has actually discovered its true humanity by taking gay men to its bosom: \"AIDS compelled a form of social integration that might have never taken place without its onslaught. Forced to choose between complete abandonment of the homosexual subculture and an awkward first encounter, America, for the most part, chose the latter.\" This optimistic opinion can be corroborated only in the most fantastic and playful of texts. Social exclusion threatens the drag queens who stop off at a small town in _To Wong Foo, Thanks for Everything, Julie Newmar_ (Beeban Kidron, 1995). \"Look at 'em. Perverts!\" exclaims the corrupt and homophobic sheriff. \"When the Founding Fathers wrote the Declaration of Independence and the Constitution, and what have you, bringing in justice for all, they didn't mean them.\" He is proved wrong: the people in the film value the impact of drag queens on their society.\n\nIt is the epidemic that has released Glenn, Joey, and their friends. Like the men in the film _The Living End_ (Gregg Araki, 1992), they are free because they believe they will die shortly anyway. \"And there was a terrifying but profoundly seductive freedom in that. They could do anything now, anything they wanted to, anything at all\" ( _Tim and Pete_ , 187). Glenn and Joey are free to engage in a range of sexual activities about which Tim and Pete are reluctant even to fantasize. Also, they are free for kamikaze sacrifices (180). So they enact what for other dissidents can only be fantasy. While gay men have not proved an insurrectionary force in the United States, others have. Mikey's idea of seizing a plane and crashing it into the building where the president will be is \"'a fantasy,'\" Joey pronounces. \"' _So what?_ ' Mikey said. 'Everything _starts_ as a fantasy!'\" (240; Baker's emphases).\n\nIn arguing that hierarchy is sexy, I am not saying that we should be abandoning politics, morality, and responsibility and plunging into reactionary and complacent relations which exploit oppressions of class, age, race, and gender. Thinking even for a moment about the reality of those oppressions should restore our commitment to fight them. Bersani, famously, evokes the anguish that may be triggered by the image of a man with his legs in the air. But you would have to be truly perverted to find this more distressing than the image of a starving child.\n\nIf lesbians and gay men had in fact succeeded in wiping out power in relationships, all we would have to do is enjoy our egalitarian practice and let everyone else in on the secret. But that is far from the case. The prevailing sex\/gender system, we have every reason to know, is geared to the production of hierarchy and, as part of that, to the production of anxious, unhappy, and violent people. It produces us and our psychic lives\u2014straights and gays\u2014and it is not going to leave us alone. Arguably, fantasy becomes specific at the points where it is most at odds, superficially at least, with reality. It is a liberal-bourgeois delusion to suppose that \"private\" space can be somehow innocent of and protected from the real world. The task is to find ways of engaging fantasy without hurting and disempowering other people.\n\nWe have to accept that crucial political commitments may fall out of alignment with our fantasies. As Judith Halberstam puts it, \"while people may well invest in values like equality and reciprocity in their political lives, they may not want those same values dominating their sexual lives.\" Bersani discovers a positive virtue in such misalignments: \"Our fantasy investments are often countered by more consciously and more rationally elaborate modes of reaching out to others, such as liking or admiring people we don't desire. In that tension lies an important moral dimension of our political engagement.\" A politics asserted over libidinal investments may be more considered, more authoritative.\n\nIf we don't acknowledge power differentials in our fantasies and our relationships, we don't begin to get a hold on exploitation\u2014including that which we perpetrate ourselves. While the political priority of resisting actual oppressions must be maintained, power imbalances in lesbian and gay personal relations may be refigured as potentially rewarding, though inevitably troubling. We should be exploring ways to assess and recombine power, sexiness, responsibility, and love.\n**5**\n\n**GENDER**\n\nThus far I have tried to sketch a materialist theory of psychic life, drawing attention to the power hierarchies that dominate our relationships. While I have isolated gender, age, class, and race as the principal vectors of power, special confusions envelop the idea of gender.\n\n**HISTORY AND THEORY**\n\nThe most disputed question in our historiography is whether there have always been lesbians and gay men, or whether we are a recent development\u2014since the nineteenth century, according to Foucault, or, in some versions, since the Stonewall Riots of 1969. The former, universalizing position is taken by Rictor Norton and Terry Castle, who discover in the molly-house subculture and masquerading women of the eighteenth century a gay and lesbian history that is continuous with the present. Note, however, that these evidences of continuity depend upon gender identities.\n\nConsider the mollies of the early eighteenth century, who set up clubs, cross-dressed, and took women's names. One contemporary account describes them as \"so far degenerated from all masculine deportment, or manly exercises, that they rather fancy themselves women, imitating all the little vanities that custom has reconciled to the female sex, affecting to speak, walk, tattle, cur[t]sy, cry, scold, and mimic all manner of effeminacy.\" This, surely, evokes a subculture organized around gender, not sexuality. According to Alan Bray, what \"most scandalised contemporary journalists writing about the molly houses was the extravagant effeminacy and transvestism they could involve.\" The focus was on people who felt themselves, or behaved as if they felt themselves, to be the \"wrong\" gender\u2014not on people having, or desiring, same-sex relations. Nonetheless, Bray reads the mollies as \"homosexual\": \"There was now a continuing culture to be fixed on and an extension of the area in which homosexuality could be expressed and therefore recognised; clothes, gestures, language, particular buildings and particular public places\u2014all could be identified as having specifically homosexual connotations.\" Norton makes the same elision.\n\nMy argument is that most cultures give primacy either to gender identity or to object-choice. One of these terms tends to serve as the primary interpretive instrument; the other is incorporated as a subordinate, and consequently incoherent, subcategory. Of course, this does not mean that the subordinate discourse will be entirely untenable; social systems are always complex, comprising residual and emergent elements. But it will be more difficult to hold, for the individual and for others.\n\nTerry Castle adduces the diaries of Anne Lister, composed between 1817 and 1824, in pursuit of her contention that the lesbian is not a recent invention. Lister and her partners refer to her repeatedly as masculine, a man, gentleman-like and having manly feelings, and to her female partners as Lister's wives or subject to her adulterous approaches. Lister fantasizes herself in men's clothes and as having a penis, and models herself on Lord Byron. For Castle, and Norton also, this is evidence that the lesbian identity existed before 1869. However, as Judith Halberstam points out, it makes better sense to regard Lister as transgendered; she even rejected the label \"sapphic,\" the contemporary term for sexual relations between women, insisting on her own masculinity. The many eighteenth-century instances surveyed in Emma Donoghue's _Passions Between Women_ indicate that women who desired other women were persistently discovered to be of irregular gender\u2014hermaphrodite, male-identified, or cross-dressed. Even romantic friendships might be structured as husband and wife, or involve cross-dressing. Donoghue wonders \"why a woman who loved women would want to pass as a husband.\" The reason is that she was male-identified; her object-choice was a consequence of that. Castle, indeed, introducing her collection of lesbian writing, remarks how, from the eighteenth century to Krafft-Ebing and Havelock Ellis, \"the wish to associate female homosexuality with physiological oddities such as distended genitalia or supposedly masculine features... seems inevitably part of a 'naive' or reflexive response to the lesbian idea.\" She supposes that the lesbian idea is prior, and then misrepresented; I think (alleged) masculine features dominated thought and action in this period. Generally, on through the nineteenth century, George Chauncey Jr. has shown, gender was the prior category. \"Investigators classified a woman as an invert because of her aggressive, 'masculine' sexual and social behavior, and the fact that her sexual object was homosexual was only the logical corollary of this inversion.\"\n\nTwo distinct aspects of the sex\/gender system are in play, then. One is structured in gender identity, and tends to look for signs of femininity in men and masculinity in women. The other is structured in object-choice, and depends upon sexual and\/or emotional commitment to another person of the same gender. To be sure, they are bound to become tangled together; nonetheless, they are analytically separate, and by no means necessarily either homologous or in a permanent relation. Gayle Rubin, coming from feminism, reaches the same position; so does Eve Sedgwick. John Fletcher declares:\n\nI want provisionally to hold apart, to separate at least analytically gender, sexuality and sexual difference, in order to interrupt the too easy assimilation of sexuality and the sexual into the question of sexual difference, male and female on the one hand, and the equally common and too ready assimilation of that sexual difference into the question of gender on the other.\n\nYet opinion is not settled. Tamsin Wilton remarks: \"The interlocutions between discourses of gender and the erotic manifest a complexity that I suggest indicates that they may not usefully be distinguished one from the other.\" William J. Spurlin makes a thorough case for this. In his essay \"How to Do the History of Male Homosexuality,\" the four premodern male traditions adduced by David Halperin (effeminacy; pederasty or \"active\" sodomy; friendship or male love; passivity or inversion) are derived empirically from particular times and places; they do not admit a primary distinction between gender identity and object-choice.\n\nThe question with which I began, about continuity and rupture in lesbian and gay histories, is awkwardly posed, therefore: it runs together gender dysphoria and dissident object-choice. The confusion in these discussions may be traced to a reluctance to distinguish between desire-to-be and desire-for. Gender identity is a kind of desire-to-be, whereas object-choice is about desire-for. As I elaborated these categories in chapter 2, they involve two quite different kinds of dissidence. In model (rg), the relative gender model, the man constitutes a double affront to convention:\n\n(rg) A man has:| desire-to-be F| desire-for M  \n---|---|---\n\nThe man has a \"wrong\" gender identity and his object-choice is \"wrong\" as well, insofar as it is homosexual. Yet insofar as there is an \"F\" and an \"M\" in the model, he may be assimilated into a conventional binary formation. In model (c), the complementarity model, the dissidence is again in the object-choice (one male desires another), but without the offense of dissident gender identity and the consolation (as it is conventionally perceived) of a shadow of heterosexual desire:\n\n(c) A man has:| desire-to-be M| desire-for M  \n---|---|---\n\n**ORIGINS AND SPECIES**\n\nWhen these matters began to attract scientific inquiry, in the nineteenth century, the reigning theoretical construct was _degeneracy_. This was a product of Darwinism\u2014a kind of reverse evolution; it was a vehicle for anxieties about class, imperialism, and European racial superiority. So far from differentiating gender identity and object-choice, degeneracy subsumed both of them, along with perversion generally, into a very broad concept of weakness, debauchery, madness, and criminality, resulting from an alleged hereditary corruption.\n\nIt was in reaction against the crudeness of degeneracy as a construct that more patient and sympathetic theorists, now often called \"sexologists,\" isolated _inversion_. They encountered self-confessed inverts who seemed admirable, and therefore attributed the condition to a congenital, but not necessarily pathological, abnormality. Both Havelock Ellis and Freud explicitly reject degeneracy as an explanation. Ellis supposed that upon conception everyone might have 50 percent of male \"germs\" and 50 percent of female ones, and that in \"the homosexual person\" and \"the psychosexual hermaphrodite\" something interfered with the business of sorting out which were to predominate; this is an elementary anticipation of Freud's idea of a universal original bisexuality. Whatever else psychoanalysis did, Foucault remarks, at least it opposed \"the political and institutional effects of the perversion-heredity-degenerescence system.\" Martin Scherzinger and Neville Hoad remark that development correlates with evolution in Freud's thought, and arrested development with degeneration.\n\nThe sexologists' investigations generally started from the most conspicuous form of dissidence: \"wrong\" gender identity. They tended to suppose that \"wrong\" object-choice would correlate, but quickly ran into instances where this seemed not to be the case. One response, as Chauncey says, was to be more careful with one's terminology. \"While 'sexual inversion' referred to an inversion in the full range of gender characteristics, 'homosexuality', precisely understood, referred only to the narrower issue of homosexual object-choice, and did not necessarily imply gender or sexual role inversion.\" Halperin, in this vein, credits Ellis and Freud with the crucial discrimination: \"That sexual object-choice might be wholly independent of such 'secondary' characteristics as masculinity or femininity.\" Ellis and Freud do make this distinction. Notwithstanding, their accounts of object-choice reinstall gender identity as a typical component.\n\nEllis was aware of the potential for confusion. In _Sexual Inversion_ (1897), he notes that Albert Moll has tidied up the terminology and recognizes only two terms: \"psychosexual hermaphroditism and homosexuality.\" This, broadly, is the discrimination that I have been broaching: between gender dissidence and \"an inclination [in a man] toward men.\" In the second edition of _Sexual Inversion_ (1901), Ellis clarifies his account of Moll's argument and remarks in a note that Moll now wants to reserve the term \"inversion\" for gender disturbance\u2014for \"those cases in which there is a complete turning around of the sexual instinct, the man feeling in every respect as a woman, the woman in every respect as a man.\" Ellis concedes that there is something to be said for Moll's distinction, but he prefers another. The term \"homosexuality\" he applies \"to the phenomena generally,\" reserving \"inversion\" for \"those cases in which the sexual attraction to the same sex seems to be deep-rooted and organic.\" This is more than terminology: Moll proposed two discrete conditions, whereas Ellis is positing two levels of intensity.\n\nSo if inversion is a \"deep-rooted and organic\" condition, as opposed to \"the phenomena generally,\" what are the signs of it? In practice, Ellis allows it to correspond to the presence of \"wrong\" gender characteristics. To be sure, he insists, the male invert need not be effeminate, he may just make a same-sex object-choice. Nonetheless,\n\nit must be said that there is a distinctly general, though not universal, tendency for sexual inverts to approach the feminine type.... Although the invert himself may stoutly affirm his masculinity, and although this femininity may not be very obvious, its wide prevalence may be asserted with considerable assurance, and by no means only among the small minority of inverts who take an exclusively passive _role_.\n\nIn other words, the gender invert is the most complete type of homosexual. Object-choice is in a continuum with the model of \"wrong\" gender identity. So with women: \"In inverted women a certain subtle masculinity or boyishness is equally prevalent.\"\n\nThe key Freudian text here, _Three Essays on the Theory of Sexuality_ (1905), is far closer to Ellis than is often recognized. Freud places same-sex object-choice at the forefront of his analysis. But, so far from maintaining a differentiation, he continues: \"People of this kind are described as having 'contrary sexual feelings,' or better, as being 'inverts,' and the fact is described as 'inversion.'\"\n\nLike Ellis, Freud dismisses fanciful theories of innate inversion, and refutes the anatomical basis claimed in such theories. However, also like Ellis, Freud cannot forsake the thought that gender identity and object-choice might line up after all. He maintains this especially, though not only, in the case of women. \"The position in the case of women is less ambiguous; for among them the active inverts exhibit masculine characteristics, both physical and mental, with peculiar frequency and look for femininity in their sexual objects\u2014though here again a closer knowledge of the facts might reveal a greater variety\" (57).\n\nThis indeterminacy persists in a note added to Freud's _Three Essays_ in 1915: \"Finally, it may be insisted that the concept of inversion in respect of the sexual object should be sharply distinguished from that of the occurrence in the subject of a mixture of sexual characters.\" Yet the ensuing sentence is equivocal once more: \"In the relation between these two factors, too, a certain degree of reciprocal independence is unmistakably present\" (57\u201358). So sexual object and gender identity are to be sharply distinguished; yet there is a residual, unspecified interaction. In a further note, added to the _Three Essays_ in 1920, Freud tries a new tack. Following Sandor Ferenczi, he posits two types of male homosexual: \"'subject homoerotics,' who feel and behave like women, and 'object homo-erotics,' who are completely masculine and who have merely exchanged a female for a male object\" (58). Only object homo-erotics, Ferenczi says, may be \"influenced psychologically,\" whereas for subject homo-erotics there can be no question of \"struggling against their inclination\" (58). Once again, the real homosexual, the one who cannot be reached through analysis, is the one who wants to identify across the gender divide. Freud, while accepting Ferenczi's distinction, notes also that many people combine elements of both.\n\nC. A. Tripp in his comprehensive survey of homosexuality, first published in 1975, dismisses psychoanalysis and declares, \"Only in popular thinking are homosexuality and inversion synonymous.\" However, prominent commentators of our own time have allowed confusion to persist. As Sedgwick confirms, Foucault's invocation of \"a personage, a past, a case history, and a childhood\" is often invoked as charting the full emergence of a modern concept of sexuality. However, Foucault says he is presenting \"the nineteenth-century homosexual\"\u2014in fact, the invert. He specifies as the founding text Karl Westphal's article of 1870 on \"contrary sexual sensations,\" in which the examples are a lesbian who dreamed she was a man, and an apparently heterosexual cross-dressing male prostitute. Indeed, \"wrong\" gender identity is central to the historical change registered by Foucault: \"Homosexuality [ _L'homosexualit\u00e9_ ] appeared as one of the forms of sexuality when it was transposed from the practice of sodomy onto a kind of interior androgyny, a hermaphroditism of the soul.\" This change is characterized \" _less by a type of sexual relations_ [that is, less by object-choice] than by a quality of sexual sensibility, a certain way of _inverting_ [ _d'intervertir_ ] the masculine and the feminine in oneself.\" Foucault is saying, in other words, that the modern homosexual is a blurred, composite figure, conceived in a confusion between object-choice and gender identity. Unfortunately, he does not explicate that confusion.\n\nGert Hekma, writing from the anthropological tradition, reviews the arguments of the classical inversion theorists (not including Freud), showing that their emphasis on inversion obscured object-choice, thus causing \"additional problems and hesitations in coming out as homosexuals\" for \"more masculine boys.\" Nowadays, however, \"the spectrum of gender possibilities has broadened to include different options\" and \"feminine styles\" are \"part of the diversity of the gay world.\" Descriptively this is fair, though perhaps a bit eager to free the post-Liberation gay man from the stigma of effeminacy. But Hekma's judicious empiricism forgoes any attempt at a theory, either of gender or sexuality. There used to be a theory\u2014inversion\u2014but it was wrong; now we just have all kinds of people doing all kinds of things.\n\nAt the other extreme, Kaja Silverman's determination to retheorize these topics within psychoanalysis plunges her back into Oedipal conjecture and an assumption that gender identity is the key to homosexuality. After all, she avers, \"human culture has to date shown itself to be stubbornly resistant to conceptualizing sexual positionality\u2014and, more recently, object-choice\u2014apart from the binary logic of gender.\" Indeed it has. Silverman is not interested in separating out gender identity and object-choice; she believes that the gay man has to accept that \"an identification with 'woman' constitutes the very basis of his identity, and\/or the position from which he desires\" (344). Lately, Silverman notes, some gay men have tended to appear masculine, but she isn't impressed: \"It is by no means clear, anyway, that even the most committed practitioner of macho homosexuality can ever succeed in entirely extirpating the 'woman' within\" (346). Even while abjuring \"global pretensions,\" and acknowledging that her theory may account only for \"certain kinds of male homosexuality,\" Silverman launches into her own reworking of the Oedipus complex as a mechanism for the acquisition of gender identity (346\u201347).\n\nThe most ambitious investigation of the thesis that the modern homosexual derives from the late nineteenth century is offered in Neil Bartlett's creative documentary, _Who Was That Man? A Present for Mr. Oscar Wilde_. Once, Bartlett says, he experienced his gayness \"in complete isolation\"; now he is \"connected with other men's lives, men living in London with me. _Or with other dead Londoners_.\" To establish this linkage, he ransacks nineteenth-century documents, especially around the Wilde scandal. An awkwardness, however, is that most of the available instances appear to involve effeminate men. This places them in specific discontinuity with contemporary gay subculture, the achievement of which, as Bartlett sees it, is to make us \"handsome, _masculine_ , demanding and unafraid of our pleasures\" (219; my emphasis). The signals for gayness are historically variable, then: in the nineteenth century they centered upon effeminacy; today they are said to involve a relatively unremarkable masculinity.\n\nBartlett was writing in that period\u2014which developed from the mid-1970s and held the field through the 1980s\u2014when it was most difficult to appreciate gay femininity: in the heyday of the \"clone\" gay image. It was the time when Gregg Blachford and Jamie Gough could write, almost without reservation, of the masculinization of the gay world. Dennis Altman observed: \"The long-haired androgynous look of the early seventies was now found among straights, and the super-macho image of the Village People disco group seemed to typify the new style perfectly.\" Richard Dyer, in an essay first published in 1981, confirms \"the current 'masculinization' of the gay male style.\" However, he saw something else in the Village People: \"all the stereotypes of ultramasculinity in a camped-up flauntingly gay way.\" Even at this point, gay effeminacy has not gone away.\n\nIn fact by no means all Bartlett's contemporary gay Londoners were macho, but in _Who Was That Man?_ he accepts the image. He himself wears Doctor Martins, 501 jeans, a check shirt, and a moustache: \"I look like, or rather hope that I look like, a lot of other gay men\" (205). Again, at the start of a section mainly transcribing nineteenth-century notions of effeminacy, he remarks how in his masculine gear he may pass as straight, while still being visible to other gay men (63). This is the image he presents on the BBC2 television program _The Late Show_ (1993). There he urges upon gay men eclectic subcultural appropriation: the system scarcely acknowledges us, but we are piecing together our own lives. He offers his own outfit as an instance. Some might say that he is not entitled to wear it\u2014he is not one of those \"regular guys.\" But he has \"earned the right,\" he says, through the (manly) confidence with which he carries it off.\n\n_Who Was That Man?_ makes much of the famous transvestites, Fanny Park and Stella Boulton, who often passed as young women until they were arrested in 1870. Bartlett sees them as demonstrating \"the existence of our culture in London,\" though the charge was \"being men and dressed in female attire\" (143, 132). Significantly, he stops short of actually identifying with them: \"I would applaud the men who wore them in their determined efforts to use their frocks to create public space for themselves,\" he says (137\u201338). Such a combination of affiliation and distance occurs again when Bartlett remarks: \"I always enjoy asking a friend, in all drunken seriousness, _how's the wife?_ We both know that there is no useful comparison between heterosexual marriage and the relationship being referred to.\" There is a point, though: \"in using the word, I recall the house at 46 Fitzroy Street\" where the police arrested a group of transvestites on August 12, 1894 (85; Bartlett's emphasis). The connection is awkward, but Bartlett believes it encodes a historical affirmation that is worth making.\n\nSuch strategic effeminacy has partly informed Bartlett's extravagant deployment of camp and drag in the theater, particularly in plays from that time, _A Vision of Love Revealed in Sleep_ and _Sarrasine_. In an interview, Simon Fraser questions him: \"But this is 1991. Is drag really important to gay men?\" Bartlett's reply is not that lots of gay men are feminine really, but that drag is an emblematic part of the culture of British gay men: \"Almost all the things that are now traditionally gay are very important for that fact alone, and they represent gay space. They are a cultural space which we can inhabit.\" The value lies in \"seeing a gay entertainer in a room full of gay people, speaking a language that no one else could understand.\" We do drag, then, not because we are really feminine, but because it's been one of our things; it is a matter of subcultural affiliation and respect.\n\nIn my view this is a fair proposition, but insofar as it subsumes gender dissonance into gayness it tends to marginalize men for whom femininity is the primary factor. In fact Bartlett's work, despite his apparent privileging of the macho image, has contributed significantly to the recovery of subcultures of effeminacy. As regards camp and drag in his stage work, he evidently has been intrigued by their theatrical potential. In Bartlett's novel _Ready to Catch Him Should He Fall_ , Mother gets Boy into drag, but at an intense moment \"his white-powdered female face and livid red lips were suddenly split open by a masculine grin of triumph.\" This grotesque juxtaposition may remind us that drag is not about women, it is about gender boundaries; theater is not about display, but appearance and reality. Boy's triumph consists partly in the quality of his masquerade, partly in his maintenance of maleness. An ideology does not require the suppression of its other, but its productive management.\n\nThe Bar in _Ready to Catch Him_ rehearses these issues. Boy is introduced as conspicuously masculine: \"Keep him strong, keep him young, and, whatever his colouring, keep him gorgeous.\" The narrator invokes an allegorical figure of \"Strength\" (14\u201315). The Bar regulars appear to belong to another generation. Their culture is insistently feminine and centered upon gender identity. They have girls' names, camp talk, and a penchant for cross-dressing. Boy is not like them. At one point he is said to imagine leaving The Bar with a husband, but the narrator checks himself: \"(Of course, Boy would never have used that word, _Husband_ , that's my word. But then, I'm old-fashioned, I mean, we used to talk like that all the time. What word do you use, then?)\" (49; Bartlett's emphasis). When Boy is being prepared to celebrate his union with O, he is dressed in drag, as a smalltown queen, and as a woman by Madame and Stella. However, he is also attired as a schoolboy, a soldier, and a black man. For the wedding Boy is not in drag; there is \"no priest and no frock, this being an actual ceremony and not some party or parody\" (207).\n\nYet, in practice, few of us want unremitting machismo. Bartlett remarks in _Who Was That Man?_ : \"I too often require of myself and my partners a female nature\u2014sexually available, domestic, a surprisingly good cook and at all times attractively dressed\u2014inhabiting a male exterior\u2014sexually aggressive, potent, financially successful, socially acceptable\" (63\u201364). In his attempt to reconcile the contradictory concurrence of (what he takes as) masculine and feminine norms, Bartlett comes unexpectedly close to the idea of Ulrichs and the sexologists: our souls may be partly female, but the male body is crucial.\n\nOf course, not all gay men became clones; camp and drag continued to thrive; clone got described as a kind of drag. The position was not coherent, but it seemed to suffice, until the mid-1990s when transgender people, by declaring themselves, made the illogicality blatant. Lately, under the regime of Queer (regarded either as a political intervention or as a capitulation to capitalism), almost all styles are welcome. Augmenting the standard menu to \"LGBTT\" prompts a recognition that we still haven't sorted out the relations between sexual orientation and gender identity.\n\n**WHAT HAPPENED AT STONEWALL**\n\nGiven the persistence of confusion over object-choice and gender identity in the thought of the most prestigious theorists of dissident sexuality, it is hardly surprising that gay men and lesbians have not readily clarified these matters. In the late nineteenth century some men, appealing to Walt Whitman, David and Jonathan, and the Theban Band, sought to establish that, so far from being effeminate, same-sex love might be quintessentially masculine. In Germany in the early twentieth century manly homosexuality was celebrated by the group around the journal _Der Eigene_. They abjured the notion that homosexuals were feminine in disposition, reasserting sexuality in chivalric love and the love of friends. In terms of the models I have proposed, they posit males desiring other males, without relinquishing their masculine gender identities:\n\n(c) A man has:| desire-to-be M| desire-for M  \n---|---|---\n\nThey manifest a conventional gender identity alongside a dissident object-choice.\n\nHowever, this was not the dominant ideology. In _Gay New York_ , Chauncey has shown that until the 1930s men were not divided by object-choice, into \"homosexuals\" and \"heterosexuals\": this was not the primary axis of identification. A man displaying a committed feminine manner got called a fairy, but \"the 'man' who responded to his solicitations\u2014no matter how often\u2014was not considered abnormal, a 'homosexual,' so long as he abided by the masculine gender conventions.\" Donald Webster Cory, writing in 1951, cited a report from a U.S. sailor who believed that \"the stranger who performed fellatio\" was \"homosexual,\" but not the man on whom it was performed. \"The performer was a 'fairy.' The compliant sailor, not.\" Still, between 1930 and 1960, gender dissidence is found by David K. Johnson and Allen Drexel to be the organizing principle of gay subculture in Chicago. A similar story is told by Edmund White's friend Lou in _The Beautiful Room Is Empty_ : homosexuals divide into boys, men, and vicious old queens. \"The boy felt a natural affinity to girls, with whom he was always exchanging makeup tips. The man had once fucked girls but now had no further use for them.\" Tragically, \"whoever succumbed to homosexual desire became immediately undesirable.\" In Britain, John Marshall has shown, gender inversion remained the dominant paradigm on into the 1970s; it \"effectively eliminated the need for a homosexual concept.\"\n\nThis analysis has consequences for the mythology of Stonewall. Who, when we liberated ourselves, came out? Not the man who presented himself as inverted, effeminate; he was always visible. The ultimate instance is Quentin Crisp. He says in _The Naked Civil Servant_ (1968) that people such as he \"must, with every breath they draw, with every step they take, demonstrate that they are feminine.\" Crisp was never _not-out_ : continually he is propositioned, harassed, and beaten, on sight and by total strangers; employers and the army reject him out of hand. The distinction between manifest and closeted gay men is urged by Peter Wildeblood, writing with unprecedented boldness after his trial and conviction in 1954: \"Everyone has seen the pathetically flamboyant pansy with the flapping wrists, the common butt of music-hall jokes and public-house stories.\" Such people, evidently, were not hiding. However, Wildeblood adds, \"Most of us are not like that. We do our best to look like everyone else, and we usually succeed.\" The object-choice men strive to distance themselves from the pansies. It is to these men, the ones who managed to look \"like everyone else,\" that the 1970s offered a new identity.\n\nThe United States produces equivalent instances. Michael Bronski observes: while \"most homosexuals could choose to 'pass,' the majority of homosexuals who formed this visible subculture were effeminate men, butch women, obvious queens, and the drags.\" Kenneth Marlowe in a popular book of 1968, _The Male Homosexual_ , distinguishes the \"effeminate\" and the \"masculine\" homosexual. The former, \"because of his physical appearance, is sometimes labelled 'queer' from the start of his life\"\u2014everyone can see at once that he's different. Meanwhile, \"the masculine homosexual,\" according to Marlowe, \"is usually referred to as the latent homosexual, the closet queen.\" Either way, he's not visible; hardly queer at all. Todd Butler, recalling New York in 1960, conjoins out-ness and effeminacy: \"It was unusual for somebody to be fully 'out' and leading a gay lifestyle to be that butch. Usually, if you met somebody who was gay and butch, they were very uptight, closety types and very, very neurotic.\" That sounds like Michael, Donald, and most of the men in Mart Crowley's _Boys in the Band_. Alan can hardly tell that he's crashed a gay party. But he knows about Emory: \"Faggot, Fairy, pansy... queer, cocksucker! I'll kill you, you goddam little mincing, swish! You goddam freak! FREAK! FREAK!\" Emory admits: \"I've known what I was since I was four years old.\" \"Everybody's always known it about _you_ , Emory,\" Michael quips, unhelpfully.\n\nIt is the straight-acting types who had a new opportunity: to come out. The 1970s are often presented as the birth of the modern gay man, his (self-)definition founded purposefully in object-choice, as a consequence of (partial) decriminalization in Britain, the Stonewall Riots, the founding of the Gay Liberation Front and other activist movements, and the burgeoning of a purposefully sexualized gay subcultural economy. The larger outcome was a reversal in the organization of the sex\/gender system. _Homosexual_ , _lesbian_ , and _gay_ got defined in terms of object-choice, and gender identity was subsumed, more or less uneasily, into that. Thus, for instance, we know that drag artistes are not invariably gay, but in practice tend to assume that they are; in _The Adventures of Priscilla, Queen of the Desert_ (Stephan Elliott, 1994), everyone is shocked that Tick (Mitzi) is planning to return to his wife and child.\n\nLesbian awareness was complicated by other priorities in the women's movement; yet, here too, the attack on butch\/femme styles and identities tended to validate object-choice rather than gender dissidence.\n\nThis has made better sense for many people, personally and politically. But, by the same token, people whose primary sense of themselves is strongly founded in gender dissidence\u2014effeminate men, butch women, transvestites, transsexuals\u2014have been marginalized. They have endured the dominance of a sex\/gender model in which they hardly figure, or only as incidental, unintelligible, out of date, embarrassing. Leslie Feinberg in hir novel _Stone Butch Blues_ shows butches being excluded by lesbians in the gay liberation movement.\n\nFor many gay men, the post-Stonewall reliance on object-choice has afforded opportunity for a denial that effeminacy is a necessary part of gayness. Andrew Sullivan is notorious for his insistence that gays are \"virtually normal\"; for him this actually means _not effeminate_. Sullivan denies that he has ever experienced insecurity about this, though he recalls that when he avoided team sports at school a girl did ask him, \"Are you sure you're not really a girl under there?\" Sissy boys are the problem, according to Sullivan. It is they who vindicate right-wingers, suffer identity conflict, and provoke explanations from psychologists. He is excited by the masculine tone of a gay party:\n\nWhile the slim and effeminate hovered at the margins, the center of the dance floor and the stage areas were dedicated to the most male archetypes, their muscles and arrogance like a magnet of self-contempt for the rest. But at the same time, it was hard also not to be struck, as I was the first time I saw it, by a genuine, brazen act of cultural defiance, a spectacle designed not only to exclude but to reclaim a gender, the ultimate response to a heterosexual order that denies gay men the masculinity that is also their own.\n\nThis craven yearning to appease the sex\/gender system that has marginalized us leads Sullivan to boost his manhood with injections of actual testosterone. Apparently this bestows a Nietzschean, Superman buzz: \"What our increasing knowledge of testosterone suggests is a core understanding of what it is to be a man, for better and worse. It is about the ability to risk for good and bad; to act, to strut, to dare, to seize.\" Somewhat defensively, Sullivan puts down gays who can't or won't normalize as \"prone to adult dysfunction and pathology,\" and as \"insecure gay adults\" who \"will always cling, to a greater or lesser extent, to the protections of gender mannerisms.\"\n\nPaul Monette takes a more thoughtful approach in _Halfway Home_. Tom has been assaulted as a child by his father and brother (Brian), both of whom see him as a sissy and a wimp. Still, in his mid-thirties, he is \"a terrible sissy when it comes to crawling things.\" He has internalized this view of himself; indeed, he has made it productive by leaving his hostile, smalltown, Irish\/Italian-Catholic origins and becoming a performance artist in California (outside Malibu). His star turn has been his camp version of Christ on the cross, \"Miss Jesus.\" As he faces death from AIDS, Tom's memories are reactivated when Brian appears, under pressure in a racketeering inquiry, with his wife (Susan) and son (Daniel).\n\nThis turns out to be the opportunity for Tom to revise his relations with manliness. This is accomplished initially through a bond with Daniel (who is seven), founded in Tom's awareness that his own childhood traumas are being reenacted in Daniel by the damaged Brian. Daniel evidently likes being with Tom\u2014a man who is not going to become violent. As they walk together Daniel falls into step; Tom is gratified to be a model of manhood. He comes to realize that manliness is a precarious masquerade: \"there was no special dispensation\" in \"the secret to being a man.\" It was not a natural gift, possessed by his father and brother and withheld from himself, but something continuously improvised and mimicked, \"a waltz on the lip of the void\" (199). This (relative) demystification of manliness frees up Tom's anxieties. He finds he is imitating Brian's \"unconscious swagger,\" and thinks:\n\nThis was how straight boys learnt to be men, mimicking and preening, stimulating the butch gene. As I trotted down in Brian's wake, I thought about Daniel following him and following me. Somewhere there had been a trade-off, gentling my brother and toughening me. Brian stopped at the bottom of the stairs while I hovered a step above him, four inches taller now. And I prayed to the nothing I didn't believe in: _Let the kid have it both ways_. (238; Monette's emphasis)\n\nPerhaps the next generation of men will be able to incorporate something of the sissy.\n\nFor this reader, the determination in _Halfway Home_ to negotiate an acceptable form of manliness betrays a persisting anxiety. Brian doesn't give much ground; excessive hopes appear to be placed in alternative counseling sessions as a way of dealing with his violence toward Daniel. The brothers admit their youthful sexual activities together, though in Brian's case it appears to have been circumstantial and temporary (he was between girlfriends), whereas for Tom it was formative and continuing. \"'You know what I used to think?'\" Brian asks. \"'That I made you gay. Like it was all my fault.' He spilled out a soft self-mocking laugh, and his fingers rustled my hair. 'Like I tempted you'\" (177). This somewhat arrogant thought is not repudiated. The idea seems to be that the men, between them, engender, or anyway foster, the gay boy; he is a by-product of their manly maneuvering, and hence not to be unfairly despised and condemned. There is little room in this for Susan, the wife and mother. As a nonmasculine influence she is barely effective; her protectiveness toward her son leads Tom to compare her to Medea (121). Tom does have a supportive lesbian and gay family of choice, but they are edged to one side while patriarchal business is transacted (compare the emphasis on the brother in Larry Kramer's play _The Normal Heart_ ).\n\nA feeling that Monette is protesting too much is strengthened when Tom manages to prove himself in more traditional fictive manner by disarming a gangster. So the gay sissy is allowed to run with the men, at least up to a point; and, correspondingly, the heteronormative system can be humanized, such that the gay man can love his birth family (as it seems he must), as well as his family of choice. The trouble is that reconciling elements of the sissy with elements of the macho still leaves the true sissy exposed\u2014and, for that matter, the true macho.\n\nWhat _Halfway Home_ does show is that manliness is learned. Again, in Joseph Hansen's novel, _Steps Going Down_ , Cutler meets a young man. \"The boy shakes hands limply, awkwardly, like a little kid who hasn't learned the knack. Or like a girl. Cutler's handshake is strong, manly. His mother made him practice. _A good, firm handshake makes people respect you_.\" Cutler's manly stance projects dominance; the young man may be correspondingly boyish, or even girlish. Masculinity is something boys acquire, if they are lucky. Hence a central dictum in queer theory, as enunciated especially by Judith Butler: _all_ gender is performative. \"Everyone is passing; some have an easier job of it than others,\" Kate Bornstein observes. Again: \"Arnold Schwarzenegger does male drag perfectly, only he doesn't seem to have much of a sense of humor about it yet.\"\n\n**THE RETURN OF THE SISSY**\n\nThe triumphant revelation that the gruff-looking man in the hard hat speaks with a high-pitched voice, loves Bette Midler, and wants to be dominated in bed is repeated again and again in gay stories, as though it encapsulated a fatal truth. The persistence of gay male femininity, in the face of such discouragement, indicates that it is fulfilling some important functions in sexual dissidence. My larger thought here is that gender hierarchy is not something that gay men and lesbians have arbitrarily got stuck on: it involves one of the basic structuring ideologies in our societies and, like class and race, it is not going to go away.\n\nWhat makes this topic so complex is the mobility, fluidity, ubiquity, and inexorability of gender typing. Historically, as I tried to show in _The Wilde Century_ , effeminacy didn't always correlate with queerness; in the time of Shakespeare and Milton it meant paying too much attention to women. In the nineteenth century anarchists were termed effeminate, and Jews. Still, today, all kinds of sensitivity, consideration, colorfulness, and exuberance may be stigmatized as failures in masculinity. Writers have often struggled to articulate sensitivity while denying effeminacy. In _A Streetcar Named Desire_ , Blanche says her young husband had \"a softness and tenderness which wasn't like a man's, although he wasn't the least bit effeminate-looking.\" Probably many boys and men who exhibit femininity are mainly concerned to disaffiliate from the grosser aspects of conventional masculinity; they may simply be registering that a real man is not a very nice person.\n\nThese uncertainties about gender may be tracked in John Rechy's writing. In _City of Night_ (1963), the narrator is a male hustler who self-consciously asserts his own macho performance, in contrast to the adoring johns and queens, without admitting his own gayness. This is a precarious stance; one of the narrator's clients rejects him when he leafs through a book: \"'really masculine men don't read!'\" However, the queens come into their own in the later, Mardi Gras chapters. Twenty years later, in _Bodies and Souls_ , Rechy is looking for more positive potential in a reconciliation of masculine and feminine qualities. Billy and Stud are hustlers and presumptively not gay, but an affair develops between them. Stud's manliness is not at issue; he's not even gay, he says. Nor is Billy \"effeminate\"; he objects to being referred to as \"she.\" Notwithstanding, Billy is said to be \"beautiful,\" and Stud \"couldn't think of anyone being that beautiful and not a girl.\" The narrator interprets: \"Billy _was_ beautiful. He had a slender blond body that turned golden instead of tan, eyes so misty at times they looked painted with water colors, and long eyelashes. It was true he was not effeminate\u2014he was gracefully boyish, looking radiantly younger than his eighteen years.\" As Stud begins to fall for Billy, he finds that \"Billy's body was not softly formed; where had he got that idea? It was slim, yes\u2014but very solid looking\" (318). In fact, \"he was becoming more masculine all the time\" (323). So they are worthy of each other. Billy is represented as retaining, miraculously, both the feminine attributes that make him initially attractive, and the masculine attributes that make him an acceptable partner. Still Rechy is unable to get masculinity, femininity, and sexual attraction into the same frame without anxiety and contradiction.\n\nIn the purportedly documentary book _The Sexual Outlaw_ (1977), Rechy's narrative voice assumes a purposeful gay liberationist stance. He asks whether gays are appreciating \"their particular _and varied_ beauty? From that of the transvestite to that of the bodybuilder? The young to the old? The effeminate to the masculine? The athletic to the intellectual?\" He praises queens as \"true hero-heroines of our time, exhibiting more courage for walking one single block in drag than a straight-looking gay to 'come out' on a comfy campus.\" However, Ben Gove points out, the effeminate queens don't get much sex in _The Sexual Outlaw_ ; they are excluded from the masculine world of promiscuous cruising. They appear what they are\u2014left over in the post-Stonewall era which they helped to inaugurate.\n\nWhether it is true that effeminate gays don't get much sex is unclear. Edmund White's writings do not support the idea that the destiny of the effeminate gay man is inevitably lonely. In _The Farewell Symphony_ , he presents his desires as feminine\u2014\"I was so besotted by Kevin. I wanted to be his wife in the most straitlaced of marriages. I wanted to cook his breakfast and bear his babies. I wanted him to be my boy-husband, my baby-master.\" However, Kevin is in love with the handsome Dennis. Nonetheless, White gets plenty of fun, friendship, and sex in diverse roles; _The Farewell Symphony_ has been criticized for excessive reporting of tricks turned. The predicament of and opportunities for the feminine boy are displayed in Joseph Mills's Glasgow story, \"Dreaming, Drag.\" We first see David with his best friend, Joan, a lesbian: he has orange hair, and is wearing a \"working class housewife's coat.\" He doesn't want to be a transsexual: he loves inhabiting the male body, and the male orgasm. On the other hand, he disidentifies with \"the men in macho drag\u2014gay or straight\u2014who were in love with the male physique and all that went with it.\" He is uncomfortable when he realizes that, between these positions, there seems to be \"no place for himself\" (151).\n\nDavid's manner had quite pleased Walter, now his ex-boyfriend: \"He thought being with a camp guy made him more masculine, less gay. He was easing himself into a hetero relationship\" (88). Walter can't stand the stigma of gayness. David gets into a fight and decides that his cuts and bruises suit better with a masculine style, so he attires himself in denim, with a leather jacket and studded belt. \"Clothes make the man,\" he remarks (90). He falls heavily for Billy, who doesn't see him as camp; not that he's against camp, he thinks it very funny: \"'But I don't find it sexually attractive. I mean I like a man\u2014otherwise why be here?'\" (106). David deals with the mismatch by dressing himself in exactly the same clothes and dying his hair the same color as Billy. He finds himself elaborating contradictory fantasies about Billy and taking the \"active\" role in bed. However, Billy doesn't want a twin either. David goes back to feminine styles, abandoning \"'all that macho stuff.... Yes, we need to _complement_ each other'\" (117\u201318; my elision, Mills's emphasis). However, Billy is not convinced. He takes up with a man who has \"'the sort of Clark Kent, well-groomed executive look'\" (123). David is left railing against \"'your little hierarchy of homosexuals... with the just-like-everyone-else-Walter type on top and the deviants like me on the bottom'\" (130; my elision). He exclaims:\n\n\"God what a freak. Billy's right. I would never go with someone as effeminate looking as me.\"... Handsome, masculine gay guys don't fall in love with camp gay guys. _Camp_ gay guys don't fall in love with their own kind.... David had tried and tried again to be \"normal,\" but he resigned himself to the fact that camp was normal for him most of the time, whatever the consequences. (138; my elisions)\n\nHowever, it is not true that David doesn't attract men, and it is doubtful whether Billy's Clark Kent can turn into Superman. David remains irrepressible, like the divas he admires, and lives to fight another day.\n\nIn Leo Bersani's view, the matter is structural: as Quentin Crisp said, gay men are attracted sexually by machismo, not camp. Bersani does admit that not all gays are the same, but he is committed to a Freudian supposition that sexuality must be reducible to a binary gender structure. This leaves gay men desiring from the position of the woman, while entertaining a simultaneous and contradictory wish to imitate \"those desiring subjects with whom we have been officially identified: other men.\" With heterosexual women, therefore, we have (reluctant) identity; with men only (respectful) imitation. As Patrick Paul Garlinger suggests, Bersani's celebration of the subversive impetus of the man with his legs in the air may be designed to manage the stigma of effeminacy. While Bersani's organic model leads him ultimately to the idea of gay men as feminine and desiring the masculine, mine anticipates also a symmetry. If\n\nA man has:| desire-to-be F| desire-for M  \n---|---|---\n\nthen it is at least plausible that there is a corresponding figure:\n\nA man has:| desire-to-be M| desire-for F  \n---|---|---\n\nIn theory at least, there is a matching partner for almost all positions.\n\nSocial surveys indicate that there certainly are men who experience desire-for camp and cross-dressing men; Tim Bergling in his book _Sissyphobia_ finds some; the \"sissy boys\" interviewed by Richard Green report sexual partners. Gove concurs in respect of Miss Destiny in Rechy's _City of Night_ , and mentions the films _Stonewall_ (Nigel Finch, 1995) and _The Adventures of Priscilla, Queen of the Desert_. Bernadette in the latter had a boyfriend who recently died (the boyfriend had a thing about transsexuals\u2014\"a sort of bent status symbol\"); at the end Bernadette teams up with Bob. Divine and Our Lady of the Flowers attract masculine types in Genet's novel. In the film comedy _To Die For_ (Peter Mackenzie Litten, 1994), Simon loves and desires his camp partner Mark, in life and in death, but is blocked in the expression of his feelings by his relationship with his father. We are given to understand that Albin in _La Cage aux Folles_ (Edouard Molinaro, 1978) and Albert in _The Birdcage_ (Mike Nichols, 1996) have been beauties in their time.\n\nEddie is young, innocent, dreamy, and effeminate in _The Fruit Machine_ (scripted by Frank Clarke, released on video in North America as _Wonderland_ ; Philip Saville, 1988). He watches musicals and romances on video with his mother, is given to what she calls \"girly mannerisms,\" and is victimized by his father. He is plainly in love with his streetwise friend and protector, Michael, who regards himself as a straight rent boy and Eddie as a \"best mate.\" They are both sixteen. When Eddie approaches the point of a declaration, Michael is embarrassed and shuts him up. Eddie claims them for femininity: they're \"just a pair of little queenettes,\" he says. But Michael counters: \"No we're not, we're lads, we're young men.\" Gove remarks the division between \"gushy queen and defensive lad,\" and that the film assumes the latter to be the more sexy. However, the viewer's sense of Eddie's erotic potential is probably enhanced when, after appearing somewhat awkward hitherto, he swims gracefully among dolphins. Michael finally admits his love when Eddie is dying.\n\nIn some of these instances, femininity may be attractive because of a conflation with boyishness. A choice of an older man between his wife and a camp boyfriend figures in John Hopkins's play _Find Your Way Home_. However, this is not always so. An older feminine person may be protective (motherly), to either a masculine or a feminine boy. This happens in the film _Stonewall_ , where Matty Dean, the new boy in town, is drawn both to the masculine, assimilationist Ethan, and the established drag queen, La Miranda. In Harvey Fierstein's _Torch Song Trilogy_ , camp drag-artiste Arnold is attractive both to Ed, who is older and so straight-acting that he insists on cultivating his affair with Laurel, and to Alan, who is a young model and hustler. Arnold is obscurely diffident about the relationship with Alan, and the text evades its outcome through the boy's violent death; we may conclude that fate is against such a liaison. Arnold goes on to foster and adopt a teenage gay boy, and the return of Ed, finally unable to resist Arnold, constitutes an explicit family of choice with Ed as father and Arnold as wife and mother. This secures an upbeat ending, though Arnold has to fight off his own homophobic mother, who assumes that his interest in the boy must be predatory (these homosexuals will stop at nothing). From a queer viewpoint, it is perhaps disappointing that Arnold's integrity depends on the implication that sexual love between a man of thirty and a boy of sixteen (now the age of consent in England) must, _by definition_ , be out of the question.\n\nIt would hardly be surprising if the attractions of male femininity were understated in our cultures. Richard Goldstein may, in many cases, be right: \"butch is the face many gay men show to each other, but not the one they reveal to their lovers.\" Gays may be reluctant to admit the attractions of being the feminine man, and the rewards of desiring him, but he is not without his admirers. Fergus vomits in _The Crying Game_ (Neil Jordan, 1992) when he sees Dil's penis\u2014the sign that the femininity that had attracted him is attached to a male body. Notwithstanding, Fergus is drawn to Dil. To keep him safe Fergus makes him look like a boy and hides him away; he says it's their honeymoon night, and surely a sexual act takes place (nothing less would placate Dil, and in the ensuing scene it is morning, Dil is asleep, and Fergus looks decidedly pensive). At the end, Dil visits Fergus in prison, along with the other wives and girlfriends though in more exuberant style. Dil's idea, for one, is that they are an ongoing couple: \"I'm counting the days.\" \"Stand By Your Man,\" the music comes up. To take care of Dil is in his nature, Fergus says.\n\nIn summary, if our dominant story encodes male femininity as misfortune, we have also a counter discourse, in which the feminine man may be a turn-on. The problem, I suspect, is not that he is unattractive, but that he has difficulty in establishing a belief in his own worth, in the face of the gross stigma that attaches in our societies to effeminacy. In _A Boy's Own Story_ , Edmund observes that it is his own dislike of himself, endorsed by the prevailing ideology, that impedes his love life. \"I see now that what I wanted was to be loved by men and to love them back but not to be a homosexual.\" The narrator in White's later work, _The Farewell Symphony_ , observes: \"what the Stonewall uprising changed was not love so much as self-esteem, upon which mutual love depends.\" In White's _The Married Man_ , Austin attracts a partner who is masterful to the point of sustaining a conventionally sexual marriage.\n\nWhile there may be a symmetry, whereby a man with desire-to-be F is matched by a man with desire-for F, the person in the subordinate position is likely to be the more vulnerable. He puts himself at the greater risk, psychologically, when he offers himself in a relationship. The problem\u2014men with an element of dissident gender identity getting to feel good about themselves\u2014still has a long way to go. James Kenneth Melson, informing his mother of his positive AIDS diagnosis, reassures her: \"'Mom, to me my sexual preference is only that. I'm not a fag, sissy or queen.'\" AIDS we can cope with, effeminacy is beyond the pale.\n\n**TRANSGENDER**\n\nThe prospect of rewarding sexual encounters may not be the priority for gender-dissident people. Todd, age seventeen: \"That's the basic problem. I want to be a woman _before_ I want to have sexual relations with a man.\" Trans people have good reason to distinguish desire-to-be and desire-for. \"Gender identity for me answers the question of who I am. Sexual preference answers the question who do I want to be romantically or sexually involved with,\" Kate Bornstein declares.\n\nIn other societies there are conventional roles for transgender people. The _fa'affines_ of Samoa are apparently biological men who dress and largely live as women; they are recognized and reared as girls and appreciated as domestics and entertainers. Their sexual partners, they claim, are 99 percent of Samoan men. Some use them casually, some are ready for relationships. Young, articulate fa'affines have visited Australia and know that in other countries gay men sleep together. However, fa'affines don't have sex with each other; they are sisters; that would be two queens. They classify themselves as \"women,\" and don't accept that the terms _gay_ , _transvestite_ , and _transsexual_ fit them. Again: _skesana_ boys, in the single-sex context of mine-labor in South Africa, glory in their femininity and assume that their roles will be confirmed by their masculine partners:\n\nMARTIN: I think in a relationship the woman must attend to her man. Like a woman she must clean the house, and he must be treated like a man.\n\nTHAMI: There must be a \"man\" and a \"woman\" in a relationship. A man must act mannish in his behaviour and his talks and walks. But a female must be queenish in every way.\n\nThe _skesana_ (like a wife) gains protection and favors in a violent and uncertain system; also, Hugh McLean and the late Linda Ngcobo add, he \"attains pleasure by flirting with power.\"\n\nThere is no comparable framework for men in Britain and the United States for men who want to be women, but there is the amazing possibility, through medicine and psychiatry, of gender reassignment. In the face of the priority accorded to object-choice in our cultures, activists have put transgender on the agenda. Leslie Feinberg in _Stone Butch Blues_ and Kate Bornstein in _Gender Outlaw_ have published powerful and successful books based on their experience. Fiction on this theme includes _Trumpet_ by Jackie Kay, _The Danish Girl_ by David Ebershoff, and _James Miranda Barry_ by Patricia Duncker. Barry is the subject also of Rachel Holmes's study, _Scanty Particulars_. We have the disconcerting photographs of Del LaGrace Volcano. Films in general release include: _To Wong Foo, Thanks for Everything, Julie Newmar_ (Beeban Kidron, 1995), _Different for Girls_ (Richard Spence, 1996), and _Ma vie en rose_ (Alain Berliner, 1997). I've already mentioned _The Crying Game_ , and _The Adventures of Priscilla, Queen of the Desert_. Typically, these films depict transvestites and transsexuals enduring hostility and winning over ordinary, decent folk.\n\nAcademic studies have appeared, and the popular media are close behind. By early 1999, Polly Toynbee was writing in _Radio Times_ (February 13\u201319) that \"programme after programme seems to be obsessed with people of confused, indeterminate or wrong sex.\" She mentions a transsexual prostitute in the BBC1 series _Paddington Green_ , a transsexual in Granada's soap opera _Coronation Street_ , and a program in the BBC1 science series, _QED_. \"There is a questioning and a redefining of sex roles going on and these cases are just the most extreme manifestations of a more general and diverse debate,\" Toynbee avers. Subsequently, Channel Four has presented a very positive, two-hour view of four female-to-male trans people in different circumstances ( _Make Me a Man_ : Katie Buchanan, 2002), and in 2003 broadcast an American documentary, _Sex Change_ , showing the details of gender-reassignment surgery. Halberstam reports the interest of American talk shows in drag kings, but is disappointed by the sensational treatment. To be sure, visibility is not necessarily power\u2014otherwise, scantily clad young women draped across automobiles would be ruling the world.\n\nThe trans phenomenon challenges customary ideas about how gender and sexuality may interact. It is not just that a male-to-female transsexual may desire either a man or a woman, or both. When a person who has been assigned at birth a male gender experiences desire-for a male, _while declaring hirself to have desire-to-be a female_ , it is impossible to say, definitively, whether that person has made a same-sex or a cross-sex object-choice. It depends on whether hir story about hirself is credited.\n\nTransgender invites a reconsideration of _The Well of Loneliness_ , which has traditionally been regarded as a classic account of \"the mannish lesbian.\" Stephen does have desire-for other women, but her main anguish and affirmation resides in her desire-to-be male. \"'Do you think that I _could_ be a man, supposing I thought very hard\u2014or prayed?'\" she asks her father. Radclyffe Hall is drawing on ideas from the sexologists\u2014she persuaded Ellis to write a preface. Jay Prosser suggests that Hall prefers Ellis, Ulrichs, and Krafft-Ebing to psychoanalysis, and also to the kinds of lesbian self-concept being developed by Natalie Barney and Ren\u00e9e Vivienne, because they focus more on the invert and thereby fit better with her own project. Stephen's father has been reading the sexologists, and it produces in him the humane sympathy that Hall wanted in her own readers. However, it remains unclear how far Stephen's desire-to-be male is due to her upbringing (her father treats her like a boy and her mother rejects her), or congenital (in her physique, \"her nature\"; 29\u201330, 165). Both versions can be supported from the text.\n\nWhile Stephen's condition is entirely mysterious to her mother and herself, all the neighbors can see it: \"they feared her; it was fear that aroused their antagonism. In her they instinctively sensed an outlaw, and theirs was the task of policing nature\" (123). What they see is aberrant gender. A hotel porter can discuss it with his wife: \"'Have you noticed her, Alice? A queer-looking girl, very tall, wears a collar and tie\u2014you know, mannish. And she seems just to change her suit of an evening\u2014puts on a dark one\u2014never wears evening dress.... I dunno, there's something about her\u2014anyhow I'm surprised she's got a young man.'\" (181; my elision). This last inference is derived from the fact that Stephen is watching out for a special letter; it doesn't occur to the porter that she might be homosexual and looking for a letter from a woman. Stephen's condition is, at once, utterly opaque and devastatingly obvious, virtually inconceivable and cruelly policed. While she has to be ignorant in order to emphasize her bewilderment and perhaps her innocence, her transgression has to be communicated as the basis for her social rejection.\n\nThis rejection is grounded in her masculine appearance; her sex life is scarcely inferred. It is her body and clothes that are the offense, not her choice of partner; even the dog recognizes Stephen's maleness and longs for \"that queer, intangible something about her that appealed to the canine manhood in him\" (382). As Laura Doan has demonstrated, this failure to register sexual dissidence is plausible historically since, before _The Well_ , boyish and mannish garb did not register any one, stable effect; if anything, they signified modernity. It was Hall's accomplishment to change this.\n\nMary, on the other hand, is all right because she looks feminine. A fashionable hostess takes a great fancy to her, but sees in Stephen \"only an unsexed creature of pose, whose cropped head and whose dress were pure affectation; a creature who aping the prerogatives of men, had lost all the charm and grace of a woman\" (465). The fact that Stephen and Mary are both lesbians hardly figures. Halberstam remarks that _The Well_ manifests an epistemology not of the closet (for Hall, like Crisp, was never not-out) but of the wardrobe: male clothing is fetishized as the badge of gender. The tragedy is that Stephen is not allowed fully to inhabit that desire-to-be by becoming the master of Morton Hall and protector of hir partner.\n\nSo with the culminating affair with Mary. Stephen's initial doubt concerns whether she should respond to the innocent love of a younger woman who is not herself aware of inversion. (This dilemma recurs in Mary Renault's novel, _The Charioteer_.) The ensuing predicament, however, is that having accepted Mary's love Stephen declines to take her seriously. Stephen manages their joint affairs and gains success as a writer, but Mary has nothing to do. She has no work; she cannot go into Society, or to Stephen's family home; Stephen doesn't share her anxieties with her\u2014hardly trusts her to go out by herself. In short, Stephen envisions herself and Mary on a husband-and-wife model, and hence their downfall. It is because they cannot play that game convincingly that Stephen pushes Mary into an opportune marriage. This was not the inevitable outcome for the invert: Hall herself seduced two women away from wedlock.\n\nThe confusions that invest _The Well_ , we may suppose, were needed and desired. We should not be looking for the right reading, but observing how this protean text has been deployed for diverse purposes by diverse constituencies. Esther Newton highlights its importance in the mid-twentieth century to women who valued the overt sexuality of the mannish lesbian as pointing beyond the Victorian romantic friendship. The dominant ideology, meanwhile, was ready to deploy gender dissidence to stigmatize object-choice, and vice versa. Today _The Well_ is still available to new constituencies. While Prosser discovers there a depiction of transsexuality, Halberstam finds a portrait of the \"masculine woman.\" Comparable arguments may be developed in respect of Gertrude Stein: \"Though she's been widely regarded as a lesbian, the fact is that she saw herself, in essence, as a man,\" Jean E. Mills observes.\n\n\"'Is that a boy or a girl?'\" people keep asking about young Jess in Leslie Feinberg's _Stone Butch Blues_. Like Neil Bartlett's Boy in _Ready to Catch Him_ , Jess finds hirself in a bar, where s\/he is supported and appreciated as a baby butch. The downside is vulnerability to raids and gross assaults by the police. Jess, like other he-shes, works in a factory, but takes up with Theresa, who has a clerical job in a university and knows about the Daughters of Bilitis, the _Ladder_ , Stonewall, and lesbian and gay pride. However, butches and femmes are not welcome among liberationist lesbians, and Jess doesn't feel there is a place for her in the women's movement. It is no simpler for Edna, a femme: \"'I know I'm not a straight woman, and lesbians won't accept me as one of them. I don't know where to go to find the butches I love or the other femmes. I feel completely misunderstood. I feel like a ghost, too, Jess'\" (214\u201315).\n\nThese new difficulties for the he-she are reverberations from the major shift I have been discussing in this chapter\u2014the moment when gender identity is decisively superseded by object-choice as the key category for understanding sexual dissidence. Like Crisp, these women appear to be left over from the old conceptual regime, where gender dissidence was the main factor. Jess in fact seems not to get much sex; everyone is too exhausted just keeping going, and stone butches hate to be touched.\n\nTo resolve the indeterminacy Jess decides to begin hormone treatments with a view to passing as a guy. Theresa cannot tolerate this: \"'I'm a femme, Jess. I want to be with a butch.... I don't want to be with a man, Jess. I won't do it'\" (151; my elision). With the treatments, Jess passes as a man quite effectively. Hir problem is that s\/he now has no social context: intimacy threatens personal exposure and lovemaking seems meretricious. S\/he has to leave a job when hir gender history is revealed. S\/he stops the hormone treatment; \"Whoever I was, I wanted to deal with it, I wanted to live it again. I wanted to be able to explain my life, how the world looked from behind my eyes\" (224). Prosser is disappointed at Jess's turning back. It shows two things: that Jess is not at home in any version of hir body\u2014radical indeterminacy is hir fate; and that society has difficulty accommodating a transsexual even when s\/he is trying to regularize hir gender.\n\nTransgender promises to unsettle established items on the agenda of queer theory and activism. The dialectic of passing versus coming out works differently for trans people, who confirm that they are indeed the person whom they believe themselves to be when they pass undetected in public. Drawing attention to their special gender characteristics defeats the purpose. However, it is difficult to build a movement when its members keep vanishing into the crowd. Sandy Stone has urged transsexuals, instead, to come out _as transsexual_ , foregrounding thereby the constructedness of all gender categories. Stone actually has two main arguments. One is that passing forecloses \"the possibility of authentic relationships.\" This was Jess's problem: she felt unable to present hirself to other people\u2014until she wrote the book, that is. The other is that by aspiring to be essentially a woman or essentially a man, the trans person colludes with the gender system that has oppressed hir in the first place; whereas refusing to pass calls into visibility and mistrust the criteria that delimit the binary sex\/gender system.\n\nThat camp and drag may have potential to subvert the sex\/gender system has often been proposed, most famously by Judith Butler, who seemed to suggest in _Gender Trouble_ that \"drag fully subverts the distinction between inner and outer psychic space and effectively mocks both the expressive model of gender and the notion of a true gender identity.\" Butler's insistence that gender can never afford stable identities is surely right. However, there are two main problems with arguments that conceptual instability leads to insight.\n\nFirst, such theory, which is a kind of poststructuralism, has supposed too readily that to demonstrate indeterminacy in a dominant construct is to expose its weakness and its vulnerability to subversion. In practice, gay pastiche and its excesses may be all too easily pigeonholed as illustrating that lesbians and gay men can only mimic true manliness and womanliness. Dominant ideologies are able to turn almost anything to their advantage; that is the sign of their dominance. Whether an instance is subversive or incorporative has to be assessed in its particular contexts. The Stonewall queens instigated gay liberation not because they were camp or wore drag\u2014there was nothing new about that\u2014but because they fought the police.\n\nSecond, many individuals are already having a difficult time with their gender identity, and don't want their mannerisms co-opted into a political argument. This point is made by Prosser, who emphasizes that many trans people experience their gender \"precisely as a disorder, a physically embodied dis-ease or dysphoria that dis-locates the self from bodily home and to which sex reassignment _does_ make all the difference.\" Transsexual passing is like gay coming out in one respect: it is not a once-and-for-all accomplishment: it may have to be renewed many times each day. Prosser calls for a \"politics of home\" which \"would not disavow the value of belonging as the basis for livable identity.\"\n\nBelonging in the gender system is not the resolution that is offered in _Stone Butch Blues_. Jess reaches a stance of peace and productivity not by changing her body or her attire, but by finding other sex and gender dissidents in New York's West Village in the queer 1980s. Despite continuing horrific street violence, s\/he finds acceptance and purpose in a new world of bookshops, drag queens, farmers' markets, Christmas, other ethnicities, AIDS campaigning, people who can talk about how they hurt and need, a gay rally where s\/he feels able to speak about gender, and a future as a political activist. In a final dream Jess envisages a new world in which innumerable people of indeterminate gender are happy together. However, the message for the present seems to be: if you want to be different, find an alternative scene. Don't try to make your home in Middle America. The dangers are grimly confirmed in _The Brandon Teena Story_ (Susan Musca and Greta Olafsdottir, 1998), and _Boys Don't Cry_ (Kimberly Peirce, 1999).\n\n**BOYS WILL BE BOYS**\n\nEve Sedgwick has warned against uncoupling homosexuality and effeminacy\u2014\"effeminophobia,\" she calls it. \"After all, 'everyone already knows' that cross-dressing usually at least alludes to homosexuality\": culturally they belong together. Further, the persecuted \"effeminate boy\" is left without the support of adult gays, who perhaps eschew him as an embarrassing remnant of their own childhood trauma. Sedgwick is right to mistrust the motives of gay men who would jettison effeminacy, but the logical and experiential impetus of transgender has surely justified the isolating of gender identity as a category. A task now is to avoid any consequent marginalizing of yet other groups. Some butch\/femme women have said to me that they experience, in the current emphasis on transsexuality, a de-legitimization\u2014an imputation that butches are settling for a halfway position, as if they lacked the courage of their convictions.\n\nMeanwhile, straight cultures today make their own stories out of our lives. A recent study of sexual bullying in an English secondary school finds that boys regarded as insufficiently masculine are called \"gay\" and \"poof.\" However, they are _not_ thereby supposed to be homosexual; the idea that real homosexuals might actually exist in the school is greeted with incredulity. \"Nearly all the boys had knowledge of the existence of homosexuality but could not relate this knowledge to their school experience or people whom they might one day meet.\" In this setting, \"gay\" signals only gender anomaly; the boys are using it to support their concern with male bonding.\n\nThe same case is put by the misogynist, homophobic, and popular white rapper, Eminem:\n\n\"Faggot\" to me doesn't necessarily mean \"gay person.\" \"Faggot\" to me means \"pussy,\" \"cissy.\" If you're a man, be a man, know what I'm saying? That's the worst thing you can say to a man. It's like calling him a girl, whether he's gay or not. Growing up, me and my friends, \"faggot\" was a common word, like \"you're being a fucking fag, man.\" Nobody really thought \"gay person.\" I don't give a shit about gay men. If they wanna be gay, then that's their fucking business.\n\nIn both these accounts, actual gay boys are invisible. The boy who suspects that he _is_ gay is not just stigmatized; the compensations of gay belonging and subcultural resources are withheld from him. Wendy Wallace confirms that _gay_ has become a general term of bullying abuse for any outsider in schools, while retaining the link with its homophobic roots. Ten-year-old Damilola Taylor was repeatedly taunted at school with the word \"gay\" before being killed in a knife attack. In _The Laramie Project_ , Mois\u00e9s Kaufman's play about the death of Matthew Shepard, the murderer says Matthew looked \"like a queer. Such a queer dude.... Yeah, like a fag, you know?\" We know.\n\nIn propounding a primary distinction between gender identity and object-choice, my goal is not to interfere in the lived experience of people; this is not a call to individuals to sort out who they want to be. Rather, I am hoping to contribute to a better analysis of what we have been doing. The case for retheorizing the extant combinations of gender identity and same-sex passion rests not on an attempt to tidy up desire, but on enabling diverse peoples to respect themselves and each other. If there have been tensions among sexual dissidents\u2014caricatures, appropriations, repudiations\u2014that is because we are accustomed to constructing sex\/gender identities contradistinctively. Reassessing these processes may help us to rebuild a more elaborate dissident coalition, beyond binary organizations of difference. The rest of this book is about ways of reading sexuality that are not organized primarily around gender difference. Yet gender, we will see, proves remarkably persistent.\n**6**\n\n**AGE**\n\n**BOYS AND EMBARRASSMENT**\n\nIn Armistead Maupin's novel _The Night Listener_ , Gabriel is embarrassed about the age disparity between himself and his partner, Jess. Fifteen years is \"not that big a difference,\" Gabriel says and, anyway, because they came out at roughly the same time they are \"the same gay age.\" This idea offers to release them from the taint of age hierarchy and narcissism: \"This meant we'd reached the same level in our personal growth... which was far more pertinent to our compatibility than our chronological difference.\" Gabriel calls it \"my marriage\" (4).\n\nHowever, hierarchy has not in fact been irrelevant; it has informed their sexual practice. Gabriel remarks that it was a big turn-on when Jess would gaze up at him \"with slavish devotion. Or he'd work my nipples like a ravenous baby, murmuring, 'Sir, yessir, yessir,' until I came with a fury\" (50). Actually Gabriel is too nice and cuddly for too much of that; an earlier relationship foundered because he was unable to play \"The Great Dark Man\" (80). Now Jess has left Gabriel, partly because he is not going to die soon of AIDS after all, partly because he has been getting into a rough, more manly, leather scene\u2014shaving off his \"baby-chick hair,\" growing a beard. Gabriel feels \"old and disconnected\" (51). Suddenly he sees Jess as \"closer to middle age than to the soft-featured boy I'd fallen in love with\" (211). A pattern emerges: previously Gabriel had a relationship with Wayne, who was at least ten years younger: \"the grownup boy who had brought me childhood again.\" Gabriel confesses to \"youthful longing\": his ultimate pain is that he'll \"never be strong enough, never be handsome enough, never be young enough, to really be a man among men\" (132).\n\nThe age hierarchy of Gabriel and Jess correlates with class hierarchy, for Gabriel was already a celebrated gay broadcaster. Here too Gabriel has indulged in an element of disavowal. \"Until now our friends had been largely mutual; we had cultivated them together as couples often do,\" he says; \"Jess, after all, had been my satellite for ten years without complaint\" (51). The unremarked discrepancy between \"mutual\" and \"satellite\" marks Gabriel's reluctance to acknowledge that his \"marriage\" has been founded in inequality.\n\nThis scenario is close to Maupin's own experience, as may be seen in a television interview in which he and Terry Anderson present themselves as the ideal couple, devoted to coping with Terry's anticipated death from AIDS. Will and Jamie constitute a comparably symmetrical couple in Maupin's story, \"Coming Home\": they say the same things, laugh the same way, sound the same on the phone, speak of themselves as married. The age disparity is put at just five years, and Jamie has an independent occupation (coppersmith), whereas Jess manages Gabriel's affairs. (Will and Jamie stories are claimed in _The Night Listener_ as disguised autobiographical writing by Gabriel; 18). If, in retrospect at least, these versions of himself and his partners appear a tad manipulative and self-deceiving, in the novel Maupin confronts this thought, allowing us to see that the relationship had not really been equal and, indeed, had been experienced as lopsided and limiting by Terry\/Jamie\/Jess.\n\nGabriel's other main relationship in _The Night Listener_ provokes another anxiety about age hierarchy. Pete, a thirteen-year-old who has been intensively abused and has AIDS, strikes up a phone dialogue, founded in his admiration for Gabriel's broadcasts. \"'I guess he sort of has a thing about me,'\" Gabriel admits (39). Gabriel becomes equally involved; they get to use the terms \"Son\" and \"Dad\" (uppercase). Other people are suspicious, especially his reactionary father: \"'Well... you're a middle-aged man, and he's... well, people could get the wrong idea, that's all.'\" This facilitates Gabriel's spirited defense: \"'The boy needs love. You don't have to be straight to do that'\" (70\u201371; Maupin's pauses). Surely this is right, but its location in _The Night Listener_ suggests Maupin's determination to establish that, even if he has to concede that his partnerships have been organized around hierarchies of age and class, this doesn't make him a pedophile. Yet he is prepared to acknowledge \"a distinct resemblance\" in his way of relating to Jess (lover) and Pete (quasi son; 182). As it transpires, he is protected by another doubt, as to whether there could actually be such a boy; perhaps he is an invention\u2014the novel doesn't resolve this. These are awkward topics\u2014the age of consent in several European countries is fourteen. _The Night Listener_ discloses two embarrassments around age hierarchy: that it is immature and can't last, and that it is ultimately pedophilic. Gabriel, left alone at the end, questions the assumption that one must get a long-term relationship to lead a dignified and contented life. Perhaps the kind of partnership he can envisage will necessarily be immature; he might live better by himself.\n\nI distinguished _gender difference_ and _gender complementarity_ in chapter 2. The former is founded in (relative) difference of gender; the latter is founded in similarity of gender, but characterized by difference within that. The gender complementarity model is founded in similarity of gender, but characterized by difference within that. In chapter 3, on fantasy, I showed how the two models may be substituted and conflated. Age constitutes one of the key hierarchies in the sex\/gender system as it is lived in our societies. It is an inevitable factor in the first power structure that all of us experience: a child with adult caregivers. Without it the infant cannot survive and grow as a human creature. Age difference affords a hierarchy that may be used protectively. We should expect it to figure intensely in our psychic lives, and not to be confined to particular circumstances, such as same-sex communities or the coming-out process. The power distribution in the age\/youth binary structure may seem less obvious than the others: notoriously, an older man can make a fool of himself over a young man, or a young woman. Notwithstanding, class for class and race for race, older people control far more wealth and institutional power than younger people. It is because this pattern may be disrupted in our societies by a premium upon youthful beauty that it gets so much attention.\n\nEve Sedgwick observes, shrewdly, that Wilde, who acquired after the trials the representative role of aesthetical, dandified, and effeminate queer, seems not to have thought of himself or his partners as inverted in gender:\n\nWilde's own eros was most closely tuned to the note of the pederastic love in process of being superseded\u2014and, we may as well therefore say, radically misrepresented\u2014by the homo\/hetero imposition.... his desires seem to have been structured intensely by the crossing of definitional lines\u2014of age, milieu, initiatedness, and physique, most notably.\n\nWhat is surprising here is Sedgwick's rather sudden assertion that such \"pederastic love [was] in process of being superseded.\" David Halperin not only takes this alleged supersession for granted, he sees little reason to regret it:\n\nAlthough love, emotional intimacy, and tenderness are not necessarily absent from the [age-structured] relationship, the distribution of erotic passion and sexual pleasure is assumed to be more or less lopsided, with the older, \"active\" partner being the _subject_ of desire and the recipient of the greater share of pleasure from a younger partner who figures as a sexual _object_ , feels no comparable desire, and derives no comparable pleasure from the contact (unless he is an invert or pathic...). The junior partner's reward must therefore be measured out in currencies other than pleasure, such as praise, assistance, gifts, or money.\n\nThis sounds to me at each point unnecessarily disapproving: in the marginalizing of intimacy in such a liaison, in the discrediting of the boy's desire, and in the assumption that genuine sexual pleasure is independent of factors such as praise and assistance. Halperin's emphases place him firmly within the dominant ideology, which prizes (purportedly) egalitarian relations. If, however, as I have suggested, age-disparate relationships are about loving versions of oneself\u2014who one is, who one was, who one would like to be\u2014they surely afford far-reaching potential for rewarding, interpersonal development.\n\nGiven their debt to Foucault, it is possible that Sedgwick and Halperin have been encouraged by a certain strand in his work to expect that models of sexuality will define, and be defined by, an epoch, characterized by distinct modes of thought, with change occurring through a sequence of large-scale epistemological shifts. Cultural materialists, drawing upon Raymond Williams, are more likely to stress uneven development, setting subordinate, residual, and emergent formations alongside dominant ones. These concepts allow that diverse models may be in play at any given time; we may identify one as dominant and another as subordinate or emergent.\n\nAge hierarchy must be profoundly embarrassing if even queer theorists want to distance themselves from it. Indeed, in today's metropolitan sex\/gender system, it is freighted with implications of immaturity, narcissism, effeminacy, pedophilia, exploitation, and humiliation. These disturbing factors are founded in relations between age and youth in society at large, where younger people are more often credited with sexual attractiveness, whereas older people often have more economic, political, and social resources. Age hierarchy therefore invites stigmatization as merely instrumental on both sides, in contrast to the reciprocity attributed to age-matched relations. The disrespect accorded age-disparate relations is evident, Simon LeVay and Elisabeth Nonas remark, in the freedom other gay men feel to court the younger partner, while age-matched couples are accorded respectful space, and a hint of the sanctity of marriage.\n\nIndeed, such is the dominance of the egalitarian model that age disparity may actually appear more defensible when it is instrumental. By this I mean that it is more acceptable (slightly) for a youngster to allow it to be supposed that he is using an older man to find his way around the scene, or because he gets taken to the opera, than it is for him to declare that he is attracted and devoted to an older person. Correspondingly, it may be more suitable for the older man to be perceived as exploitative\u2014on to a good thing\u2014than for him to appear to take the boy seriously as a partner. It can work, Edmund White's narrator avers in _The Married Man_ , if you get the chemistry right. Austin, unfortunately, \"was incapable of picking out the talented tenth, the blessed exception, that nearly unique boy who admired experience and accomplishment more than an uncreased face and a tympanum-tight tummy. Nor could he spot that one guy in a hundred who was age-blind and didn't judge another man as a commodity.\"\n\nThis chapter dwells upon the impediments in age difference; my habitual mode is critical, and inclined to discover difficulties. Notwithstanding, we have many invocations of the joys of age-disparate liaisons. Some of the deceptively diffident poems of Gregory Woods celebrate such liaisons, appealing sometimes to the gay tradition\u2014Orpheus, Alexander, Whitman, Wilde, Rimbaud, Henry James. This untitled, twelve-line poem finds the poet on the island of Sirmio (I deduce), where the ancient Roman lyricist had a villa:\n\nUnder where Catullus\n\ntoyed with reality\n\nin his cushioned saloons\n\na decorum of very\n\nreasonable cabins\n\noversees the bathers.\n\nBut we beyond the rocks,\n\na slippery broker\n\nof boyhood and I\n\nwith water up trunks down\n\nto our knees, negotiate\n\nthe space between our ages.\n\nCatullus's ancient, luxurious villa was above; the modern, decent cabins are below, as if policing the beach scene. But the poet and his boy have located a third place, beyond the rocks. The imagery has a financial edge; perhaps money is being discussed. Yet the scenario is not primarily about money; the boy is slippery because of his wet skin, not dishonesty, and their trunks are down, not their futures. It is age disparity and sexual attraction that are being negotiated, as they once were by Catullus and his boys. One way or another, we will find a space.\n\n**ADJUSTMENTS**\n\nI discussed Paul Monette's novel _Halfway Home_ in chapter 5. Tom, the narrator, forges a great relationship with his nephew, Daniel, a classic desire-to-be: \"I noticed how he set his pace to mine as we walked across to the store. I would have done anything for him just then, for he made me feel like I was a fellow to be emulated, as he studied his way to becoming a man\" (139). Monette makes the boy as young as he can, given the kind of sensitivity and awareness that he wants him to have, and thus unavailable for sexual advances (he is seven). Nonetheless, Tom is fearful that he will be perceived as corrupting Daniel. He comes upon him doing a jigsaw puzzle he has found in the house: the picture is of Michelangelo's statue of David. Tom is terrified that Daniel's parents might enter and suppose that it was his idea. (Indeed, Susan is very suspicious and aggressive.) \"Instantly I knew, sitting like a giant beside this little boy, what I was really afraid of. That Daniel would turn out gay, and they would blame me and curse my infected ghost.\" And this was \"the old self-hatred\": \"Because what I really meant was that I didn't _want_ him to be gay, to run that gauntlet of misery and solitude. Where the hell was all my pride that had marched in a hundred parades?\" Meanwhile Tom, against his expectations, reorients his own age allegiance. He falls in love with Gray, the older man who has been quietly protecting him. \"His being fifty had no downside; he was simply a full-grown man. And lying there lazily under the comforter, I took the most wanton joy in being the younger one\" (156).\n\nIn Jack Dickson's Glasgow thriller, _Oddfellows_ , Joe (aged thirty-one) is in turn the younger and the older partner. He feels indebted to Billy, who gave him a job and a place to live when he was thrown out of the army. Billy is a club owner, engaged in endless negotiations with other entrepreneurs, and with the police (over drugs and murder). He demands brutal sexual episodes, but this power play suits Joe, whose fantasy investments center upon being humiliated and beaten in the army. He is less happy when Billy behaves as if he owns him. He gives him an expensive watch, and insists that he give up work, and hence his financial independence; \"The gold strap on his wrist suddenly felt very tight.\" Billy orders Joe around and strikes him in public. Even more distressing is a growing awareness that Billy is violently abusing youngsters. Joe leaves, but Billy doesn't like being thwarted and the outcome is violent.\n\nJoe is specially sensitive to the treatment of boys because of his strong feeling for his nephew, Sean, who is fifteen\u2014even as Joe was drawn to Sean's father (now dead). It becomes clear that Sean is gay and devoted to Joe, while Joe finds himself moved sexually by Sean's boyish body. Eventually, in the final episode of the book, they declare their love. However, they cannot pursue the relationship, Joe insists. They are too close; Sean has to sort things out for himself; he needs another kid to explore himself with; Joe would be holding him back. The combination of brother, father, and lover all rolled into one is \"An ideal, a fantasy... not made for the real world, a world in which Sean had to learn to function\" (301; Dickson's pause).\n\nJoe tells himself that Sean deserves the truth, but he lies when he tells him that he doesn't want to be the boy's lover: \"'Ah'll be here fur ye, but ah'll no' be part o' yer life... ah don't want tae be.' Amongst all the lies he'd ever told himself, that was the hardest\" (Dickson's pause). We understand that Joe is disguising his feelings and putting Sean off for his own good. \"From somewhere deep inside Joe found the strength to say what he knew he had to\" (302). Why does he have to? Because of normative assumptions about youth, maturity, and their proper development; whether these matters might be different for gay boys is not considered. Joe accedes to conventional notions of equality and manhood. He is going to move in with Andy, who is of a similar age and background: \"'maybe mates ur the maist important thing. Maybe ah'll love him as a mate, maybe as somewan ah want tae spend the resta ma life wi' '\" (302). Also, Joe confirms his rejection of Sean by an appeal to manliness: he asks him whether he is a wee boy with a crush on his uncle, or a man? \"As one pair of blue eyes stared deep into another, silent agreement passed between two men. Men. Not a kid and a man\" (302\u2013303). Despite this sudden maturation, Sean remains off limits for sexual love.\n\nWith Sean as well as with Billy, Joe has to reject the hierarchical option\u2014the one because it is too close, the other because it betrays any possibility of closeness\u2014despite passionate involvement. Or perhaps because of it. There is, after all, a tradition of gay renunciation. In _The Well of Loneliness_ (1928), Stephen introduces Mary to lesbian experience and subculture, but finally pushes her out into marriage. In Mary Renault's _The Charioteer_ (1953), Laurie has to reject the innocent Andrew in order to allow him the opportunity to grow out of his queerness.\n\nThe dangers of age-disparate liaisons are stressed again in Simon Lovat's psychological thriller, _Disorder and Chaos_. Here gays generally have a hard time, except the couple, Derek and Bob. They wear each other's clothes, or identical denim outfits; Keith calls them \"The Bookends.\" He \"notices how similar they look now. Twin haircuts, twin mustaches, twin forced jollity in their glinting eyes. Or is it simply that he's forgotten how much they operate as a two-cylinder machine, now that he sees so much less of them?\" \"'We don't take the clone thing seriously,' says Bob\" (25). The Bookends admonish Keith about his liking for youngsters: \"'These age gaps are a disaster area.... What you need is someone who has finished growing up. Someone who threw away their L plates years ago'\" (15; my elision). Keith does come to accept that his liaisons follow a doomed pattern, as his dependence upon Nick, a disturbed, devious, and dishonest sixteen-year-old, degenerates. Buying clothes, for instance: \"'I'm not a doll, or a kid,' Nick had said, flying into an instant rage. 'It'd make me feel under control, under your thumb'\" (133). Keith's infatuation leads him to prison. Upon release, he appears to forsake any sex life, as he generates a new obsession, with the infant whom he believes to be his son (he has donated sperm to a lesbian couple). He decides to abduct the boy; he will be very gentle with him, \"take care of him, spoil him, love him\" (225). As in the treatments of Monette and Maupin, there is an ominous though unspecified overlap between sexual and filial emotions.\n\nMeanwhile Lenny, who is married, has been suppressing his gayness. He is helped along by an experienced man, though he remains cowed by his senior partner in dentistry. He uses contact advertisements; most of the replies are from married men, and Lenny is inclined to reject them; but he too is married, so what can he expect? Like Owen in _The Lost Language of Cranes_ , he sensibly looks for someone like himself. Lenny is suddenly freed from conventional obligations by the strange behavior of his daughter, Monica. She proves highly tolerant of his gayness, but this is perhaps part of her general weirdness. She is into black magic and, it transpires, has a fatal fixation upon serial killer Myra Hindley. Lenny's concluding reflections apparently point the moral: \"subterfuge, half-truths, and lies\" create \"disorder and chaos\" (249)\u2014referring to the title of the book. On this criterion, only Derek and Bob match up. It's a dangerous world, Lovat seems to be saying, and other kinds of liaison are asking for trouble.\n\n**MENTORING**\n\nThe social system wants its young people socialized, so that they can contribute to the workforce and the rearing of the next generation. For the most part, though, it does not want them socialized into gay subculture. Even for the ancient Greeks, Foucault points out, this was a tricky matter. \"Because if there were no problem, they would speak of this kind of love in the same terms as love between men and women. The problem was that they couldn't accept that a young boy who was supposed to become a free citizen could be dominated and used as an object for someone's pleasure.\" The interface between these two desiderata, which are often located in the same individuals and institutions, is probably more heavily policed since gay liberation than it was before.\n\nA gay man may justify to himself and others his sexual attentions to boys by dwelling upon the emotional and practical support he is providing. David Leeming in his biography of James Baldwin writes of him as obliged by a \"puritanical streak\" to \"deny the merely carnal\" by placing himself as \"a father figure, financial sponsor, and teacher for a much younger individual.\" So with a North African \"street boy\" Baldwin insisted almost immediately on \"formalizing the relationship on somewhat paternalistic grounds, by meeting the boy's family.\" Compare _The Swimming-Pool Library_ , where Charles maintains a paternalist idea of his role as colonial administrator. Will is less convinced: \"I wanted to save Arthur. At least, I think that's what I wanted to do to him. It was a strange conviction I had, that I could somehow make these boys' lives better, as by a kind of patronage\u2014especially as it never worked out that way.\" Keith in Lovat's _Disorder and Chaos_ tells himself that he is helping Nick to get himself together. \"He knows it's a delusion but it holds up, providing he does not scrutinize it too much.\"\n\nUncles and nephews, I notice, are a recurring feature. Here the man has mentoring status already, perhaps along with some incestuous thrill. They occur in _Halfway Home_ , _Oddfellows_ , and _Wasted_ by Aiden Shaw; Edmund in _The Farewell Symphony_ gets to look after his attractive nephew, though he's not gay. In Larry Kramer's _Faggots_ , Richard has an anxiety attack when he is accosted in a club by his nephew, Wyatt. They compare sizes\u2014Wyatt is well-endowed\u2014they hasten \"to join each other in family togetherness.\" Wyatt should be \"'overcome with Jewish guilt,'\" Richard objects. But Wyatt thinks Richard \"'really should get some help.'\" Again: when Luke leaves home and comes to London to be gay, he naturally calls on his gay uncle Martin\u2014so _This Island's Mine_ by Philip Osment. However, in this virtuous Gay Sweatshop theater company play there is no embarrassing familial romance: it is assumed that uncle and nephew will each find a partner of his own age.\n\nA gay man may be confident enough to combine his official and his subcultural mentoring. George in Christopher Isherwood's novel _A Single Man_ is coming to terms with the death of his partner, Jim, in a traffic accident. The emphasis is on their equal partnership, but we may deduce that Jim was the younger, as Isherwood's lovers always were (\"he treats his exclusive interest in very young men as entirely natural,\" Paul Robinson remarks). Now George is feeling old. In Tennyson's poem \"Tithonus,\" which he expounds to his students, he is both Eos, the lover of boys, and Tithonus, the repulsively aging former boy.\n\nNonetheless, George is looking around for a new partner. Kenny, one of his students, seems to have conceived a special interest in him. In a bar George experiences with Kenny something like a Platonic dialogue, a \"symbolic encounter\"; Kenny appears \"beautiful. _Radiant with rapport_.\" Kenny likes authority; he thinks respect and friendship are most likely between males of different ages; it pleases him to call George \"Sir\" (133). They swim in the ocean and the rapport becomes physical. Moving indoors, to George's home, calls forth a return of responsibility: he refuses Kenny's invitation to shower together. However, the conversation becomes\n\npositively flirty, on both sides. Kenny's blanket, under the relaxing influence of the talk and beer, has slipped, baring an arm and a shoulder and turning itself into a classical Greek garment, the chlamys worn by a young disciple\u2014the favourite, surely\u2014of some philosopher. At this moment he is utterly, dangerously charming. (143)\n\nInveighing against a society which prefers \"flirtation instead of fucking,\" George finds himself uttering a sexual proposition. Kenny \"grins, dazzlingly,\" but George passes out (150). We will never know what the enigmatic Kenny might have done.\n\nThe Platonic dialogues which George invokes are a characteristic focus for the official and sexual roles of the mentor. Socrates' strategy is to tease the boy with talk of love, while maintaining the idea that the teacher is on a higher plane, above physical expression. Even if the boy is keen, as Alcibiades is in the _Symposium_ , it may be beneath the Socratic teacher's dignity to respond. Plato has been a conduit for same-sex passion and, simultaneously, a way of disavowing any such concern. Platonic ideas of intense friendship were made available in the Renaissance by Marsilio Ficino, revived in the eighteenth century by Johann Winckelmann, and deployed in the formative stages of modern gay self-construction by Walter Pater. These are the terms for Wilde's famous speech from the dock on the love that dare not speak its name: it is David and Jonathan, Plato, Michelangelo, and Shakespeare; \"It is that deep, spiritual affection that is as pure as it is perfect.... It is beautiful, it is fine, it is the noblest form of affection.\" Unfortunately for Wilde, this exalted vision seemed not to embrace the casual sexual and financial liaisons of which he was accused.\n\nIn _Death in Venice_ , Aschenbach's growing infatuation with Tadzio is accompanied by quotations from the _Phaedrus_ , which keep floating into Aschenbach's head. As Jonathan Dollimore shows, Mann is on Freudian territory: the ambivalent innocence of the Socratic dialogue is exposed as sublimation. At a key moment in Mary Renault's novel _The Charioteer_ , Laurie approaches and then evades self-disclosure when he shows the innocent Andrew his copy of Plato's _Phaedrus_. \"'I haven't read this one,'\" Andrew says. \"'I thought it was the _Phaedo_ for a minute, we did that at school.'\" The _Phaedo_ was a safe text, the _Phaedrus_ not. \"'What's it about?'\" Andrew asks, allowing Laurie a second chance to reveal or conceal himself: \"'Well, primarily, it's about the laws of rhetoric.'\" There were acceptable and unacceptable Platonic texts, and acceptable and unacceptable ways of addressing them.\n\nA striking deployment of the ambiguity of Plato is effected by Allan Bloom, whose best-selling book _The Closing of the American Mind_ chimed in with a reactionary turn in literary and cultural education in the United States in the late 1980s. Bloom's distinctive pitch is a yearning for a relationship between the teacher and student such as he believes Socrates shared with the young male aristocrats of Athens. Since the 1960s, he believes, sexual relations have been routinized and corrupted by Freudianism, the decline of the leisure class, and the women's and gay liberation movements. Socrates had the right idea: \"The longing for his conversations with which he infected his companions, and which was intensified after his death and has endured throughout the centuries, proved him to have been both the neediest and most grasping of lovers, and the richest and most giving of beloveds.\" Such relationships work best, in Bloom's account, with upper-class students\u2014\"they have money and hence leisure and can appreciate the beautiful and useless\" (279).\n\nSurprisingly (it is the surprising part of the book), Bloom allows the sexual teasing to become manifest. The survival of Socratic relations is most likely with students who\n\nhave not settled the sexual problem, who are still young, even look young for their age.... A youngster whose sexual longings consciously or unconsciously inform his studies has a very different set of experiences from one in whom such motives are not active. A trip to Florence or to Athens is one thing for a young man who hopes to meet his Beatrice on the Ponte Santa Trinit\u00e0 or his Socrates in the Agora, and quite another for one who goes without such aching need. (134; my elision)\n\nIn other words, the unformed young man is likely to fall either for a nice young lady from Smith, or for his male professor in the classroom. Bloom again courts sexual interpretation when he writes that his students \"wanted to find out what happened to Glaucon during his wonderful night with Socrates\" (332). Perhaps we should go back to sublimation, a \"naive and good-natured\" freshman suggests; \"I was charmed by the lad's candor but could not regard him as a serious candidate for culture\" (234). Bloom cannot get sexuality properly into, or out of, his classes.\n\nFurther light is thrown upon Bloom's stance in the novel which his old friend Saul Bellow based on his life and ideas, _Ravelstein_. The narrator is a journalist, straight; he has known Abe for a good while and Abe wants him to write his biography. The novel recalls meetings at which they discuss this, Abe's death from AIDS-related illness, and the narrator's own sickness and difficulty in writing. In the course of these lucubrations the narrator has in effect written the biography.\n\nAbe \"was considered, to use a term from the past, an invert. Not a 'gay.' He despised campy homosexuality and took a very low view of 'gay pride.'\" His apartment scarcely hints at his sexual preference: \"One had no reason, in any respect, to suspect him of irregularities of the commoner sort\u2014the outlandish seductive behaviors of old-fashioned gay men. He couldn't bear the fluttering of effeminate men\" (99). He is upset when a nurse may be overheard saying that it's time for his AZT. Nonetheless, \"He was doomed to die because of his irregular sexual ways,\" the narrator declares. \"About these he was entirely frank with me, with all his close friends\" (160). Not much of the sexuality gets into the novel, though; \"There were times when I simply didn't know what to make of his confidences,\" the narrator confesses (160).\n\nLike Bloom in _The Closing of the American Mind_ , Abe depends heavily on Aristophanes' speech in Plato's _Symposium_ , about how each of us pines for his or her other half, to restore an original complete whole (24). Whether Abe's maneuverings sometimes led to a sexual liaison is not indicated. His sexual feelings have increased, he tells the narrator, and \"'some of these kids have a singular sympathy with you'\" (143). The stronger impression, in Bellow's novel, is that Abe has romantic-paternal relations with his students while cultivating less exalted sexual relations in other quarters. When he needs a check made out to cash so that his companion will not know of it, the narrator supposes that this is a payoff for some sexual indiscretion (143).\n\nThere is a further force in Abe's life, one not avowed in _The Closing of the American Mind_ : Nikki, his \"companion\" (5). \"He would sometimes lower his voice in speaking of Nikki, to say that there was no intimacy between them. 'More father and son'\" (69); presumably this was not so initially. Nikki is from Singapore, in his early thirties and \"boyish still\" (5). The narrator is notably ill at ease around him.\n\nNobody questioned the strength of Nikki's attachment to Abe. Nikki was perfectly direct\u2014direct, by nature, a handsome, smooth-skinned, black-haired, Oriental, graceful, boyish man. He had an exotic conception of himself. I don't mean that he put on airs. He was never anything but natural. This prot\u00e9g\u00e9 of Abe's, I thought\u2014or used to think\u2014was somewhat spoiled. I was wrong, there, too. Brought up like a prince, yes. Even before the famous book that sold a million copies was written, Nikki was better dressed than the Prince of Wales. (68)\n\nNikki was attached, direct, and natural; but he did like to spend money. A suspicion is floated, that he is on the scrounge and Abe is something of a dupe. The narrator carefully distances himself from Abe's high opinion of Nikki's talents: \"in Nikki, Abe saw a brilliant young man who had every right to assert himself.\" As well as clothes, Nikki gets a BMW car: \"'He feels he should have something outstanding and entirely his own,'\" Abe explains; \"'It's only natural'\" (71, 74).\n\nThe gay reader need not share the narrator's prejudice against Nikki. The narrator's wife is also much younger\u2014one of Abe's former students, in fact\u2014but her credentials and her motives are not interrogated, and nor are the narrator's. Nikki returns from Geneva when Abe is ill, and abandons his studies to look after him. His devotion when Abe is in hospital is exemplary. We might recognize here a successful mentoring partnership, beyond the artificial teasing with the students.\n\nOf course, this is a notable example of the conflation of roles: Nikki's position is as much to do with his race as his age. \"He had his own kind of princely Asiatic mildness, but if you were to offend him Nikki would tear your head off,\" the narrator remarks (145). Damn'd inscrutable, these orientals; never know where you are with them. There is an arrogance and condescension in the attitude of the older men:\n\nNikki was training in a Swiss hotel school. I can't say more than that because I'm not the ideal person to recall the minute particulars but Nikki was an accredited ma\u00eetre d.' He was ready to go into fits of laughter when he modeled the cutaway coat of his trade for Abe and me, and put on his professional dignities. (18\u201319)\n\nNikki is expected to mock his own ambitions for the amusement of his backers; I suppose the cutaway coat allowed him to display his bottom. I don't know whether to be angry at this debasing of Nikki, or to rejoice in his resourcefulness. Yet perhaps his display is no more demeaning than Bloom cavorting intellectually for his upper-class students, who know that their father and the military-industrial complex are paying for all this.\n\nEve Sedgwick studied with Bloom, though there is no indication that he gave special attention even to such an exceptional woman. What Bloom has produced, Sedgwick says, is \"an ingenuously faithful and candid representation of... the stimulation and glamorization of the energies of male-male desire\" in teaching the humanities. It is an eloquent analysis of \"the prestige, magnetism, vulnerability, self-alienation, co-optability, and perhaps ultimately the potential for a certain defiance that adhere in the canonical culture of the closet.\" That is well put. However, we don't need to assume that the erotics of teaching can thrive only in the closet. The fully-out teacher is relegitimated with the fully-out student; everyone knows the score. Now the teacher may exchange glances with a grown-up boy or girl, without fearing or hoping that they are going to be seduced in ignorance, or against their will.\n\nAn air of desperation often lurks around mentoring in modern times, starting with Wilde and Bosie. Disappointment may be avoided at the price of death. Thomas Mann's _Death in Venice_ is retraced in Gilbert Adair's novel _Love and Death on Long Island_ (filmed by Richard Kwietniowski in 1997): the established author Giles De'Ath, a reclusive widower, becomes obsessed with a young actor in a trashy film. He tells his agent he is developing a new theme: the discovery of beauty where no one ever thought of looking for it. He dies. Lecturer Ivo is, in effect, killed twice when his affair with a student collapses in Barbara Vine's psychosexual thriller _No Night Is Too Long_ (adapted for BBC television with a screenplay by Kevin Elyot in 2002). In Isherwood's _A Single Man_ , mortality disrupts a potential relationship before its enigmatic promise can be explored. Patricia Duncker in _Hallucinating Foucault_ makes the charismatic great author die before the continuance of his liaison with a student can be tested. Edmund White in _The Married Man_ shows Austin's care for Julien, who is half his age, cut off by AIDS. The love affair between Tonio, a dancer, and Jack, a therapist, is tormented and energized by Tonio's impending death in the film _Alive and Kicking_ (Jim Sharman, 1996; screenplay by Martin Sherman).\n\n**THE MEN AND THE BOYS**\n\nThe strong presence of the coming-out story in gay fiction is very understandable: gay people are born into largely hostile social contexts, and finding their way to an alternative home base is rarely easy. Guides and initiators tend to be important. However, the structure of narcissistic complementarity especially invites a crossover between desire-for and desire-to-be. The boy's desire-for the man may well be involved with a desire-to-be the man, leading him to wish to _graduate out_ of boyhood. Eric in Baldwin's _Another Country_ is disconcerted when Yves says he wants to make his future in New York. \"'I can find my way. Do you really think that I want to be protected by you for ever?... it cannot go on for ever, I also am a _man_... my youth. It cannot last forever.'\" Eric \"knew what Yves meant and he knew that what Yves said was true.\" In many coming-out stories age hierarchy is presented as a tactical convenience. A satisfactory outcome usually implies the boy's release from such artificial supports, and his readiness for rewarding sexual relations on the egalitarian model. The mentor must, by definition, be cast off. One classic instance is Baldwin's _Giovanni's Room_ (1956): David finds himself at the expense of Giovanni's life. Another is Joe Orton and Kenneth Halliwell.\n\nIn Kenneth Martin's 1950s novel _Aubade_ , Paul, who has just finished high school, is pleased when he is taken up by Gary, who is a graduate from medical school, securely middle class, with a car. Paul's desire-for Gary is tied up with desire-to-be him: \"He's exactly the way I'd like to be when I'm his age, thought Paul.\" Nonetheless, when Gary declares his love and tries to kiss him Paul flees in terror. Eventually Paul acknowledges his own passion, but the summer is over and the guide must be discarded. Paul puts it all down to experience: \"'It was part of growing up, wasn't it?'\" (155). Gary has a definition of the homosexual which excludes Paul and himself:\n\n\"Do you know what homosexuality is? It's wanting to fiddle with every little boy you see. It's standing on the pier waiting for the next boatload of sailors to come in. It's giving women an inferiority complex. It's standing taking peeks in a man's toilet. I'm not like that, Paul. I love you, I love you.\" (128)\n\nSo no nasty queers need apply. The typical coming-out novel is strewn with repudiations not only of the mentor but of other \"bad\" types whom the boy brushes with but fortunately evades\u2014effeminate bitchy queens, fearful closet types, disgusting (tearoom) cottagers, often any organized gay scene at all. The emergence of Laurie in Renault's _The Charioteer_ depends on such exclusions. I showed in chapter 4 how Philip in _The Lost Language of Cranes_ rejects the closeted secrecy of his father, the old-style Wildean mannerisms of Derek and Geoffrey, pornographic movies, cruising, and bars with back rooms. Eventually he settles down with Brad, an old school friend who is of the same age, class, race, and educational background.\n\nSuch repudiations are sanctioned as necessary to the process of self-discovery, as the young man sheds the accretions of an oppressed history and steps out into the bright sunshine of an accomplished gay freedom. Charles, a minor character in Timothy Ireland's novel _Who Lies Inside_ , wears makeup and dresses flamboyantly in the local pub; he gets beaten up. This enables the emergent young Martin to feel revulsion and pity: \"He always gave me the creeps somehow,\" says our young hero. Charles fails Martin when he turns to him for help. Eventually Martin finds love with Richard, an attractive boy from his own class at school whom he has been enticing and rejecting since the start. Martin is still anxious about being \"homosexual,\" but Richard (who has himself benefited from experience with an older man) reassures him that there is no need for labels, since they are all persons.\n\nIn _Coming Out_ , an East German film directed by Heiner Carow (1988), Philipp, a schoolteacher, allows himself to be drawn into a relationship with Tanja, a colleague, and marries her when she becomes pregnant. However, he is in denial about an earlier queer episode, and has found his way to a gay bar (he was just wanting to buy cigarettes, he says). Pretending that he is unattached, Philipp allows a romantic gay boy, Matthias, to fall in love with him. Tanja and Matthias find out about each other and Philipp loses both of them. He is left to cruise an alienated scene. Asked by his mother why he must be this way, Philipp declares that it is nature, and he would be wrong to pretend and lie to himself. This thought is tardy, however, the harm to Matthias is already done. The film tends to suggest that gays must be lonely and unhappy, but it shows also that people should take more care with each other.\n\nThe selfishness of the neophyte is sharply analyzed again in Guy Willard's novel _Mirrors of Narcissus_ (discussed also in chapter 3). Guy has been dismissive of middle-aged men who pursue boys, but eventually he responds when propositioned by a well-preserved professor, Harry Golden:\n\nHe was old enough to be my father, yet I found him strangely attractive. Occasionally, when I thought about him, my feelings for him were distinctly sexual\u2014perhaps inevitable, given his intelligence and strong character. I'd always been drawn to dominant types. Though he was past his prime as far as looks were concerned, his personality almost made physical attractiveness seem unimportant. And there was the seductive thought of his power\u2014he was a full professor while I was just a freshman boy.\n\nThe disparity in their ages has advantages, Guy finds. It frees him from competitive impulses and shame, and from \"the need to adopt a tough, masculine facade. Instinctively I knew he wouldn't see my desire as a sign of weakness\" (160). In a typical conflation of roles, Guy reads his submission as feminine: \"Within the protective clasp of his big strong arms, pressed chest-to-chest in an intimate hug, I felt the stirrings of a tender submission. The secret little girl inside of me came alive and blossomed, gloriously\" (162). He feels \"exalted,\" but Golden is now redundant; he appears \"helpless and weak\" without his glasses, and has no further part in Guy's story (164). Thus fortified, Guy feels able to embrace his homosexuality, to abandon his girlfriend, and to force his attentions upon his friend.\n\nWhile the situations of lesbians are quite distinct\u2014for instance, there seems to be less of a premium on youth\u2014the thrills and spills of the coming-out process may be similar. In Jill Posener's classic Gay Sweatshop play, _Any Woman Can_ , predatory and stereotyped older lesbians are the problem for the bright and determined Ginny, rather than straight society. Rising from the ashes of her former selves, she discards the nervous and the exploitative and, through her own dramatic coming out, becomes an effective model for other women. Jeanette, in Jeanette Winterson's _Oranges Are Not the Only Fruit_ , is supported by several women in her contorted Christian community and responds with affection and respect. However, Miss Jewsbury, who has seduced her, is rejected. \"I don't know why I didn't thank her, or even say goodbye.\"\n\nIf boys ultimately have time on their side, men generally have control of representation. From Shakespeare's _Sonnets_ through J. R. Ackerley's _We Think the World of You_ (1960) to _The Swimming-Pool Library_ , we get the story predominantly from the viewpoint of the older man. This replicates the situation of the male suitor and the idealized female object of his attentions, in the classic heterosexual love lyric from Dante and Petrarch through to the present. A prime reason why we don't get the boy's story is that by the time he is in a position to write it, certainly publish it, he has become a man. The story he then writes, even if it is presented as if through the eyes of the youngster, is likely to manifest the viewpoint of experience.\n\nEven in fiction written ostensibly out of the perceptions of the younger person, we may in fact be getting an adult's fantasy. In P-P Hartnett's _I Want to Fuck You_ , Handa San (PE teacher) and Takeo (schoolboy, thirteen) are drawn sexually to each other. In one chapter Handa San's thoughts are indicated:\n\nHanda San wasn't the only teacher in the school who didn't want to think of himself as a _paedophile_ or _ephebophile_. He hoped the shivering attention he paid Takeo was just a phase he was going through. After all, he consoled himself, he didn't actually want to fuck the boy, get the boy on his knees or anything like that. He liked his students, was genuinely interested in their variety of character and outlook.\n\nThat is a plausible representation of Handa San's thought. Compare this passage, from a chapter focusing on young Takeo:\n\nTakeo dragged shirt and vest over his ears together and folded them as a complex on the back of his homework chair. These warm clothes gave off a fragrance many would buy if bottled. Over that same chair Takeo layered his socks, smoothed flat into two-dimensional neatness. Finally, he undid the belt of his trousers, wriggling out of them with extravagant movements of hips and behind. He slithered out of his pants exposing his total, subtle naked body to the paired full length mirrors on the inside of the wardrobe door. (54\u201355)\n\nThe first sentence there could be Takeo's account of himself; so could the third. But the sentence in between\u2014\"These warm clothes gave off a fragrance many would buy if bottled\"\u2014is the perception of experience, not innocence. The narrator invites a voyeuristic interest in how Takeo takes off his trousers; it is not only the mirrors that have Takeo under surveillance. In short, the narrator is close to Handa San.\n\nThe anonymously published novel _The Scarlet Pansy_ offers other examples. Again, it may appear at first sight that the viewpoint of the novice is represented, but actually the appeal is to the experienced reader. The young protagonist is revolted by the approaches of older homosexual men: \"Randall turned this episode over in his mind. Was he always to be pursued by some man? Then the only hope of escape was the cultivation of the utmost reserve.\" The narrator seems to have entered Randall's viewpoint. However, the passage continues: \"Quite unconsciously he was laying the foundation for one of the greatest charms of any person. He suppressed himself; a faint smile took the place of laughter, and thus he forever escaped that prevalent bane, the society grin.\" The author has the experienced view, and invites the reader to share it.\n\nThe appropriative power of these narrators mimics the personal advantage of the experienced man. If the boy's desire-for the man is destabilized by his desire-to-be the man, the feeling of the man for the boy may be complicated by a desire-to-be him. We may suspect, indeed, that interest in coming-out stories has less to do with succoring oppressed youth and asserting gay rights, than with the fantasies of older writers and readers, who may be moved by an unstable conflation of desire-for and desire-to-be such a boy. Edmund White has recognized this. In the preface to _A Boy's Own Story_ , he positions his present, writing self against his younger self: \"If I'd hated myself as a boy and adolescent, I now felt an affection for the miserable kid I'd once been, a retrospective kindliness one might call 'the pederasty of autobiography.'\" Again, in the novel, White's narrator finds he has come to like the fifteen-year-old he once was, \"even desire him\"; it is a \"retrospection three parts sentimental and one part erotic\" (158).\n\nIn _The Folding Star_ , Alan Hollinghurst makes a narrative virtue out of the young man's silence and distance. Edward, aged thirty-two, experiences a sexual infatuation with a mysterious youngster whom he is supposed to be teaching\u2014Luc, age seventeen. He spies upon Luc and his young friends; he systematically steals his underwear. \"'He thinks of me as a friend,'\" Edward avers. \"'How on earth would you know what he thinks. You haven't got a clue what goes on inside his head,'\" Luc's young friend Sibylle retorts (370). The reader is no better informed; Luc's behavior is surprising at every turn, and not fully explained even at the end of the book. The folding star of the title evokes Milton's \"Lycidas\": \"'It's when you know you've got to put the sheep all safely in the fold'\" (247). However (as in _The Swimming-Pool Library_ ), this pastoral impulse is accompanied, for Edward, by sexual pursuit: he is the wolf as much as the shepherd.\n\nThe local saint is Narcissus, though Edward says he doesn't \"' believe in the narcissist theory of gay attraction; I've always loved it with people who are different from me'\" (156). However, as I argued in chapter 2, in narcissistic desire a man loves not so much himself as an idealized displacement of himself\u2014what he is, what he was, what he would like to be, someone who was once part of himself. Edward acknowledges such an impulse when he wonders whether in placing no-longer-fashionable poems before Luc he is betraying an \"impulse to keep him back with me in a shared childhood\" (115). His determination to script Luc as having gay interests positions Luc as replicating Edward's own youthful sex life.\n\nThe central theme appears to be the fatal power of sexual fascination. As Edward's quest for the boy moves to a conclusion, so does the story of Paul, the museum curator. He recalls how as a boy during the German Occupation he had a liaison with a young man, not realizing that he was in the fascist militia. Thinking of Jewish children hidden away during the war years, Paul ruminates: \"'Personally I wouldn't want to place so much trust in a frightened or bereaved teenager\u2014but what could they do when it was their only chance?'\" (386). Again: a youngster \"'picks up an older person's life and then\u2014he is distracted, self-absorbed, over-zealous, or perhaps quite unreflecting, he's no idea what he's doing\u2014lets it drop'\" (414). Perilous creatures, these boys.\n\n**VISIONS AND ACCOMMODATIONS**\n\nI choose to close this chapter with positive views of its themes. In _A Fairly Honourable Defeat_ , published in 1970, Iris Murdoch explores some of the pitfalls that beset age hierarchy. Rupert and Hilda are married. Simon, Rupert's young brother, after sowing some wild oats, has settled into coupledom with Axel, Rupert's old friend. Simon is twenty-nine, Axel forty-two. In an initial, notably complacent conversation, Hilda displays casual liberal prejudices. Perhaps Simon is not really homosexual, queers are always a bit sly, they don't like being reminded of normal relationships or happy marriages, queer friendships are so unstable, Simon is so much younger. Rupert disputes such generalizations. Axel bullies Simon, Hilda thinks; \"'Some people like to be bullied,'\" Rupert responds; the novel is a study in the abuse of and retrieval of dominance and subordination. \"'Thank heavens our relationship is democratic,'\" Hilda declares\u2014complacently, as it transpires (15\u201318).\n\nSimon does look queer, he cheerfully acknowledges (198); he is in charge of interior decoration at their flat. It is out of character, friends agree, for him not to like opera because he's such a feminine person, as may be observed in his decor. Rupert and Hilda's son, Peter, jeers at Simon for being more female than Shakespeare's fairies. (Peter is very insecure.)\n\nAxel is straight-acting to the point of total inscrutability, even to Simon in their early meetings. He complains at Simon's aftershave lotion\u2014\"'Try to remember you're male, not female, will you?'\" Axel won't have any \"camp\" (in quotation marks) or _risqu\u00e9_ jokes (36). When they first drew together, Axel hesitated because he believed himself to be naturally monogamous, whereas Simon \"was by nature frivolous, inconstant, evasive, impulsive, irrational, shallow\" (202). Actually, Simon is seeking someone to give his heart to; \"Yet he enjoyed some of his adventures and liked the jokey parochial atmosphere of the gay bars which he had been used to frequent\" (37). Sometimes Simon himself fears he may be too trivial a person for Axel. His declarative outbursts only provoke Axel to withdraw. For Simon, \"There was at every moment total vulnerability. There was a dangerous thrilling trembling inner circuit of the soul\" (39). He suggests that they are like Apollo and Marsyas.\n\nIn fact Axel is chronically jealous and insecure; it is surprising that Simon doesn't quite realize this, and lose respect for him. He won't let Simon learn to drive, tries to stop him drinking and becoming exuberant at parties. He accuses Simon of being irrational, but enjoys putting him through the drama of accusation, reproach, and separation. Then he commands him to be independent. \"'You mustn't let me influence you so much, dear boy'\" (87).\n\nEnter Julius, another old friend of Rupert. He worked on biological warfare for the United States, but gave it up because he became bored. (Eventually it emerges that he was in Belsen; he considers himself an instrument of justice.) He denies that goodness is important: \"'we know what moves people, dear Rupert. Fears, passions of all kinds. The desire for power, for instance. Few questions are more important than: who is the boss?'\" \"'Though of course some people prefer to be bossed!'\" \"'Yes, yes. It's all a question of choosing one's technique'\" (225).\n\nJulius is a dedicated controller: he experiments with people and it gives him the pleasure of confirming his amoral principles. The other disturber of the peace is Morgan, Hilda's sister, who is even more dangerous because out of control. Julius makes a wager with Morgan that he can get Simon away from Axel. The \"fairly honourable defeat\" of the title is Julius's failure to part Axel and Simon while managing to maneuver Rupert and Morgan into an affair, with purloined letters and Iago-like insinuations.\n\nJulius traps Simon into moments of complicity and small lapses in total candor; he plays on Simon's fear that Axel will see him through Julius's eyes: \"Axel would suddenly see how flimsy Simon was, how unsophisticated, how lacking in cleverness and wit, how hopelessly ignorant about important things such as Mozart and truth functions and the balance of payments\" (77). Julius, alternately flirting and commanding, exploits Simon's tendency to be dominated\u2014effortlessly because, unlike Axel, he has little at stake. Simon is frightened by his attraction to Julius. \"But then, thought Simon, I have never really been able to distinguish between fear and sexual desire\" (160).\n\nJulius predicts a sorry end for Simon's partnership with Axel:\n\n\"You choose at present to give in. But every time you give in you notice it. Later perhaps you will make Axel's life a misery. Then gradually the balance will tilt. You will get tired of being Axel's lapdog. You are not at all monogamous really, my dear Simon. You miss your adventures, you know you do. And you will find out one day that you want to play Axel to some little Simon. The passage of time brings about these shifts automatically, especially in relationships of your kind.\" (269\u201370)\n\nThese are perhaps characteristic pitfalls of age hierarchy. However, the narration suggests that they need not be fatal.\n\nSimon eventually tells Axel all about Julius. Axel confesses that he was miserable when he thought he would have to end their affair; he admits that his reserve has kept Simon insecure, that he is guilty of a failure of love and trust. Simon appreciates that Axel does love and need him. \"'A little bullying between lovers needn't matter. But I've always withheld a bit of myself,'\" Axel admits (434). The important distinction, in Murdoch's view, is between fumbling good will and the destructiveness of Julius and Morgan. The pattern remains, but it is interpreted with trust. At the end Simon seems justified in feeling \"the warm anticipation of a new happiness\" (437).\n\nMurdoch's account is surprisingly modern; clones, AIDS, queer, and lifestyle lie between us and _A Fairly Honourable Defeat_ , but she comes through with a valuable analysis. Perhaps this is because she is not troubled by the age-disparate liaison as such: she takes its potential for granted, and is thus free to pursue wider and deeper psychological and ethical dimensions. Compare other mid-century fictions\u2014Stephen Spender's _The Temple_ , Angus Wilson's _Hemlock and After_ , Barbara Pym's _A Glass of Blessings_ , the film _Victim_ (Basil Dearden, 1961). In _We Think the World of You_ , Ackerley's protagonist does his best to keep things going with his boyfriend, but finds he gets on better with the dog.\n\nThe most striking affirmation of age-disparity occurs in Neil Bartlett's _Ready to Catch Him Should He Fall_. Bartlett doesn't shrink from awkward aspects of the theme. He invites readers to occupy the position of the man, and to share in the constitution of the ideal boy. Perhaps the narrator's description is wrong for some readers: \"do go back, and amend my description of Boy so that he is, is some way, if you see what I mean, your type. Make him fit the bill; imagine for him the attributes that you require.\" At the same time, the image will derive from and belong to the subculture. Boy is destined to discover himself in a historic gay identity, self-consciously bestowed by Madame, who owns The Bar, and her clientele. He needs \"application, study, repetition, diligent imitation and sincere admiration of his peers\" (33). He is the figure we (men) all hold in our imaginations. He is in a line with Chance Wayne played by Paul Newman, Alec Scudder in _Maurice_ , Boy Barrett in _Victim_ , Bosie Douglas\u2014drawn purposefully by Bartlett from several generations. The idea of Boy is a ratification of gay history, and hence of gay existence.\n\nThere is no question that Boy, in assuming this identity, is doing what he wants to do, though he is only nineteen (at the time of publication the gay age of consent was twenty-one in England). He is already exhausted with questing for The Bar; already imagining sexual practices that men might pursue together. He only ever goes home with older men. \"One thing Boy never said, the line of Paul] Newman's he would never have used, was _don't call me Boy_. He loved to be called Boy\" (13). It is a position of honor, not of inferiority. He wants one thing more than to be one of the men in The Bar: \"to be reassured that he might somehow remain a boy for ever\" (38). (I discuss the gendering of Boy and The Bar in [chapter 5.)\n\nO is The Older Man\u2014forty-five at least. No celibate, but very self-contained; \"you never saw him following anyone, gazing after someone or persuading them to come home with him. Asked exactly how O took people home, you'd have to say that O just summoned his men to him somehow\" (68). O's stance toward Boy is protective (hence the title of the novel), but sex between them is violent\u2014though Boy does bruise easily. \"I should say here that Boy never once wanted O to stop, and that he was used to sometimes being frightened by what O wanted to do, and by what he made Boy himself feel that he wanted to do, things he hadn't ever known that he wanted to do\" (141\u201342). As in other novels we have considered, and especially in Paul Russell's _Boys of Life_ (discussed in chapter 7) and Duncker's _Hallucinating Foucault_ , innovative and fierce power play in bed is prized by the boy as evidence of the man's general commitment to life and particular appreciation of the boy.\n\nWhat follows this intense courtship is less remarkable. Boy and O affirm their relationship in a version of the conventional marriage ceremony. Some of The Bar people are offended by this; it is the one point in the novel at which there is any subcultural dissent. Boy and O form \"The model couple in the model flat, Boy at home all day with the appliances and O out to work\" (225). For Murdoch, the key question is about how men adapt their fantasy desires and personae to the business of getting along together. This is admitted in _Ready to Catch Him_ : \"what really matters is what happens when two people try to hold things together\" (309). Boy \"spends the evening or night with a younger man sometimes now, a man younger than himself, a boy really\" (309). We are not told quite how this is managed. Is Boy's desire-to-be O translating into a desire-for a boy of his own? Will O meet a new Boy? Are the differences between O and Boy fading away, leaving an egalitarian couple?\n\nReviewing _Ready to Catch Him_ , Adam Mars-Jones criticized its endorsement of The Bar and its insistence on the special, poetic quality of gay sex; they correlate, Mars-Jones says, with a separatist as opposed to an assimilationist aesthetic. He contests the idea that \"it is by exhaustively exploring their fantasies that gay people best prepare themselves to take their place in the world.\" This is surely somewhat abrupt; it is hard to imagine a commentator writing in the _Independent_ that black people should not investigate and value their histories. Bartlett has justified the introspective stance of the novel as an attempt to consolidate gay subculture in a context where AIDS was taking the lives of friends and legitimating homophobic assaults. \"You couldn't walk down the street or open a newspaper without flinching, because there would be some new graffito about AIDS\u2014on the wall or as a headline in the best selling newspaper.\" This is not all, however. Although we prefer to regard our fantasies as private, the public iconography of art, advertising, and pornography demonstrates conclusively that they are acquired through a collaborative process. The Bar is a metaphor for the communal mechanisms through which we negotiate desire-for and desire-to-be. \"I apologize,\" says the narrator, \"if this description of Boy sounds to you like some fantasy and not a real person\" (15). Bartlett's aesthetic is all about peak experiences, moments at which fantasy miraculously catches up with actuality. Boy is indeed a fantasy, a luminous evocation of all the boys who have been loved since Antinous enchanted the Emperor Hadrian in A.D. 120.\n**7**\n\n**CLASS**\n\nDifference in disposable assets is happily acknowledged in the Pet Shop Boys' \"Rent.\" The song is offered from the point of view of a kept boy. Financial status is not effaced: \"You bring me food, I need it, you give me love, I feed it.\" To the contrary, it is claimed as integral with love:\n\nNow look at the two of us, in sympathy with everything we see\n\nI never want anything, it's easy, you buy whatever I need\n\nLook at my hopes, look at my dreams, the currency we've spent\n\nI love you, you pay my rent. I love you, you pay my rent.\n\nTheir sympathy is founded in shared consumption, rendered the more passionate and pleasurable by the fact that there is no hesitation about commitment. \"I love you. \/ You pay the rent\": no syntax. It is not \"I love you, therefore you pay the rent.\" Nor is it \"I love you because you pay the rent.\" The personal feeling and the provision of security occur together. The currency they have spent is both money and psychic investment. The liaison works; as well as sympathy there is \"sometimes ecstasy.\"\n\nOne of Eve Sedgwick's pregnant remarks is that many dimensions of sexual choice appear not to have a \"distinctive, explicit definitional connection with gender; indeed, some dimensions of sexuality might be tied, not to gender, but _instead_ to differences or similarities of class or race.\" Sex, in other words, may be organized around hierarchies other than gender; \"to assume the distinctiveness of the _intimacy_ between sexuality and gender might well risk assuming too much about the definitional _separability_ of either of them from determinations of, say, class or race.\" In earlier chapters I located the cross-class liaison as a version of the complementarity model, in which sameness of gender is complicated by other differences. Our experience of class evokes intense idealizations and enduring humiliations. I am taking \"class\" approximately, as comprising hierarchies of wealth, income, status, educational attainment, and cultural sophistication, along with their markers in attire, decor, and general lifestyle. While I disagree with Trotskyists, who proclaim the ineluctable priority of class struggle, I believe that class difference is everywhere in our psychic lives, as it is in our social system.\n\n**TURNING ON TO CLASS**\n\nThe Wilde trials, I and others have suggested, were crucial in establishing the stereotype of the queer man which dominated until gay liberation in the 1970s. At the trials, the entire, vaguely disconcerting nexus of effeminacy, leisure, idleness, immorality, luxury, insouciance, decadence, and aestheticism, which Wilde was perceived, variously, as instantiating, was transformed into a brilliantly precise image. The principal twentieth-century stereotype entered our cultures: not just the homosexual, as the lawyers and medics would have it, but the queer. A comparable effect was produced for lesbianism by the prosecution of Radclyffe Hall's _Well of Loneliness_ in 1929.\n\nWilde's effeminate manner was linked as much to class as to gender. Already by the end of the eighteenth century, the aristocracy was positioned as feminized. The newly dominant middle class justified itself through an ideology of work, manly purity, purpose, and responsibility. The leisure-class male was identified, correspondingly, with effeminate idleness and immorality; his options were to repudiate this identification, or to embrace it. Wilde affected a feminine stance in order to claim a class position, while exercising the male authority of accomplishment in public life; the combination of these strategies evidently made him impressive to the boys whose acquaintance he cultivated.\n\nThe Wildean, cultured gent and his bit of rough trade became the dominant image of the queer for the twentieth century, up until the 1970s. It was less an individual experience than a subcultural myth. As Foucault puts it, \"it was in the 'bourgeois' or 'aristocratic' family\" that sexuality was \"first problematized,\" whereas \"the working classes managed for a long time to escape the deployment of 'sexuality.'\" He disputes the idea that the surveillance of sexuality was inflicted upon the lower orders by the ruling classes: \"Rather it appears to me that they first tried it on themselves.\" The lower-class partner might be presented as a secretary or manservant; John Addington Symonds did this, so did Somerset Maugham and N\u00f6el Coward. Otherwise, it was difficult for two men to live together; Terence Rattigan installed his lovers in nearby apartments. (Of course, that cost money.) When she lectured in the United States, Gertrude Stein presented Alice B. Toklas as her secretary.\n\nIn the United States also, it was upper-class men who first got the idea of \"being a homosexual.\" George Chauncey Jr. shows that working-class men, particularly those linked to \"masculine\" milieux\u2014sailors, laborers, hoboes, and other transient workers\u2014might engage in same-sex activity across class without having to categorize themselves as \"queer.\" After all, they might be, or might appear to be, motivated largely by social deference and financial advantage; probably they were married. As Murray Healy observes, this is not to say that there were no working-class pubs, cruising grounds, or gay identity; rather that these resources were rare, and other options more available, more visible, more attractive.\n\nThis way of regarding cross-class relations was not confined to homosexuals: to a striking extent it replicated wider class and sex\/gender patterns. It was in practice almost acceptable for an upper-class man to have as a mistress, or to have casual sex with, a female of a lower class (typically a servant, a shopgirl, or a secretary), or to employ a sex-worker; it was almost expected. What he was not supposed to do was foul up his own social stratum by forming extramarital liaisons with women whom he might meet there. Freud actually imagined this as a universal trait. In an essay \"On the Universal Tendency to Debasement in the Sphere of Love,\" he posits that a man\u2014any man\u2014has difficulty in combining \"the _affectionate_ and the _sensual_ current.\" He can't satisfy his desires with a woman he respects; hence \"his need for a debased sexual object, a woman who is ethically inferior, to whom he need attribute no aesthetic scruples, who does not know him in his other social relations and cannot judge him in them\" (254).\n\nThe cross-class queer liaison worked similarly. It is often remarked that some of the objection to Wilde's behavior was that he had crossed class barriers. For example, he was questioned about Alphonse Conway: did Wilde buy him a suit \"In order that he might look more like an equal?\" The furor over _Lady Chatterley's Lover_ was similarly framed. Famously, prosecuting counsel asked the jury whether they would want their wives or servants to read the book. Actually, the offense was not that Wilde had cultivated cross-class liaisons, but that he had openly paraded them\u2014as Lawrence did in fiction. The more disgraceful connection was between Wilde and Douglas; this was too sensitive to be addressed in court. The crucial opposition was not between heterosexuality and homosexuality, then, but between legitimate and illegitimate relations, defined in terms of class.\n\nIt is clear that, for many people who expressed their sexualities in this way, the cross-class liaison was not just a convenience: it was a turn-on. For many middle- and upper-class men, lower-class people were sexy _as such_. For most middle-class men, after all, servants had tendered the principal physical, affective, and intimate support in infancy. In the 1850s, Arthur Munby, a lawyer and civil servant, established a liaison with Hannah Cullwick, a servant; the impetus for each of them is clear in their diaries. They developed his lustful overseeing of her, scrubbing the floor on her knees, into a scenario in which she got herself dirty and undertook rough, slavish tasks and manners for his pleasure. Peter Stallybrass and Allon White comment: \"The opposition of working-class maid and upper-class male, then, depended upon a physical and social separation which was constitutive of desire.\" As they note, Freud's \"Wolfman\" retrieves an intense early experience involving a maid in a scrubbing posture. Hannah was not just a figure of \"lowness,\" however: she was also a figure of comfort and power. Her diaries indicate clearly that she is entirely happy with her subordinate position. She is both glad and sorry when Munby says that he almost tells a friend about them: \"glad' cause it show'd that M. does love me, & sorry' cause I don't want to disgrace him & I canna bear for anyone to think I want to be anything but what I am to him. And so I want no one to know.\" She is not aspiring to change class; she likes her work and is proud of it: \"But tho' I'm never so happy as when I'm with him or working for _him_ , yet I want to be still a servant & working so as to be independent & get my own living.\" At Munby's behest they do marry, but Hannah feels it has \"little to do with our _love_ & our union\" (252; Cullwick's emphases).\n\nThe mysteries of lower-class life may hold a fascination for the middle-class man. Crime thrillers, from _The Heart in Exile_ by Rodney Garland (1953) through _Skinflick_ by Joseph Hansen (1979) to _Doing Business_ by Jeremy Beadle (1990), have depended on the middle-class man being sucked into a mysterious underworld of rent boys, hustlers, and beach bums. He mediates the lower orders to presumptively middle-class readers.\n\nThe happy ending to Garland's _Heart in Exile_ is secured when the protagonist, a psychiatrist, finds after a period of loneliness and gloomy exploration of queer subculture that he is drawn to his manservant.\n\nI confess that the attraction was much stronger when I saw him doing the sort of work I would never had dreamed of asking him to do. When my charwoman left, he insisted on scrubbing the kitchen floor, kneeling on the rubber mat, bending over the mop in his singlet. One saw the servant's humility in the attitude. But one also saw the broad shoulders, the arched back with the freckled skin under the rebellious hair, and he would look up as I entered and give me a beautiful smile of his brown dog eyes and white teeth.\n\nThe excitement here appears to arise from a coalescence of subordination and strength. On the one hand, Jeffrey Weeks comments, we are seeing \"a form of sexual colonialism, a view of the lower classes as a source of 'trade.' On the other we may have a sentimental rejection of one's own class values and a belief in reconciliation through sexual contact.\"\n\nStephen Spender tells in his autobiography how in the mid-1930s he took up with a young man whom he calls Jimmy Younger: \"I asked him to live in my flat and work for me.\" The contrast in their background, Spender more or less admits, was not just an inconvenience; it was exciting:\n\nFor the differences of class and interest between Jimmy and me certainly did provide some element of mystery which corresponded almost to a difference of sex. I was in love, as it were, with his background, his soldiering, his working-class home. Nothing moved me more than to hear him tell stories of the Cardiff streets, of Tiger Bay.... At such moments, too, I was very close to certain emotions awakened in childhood by the workers, who to us seemed at the same time coarse, unclean, and yet with something about them of forbidden fruit, and also of warm-heartedness which suddenly flashed across the cold gulf of class, secret and unspoken. (158\u201359; my elision)\n\nThere were tensions, however. Jimmy \"was accustomed to be treated rough, and he expected that I would behave like his past employers. When I did not do so he was disconcerted and felt that in some way I was gaining power over him as no one had done before.\" He said, \"'You are very nice to me, but I feel that I am becoming your property'\" (151).\n\nAs in the Pet Shop Boys' \"Rent,\" it seems plain that the lower-class person may gain more than a meal ticket: there is a romance about the affluent, an aura about the powerful, that may make them sexy. However, indignation and resentment are equally likely: a role reversal (such as I discussed in chapter 3) might place the lower-class man in control. That is the theme of Robin Maugham's novella, _The Servant_ , filmed by Joseph Losey in 1963.\n\nContemporary viewers are invited to recognize a liaison from the 1960s in John Maybury's film about the painter Francis Bacon, _Love Is the Devil_ (1998). George (Daniel Craig) arrives in Francis's studio through the skylight, as a burglar, looking notably proletarian in a donkey jacket. \"Come to bed, and you can have whatever you want,\" says Francis (Derek Jacobi). George's first question when we see them in bed is: \"You actually make money out of painting?\" He's pleased when Francis says he may use him as a subject; he allows Francis to buy him new clothes. George is ill at ease among Francis's posh-bohemian friends, however. They have common ground in boxing, but George's old East End friends despise his new connections. They warn him that he will be used and dropped. Sexually, Francis likes to be submissive, to relinquish control; in the relationship he holds all the cards (this is a typical reversal of class roles).\n\nFrancis gets bored with George, who, having no occupation or space of his own, becomes importunate, obstreperous, maudlin, suicidal. Francis tells himself that his work leaves him no room for relationships; that he is powerless to protect George from his dreams. He apologizes for him to his friends. His impatience and unkindness are defended as the artist's necessary response to his demons. Francis takes George as subject; the pictures in which he figures are especially admired, but Francis can't tolerate him in the studio. Friends warn him that George needs help, but he can't be bothered. George takes pills and alcohol while Francis is feted. George's body seems to grow and fade, miasmically, in and out of the artworks (now famous and immensely valuable) that he has made possible.\n\n**THE PERSISTENCE OF CLASS**\n\nIt is a pleasant trope of conservatives that class is no longer significant in our societies. At the same time, the gap between the rich and the poor is wider than ever in Britain and the United States.\n\nIt might be thought that the idealized relationship between two boys of the same age, who live in adjacent flats on a working-class estate, in a play designed explicitly to contribute to the campaign for an equal age of consent, would be free of class hierarchy. However, Jonathan Harvey in his play _Beautiful Thing_ makes Jamie's mother upwardly mobile (she is promoted to manage a pub) whereas Ste's father is definitely rough (drunken and violent). Jamie's care matches Ste's endurance. Further, in familiar ways, class correlates with gender: Ste is good at football, whereas Jamie reads about soap stars in his mother's magazine. (Note, however, that, unlike earlier coming-out narratives, _Beautiful Thing_ does give a positive role to gay institutions\u2014such as the magazine _Gay Times_ , and a gay pub.)\n\nWhile it seems appropriate to elaborate the concept of class so as to register social, educational, economic, and political inequalities, this need not entail any abandonment of a more traditional Marxist sense of class as about economic, political, and social control. In the last analysis, all power is about command over the means to life. If we have neglected that thought in metropolitan contexts, it is because we have been immensely fortunate in the second half of the twentieth century, to the point where many of us have stopped worrying about food and shelter, and centered our anxieties upon the attainment of a new update of our sound system. Nonetheless, the intense commitments that we call \"love\" may, ultimately, be intricately mediated versions of a will to survive, ontologically as well as materially. This may lead us into interpersonal opportunities which seem to afford a reassuring exercise of our own power. Equally, it may draw us into the orbit of people who appear to be powerful and may protect us.\n\nThe importance of class is indicated in an abiding image in gay pornography and popular fiction: the maintenance man who comes into your house to fix the plumbing and ends up in the shower with you. A thoughtful version of this scenario\u2014told from the viewpoint of the working man\u2014is offered by Jay Quinn in his story, \"The Kitchen Table.\" Phil regards himself as straight, and has been in prison; he is employed by Trace to work with him on his house. As they share their exertions, Phil finds he is increasingly drawn into sexual fantasies about Trace (they are the same age and both presumptively white). \"He noticed how his upper arms and thighs hardened from an office worker's slack fullness to the firmness suited to their new function.\" Hitherto, Phil has been hostile to gays who have approached him, but generally he takes things as they come; he is scarcely troubled by his attraction to Trace. They develop a physical rapport. Unusually for an American story, class difference is made explicit. When a building inspector is not satisfied with Phil's work, he nonetheless defends Phil against Trace. When it comes down to it, Trace observes, \"'A working man'll side with one of his own kind over somebody in a three-piece suit every time'\" (188). Despite his business habits of decision and command, a photograph of Trace and his deceased lover shows Trace as the one who is held, supportively, from behind: \"Trace was relaxed into the man's broad chest, held lovingly by his large hand on his bare stomach\" (199). This is evidently the role demarcation Trace and Phil will now assume. They have a relationship of mutual respect, in which Trace's bourgeois introversion is matched by Phil's work-a-day steadiness.\n\nTo be sure, class difference may be only relative. The English novel that seems designed to set aside traditional hierarchies, representing a partnership between two ordinary blokes, is Tom Wakefield's _Mates_. Cyril and Len meet up in the army in 1954 and stay together for a lifetime. Cyril is pleased that their fellow soldiers don't regard them as conventional queers. He is afraid they will somehow intuit his sexuality.\n\nThey nodded to him as they usually did, and he nodded back. It was a relief to find out that he didn't look any different. He turned to look at Len. His friend Len didn't lisp, wear scent or make-up. That's what he was supposed to do, according to all the newspapers of the day. So what if he did? Sod them, sod them all. No, not all of them. The tea-lady was all right.\n\nCyril doesn't envisage that he might himself wear scent; Len may be relatively feminine, but neither of them cultivates an upper-class, Wildean manner.\n\nYet even this relationship is framed in terms of class. The six weeks of basic training make them appear equal, but hierarchy reappears when they receive different postings because Cyril has good examination results whereas Len, though bookish, is unambitious. Cyril is more confident socially and sexually; he takes up with a local group of artistic and professional men; becomes a head teacher. Len remains a conscientious clerk and does the cooking and the shopping. Finally, Len is unable to assert himself when Cyril dies, and gets pushed out of their home by Cyril's businessman brother.\n\nClass may figure largely a matter of cultural capital. Austin in Edmund White's _The Married Man_ is a successful American writer living in Paris; as such he has both money (earned) and prestige. He is taken with Julien partly because of his status and aura as minor aristocracy. Austin writes about the furniture of upper-class families and as a boy daydreamed of finding that his mother or father was descended from Huguenot nobility. Apparently it is in such elevated circles that Austin can best be himself: \"Only in upper-class French life had Austin found the exact shade of inclusion he had craved for. Perhaps it was natural in a society where a king had been surrounded by cute boys, his _mignons_ , and in which the brother of Louis XIV, ' _Monsieur_ ,' had maintained an all-male shadow court.\" Julien entertains Austin with fantastic tales of his eccentric family; when he doesn't want to talk about something he will \"smile and turn his head slightly to one side with a sort of royal unreachability\" (129). Austin is charmed. After Julien's death, at the end of the book, Austin and the reader learn that all this was fantasy and bluff: Julien comes from a peasant family. Whether French people were deceived by him, or cared very much, is not disclosed.\n\nThe history, albeit dimly perceived in recent years, of preliberation liaisons as typically cross-class and exploitative, has led lesbian and gay people to repudiate or understate the influence of class in our affairs. Gender, race, and sexuality have seemed the more promising banners under which to unite. Sally Munt describes how commitment to lesbian feminism made her feel that talking about class \"was to be labelled a spoiler, a guilt-tripper, a Manichean thinker, a fifth columnist.\" As Munt adds, the occlusion of class has been more powerful in the United States, where Marxism has become hard to think about. A collection of essays reasserting the value of Marxism to Shakespeare studies begins by repudiating any supposed priority of class, as opposed to race and gender; the collection makes almost no further mention of class at all. The same thing happens in Ken Plummer's collection, _Modern Homosexualities_ , Leslie J. Moran points out.\n\nAn intriguing interaction between different kinds of capital is described by Carol M. Ward in her account of a two-year relationship between Rita Mae Brown and Martina Navratilova in 1979\u20131981. It might be thought that the tennis player possessed considerable financial and cultural capital. However, Brown was strongly aware of her own status as a successful novelist and intellectual. She saw Navratilova as \"a nice young girl in a limiting profession; where you can make a lot of money but know very little.\" Correspondingly, Navratilova felt that her career as a sportswoman was being held in low esteem. She wanted to learn from Brown\u2014to read books, visit museums, and talk about politics\u2014but not at the price of her own self-valuation. According to Navratilova, the relationship broke up because Brown's attitude undermined her ambitions in tennis: she became ambivalent about her own aims. Brown believes that the relationship foundered when Navratilova took up with basketball star Nancy Lieberman; Navratilova says Lieberman was only coaching her\u2014helping her to recover her will to win. After a violent breakup, the two women became friends.\n\nDorothy Allison's experience leads her to believe that class is highly active among Americans, in the constitution of both sexualities and political attitudes. She relates her masochistic desires, and her use of dildos and leather, to abuse by her stepfather, and to the white trash culture in which she grew up; she tells a fictionalized version of the story in _Bastard Out of Carolina_. \"What I know for sure is that class, gender, sexual preference, and prejudice\u2014racial, ethnic, and religious\u2014form an intricate lattice that restricts and shapes our lives, and that resistance to hatred is not a simple act,\" she says. Hence the hostility of middle-class feminists and lesbians toward her practices. They are uncomfortable with her butch, working-class partners: \"The kind of woman I am attracted to is invariably the kind of woman who embarrasses respectably middle-class, politically aware lesbian feminists.\" The task, Allison says, is \"to understand how we internalize the myths of our society even as we resist them.\"\n\nIn fact, probably because of general pressures toward embourgeoisement in personal relationships and the particular effects of the targeting of the pink economy, we have scarcely sought to imagine a subculture in which class would be truly a matter of indifference. Instead, we have complacently supposed that gay people will become, almost by definition, middle class\u2014in lifestyle and aspirations, if not in background and income.\n\nClass hierarchy may be obscured through processes of substitution and conflation which make it easy to read class as gender or age difference. Texts which I discuss in other chapters as structured by gender, age, and race usually include a significant element of class difference\u2014so Maupin, _The Night Listener_ ; Dickson, _Oddfellows_ ; Hall, _The Well of Loneliness_ ; Baker, _Tim and Pete_ ; Feinberg, _Stone Butch Blues_ ; Monette, _Halfway Home_ ; Hollinghurst, _The Swimming-Pool Library_. When Malone in Andrew Holleran's _Dancer from the Dance_ is living with Frankie, they begin to look like each other (they dress similarly), \"like all homosexual lovers\"\u2014except for \"that unmistakable difference\": race. (They are from Irish and Italian families.) \"When they lay tangled in each other's limbs by day or night, the pale, golden form, and the swarthy, dark-eyed, one, the northern and southern race joined at last. In fact what is clear, but unmarked, is that their class and educational backgrounds are entirely dissimilar\u2014Malone went to Yale Law School whereas Frankie labors, maintaining the subway system. This is far more relevant than race as such; when the romance wears off they have little to talk about, so they quarrel and part.\n\nClass is not secondary. If a gay man gets off on wearing the gear of a construction worker, or desiring someone else who does that, it is not helpful to read this as a gender phenomenon and translate it into his relations with his father. It is more reasonable to recognize that there is a historical figure\u2014the construction worker. If he appears sexy and dominant\u2014tough, highly skilled, and inured to danger\u2014it is because behind the fantasy lies an actuality in which there is hard, difficult, and perilous labor, probably worsened by stressful working conditions and management resistance to unionization. If he appears sexy and ready to serve, it is because he cannot afford to risk dismissal. It is in such a world that the middle-class gay man invests, financially and psychologically, in real estate and decor.\n\nThe corollary of the centrality of class in gay affairs has been intense evocations of man-to-man equality. The apostle of the liaison that transcends class, in the United States especially, has been Walt Whitman. David Bergman shows how the gay critic F. O. Matthiessen, drawing partly on Edward Carpenter, established Whitman as a poet of organic unity who believed that personal fulfillment might be continuous with social unity. Whitman's reputation as the model for an ideology of opportunity, democracy, and rights, crosses, perhaps too easily, into gay culture as an ideal of comradely, manly, sexual democracy. A memoir of Mark Bingham, for instance, presents him as a regular guy, a rugby player and business executive; a former lover recalls how, when Bingham was coming out, he would read him Whitman. The reason for the memoir is that Bingham died on the hijacked 9\/11 plane that missed its target. Surely he must have been one of the passengers who rushed the hijackers; attention to Whitman certifies his heroic potential. In fact, Whitman's adhesive partners were lower class and younger, and his life and work may be seen as displaying damaging hints of effeminacy.\n\nIn practice, democracy and freedom have immense difficulty evading the demands of wealth and beauty. Money was never absent from the elaborate cruising opportunities of the 1970s. The idea that a kind of democracy flourished on Fire Island is evoked but then given further twists in Ethan Mordden's story, \"And Eric Said He'd Come\":\n\nFor here we find gay stripped to its essentials. The beautiful are more fully exposed here, the trolls more cast out than anywhere else\u2014thus their pride and passion. The beguiling but often irrelevant data of talent and intelligence that can seem enticing in the city are internal contradictions in a place without an opera house or a library. Only money and charm count. Professional advantages are worthless, for, in a bathing suit, all men have the same vocation. Yet there are distinctions of rank. Those who rent are the proletariat, those who own houses are the bourgeoisie, and houseboys form the aristocracy.\n\nAt first it appears that only sex appeal counts; then we learn that money matters after all; indeed, the island has its own version of the class system.\n\nJames Kenneth Melson had a mixed experience of the prestigious Studio 54 disco, he says in his autobiography. He proved that a boy from Ohio could gain entrance just by looking good. Once inside, there was no problem: \"the atmosphere was devoid of the pretenses, the 'attitude,' that prevailed among those left standing outside. Everyone could let his hair down; royalty would dance with rock stars, Eurotrash with debutantes, and pro athletes with the likes of Disco Sally and Rollerena, two of the notables of the 'outrageous' category.\" Better still, this democratic atmosphere seemed to facilitate Melson's quest to seek out the most wealthy and influential men. However, they expected him to comply with their sexual demands, and attempts to pursue upper-class acquaintances beyond the club scene and the bedroom left him rapidly and comprehensively snubbed. He found qualifications and a job on Wall Street a better route to success.\n\nThe implicit underpinning for the Whitmanesque, gay egalitarian ideology is supplied by the concept of \"America.\" As Steven Epstein observes, the appeal is to \"the rules of the modern American pluralist myth, which portrays a harmonious competition among distinct social groups.\" On this basis, lesbians and gay men have constituted themselves as something like an ethnic group claiming rights. How far that pluralistic myth is to be trusted is a question far wider than queer politics: it is about how much we should expect from the institutions through which capitalism and heteropatriarchy are reproduced.\n\nPete in Baker's _Tim and Pete_ offers an explanation for and reaffirmation of the general idealism and the aspiration to transcend class that gay men have associated with a \"Whitmanic\" feeling. Pete relates it to \"'a lot of people feeling good about themselves for the first time in their lives. That was the best time, really, the early gay lib days. There was a bohemian spirit, you know. In a sense it was still the sixties.'\" The image may still be potent. Whitman would have been at home in the semimystical gatherings of the Radical Faeries in 1990, Paul Monette avers.\n\n**WHO SPEAKS?**\n\nPhil in \"The Kitchen Table\" and Melson in his autobiography tell their stories from the viewpoint of the relatively lower-class man. Generally, as we see elsewhere in this book, the feelings of the partner in the subordinate position are less well documented. For instance, we can only speculate about the subsequent sex lives of the boys who featured in Wilde's trials. The unusual factor in Munby's household is that we have Hannah Cullwick's diaries.\n\nNeal Drinnan's _Glove Puppet_ ingeniously refocuses the perils of the cross-class liaison. The novel is written from the viewpoint of Johnny, also called Vaslav, who is now twenty; he interprets his early life, promising the reader some lurid details. Johnny comes from a classically deprived background: he has never known his father, his mother is a sex worker and a drug addict. When she dies suddenly on a railway station he is scooped up by Shamash, a gay ballet dancer whose son has recently died, and taken to Australia. There, at the age of seven, he is named Vaslav and passed off as Shamash's son. He lives happily and luxuriously, and adopts the concepts and values of an artistic, bohemian community. From the age of eleven he finds that his feelings of filial affection for Shamash are complicated by an intense sexual awareness. Shamash tries to damp this down, but at age fourteen Vaslav encourages Ashley, Shamash's partner, to seduce him. When Vaslav is sixteen, he and Shamash become lovers.\n\nAs we saw with age hierarchy, purported accounts of sexual experience from the subordinate viewpoint may offer to titillate the reader, even while inviting condemnation on the ground of exploitation. This process is actually incorporated into _Glove Puppet_ , for Johnny\/Vaslav indicates that he expects to make a lot of money from his book. \"People love sex freaks, trash fucks, dirty young beauty; fresh filth-statutory rape-date rape, boy pussy surprise.\" The reader has been warned\u2014or is it enticed?\n\nAn adult having sex with a boy of fourteen is presented as the moral issue of _Glove Puppet_. Although it is clear that Vaslav was hell-bent on gaining sexual experience, he maintains consistently that Ashley was wrong to exploit him. However, another vein of thought in the book suggests that, in one way or another, Johnny was always bound to be bad, sooner or later and whoever he was with, because of his class origins. Hence his relatively early development: \"hormonally I developed early in that white trashy way that really was my genetic inheritance\" (72). His enjoyment of Shamash's lifestyle is complicated by his assumption that really he is Johnny, the rough boy on the make, bad by definition.\n\nHe [Johnny] was winking at me from where he slouched by the garbage dump in the council estate. In my mind's eye he was making lewd gestures just like those trashy boys in the porn movies, his hand outlining the erection in his torn jeans, his other hand fingering his mouth for saliva, for lube. He was mustering his strength, weighing his sex because that's all he had to sell, that was his only ticket out of the council estate and at that stage his only ambition. (115)\n\nThis enticing but threatening creature has only one object in view: a ticket out. Any signs of untainted love and trust are merely deceptive. Johnny \"could never be tamed or cultured. He was like his mother, hardly a person at all, just someone to do stuff to, something to fuck,\" like a boy in a pornographic magazine (80).\n\nWhen the narration of the novel catches up with present time, Vaslav is entrenched in sauna, pornographic, drug, and sex-worker scenes; he is depressed and has tried to take his own life; yet he is surviving. Shamash has confessed to sodomizing his own son and is in prison. This is on the supposition that Shamash is Vaslav's father; the reader knows better. The issue is not incest, nor even underage sex (in most metropolitan countries sex at sixteen is legal), but the destructive intrusion of the lower-class boy. The reader may or may not choose to accept Vaslav's repeated claims that the truth of gay life is revealed in his corrupt but thrilling story: \"We might frighten ourselves, us fags, but we are what we are and we do experience some extraordinary sensations in our endless pursuit of whatever it is we are looking for\u2014momentary oblivion or eternal rest\" (173).\n\nA powerful novel which takes seriously the potential and the predicament of being the lower-class and younger lover is Paul Russell's _Boys of Life_. Tony is seduced as a sixteen-year-old in Owen, Kentucky, by Carlos, a bohemian, avant-garde filmmaker who is passing through. Carlos is charismatic, worldly wise, and very good at sex; he gives Tony the whiskey he craves, and promises him a role in a film. Despite or because of his rural innocence, Tony already has same-sex experience. His initial response to Carlos is a combination of fascination and trepidation. Thereafter, all the sex is marvelous. Writing about the first fuck still breaks Tony up, ten years later: \"suddenly I was so upset about everything, I couldn't stand it. I started crying, sobbing like some crazy drunk to think how that's all gone, nothing like that's ever going to happen again and the only thing I can do is try and remember it.\" The sexual absorption appears reciprocal, though Carlos is not interested in being fucked. \"I guess you could say he was greedy with me, but I didn't care one bit once he started in on me\" (72). Carlos's exercise of power is sexy.\n\nSo, according to the narration, offered as Tony's retrospective account, it could not fairly be alleged that Tony was seduced into sexual experience which he did not want. In fact, the problem is the other way around: Tony is totally infatuated. In New York he spends the days waiting for Carlos to come home\u2014\"I'd hear him tramping up the stairs. He never knew how happy I was to see him come in that door\" (62). Nor has Carlos detained Tony from heterosexual experience. He doesn't think of himself initially as queer or gay, and he does look at girls in the street\u2014though they tend to be of boyish appearance. \"Maybe it was knowing somewhere inside me I really was a queer that made me look at those women the way I did; maybe I was saying good-bye to something, even though I didn't know that's what I was doing\" (64\u201365).\n\nWhen he is abandoned by Carlos, and bored with making movies and cruising the bars, Tony allows Monica (who is somewhat boyish) to fall in love with him, take him down to Tennessee, and marry him. So eventually he picks up where he might have been if he'd just stayed among his own class in Owen\u2014\"finally back on track\" (219). However, he doesn't find Monica or Tennessee sexy or interesting.\n\nAfter two months in New York, Tony intuits that Carlos has no real interest in his personal development. \"I grew up on Carlos right then\u2014not that I ever thought he was going to get me through stuff, but I do think I hooked up with him back in Owen because I could see I needed some kind of help. And now that fell apart\" (91). He accuses Carlos of using him. It is at this point that Carlos cements Tony's attachment by starting work on the movie with him as star. What hurts Tony crucially is the discovery that he is neither the first nor the last of Carlos's boys. Tony's dawning realization that he has been displaced and he and Carlos will never fuck together again is very painful to read. Carlos claims later that he intended to set Tony free: \"'I wanted you to learn that you wanted to go away.... I sent you away, I let you grow up'\" (295; my elision). However, it sounds like special pleading. Carlos does this with everyone, we learn; it is how he works as an artist. He needs to break his performers to get them to realize their potential for the camera.\n\nTony, when he is angry, accuses Carlos of being manipulative and exploitative. However, a thought running through the narration is that people generally do what they want to do; if it had not been Carlos for Tony, it would have been somebody else. This thought governs Carlos's film technique, which relies upon improvisation to the point where the performers' fantasies emerge. Tony is adamant that the sexual practices in the films were the decisions of the actors, not under compulsion or hypnosis (177, 179). This is perhaps all very well, but if each person is responsible for his own actions, consent must be informed. When Seth tells Tony about Carlos's other boys he adds: \"'I just think people have got a right to know certain things. So they can make their own decisions, if you know what I mean. But to do that they've got to know what's what'\" (150).\n\nThese questions are sharpened toward the end of the novel, as Carlos pushes his experiments with sex and power to new limits. It is disclosed that Carlos has seduced also Tony's younger brother, Ted, and that they pursued the reality of art and the freedom of the actor into increasingly violent and finally fatal courses. Carlos repeats the libertarian motif of the book: people do what they want to do, it takes two to fuck. Now Tony challenges this. \"I hated this man. I hated how he stepped into my life and ruined everything he touched and then just walked out without ever looking back\" (301).\n\nA recurring topic, on the other hand, is where Tony might have been otherwise. Mainly through people he meets around Carlos, he becomes politically aware and personally sensitive, in ways that, he says, could not have developed in Owen, Kentucky. \"'Carlos lifted me out of all that'\" (117). Intellectually, imaginatively, and morally, Tony becomes superior to the system that is going to incarcerate him indefinitely. His defense lawyer argued that Carlos was a monster; Tony is writing to set the record straight. The problem, according to _Boys of Life_ , is not class or age hierarchy, and certainly not the sex that goes along with them. It is Carlos's insatiable appetite for new partners, and his fluency in devising theories to justify it.\n\n**HUSTLING**\n\nRichard von Krafft-Ebing, _Psychopathia Sexualis_ , case number 146: \"I felt myself drawn exclusively towards powerful, youthful and entirely masculine individuals.... Since my desires are limited to persons of the lower social order, I could always find someone who could be had for money.\"\n\nIn _Bruiser_ by Richard House, Paul, an anxious Englishman, is attracted by the readiness of Adrian, a young waiter, to gratify him sexually while he pays the bills. \"Despite moments of tenderness between us, it's my money that keeps him with me,\" he admits to himself. However, this arrangement is satisfactory for Paul: it gives him a new confidence to express himself sexually. He forsakes his bourgeois caution and sets out to drive from Chicago to Mexico, without proper papers, banking arrangements, or travel information, and terrified by the thought that Adrian is HIV-positive. The boy, removed from his accustomed context and subject to Paul's priorities, becomes bored and captious. His erratic and enigmatic behavior initially intrigues Paul, but then increasingly undermines his sense of prudence, order, and purpose. If initially Paul was exploiting Adrian's dependency, he becomes his caregiver, trapped by the boy's insufficiency and his own middle-class sense of responsibility. \"He looks like an old man, his face puffy, cheeks and jawline swollen; it is hard to understand how it all came to this\" (170). Surprisingly, the novel ends with them together, escaping other commitments as they had wanted.\n\nThe outcome is less fortunate in _Steps Going Down_ by Joseph Hansen. Darryl Cutler has always lived off older men, and now is waiting for the death of his lover\/employer. When he falls for a casual pickup, Chick Pelletier, he meets his match in vanity, selfishness, and brutality. Chick is already being kept by an older partner, and has a girlfriend as well. He demands that Cutler get him a starring part in a film, and spends his money copiously as a way of compensating for his dependence. He abuses Cutler verbally and physically. But Cutler is infatuated, and is drawn further into fraud and murder. When Chick leaves him Cutler attracts the devotion of Eduardo, a Mexican delivery boy, but Chick tips off the immigration service. \"Find some nice guy with a job and a bank account, someone your own age\" (74), an acquaintance warns him. That seems to be the moral.\n\nThe hustler is sometimes a romanticized figure in gay writing. Mike (River Phoenix), blessed with beauty but not quite able to function by himself, spends time in a supportive group of street boys in _My Own Private Idaho_ (Gus Van Sant, 1991). This is a consolation also in Pai Hsienyung's novel, _Crystal Boys_. Phil Andros meets all kinds of interesting people as he peddles himself around Chicago and San Francisco; indeed, he finds that he is himself already a legendary figure. The clients are respectful, given to swapping literary quotes, and many are so sexy themselves that it is unclear why they need a hustler's services. The cops are sexy as well in their uniforms, and happy to join in. Unsurprisingly, such celebration of the hustler occurs toward the pornographic end of the spectrum of writing. As I argued in chapter 4, the function of pornography is to present images of sexual relations that are otherwise impermissible, or barely permissible. Phil Andros's adventures propose a liberty and a success that most of us can only fantasize about. To be sure, if hustling were truly so rewarding we would all be doing it; but we may like to imagine that it may be so.\n\nHustlers are likely exponents of the value of multiple and anonymous relations, and may well be explicit about the overwhelming of difference that they enact. Phil Andros's collection _Below the Belt and Other Stories_ is dedicated to \"Twelve Johns, eleven Dons, four Kennys, nine Jims, two Ikes, three Sams, two Scotts, four Garys, four Roberts, one Dean, and lots of other good guys too numerous to mention\" (v). Porn star Scott O'Hara writes in his memoir, _Autopornography_ , of a particular sexual encounter:\n\nI can't say much about that night, except that it was the most perfect night in my memory. What makes one particular night stand out so, from among the hundreds of nights of good sex that I've had? Cynically, I have to say that it's largely due to the fact that I never saw Colm again. We never had a chance to become familiarly contemptible (or is that contemptibly familiar?) toward each other.\n\nWhat made Colm special was his anonymity.\n\nJohn Rechy's narrator in _City of Night_ presents a more acerbic view of the scores (clients), hustlers, and queens. He finds his way from Texas to Times Square: \"like a possessive lover\u2014or like a powerful drug\u2014it lured me. FASCINATION! I stopped working.... And I returned, dazzled, to this street. The giant sign winked its welcome: FASCINATION!\" He resists attempts to settle him, either as a lover or as an employee. The seductions of the scene are continuous with the dangers: \"Life is lived on the brink of panic on the streets\" (150)\u2014panic generated by the vice squad, and by the prospect of finding that you are no longer young and desirable. The narrator's relation to all this is ambivalent. The others are presented as trapped in the scene, whereas the narrator, even while working it, is an observer. Which we are to take as the dominant image is unclear.\n\nEventually the narrator is almost persuaded by a client, Jeremy, to admit that really he wants to be loved, and has been using the passing of money as a repeated, because unsatisfactory, reassurance that he is wanted. Money will not be irrelevant, of course: Jeremy is offering full support in New York. However, this would require the narrator not just to trust himself in someone else's hands, but to admit that his response to gay sex is personal and not merely professional\u2014that he is gay. He decides that love is a myth to which he should not surrender (after all, Jeremy has not found it himself); he prefers the reality of the \"grinding streets\" (370). There is no suggestion that he has been damaged, any more than anyone else in a godless world.\n\nHugo in Oscar Moore's _A Matter of Life and Sex_ is attracted by the romance of hustling:\n\nHe wanted to be let into the game, to join in the tawdry spectacle, dressed up with tinsel and a smile, sinister and all-knowing, feigning naivety, feigning experience, battered by fate and by pimps, teetering on the brink of the gutter and drugs, living in a nightlife world of sex, violence and cash in the hand.\n\nHe is from a suburban background and goes to university, but exploiting his sexual attractions seems more fun. However, the streets have changed and there is no \"fraternity of hustlers\" such as he has read of in fiction. The scene is managed through an agency and is in hotel bedrooms. Initially, Hugo enjoys being attractive and skillful, but constant faking destroys his spontaneity. He makes no friends, and fails to find a sugar daddy or to meet a celebrity. Part of the lore of hustling is that you insist on your limits, what you will and will not do, and avoid dangerous situations. However, Hugo is forced. This happens also in _Close to the Knives_ , David Wojnarowicz's novel. Both protagonists tell their stories with their deaths from AIDS in view.\n\nAIDS and other sexually transmitted diseases apart, it is not obvious that the employment of sex workers is doing much _distinctive_ harm to anyone\u2014by this, I mean much harm that is not already endemic in the lower-paid, unregulated regions of market capitalism. The exploitation and humiliation is hardly less if you are working in a manufacturing sweatshop, and there may be a better prospect of the occasional thrill. Donald J. West in his study, _Male Prostitution_ , finds a pattern of social malaise among the boys, but concludes that \"they would have been problem personalities in any event and that involvement in street work was incidental to the disaster-laden course of their lives.\" The exchange of sexual favors for money is, perhaps, less a perversion of egalitarian, companionable relations, than a counterpart of them.\n\nNotwithstanding, sex that is paid for is firmly signaled as second-rate in much of our fiction. In Michael Arditti's much admired novel _Easter_ , readers learn to dislike the Archdeacon; he is hostile to the sincere, conscientious, and troubled vicar and curate. So when we find that on Good Friday he conducts a masochistic scenario centered upon himself as Christ and artfully elaborating the biblical narrative, we are hardly surprised. The corruption that we intuited is exposed. Ronan, the young man who is paid to flog and humiliate the Archdeacon, is supplied by Harry, a typically sinister professional, who was corrupted when a church server; Ronan is black, inexperienced, and has no money. \"'Do you think I want [to] do this? I'm telling you I'm skint, man. I'm going mad stuck in three rooms with my mum and sisters.'\" He embodies a normative perspective upon the Archdeacon's doings. \"Ronan no longer knows what to think. Nothing Harry said has prepared him for such perversity.\" To the Archdeacon, he's one of Pilate's thugs. \"'You see the broken, bruised, bloody body in front of you; so what do you want to do to it?'\" \"'Cover it up?'\" Ronan suggests, artlessly. \"'No, you fool: flog it! Don't you know the gospel story?'\" When the Archdeacon ejaculates with a cry of \"'It is finished,'\" Ronan rebels: \"'You're completely round the twist.... I'm not staying here'\" (339; my elision). It is left to the Archdeacon's mother, in an unscripted conflation of Mary Magdalene and Mary the mother of Jesus, to get him down from the cross.\n\nYet is this all so contemptible? If it is not your thing then it will appear weird, but other people's fixations are always like that. Of course, it is hypocritical of the Archdeacon to use the imagery of religion, in the face of its traditional doctrines of abstinence, for sexual excitement. But if fantasy is dependent on a supply of provocative materials from the power structures in which we live, as I have been arguing, it will feature substitutions and reversals of authoritative imagery (such as that of Christianity). Must it be that fantasies featuring heterosexuals in the missionary position and trying for a baby are authentic, while more adventurous scenarios are ridiculous or evil? To whose advantage is that? Should we not, in fact, be wary of satire as a form (on the cover of my edition, Muriel Spark says, \"Arditti writes about Western Christianity with pungency and satirical frankness\")? Does satire not, often, insinuate a taken-for-granted set of normative assumptions?\n\n**HOUSE PARTIES**\n\nA versatile location for the cross-class liaison is the weekend house party. Here the upper-class man will feel at home, and probably keen to show off the sexy man he has captured. The lower-class partner will feel like an outsider and ill at ease, so their relationship will be tested. At the same time, the outsider may offer a challenge to the other occupants, drawing attention to their own deceptions and insecurities. The class intruder is a wild card; he doesn't know the rules, you don't know what he might do (the plays of Harold Pinter are ultimately about this). The outcome, typically, finds some couples fading in commitment and others growing. Most often the outsider is cast off\u2014illustrating the dominance of intraclass pressure over sheer sexiness.\n\nIn _Love! Valour! Compassion!_ , Terrence McNally's play (filmed by Joe Mantello in 1997), everyone in the house responds to the attractions of Ramon, a young dancer of Puerto Rican origin and little education, whom John has brought with him to a gathering of old friends involved in the making of stage musicals. His strangeness is immediately registered: \"A Third World boyfriend. So John Jeckyll has gone PC.\" They discuss funding for the arts, and the need for a Diaghilev; but he would expect sexual favors. The only thing an artist should do for free is make love, Ramon declares. John, who is not a happy man, feels sidelined, and tries clumsily to assert his proprietary rights. \"Can we go upstairs and fuck?\" he demands. \"I didn't appreciate that fucking remark in front of your friends,\" Ramon complains later (29). The situation is delicate, he reminds John: \"Look, I'm sort of out of my element this weekend.... You're all old friends. You work together. You have a company. I'm just somebody you brought with you\" (33; my elision). Notwithstanding, John corrects Ramon's vocabulary and calls him \"Chiquita\" in front of the others. We learn later that John's sexuality is fixated upon a master-and-servant scenario, deriving from an early experience with an Irish boy who worked for his father.\n\nHowever, Ramon is already asserting his independence by setting out to seduce Bobby, who is blind and the partner of Gregory who owns the house. This leads to the breakup of John and Ramon, and eventually of Gregory and Bobby. However, it is when he accepts that it is Ramon who must dance the work that he can write but no longer perform that Gregory is released from his work block. Not all affairs are doomed to fail. Arthur and Perry reaffirm their fourteen-year relationship; but they are class-matched\u2014an accountant and a lawyer. Buzz and James, both of whom are living with AIDS, fall in love. The action is finally overshadowed by AIDS, which is said to be a genocide, destroying gay life. Actually, _Love! Valour! Compassion!_ displays the subculture as disconcertingly resilient.\n\nLyle takes Robert to meet his old friends John and Marian, in Peter Cameron's novel _The Weekend_. Lyle lives in a Brownstone house he has inherited from Tony (who was a travel writer), and has written a successful book on painting. Lyle was a visiting speaker and Robert was employed to drive him to the airport: that is how they met. Robert is a struggling young artist and a waiter in an Indian restaurant (he doesn't get on with his father, who is Indian, lives in Delhi, and makes money manufacturing counterfeit sportswear). Lyle is able to offer Robert a studio; he is attracted by his malleability\u2014\"'He listens to people; he really listens.'\" Robert listens mainly to Lyle:\n\nHe found almost everything about Lyle sexy: his body, his mind, his talk, the way he climbed stairs, the way his fingers gripped a fork, blushing with tension, the way he smelled and tasted, the impossibly soft way his back and neck and shoulders congregated, the spot there, the crux of him, naked and lickable. (127)\n\nRobert perceives Lyle as powerful, not in this or that respect, but generally, through his entire person; even his points of softness and vulnerability are powerful. Unfortunately, Lyle doesn't have much wisdom.\n\nJohn and Marian (not \"Mary-Anne,\" she insists) are too wealthy to have to work. They are old friends of Lyle, Tony was John's half brother, and he died of AIDS; so a lot is at stake when Lyle brings Robert to visit on the anniversary of Tony's death. John does his best to absent himself; under Marian's scrutiny, Robert appears not to belong. When he says \"'We've just lay about all afternoon,'\" Marian wants to correct him. \" _Lain_ about,\" she wants to say (103); when Lyle has a black eye from walking into a tree, John asks if Robert has hit him; his white shirt is presumed to be one he wears as a waiter. The symptomatic incident is at dinner, when Robert pulls a grape from the bunch, unaware that there are special scissors for grapes (they had belonged to Marian's grandmother). Laura, who is Italian and raunchy, takes Robert's part: \"'Oh, don't tame him!' Laura suddenly cried. 'Let him eat grapes with his fingers if he wants! Let us all be free of these stupid affectations!'\" (180). Robert, who has overheard Marian saying that she doesn't like him, feels that Lyle is siding against him and leaves abruptly for the city.\n\nNeither Cameron nor his characters quite say that class is the issue. \"'He's all wrong for Lyle,'\" Marian declares (156); \"'we are not well suited,'\" Lyle accepts (235). On the defensive, Lyle denies that the feeling between Robert and himself can properly be called love: \"'If you loved me\u2014if what you feel is love\u2014love would be a very cheap and common thing'\" (189). However determined the effacement, these are class terms. My sense is that Robert offered a challenge which the others failed to meet. It is unclear, at the end, whether Lyle and he can rescue anything.\n\nThe stranger is already within the gates in Alan Hollinghurst's novel _The Spell_. Justin used to be with Alex but is now with Robin; he invites Alex down to Robin's house in the country. Alex is a civil servant and Robin is an architect who tidies up country houses; both have a lot more money and status, and are far more established in the world than the somewhat younger men to whom they are attracted. These are principally Justin, who wants to be an actor; Danny, who is Robin's son but works as gallery attendant; and Terry, the local factotum with an eye for the main chance (a cross between Alec Scudder in Forster's _Maurice_ and Ron Wrigley in Angus Wilson's _Hemlock and After_ ). In a sequence of episodes alternating between London and weekends at Robin's house, Alex becomes accustomed to Justin's defection and falls for Danny, who becomes bored and leaves him.\n\nJustin has a starkly instrumental idea of relationships. \"'You're like me, darling, you need someone older to look after you,'\" he tells Danny. \"'I know Alex is rather shy and sensitive, but he's got plenty of money and a comfortable house and a sports car\u2014and in bed... well\u2014.'\" Actually, it emerges, being kept is not very good for Justin. He is bored all day, and finds himself unable to get interested in housework (which, of course, would confirm his subservient status). At one point Robin's approach affects him \"like a secretary briefly disarranged by an importunate boss\" (124). Danny maintains his own flat and gets a job so as not to be \"a kept boy\"; however, it is only casual work (141). Both Justin and Danny feel justified in undertaking flings with other, similarly placed young men. Justin tells himself \"how outrageous it was of Robin to leave him locked up here, like a slave, a mistress with no life of her own\" (93). Terry is available: \"'I can slip in through the back gate,'\" he says (48).\n\nThe answer, it appears, is to avoid relations of manifest, one-way dependence. Robin and Justin get onto a better footing eventually: the trigger is Justin's inheriting money of his own, so that he is no longer beholden. Alex takes up with Nick, who is of similar background and interests to himself and slightly older. The (relatively) egalitarian liaison wins out, then, though Alex finds that the spell of Danny lingers.\n**8**\n\n**RACE**\n\n**HUMILIATIONS AND AFFIRMATIONS**\n\nHierarchies of gender, age, and class afford a range of attractive scenarios for sexual power play. Age difference, for instance, may plug into fathers and sons, and teachers and pupils; these may be entirely amiable relations, though they may also afford scope for the enactment of violent and exploitative fantasies. The scenarios that race sustains are centrally and indelibly embedded in slavery, colonialism, apartheid, and servitude; they reference actual practices of torture and ill-treatment of such extremity that they have been outlawed in many countries. Historically, even benevolent relations between white people and people of color have characteristically involved condescension and stereotyping. It is still hard to think of intimate relations between black and white men without invoking a heritage of dominance and subordination.\n\nIn an encyclopedia article on \"Black Gay Americans,\" Ward Houser posits a range of responses to this heritage, often involving role substitutions, reversals, and loops. For some it seems appropriate for a white male to take a \"female\" role in a mixed-race liaison, because this may compensate for the historic dominance of white men. Others, \"being more comfortable in the submissive role, generalize from their experience of whites as holding the major power positions of American society to perceive white males as particularly sexually powerful, and so are attracted to them.\" Correspondingly, some whites may \"feel more comfortable dominating\" effeminate black gays, whereas others are drawn to more \"macho\" black men because they are supposed to be more virile. (We might hesitate there at the expression \"feel more comfortable.\" A good deal of the published evidence suggests that such preferences may be passionately driven; as I observe in chapter 3, fantasy is as likely to involve compulsion as freedom.) Even those who are not much affected by those hierarchical factors, Houser says, are likely to be involved in \"the attractiveness of the 'different,' curiosity, class differences, rebellion against social custom, or a belief that race should not be a factor in discriminating between potential sexual partners.\"\n\nThere is no reason to suppose that gay men and lesbians will be unusually progressive about other disadvantaged groups. In the 1970s and 1980s, Puerto Ricans seem to have been regarded in a notably functional light on the New York gay scene. In _The Farewell Symphony_ Edmund White remarks casually, of a man he sleeps with: \"he was himself bewitched by Puerto Ricans, as who was not\"; the question is so rhetorical that it doesn't get a question mark. Midwestern boys, in particular, are said to find Puerto Ricans exotic. In Andrew Holleran's _Dancer from the Dance_ , Puerto Rican boys are ubiquitous, beautiful, and said to have \"big cocks.\" They unload boxes at a store, exterminate roaches, and carry messages; they play handball in empty lots; they are never within the social orbit of the narrator and his friends. Malone \"loved those boys, as did I, to be among them was enough; he was in thrall to them, he was in the thrall of Puerto Ricans\" (188). Again: \"My dick would straighten out like a divining rod, forcing me to follow more than twenty blocks in fruitless pursuit\" when a Puerto Rican passes, we are told in David B. Feinberg's _Eighty-Sixed_. There is no sense that these enthusiasms might be personally and politically problematic; no concern about how it might feel to be the object of such casual attentions.\n\nTo be sure, sexual attraction frequently involves objectification. But race supplies the most ready-made, and hence the most crude, repertoire\u2014and all no more than _skin deep_. In Lyle Glazier's story, \"Chester,\" the narrator, who prefers black men, remarks of a partner: \"I thought he had no formal education, no book learning, no academic interest in literature, music, the fine arts. I erased our difference. I was engulfed by his brown warmth. He was pure sexuality\u2014gentle, placid, as open to love as the earth is open to the sun. I loved his brown against my white.\" The narrator tries to make his partner a blank, erasing every attribute but two: sex and skin color\u2014and finally just skin color. Further, racial differences are usually permanent. As Rhonda Cobham remarks, the youth in ancient Greece was expected to accept homosexual advances while a boy, on the assumption that this was an interim stage before he became himself a citizen. But you don't grow out of racial subordination, even if you change class (in the language of race, you remain a \"boy\"). Indeed, the white mentor may feel threatened at the prospect of his black prot\u00e9g\u00e9 rising in the social scale. In Paul Thomas Cahill's story, \"The Reunion,\" Rodger (white) has left Julius (black) because he couldn't cope with his being successful: \"'I couldn't be in charge, couldn't take care of you as I thought I was expected to, not when we were... equals.'\"\n\nJames Baldwin in _Another Country_ depicts Rufus as driven to violence, despair, and suicide by his inability to handle the racial milieu in which he is trapped. Rufus's sister Ida is tougher, at the cost of total separation: \"'If any _one_ white person gets through to you, it kind of destroys your\u2014single-mindedess. They say that love and hate are very close together. Well, that's a fact.'\" Black Panther Eldridge Cleaver, in his notorious attack on Baldwin in _Soul on Ice_ , alleged that _Another Country_ displays \"the most grueling, agonizing, total hatred of the blacks, particularly of [Baldwin] himself, and the most shameful, fanatical, fawning, sycophantic love of the whites.\" In Cleaver's view, \"many Negro homosexuals, acquiescing in this racial death wish, are outraged and frustrated because in their sickness they are unable to have a baby by the white man.\" Actually, _Another Country_ is not incompatible with Cleaver's hostile analysis: it amounts to an extended demonstration that cross-racial affairs are irretrievably doomed, because of hang-ups such as Cleaver posits\u2014and, anyway, are scarcely tolerated by people in New York (let alone the South). The good sex, in Baldwin's novel, is between white men. Notwithstanding, Huey Newton, another Panther, envisaged an alliance between the black, feminist, and gay movements, and labeled Cleaver himself a repressed homosexual. These concerns are still politically active in the United States around Minister Louis Farrakhan and the Nation of Islam.\n\nThere may be scarcely less objectification when the conjuncture of sexuality and race is offered as the ground of spectacular harmony. In the films _My Beautiful Laundrette_ (Stephen Frears, 1985) and _The Wedding Banquet_ (Ang Lee, 1993), a miraculously egalitarian, racially blind gay relationship is presented as a magical opportunity for the overthrow of (merely) cultural misunderstanding. The boys in _My Beautiful Laundrette_ inhabit a world in which \"Asian\" and \"skinhead\" are the most antipathetic terms, but between the two of them race is unremarkable. In _The Wedding Banquet_ , Wai-Tung's Taiwanese family produces intense cultural disruption in the relationship of Wai-Tung and Simon, but the two principals appear ideally harmonious in every other respect, and totally unaware of any complication in the conjunction of sexuality and racial difference. Racial blindness appears comparably in Marshall Moore's story \"Everybody Loves the Mus\u00e9e d'Orsay.\" In this instance a Sino\/American couple are living happily in Malaysia, when the American mother and the demands of family intrude clumsily upon them. Again, race itself appears to be merely incidental for the two men. _The Crying Game_ (Neil Jordan, 1992) presents transvestism as the big shock, while making racial difference of little account between Fergus and Dil. This instance is complicated by the fact that the white man, Fergus, is Irish\u2014of a subjugated people\u2014whereas Jodi, who is black and from Antigua (formally independent) has signed up to fight for the imperial power. The simply English person is Dil, who is of mixed race.\n\n**IMPERIAL RESIDUES**\n\nThe alleged inadequacies of colonial subjects position them as the inferiors that witness to European superiority. Fiction indicates that colonial Europeans spent a good part of their time producing anxious, self-justifying stories about the relationship between the natives and themselves. If colonial fiction \"can demonstrate that the barbarism of the native is irrevocable, or at least very deeply ingrained, then the European's attempt to civilize him can continue indefinitely, the exploitation of his resources can proceed without hindrance, and the European can persist in enjoying a position of moral superiority,\" Abdul JanMohamed observes.\n\nIn the midst of apartheid, South Africa produced stories exploring the rewards and difficulties of black-white liaisons. David's initial defining experience, at the start of John Sandys's novel _Against the Tide_ , is on the beach, with \"a golden-brown gypsy boy with dark, curly hair.\" David is in his late twenties and can't settle in England after World War II, personally or to an occupation; there are said to be more opportunities in South Africa. On arrival there he immediately gets invited to a party of same-sex \"Coloureds.\" They have \"dark-skinned shining bodies, beautiful to behold, muscular, deep-chested, narrow-hipped, long-legged\" (58); they are \"young, virile, gentle and uninhibited, with music in their souls and laughter in their hearts\" (61). However, David is quickly told that mixing is not tolerated. He despises the local white men he sees at gay parties, and the married couple who take him in and try to seduce him.\n\nDavid becomes a commercial traveler; it is love at first sight when he is assigned a Zulu driver. \"Ugi was the golden-brown gypsy boy and he was in love. He asked God to bless them both\" (82). Ugi, we learn, fell in love immediately with David. Their meetings are dangerous and desperate. David is distraught when he learns that Ugi has a wife and children. However, Ugi introduces him to Guy, another white man with a black lover, who explains that marriage is required by local custom. Guy draws David into a circle of clandestine black-white couples, but even here he dislikes the whites.\n\n\"'Why do they refer to a grown man\u2014a husband and a father as 'boy'?'\" David demands (126). Nonetheless, his relationship with Ugi depends on difference. \"Ugi gave a wide smile. 'Me happy to be David's boy,' he said proudly'\" (126\u201327). Ugi uses the name given to him by white employers; David himself thinks of him as \"this beautiful child\" (87). Ugi apparently has a preference for male lovers: he thinks of himself, in the idiom of his people, as a \"woman-man\" (140).\n\nHowever, Ugi has educated himself and gives David the best advice on how to establish himself financially. He sees a market opportunity, making and selling lampshades; he becomes the creative force in the business. Compare Alice Walker's _The Color Purple_ , where Shug and Celie secure their future by setting up a business making pants: it appears that small-scale capitalist development is the key to security, advancement, and affluence. David and Ugi pay no attention to the political situation, and consequently are taken by surprise when business confidence is destroyed by the Sharpeville Massacre. Eventually they are driven from South Africa, but their love survives and they will start again in England.\n\nIn _Against the Tide_ , the South African political situation figures mainly as threatening David's personal and business affairs. In Stephen Gray's _Time of Our Darkness_ , the personal bears a political message. This novel is set in South Africa during the school boycott which led eventually to the release of Nelson Mandela and the end of white rule. Pete has a fading gay relationship with Andr\u00e9, who is Afrikaans and an airline pilot. Pete is a teacher at a private school and well aware of the rule, \"never to lay a hand in amorous expectation on a pupil.\" Disley, the token scholarship black student in the class, age thirteen, arrives at his house, evidently in need of support. The boy seduces Pete, and they become lovers. Like Ugi, Disley has an independent understanding of sex between men\u2014he knows already about single-sex hostels, mine compounds, Andr\u00e9 (Pete's partner), and Andr\u00e9's pickup.\n\nIn the metropolitan, egalitarian mode, _Time of Our Darkness_ is inclined to play down the sexual significance of hierarchy between Pete and Disley. \"We were more than equal in the dark,\" Pete declares, referring, I suppose, to genital endowment (78). He does not admit to any special thrill from age and class difference, and appears not to have entertained any prior idea of sex with black Africans.\n\nI was brought up not to touch black skin. Black skin was unhealthy, scaly like a reptile's, gave you TB. A whole country has been divided on that prejudice. When I was a child my mother pulled me out of the reach of the nanny, feeding me herself, bathing me. When I was at primary school we'd run down the corridor, make a circle around the cleaner on her hands and knees, reassemble, not having touched black flesh. (138).\n\nFor Pete, skin color is a revelation rather than an abiding fascination. \"Do I need to describe the sensation that I experienced as the blackness went out of Disley's skin for me and I felt the person beneath. [Again, there is no question mark.] All of him\" (138). In this passage Disley's skin is sexy not because it is black, but because, for Pete, it is no longer that. In a typically egalitarian gesture, Pete erases the blackness of Disley's skin and finds \"the person beneath. All of him.\" Even in such an extreme situation, difference is said to be unimportant.\n\nYet, contradictorily, Pete does proceed to find positive value in Disley's skin. He imagines a visit from the police:\n\nLet the man in the raincoat come. I had a few things to tell him: about black skin, about how it felt, the texture, the grain. And about loss. And about deprivation and humiliation. I was sick with it, as my white, police-supporting countrymen were, but they were sick with their aversion for black skin. (138)\n\nNow we have positive qualities of texture and grain; Disley's skin is relevant after all, it does contribute to an erotics of difference.\n\nAccording to Gray, in an interview, the age difference in his novel is symbolic. His idea was that \"the entire impetus of the uprising in South Africa in the mid-1980s, during which children assumed the role of adults and adults became, to say the least, vindictively childish, should be acted out literally.\" The message is that the country has been plunged into turmoil by a pointless phobia, and that children, by refusing to learn Afrikaans in schools, are leading the revolt. Actually, black skin has its own attractions and the children are wise. For the time being\u2014the time of darkness\u2014the system wins: Disley achieves not a gay relationship but martyrdom in the cause of his people.\n\nEvidently, South Africans have not finished with this scenario. A more gloomy version appears in \"A Son's Story\" by Tony Peake. Paul (white) is having a loving sexual relationship with young Memphis (black), who looks after the garden, and whom Paul is coaching in English literature. The relationship is destroyed when a grossly violent dispute between ANC and Inkatha factions draws Memphis back to his family home. In his distress, and perhaps resentment, Memphis tears up all the literary books. Unfortunately, we learn nothing more precise about his feelings, since he is given almost nothing to say. This is partly because the action is seen from the viewpoint of an English visitor, a born-again Christian\u2014a rather easy target.\n\nA traditional skill of the fiction writer is entering into unfamiliar consciousnesses. Yet Sandys in _Against the Tide_ makes little attempt to explain Ugi's thoughts to the reader. We don't know what the two men talk about when they are together, except business and how beautiful and sexy each finds the other. When David (upset by Ugi's visit to his wife and child) is unfaithful, we are told Ugi's response: \"Ugi had been badly hurt, and his first impulse was to leave David, to live only for his wife and children, to give up being a woman-man. But he loved David deeply, and when he surfaced after the initial shock he sensed that with time and patience the memory of that awful day would blur, and he would be able to forgive\" (140). This is helpful, but perfunctory compared to the central narrative role of David's thoughts. After twelve years they go to Europe together, and David ponders Ugi's difference:\n\nDavid lay, thinking about Ugi's early days and the way he had changed. \"I wish I knew what he thinks\u2014how he thinks. To be taken out of a tribal society and thrust into this strange way of life is a challenge which has needed a superhuman adjustment, yet he seems to succeed without effort. I am so proud of him...\" (150)\n\nDavid allows Ugi to see that he is thinking along these lines. Ugi objects, forcibly: \"'I see David is still conscious that I am black.... I am no novelty. I am not different.' He stared hard at David. 'I have the same emotions, the same feelings, I have the same red blood.'\" \"David was hurt, and ashamed into silence\" (151; my elisions). This is perhaps the only point at which a distance is allowed between David and the narration. Still, David feels able to pronounce that Ugi \"'now has a well-developed mind and a personality'\" (165), as though he had previously lacked these attributes.\n\nSimilarly, in _Against the Tide_ , it would be marvelous to know what Disley thinks about it all. We may infer, with Pete, that Disley is impressed with his lifestyle; he has thrown in his lot with education (while the rest of his township is rioting); Pete can take him forward. Pete's worry is that Disley will underestimate the gap between their social positions, and will imagine that by imitating Pete's amiable manner he can progress to a comparable state of affluence and self-determination. The predicament is manifest when Disley is returning to his township: there is no point in Pete giving him a hair drier because they have no electricity. However, Disley's self-possession, boosted by his success at school, enables him to inspire his people. If the narrator in _Time of Darkness_ does not presume to take us into Disley's consciousness, the reader is encouraged to share a respect for the other dimensions in his life, in the township and in a rural area that they visit.\n\nPerhaps it is not possible, still, for a white South African to presume to interpret the thoughts of a black person. J. M. Coetzee, in his prize-winning novel _Disgrace_ , doesn't attempt this; we get only the viewpoint of the white, male protagonist. A black African has a large influence on the outcome of the action, but his thoughts and feelings remain mysterious and ominous. Perhaps such reticence is ultimately a sign of respect: what white people have done to black people in South Africa makes it impossible for one to speak for, or through, the other. Yet it must also undermine any humanistic or paternal argument, that all peoples are, or can become, one. Actually, as Patricia Duncker points out, Coetzee has another character of restricted representation, a white lesbian. She also has no independent consciousness. She does have a lot of conversation with the protagonist, but, being traumatized, she is unable or unwilling to explain how she feels. This is not just a matter of fair shares, so to speak. If Coetzee were to assay the consciousness of these others, he would have to consider more carefully the claims these people might reasonably make upon each other and upon South Africa.\n\nHomosexuality is shown as enhancing the opportunity for exploitation and humiliation in V. S. Naipaul's novella, _In a Free State_. Bobby, who is white, works as some kind of civil servant in an independent East African country, in which the president and the military are in the process of overthrowing the more traditional rule of the king. Bobby does this work not because it allows him to lord it over black African people, but because it enables him to abase himself before them in ways that are actually patronizing and condescending, and calculated to shore up his fragile emotional stability. He had a nervous breakdown; \"'Africa saved my life.'\" Bobby tells himself that he is recognizing the humanity of the African, but becomes crazily enraged when confronted with irrefutable indifference or hostility toward himself. He seeks to ingratiate himself with Africans by wearing what is called a native-style shirt; this annoys an army officer and he is severely beaten up.\n\nBobby acknowledges an effeminate side to his nature. For him Africa is empty spaces, long drives on open roads. \"'You want lift? You big boy, you no go school? No, no, you no frighten. Look, I give you shilling. You hold my hand. Look, my colour, your colour. I give you shilling buy schoolbooks. Buy books, learn read, get big job. When I born again I want your colour'\" (109). The key moment is the color juxtaposition, the touching of the black and white skin. Unfortunately for Bobby it doesn't always work. He chats up a South African in a hotel bar: \"'If I come into the world again I want to come with your colour.' His voice was low... his fingers moved until they were over one of the Zulu's.... Then, without moving his hand or changing his expression, the Zulu spat in Bobby's face\" (107; my elisions).\n\nAlthough Naipaul dwells upon the inadequacies of African regimes, he appears to share the opinion of President Sam Nujoma of Namibia and President Robert Mugabe of Zimbabwe, that homosexuality is a corrupt European import. His main theme, in this offensive and tremendously written book, is that relations with Africans are impossible. When they are not threatening or actually brutal, they are ridiculous: \"The frightened boy brought in the soup plate by plate, pressing his thumbs on the rims. He walked with a stoop, raising his knees high; his big feet, loosely hinged at the ankles, flapped up and down\" (134). Above all, for Naipaul, these people are black; the narrative returns obsessively to their blackness. The only man more objectionable than the black African is the white man who takes him seriously, and no one but a pathetic, damaged queer is going to do that. Naipaul's story shows, in effect, what happens when you allow the native to get the whip hand.\n\nAngus Wilson takes third world peoples more seriously, though his novel _As If by Magic_ is not free of suggestions that they may be amusing, irrational, mysterious, and dangerous. However, it does offer a significant challenge, from within a gay framework, to the complacency of empire. Hamo Langmuir, a famous plant geneticist, tall and clumsy, has such exacting fantasy requirements that he has driven away Leslie, his devoted partner, because at twenty-five he is too old and has to shave. Hamo has found a network of rent boys, but is unable to cope with any sexual relation that might become personal. His sexual opportunities are transformed when he is sent on a world tour to observe the success of \"Magic,\" the new rice which he has bred, and which is transforming the agro-economy of the developing world. He passes in a haze through Japan, Indonesia, and South Asia, becoming increasingly bold in pursuit of the Fairest Youth, and grossly disrespectful toward his courteous and kind hosts, whose folly consists merely of embeddedness in a social milieu in which there are manners, decencies, and ongoing expectations. \"'Mr Langford, we are here a _small_ community. You have come among us without respect.'\"\n\nAbove all, it is difficult for the boys whom Hamo has accosted to carry on as before in their communities; he imagines he can make it all right by writing checks. His remark, that if a Singhalese serving boy, Muthu, gets into trouble he will take him to England, leads the boy to run away from a family where he has been well treated. \"'What can you do with the boy in England? He is a good boy. But he is ignorant. What will he do there?'\" his mistress, Mrs. Dissawardene asks (282). This is a pertinent question\u2014we know that Hamo's attraction to Muthu will last only while he is young. He catches fleeting glimpses of Muthu everywhere he goes (to him, one beautiful youth is much like another), and makes desperate attempts to find him. People die; the forces in play are entirely beyond Hamo's control.\n\nThese interferences in local mores are continuous with the intrusion effected by Magic rice, which is so productive that small farms on poor land are no longer viable. Muthu, for instance, might have returned to his village, but their land is marginal and Magic means that they wouldn't be able to feed him. A homophobic official accuses:\n\nI suggest that instead of insulting your host by buggering his servants, you look for your leavings in the slums of the cities of Asia. Luckily your very valuable scientific rationalisation of our local agriculture has made sure that the bazaars and public places are filled with the scum overflowing from the waters of hopeless paddy fields. If anyone has such low tastes, they are always ready to oblige for the price of a bowl of rice. (176)\n\nHamo submits a report calling for social conditions to be taken into account in agrarian innovation, but his employers ignore it. I admire _As If by Magic_ for the sharp light it throws upon metropolitan gay mores, from within a gay perspective that may still endorse the quest for the Fairest Youth and his potential to respond. Also, it illuminates the continuity between psychosexual and sociopolitical hierarchies: Hamo's private quest is implicated in his professional practice.\n\nOn the opening page of Alan Hollinghurst's novel, _The Swimming-Pool Library_ , Will Beckwith, the protagonist, admires a black maintenance worker on the Underground: \"The black was looking at his loosely cupped hands: he was very aloof, composed, with an air of massive, scarcely conscious competence\u2014I felt more than respect, a kind of tenderness for him.\" There is a similar moment in the closing pages of the novel.\n\nWill's new boyfriend, we learn at once, is black, seventeen, and from Stratford East (Will is twenty-five). Will is besotted: \"Oh, the ever-open softness of black lips; and the strange dryness of the knots of his pigtails, which crackled as I rolled them between my fingers, and seemed both dead and half-erect\" (3). Apparently the enthusiasm is reciprocated: \"Then he would give up and fall recklessly on top of me on the sofa, panting in my face, kissing me, full of clumsy humour and longing\" (13). Will is in charge. The relationship takes place entirely on his territory, and he expresses his passion with playful spankings.\n\nWill admits to his friend James that his world and Arthur's are too far apart for a love relationship; \"'It must be just an infatuation'\" (20). Nonetheless, he does think of himself as \"in love\" with Arthur\u2014the more so because of his (alleged) limitations:\n\nEven when he spoke, in his basic, unimaginative way, I felt almost sick with desire and compassion for him. Indeed, the fact that he had not mastered speech, that he laboured towards saying the simplest things, that his vocal expressions were prompted only by the strength of his feelings, unlike the camp, exploitative, ironical control of my own speech, made me want him more.... But in sex he lost his awkwardness.... It was a kind of gift for giving. (64; my elisions)\n\nThe dichotomy Will produces here is not ungenerous\u2014his own mode of speech is not more attractive. Yet his is the language privileged in the book's narration. An antiracist will surely wonder whether Will might not have found more various qualities in Arthur if he had not been so ready to stereotype him.\n\nIn fact, Arthur is not quite as dumb as Will likes to believe. He finds Will's speech strange and funny: \"Odd words seemed to amuse or offend him, and he gave urchin imitations of my speech. 'Arse-hale,' he would drawl. 'Get orf my arse-hale'\" (presumably we should be hearing the entire narration in this upper-class accent). Will is quite disconcerted by such role reversals: \"Sometimes I laughed graciously too, and did even posher imitations of his mimicry, knowing no one was listening. Sometimes I caught him and gave him what he was asking for\" (14). Will is unnerved until he has reasserted his mastery.\n\nWhen Arthur becomes trapped in the flat by some obscure, drug-related gangsterism in his family, their incompatibility becomes overwhelming for Will. They fall into a fluctuation between hostility, sexual violence, and sentimentality: \"Now it became a murky business, a coupling in which we both exploited each other, my role as protector mined by the morbid emotion of protectiveness. I saw him becoming more and more my slave and my toy, in a barely conscious abasement which excited me even as it pulled me down\" (31). It is unclear whether this is to be seen as an inevitable, natural corruption, or the consequence of Arthur's artificial confinement.\n\nWhat, we may ponder, should Will and Arthur have been doing? They might have avoided each other, on the axiom that mixed-race couples are inevitably exploitative. They might each have retained their racial interest while looking out for someone of their own class and age, reasoning that taking on several hierarchies at once is too ambitious. They might have gone into therapy together, so that Will might learn to outgrow his domineering impulses and Arthur might receive assertiveness training. But would that have helped? Will positions Arthur as a simple and primitive counterpart because that is what turns him on. As James notes in his diary, \"'yet again he had picked on someone vastly poorer & dimmer than himself\u2014younger too. I don't think he's ever made it with anyone with a degree'\" (218). Is it wrong, or a mistake, to try to relate to someone of a different educational background? I like to think that they might have proceeded in something like the way they do in the novel, but with Will taking as much trouble to understand, engage, and please Arthur as Arthur is probably taking over him.\n\nA partly comparable pattern is disclosed by Charles, who is eighty-three. His diaries of interwar life record an alternation between idealization of young black men and casual cottaging (sex in public toilets); the love of his life was a Sudanese boy, Taha, whom he brought back to London as his servant. Charles has regarded this as a noble commitment, but Will suggests that it is paternalistic, seeing adults as children (this is rich, coming from Will). Charles reasserts that there was in the colonial service \"'this absolute adoration of black people.... I've always had to be among them, you know, negroes'\" (242; my elision). Whether Charles's mode is preferable, personally, politically, and ethically, is hard to say. Anyway, it appears to have degenerated in modern conditions: he now uses his hold over his retinue of beneficiaries to involve them in pornographic movies.\n\nRacial hierarchy is not just Will's and Arthur's thing. It is the outcome of historic imperial relations. Will sees Abdul, the chef, \"abstractly sharpening his knife on the steel and gazing at me as if I were a meal\" (42). Eventually, Abdul is to have Will across a chopping table in the kitchen, tenderized with hard slaps and lubricated with corn oil. This is the only time Will is the insertee; Abdul is Taha's son; the empire fucks back. Again, Britain may retain the Falkland\/Malvinas Islands, but imperial corruption is suggested by the exchange with Gabriel, a wealthy Argentinian, who is visiting London to buy pornography which he can't get in his own, less decadent country.\n\nDavid Alderson rejects the idea that _The Swimming-Pool Library_ is \"an attempt to lay bare the roots of present-day sexual projections in historically grounded power relations.\" I agree that the novel cannot be enlisted for an anticolonial agenda; it does not sustain a coherent critique of imperial relations. Rather, as Alderson shows, it tends nostalgically to set an imperialist and preliberation past against a degraded present, now (at the time of narration) menaced additionally by AIDS. Nonetheless, the novel does disclose structural relations between imperialism and racism, in the formative mid-twentieth century and on into the present. Also, it exposes Will's delusion that he can be free of history. His self-centered personal life is of a piece with his general discovery that he has been arrogant in supposing that he is free, autonomous, and without responsibility. Hollinghurst does not condemn imperialism, together with its interpersonal outgrowths; he explores the seductive and distasteful erotic opportunities for people living in its wake, and locates the sexual energies which they may still release.\n\nWhat the protagonists find in _As If by Magic_ and _The Swimming-Pool Library_ is that they are less in control of their destinies than they had thought. This theme is to be expected when the decline of empire is at stake: western Europeans do not rule the world any more. It is also continuous with the argument of this book, that our psychic life is organized along the lines of the main hierarchies that determine our economic, social, and political lives at large. Hamo and Will make their personal choices, but they are structured by the imperial scenarios within which they operate. Hamo's fantasies are those of the European intruder upon South Asia, from the eighteenth century to the sexual tourism of today. Will's desires are continuous with those which Charles exercised in the Sudan.\n\n**SISTERS AND BROTHERS**\n\nFor many lesbians and gay men of color, cross-racial relations such as I have been describing spell bad news. The endemic complications and hesitations emerge in a candid and courageous late-night television discussion among three lesbians of color and three gay men of color, _Doing It with You Is Taboo_ (SOI for Channel 4, 1993). Some of the contributors are in or ready to contemplate relationships with whites, some not. Either way, two main alternative propositions emerge: (1) black\/white couples are to be avoided because between such people race is bound to be important, and destructive; (2) a black\/white couple is fine because, for the speaker and his or her partner at least, race is insignificant. A third proposition, that black\/white difference might be both central and rewarding, is not entertained.\n\nThis pattern has occurred elsewhere in this book: difference is all right so long as it doesn't make any difference. It appears again in the British compilation, _Lesbians Talk: Making Black Waves_ (1993). Relationships with white women are vastly problematic because, with the best will in the world, hegemonic white racism is so insidious. Some of the women canvassed eschew them altogether. Yet such relations may be \"healthy,\" the editors say. One woman concurs: \"I don't think making love to a White woman is any different from making love to a Black woman. All my life in this country I've had four beautiful relationships. Three were with White women and the fourth was with a Black woman.\" For another, conversely, \"the issue of racism was always there. You cannot escape from racism.\" The tendency of my argument, I realize, has been to endorse any kind of sexual agenda that promises to work. However, at certain points the psychological and political risks may be too great; these troubled accounts of cross-race liaisons reveal an intensity that bespeaks not only decades of personal discrimination but the histories of peoples.\n\nFictional and autobiographical treatments rehearse the problems. In Steven Corbin's novel _Fragments That Remain_ , Skylar (black) and Evan (white) are probably breaking up, mainly because of continuous bickering about race. \"I think Evan likes black people for all the wrong reasons,\" Skylar observes. So what reasons are these? Judging from Evan, they appear to be taking the opportunity to be patronizing and self-congratulatory about one's liberalism, while maintaining a subdued but persistent level of racist sniping. Skylar declares, in anger, that Evan's attraction to blacks is like \"the master leering at the slave\" (173\u201374).\n\nProgress between the two men occurs when Evan sees that Skylar has a point: \"'I do love black people, their passion, their exoticism, their colorfulness, but maybe I don't look beyond that'\" (235). What would one see \"beyond\"? It might be the fullness of the individual, but Corbin encourages his reader to think rather of the lived realities of racial prejudice, at large and in the relationship. The main point, Corbin is perhaps saying, is not whether people's sexual fantasies follow racial divisions, but that each person should give adequate recognition to the situation of the other. This means the white man attempting to get some distance from an almost inevitable white viewpoint.\n\nThat is surely possible at a rational level. But what about the underlying fantasies? For if, after all, racial difference is the basis of the attraction, that difference is manifest, in our societies, as inequality. Unfortunately, Corbin gives little indication of what Skylar and Evan fantasize about or do in bed, so it is hard to envisage how far and in what ways their interaction depends on hierarchy. Corbin protects Skylar from the sissy-stigma by having him tell his brother \"a million times, all gay men don't get fucked. He's said it so much, I believe he's one of them\" (260). In fact it is not easy to see what Skylar, who is unusually self-possessed about both gayness and race, gets from the relationship with Evan. While all the characters, including Skylar and Evan, make moral progress during the action of the novel, it remains unlikely, at the end, that Skylar and Evan can live together again.\n\nIn Larry Duplechan's novel _Eight Days a Week_ , Johnnie Ray Rousseau is a singer, taking after his mother, in the shadow of his handsome brother. He experiences himself as unattractive and takes Barbra Streisand as his \"hero\" (note that Streisand's main film romances show Jews connecting with gentiles). Johnnie's desire-to-be produces a corresponding desire-for: he fantasizes that he might attract a blond hulk\u2014\"a lover. A husband. Some impossible combination of Tab Hunter, Rick Nelson and Steve Reeves.\" At school Johnnie attaches himself to sporty white boys, abhorring his own \"flat nose and (to me) overlarge lips\" (25). A friend tells him that he has become \"'the all-too-willing victim of America's white-supremacist, master-race plantation mentality.'\" Whatever the reason, Johnnie replies, he still likes blonds (28).\n\nHe seems to have found what he wants in Keith Keller. Johnnie's adulation of whites tends to place him in the \"sissy\" role, but he takes up bodybuilding and is delighted when Keith likes to be fucked in turn. It transpires nonetheless that Keith really wants Johnnie in the role of housewife. He starts calling Johnnie \"K. T.,\" because he looks like Tutankhamen (\"King Tut,\" but also \"Katie,\" I think; 186). Keith is unable to cope with Johnnie's musician's lifestyle, and his possessiveness breaks up the relationship. However, Keith is not altogether in the wrong. Johnnie admits that his response to the standard message in the contact ads\u2014\"'No fats, fems or blacks, please'\"\u2014coming on top of his own initial lack of confidence, is to welcome, flirtatiously, every available instance of white sexual interest in him; he regards it as a triumph over the prevailing racist disconfirmation. \"The prettier, the blonder, the more Aryan the man, the bigger (if not more permanent) the sense of victory at the sight of the man's fair head bobbing between my thighs. I suppose I truly am, as Snookie often tells me, a sick, sick woman\" (197\u201398). Even at the moment of his proclaimed victory Johnnie types himself, self-deprecatingly, as female: race and gender line up in a classic conflation of subordinate roles. He cannot extract himself from disconfirmation and stigma while he stays within the hierarchical framework. Nonetheless, Johnnie moves on to a valuable relationship with another white man, so it appears that cross-racial affairs are not doomed.\n\nIf feelings of racial inferiority and superiority may disturb cross-racial liaisons, they may also interfere with black-on-black relations. Audre Lorde's inspirational commitment to black women, as she presents it in _Zami_ , does not occur spontaneously. Among school friends she \"never mentioned how enticing and frightening I found their strange blonde- and red- and chestnut-colored secrets that peeked out from beneath their pulled-up half-slips.\" Lorde's early affairs are with white women. Even in Mexico, where she feels suddenly at home, her main relationship is with a white woman. \"'How beautiful and brown you are,'\" Eudora says (144). Back in New York in the 1950s, there appears to be little choice; the only organized scene is white. \"It seemed that loving women was something that other Black women just didn't do\" (155). The few that did found themselves \"sleeping with the same white women. We recognized ourselves as exotic sister-outsiders who might gain little from banding together\" (153). \"[W]e seldom looked into each other's Black eyes lest we see our own aloneness and our own blunted power mirrored in the pursuit of darkness\" (197\u201398). Lorde and her friends colluded in this scene, though they resented the racist admission policies of the bars, because it was the only one they could find. Muriel, her lover, admired by Lorde for her \"paleness\" (159), believed that lesbians were all equal in their outsiderhood: \"'We're all niggers,' she used to say, and I hated to hear her say it\" (177).\n\nLorde makes progress as a lesbian and a black woman through a culminating affair with Afrekete (Kitty): \"her chocolate skin and deep, sculptured mouth reminded me of a Benin bronze\" (214). She evokes a Caribbean island in Harlem:\n\n_And I remember Afrekete, who came out of a dream to me always being hard and real as the fire hairs along the underedge of my navel. She brought me live things from the bush, and from her farm set out in cocoyams and cassava_ \u2014those magical fruit which Kitty bought in the West Indian markets along Lenox Avenue in the 140s or in the Puerto Rican _bodegas_ within the bustling market over on Park Avenue. (218; Lorde's emphasis)\n\nThis is both impressive and limiting. Some black women may find such determined cultivation of the exotic a high price for a color-affirmative relationship. In the early 1970s, Lorde \"began to live together permanently\" with her daughter, her son, and her white lover, Frances.\n\nLorde's achievement is to write positively of black-on-black lovers without suppressing the complications. She enlists her mother as an honorary dyke, on the ground that she was a \"powerful woman\" (6), but is this enough? She seems a monstrous abuser to me. Despite Lorde's famous skepticism about dismantling the house with the master's tools, the two projects she declares at the start of _Zami_ \u2014to be both male and female, and to replace the mother-father-child triad with grandmother-mother-daughter\u2014aspire to co-opt the power system of the heterosexual family. Anna Wilson finds here \"a crucial lack of alternative conceptualisations through which to imagine community.\" In fact, Katie King notes, despite her lack of comment on butch\/femme roles, Lorde characteristically assumes the dominant and protective role of a \"top.\" In her relationship with Muriel she adds fifteen years to her age in order to secure this role ( _Zami_ , 160, 165).\n\nLesbians of color continue to confront these questions. Ekua Omosupe ponders:\n\nwhen my white lover sees my Black face, does she read in it that I am the mythological, strong Black woman who is more stick than flesh? Am I the dark exotic? Am I the testimony to herself and others that she is not racist, but quite liberal? Does she see the face of her family's maid who was paid to love her and to take care of her because mother and father were too busy to be bothered?\n\nMy argument is that we have to entertain the prospect that the answer to at least some of these questions is Yes; that such residues are inevitable; and that they may not always be unfortunate. Jackie Goldsby tells how she and her partner have carefully examined their motives to make sure that neither of them is \"succumbing to internalized racism. We say this, even to ourselves, even though we know differently: where, in the context of lesbian political discourse on race, can we acknowledge that our knowingly crossing boundaries of race and class _is_ part of our desire for each other?\"\n\nFor many African American gay men, the most significant pressures come from their own black communities. William G. Hawkeswood reports that gay black men in Harlem generally live there \"because they prefer black men as friends and lovers and because they prefer to live closer to family and other relatives. Thus they avoid prolonged contact with whites,\" and rarely experience racism. They enjoy a generally supportive environment, despite occasional instances of verbal abuse, mostly from youths. \"Generally they feel that gays are more tolerated and better accepted in Harlem than they are in mainstream America,\" partly because there are more pressing issues (poverty, unemployment, poor education, teenage pregnancies, drugs, AIDS). This may be true. However, many of the contributors to major anthologies of African American gay experience\u2014 _In the Life_ , edited by Joseph Beam, and _Brother to Brother_ , edited by Essex Hemphill\u2014record anguish, alienation, and damage in their childhood and youth, as they suffered catcalls of sissy and faggot from within the black community. Indeed, they reveal a culture engrossed with the precariousness of male gendering. Reginald T. Jackson remarks that the taunt, \"faggot,\" preceded any knowledge of homosexuality: had he known that this might involve \"the love of a man. A black man,\" then he would gladly have endured the name-calling. In the terms broached earlier in this book, gender identification precedes object-choice; the stigma attaches to effeminacy.\n\nMarlon Riggs in his film _Tongues Untied_ (1989), written around Riggs's poem of the same name, expounds a personal negotiation of sexuality and racial difference. Riggs experienced oppression in his own community as a sissy, more immediately than oppression as a black man. The solution was part of the problem:\n\nA whiteboy came to my rescue.\n\nBeckoned with gray\/green eyes, a soft Tennessee drawl.\n\nSeduced me out of my adolescent silence.\n\nRiggs found his way to San Francisco\u2014the Castro, the gay district:\n\nI learned the touch and taste of snow.\n\nCruising whiteboys, I played out\n\nadolescent dreams deferred.\n\nPatterns of black upon white upon black upon white\n\nmesmerized me. I focused hard, concentrated deep.\n\nRiggs was playing out racial relations as sexual relations. But it dawned upon him that this was not making him the person he sought to be: in the Castro he was \"an invisible man\"; the proffered black images were racist stereotypes. \"I was a nigga, still.\" So he \"went in search of something better,\" he says, listening to \"Rhythms of blood, culture, \/ history, and race\": black men must love one other (203\u2013204). A concluding section of the film links civil rights marches with a black gay demonstration. The slogan on a banner is from an essay by Joseph Beam: \"Black men loving Black men is the revolutionary act of the eighties.\"\n\nHowever, Riggs admits, the \"absence of black images\" occurred not only in the books, posters, and films dominated by white people, but in his \"own fantasies\" ( _Tongues Untied_ , 202\u2013203). The film became controversial within black and gay communities when commentators noted that Riggs's personal life appeared to contradict his argument: his own partner was white. B. Ruby Rich objects: \"In a film full of the courage of coming out of the closet on the subject of queerness, it looked as though Riggs had stayed in the closet on the subject of race (as object of affection, not identity).\" In the credits Jack Vincent, the controversial partner, is thanked for his loving support; far less is at stake for the white man\u2014he can afford to be sympathetic.\n\nQuestions have been raised also about the gendering of gay men in _Tongues Untied_. Goldsby remarks that drag queens are depicted as pathetic, lonely figures, to be displaced by more macho images. Evidently, Riggs is under pressure from \"Afrocentric\" commentators who despise effeminacy:\n\nBecause of my sexuality, I cannot be black. A strong, proud, \"Afrocentric\" black man is resolutely heterosexual, not _even_ bisexual. Hence I remain a Negro. My sexual difference is considered of no value; indeed, it is a testament to weakness, passivity, the absence of real guts\u2014balls. Hence I remain a sissy, punk, faggot. I cannot be a black gay man because by the tenets of black macho, black gay man is a triple negation. I am consigned, by these tenets, to remain a Negro faggot.\n\nThe energy in this passage derives perhaps from an unstable combination of resistance to the way black manhood has been defined, and resentment at exclusion from that manhood. Is it that Riggs wants Afrocentrism to incorporate the sissy, or that he objects to being labeled a sissy? Gay-oriented commentators insist that emphasizing manliness is not the answer. Kobena Mercer calls it \"another turn of the screw of oppression... when black men subjectively internalise and incorporate aspects of the dominant definitions of masculinity in order to contest the conditions of dependency and powerlessness which racism and racial oppression enforce.\" Phillip Brian Harper, in comparable vein, rejects the \"imagined solution\": \"a proper affirmation of black male authority.\"\n\nFinally I note that Riggs's appeal to sameness as the resolution of the dilemmas of racism converges upon a leading theme of this book. He has rediscovered and reaffirmed, in his case for black men loving black men, not a distinctive African American way of relating, but a purer, sharper version of the egalitarian ideology, which asserts that the most productive relations will be founded in sameness. The Castro, Riggs sees, proclaims the community of gayness and the unity of gay experience, while actually reinscribing the racist assumptions of American society generally; the task proposed in the film, then, is to make the egalitarian ideology succeed among black men. Yet Riggs's work is more important for the struggle it exposes, than for the watchword with which it ends.\n\n**SLAVES AND MASTERS**\n\nAs I have said, S\/M is best regarded, not as a speciality, but as continuous with the hierarchies that we all experience. Invocation of cross-racial masters and slaves does, however, raise the odds. In \"Beneath the Veneer,\" in a recent collection of gay African American writing, Kheven L. LaGrone records his shock upon seeing at a party a white man leading a black man by a leash. Have black people forgotten the role of white supremacists in history? LaGrone demands. No they haven't: that is what the leash is about.\n\nIsaac Julien has made diverse approaches to these topics. His film _Looking for Langston_ (1988) suggests that sexual relations between blacks and whites were regarded quite positively in Harlem in the 1920s and 1930s. This interpretation has been supported by Kevin J. Mumford, who finds that there was \"a kind of affinity between homosexuals, black\/white sex districts, and African-American culture more generally.\" In a poem from around the same time as _Looking for Langston_ , \"Gary's Tale,\" Julien observes that race has infiltrated the psyche and infects every relationship:\n\nBecause the last fight, the last battle, territory, will be with one's self, the most important terrain, the psyche.\n\nThe mind will be the last neo-colonialised space to be decolonialised, this I know because I have been there, backwards and forwards.\n\nFor Julien, this is not a reason to avoid cross-racial contact, or to give up hopes of positive change. In _Young Soul Rebels_ (1991), the killer is a white boy who can't handle his attraction to black males. However, the film ends in utopian style, with a sexual liaison between Caz (black male) and Billibud (white male), another between Chris (black male) and Tracy (black female), a main-man comradeship between Caz and Chris, a special friendship between Tracy and Jill (white female), and everybody learning to funk.\n\nJulien's short film _The Attendant_ (1991) is set in an art gallery. The Attendant (Thomas Baptiste) is black; a visitor is white and wearing leather gear; they are drawn to each other. The gallery closes; the Attendant reimagines the paintings, replacing the original figures with men such as himself and the visitor, wearing contemporary leather gear. He focuses especially on an admired nineteenth-century painting by F. A. Biard, \"Scene on the African Coast,\" showing episodes in the slave trade. Two of the paintings become erotic cartoons by Tom of Finland. Other tableaux are composed of camp go-go boys. The Attendant whips the visitor. The Attendant sings in a theater\u2014\"Dido's Lament\" from Purcell's _Dido and Aeneas_ (Dido was abandoned on an African coast by an imperial visitor).\n\nJulien has been criticized for depicting a black man with his psyche organized around imagery of historic humiliation. He has replied that images cannot be trapped in one meaning: the implements of slavery have been transformed into \"sexualised, stylized fetish clothing for the queer body. The imperialist slave iconography is appropriated and repositioned.\" Gay subculture, including black gay subculture, makes its own use of the imagery. Furthermore, Julien adds, there are diverse ways of reading the film: it might be a parody, or transgressive; or it might be \"moralistically read into the cheap sociology of a pathological, black\/self-hating discourse.\" The sickness resides in puritanical rejection, not in gay sex.\n\nThe problem with Julien's argument is that he wants to assert an open plurality of readings while disallowing some readings. He is on stronger ground when he asserts that morbidity resides with those who would deny \"the psychic reality of black\/white desire.\" These desires exist, and they may be more appealing than the restricted and restricting vision of puritans. \"The out black snow queen draws attention to the fact of black desire for the white subject and contests pathological racial identities, the products of Afrocentric readings,\" Julien adds (125). He regrets what he regards as Riggs's retreat into Afrocentrism.\n\nHowever, Riggs in fact bears witness on both sides of the argument. We may perhaps envisage a dynamic process; one in which Afrocentrism might be an appropriate stage on the route to self-affirmation, while investment in cross-racial imagery might be productive of political insights, but not destructive of the relations that promote them. The goal might be a point, not where one or the other is the right answer, but where they can inform each other. Of course, as we have seen in Corbin's _Fragments That Remain_ and in Duplechan's _Eight Days a Week_ , this depends as much on the white partner as the black. However, it is likely to be the latter who places himself most at risk.\n\nThese issues echo powerfully in the journals of Gary Fisher, published at the instigation of Eve Sedgwick after his death in 1994. Fisher tells of his fantasies and his encounters, and of the link between them and race relations in the United States. He writes a letter to a sexual partner, Master Park ( _Slavery Defended_ is a classic collection of \"Views of the Old South\"): \"Here's that letter you wanted. I'm laying here sideways in the bed with _Slavery Defended_ opened to about midway, sampling the arguments and thinking about how good it felt to serve. Not that it matters, but I enjoyed Thursday immensely, particularly the sleaze and humiliation of some of it\" (extended pain he likes less). \"The racial humiliation is a huge turn-on. I enjoy being your nigger, your property and worshipping not just you, but your whiteness.\"\n\nThis is not a generalized masochism, nor just an individual thrill; it is about race, slavery, history, and skin color. In an earlier journal entry Fisher asks himself:\n\nHave I tried to oppress myself\u2014as a black man and as a (passive) homosexual man\u2014purely for the pleasure of it, or does that oppression go right to the point of my perceived weaknesses[?]... Can I divorce sexuality from power in the real world or do I want to[?]... I want to, in effect, give in to a system that wants to (has to) oppress me. (187; my question marks and elisions)\n\nIs his submission to whiteness a personal pleasure, he asks, or does it bespeak a personal weakness? Is it not, rather, systemic?\u2014the operations of real-world ideology in the psyche? Fisher is playing the part that is written for him in the racial script of the United States. Again: \"Blackness is a state of frustration. There's no way out of this racial depression (I don't feel the frustration personally, but as part of a people I know that I'm being fucked, abused)\" (199). Fisher experiences humiliating scenarios not just personally, but as the structural, racial exploitation which he knows to be their origin. In a manner of speaking, therefore, the fantasy is _not him_ : \"What is this fantasy that cuts across all of me, racial, intellectual, moral, spiritual, sexual...?\" (213; Fisher's hesitation).\n\nAs Sedgwick remarks in her Afterword to _Gary in Your Pocket_ , it has seemed necessary in our time to handle S\/M as a stigmatized minority practice, emphasizing the _dis_ linkages between sexual fantasy and activity, on the one hand, and \"the social realities of power and violence\" on the other. We are only playing, S\/Mers are expected to say, our games are entirely separate from real-world violence. A theme of this book is that such a dislinkage is impossible to maintain and ultimately misleading. Yet, as elsewhere, we reach a point at which a violent disjunction becomes so inflammatory that we have to revert to the domains of ethical and political responsibility. Fisher is troubled: \"Crandall calls me his nigger. He's much rougher with me than Ed and talks so genuinely I wonder if he doesn't believe what he's saying to me. It disturbs me to write what he says about carving 'White Power' into my flesh\" (217\u201318). In fact it is a question, how far abuse is what Fisher wants: \"I love snuggling up to him. We're on our sides. He has one arm beneath me and another round me. It's so warm.... I feel so safe next to him\" (157; my elision). Again: \"I want a loving master\/daddy\" (217). The desire for love and protection from a version of the abuser seems the key to this; the risk\u2014the thrill\u2014is in placing your trust in a person who might damage you. \"So, particularly (especially or primarily) a white man, when he holds and protects me from others like him, brings me an excitement which strangely and uniquely parallels that which he causes me when he threatens or frightens me\" (237).\n\nThis scenario seems to correlate with a persistent trope in black and white American gay writing: the violent and abusive father who gains excitement from punishing his son. Everyone he knows suffered some element of abuse in childhood, David Wojnarowicz says. He describes how his own father systematically beat all his family, and ordered David to play with his penis. He derives from this an abiding sexual preference: \"I have always been attracted to dangerous men, men whose gestures intimated the possibilities of violence, and I have always seduced them into states of gentle grace with my hands and lips.\" Rechy's narrator reports similarly in _City of Night_. Tom in Monette's _Halfway Home_ was tormented by his father and his brother, Brian, at the same time as being sexually drawn to the latter. This childhood experience has inhibited his love life, Tom believes: \"I admit I have mixed them up, Brian and the man I never had.\" Now, as he makes a new relationship with Gray, a calm, caring, older man, he revises his model of manliness: \" _This is what a brother is_.\" The eponymous heroes in _Tim and Pete_ both had hostile fathers; Joey's father was \"a monster.\" Gary Fisher reports a violent and abusive father. If his preferred mode of sexual expression mimics the relationship with his father, it also repudiates it. \"The ultimate slap was to let it loose that I'd let some other man, not my dad, have me.\" He does \"finally meet a white man not twisted by the color thing,\" but their prospects are spoiled when Fisher declares his HIV status ( _Gary in Your Pocket_ , 245\u201346).\n\n**PHOBIA**\n\nI have tried elsewhere to distinguish phobic and structural racism. For only some individuals is racism phobic; for the rest of us, it gets structured into the language, into the prevailing stories through which the society seeks to understand itself. It becomes \"common sense.\" Structural racism is certainly not innocent; it affords a sympathetic milieu into which phobic racism may expand, and in some cases the two virtually merge. However, in other circumstances it is at least ameliorable: when pointed out it may be worked upon and changed.\n\nPhobic racism, on the other hand, seeks to secure not just the economic, political, and general psychic well-being of the white man; the racial other is invoked as a way of handling profound personal inadequacy. The phobic racist cannot leave the topic\u2014he or she looks for people of other races in order to exercise phobic feelings. Probably there is a more or less permanent phobic minority, the recruiting ground for fascist movements. At a certain point, enthrallment becomes obsessive, grotesque, violent, intolerable. I may seem in danger here of exonerating the system, by making it responsible for habitual structures, while phobic intensity is ascribed to individuals. However, it seems to me that phobia is usually dysfunctional, not actually sustaining the system. Its irrationalism is too unpredictable, too loose, too dangerous.\n\nHomophobia, similarly, is a term that should not be used casually, for general anxiety and prejudice. It is best reserved for implacable, hostile fascination. Of course, racial and sexual phobias may occur together. The serial killer Jeffrey Dahmer was so drawn to his male victims, nearly all of whom were black or Asian, that he ate parts of their bodies. This certainly attests to a consuming desire-for; we might regard it as a monstrous parody of desire-to-be (the ingested person).\n\nThe gay tradition includes powerful and equivocal evocations of phobia. In Tennessee Williams' story, \"Desire and the Black Masseur,\" Anthony Burns, a nebulous and ineffective clerk (white) has been waiting, obscurely, for something to swallow him up. He finds it in the rough treatment of a black masseur (who isn't given a name).\n\nThe knowledge grew quickly between them of what Burns wanted, that he was in search of atonement, and the black masseur was the natural instrument of it. He hated white-skinned bodies because they abused his pride. He loved to have their white skin prone beneath him, to bring his fist or the palm of his hand down on its passive surface. He had barely been able to hold this love in restraint, to control the wish that he felt to pound more fiercely and use the full of his power. But now at long last the suitable person had entered his orbit of passion. In the white-collar clerk he had located all that he longed for.\n\nThis might be a mutually rewarding project\u2014\"The giant loved Burns, and Burns adored the giant\" (221)\u2014but the violence of the massage increases. While a preacher across the street invokes the atonement of Jesus, the masseur devours the body of Burns; he moves to another town and new customers.\n\nAll this, we are told, illustrates something about \"the earth's whole population\" as it \"twisted and writhed beneath the manipulation of night's black fingers and the white ones of day\" (223). The referent seems to be both color and humanity; guilt at historic racial exploitation is made to coincide with the Christian notion that everyone needs atonement. David Bergman emphasizes the racial implications: \"Like Melville, Williams believes that an egalitarian relation between so-called civilized men and their primitive brothers can be achieved only through an act of cannibalism in which the civilized will be consumed.\" Unfortunately, this is at the expense of reconfirming the black primitive. \"'What do you think this is? A jungle?'\" the masseur's boss demands, not altogether unreasonably, when Burns's leg is fractured (221).\n\nGetting eaten is the fate also of Sebastian in Williams' play _Suddenly Last Summer_ (1958, filmed by Joseph L. Mankiewicz in 1959), where the agon of homosexuality again seems to coalesce with the supposed human condition. In a mythic Latin country, Sebastian is consumed by the native boys whom he has courted\u2014\"a flock of featherless little black sparrows,\" with gobbling birds on the Galapagos Islands, that prey upon the newly hatched turtles as they try to reach the sea, making \"the sky almost as black as the beach.\" How far Williams is in control of this scenario is in my view a question. It seems to me that he is making a melodramatic projection of queer guilt onto the universe, and positioning black males as angels of death.\n\nJames Robert Baker's novel _Testosterone_ is, like _Tim and Pete_ , a quest which takes the reader back and forth on the L.A. freeways. Whereas Tim and Pete mean well, Dean's mission is partly to get back together with Pablo Ortega so that they can share an ideal future\u2014but mainly to kill him. This is because Pablo abruptly terminated their relationship. Dean doesn't consider whether he might have turned Pablo off in some way (Dean emerges as very strange), but interprets Pablo as an emotional serial killer. He tracks down Pablo's other lovers and finds ample confirmation of his malevolence. How much of this is true the reader cannot tell, since we have only Dean's word for it and he is evidently obsessive. He believes Pablo burned his house down, and took his dog for medical experiments or black magic. Anne, who says she had an affair with Pablo, claims he was a notorious doctor who tortured political prisoners in Chile in 1987, and Dean believes this, although Pablo would have been about twenty at the time. Anne retracts this story, but Dean goes on believing it. Then he credits another story, that Pablo was part of a notorious drug gang in Mexico, also in 1987. He becomes out of control and violent with people who frustrate him in any way. He attacks a man who looks rather like Pablo (he's not wearing his glasses). With increasing abandon, he imagines what Pablo might be saying or thinking.\n\nWhat gradually becomes apparent is that Dean is preoccupied with Pablo's Latino race. Initially he reassures his friends that Pablo is not stereotypical: \"He's Latino, but he doesn't have an accent (like your typical dumb beaner).\" Indeed, the stereotypes are reversed: Pablo is cool and dispassionate, Dean is spontaneous, expressive. This becomes an accusation: \"I think his brown skin fooled me at first, so I didn't realize how much he was really like my father\" (16). A sense of Dean as phobic develops when he laments the old days, when Southern California was a white boy's utopia, before \"the killers of color shot you in the head and took your T-bird away.\" He has been noting contact ads in which Latinos offer to mistreat white boys, whereas the other way around would not be allowed because of \"PC.\" \"But I've made an effort through all this not to get racist. And I haven't. I really don't think I have\" (51\u201352). This is a love\/hate thing: Pablo's \"skin was really amazing. So smooth and warm\" (58); Dean imagines himself having a \"faithful, young, smooth, brown-skinned fellow lifeguard\/slave\" (28).\n\nThe idea that Pablo has killed Dean's dog, either for scientific experiment or for Palo Mayombe black-magic rituals, depends on superstitious, racist notions of what Latino people are like: \"Latino's do have this different feeling about cruelty to animals. I know that's a blanket statement.... Call me a racist if you want to, but it fucking happens. That's why it's taking a great effort on my part not to think all Latinos are sick\" (94; my elision). Pablo becomes the dog: \"He's like a mad cornered South American dog\" (84). Dean believes a hairdresser who says that killing Pablo would not be enough to remove his curse; he takes the fact that the Sam Peckinpah horror movie _Bring Me the Head of Alfredo Garcia_ is showing on TV as a sign that he must behead Pablo. He steals a machete and an ice chest to store the head. It is Dean, the white man from a Presbyterian background, who turns out to be the superstitious killer. The last sentence of the book shows Dean fixed in his obsession: \"The smooth warmth of his brown skin\" (200).\n\n**SHAME**\n\nEve Sedgwick put the concept of _shame_ into gay circulation in 1993, in an article heading up the new journal _GLQ_ ( _Gay and Lesbian Quarterly_ ). Her title, \"Queer Performativity: Henry James's _The Art of the Novel_ ,\" announced a recognition of Judith Butler's work (performativity had been Butler's key contribution at that date), while moving, via James, toward a more inward, literary, intimate, and intense register. Sedgwick presents shaming as a condition of queer sexuality, and as constitutive of queer identity. Butler, in response, acknowledged an element of shame in performing queerness, while drawing the argument into collective and practical aspects of identity management and activism.\n\nI had not seen much potential for the development of shame; as Sedgwick remarks, it seems negative to dwell on the terms of our stigma. Then I noticed a line in Tennessee Williams' play, _Sweet Bird of Youth_ , and its quotation in Neil Bartlett's novel, _Ready to Catch Him Should He Fall_. The Princess (an aging movie star) has been fencing verbally with Chance (a still-youthful hustler). In the last line of the scene she appeals to him: \"Now get a little sweet music on the radio and come here to me and make me almost believe that we're a pair of young lovers _without any shame_.\" Good lovemaking retrieves youth and confers emancipation from shame\u2014 _almost_. As I conclude this chapter on race\u2014the most sensitive, threatening, and exciting hierarchy\u2014it seems to me that, while shame is indeed loaded upon us by our societies, that does not altogether account for the cultural energy that attends it. In a parallel movement (for social structures, once more, are continuous with psychic structures) we generate shame among ourselves.\n\nAs I have said, people on the downside of the binary model are under the greater pressure. They put more at risk. Yet to exercise dominance also is to expose oneself; even the relatively powerful place themselves in jeopardy\u2014of rejection, humiliation, and the sudden inundation of unwelcome self-knowledge. Shame, I suggest, derives from _awareness of exploitation_ \u2014both for the boy who turns over, and for the man who takes advantage of his willingness. It is a product of the hierarchies I have been discussing; it is integral with sexual passion, therefore; it is sexy. If this view is correct, the shame of exploiting and being exploited will be hard to exclude or evade. Maybe there are magical remedies\u2014youth and beauty, according to Williams and Bartlett. And something for the older and less beautiful person to contribute\u2014perhaps strength, accomplishment, style, generosity, validation, trust.\n**9**\n\n**FICTION**\n\nI started out on this book with some ideas about desire and power. However, the great pleasure in the work has been encountering so many marvelous lesbian- and gay-themed novels and films. I have been delighted with the seriousness and humor, the inventiveness and responsibility, the imagination and abundance that have been revealed. As I say in chapter 1, my use of such resources is not designed to deduce facts from fiction, but to investigate the kinds of representations that have been circulating. My aim has been to explore how our experiences have fed into books, films, and cultural commentary, and how we, in turn, have read and pondered them, recognized and reframed ourselves through them. The pleasure in reading has derived partly from artistic achievement, but mainly it has been about registering the kinds of people that we have become, or aspired to become. If you read Leavitt or Monette differently from me, this will almost certainly be, not just a literary judgment, but a way of thinking about sexuality and power. We may have our most ambitious conversations and contests through fiction.\n\nI am scarcely concerned, therefore, with whether a book is likely to become part of a literary canon. At the same time, if it transpires that the processes of subcultural exploration and self-recognition are effective, in part at least, and other readers and commentators frame their ideas through some of these texts, the outcome will _in practice_ be something rather like a _lesbian and gay canon_. Now, insofar as canon means \"law, rule, edict,\" that is not what I have in mind. However, \"canonical\" also means books that are accepted as \"genuine and inspired,\" and those we have.\n\nIn effect, this is what gay people have always done. There was little lesbian fiction around when Alison Hennegan was young, but she cultivated her own investment in the classics. A passionate identification with Achilles, distraught at the death of Patroclus, helped \"by offering me a world free from the assumption that human completeness exists solely in the fusion of male and female.\" Hennegan went on to Petronius, Horace, and Virgil, and subsequently to Siegfried Sassoon and Wilfred Owen, gay poets of the Great War. \"I did as most young gay and lesbian readers of the time\u2014older ones, too, for that matter\u2014had to do. I created my own 'popular fiction,' developed my own much cherished canon.\"\n\nMy ambition to make the discussion of texts subculturally effective has been challenged by Stanley Fish, who takes me as exemplifying the approach of \"The Cultural Critic.\" This arises out of my book _Faultlines_ , where I agree with Fish that it is the academic profession that determines which readings will pass as plausible, but accuse him of wanting to head off such political potential as English literature may have. Fish responds by reiterating his assertion, that it must be futile for people working with literature to aspire to political influence:\n\nChanging the mode of literary analysis or changing the object of literary analysis or changing the name of literary analysis will not change the material effectiveness of literary analysis and make it into an instrument of political action. That kind of change, if it is ever to occur, will require wholesale _structural_ changes of which literary analysts might take advantage, but which they could never initiate.\n\nAt one level this is plainly correct. Literary criticism is designed, precisely, to head off any real-world engagement that literary intellectuals might seek. The application of the categories \"art\" and \"literature\" amounts to a way of distracting us from contemporary issues (which we might do something about), by asserting that the important writing will be \"universal\" (and hence probably beyond remedy). The overthrow of such deep-set ideas would indeed require a notable structural change. However, feminists, for instance, have succeeded in writing novels and criticism that both create and respond to urgent political issues. I discern in Fish's pronouncement a typical conservative strategy, whereby the terms for significant action are set at such a demanding level that they will never occur. Revolution appears unlikely, so it is not worth trying to change anything. In my view, writing of all kinds may change the boundaries of the plausible, and hence of the effective scope for action, and it may be especially valuable in subcultural formations.\n\nThe most considered discussion from a radical perspective of minority reading has been made by John Guillory in _Cultural Capital_. He believes that the goal of minority groups must be to place a sample of their culture in mainstream venues\u2014\"opening out the canon.\" Such a process, Guillory says, falls within the American tradition of liberal pluralism: through _representation_ , groups who have been excluded from full citizenship expect to make their presence felt. Insofar as this has the implicit aim of redressing the inadequacies of public political process, and of universities in particular, Guillory argues, it can be only an imaginary politics. Installing minority groups in canonical venues gestures toward a national oppression which it cannot affect.\n\nPartly for this reason, Guillory adds, the valorization of noncanonical texts tends to depend upon theoretically unsophisticated notions, such as the identity and authentic experience of the author, and reading is imagined, naively, as a transparent process. It is not inevitable, however, that the purposive invocation of texts by and about marginalized groups will succumb to these theoretical pitfalls. Compare the situation of racial minorities. Commentators such as Stuart Hall, bell hooks, and Paul Gilroy have pointed out that there can be no essential grounding for racial and ethnic identities. In an initial stage, Hall notes, black people sought to make their own images, challenging hegemonic versions of themselves; but today it is understood that representation is formative\u2014active, constitutive\u2014rather than mimetic. \"Black,\" according to Hall, \"is essentially a politically and culturally _constructed_ category, which cannot be grounded in a set of fixed transcultural or transcendental racial categories and which therefore has no guarantees in Nature.\" It is the same with lesbian, gay, and bisexual. There was a time for positive images\u2014indeed, for visibility of any kind. Today a developed theory of ideology, the human subject, and cultural production affords opportunities for new engagements with textuality, and with the anxieties and hopes of a subcultural constituency.\n\nIndeed, it is awareness of identity as constituted that affords opportunities for intervention. One inference from antiessentialist theory should be that we cannot simply throw off our current constructions. We are consequences of our histories\u2014those that have been forced upon us and those that we have made ourselves. At the same time, it is because we believe that culture constructs the scope for our identities that we may believe those identities to be contingent and provisional, and therefore may strive to revise our own self-understanding and representation. If gay subculture is effective for its constituency, it is not because it evades poststructuralist insights, but because it responds to them.\n\nThe concept of \"cultural capital\" from which Guillory begins (in my view a very useful concept) leads him to present a static impression of how a group inhabits a culture. While it is true that mainstream cultures are under pressure to incorporate their treasures into authoritative state and national monoliths, subordinated groups have more urgent and particular concerns. For lesbians and gay men, though some ground may be gained by remarking how traditionally canonical authors have displayed a significant streak of homoeroticism, it is not important simply to possess this or that statusful icon.\n\nThe reply to Guillory is that the success of subcultural intervention may not reside primarily, or even at all, in infiltrating the mainstream. Indeed, insofar as that occurs, a consequence is usually distortion or appropriation; we are used to coping with hostile and patronizing images. The quality of our canon is not to be measured by the extent to which it impresses the mainstream, but in terms of its contribution to shared self-understanding. The point is to mark out a space in which to compare stories, for consolation, insight, and new imaginative awareness.\n**NOTES**\n\n**1. INTRODUCTION**\n\n. Reginald Shepherd, \"On Not Being White,\" in Joseph Beam, ed., _In the Life_ (Boston: Alyson, 1986), 53\u201354. Subsequent references are in the text. On white-on-black fixation, see Christopher Cutrone, \"The Child with a Lion: The Utopia of Interracial Intimacy,\" _GLQ_ ( _Gay and Lesbian Quarterly_ ) 6 (2000): 249\u201385.\n\n. Jonathan Dollimore, _Sexual Dissidence_ (Oxford: Clarendon, 1991), 65.\n\n. See R. D. Laing and D. G. Cooper, _Reason and Violence_ (New York: Pantheon, 1971), 22\u201325.\n\n. See Alan Sinfield, _Cultural Politics\u2014Queer Reading_ (1994), 2d ed. (London: Routledge, 2005), chs. 1 and 2.\n\n. Louis Althusser, _Lenin and Philosophy_ , trans. Ben Brewster (London: New Left, 1997), 160\u201365.\n\n. Althusser, _Lenin and Philosophy_ , 190\u201391, 200.\n\n. Herbert Marcuse, _Eros and Civilization_ (New York: Random House, 1955).\n\n. Michel Foucault, _The History of Sexuality_ , vol. 1: _An Introduction_ , trans. Robert Hurley (New York: Vintage, 1978), 11. I return to Foucault and power in chapter 4.\n\n. Joan Riviere, \"Womanliness as a Masquerade\" (1929), in Victor Burgin, James Donald, and Cora Kaplan, eds., _Formations of Fantasy_ (London: Methuen, 1986), 37\u201338. See Vicky Lebeau, _Psychoanalysis and Cinema_ (London: Wallflower, 2001), 101\u201315.\n\n. Michael Warner, \"Homo-Narcissism; Or, Heterosexuality,\" in Joseph A. Boone and Michael Cadden, eds., _Engendering Men_ (New York: Routledge, 1990), 200.\n\n. Lynne Segal, _Straight Sex_ (London: Virago, 1994), 135.\n\n. For a defense and referencing of recent work, see Christopher Lane, \"Psychoanalysis and Sexual Identity,\" in Andy Medhurst and Sally R. Munt, eds., _Lesbian and Gay Studies_ (London: Cassell, 1997).\n\n. Alan Sinfield, _Out on Stage_ (New Haven and London: Yale University Press, 1999).\n\n. Bertolt Brecht, \"A Short Organum for the Theatre,\" in John Willett, ed. and trans., _Brecht on Theatre_ (London: Eyre Methuen, 1978), 186.\n\n. Jim Grimsley, interview in _Gay Times_ 288 (September 2002): 77.\n\n. Alan Sinfield, _Faultlines_ (Berkeley: University of California Press, 1992; Oxford: Clarendon Press, 1992), ch. 2.\n\n. The terms _erotic dissidence_ , _dissident sexuality_ , and _sexual dissidence_ are used for forbidden and\/or stigmatized sex by Gayle S. Rubin in 1982. See Rubin, \"Thinking Sex: Notes for a Radical Theory of the Politics of Sexuality,\" in Henry Abelove, Mich\u00e8le Aina Barale, and David M. Halperin, eds., _The Lesbian and Gay Studies Reader_ (New York: Routledge, 1993), 18, 22, 23. _Sexual Dissidence_ is, of course, the title of Jonathan Dollimore's book.\n\n. Alan Sinfield, _Gay and After_ (London: Serpent's Tail, 1998), ch. 3; Sinfield, \"The Production of Gay and the Return of Power,\" in Richard Phillips, Diane Watt, and David Shuttleton, eds., _De-centring Sexualities_ (London: Routledge, 2000); Sinfield, \"Transgender and Les\/bi\/gay Identities,\" in David Alderson and Linda Anderson, eds., _Territories of Desire in Queer Culture_ (Manchester: Manchester University Press, 2000).\n\n. Sinfield, _Gay and After_ , 91, 103.\n\n**2. TAXONOMIES**\n\n. David M. Halperin, _How to Do the History of Homosexuality_ (Chicago: University of Chicago Press, 2002), 109\u2013110.\n\n. Halperin, _How to Do the History_ , 133\u201334.\n\n. David Halperin, \"Pal o' Me Heart,\" _London Review of Books_ , May 22, 2003, 32\u201333. See Jamie O'Neill, _At Swim, Two Boys_ (London: Scribner, 2001).\n\n. Halperin, _How to Do the History_ , 110, 134.\n\n. Bruce R. Smith, _Homosexual Desire in Shakespeare's England_ (Chicago: University of Chicago Press, 1991); G. S. Rousseau, _Perilous Enlightenment_ (Manchester: Manchester University Press, 1991), 9\u201313.\n\n. See David F. Greenberg, _The Construction of Homosexuality_ (Chicago: University of Chicago Press, 1988); Stephen O. Murray, \"The 'Underdevelopment' of Modern\/Gay Homosexuality in MesoAmerica,\" in Kenneth Plummer, ed., _Modern Homosexualities_ (London: Routledge, 1992); Gilbert Herdt, _Same Sex, Different Cultures_ (Boulder, Colo.: Westfield, 1997).\n\n. Jeffrey Weeks, _Sexuality and Its Discontents_ (London: Routledge, 1985), 90.\n\n. Eve Kosofsky Sedgwick, _Epistemology of the Closet_ (Hemel Hempstead: Harvester Wheatsheaf, 1991), 23 (Sedgwick's emphasis).\n\n. Kenneth Lewes, _The Psychoanalytic Theory of Male Homosexuality_ (London: Quartet, 1989), 35\u201342; C. A. Tripp, _The Homosexual Matrix_ (1975) 2d ed. (New York: Meridian, 1987), 72\u201373; Kaja Silverman, _Male Subjectivity at the Margins_ (New York: Routledge, 1992), ch. 8 (Silverman's third category has two subdivisions).\n\n. Sigmund Freud, _Three Essays on the Theory of Sexuality_ (1905), in _The Penguin Freud Library_ , vol. 7: _On Sexuality_ (Harmondsworth: Penguin, 1977), 56.\n\n. Sigmund Freud, _Leonardo da Vinci_ (1910), in _The Penguin Freud Library_ , vol. 14: _Art and Literature_ (Harmondsworth: Penguin, 1985). For a critique of this essay, see Earl Jackson Jr., _Strategies of Deviance_ (Bloomington: Indiana University Press, 1995), 53\u201373.\n\n. E. M. Forster, _Maurice_ (1971) (Harmondsworth: Penguin, 1972), 208.\n\n. Silverman, _Male Subjectivity at the Margins_ , 344. However, Silverman also gestures toward the importance of class, age, and nationality, remarking: \"Because their object-choice defies the libidinal logic of conventional masculinity, gay men are frequently viewed through the alternative screen of femininity\" (353).\n\n. Sheila Jeffreys, \"Butch and Femme: Now and Then,\" in Lesbian History Group, _Not a Passing Phase_ (London: Women's Press, 1993).\n\n. Teresa de Lauretis, _The Practice of Love_ (Bloomington: Indiana University Press, 1994), 240.\n\n. Elizabeth Grosz, _Space, Time, and Perversion_ (New York: Routledge, 1995), ch. 10.\n\n. Rita Mae Brown, _Rubyfruit Jungle_ (1973) (London: Corgi, 1978), 147. See Jonathan Dollimore, _Sexual Dissidence_ (Oxford: Clarendon, 1991), 52\u201355.\n\n. Judith Butler, \"Imitation and Gender Insubordination,\" in Diana Fuss, ed., _Inside\/Out_ (New York: Routledge, 1991), 20.\n\n. Freud's letter is quoted in Lewes, _The Psychoanalytic Theory of Male Homosexuality_ , 32.\n\n. Michael Warner, \"Homo-Narcissism; Or, Heterosexuality,\" in Joseph A. Boone and Michael Cadden, eds., _Engendering Men_ (New York: Routledge, 1990), 192.\n\n. Carole-Anne Tyler, \"Boys Will Be Girls: The Politics of Gay Drag,\" in Diana Fuss, ed., _Inside\/Out_ , 34.\n\n. Sigmund Freud, \"Female Sexuality\" (1931), in Freud, _The Penguin Freud Library_ , vol. 7, _On Sexuality_ , 376\u201377 (my elision).\n\n. Sigmund Freud, \"On Narcissism: An Introduction\" (1914), in _The Penguin Freud Library_ , vol. 11: _On Metapsychology_ (Harmondsworth: Penguin, 1984), 84. An editor's note directs readers to Freud's essay on Leonardo.\n\n. Freud, \"On Narcissism,\" 81\u201382 (my elision). Tim Dean argues that narcissism need not exclude otherness: Dean, \"Homosexuality and the Problem of Otherness,\" in Tim Dean and Christopher Lane, eds., _Homosexuality and Psychoanalysis_ (Chicago: University of Chicago Press, 2001).\n\n. Lewes, _The Psychoanalytic Theory of Male Homosexuality_ , 72.\n\n. Audre Lorde, _Zami: A New Spelling of My Name_ (1982) (London: Pandora, 1996), 160, 165.\n\n. Freud, \"Group Psychology and the Analysis of the Ego,\" in _The Penguin Freud Library_ , vol. 12: _Civilization, Society, and Religion_ (Harmondsworth: Penguin, 1985), 134.\n\n. Freud, \"Group Psychology,\" 134.\n\n. Sedgwick, _Epistemology of the Closet_ , 62.\n\n. Judith Butler, _Bodies That Matter_ (New York: Routledge, 1993), 239.\n\n. Wayne Koestenbaum, _The Queen's Throat_ (New York: Poseidon, 1993), 18.\n\n. John Fletcher, \"Freud and His Uses: Psychoanalysis and Gay Theory,\" in Simon Shepherd and Mick Wallis, eds., _Coming On Strong_ (London: Unwin Hyman, 1989), 99. See Warner, \"Homo-Narcissism; Or, Heterosexuality,\" in Boone and Cadden, eds., _Engendering Men_ , 197\u201398.\n\n. Sedgwick, _Epistemology of the Closet_ , 86\u201390.\n\n. Ned Cresswell, _A Hollywood Conscience_ (Brighton: Millivres, n.d.), 202.\n\n. James Robert Baker, _Tim and Pete_ (1995) (London: Fourth Estate, 1996), 38 (my elision). Further on this novel, see below and ch. 4.\n\n. Quentin Crisp, _The Naked Civil Servant_ (1968) (New York: Plume, 1977), 56.\n\n. Alan Sinfield, _Gay and After_ (London: Serpent's Tail, 1998), ch. 3.\n\n. Jay Prosser explains that \"transgender\" was used initially to denote a stronger commitment to living as a woman than \"transvestite\" or \"cross-dresser,\" and without the implications for sexuality in \"transsexual.\" However, the tendency now is to use \"transgender\" in a coalitionary politics, to include all those subjects. In this essay I do the latter while retaining an emphasis from the former. See Jay Prosser, \"Transgender,\" in Andy Medhurst and Sally R. Munt, eds., _Lesbian and Gay Studies_ (London: Cassell, 1997); Prosser, _Second Skins_ (New York: Columbia University Press, 1998), 176.\n\n. Leslie Feinberg, _Stone Butch Blues_ (Ithaca: Firebrand, 1993), 143.\n\n. Judith Halberstam, \"F2M: The Making of Female Masculinity,\" in Laura Doan, ed., _The Lesbian Postmodern_ (New York: Columbia University Press, 1994), 212.\n\n. Don Kulick, \"A Man in the House: The Boyfriends of Brazilian _Travesti_ Prostitutes,\" _Social Text_ 52\u201353 (1997): 133\u201360.\n\n. Prosser, _Second Skins_ , ch. 5; Kate Bornstein, _Gender Outlaw_ (New York: Routledge, 1994), 4. See Alan Sinfield, \"Transgender and Les\/bi\/gay Identities,\" in David Alderson and Linda Anderson, eds., _Territories of Desire in Queer Culture_ (Manchester: Manchester University Press, 2000).\n\n. Feinberg, _Stone Butch Blues_ , 147.\n\n. See David Valentine and Riki Anne Wilchins, \"One Percent on the Burn Chart: Gender, Genitals, and Hermaphrodites with Attitude,\" _Social Text_ 52\u201353 (1997): 215\u201322; Anne Fausto-Sterling, \"How to Build a Man,\" in Vernon A. Rosario, ed., _Science and Homosexualities_ (New York: Routledge, 1997); Cheryl Chase, \"Hermaphrodites with Attitude: Mapping the Emergence of Intersex Political Activism,\" _GLQ_ ( _Gay and Lesbian Quarterly_ ) 4 (1998): 189\u2013211; Iain Morland, \"Is Intersexuality Real?\" _Textual Practice_ 15 (2001): 527\u201347.\n\n. Clifford Geertz, _Local Knowledge_ (New York: Basic Books, 1983).\n\n. See Sally R. Munt, ed., _Butch\/Femme_ (London: Cassell, 1998), 1, 41, 105, 143, 154, 159.\n\n. Judith Butler, _Gender Trouble_ (New York and London: Routledge, 1990), 123.\n\n. Halberstam, \"F2M: The Making of Female Masculinity,\" 220.\n\n. Stephen Maddison, _Fags, Hags, and Queer Sisters_ (London: Macmillan, 2000), 191\u201392. This topic is much disputed; see Zachary I. Nataf, _Lesbians Talk Transgender_ (London: Scarlet Press, 1996), 35\u201354.\n\n. Sigmund Freud, \"A Case of Homosexuality in a Woman\" (1920), in _The Penguin Freud Library_ , vol. 9: _Case Histories II_ (Harmondsworth: Penguin, 1979), 383\u201384.\n\n. Freud, \"Group Psychology,\" 138.\n\n. Judith Halberstam, _Female Masculinity_ (Durham: Duke University Press, 1998), 57.\n\n. Maddison, _Fags, Hags, and Queer Sisters_ , 12.\n\n. Halberstam, _Female Masculinity_ , 57\u201358.\n\n. See Alan Sinfield, _The Wilde Century_ (New York: Columbia University Press, 1994), ch. 6; Sinfield, _Out on Stage_ (New Haven and London: Yale University Press, 1999), 154\u201377.\n\n. Prosser, _Second Skins_ , 165.\n\n. Baker, _Tim and Pete_ , 35, 37.\n\n. Felice Picano, _Like People in History_ (1995) (London: Abacus, 1996), 50.\n\n. Edmund White, _The Beautiful Room Is Empty_ (London: Picador, 1988), 104.\n\n. Teresa de Lauretis, \"The Essence of the Triangle; Or, Taking the Risk of Essentialism Seriously: Feminist Theory in Italy, the U.S., and Britain,\" in Naomi Schor and Elizabeth Weed, eds., _The Essential Difference_ (Bloomington: Indiana University Press, 1994), 24.\n\n. Freud, _Three Essays_ , 56.\n\n. See Arlene Stein, _Sex and Sensibility_ (Berkeley: University of California Press, 1997); Lisa Power, \"Forbidden Fruit,\" in Mark Simpson, ed., _Anti-Gay_ (London: Cassell, 1996).\n\n. Aiden Shaw, _Wasted_ (London: Gay Men's Press, 2001), 240.\n\n. Sarah Schulman, _Empathy_ (1992) (London: Sheba Feminist Press, 1993), 4.\n\n. David T. Evans, _Sexual Citizenship_ (London: Routledge, 1993), 45. See Donald Morton, \"Queerity and Ludic Sado-Masochism: Compulsory Consumption and the Emerging Post-al Queer,\" in Mas'ud Zavarzadeh, Terese L. Ebert, and Donald Morton, eds., _Post-ality: Marxism and Postmodernism_ (Washington, D.C.: Maisonneuve Press, 1995), 189\u2013215; Sinfield, _Gay and After_ , ch. 9.\n\n. Feinberg, _Stone Butch Blues_ , 148.\n\n**3. FANTASY**\n\n. Esther Newton and Shirley Walton, \"The Misunderstanding: Toward a More Precise Sexual Vocabulary,\" in Carole S. Vance, ed., _Pleasure and Danger_ (London: Routledge, 1984), 250.\n\n. Vicky Lebeau, _Psychoanalysis and Cinema_ (London: Wallflower, 2001), 29.\n\n. Leo Bersani, _Homos_ (Cambridge: Harvard University Press, 1995), 103\u2013104.\n\n. See Leslie J. Moran, _The Homosexual(ity) of the Law_ (London: Routledge, 1996), 180\u201391.\n\n. Laura Mulvey, \"Visual Pleasure and Narrative Cinema,\" in Antony Easthope, ed., _Contemporary Film Theory_ (London: Longman, 1993).\n\n. For Mulvey's later comments and a full debate, see Easthope, ed., _Contemporary Film Theory_. For an independent-minded assessment, see Brett Farmer, _Spectacular Passions_ (Durham: Duke University Press, 2000). Teresa de Lauretis points out that there are important differences between film and fantasy: de Lauretis, _The Practice of Love_ (Bloomington: Indiana University Press, 1994), 148.\n\n. Jean Laplanche and Jean-Bertrand Pontalis, \"Fantasy and the Origins of Sexuality,\" in Victor Burgin, James Donald, and Cora Kaplan, eds., _Formations of Fantasy_ (London: Methuen, 1986), 26, 22\u201323 (their emphases).\n\n. Sigmund Freud, \"'A Child Is Being Beaten'\" (1919), in _The Penguin Freud Library_ , vol. 10: _On Psychopathology_ (Harmondsworth: Penguin, 1979).\n\n. James Robert Baker, _Boy Wonder_ (1988) (London: Fourth Estate, 1996), 40 (Baker's emphases).\n\n. Chris Straayer, _Deviant Eyes, Deviant Bodies_ (New York: Columbia University Press, 1996), 18\u201322.\n\n. Dorothy Allison, _Skin_ (London: Pandora, 1995), 109\u201310.\n\n. Constance Penley, \"Feminism, Psychoanalysis, and the Study of Popular Culture,\" in Lawrence Grossberg, Cary Nelson, and Paula A. Treichler, eds., _Cultural Studies_ (New York: Routledge, 1992), 489. _Star Trek_ is discussed also in Joanna Russ, _Magic Mommas, Trembling Sisters, Puritans, and Perverts_ (Freedom, Calif.: Crossing Press: 1985), and in Allison, _Skin_ , 95\u201397.\n\n. Richard Dyer, _Heavenly Bodies_ (London: Macmillan, 1987), 168, 155\u201356.\n\n. Stephen Maddison, _Fags, Hags, and Queer Sisters_ (London: Macmillan, 2000), 6.\n\n. David Wojnarowicz, _Close to the Knives_ (1991) (London: Serpent's Tail, 1992), 26.\n\n. Lynne Segal, _Straight Sex_ (London: Virago, 1994), 233; see 233\u201345.\n\n. Cora Kaplan, \"'A Cavern Opened in My Mind': The Poetics of Homosexuality and the Politics of Masculinity in James Baldwin,\" in Marcellus Blount and George P. Cunningham, eds., _Representing Black Men_ (London: Routledge, 1996), 32.\n\n. Cheryl Clarke, \"Living the Texts _Out_ : Lesbians and the Uses of Black Women's Traditions,\" in Stanlie M. James and Abena P. A. Busia, eds., _Theorizing Black Feminisms_ (New York: Routledge, 1993), 214.\n\n. Bia Lowe, \"Mothers and Others, But Also Brothers,\" in Joan Nestle and John Preston, eds., _Sister and Brother_ (London: Cassell, 1994), 127\u201328.\n\n. Sue-Ellen Case, \"Tracking the Vampire,\" _Differences_ 3.2 (Summer 1991): 1\u201320 (quote at 1).\n\n. Valerie Walkerdine, \"Video Replay: Families, Films, and Fantasy,\" in Burgin, Donald, and Kaplan, eds., _Formations of Fantasy_ , 169 (Walkerdine's emphasis; my elision).\n\n. Kaplan, \"'A Cavern Opened in My Mind,'\" 30.\n\n. Segal, _Straight Sex_ , 234.\n\n. Pai Hsien-yung, _Crystal Boys_ (1990), trans. Howard Goldblatt (San Francisco: Gay Sunshine Press, 1995), 27, 100. See Alan Sinfield, _Gay and After_ (London: Serpent's Tail, 1998), 59\u201368.\n\n. Alec Waugh, _Public School Life_ (London: Collins, 1922), 137\u201338.\n\n. See Alan Sinfield, _Out on Stage_ (New Haven and London: Yale University Press, 1999), 124\u201325.\n\n. Freud, _Three Essays on the Theory of Sexuality_ (1905), in _The Penguin Freud Library_ , vol. 7: _On Sexuality_ (Harmondsworth: Penguin, 1977), 56.\n\n. Michel Foucault, _The Uses of Pleasure_ , trans. Robert Hurley (Harmondsworth: Viking, 1986), 200.\n\n. See David M. Halperin, _How to Do the History of Homosexuality_ (Chicago: Chicago University Press, 2002), 188\u201389.\n\n. Edmund White, _The Farewell Symphony_ (London: Chatto, 1997), 402.\n\n. Jacqueline Rose, _The Haunting of Sylvia Plath_ (London: Virago, 1991), 210.\n\n. Reginald Shepherd, \"On Not Being White,\" in Joseph Beam, ed., _In the Life_ (Boston: Alyson, 1986), 53\u201354.\n\n. Gary Fisher, _Gary in Your Pocket_ , ed. Eve Kosofsky Sedgwick (Durham and London: Duke University Press, 1996), 203.\n\n. Earl Jackson Jr., _Strategies of Deviance_ (Bloomington: Indiana University Press, 1995), 173, 132.\n\n. See Segal, _Straight Sex_ , 282\u201397.\n\n. Robert Chesley, _Jerker, or the Helping Hand_ (1986), in Chesley, _Hard Plays \/ Stiff Parts_ (San Francisco: Alamo Square, 1990), 112 (Chesney's emphases [ _sic_ ]).\n\n. Edmund White, _A Boy's Own Story_ (1982) (London: Picador, 1983), 162.\n\n. Jonathan Dollimore, \"Bisexuality, Heterosexuality, and Wishful Theory,\" _Textual Practice_ 10 (1996): 523\u201339 (quote at 529; Dollimore's emphasis). For a slightly different version, see Dollimore, _Sex, Literature, and Censorship_ (Cambridge, Eng.: Polity, 2001), 28-9.\n\n. Guy Willard, _Mirrors of Narcissus_ (London: Millivres, 2000), 20\u201321, 25 (my elision).\n\n. Constance Penley, \"Time Travel, Primal Scene, and the Critical Dystopia,\" in James Donald, ed., _Fantasy and the Cinema_ (London: British Film Institute, 1989), 202.\n\n. De Lauretis, _The Practice of Love_ , 140.\n\n. Judith Butler, _Gender Trouble_ (New York and London: Routledge, 1990), 149.\n\n. Henning Bech, _When Men Meet_ , trans. Teresa Mesquit and Tim Davies (Cambridge, Eng.: Polity, 1997), 217 (Bech's emphases; my elision).\n\n. John Clarke, \"Style,\" in Stuart Hall and Tony Jefferson, eds., _Resistance Through Rituals_ (London and Birmingham: Hutchinson\/Centre for Contemporary Cultural Studies, 1976), 177.\n\n. R. D. Laing, _Self and Others_ (Harmondsworth: Penguin, 1971), 136.\n\n. Dollimore, _Sex, Literature, and Censorship_ , 56.\n\n. Andrew Holleran, _Dancer from the Dance_ (1978) (London: Cape, 1979), 146.\n\n. Segal, _Straight Sex_ , 161.\n\n. Jeanette Winterson, _The PowerBook_ (2000) (London: Vintage, 2001), 4\u20135.\n\n. Mark Ravenhill, _Shopping and Fucking_ (London: Methuen, 1996), 83.\n\n. Dennis Cooper, _Frisk_ (1991) (London: Serpent's Tail, 1992), 44.\n\n. Oscar Wilde, _The Picture of Dorian Gray_ (1891) (Oxford: Oxford University Press, 1981), 18.\n\n. Claudia Card, _Lesbian Choices_ (New York: Columbia University Press, 1995), 231\u201335.\n\n. Dennis Cooper, _Try_ (London: Serpent's Tail, 1994), 185 (Cooper's pause).\n\n. Dennis Cooper, _Guide_ (1997) (London: Serpent's Tail, 1998), 155.\n\n**4. POWER**\n\n. Jean Genet, _The Balcony_ (1957), trans. Bernard Frechtman (London: Faber, 1966), 47 (my elision).\n\n. Paul Monette, _Becoming a Man_ (1992) (London: Abacus, 1994), 66\u201368 (Monette's emphasis).\n\n. Paul Monette, _Borrowed Time_ (New York: Avon, 1988), 13 (Monette's emphasis).\n\n. _Becoming a Man_ , 175.\n\n. Theodore Redpath, ed., _The Songs and Sonets of John Donne_ (London: Methuen, 1967), 2. Donne is usually thought of as a heterosexual poet, but see George Klawitter, \"Verse Letters to T. W. from John Donne: 'By You My Love Is Sent,'\" in Claude J. Summers, ed., _Homosexuality in Renaissance and Enlightenment England_ (New York: Harrington Park, 1992).\n\n. Walt Whitman, _The Complete Poems_ , ed. Francis Murphy (Harmondsworth: Penguin, 1996), 162.\n\n. Michel Foucault, _The History of Sexuality_ , vol. 1: _An Introduction_ , trans. Robert Hurley (New York: Vintage, 1978), 11; and 92\u201398.\n\n. Michel Foucault, _Discipline and Punish_ , trans. Alan Sheridan (Harmondsworth: Penguin, 1979), 27. See Gail Mason, _The Spectacle of Violence_ (New York: Routledge, 2002), ch. 6; Judith Butler, _Gender Trouble_ (New York and London: Routledge, 1990), 91\u2013106.\n\n. Teresa de Lauretis, _The Practice of Love_ (Bloomington: Indiana University Press, 1994), 146 (my elision).\n\n. Geoffrey Gorer, _Sex and Marriage in England Today_ (London: Nelson, 1971), 62, 65.\n\n. See Lillian Faderman, _Odd Girls and Twilight Lovers_ (New York: Penguin, 1992); George E. Haggerty, _Men in Love_ (New York: Columbia University Press, 1999).\n\n. David M. Halperin, _How to Do the History of Homosexuality_ (Chicago: University of Chicago Press, 2002), 133\u201334.\n\n. Jeffrey Weeks, Brian Heaphy, and Catherine Donovan, _Same Sex Intimacies_ (London: Routledge, 2001), 105.\n\n. Lynda Hart and Joshua Dale, \"Sadomasochism,\" in Andy Medhurst and Sally R. Munt, eds., _Lesbian and Gay Studies_ (London: Cassell, 1997), 345\u201346.\n\n. David Leavitt, _The Lost Language of Cranes_ (1986) (Harmondsworth: Penguin, 1987), 169.\n\n. See Larry Gross, _Up from Invisibility_ (New York: Columbia University Press, 2001), 146\u201347.\n\n. A. M. Homes, _Jack_ (1989) (Harmondsworth: Penguin, 1991).\n\n. Anthony McDonald, _Adam_ (London: Gay Men's Press, 2003).\n\n. Michael Cunningham, _Flesh and Blood_ (New York: Farrar, Straus, and Giroux, 1995), 302, 303, 305 (my elision).\n\n. Edmund White, _The Farewell Symphony_ (London: Chatto, 1997), 414\u201315, 298.\n\n. Samuel R. Delany, _The Motion of Light in Water_ (1988) (London: Paladin, 1990), 267 (Delany's emphasis).\n\n. David Wojnarowicz, _Close to the Knives_ (London: Serpent's Tail, 1992), 17 (my elision).\n\n. Ben Gove, _Cruising Culture_ (Edinburgh: Edinburgh University Press, 2000), 156\u201359.\n\n. Dennis Altman, _The Homosexualization of America_ (Boston: Beacon Press, 1982), 79\u201380; Leo Bersani, \"Is the Rectum a Grave?\" in Douglas Crimp, ed., _AIDS: Cultural Analysis, Cultural Activism_ (Cambridge: MIT Press, 1988), 206.\n\n. Andrew Holleran, _Dancer from the Dance_ (1978) (London: Cape, 1979), 40\u201343.\n\n. White, _The Farewell Symphony_ , 298, 416.\n\n. Neil Bartlett, _Who Was That Man?_ (London: Serpent's Tail, 1988), 220.\n\n. Larry Kramer, _Faggots_ (1978) (London: Minerva, 1990), 382.\n\n. Oscar Moore, _A Matter of Life and Sex_ (1991) (Harmondsworth: Penguin, 1992).\n\n. William M. Hoffman, _As Is_ (1985), in Michael Feingold, ed., _The Way We Live Now_ (New York: Theatre Communications Group, 1990), 25. The film was directed by Michael Lindsay-Hogg in 1986.\n\n. Thom Gunn, _The Man with Night Sweats_ (London: Faber, 1992), 80.\n\n. Bersani, \"Is the Rectum a Grave?\" in Crimp, ed., _AIDS: Cultural Analysis_ , 215.\n\n. Leo Bersani, _Homos_ (Cambridge: Harvard University Press, 1995), 162, 170\u201371 (Bersani's emphases). Cf. Alan Sinfield, _Gay and After_ (London: Serpent's Tail, 1998), ch. 7.\n\n. Leo Bersani, \"Genital Chastity,\" in Tim Dean and Christopher Lane, eds., _Homosexuality and Psychoanalysis_ (Chicago: University of Chicago Press, 2001), 365 (Bersani's emphasis).\n\n. Sigmund Freud, \"From the History of an Infantile Neurosis\" (1918 [1914]), in _The Penguin Freud Library_ , vol. 9: _Case Histories II_ (Harmondsworth: Penguin, 1979), 280.\n\n. Freud, _Three Essays on the Theory of Sexuality_ (1905), in _The Penguin Freud Library_ , vol. 7: _On Sexuality_ (Harmondsworth: Penguin, 1977), 61\u201362 (Freud's emphases).\n\n. Freud, \"'A Child Is Being Beaten,'\" (1919) in _The Penguin Freud Library_ , vol. 10: _On Psychopathology_ (Harmondsworth: Penguin, 1984), 180.\n\n. Ibid., 184\u201385 (Freud's emphasis).\n\n. Sigmund Freud, \"The Economic Problem of Masochism\" (1924), in _The Penguin Freud Library_ , vol. 11: _On Metapsychology_ (Harmondsworth: Penguin, 1984), 419; \"'A Child Is Being Beaten,'\" 166.\n\n. Luce Irigaray, _Je, Tu, Nous_ (New York: Routledge, 1993), 12 (Irigaray's emphasis).\n\n. Jean Laplanche, _New Foundations for Psychoanalysis_ , trans. David Macey (Oxford: Blackwell, 1989); for Fletcher, see below. For some of the take-up of Laplanche's concept of implantation, see Elizabeth Cowie, \"The Seductive Theories of Jean Laplanche: A New View of the Drive, Passivity, and Femininity,\" in John Fletcher and Martin Stanton, eds., _Jean Laplanche_ (London: Institute of Contemporary Art, 1992); Catherine Belsey, _Desire: Love Stories in Western Culture_ (Oxford: Blackwell, 1994), 51\u201352; Lynne Segal, _Why Feminism?_ (Cambridge: Polity, 1999), 184\u201385.\n\n. Jean Laplanche, \"Implantation, Intromission,\" in Laplanche, _Essays on Otherness_ , ed. John Fletcher (London: Routledge, 1999), 135 (Laplanche's emphasis).\n\n. John Fletcher, \"Gender, Sexuality, and the Theory of Seduction,\" _Women: A Cultural Review_ 11 (2000): 102, 104 (Fletcher's emphasis).\n\n. \"Interview: Jean Laplanche Talks to Martin Stanton,\" in John Fletcher and Martin Stanton, eds., _Jean Laplanche_ , 10.\n\n. Fletcher, \"Gender, Sexuality,\" 104.\n\n. Ibid., 106 (Fletcher's emphasis).\n\n. John Fletcher, \"Recent Developments in the General Theory of Primal Seduction,\" _New Formations_ 48 (2002\u20132003): 5\u201325 (quote at 9). See Dominique Scarfone, \"'It was _not_ my mother': From Seduction to Negation,\" _New Formations_ 48 (2002\u20132003): 69\u201376.\n\n. Laplanche, \"Implantation, Intromission,\" 136.\n\n. Sheila Jeffreys, \"Butch and Femme: Now and Then,\" in Lesbian History Group, _Not a Passing Phase_ (London: Women's Press, 1993), 178.\n\n. Amber Hollibaugh and Cherr\u00ede Moraga, \"What We're Rollin' Around in Bed With: Sexual Silences in Feminism,\" in Ann Snitow, Christine Stansell, and Sharon Thompson, eds., _Desire: The Politics of Sexuality_ (London: Virago, 1984), 410\u201311.\n\n. Jeffreys, \"Butch and Femme,\" 184 (my elision).\n\n. Hollibaugh and Moraga, \"What We're Rollin' Around in Bed With,\" 406.\n\n. Cherri\u00e9 Moraga, _Loving in the War Years_ (Boston: South End Press, 1983), 125\u201326 (Moraga's emphasis).\n\n. Judith Halberstam, \"Sex Debates,\" in Medhurst and Munt, eds., _Lesbian and Gay Studies_ , 335.\n\n. Esther Newton and Shirley Walton, \"The Misunderstanding: Toward a More Precise Sexual Vocabulary,\" in Carole S. Vance, ed., _Pleasure and Danger_ (London: Routledge, 1984), 247.\n\n. Ursula Zilinsky, _Middle Ground_ (1968) (London: Gay Men's Press, 1987), 146.\n\n. Sir Philip Sidney, _The Old Arcadia_ , ed. Katherine Duncan-Jones (Oxford: Oxford University Press, 1985), 167.\n\n. Christopher Ricks, ed., _The Poems of Tennyson_ (London: Longmans, 1969): _In Memoriam_ , sec. 25 and 42.\n\n. W. H. Auden, _Collected Shorter Poems, 1927\u20131957_ (London: Faber, 1969), 107\u2013108.\n\n. Lynne Segal, _Straight Sex_ (London: Virago, 1994), 248.\n\n. John Rechy, _The Sexual Outlaw_ (1977) (London: W. H. Allen, 1978), 68 (my elision).\n\n. \"Michel Foucault: Sex, Power, and the Politics of Identity,\" interview with Bob Gallagher and Alexander Wilson, _The Advocate_ 400 (August 7, 1984): 30; quoted in Bersani, _Homos_ , 88.\n\n. Bersani, _Homos_ , 88.\n\n. Jeffrey Weeks, _Sexuality and Its Discontents_ (London: Routledge, 1985), 44; Bersani, \"Is the Rectum a Grave?\" in Crimp, ed., _AIDS: Cultural Analysis_ , 220\u201321.\n\n. Claudia Card, _Lesbian Choices_ (New York: Columbia University Press, 1995), 221.\n\n. James Robert Baker, _Tim and Pete_ (1995) (London: Fourth Estate, 1996), 150.\n\n. See Sadie Plant, _The Most Radical Gesture_ (London: Routledge, 1992), 143\u201347.\n\n. See Graham White, \"Direct Action, Dramatic Action: Theatre and Situationist Theory,\" _New Theatre Quarterly_ 9.36 (November 1993): 329\u201340 (see 337).\n\n. See Peter Dickinson, \"'Go-go Dancing on the Brink of the Apocalypse': Representing AIDS,\" in Richard Dellamora, ed., _Postmodern Apocalypse_ (Philadelphia: University of Pennsylvania Press, 1995).\n\n. Wojnarowicz, _Close to the Knives_ , 81.\n\n. Sinfield, _Gay and After_ , ch. 2 and 6. See Dennis Altman, _AIDS and the New Puritanism_ (London: Pluto, 1986), ch. 8: \"A Very American Epidemic?\"\n\n. Andrew Sullivan, _Love Undetectable_ (London: Chatto and Windus, 1998), 18.\n\n. Bersani, \"Is the Rectum a Grave?\" in Crimp, ed., _AIDS: Cultural Analysis_ , 212.\n\n. Halberstam, \"Sex Debates,\" in Medhurst and Munt, eds., _Lesbian and Gay Studies_ , 333.\n\n. Bersani, _Homos_ , 64.\n\n**5. GENDER**\n\n. From Ned Ward, _The History of the London Clubs_ (1709), printed in Ian McCormick, ed., _Secret Sexualities_ (London: Routledge, 1977), 131.\n\n. Alan Bray, _Homosexuality in Renaissance England_ , 2d ed. (London: Gay Men's Press, 1988), 86.\n\n. Ibid., 92.\n\n. Rictor Norton, _Mother Clap's Molly House_ (London: Gay Men's Press, 1992).\n\n. Terry Castle, _The Apparitional Lesbian_ (New York: Columbia University Press, 1993), ch. 5; Rictor Norton, _The Myth of the Modern Homosexual_ (London: Cassell, 1997), 196\u2013202.\n\n. Judith Halberstam, _Female Masculinity_ (Durham: Duke University Press, 1998), 65\u201373.\n\n. Emma Donoghue, _Passions Between Women_ (London: Scarlet Press, 1993), 61.\n\n. Terry Castle, ed., _The Literature of Lesbianism_ (New York: Columbia University Press, 2003), 20.\n\n. George Chauncey Jr., \"From Sexual Inversion to Homosexuality: Medicine and the Changing Conceptualization of Female Deviance,\" _Salmagundi_ 58\u201359 (1982\u201383): 114\u201346 (quote at 123).\n\n. Gayle S. Rubin, \"Thinking Sex: Notes for a Radical Theory of the Politics of Sexuality,\" in Henry Abelove, Mich\u00e8le Aina Barale, and David M. Halperin, eds., _The Lesbian and Gay Studies Reader_ (New York: Routledge, 1993), 33; Eve Kosofsky Sedgwick, _Epistemology of the Closet_ (Hemel Hempstead: Harvester Wheatsheaf, 1991), 27\u201335.\n\n. John Fletcher, \"Gender, Sexuality, and the Theory of Seduction,\" _Women: A Cultural Review_ 11 (2000): 95\u2013108, 95\u201396.\n\n. Tamsin Wilton, \"Which One's the Man?\" in Diane Richardson, ed., _Theorising Heterosexuality_ (Buckingham: Open University Press, 1996), 137; William J. Spurlin, \"Sissies and Sisters: Gender, Sexuality, and the Possibilities of Coalition,\" in Mandy Merck, Naomi Segal, and Elizabeth Wright, eds., _Coming Out of Feminism?_ (Oxford: Blackwell, 1998).\n\n. David M. Halperin, _How to Do the History of Homosexuality_ (Chicago: University of Chicago Press, 2002), ch. 4.\n\n. See Jennifer Terry, _An American Obsession_ (Chicago: University of Chicago Press, 1999); Alan Sinfield, _The Wilde Century_ (New York: Columbia University Press, 1994), 93\u201397.\n\n. Havelock Ellis and John Addington Symonds, _Sexual Inversion_ (London: Wilson and Macmillan, 1897; New York: Ayer, 1994), 136\u201337; Sigmund Freud, _Three Essays on the Theory of Sexuality_ (1905), in _The Penguin Freud Library_ , vol. 7: _On Sexuality_ (Harmondsworth: Penguin, 1977), 48\u201349. Only the suppressed first edition of _Sexual Inversion_ bore Symonds' name, so in my text I follow the convention of referring to Ellis as the author.\n\n. Ellis and Symonds, _Sexual Inversion_ (1897), 133.\n\n. Michel Foucault, _The History of Sexuality_ , vol. 1: _An Introduction_ , trans. Robert Hurley (New York: Vintage, 1978), 119.\n\n. Martin Scherzinger and Neville Hoad, \"A\/Symmetrical Reading of _Inversion_ in Fin-de-Si\u00e8cle Music, Musicology, and Sexology,\" in C. Lorey and J. Plews, eds., _Queering the Canon_ (New York: Camden House, 1998).\n\n. Chauncey, \"From Sexual Inversion to Homosexuality,\" 124. See Halberstam, _Female Masculinity_ , 75\u201383.\n\n. David M. Halperin, _One Hundred Years of Homosexuality_ (New York and London: Routledge, 1990), 16. A similar claim is made by Arnold Davidson, \"How to Do the History of Psychoanalysis: A Reading of Freud's _Three Essays on the Theory of Sexuality_ ,\" in Fran\u00e7ois Meltzer, ed., _The Trial(s) of Psychoanalysis_ (Chicago: University of Chicago Press, 1988), and by Jeffrey Weeks, _Sexuality and Its Discontents_ (London: Routledge, 1985), 153\u201354.\n\n. Ellis and Symonds, _Sexual Inversion_ (1897), 32.\n\n. Havelock Ellis, _Studies in the Psychology of Sex_ , vol. 2: _Sexual Inversion_ (Philadelphia: F. A. Davis, 1901); quoted from Aron Krich, ed., _The Sexual Revolution: Pioneer Writings on Sex: Krafft-Ebing, Ellis, Freud_ (New York: Delta, 1964), 152, 156. These passages are revised and elaborated in Ellis's third edition of 1915: see Ellis, _Studies in the Psychology of Sex_ , vol. 2, part 2 (New York: Random House, 1936), 2\u20134.\n\n. Ellis and Symonds, _Sexual Inversion_ (1897), 119 (my elision; Ellis and Symonds' emphasis).\n\n. Ellis and Symonds, _Sexual Inversion_ (1897), 120. Krafft-Ebing actually lists four stages or degrees: attraction to the same sex without effect on the manliness of a man; change of character, the man feeling himself to be a woman; sensation of physical transformation; delusion of sexual change. Richard von Krafft-Ebing, _Psychopathia Sexualis_ (1886; 12th ed., 1903), trans. Franklin S. Klaf (New York: Scarborough, 1978), 190, 195, 200, 216.\n\n. Freud, _Three Essays_ , 46. Freud's persistence in a gendered model of homosexuality is demonstrated by Christopher Craft, _Another Kind of Love_ (Berkeley: University of California Press, 1994), 36\u201343. See also Sinfield, _The Wilde Century_ , ch. 7.\n\n. C. A. Tripp, _The Homosexual Matrix_ (1975), 2d ed. (New York: Meridian, 1987), 20, 71\u201374.\n\n. Sedgwick, _Epistemology of the Closet_ , 45\u201347, 157\u201359.\n\n. Foucault, _The History of Sexuality_ 1:43 (my emphases).\n\n. Gert Hekma, ' \"A Female Soul in a Male Body': Sexual Inversion as Gender Inversion in Nineteenth-Century Sexology,\" in Gilbert Herdt, ed., _Third Sex, Third Gender_ (New York: Zone, 1994), 236, 238.\n\n. Kaja Silverman, _Male Subjectivity at the Margins_ (New York: Routledge, 1992), 342.\n\n. Neil Bartlett, _Who Was That Man? A Present for Mr. Oscar Wilde_ (London: Serpent's Tail, 1988), xx (Bartlett's emphasis).\n\n. Gregg Blachford, \"Male Dominance and the Gay World,\" in Kenneth Plummer, ed., _The Making of the Modern Homosexual_ (London: Hutchinson, 1981); Jamie Gough, \"Theories of Sexual Identity and the Masculinization of the Gay Man,\" in Simon Shepherd and Mick Wallis, eds., _Coming on Strong_ (London: Unwin Hyman, 1989).\n\n. Dennis Altman, _The Homosexualization of America_ (Boston: Beacon Press, 1982), 1.\n\n. Richard Dyer, _Only Entertainment_ (London: Routledge, 1992), 165\u201366.\n\n. Simon Fraser, \"Visions of Love,\" interview with Neil Bartlett, _Rouge_ 8 (October-December 1991): 20\u201322 (quote at 21). See Alan Sinfield, \"'The Moment of Submission': Neil Bartlett in Conversation,\" _Modern Drama_ 39 (1996): Special Issue on Lesbian\/Gay\/Queer Drama, ed. Hersh Zeifman, 211\u201321 (see 215).\n\n. Neil Bartlett, _Ready to Catch Him Should He Fall_ (London: Serpent's Tail, 1990), 162.\n\n. See Harry Oosterhuis and Hubert Kennedy, eds., _Homosexuality and Male Bonding in Pre-Nazi Germany_ (New York: Harrington Park Press, 1991).\n\n. George Chauncey, _Gay New York_ (New York: Basic Books, 1994), 13.\n\n. Quoted in Donald Webster Cory, _The Homosexual in America_ (1951), with a retrospective foreword (New York: Arno Press, 1975), 188.\n\n. David K. Johnson, \"The Kids of Fairytown: Gay Male Culture on Chicago's Near North Side in the 1930s,\" and Allen Drexel, \"Before Paris Burned: Race, Class, and Male Homosexuality on the Chicago South Side, 1935\u20131960,\" both in Brett Beemyn, ed., _Creating a Place for Ourselves_ (New York: Routledge, 1997).\n\n. Edmund White, _The Beautiful Room is Empty_ (London: Picador, 1988), 102-3, 33. The latter thought recurs: see 36 and 71.\n\n. John Marshall, \"Pansies, Perverts, and Macho Men: Changing Conceptions of Homosexuality,\" in Plummer, ed., _The Making of the Modern Homosexual_ , 135. See also Sinfield, _The Wilde Century_ , ch. 6.\n\n. Quentin Crisp, _The Naked Civil Servant_ (1968) (New York: Plume, 1977), 21.\n\n. Peter Wildeblood, _Against the Law_ (London: Weidenfeld, 1955), 7.\n\n. Michael Bronski, _Culture Clash_ (Boston: South End Press, 1984), 79\u201380.\n\n. Kenneth Marlowe, _The Male Homosexual_ (Los Angeles: Medco, 1968), 12\u201313, 18.\n\n. Hall Carpenter Archives and Gay Men's Oral History Group, _Walking After Midnight_ (London: Routledge, 1989), 87.\n\n. Mart Crowley, _The Boys in the Band_ (New York: French, 1968), 45, 87 (Crowley's emphasis; my elision).\n\n. Leslie Feinberg, _Stone Butch Blues_ (Ithaca: Firebrand, 1993), 11, 135\u201336.\n\n. Andrew Sullivan, _Virtually Normal_ , 2d ed. (London: Picador, 1996), 4.\n\n. Andrew Sullivan, _Love Undetectable_ (London: Chatto and Windus, 1998), 12\u201313.\n\n. Andrew Sullivan, \"Mainlining Manhood,\" _Guardian Saturday Review_ , April 8, 2000, 1\u20133.\n\n. Sullivan, _Love Undetectable_ , 153.\n\n. Paul Monette, _Halfway Home_ (New York: Crown, 1991), 246.\n\n. Joseph Hansen, _Steps Going Down_ (1985) (London: Arlington, 1986), 24 (Hansen's emphasis).\n\n. Judith Butler, _Gender Trouble_ (New York and London: Routledge, 1990), 25; Kate Bornstein, _Gender Outlaw_ (New York: Routledge, 1994), 125, 138.\n\n. Sinfield, _The Wilde Century_ , 75\u201378.\n\n. Tennessee Williams, _A Streetcar Named Desire_ (1947), in Williams, _\"Sweet Bird of Youth,\" \"A Streetcar Named Desire,\" \"The Glass Menagerie,\"_ ed. E. Martin Browne (Harmondsworth: Penguin, 1962), 182\u201383.\n\n. John Rechy, _City of Night_ (1963) (London: MacGibbon and Kee, 1964), 34.\n\n. John Rechy, _Bodies and Souls_ (1983) (London: Star Books, 1985), 302\u2013305 (Rechy's emphasis).\n\n. John Rechy, _The Sexual Outlaw_ (1977) (London: W. H. Allen, 1978), 243 (Rechy's emphasis).\n\n. Ben Gove, _Cruising Culture_ (Edinburgh: Edinburgh University Press, 2000), 43\u201346.\n\n. Edmund White, _The Farewell Symphony_ (London: Chatto, 1997), 301.\n\n. Joseph Mills, \"Dreaming, Drag,\" in Mills, _Obsessions_ (Brighton: Millivres, 1998), 74.\n\n. Leo Bersani, \"Is the Rectum a Grave?\" in Douglas Crimp, ed., _AIDS: Cultural Analysis, Cultural Activism_ (Cambridge: MIT Press, 1988), 208\u2013209.\n\n. Leo Bersani, _Homos_ (Cambridge.: Harvard University Press, (1995), 60\u201361.\n\n. See Patrick Paul Garlinger, \"'Homo-ness' and the Fear of Femininity,\" _Diacritics_ 29 (1999): 57\u201371.\n\n. Tim Bergling, _Sissyphobia: Gay Men and Effeminate Behavior_ (New York: Harrington Park, 2001), 9; Richard Green, _The \"Sissy Boy Syndrome\" and the Development of Homosexuality_ (New Haven: Yale University Press, 1987), 141\u201343, 159\u201360, 169, 191.\n\n. Gove, _Cruising Culture_ , 64\u201372 (at 50).\n\n. Jean Genet, _Our Lady of the Flowers_ (1943), trans. Bernard Frechtman (London: Panther, 1966).\n\n. Ben Gove, \"Framing Gay Youth,\" _Screen_ 37 (1996): 174\u201392 (at 186\u201387).\n\n. John Hopkins, _Find Your Way Home_ (1970) (Harmondsworth: Penguin, 1971).\n\n. See Stephen Maddison, _Fags, Hags and Queer Sisters_ (London: Macmillan, 2000), 1\u20136.\n\n. Harvey Fierstein, _Torch Song Trilogy_ (1979) (London: Methuen, 1984).\n\n. Richard Goldstein, _The Attack Queers_ (London: Verso, 2002), 78.\n\n. Edmund White, _A Boy's Own Story_ (1982) (London: Picador, 1983), 169.\n\n. White, _The Farewell Symphony_ , 34.\n\n. James Kenneth Melson, _The Golden Boy_ (New York: Harrington Park, 1992), 191.\n\n. Richard Green, _The \"Sissy Boy Syndrome\" and the Development of Homosexuality_ (New Haven: Yale University Press, 1987), 124; Todd's emphasis.\n\n. Bornstein, _Gender Outlaw_ , 191.\n\n. See _Paradise Bent_ , a documentary film produced and directed by Heather Croall (ReAngle Pictures, 1999).\n\n. Hugh McLean and Linda Ngcobo, \"Abangibhamayo bathi ngimnandi (Those who fuck me say I'm tasty): Gay Sexuality in Reef Townships,\" in Mark Gevisser and Edwin Cameron, eds., _Defiant Desire_ (London: Routledge, 1995), 164\u201365.\n\n. See Del LaGrace Volcano and Judith \"Jack\" Halberstam, _The Drag King Book_ (London: Serpent's Tail, 1999).\n\n. Judith Halberstam, \"What's That Smell? Queer Temporalities and Subcultural Lives,\" _International Journal of Cultural Studies_ 6 (2003): 313\u201333.\n\n. Radclyffe Hall, _The Well of Loneliness_ (London: Falcon Press, 1949), 29 (Hall's emphasis).\n\n. Jay Prosser, _Second Skins_ (New York: Columbia University Press, 1998), 155\u201356.\n\n. For substantial evidence of the confused and formative reception of _The Well_ , see Laura Doan, _Fashioning Sapphism_ (New York: Columbia University Press, 2001), and Laura Doan and Jay Prosser, eds., _Palatable Poison_ (New York: Columbia University Press, 2001).\n\n. Halberstam, _Female Masculinity_ , 98.\n\n. Esther Newton, \"The Mythic Mannish Lesbian: Radclyffe Hall and the New Woman,\" in Martin Duberman, Martha Vicinus, and George Chauncey Jr., eds., _Hidden from History_ (New York: Meridian, 1990).\n\n. Prosser, _Second Skins_ , ch. 4; Halberstam, _Female Masculinity_ , 110.\n\n. Jean E. Mills, \"Gertrude Stein Took the War Like a Man,\" _The Gay and Lesbian Review_ 10.2 (March-April 2003): 16\u201317.\n\n. Feinberg, _Stone Butch Blues_ , 13.\n\n. Sandy Stone, \"The 'Empire' Strikes Back: A Posttranssexual Manifesto,\" in Julia Epstein and Kristina Straub, eds., _Bodyguards_ (New York: Routledge, 1991), 298.\n\n. Butler, _Gender Trouble_ , 137. Butler revises this argument in her book _Bodies That Matter_ (New York: Routledge, 1993), 125. See further Moe Meyer, _The Politics and Poetics of Camp_ (London: Routledge, 1994), and Fabio Cleto, _Camp: Queer Aesthetics and the Performing Subject_ (Edinburgh: Edinburgh University Press, 1999).\n\n. Prosser, _Second Skins_ , 204\u2013205 (Prosser's emphasis).\n\n. Eve Kosofsky Sedgwick, _Tendencies_ (London: Routledge, 1994), 221, 157\u201358.\n\n. Neil Duncan, _Sexual Bullying_ (London: Routledge, 1999), 107\u2013108. Girls in this school were more tolerant of male and female homosexuality, and tended to support each other in the face of accusations of lesbianism from boys (121\u201324).\n\n. Richard Smith, \"Pretty Hate Machine,\" _Gay Times_ 263 (August 2000): 19\u201320.\n\n. Wendy Wallace, \"Is This Table Gay?\" _Times Educational Supplement_ , January 19, 2001, 9\u201310.\n\n. Mois\u00e9s Kaufman, _The Laramie Project_ (New York: Vintage, 2001), 90 (my elision).\n\n**6. AGE**\n\n. Armistead Maupin, _The Night Listener_ (London: Bantam, 2000), 42\u201343 (my elision).\n\n. _The Long Goodbye_ , featuring Maupin and his partner Terry Anderson, BBC television, June 1, 1995. The series is about bereavement, in anticipation of Anderson's death.\n\n. Armistead Maupin, \"Coming Home,\" in Edmund White, ed., _The Faber Book of Gay Short Fiction_ (London: Faber, 1991), 355\u201356.\n\n. Eve Kosofsky Sedgwick, _Tendencies_ (London: Routledge, 1994), 57\u201358 (my elision).\n\n. David M. Halperin, _How to Do the History of Homosexuality_ (Chicago: University of Chicago Press, 2002), 115\u201316 (Halperin's emphases; my elision). Halperin, commenting on a draft from the present work, says that he means in this passage only \"a largely lop-sided or non-reciprocal pattern of desire and pleasure\" (ibid., 190). However, this strikes me as tautologous: when he writes of anomic relations, he only means anomic relations. This is the only place in his essay where Halperin addresses age difference.\n\n. Raymond Williams, _Problems in Materialism and Culture_ (London: New Left Books, 1980), 38-42.\n\n. Simon LeVay and Elisabeth Nonas, _City of Friends_ (Cambridge: MIT Press, 1995), 30\u201331. See Barry D. Adam, \"Age Preferences Among Gay and Bisexual Men,\" _GLQ_ ( _Gay and Lesbian Quarterly_ ) 6 (2000): 423\u201333.\n\n. Edmund White, _The Married Man_ (London: Chatto, 2000), 11.\n\n. Gregory Woods, _We Have the Melon_ (Manchester: Carcanet, 1992), 60. See also Woods's Carcanet volumes, _May I Say Nothing_ (1998) and _The District Commissioner's Dreams_ (2002).\n\n. Paul Monette, _Halfway Home_ (New York: Crown, 1991), 108 (Monette's emphasis).\n\n. Jack Dickson, _Oddfellows_ (Brighton: Millivres, 1997), 50.\n\n. Simon Lovat, _Disorder and Chaos_ (Brighton: Millivres, 1996), 159.\n\n. Paul Rabinow, ed., _The Foucault Reader_ (Harmondsworth: Penguin, 1984), 344\u201345.\n\n. David Leeming, _James Baldwin_ (Harmondsworth: Penguin, 1995), 286\u201387.\n\n. Alan Hollinghurst, _The Swimming-Pool Library_ (New York: Random House, 1988), 284.\n\n. Lovat, _Disorder and Chaos_ , 132.\n\n. Larry Kramer, _Faggots_ (1978) (London: Minerva, 1990), 238\u201343.\n\n. Philip Osment, _This Island's Mine_ , in Osment, ed., _Gay Sweatshop: Four Plays and a Company_ (London: Methuen, 1989).\n\n. Paul Robinson, _Gay Lives_ (Chicago: University of Chicago Press, 1999), 100.\n\n. Christopher Isherwood, _A Single Man_ (1964) (London: Minerva, 1991) 130\u201331 (Isherwood's emphasis).\n\n. H. Montgomery Hyde, ed., _The Trials of Oscar Wilde_ (London: Hodge, 1948), 236 (my elision).\n\n. See Jonathan Dollimore, _Death, Desire and Loss in Western Culture_ (London: Routledge, 1998), ch. 19.\n\n. Mary Renault, _The Charioteer_ (1953) (London: New English Library, 1990), 114. See Sinfield, _The Wilde Century_ (New York: Columbia University Press, 1994), 143\u201345.\n\n. Allan Bloom, _The Closing of the American Mind_ (New York: Simon and Schuster, 1987), 132\u201333.\n\n. Saul Bellow, _Ravelstein_ (Harmondsworth: Penguin, 2000), 160.\n\n. Eve Kosofsky Sedgwick, _Epistemology of the Closet_ (Hemel Hempstead: Harvester Wheatsheaf, 1991), 56\u201357 (my elision).\n\n. Patricia Duncker, _Hallucinating Foucault_ (London: Serpent's Tail, 1996).\n\n. James Baldwin, _Another Country_ (London: Michael Joseph, 1963), 181\u201382 (my elision, Baldwin's emphasis).\n\n. Kenneth Martin, _Aubade_ (1957) (London: Gay Men's Press, 1989), 104.\n\n. Timothy Ireland, _Who Lies Inside_ (London: Gay Men's Press, 1984), 17.\n\n. Guy Willard, _Mirrors of Narcissus_ (London: Millivres, 2000), 121.\n\n. Jill Posener, _Any Woman Can_ (1975), in Jill Davis, ed., _Lesbian Plays_ (London: Methuen, 1987).\n\n. Jeanette Winterson, _Oranges Are Not the Only Fruit_ (1985) (New York: Vintage, 1996), 153.\n\n. P-P Hartnett, _I Want to Fuck You_ (London: Pulp Faction, 1998), 25.\n\n. Anonymous, _The Scarlet Pansy_ (New York: Badboy, 1992), 42).\n\n. Edmund White, _A Boy's Own Story_ (1982) (London: Picador, 1983), preface. See Ben Gove, \"Framing Gay Youth,\" _Screen_ 37 (1996): 174\u201392.\n\n. Iris Murdoch, _A Fairly Honourable Defeat_ (1970) (Harmondsworth: Penguin, 1972).\n\n. Neil Bartlett, _Ready to Catch Him Should He Fall_ (London: Serpent's Tail, 1990), 14.\n\n. Adam Mars-Jones, \"Camp for Internal Exiles,\" _The Independent on Sunday_ , October 14, 1990, Sunday Review 32. For this reference, and many rewarding exchanges about Bartlett, I am indebted to Linda Logie.\n\n. Alan Sinfield, \"'The Moment of Submission': Neil Bartlett in Conversation,\" _Modern Drama_ 39 (1996): Special Issue on Lesbian\/Gay\/Queer Drama, ed. Hersh Zeifman, 211\u201321 (quote at 212\u201313).\n\n**7. CLASS**\n\n. Eve Kosofsky Sedgwick, _Epistemology of the Closet_ (Hemel Hempstead: Harvester Wheatsheaf, 1991), 31 (Sedgwick's emphases).\n\n. On Trotskyism, see Simon Edge, _With Friends Like These_ (London: Cassell, 1995).\n\n. See Jeffrey Weeks, _Coming Out_ (London: Quartet, 1977), 21; Ed Cohen, _Talk on the Wilde Side_ (New York: Routledge, 1993), 145\u201348; Alan Sinfield, _The Wilde Century_ (New York: Columbia University Press, 1994).\n\n. See Alan Sinfield, _Literature, Politics, and Culture in Postwar Britain_ , 2d ed. (London: Athlone, 1997), ch. 5; Sinfield, _Gay and After_ (London: Serpent's Tail, 1998), 95\u201399.\n\n. Michel Foucault, _The History of Sexuality_ , vol. 1: _An Introduction_ , trans. Robert Hurley (New York: Vintage, 1978), 120\u201321.\n\n. George Chauncey, _Gay New York_ (New York: Basic Books, 1994), 118\u201321. See chapter 5, this volume.\n\n. Murray Healy, _Gay Skins_ (London: Cassell, 1996), 16\u201336.\n\n. Sigmund Freud, \"On the Universal Tendency to Debasement in the Sphere of Love\" (1912), in _The Penguin Freud Library_ , vol. 7: _On Sexuality_ (Harmondsworth: Penguin, 1977), 248 (Freud's emphases).\n\n. H. Montgomery Hyde, ed., _The Trials of Oscar Wilde_ (London: Hodge, 1948), 138.\n\n. C. H. Rolph, ed., _The Trial of Lady Chatterley_ (Harmondsworth: Penguin, 1961), 17.\n\n. Peter Stallybrass and Allon White, _The Politics and Poetics of Transgression_ (London: Methuen, 1986), 156, 153. See Leonore Davidoff, \"Class and Gender in Victorian England: The Diaries of Arthur J. Munby and Hannah Cullwick,\" _Feminist Studies_ 5 (1979): 87\u2013141.\n\n. Liz Stanley, ed., _The Diaries of Hannah Cullwick, Victorian Maidservant_ (London: Virago, 1984), 193.\n\n. Rodney Garland, _The Heart in Exile_ (1953) (Brighton: Millivres, 1995), 179.\n\n. Jeffrey Weeks, \"Inverts, Perverts, and Mary-Annes: Male Prostitution and the Regulation of Homosexuality in England in the Nineteenth and Early Twentieth Centuries,\" in Salvatore J. Licata and Robert P. Petersen, eds., _Historical Perspectives on Homosexuality_ (New York: Haworth Press, 1985), 121.\n\n. Stephen Spender, _World within World_ (London: Readers Union, 1953), 151.\n\n. Jonathan Harvey, _Beautiful Thing_ (1993), in Michael Wilcox, ed., _Gay Plays 5_ (London: Methuen, 1994); filmed by Hettie Macdonald (1995).\n\n. Jay Quinn, \"The Kitchen Table,\" in Quinn, ed., _Rebel Yell 2_ (New York: Harrington Park, 2002), 189.\n\n. Tom Wakefield, _Mates_ (London: Gay Men's Press, 1983), 18.\n\n. Edmund White, _The Married Man_ (London: Chatto, 2000), 88.\n\n. Sally R. Munt, \"Introduction,\" in Munt, ed., _Cultural Studies and the Working Class_ (London: Cassell, 2000), 9. See Mary McIntosh, \"Class,\" in Andy Medhurst and Sally R. Munt, eds., _Lesbian and Gay Studies_ (London: Cassell, 1997).\n\n. Jean E. Howard and Scott Cutler Shershow, \"Introduction: Marxism Now, Shakespeare Now,\" in Howard and Shershow, eds., _Marxist Shakespeares_ (London: Routledge, 2001), 7\u20138. See also, in the same collection, Richard Halpern, \"An Impure History of Ghosts: Derrida, Marx, Shakespeare,\" 43.\n\n. Leslie J. Moran, \"Homophobic Violence: The Hidden Injuries of Class,\" in Munt, ed., _Cultural Studies and the Working Class_ , 211\u201312, with reference to Ken Plummer, ed., _Modern Homosexualities_ (London: Routledge, 1992), 22.\n\n. Carol M. Ward, _Rita Mae Brown_ (New York: Twayne, 1993), 6. Ward's account is based on published interviews.\n\n. Dorothy Allison, _Bastard Out of Carolina_ (London: Flamingo, 1993).\n\n. Dorothy Allison, _Skin_ (London: Pandora, 1995), 23\u201324. Compare the arguments of Cherr\u00ede Moraga, discussed in chapter 4 above.\n\n. Andrew Holleran, _Dancer from the Dance_ (1978) (London: Cape, 1979), 83.\n\n. David Bergman, _Gaiety Transfigured_ (Madison: University of Wisconsin Press, 1991), 96\u2013102. Other gay critics who have contributed to Whitman's importance, include Thomas Yingling, \"Homosexuality and Utopian Discourse in American Poetry,\" in Betsy Erkkila and Jay Grossman, eds., _Breaking Bounds_ (New York: Oxford University Press, 1996); Gregory Woods, _A History of Gay Literature_ (New Haven: Yale University Press, 1998), 154\u201359, 176\u201380.\n\n. Jon Barrett, _Mark Bingham_ (Los Angeles: Advocate, 2002), 71.\n\n. See Alan Sinfield, _Cultural Politics\u2014Queer Reading_ (1994), 2d ed. (London: Routledge, 2005), 35\u201336.\n\n. Ethan Mordden, _I've a Feeling We're Not in Kansas Anymore_ (1983) (New York: St. Martin's, 1985), 105.\n\n. James Kenneth Melson, _The Golden Boy_ (New York: Harrington Park, 1992), 58.\n\n. Steven Epstein, \"Gay Politics, Ethnic Identity,\" in Edward Stein, ed., _Forms of Desire_ (New York: Routledge, 1992), 282. See Sinfield, _Gay and After_ , ch. 2.\n\n. James Robert Baker, _Tim and Pete_ (1995) (London: Fourth Estate, 1996), 142.\n\n. Paul Monette, _Becoming a Man_ (1992) (London: Abacus, 1994), 19.\n\n. Neal Drinnan, _Glove Puppet_ (Camberwell, Victoria, Australia: Penguin, 1998), 158.\n\n. Paul Russell, _Boys of Life_ (1991) (New York: Plume, 1992), 52.\n\n. Richard von Krafft-Ebing, _Psychopathia Sexualis_ (1886; 12th ed., 1903), trans. Franklin S. Klaf (New York: Scarborough, 1978), 250 (my elision).\n\n. Richard House, _Bruiser_ (London: Serpent's Tail, 1997), 77.\n\n. Joseph Hansen, _Steps Going Down_ (1985) (London: Arlington, 1986), 74.\n\n. Phil Andros, _Below the Belt and Other Stories_ (1982) (Boston, Mass.: Perinium Press, 1992). These stories were written by Samuel M. Steward.\n\n. Scott O'Hara, _Autopornography_ (New York: Harrington Park, 1997), 151.\n\n. John Rechy, _City of Night_ (1963) (London: MacGibbon and Kee, 1964), 32 (Rechy's elision).\n\n. Oscar Moore, _A Matter of Life and Sex_ (1991) (Harmondsworth: Penguin, 1992), 162.\n\n. Donald J. West in association with Buz de Villiers, _Male Prostitution_ (New York: Harrington Park, 1993), 162.\n\n. Michael Arditti, _Easter_ (London: Arcadia, 2000), 335.\n\n. See Alan Sinfield, _Out on Stage_ (New Haven and London: Yale University Press, 1999), 180\u201385.\n\n. Terrence McNally, _Love! Valour! Compassion!_ (1994) (New York: Plume, 1995), 35.\n\n. Peter Cameron, _The Weekend_ (1994) (London: Fourth Estate, 1996), 112.\n\n. Alan Hollinghurst, _The Spell_ (London: Chatto, 1998), 47 (Hollinghurst's pause).\n\n**8. RACE**\n\n. Ward Houser, \"Black Gay Americans,\" in Wayne R. Dynes, ed., _The Encyclopaedia of Homosexuality_ (Chicago and London: St. James Press, 1990), 149\u201350.\n\n. Edmund White, _The Farewell Symphony_ (London: Chatto, 1997), 25, 90.\n\n. Andrew Holleran, _Dancer from the Dance_ (1978) (London: Cape, 1979), 54.\n\n. David B. Feinberg, _Eighty-Sixed_ (1989) (London: Gay Men's Press, 1991), 4.\n\n. Lyle Glazier, \"Chester,\" in Michael J. Smith, ed., _Black Men\u2014White Men_ (San Francisco: Gay Sunshine Press, 1983), 101.\n\n. Rhonda Cobham, \"Jekyll and Claude: The Erotics of Patronage in Claude McKay's _Banana Bottom_ ,\" in Cindy Patton and Benigno S\u00e1nchez-Eppler, eds., _Queer Diasporas_ (Durham: Duke University Press, 2000).\n\n. Paul Thomas Cahill, \"The Reunion,\" in Smith, ed., _Black Men\u2014White Men_ , 183 (Cahill's pause).\n\n. James Baldwin, _Another Country_ (London: Michael Joseph, 1963), 337.\n\n. Eldridge Cleaver, _Soul on Ice_ (London: Panther, 1970), 97, 100. See David Bergman, _Gaiety Transfigured_ (Madison: University of Wisconsin Press, 1991), ch. 9; Lee Edelman, _Homographesis_ (New York: Routledge, 1994), ch. 3.\n\n. Georges-Michel Sarotte, _Like a Brother, Like a Lover_ , trans. Richard Miller (New York: Anchor\/Doubleday, 1978), 97.\n\n. Marshall Moore, \"Everybody Loves the Mus\u00e9e d'Orsay,\" in Jay Quinn, ed., _Rebel Yell 2_ (New York: Harrington Park, 2002).\n\n. See bell hooks, _Outlaw Culture_ (New York: Routledge, 1994); Lola Young, \"'Nothing Is As It Seems': Re-viewing _The Crying Game_ ,\" in Pat Kirkham and Janet Thumin, eds., _Me Jane: Masculinity, Movies, and Women_ (London: Lawrence and Wishart, 1995).\n\n. Abdul JanMohamed, \"The Economy of Manichean Allegory: The Function of Racial Difference in Colonialist Literature,\" _Critical Inquiry_ 12 (1985): 59\u201387 (quote at 62).\n\n. John Sandys, _Against the Tide_ (Penzance, Cornwall: United Writers Publications, 1984), 7.\n\n. Stephen Gray, _Time of Our Darkness_ (London: Frederick Muller, 1988), 76.\n\n. Quoted in Shaun de Waal, \"A Thousand Forms of Love: Representations of Homosexuality in South African Literature,\" in Mark Gevisser and Edwin Cameron, eds., _Defiant Desire_ (London: Routledge, 1995), 240.\n\n. Tony Peake, \"A Son's Story,\" in Peter Burton, ed., _The Mammoth Book of Gay Short Stories_ (London: Robinson, 1997).\n\n. Patricia Duncker, _Writing on the Wall_ (London: Pandora, 2002), 167\u201371. See J. M. Coetzee, _Disgrace_ (London: Secker and Warburg, 1999).\n\n. V. S. Naipaul, _In a Free State_ (1971) (Harmondsworth: Penguin, 1973), 106.\n\n. Angus Wilson, _As If by Magic_ (London: Secker and Warburg, 1973), 283 (Wilson's emphasis).\n\n. Alan Hollinghurst, _The Swimming-Pool Library_ (New York: Random House, 1988), 1. For a discussion of other aspects, see Alan Sinfield, \"Culture, Consensus, and Difference: Angus Wilson to Alan Hollinghurst,\" in Alistair Davies and Alan Sinfield, eds., _British Culture of the Postwar_ (London: Routledge, 2000).\n\n. David Alderson, \"Desire as Nostalgia: The Novels of Alan Hollinghurst,\" in Alderson and Linda Anderson, eds., _Territories of Desire in Queer Culture_ (Manchester: Manchester University Press, 2000), 33.\n\n. Valerie Mason-John and Ann Khambatta, eds., _Lesbians Talk: Making Black Waves_ (London: Scarlet Press, 1993), 30. See further B. Ruby Rich, \"When Difference Is (More Than) Skin Deep,\" in Martha Gever, John Greyson, and Pratibha Parmar, eds., _Queer Looks_ (New York: Routledge, 1993), 319\u201320; Biddy Martin, \"Sexualities Without Gender and Other Queer Utopias,\" _Diacritics_ 24.2\u20133 (1994): 104\u201321; see 114\u201315.\n\n. Steven Corbin, _Fragments That Remain_ (1985) (London: Gay Men's Press, 1993), 69.\n\n. Chris Straayer regrets a similar lack of explicitness in Marlon Riggs's _Tongues Untied_ (on which see below): Straayer, _Deviant Eyes, Deviant Bodies_ (New York: Columbia University Press, 1996), 171.\n\n. Larry Duplechan, _Eight Days a Week_ (Boston: Alyson, 1985), 17.\n\n. See further Wei-cheng Raymond Chu, \"Some Ethnic Gays Are Coming Home; Or, the Trouble with Interraciality,\" _Textual Practice_ 11 (1997): 219\u201336; Darieck Scott, \"Jungle Fever? Black Gay Identity Politics, White Dick, and the Utopian Bedroom,\" _GLQ_ ( _Gay and Lesbian Quarterly_ ) 1 (1994): 299\u2013321.\n\n. Audre Lorde, _Zami: A New Spelling of My Name_ (1982) (London: Pandora, 1996), 100.\n\n. See Audre Lorde, _Sister Outsider_ (Freedom, Calif.: Crossing Press, 1984), 74\u201377.\n\n. Anna Wilson, \"Audre Lorde and the African-American Tradition: When the Family Is Not Enough,\" in Sally Munt, ed., _New Lesbian Criticism_ (Hemel Hempstead: Harvester Wheatsheaf, 1992), 87.\n\n. Katie King, \"Audre Lorde's Lacquered Layerings: The Lesbian Bar as a Site of Literary Production,\" in Munt, ed., _New Lesbian Criticism_ , 71.\n\n. Ekua Omosupe, \"Black\/Lesbian\/Bulldagger,\" _Differences_ 3.2 (Summer 1991): 101\u201311 (quote at 104).\n\n. Jackie Goldsby, \"What It Means to Be Colored Me,\" _Outlook: National Gay and Lesbian Quarterly_ 9 (Summer 1990): 11; quoted in Rich, \"When Difference Is (More Than) Skin Deep,\" in Gever, Greyson, and Parmar, eds., _Queer Looks_ , 327 (Goldsby's emphasis)\n\n. William G. Hawkeswood, _One of the Children_ (Berkeley: University of California Press, 1996), 155\u201357.\n\n. Joseph Beam, ed., _In the Life_ (Boston: Alyson, 1986); Essex Hemphill, ed., _Brother to Brother_ (Boston: Alyson, 1991).\n\n. Reginald T. Jackson, \"The Absence of Fear,\" in Hemphill, ed., _Brother to Brother_ , 207.\n\n. Quoted from Marlon Riggs's poem, \"Tongues Untied,\" in Hemphill, ed., _Brother to Brother_ , 202. The poem affords the backbone to the film.\n\n. Joseph Beam, \"Brother to Brother: Words from the Heart,\" in Beam, ed., _In the Life_ , 240.\n\n. See Ron Simmons, \" _Tongues Untied_ : An Interview with Marlon Riggs,\" in Hemphill, ed., _Brother to Brother_.\n\n. Rich, \"When Difference Is (More Than) Skin Deep,\" in Gever, Greyson, and Parmar, eds., _Queer Looks_ , 333.\n\n. Jackie Goldsby, \"Queens of Language: _Paris Is Burning_ ,\" in Gever et al., eds, _Queer Looks_ , 114.\n\n. Marlon Riggs, \"Black Macho Revisited,\" in Hemphill, ed., _Brother to Brother_ , 254 (Riggs's emphasis).\n\n. Kobena Mercer and Isaac Julien, \"Race, Sexual Politics, and Black Masculinity: A Dossier,\" in Rowena Chapman and Jonathan Rutherford, eds., _Male Order_ (London: Lawrence and Wishart, 1988), 112 (my elision). See Lynne Segal, _Slow Motion_ (London: Virago, 1990), ch. 7.\n\n. Phillip Brian Harper, _Are We Not Men?_ (New York and Oxford: Oxford University Press, 1996), x.\n\n. Kheven L. LaGrone, \"Beneath the Veneer,\" in Charles Michael Smith, ed., _Fighting Words_ (New York: Avon, 1999). On continuities in images of slavery, see David Marriott, _On Black Men_ (Edinburgh: Edinburgh University Press, 2000). The issue is flagged in Lynda Hart and Joshua Dale, \"Sadomasochism,\" in Andy Medhurst and Sally R. Munt, eds., _Lesbian and Gay Studies_ (London: Cassell, 1997), 351.\n\n. Kevin J. Mumford, _Interzones: Black\/White Sex Districts in Chicago and New York_ (New York: Columbia University Press, 1997), 73.\n\n. Mercer and Julien, \"Race, Sexual Politics, and Black Masculinity,\" in Chapman and Rutherford, eds., _Male Order_ , 128. This Gary has no connection with Gary Fisher (on whom see below).\n\n. See Marcus Wood, _Blind Memory_ (Manchester: Manchester University Press, 2000), 43\u201346.\n\n. Isaac Julien, \"Confessions of a Snow Queen: Notes on the Making of _The Attendant_ ,\" _Critical Quarterly_ 36.1 (Spring 1994): 120\u201326 (quotes at 123).\n\n. Gary Fisher, _Gary in Your Pocket_ , ed. Eve Kosofsky Sedgwick (Durham and London: Duke University Press, 1996), 230\u201331. See Eric L. McKitrick, _Slavery Defended: The Views of the Old South_ (Englewood Cliffs, N.J.: Prentice-Hall, 1963).\n\n. Fisher, _Gary in Your Pocket_ , 282.\n\n. David Wojnarowicz, _Close to the Knives_ (1991) (London: Serpent's Tail, 1992), 255, 266, 271.\n\n. John Rechy, _City of Night_ (1963) (London: MacGibbon and Kee, 1964), 15\u201321.\n\n. Paul Monette, _Halfway Home_ (New York: Crown, 1991), 32, 41 (Monette's emphasis).\n\n. James Robert Baker, _Tim and Pete_ (1995) (London: Fourth Estate, 1996), 35\u201336, 166.\n\n. From a letter from Fisher to his sister, quoted in Eve Kosofsky Sedgwick, \"Gary Fisher in Your Pocket,\" in Joshua Oppenheimer and Helena Reckitt, eds., _Acting on AIDS_ (London: Serpent's Tail and ICA, 1997), 414.\n\n. See Alan Sinfield, _Literature, Politics, and Culture in Postwar Britain_ , 2d ed. (London: Athlone, 1997), 121\u201324; Erroll Lawrence, \"Just Plain Common Sense: The 'Roots' of Racism,\" in Centre for Contemporary Cultural Studies, _The Empire Strikes Back_ (London: Hutchinson, 1982).\n\n. See Marriott, _On Black Men_ , 34\u201341.\n\n. Tennessee Williams, \"Desire and the Black Masseur,\" in Williams, _Collected Short Stories_ (New York: Ballantine, 1985), 220.\n\n. Bergman, _Gaiety Transfigured_ , 156\u201357. Bergman finds cannibalism linked with homosexuality also by Yukio Mishima, Herman Melville, Freud, and Tobias Schneebaum.\n\n. Tennessee Williams, _Suddenly Last Summer_ (1958), in Williams, _\"Orpheus Descending,\" \"Something Unspoken,\" \"Suddenly Last Summer\"_ (Harmondsworth: Penguin, 1961), 188, 142.\n\n. James Robert Baker, _Testosterone_ (Los Angeles: Alyson Books, 2000), 7.\n\n. Eve Kosofsy Sedgwick, \"Queer Performativity: Henry James's _The Art of the Novel_ ,\" _GLQ_ ( _Gay and Lesbian Quarterly_ ) 1 (1993): 1\u201316; Judith Butler, \"Critically Queer,\" ibid., 17\u201332. See further Sally R. Munt, _Heroic Desire_ (London: Cassell, 1998), ch. 4; Michael Warner, _The Trouble with Normal_ (New York: Free Press, 1999), ch. 1; Douglas Crimp, \"Mario Montez, For Shame,\" in Stephen M. Barber and David L. Clark, eds., _Regarding Sedgwick_ (New York: Routledge, 2002).\n\n. Tennessee Williams, _Sweet Bird of Youth_ (1959), in Williams, _\"Sweet Bird of Youth,\" \"A Streetcar Named Desire,\" \"The Glass Menagerie\"_ (Harmondsworth: Penguin, 1962), 42 (my emphasis). See Neil Bartlett, _Ready to Catch Him Should He Fall_ (London: Serpent's Tail, 1990), 13.\n\n**9. FICTION**\n\n. _The Shorter Oxford English Dictionary_ , 3d ed. (Oxford: Clarendon, 1964). The rules and books specified in the dictionary are those of the Christian churches. See Reed Woodhouse, _Unlimited Embrace: A Canon of Gay Fiction, 1945\u20131995_ (Amherst: University of Massachusetts Press, 1998).\n\n. Alison Hennegan, \"On Becoming a Lesbian Reader,\" in Susannah Radstone, ed., _Sweet Dreams_ (London: Lawrence and Wishart, 1988), 169\u201371.\n\n. Stanley Fish, _Professional Correctness_ (Oxford: Clarendon, 1995), 2; Alan Sinfield, _Faultlines_ (Berkeley: University of California Press, 1992; Oxford: Clarendon Press, 1992), 288\u201390.\n\n. Fish, _Professional Correctness_ , 44\u201345 (his emphasis).\n\n. John Guillory, _Cultural Capital_ (Chicago: University of Chicago Press, 1993), 1\u201314.\n\n. Stuart Hall, \"New Ethnicities,\" in James Donald and Ali Rattansi, eds., _\"Race,\" Culture, and Difference_ (London: Sage, 1992), 254 (Hall's emphasis). See Alan Sinfield, _Gay and After_ (London: Serpent's Tail, 1998), chs. 1 and 2; Carrie Tirado Bramen, \"Why the Academic Left Hates Identity Politics,\" _Textual Practice_ 16 (2002): 1\u201311.\n**INDEX**\n\nabuse\n\nAckerley, J.R.\n\nACT UP\n\nAdair, G.\n\n_Adventures of Priscilla, Queen of the Desert, The_\n\nAfrican Americans\n\nage\n\nAIDS\n\n_AKA_\n\nAlderson, D.\n\n_Alive and Kicking_\n\nAllen, C.\n\nAllison, D.\n\nAlthusser, L.\n\nAltman, D\n\nAndros, P.\n\nanthropology\n\nArditti, M.\n\narrested development\n\n_Attendant, The_\n\nAuden, W. H.\n\nBacon, F.\n\nBaker, J. R.\n\nBaldwin, J.\n\nBarney, N.\n\nBartlett, N.\n\nBawer, B.\n\nBeadle, J.\n\nBeam, J.\n\n_Beautiful Thing_\n\nBech, H.\n\nBellow, S.\n\nBergling, T.\n\nBergman, D.\n\nBersani, L.\n\n_Bigger Splash, A_\n\n_Birdcage, The_\n\nbisexuals\n\nBlack Panthers\n\nBloom, A.\n\nBornstein, K.\n\nBoulton, S.\n\n_Boys Don't Cry_\n\n_Brandon Teena Story, The_\n\nBray, A.\n\nBrecht, B.\n\nBritain\n\nBronski, M.\n\nBrown, R. M.\n\nbullying\n\nBush, G.\n\nbutch\/femme\n\nButler, J.\n\nButler, T.\n\nCahill, P. T.\n\nCalifia, P.\n\nCameron, P.\n\ncamp\n\ncanon\n\ncapitalism\n\nCard, C.\n\n_Carousel_\n\nCase, S.-E.\n\nCastle, T.\n\nCatullus\n\n_Chasing Amy_\n\nChauncey, G. Jr.\n\nChesney, R.\n\nClarke, C.\n\nClarke, F.\n\nClarke, J.\n\nclass\n\nCleaver, E.\n\nCobham, R.\n\nCoetzee, J. M.\n\ncoming out\n\n_Coming Out_\n\nconflation of roles\n\nCooper, D.\n\nCorbin, S.\n\nCory, D. W.\n\nCoward, N.\n\nCresswell, N.\n\nCrisp, Q.\n\nCrowley, M.\n\n_Crying Game, The_\n\nCullwick, H.\n\ncultural materialism\n\ncultural capital\n\nCunningham, M.\n\nDahmer, J.\n\nDale, J.\n\nde Lauretis, T.\n\nDebord, G.\n\nDelany, S. R.\n\nDerrida, J.\n\ndesire-for\/desire-to-be\n\nDickson, J.\n\n_Different for Girls_\n\nDoan, L.\n\n_Doing It with You Is Taboo_\n\nDollimore, J.\n\nDonne, J.\n\nDonoghue, E.\n\nDonovan, C.\n\ndrag\n\nDrexel, A.\n\nDrinnan, N.\n\nDuncker, P.\n\nDuplechan, L.\n\nDworkin, A.\n\nDyer, R.\n\n_Early Frost, An_\n\nEbershoff, D.\n\negalitarian ideology\n\nEllis, H.\n\nEminem\n\n_Entre Nous_\n\n_fa'affines_\n\nfantasy\n\nFarrakhan, L.\n\nFeinberg, D. B.\n\nFeinberg, L.\n\nfemininity\n\nFerenczi, S.\n\nFicino, M.\n\nFierstein, H.\n\nFish, S.\n\nFisher, G.\n\nFletcher, J.\n\nForster, E. M.\n\nFoucault, M.\n\nFraser, S.\n\nFreud, S.\n\n_Fruit Machine, The_\n\nGarland, J.\n\nGarland, R.\n\nGarlinger, P. P.\n\ngaze, the\n\nGeertz, C.\n\ngender identity\n\nGenet, J.\n\nGide, A.\n\nGilroy, P.\n\nGlazier, L.\n\nGoldsby, J.\n\nGorer, G.\n\nGove, B.\n\nGray, S.\n\nGreeks\n\nGreen, R.\n\nGreenberg, D.\n\nGrimsley, J.\n\nGrosz, E.\n\nGuillory, J.\n\nGunn, T.\n\nHalberstam, J.\n\nHall, R.\n\nHalperin, D.\n\nHammerstein, O.\n\nHansen, J.\n\nHarper, P. B.\n\nHart, L.\n\nHartnett, P-P.\n\nHarvey, J.\n\nHawkeswood, W. G.\n\nHazan, J.\n\nHealy, M.\n\nHeaphy, B.\n\nHekma, G.\n\nHemphill, E.\n\nHennegan, A.\n\nHerdt, G.\n\nhierarchies\n\nHoad, N.\n\nHockney, D.\n\nHoffman, W. F.\n\nHolleran, A.\n\nHollibaugh, A.\n\nHollinghurst, A.\n\n_Hollow Reed_\n\nHolmes, R.\n\nHomes, A. M.\n\nhomophobia\n\nhooks, b.\n\nHopkins, J.\n\nHouse, R.\n\nHouser, W.\n\nHsien-yung, P.\n\nhustling\n\nidentity\n\nimperialism\n\nintersex\n\ninversion\n\nIreland, T.\n\nIrigaray, L.\n\nIsherwood, C.\n\nJackson, E., Jr.\n\nJackson, R. T.\n\nJanMohamed, A.\n\nJeffreys, S.\n\nJohnson, D. K.\n\nJulien, I.\n\nKaplan, C.\n\nKaufman, M.\n\nKay, J.\n\nKhambatta, A.\n\nKoestenbaum, W.\n\nKrafft-Ebing, R. von\n\nKramer, L.\n\nKulick, D.\n\n_La Cage aux Folles_\n\nLaGrone, K. L.\n\nLaing, R. D.\n\nLaplanche, J.\n\nLatin Americans\n\nLawrence, D. H.\n\nLeavitt, D.\n\nLeeming, D.\n\nlesbians\n\nLeVay, S.\n\nL\u00e9vi-Strauss, C.\n\nLewes, K.\n\nLister, A.\n\nliterary criticism\n\n_Living End, The_\n\nLlangollen, Ladies of\n\n_Looking for Langston_\n\nLorde, A.\n\nLovat, S.\n\n_Love and Death on Long Island_\n\n_Love Is the Devil_\n\n_Love! Valour! Compassion!_\n\nLowe, B.\n\n_Ma vie en rose_\n\nMacKinnon, C. A.\n\nMaddison, S.\n\nMains, G.\n\n_Make Me a Man_\n\nMann, T.\n\nMarcuse, H.\n\nMarlowe, K.\n\nMarshall, J.\n\nMars-Jones, A.\n\nMartin, K.\n\nmasculinity\n\nmasochism\n\nmentoring\n\nMason-John, V.\n\nMaugham, R.\n\nMaugham, S.\n\nMaupin, A.\n\nMcLean, H.\n\nMcNally, T.\n\nMelson J. K.\n\nMercer, K.\n\nMills, J.\n\nMills, J. E.\n\nMoll, A.\n\nmollies\n\nMonette, P.\n\nMoore, M.\n\nMoore, O.\n\nMoraga, C.\n\nMoran, L. J.\n\nMulvey, L.\n\nMumford, K. J.\n\nMunby, A.\n\nMunt, S.\n\nMurdoch, I.\n\nMurray, S. O.\n\n_My Beautiful Laundrette_\n\n_My Own Private Idaho_\n\nNaipaul, V. S.\n\nnarcissism\n\nnarrative\n\nNavratilova, M.\n\nNewton, E.\n\nNewton, H.\n\nNgcobo, L.\n\n_No Night Is Too Long_\n\nNonas, E.\n\nNorton, R.\n\nobject-choice\n\nOedipus complex\n\nO'Hara, S.\n\nO'Neill, J.\n\nOmosupe, E.\n\nOrton, J.\n\nOsment, P.\n\n_Paradise Bent_\n\nPark, F.\n\nPater, W.\n\nPeake, T.\n\npedophilia\n\nPenley, C.\n\nPet Shop Boys\n\n_Philadelphia_\n\nphobia\n\nPicano, F.\n\nPlato\n\nPlummer, K.\n\nPontalis, J. B.\n\npornography\n\nPosener, J.\n\npoststructuralism\n\npromiscuity\n\nProsser, J.\n\nPym, B.\n\nQuinn, J.\n\nrace\n\nracism\n\nRattigan, T.\n\nRavenhill, M.\n\n_Rebel Without a Cause_\n\nRechy, J.\n\nRenault, M.\n\nreproduction\n\nRich, B. R.\n\nRiggs, M.\n\nRiviere, J.\n\n_Rocky II_\n\nRodgers, R.\n\nroles\n\nRousseau, G.\n\nRussell, P.\n\nSandys, J.\n\nSartre, J.-P.\n\n_Scarlet Pansy, The_\n\nScherzinger, M.\n\nSchulman, S.\n\nSedgwick, E. K.\n\nseduction\n\nSegal, L.\n\nSegal, N.\n\nserial killers\n\n_Servant, The_\n\nsexiness\n\nShakespeare, W.\n\nshame\n\nShaw, A.\n\nShepard, M.\n\nShepherd, R.\n\nShilts, R.\n\nSidney, P.\n\nSilverman, K.\n\nSituationism\n\n_skesana_ boys\n\nSmith, B. R.\n\nSouth Africa\n\nSpark, M.\n\nSpender, S.\n\nSpurlin, W. K.\n\nStallybrass, P.\n\nStanton, M.\n\n_Star Trek_\n\nStein, G.\n\nStone, S.\n\n_Stonewall_\n\nStonewall Riots\n\nStraayer, C.\n\nStreisand, B.\n\nsubculture\n\n_Suddenly Last Summer_\n\nSullivan, A.\n\nSymonds, J. A.\n\nS\/M\n\n_Talented Mr. Ripley, The_\n\n_Taxi zum Klo_\n\nTaylor, D.\n\nTennyson, A.\n\n_To Die For_\n\ntoilets\n\n_Tongues Untied_\n\n_To Wong Foo, Thanks for Everything, Julie Newmar_\n\nToynbee, P.\n\ntransgender, transsexuals\n\nTripp, C. A.\n\nTyler, C.-A.\n\nUnited States of America\n\nVaneigem, R.\n\nVicinus, M.\n\n_Victim_\n\nVivienne, R.\n\nVolcano, Del LaGrace\n\n_Voyage en Douce_\n\nWakefield, T.\n\nWalker, A.\n\nWalkerdine, V.\n\nWallace, W.\n\nWalton, S.\n\nWard, C. M.\n\nWarner, M.\n\nWaugh, A.\n\n_Wedding Banquet, The_\n\nWeeks, J.\n\nWest, D. J.\n\nWhite, A.\n\nWhite, E.\n\nWhitman, W.\n\nWilde, O.\n\nWildeblood, P.\n\nWillard, G.\n\nWilliams, R.\n\nWilliams, T.\n\nWilson, Angus\n\nWilson, Anna\n\nWilton, T.\n\nWinckelmann, J.\n\nWinterson, J.\n\nWojnarowicz, D.\n\n_Wonderland_\n\nWoods, G.\n\nWoolf, V.\n\nZilinsky, U.\n**BETWEEN MEN \u223c BETWEEN WOMEN**   **LESBIAN, GAY, AND BISEXUAL STUDIES**\n\nTerry Castle and Larry Gross, Editors\n\nRichard D. Mohr, _Gays\/Justice: A Study of Ethics, Society, and Law_\n\nGary David Comstock, _Violence Against Lesbians and Gay Men_\n\nKath Weston, _Families We Choose: Lesbians, Gays, Kinship_\n\nLillian Faderman, _Odd Girls and Twilight Lovers: A History of Lesbian Life in Twentieth-Century America_\n\nJudith Roof, _A Lure of Knowledge: Lesbian Sexuality and Theory_\n\nJohn Clum, _Acting Gay: Male Homosexuality in Modern Drama_\n\nAllen Ellenzweig, _The Homoerotic Photograph: Male Images from Durieu\/Delacroix to Mapplethorpe_\n\nSally Munt, editor, _New Lesbian Criticism: Literary and Cultural Readings_\n\nTimothy F. Murphy and Suzanne Poirier, editors, _Writing AIDS: Gay Literature, Language, and Analysis_\n\nLinda D. Garnets and Douglas C. Kimmel, editors, _Psychological Perspectives on Lesbian and Gay Male Experiences_ (2nd edition)\n\nLaura Doan, editor, _The Lesbian Postmodern_\n\nNoreen O'Connor and Joanna Ryan, _Wild Desires and Mistaken Identities: Lesbianism and Psychoanalysis_\n\nAlan Sinfield, _The Wilde Century: Effeminacy, Oscar Wilde, and the Queer Moment_\n\nClaudia Card, _Lesbian Choices_\n\nCarter Wilson, _Hidden in the Blood: A Personal Investigation of AIDS in the Yucat\u00e1n_\n\nAlan Bray, _Homosexuality in Renaissance England_\n\nJoseph Carrier, _De Los Otros: Intimacy and Homosexuality Among Mexican Men_\n\nJoseph Bristow, _Effeminate England: Homoerotic Writing After 1885_\n\nCorinne E. Blackmer and Patricia Juliana Smith, editors, _En Travesti: Women, Gender Subversion, Opera_\n\nDon Paulson with Roger Simpson, _An Evening at The Garden of Allah: A Gay Cabaret in Seattle_\n\nClaudia Schoppmann, _Days of Masquerade: Life Stories of Lesbians During the Third Reich_\n\nChris Straayer, _Deviant Eyes, Deviant Bodies: Sexual Re-Orientation in Film and Video_\n\nEdward Alwood, _Straight News: Gays, Lesbians, and the News Media_\n\nThomas Waugh, _Hard to Imagine: Gay Male Eroticism in Photography and Film from Their Beginnings to Stonewall_\n\nJudith Roof, _Come As You Are: Sexuality and Narrative_\n\nTerry Castle, _Noel Coward and Radclyffe Hall: Kindred Spirits_\n\nKath Weston, _Render Me, Gender Me: Lesbians Talk Sex, Class, Color, Nation, Studmuffins..._\n\nRuth Vanita, _Sappho and the Virgin Mary: Same-Sex Love and the English Literary Imagination_\n\nren\u00e9e c. hoogland, _Lesbian Configurations_\n\nBeverly Burch, _Other Women: Lesbian Experience and Psychoanalytic Theory of Women_\n\nJane McIntosh Snyder, _Lesbian Desire in the Lyrics of Sappho_\n\nRebecca Alpert, _Like Bread on the Seder Plate: Jewish Lesbians and the Transformation of Tradition_\n\nEmma Donoghue, editor, _Poems Between Women: Four Centuries of Love, Romantic Friendship, and Desire_\n\nJames T. Sears and Walter L. Williams, editors, _Overcoming Heterosexism and Homophobia: Strategies That Work_\n\nPatricia Juliana Smith, _Lesbian Panic: Homoeroticism in Modern British Women's Fiction_\n\nDwayne C. Turner, _Risky Sex: Gay Men and HIV Prevention_\n\nTimothy F. Murphy, _Gay Science: The Ethics of Sexual Orientation Research_\n\nCameron McFarlane, _The Sodomite in Fiction and Satire, 1660\u20131750_\n\nLynda Hart, _Between the Body and the Flesh: Performing Sadomasochism_\n\nByrne R. S. Fone, editor, _The Columbia Anthology of Gay Literature: Readings from Western Antiquity to the Present Day_\n\nEllen Lewin, _Recognizing Ourselves: Ceremonies of Lesbian and Gay Commitment_\n\nRuthann Robson, _Sappho Goes to Law School: Fragments in Lesbian Legal Theory_\n\nJacquelyn Zita, _Body Talk: Philosophical Reflections on Sex and Gender_\n\nEvelyn Blackwood and Saskia Wieringa, _Female Desires: Same-Sex Relations and Transgender Practices Across Cultures_\n\nWilliam L. Leap, ed., _Public Sex\/Gay Space_\n\nLarry Gross and James D. Woods, eds., _The Columbia Reader on Lesbians and Gay Men in Media, Society, and Politics_\n\nMarilee Lindemann, _Willa Cather: Queering America_\n\nGeorge E. Haggerty, _Men in Love: Masculinity and Sexuality in the Eighteenth Century_\n\nAndrew Elfenbein, _Romantic Genius: The Prehistory of a Homosexual Role_\n\nGilbert Herdt and Bruce Koff, _Something to Tell You: The Road Families Travel When a Child Is Gay_\n\nRichard Canning, _Gay Fiction Speaks: Conversations with Gay Novelists_\n\nLaura Doan, _Fashioning Sapphism: The Origins of a Modern English Lesbian Culture_\n\nMary Bernstein and Renate Reimann, eds., _Queer Families, Queer Politics: Challenging Culture and the State_\n\nRichard R. Bozorth, _Auden's Games of Knowledge: Poetry and the Meanings of Homosexuality_\n\nLarry Gross, _Up from Invisibility: Lesbians, Gay Men, and the Media in America_\n\nLinda Garber, _Identity Poetics: Race, Class, and the Lesbian-Feminist Roots of Queer Theory_\n\nRichard Canning, _Hear Us Out: Conversations with Gay Novelists_\n\nDavid Bergman, _The Violet Hour: The Violet Quill and the Making of Gay Culture_\n\nKatherine Sender, _Business, Not Politics: The Making of the Gay Market_\n","meta":{"redpajama_set_name":"RedPajamaBook"}}
+{"text":" \n**Machine Learning with Swift\n\n**\n\nArtificial Intelligence for iOS\n\nAlexander Sosnovshchenko\n\n****BIRMINGHAM - MUMBAI****\n\n# Machine Learning with Swift\n\nCopyright \u00a9 2018 Packt Publishing\n\nAll rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, without the prior written permission of the publisher, except in the case of brief quotations embedded in critical articles or reviews.\n\nEvery effort has been made in the preparation of this book to ensure the accuracy of the information presented. However, the information contained in this book is sold without warranty, either express or implied. Neither the author, nor Packt Publishing or its dealers and distributors, will be held liable for any damages caused or alleged to have been caused directly or indirectly by this book.\n\nPackt Publishing has endeavored to provide trademark information about all of the companies and products mentioned in this book by the appropriate use of capitals. However, Packt Publishing cannot guarantee the accuracy of this information.\n\n**Commissioning Editor:** Veena Pagare  \n**Acquisition Editor:** Vinay Argekar  \n**Content Development Editor:** Mayur Pawanikar  \n**Technical Editor:** Dinesh Pawar  \n**Copy Editor:** Vikrant Phadkay, Safis Editing  \n**Project Coordinator:** Nidhi Joshi  \n**Proofreader:** Safis Editing  \n**Indexer:** Pratik Shirodkar  \n**Graphics:** Tania Dutta  \n**Production Coordinator:** Arvindkumar Gupta\n\nFirst published: February 2018\n\nProduction reference: 1270218\n\nPublished by Packt Publishing Ltd.  \nLivery Place  \n35 Livery Street  \nBirmingham  \nB3 2PB, UK.\n\nISBN 978-1-78712-151-5\n\nwww.packtpub.com\n\nmapt.io\n\nMapt is an online digital library that gives you full access to over 5,000 books and videos, as well as industry leading tools to help you plan your personal development and advance your career. For more information, please visit our website.\n\n# Why subscribe?\n\n  * Spend less time learning and more time coding with practical eBooks and Videos from over 4,000 industry professionals\n\n  * Improve your learning with Skill Plans built especially for you\n\n  * Get a free eBook or video every month\n\n  * Mapt is fully searchable\n\n  * Copy and paste, print, and bookmark content\n\n# PacktPub.com\n\nDid you know that Packt offers eBook versions of every book published, with PDF and ePub files available? You can upgrade to the eBook version at www.PacktPub.com and as a print book customer, you are entitled to a discount on the eBook copy. Get in touch with us at `service@packtpub.com` for more details.\n\nAt www.PacktPub.com, you can also read a collection of free technical articles, sign up for a range of free newsletters, and receive exclusive discounts and offers on Packt books and eBooks.\n\n# Contributors\n# About the author\n\n**Alexander Sosnovshchenko** has been working as an iOS software engineer since 2012. Later he made his foray into data science, from the first experiments with mobile machine learning in 2014, to complex deep learning solutions for detecting anomalies in video surveillance data. He lives in Lviv, Ukraine, and has a wife and a daughter.\n\nThanks to Dmitrii Vorona for moral support, invaluable advice, and code reviews; Nikolay Sosnovshchenko and Oksana Matskovich for the help with pictures of creatures and androids; David Kopec and Matthijs Hollemans for their open source projects; Mr. Jojo Moolayil for his efforts and expertise as a contributing author and reviewer; and my family for being supportive and patient.\n\n# About the reviewers\n\n**Jojo Moolayil** is an artificial intelligence, deep learning, and machine learning professional with over 5 years of experience and is the author of _Smarter Decisions \u2013 The Intersection of Internet of Things and Decision Science_. He works with GE and lives in Bengaluru, India. He has also been a technical reviewer about various books in machine learning, deep learning, and business analytics with Apress and Packt.\n\nI would like to thank my family, friends, and mentors.\n\n**Cecil Costa** , also known as Eduardo Campos in Latin American countries, is a Euro-Brazilian freelance developer who has been learning about computers since he got his first PC in 1990. Learning is his passion, and so is teaching; this is why he works as a trainer. He has organized both on-site and online courses for companies. He is also the author of a few Swift books.\n\nI'd like to thank Maximilian Ambergis for creating the delete key; it has been very useful for me!\n\n# Packt is searching for authors like you\n\nIf you're interested in becoming an author for Packt, please visit authors.packtpub.com and apply today. We have worked with thousands of developers and tech professionals, just like you, to help them share their insight with the global tech community. You can make a general application, apply for a specific hot topic that we are recruiting an author for, or submit your own idea.\n\n# Table of Contents\n\n  1. Title Page\n  2. Copyright and Credits\n    1. Machine Learning with Swift\n  3. Packt Upsell\n    1. Why subscribe?\n    2. PacktPub.com\n  4. Contributors\n    1. About the author\n    2. About the reviewers\n    3. Packt is searching for authors like you\n  5. Preface\n    1. Who this book is for\n    2. What this book covers\n    3. To get the most out of this book\n      1. Download the example code files\n      2. Download the color images\n      3. Conventions used\n    4. Get in touch\n      1. Reviews\n  6. Getting Started with Machine Learning\n    1. What is AI?\n    2. The motivation behind ML\n    3. What is ML ?\n    4. Applications of ML\n      1. Digital signal processing (DSP)\n      2. Computer vision\n      3. Natural language processing (NLP)\n      4. Other applications of ML\n    5. Using ML to build smarter iOS applications\n    6. Getting to know your data\n      1. Features\n        1. Types of features\n        2. Choosing a good set of features\n      2. Getting the dataset\n      3. Data preprocessing\n    7. Choosing a model\n      1. Types of ML algorithms\n      2. Supervised learning\n      3. Unsupervised learning\n      4. Reinforcement learning\n      5. Mathematical optimization\u00a0\u2013 how learning works\n      6. Mobile versus server-side ML\n      7. Understanding mobile platform limitations\n    8. Summary\n    9. Bibliography\n  7. Classification \u2013 Decision Tree Learning\n    1. Machine learning toolbox\n    2. Prototyping the first machine learning app\n      1. Tools\n      2. Setting up a machine learning environment\n    3. IPython notebook crash course\n    4. Time to practice\n    5. Machine learning for extra-terrestrial life explorers\n    6. Loading the dataset\n    7. Exploratory data analysis\n    8. Data preprocessing\n      1. Converting categorical variables\n      2. Separating features from labels\n      3. One-hot encoding\n      4. Splitting the data\n    9. Decision trees everywhere\n    10. Training the decision tree classifier\n      1. Tree visualization\n      2. Making predictions\n      3. Evaluating accuracy\n      4. Tuning hyperparameters\n      5. Understanding model capacity trade-offs\n    11. How decision tree learning works\n      1. Building a tree automatically from data\n      2. Combinatorial entropy\n      3. Evaluating performance of the model with data\n        1. Precision, recall, and F1-score\n        2. K-fold cross-validation\n        3. Confusion matrix\n    12. Implementing first machine learning app in Swift\n    13. Introducing Core ML\n      1. Core ML features\n      2. Exporting the model for iOS\n      3. Ensemble learning random forest\n      4. Training the random forest\n      5. Random forest accuracy evaluation\n      6. Importing the Core ML model into an iOS project\n      7. Evaluating performance of the model on iOS\n        1. Calculating the confusion matrix\n      8. Decision tree learning pros and cons\n    14. Summary\n  8. K-Nearest Neighbors Classifier\n    1. Calculating the distance\n      1. DTW\n      2. Implementing DTW in Swift\n    2. Using instance-based models for classification and clustering\n    3. People motion recognition using inertial sensors\n    4. Understanding the KNN algorithm\n      1. Implementing KNN in Swift\n    5. Recognizing human motion using KNN\n      1. Cold start problem\n      2. Balanced dataset\n      3. Choosing a good k\n    6. Reasoning in high-dimensional spaces\n    7. KNN pros\n    8. KNN cons\n    9. Improving our solution\n      1. Probabilistic interpretation\n      2. More data sources\n      3. Smarter time series chunking\n      4. Hardware acceleration\n      5. Trees to speed up the inference\n      6. Utilizing state transitions\n    10. Summary\n    11. Bibliography\n  9. K-Means Clustering\n    1. Unsupervised learning\n    2. K-means clustering\n    3. Implementing k-means in Swift\n      1. Update step\n      2. Assignment step\n    4. Clustering objects on a map\n    5. Choosing the number of clusters\n    6. K-means clustering\u00a0\u2013 problems\n    7. K-means++\n    8. Image segmentation using k-means\n    9. Summary\n  10. Association Rule Learning\n    1. Seeing association rules\n    2. Defining data structures\n    3. Using association measures to assess rules\n      1. Supporting association measures\n      2. Confidence association measures\n      3. Lift association measures\n      4. Conviction association measures\n    4. Decomposing the problem\n    5. Generating all possible rules\n    6. Finding frequent item sets\n    7. The Apriori algorithm\n    8. Implementing Apriori in Swift\n    9. Running Apriori\n    10. Running Apriori on real-world data\n    11. The pros and cons of Apriori\n    12. Building an adaptable user experience\n    13. Summary\n    14. Bibliography\n  11. Linear Regression and Gradient Descent\n    1. Understanding the regression task\n    2. Introducing simple linear regression\n      1. Fitting a regression line using the least squares method\n        1. Where to use GD and normal equation\n        2. Using gradient descent for function minimization\n      2. Forecasting the future with simple linear regression\n    3. Feature scaling\n    4. Feature standardization\n      1. Multiple linear regression\n    5. Implementing multiple linear regression in Swift\n      1. Gradient descent for multiple linear regression\n        1. Training multiple regression\n        2. Linear algebra operations\n      2. Feature-wise standardization\n        1. Normal equation for multiple linear regression\n      3. Understanding and overcoming the limitations of linear regression\n    6. Fixing linear regression problems with regularization\n      1. Ridge regression and Tikhonov regularization\n        1. LASSO regression\n      2. ElasticNet regression\n    7. Summary\n    8. Bibliography\n  12. Linear Classifier and Logistic Regression\n    1. Revisiting the classification task\n      1. Linear classifier\n      2. Logistic regression\n    2. Implementing logistic regression in Swift\n      1. The prediction part of logistic regression\n      2. Training the logistic regression\n      3. Cost function\n    3. Predicting user intents\n      1. Handling dates\n    4. Choosing the regression model for your problem\n    5. Bias-variance trade-off\n    6. Summary\n  13. Neural Networks\n    1. What are artificial NNs anyway?\n    2. Building the neuron\n      1. Non-linearity function\n        1. Step-like activation functions\n        2. Rectifier-like activation functions\n    3. Building the network\n    4. Building a neural layer in Swift\n    5. Using neurons to build logical functions\n    6. Implementing layers in Swift\n    7. Training the network\n      1. Vanishing gradient problem\n      2. Seeing biological analogies\n    8. Basic neural\u00a0network subroutines (BNNS)\n      1. BNNS example\n    9. Summary\n  14. Convolutional Neural Networks\n    1. Understanding users emotions\n    2. Introducing computer vision problems\n    3. Introducing convolutional neural networks\n    4. Pooling operation\n    5. Convolution operation\n      1. Convolutions in CNNs\n    6. Building the network\n      1. Input layer\n      2. Convolutional layer\n      3. Fully-connected layers\n      4. Nonlinearity layers\n      5. Pooling layer\n      6. Regularization layers\n        1. Dropout\n        2. Batch normalization\n    7. Loss functions\n    8. Training the network\n    9. Training the CNN for facial expression recognition\n    10. Environment setup\n    11. Deep learning frameworks\n      1. Keras\n    12. Loading the data\n    13. Splitting the data\n    14. Data augmentation\n    15. Creating the network\n    16. Plotting the network structure\n    17. Training the network\n    18. Plotting loss\n    19. Making predictions\n    20. Saving the model in HDF5 format\n    21. Converting to Core ML format\n    22. Visualizing convolution filters\n    23. Deploying CNN to iOS\n    24. Summary\n    25. Bibliography\n  15. Natural Language Processing\n    1. NLP in the mobile development world\n    2. Word Association game\n    3. Python NLP libraries\n    4. Textual corpuses\n    5. Common NLP approaches and subtasks\n      1. Tokenization\n      2. Stemming\n      3. Lemmatization\n      4. Part-of-speech (POS) tagging\n      5. Named entity recognition (NER)\n      6. Removing stop words and punctuation\n    6. Distributional semantics hypothesis\n    7. Word vector representations\n    8. Autoencoder neural networks\n    9. Word2Vec\n    10. Word2Vec in Gensim\n    11. Vector space properties\n    12. iOS application\n      1. Chatbot anatomy\n      2. Voice input\n      3. NSLinguisticTagger and friends\n      4. Word2Vec on iOS\n      5. Text-to-speech output\n      6. UIReferenceLibraryViewController\n      7. Putting it all together\n    13. Word2Vec friends and relatives\n    14. Where to go from here?\n    15. Summary\n  16. Machine Learning Libraries\n    1. Machine learning and AI APIs\n    2. Libraries\n    3. General-purpose machine learning libraries\n      1. AIToolbox\n      2. BrainCore\n      3. Caffe\n      4. Caffe2\n      5. dlib\n      6. FANN\n      7. LearnKit\n      8. MLKit\n      9. Multilinear-math\n      10. MXNet\n      11. Shark\n      12. TensorFlow\n      13. tiny-dnn\n      14. Torch\n      15. YCML\n    4. Inference-only libraries\n      1. Keras\n      2. LibSVM\n      3. Scikit-learn\n      4. XGBoost\n    5. NLP libraries\n      1. Word2Vec\n      2. Twitter text\n    6. Speech recognition\n      1. TLSphinx\n      2. OpenEars\n    7. Computer vision\n      1. OpenCV\n      2. ccv\n      3. OpenFace\n      4. Tesseract\n    8. Low-level subroutine libraries\n      1. Eigen\n      2. fmincg-c\n      3. IntuneFeatures\n      4. SigmaSwiftStatistics\n      5. STEM\n      6. Swix\n      7. LibXtract\n      8. libLBFGS\n      9. NNPACK\n      10. Upsurge\n      11. YCMatrix\n    9. Choosing a deep learning framework\n    10. Summary\n  17. Optimizing Neural Networks for Mobile Devices\n    1. Delivering perfect user experience\n    2. Calculating the size of a convolutional neural network\n    3. Lossless compression\n    4. Compact CNN architectures\n      1. SqueezeNet\n      2. MobileNets\n      3. ShuffleNet\n      4. CondenseNet\n    5. Preventing a neural network from growing big\n    6. Lossy compression\n      1. Optimizing for inference\n        1. Network pruning\n        2. Weights quantization\n        3. Reducing precision\n        4. Other approaches\n          1. Facebook's approach in Caffe2\n      2. Knowledge distillation\n      3. Tools\n    7. An example of the network compression\n    8. Summary\n    9. Bibliography\n  18. Best Practices\n    1. Mobile machine learning project life cycle\n      1. Preparatory stage\n        1. Formulate the problem\n        2. Define the constraints\n        3. Research the existing approaches\n        4. Research the data\n        5. Make design choices\n      2. Prototype creation\n        1. Data preprocessing\n        2. Model training, evaluation, and selection\n        3. Field testing\n      3. Porting or deployment for a mobile platform\n      4. Production\n    2. Best practices\n      1. Benchmarking\n      2. Privacy and differential privacy\n      3. Debugging and visualization\n      4. Documentation\n    3. Machine learning gremlins\n      1. Data kobolds\n        1. Tough data\n        2. Biased data\n        3. Batch effects\n      2. Goblins of training\n      3. Product design ogres\n        1. Magical thinking\n        2. Cargo cult\n        3. Feedback loops\n        4. Uncanny valley effect\n    4. Recommended learning resources\n      1. Mathematical background\n        1. Machine learning\n        2. Computer vision\n        3. NLP\n    5. Summary\n\n# Preface\n\nMachine learning, as a field, promises to bring increasing intelligence to software by helping us learn and analyze information efficiently and discover certain things that humans cannot. We'll start by developing lasting intuition about the fundamental machine learning concepts in the first section. We'll explore various supervised and unsupervised learning techniques in the second section. Then, the third section, will walk you through deep learning techniques with the help of common real-world cases.  \nIn the last section, we'll dive into hardcore topics such as model compression and GPU acceleration, and provide some recommendations to avoid common mistakes during machine learning application development. By the end of the book, you'll be able to develop intelligent applications written in Swift that can learn for themselves.\n\n# Who this book is for\n\nThis book is for iOS developers who wish to create intelligent iOS applications, and data science professionals who are interested in performing machine learning using Swift. Familiarity with some basic Swift programming is all you need to get started with this book.\n\n# What this book covers\n\nChapter 1, _Getting Started with Machine Learning_ , teaches the main concepts of machine learning.\n\nChapter 2, _Classification \u2013 Decision Tree Learning_ , builds our first machine learning application.\n\nChapter 3, _K-Nearest Neighbors Classifier_ , continues exploring classification algorithms, and we learn about instance-based learning algorithms.\n\nChapter 4, _K-Means Clustering_ , continues with instance-based algorithms, this time focusing on an unsupervised clustering task.\n\nChapter 5, _Association Rule Learning_ , explores unsupervised learning more deeply.\n\nChapter 6, _Linear Regression and Gradient Descent_ , returns to supervised learning, but this time we switch our attention from non-parametric models, such as KNN and k-means, to parametric linear models.\n\nChapter 7, _Linear Classifier and Logistic Regression_ , continues by building different, more complex models on top of linear regression: polynomial regression, regularized regression, and logistic regression.\n\nChapter 8, _Neural Networks_ , implements our first neural network.\n\nChapter 9, _Convolutional Neural Networks_ , continues NNs, but this time we focus on convolutional NNs, which are especially popular in the computer vision domain.\n\nChapter 10, _Natural Language Processing_ , explores the amazing world of human natural language. We're also going to use neural networks to build several chatbots with different personalities.\n\nChapter 11, _Machine Learning Libraries_ , overviews existing iOS-compatible libraries for machine learning.\n\nChapter 12, _Optimizing Neural Networks for Mobile Devices_ , talks about deep neural network deployment on mobile platforms.\n\nChapter 13, _Best Practices_ , discusses a machine learning app's life cycle, common problems in AI projects, and how to solve them. \n\n# To get the most out of this book\n\nYou will need the following software to be able to smoothly sail through this book:\n\n  * Homebrew 1.3.8 +\n  * Python 2.7.x\n  * pip 9.0.1+\n  * Virtualenv 15.1.0+\n  * IPython 5.4.1+\n  * Jupyter 1.0.0+\n  * SciPy 0.19.1+\n  * NumPy 1.13.3+\n  * Pandas 0.20.2+\n  * Matplotlib 2.0.2+\n  * Graphviz 0.8.2+\n  * pydotplus 2.0.2+\n  * scikit-learn 0.18.1+\n  * coremltools 0.6.3+\n  * Ruby (default macOS version)\n  * Xcode 9.2+\n  * Keras 2.0.6+ with TensorFlow 1.1.0+ backend\n  * keras-vis 0.4.1+\n  * NumPy 1.13.3+\n  * NLTK 3.2.4+\n  * Gensim 2.1.0+\n\nOS required:\n\n  * macOS High Sierra 10.13.3+\n  * iOS 11+ or simulator\n\n# Download the example code files\n\nYou can download the example code files for this book from your account at www.packtpub.com. If you purchased this book elsewhere, you can visit www.packtpub.com\/support and register to have the files emailed directly to you.\n\nYou can download the code files by following these steps:\n\n  1. Log in or register at www.packtpub.com.\n  2. Select the SUPPORT tab.\n  3. Click on Code Downloads & Errata.\n  4. Enter the name of the book in the Search box and follow the onscreen instructions.\n\nOnce the file is downloaded, please make sure that you unzip or extract the folder using the latest version of:\n\n  * WinRAR\/7-Zip for Windows\n  * Zipeg\/iZip\/UnRarX for Mac\n  * 7-Zip\/PeaZip for Linux\n\nThe code bundle for the book is also hosted on GitHub at <https:\/\/github.com\/PacktPublishing\/Machine-Learning-with-Swift>. In case there's an update to the code, it will be updated on the existing GitHub repository. The author has also hosted the code bundle on his GitHub repository at: <https:\/\/github.com\/alexsosn\/SwiftMLBook>.\n\nWe also have other code bundles from our rich catalog of books and videos available at **<https:\/\/github.com\/PacktPublishing\/>**. Check them out!\n\n# Download the color images\n\nWe also provide a PDF file that has color images of the screenshots\/diagrams used in this book. You can download it here: <https:\/\/www.packtpub.com\/sites\/default\/files\/downloads\/MachineLearningwithSwift_ColorImages.pdf>.\n\n# Conventions used\n\nThere are a number of text conventions used throughout this book.\n\n`CodeInText`: Indicates code words in text, database table names, folder names, filenames, file extensions, pathnames, dummy URLs, user input, and Twitter handles. Here is an example: \"The library we are using for datasets loading and manipulation is `pandas`.\"\n\nA block of code is set as follows:\n\n    let bundle = Bundle.main \n    let assetPath = bundle.url(forResource: \"DecisionTree\", withExtension:\"mlmodelc\")\n\nWhen we wish to draw your attention to a particular part of a code block, the relevant lines or items are set in bold:\n\n    let metricsSKLRandomForest = evaluateAccuracy(yVecTest: **groundTruth** , predictions: predictionsSKLRandomForest) \n    print(metricsSKLRandomForest)\n\nAny command-line input or output is written as follows:\n\n    **> pip install -U numpy scipy matplotlib ipython jupyter scikit-learn pydotplus coremltools**\n\n**Bold** : Indicates a new term, an important word, or words that you see onscreen. For example, words in menus or dialog boxes appear in the text like this. Here is an example: \"In the interface, the user selects the type of motion he wants to record, and presses the Record button.\"\n\nWarnings or important notes appear like this.\n\nTips and tricks appear like this.\n\n# Get in touch\n\nFeedback from our readers is always welcome.\n\n**General feedback** : Email `feedback@packtpub.com` and mention the book title in the subject of your message. If you have questions about any aspect of this book, please email us at `questions@packtpub.com`.\n\n**Errata** : Although we have taken every care to ensure the accuracy of our content, mistakes do happen. If you have found a mistake in this book, we would be grateful if you would report this to us. Please visit www.packtpub.com\/submit-errata, selecting your book, clicking on the Errata Submission Form link, and entering the details.\n\n**Piracy** : If you come across any illegal copies of our works in any form on the Internet, we would be grateful if you would provide us with the location address or website name. Please contact us at `copyright@packtpub.com` with a link to the material.\n\n**If you are interested in becoming an author** : If there is a topic that you have expertise in and you are interested in either writing or contributing to a book, please visit authors.packtpub.com.\n\n# Reviews\n\nPlease leave a review. Once you have read and used this book, why not leave a review on the site that you purchased it from? Potential readers can then see and use your unbiased opinion to make purchase decisions, we at Packt can understand what you think about our products, and our authors can see your feedback on their book. Thank you!\n\nFor more information about Packt, please visit packtpub.com.\n\n# Getting Started with Machine Learning\n\nWe live in exciting times. **Artificial intelligence** ( **AI** ) and **Machine Learning** ( **ML** ) went from obscure mathematical and science fiction topics to become a part of mass culture. Google, Facebook, Microsoft, and others competed to become the first to give the world general AI. In November 2015, Google open sourced its ML framework with TensorFlow, which is suitable for running on supercomputers as well as smartphones, and since then has won a broad community. Shortly afterwards, other big companies followed the example. The best iOS app of 2016 (Apple Choice), viral photo editor Prisma owes its success entirely to a particular kind of ML algorithm: **convolutional neural network** ( **CNN** ). These systems were invented back in the nineties but became popular only in the noughties. Mobile devices only gained enough computational power to run them in 2014\/2015. In fact, artificial neural networks became so important for practical applications that in iOS 10 Apple added native support for them in the metal and accelerate frameworks. Apple also opened Siri to third-party developers and introduced GameplayKit, a framework to add AI capabilities to your computer games. In iOS 11, Apple introduced Core ML, a framework for running pre-trained models on vendors' devices, and Vision framework for common computer vision tasks.\n\nThe best time to start learning about ML was 10 years ago. The next best time is right now.\n\nIn this chapter, we will cover the following topics:\n\n  * Understanding what AI and ML is\n  * Fundamental concepts of ML : model, dataset, and learning\n  * Types of ML tasks\n  * ML project life cycle\n  * General purpose ML versus mobile ML\n\n# What is AI?\n\n\"What I cannot create, I do not understand.\"\n\n\u2013 _Richard Feynman_\n\nAI is a field of knowledge about building intelligent machines, whatever meaning you assign to the word _intelligence_. There are two different AI notions among researchers: strong AI and weak AI.\n\nStrong AI, or **artificial general intelligence** ( **AGI** ), is a machine that is fully capable of imitating human-level intelligence, including consciousness, feelings, and mind. Presumably, it should be able to apply successfully its intelligence to any tasks. This type of AI is like a horizon\u2014we always see it as a goal but we are still not there, despite all our struggles. The significant role here plays the **AI effect** : the things that were yesterday considered a feature of strong AI are today accepted as granted and trivial. In the sixties, people believed that playing board games like chess was a characteristic of strong AI. Today, we have programs that outperform the best human chess players, but we are still far from strong AI. Our iPhones are probably an AI from the eighties perspective: you can talk to them, and they can answer your questions and deliver information on any topic in just seconds. So, keeping strong AI as a distant goal, researchers focused on things at hand and called them **weak AI** : systems that have some features of intelligence, and can be applied to some narrow tasks. Among those tasks are automated reasoning, planning, creativity, communication with humans, a perception of its surrounding world, robotics, and emotions simulation. We will touch some of these tasks in this book, but mostly we will focus on ML because this domain of AI has found a lot of practical applications on mobile platforms in the recent years.\n\n# The motivation behind ML\n\nLet's start with an analogy. There are two ways of learning an unfamiliar language:\n\n  * Learning the language rules by heart, using textbooks, dictionaries, and so on. That's how college students usually do it.\n  * Observing live language: by communicating with native speakers, reading books, and watching movies. That's how children do it.\n\nIn both cases, you build in your mind the language model, or, as some prefer to say, develop a sense of language.\n\nIn the first case, you are trying to build a logical system based on rules. In this case, you will encounter many problems: the exceptions to the rule, different dialects, borrowing from other languages, idioms, and lots more. Someone else, not you, derived and described for you the rules and structure of the language.\n\nIn the second case, you derive the same rules from the available data. You may not even be aware of the existence of these rules, but gradually adjust yourself to the hidden structure and understand the laws. You use your special brain cells called **mirror neurons** , trying to mimic native speakers. This ability is honed by millions of years of evolution. After some time, when facing the wrong word usage, you just feel that something is wrong but you can't tell immediately what exactly.\n\nIn any case, the next step is to apply the resulting language model in the real world. Results may differ. In the first case, you will experience difficulty every time you find the missing hyphen or comma, but may be able to get a job as a proofreader at a publishing house. In the second case, everything will depend on the quality, diversity, and amount of the data on which you were trained. Just imagine a person in the center of New York who studied English through Shakespeare. Would he be able to have a normal conversation with people around him?\n\nNow we'll put the computer in place of the person in our example. Two approaches, in this case, represent the two programming techniques. The first one corresponds to writing ad hoc algorithms consisting of conditions, cycles, and so on, by which a programmer expresses rules and structures. The second one represents ML , in which case the computer itself identifies the underlying structure and rules based on the available data.\n\nThe analogy is deeper than it seems at first glance. For many tasks, building the algorithms directly is impossibly hard because of the variability in the real world. It may require the work of experts in the domain, who must describe all rules and edge cases explicitly. Resulting models can be fragile and rigid. On the other hand, this same task can be solved by allowing computers to figure out the rules on their own from a reasonable amount of data. An example of such a task is face recognition. It's virtually impossible to formalize face recognition in terms of conventional imperative algorithms and data structures. Only recently, the task was successfully solved with the help of ML .\n\n# What is ML ?\n\nML is a subdomain of AI that has demonstrated significant progress over the last decade, and remains a hot research topic. It is a branch of knowledge concerned with building algorithms that can learn from data and improve themselves with regards to the tasks they perform. ML allows computers to deduce the algorithm for some task or to extract hidden patterns from data. ML is known by several different names in different research communities: predictive analytics, data mining, statistical learning, pattern recognition, and so on. One can argue that these terms have some subtle differences, but essentially, they all overlap to the extent that you can use the terminology interchangeably.\n\nAbbreviation ML may refer to many things outside of the AI domain; for example, there is a functional programming language of this name. Nevertheless, the abbreviation is widely used in the names of libraries and conferences as referring to ML . Throughout this book, we also use it in this way.\n\nML is already everywhere around us. Search engines, targeted ads, face and voice recognition, recommender systems, spam filtration, self-driven cars, fraud detection in bank systems, credit scoring, automated video captioning, and machine translation\u2014all these things are impossible to imagine without ML these days.\n\nOver recent years, ML has owed its success to several factors:\n\n  * The abundance of data in different forms (big data)\n  * Accessible computational power and specialized hardware (clouds and GPUs)\n  * The rise of open source and open access\n  * Algorithmic advances\n\nAny ML system includes three essential components: data, model, and task. The data is something you provide as an input to your model. A model is a type of mathematical function or computer program that performs the task. For instance, your emails are data, the spam filter is a model, and telling spam apart from non-spam is a task. The _learning_ in ML stands for a process of adjusting your model to the data so that the model becomes better at its task. The obvious consequences of this setup is expressed in the piece of wisdom well-known among statisticians, \"\"Your model is only as good as your data\"\".\n\n# Applications of ML\n\nThere are many domains where ML is an indispensable ingredient, some of them are robotics, bioinformatics, and recommender systems. While nothing prevents you from writing bioinformatic software in Swift for macOS or Linux, we will restrict our practical examples in this book to more mobile-friendly domains. The apparent reason for this is that currently, iOS remains the primary target platform for most of the programmers who use Swift on a day-to-day basis.\n\nFor the sake of convenience, we'll roughly divide all ML applications of interest for mobile developers into three plus one areas, according to the datatypes they deal with most commonly:\n\n  * Digital signal processing (sensor data, audio)\n  * Computer vision (images, video)\n  * Natural language processing (texts, speech)\n  * Other applications and datatypes\n\n# Digital signal processing (DSP)\n\nThis category includes tasks where input data types are signals, time series, and audio. The sources of the data are sensors, HealthKit, microphone, wearable devices (for example, Apple Watch, or brain-computer interfaces), and IoT devices. Examples of ML problems here include:\n\n  * Motion sensor data classification for activity recognition\n  * Speech recognition and synthesis\n  * Music recognition and synthesis\n  * Biological signals (ECG, EEG, and hand tremor) analysis\n\nWe will build a motion recognition app in Chapter 3, _K-Nearest Neighbors Classifier_.\n\nStrictly speaking, image processing is also a subdomain of DSP but let's not be too meticulous here.\n\n# Computer vision\n\nEverything related to images and videos falls into this category. We will develop some computer vision apps in Chapter 9, _Convolutional Neural Networks_. Examples of computer vision tasks are:\n\n  * **Optical character recognition** ( **OCR** ) and handwritten input\n  * Face detection and recognition\n  * Image and video captioning\n  * Image segmentation\n  * 3D-scene reconstruction\n  * Generative art (artistic style transfer, Deep Dream, and so on)\n\n# Natural language processing (NLP)\n\nNLP is a branch of knowledge at the intersection of linguistics, computer science, and statistics. We'll talk about most common NLP techniques in Chapter 10, _Natural Language Processing_. Applications of NLP include the following:\n\n  * Automated translation, spelling, grammar, and style correction\n  * Sentiment analysis\n  * Spam detection\/filtering\n  * Document categorization\n  * Chatbots and question answering systems\n\n# Other applications of ML\n\nYou can come up with many more applications that are hard to categorize. ML can be done on virtually any data if you have enough of it. Some peculiar data types are:\n\n  * Spatial data: GPS location (Chapter 4, _K-Means_ _Clustering_ ), coordinates of UI objects and touches\n  * Tree-like structures: hierarchy of folders and files\n  * Network-like data: occurrences of people together in your photos, or hyperlinks between web pages\n  * Application logs and user in-app activity data (Chapter 5, _Association Rule Learning_ )\n  * System data: free space disk, battery level, and similar\n  * Survey results\n\n# Using ML to build smarter iOS applications\n\nAs we know from press reports, Apple uses ML for fraud detection, and to mine useful data from beta testing reports; however, these are not examples visible on our mobile devices. Your iPhone itself has a handful of ML models built into its operating system, and some native apps helping to perform a wide range of tasks. Some use cases are well known and prominent while others are inconspicuous. The most obvious examples are Siri speech recognition, natural language understanding, and voice generation. Camera app uses face detection for focusing and Photos app uses face recognition to group photos with the same person into one album. Presenting the new iOS 10 in June 2016, Craig Federighi mentioned its predictive keyboard, which uses an LSTM algorithm (a type of recurrent neural network) to suggest the next word from the context, and also how Photos uses deep learning to recognize objects and classify scenes. iOS itself uses ML to extend battery life, provide contextual suggestions, match profiles from social networks and mail with the records in Contacts, and to choose between internet connection options. On Apple Watch, ML models are employed to recognize user motion activity types and handwritten input.\n\nPrior to iOS 10, Apple provided some ML APIs like speech or movement recognition, but only as black boxes, without the possibility to tune the models or to reuse them for other purposes. If you wanted to do something slightly different, like detect the type of motion (which is not predefined by Apple), you had to build your own models from scratch. In iOS 10, CNN building blocks were added in the two frameworks at once: as a part of Metal API, and as a sublibrary of an Accelerate framework. Also, the first actual ML algorithm was introduced to iOS SDK: the decision tree learner in the GameplayKit.\n\nML capabilities continued to expand with the release of iOS 11. At the WWDC 2017, Apple presented the Core ML framework. It includes API for running pre-trained models and is accompanied by tools for converting models trained with some popular ML frameworks to Apple's own format. Still, for now it doesn't provide the possibility of training models on a device, so your models can't be changed or updated in runtime.\n\nLooking in the App Store for the terms _artificial intelligence_ , _deep learning_ , _ML_ , and similar, you'll find a lot of applications, some of them quite successful. Here are several examples:\n\n  * Google Translate is doing speech recognition and synthesis, OCR, handwriting recognition, and automated translation; some of this is done offline, and some online.\n  * Duolingo validates pronunciation, recommends optimal study materials, and employs Chatbots for language study.\n  * Prisma, Artisto, and others turn photos into paintings using a neural artistic style transfer algorithm. Snapchat and Fabby use image segmentation, object tracking, and other computer vision techniques to enhance selfies. There are also applications for coloring black and white photos automatically.\n  * Snapchat's video selfie filters use ML for real-time face tracking and modification.\n  * Aipoly Vision helps blind people, saying aloud what it sees through the camera.\n  * Several calorie counter apps recognize food through a camera. There are also similar apps to identify dog breeds, trees and trademarks.\n  * Tens of AI personal assistants and Chatbots, with different capabilities from cow disease diagnostics, to matchmaking and stock trading.\n  * Predictive keyboards, spellcheckers, and auto correction, for instance, SwiftKey.\n  * Games that learn from their users and games with evolving characters\/units.\n  * There are also news, mail, and other apps that adapt to users' habits and preferences using ML .\n  * Brain-computer interfaces and fitness wearables with the help of ML recognize different user conditions like concentration, sleep phases, and so on. At least some of their supplementary mobile apps do ML .\n  * Medical diagnostic and monitoring through mobile health applications. For example, OneRing monitors Parkinson's disease using the data from a wearable device.\n\nAll these applications are built upon the extensive data collection and processing. Even if the application itself is not collecting the data, the model it uses was trained on some usually big dataset. In the following section, we will discuss all things related to data in ML applications.\n\n# Getting to know your data\n\nFor many years, researchers argued about what is more important: data or algorithms. But now, it looks like the importance of data over algorithms is generally accepted among ML specialists. In most cases, we can assume that the one who has better data usually beats those with more advanced algorithms. Garbage in, garbage out\u2014this rule holds true in ML more than anywhere else. To succeed in this domain, one need not only have data, but also needs to know his data and know what to do with it.\n\nML datasets are usually composed from individual observations, called samples, cases, or data points. In the simplest case, each sample has several features.\n\n# Features\n\nWhen we are talking about features in the context of ML , what we mean is some characteristic property of the object or phenomenon we are investigating.\n\nOther names for the same concept you'll see in some publications are explanatory variable, independent variable, and predictor.\n\nFeatures are used to distinguish objects from each other and to measure the similarity between them.\n\nFor instance:\n\n  * If the objects of our interest are books, features could be a title, page count, author's name, a year of publication, genre, and so on\n  * If the objects of interest are images, features could be intensities of each pixel\n  * If the objects are blog posts, features could be language, length, or presence of some terms\n\nIt's useful to imagine your data as a spreadsheet table. In this case, each sample (data point) would be a row, and each feature would be a column. For example, Table 1.1 shows a tiny dataset of books consisting of four samples where each has eight features.\n\nTable 1.1: an example of a ML dataset (dummy books):\n\n**Title** | **Author's name** | **Pages** | **Year** | **Genre** | **Average readers review score** | **Publisher** | **In stock**\n\n---|---|---|---|---|---|---|---\n\nLearn ML in 21 Days | Machine Learner | 354 | 2018 | Sci-Fi | 3.9 | Untitled United | False\n\n101 Tips to Survive an Asteroid Impact | Enrique Drills | 124 | 2021 | Self-help | 4.7 | Vacuum Books | True\n\nSleeping on the Keyboard | Jessica's Cat | 458 | 2014 | Non-fiction | 3.5 | JhGJgh Inc. | True\n\nQuantum Screwdriver: Heritage | Yessenia Purnima | 1550 | 2018 | Sci-Fi | 4.2 | Vacuum Books | True\n\n# Types of features\n\nIn the books example, you can see several types of features:\n\n  * **Categorical or unordered** : Title, author, genre, publisher. They are similar to enumeration without raw values in Swift, but with one difference: they have levels instead of cases. Important: you can't order them or say that one is bigger than another.\n  * **Binary** : The presence or absence of something, just true or false. In our case, the _In stock_ feature.\n  * **Real numbers** : Page count, year, average reader's review score. These can be represented as float or double.\n\nThere are others, but these are by far the most common.\n\nThe most common ML algorithms require the dataset to consist of a number of samples, where each sample is represented by a vector of real numbers (feature vector), and all samples have the same number of features. The simplest (but not the best) way of translating categorical features into real numbers is by replacing them with numerical codes (Table 1.2).\n\nTable 1.2: dummy books dataset after simple preprocessing:\n\n**Title** | **Author's name** | **Pages** | **Year** | **Genre** | **Average readers review score** | **Publisher** | **In stock**\n\n---|---|---|---|---|---|---|---\n\n0.0 | 0.0 | 354.0 | 2018.0 | 0.0 | 3.9 | 0.0 | 0.0\n\n1.0 | 1.0 | 124.0 | 2021.0 | 1.0 | 4.7 | 1.0 | 1.0\n\n2.0 | 2.0 | 458.0 | 2014.0 | 2.0 | 3.5 | 2.0 | 1.0\n\n3.0 | 3.0 | 1550.0 | 2018.0 | 0.0 | 4.2 | 1.0 | 1.0\n\nThis is an example of how your dataset may look before you feed it into your ML algorithm. Later, we will discuss the nuts and bolts of data preprocessing for specific applications.\n\n# Choosing a good set of features\n\nFor ML purposes, it's necessary to choose a reasonable set of features, not too many and not too few:\n\n  * If you have too few features, this information may be not sufficient for your model to achieve the required quality. In this case, you want to construct new ones from existing features, or extract more features from the raw data.\n  * If you have too many features you want to select only the most informative and discriminative, because the more features you have the more complex your computations become.\n\nHow do you tell which features are most important? Sometimes common sense helps. For example, if you are building a model that recommends books for you, the genre and average rating of the book are perhaps more important features than the number of pages and year of publication. But what if your features are just pixels of a picture and you're building a face recognition system? For a black and white image of size 1024 x 768, we'd get 786,432 features. Which pixels are most important? In this case, you have to apply some algorithms to extract meaningful features. For example, in computer vision, edges, corners, and blobs are more informative features then raw pixels, so there are plenty of algorithms to extract them ( _Figure 1.1_ ). By passing your image through some filters, you can get rid of unimportant information and reduce the number of features significantly; from hundreds of thousands to hundreds, or even tens. The techniques that helps to select the most important subset of features is known as **feature selection** , while the **feature extraction** techniques result in the creation of new features:\n\nFigure 1.1: Edge detection is a common feature extraction technique in computer vision. You can still recognize the object on the right image, despite it containing significantly less information than the left one.\n\nFeature extraction, selection, and combining is a kind of the art which is known as **feature engineering**. This requires not only hacking and statistical skills but also domain knowledge. We will see some feature engineering techniques while working on practical applications in the following chapters. We also will step into the exciting world of **deep learning** : a technique that gives a computer the ability to extract high-level abstract features from the low-level features.\n\nThe number of features you have for each sample (or length of feature vector) is usually referred to as the **dimensionality** of the problem. Many problems are high-dimensional, with hundreds or even thousands of features. Even worse, some of those problems are sparse; that is, for each data point, most of the features are zero or missed. This is a common situation in recommender systems. For instance, imagine yourself building the dataset of movie ratings: the rows are movies and columns are users, and in each cell, you have a rating given by the user of the movie. The majority of the cells in the table will remain empty, as most of the users will never have watched most of the movies. The opposite situation is called **dense,** which is when most values are in place. Many problems in natural language processing and bioinformatics are high-dimensional, sparse, or both.\n\nFeature selection and extraction help to decrease the number of features without significant loss of information, so we also call them **dimensionality reduction algorithms**.\n\n# Getting the dataset\n\nDatasets can be obtained from different sources. The ones important for us are:\n\n  * Classical datasets such as Iris (botanical measurements of flowers composed by R. Fisher in 1936), MNIST (60,000 handwritten digits published in 1998), Titanic (personal information of Titanic passengers from Encyclopedia Titanica and other sources), and others. Many classical datasets are available as part of Python and R ML packages. They represent some classical types of ML tasks and are useful for demonstrations of algorithms. Meanwhile, there is no similar library for Swift. Implementation of such a library would be straightforward and is a low-hanging fruit for anyone who wants to get some stars on GitHub.\n  * Open and commercial dataset repositories. Many institutions release their data for everyone's needs under different licenses. You can use such data for training production models or while collecting your own dataset.\n\nSome public dataset repositories include:\n\n  *     * The UCI ML repository: https:\/\/archive.ics.uci.edu\/ml\/datasets.html\n\n    * Kaggle datasets: <https:\/\/www.kaggle.com\/datasets>\n\n    * data.world, a social network for dataset sharing: https:\/\/data.world\n\nTo find more, visit the list of repositories at KDnuggets: <http:\/\/www.kdnuggets.com\/datasets\/index.html>. Alternatively, you'll find a list of datasets at Wikipedia: <https:\/\/en.wikipedia.org\/wiki\/List_of_datasets_for_machine_learning_research>.\n\n  * **Data collection (acquisition)** is required if no existing data can help you to solve your problem. This approach can be costly both in resources and time if you have to collect the data ad hoc; however, in many cases, you have data as a byproduct of some other process, and you can compose your dataset by extracting useful information from the data. For example, text corpuses can be composed by crawling Wikipedia or news sites. iOS automatically collects some useful data. HealthKit is a unified database of users' health measurements. Core Motion allows getting historical data on user's motion activities. The ResearchKit framework provides standardized routines to assess the user's health conditions. The CareKit framework standardizes the polls. Also, in some cases, useful information can be obtained from app log mining. \n    * In many cases, to collect data is not enough, as raw data doesn't suit many ML tasks well. So, the next step after data collection is data labeling. For example, you have collected dataset of images, so now you have to attach a label to each of them: to which category does this image belong? This can be done manually (often at expense), automatically (sometimes impossible), or semi-automatically. Manual labeling can be scaled by means of crowdsourcing platforms, like Amazon Mechanical Turk.\n  * Random **data generation** can be useful for a quick check of your ideas or in combination with the TDD approach. Also, sometimes adding some controlled randomness to your real data can improve the results of learning. This approach is known as **data augmentation**. For instance, this approach was taken to build an optical character recognition feature in the Google Translate mobile app. To train their model, they needed a lot of real-world photos with letters in different languages, which they didn't have. The engineering team bypassed this problem by creating a large dataset of letters with artificial reflections, smudges, and all kinds of corruptions on them. This improved the recognition quality significantly.\n  * **Real-time data sources** , such as inertial sensors, GPS, camera, microphone, elevation sensor, proximity sensor, touch screen, force touch, and Apple Watch sensors can be used to collect a standalone dataset or to train a model on the fly.\n\nReal-time data sources are especially important for the special class of ML models called **online ML** , which allows models to embed new data. A good example of such a situation is spam filtering, where the model should dynamically adapt to the new data. It's the opposite of batch learning, when the whole training dataset should be available from the very beginning.\n\n# Data preprocessing\n\nThe useful information in the data is usually referred to as a **signal**. On the other hand, the pieces of data that represent errors of different kinds and irrelevant data are known as **noise**. Errors can occur in the data during measurements, information transmission, or due to human errors. The goal of data cleansing procedures is to increase the signal\/noise ratio. During this stage, you will usually transform all data to one format, delete entries with missed values, and check suspicious outliers (they can be both noise and signal). It is widely believed among ML engineers, that the data preprocessing stage usually consumes 90% of the time allocated for the ML project. Then, algorithm tweaking consumes another 90% of time. This statement is a joke only partially (about 10% of it). In Chapter 13, _Best Practices_ , we are going to discuss common problems with the data and how to fix them.\n\n# Choosing a model\n\nLet's say you've defined a task and you have a dataset. What's next? Now you need to choose a model and train it on the dataset to perform that task.\n\nThe model is the central concept in ML . ML is basically a science of building models of the real world using data. The term _model_ refers to the phenomenon being modeled, while _map_ refers to the real territory. Depending on the situation, it can play a role of good approximation, an outdated description (in a swiftly changing environment), or even self-fulfilled prophecy (if the model affects the modeled object). \"\"All models are wrong, but some are useful\"\" is a well-known proverb in statistics.\n\n# Types of ML algorithms\n\nML models\/algorithms are often divided into three groups depending on the type of input:\n\n  * Supervised learning\n  * Unsupervised learning\n  * Reinforcement learning\n\nThis division is rather vague because some algorithms fall into two of these groups while others do not fall into any. There are also some middle states, such as semi-supervised learning.\n\nAlgorithms in these three groups can perform different tasks, and hence can be divided into subgroups according to the output of the model. _Table 1.3_ shows the most common ML tasks and their classification.\n\n# Supervised learning\n\nSupervised learning is arguably the most common and easy-to-understand type of ML . All supervised learning algorithms have one prerequisite in common: you should have a labeled dataset to train them. Here, a dataset is a set of samples, plus an expected output (label) for each sample. These labels play the role of supervisor during the training.\n\nIn different publications, you'll see different synonyms for labels, including dependent variable, predicted variable, and explained variable.\n\nThe goal of supervised learning is to get a function that for every given input returns a desired output. In the most simplified version, a supervised learning process consists of two phases: training and inference. During the first phase, you train the model using your labeled dataset. On the second phase, you use your model to do something useful, like make predictions. For instance, given a set of labeled images (dataset), a neural network (model) can be trained to predict (inference) correct labels for previously unseen images.\n\nUsing supervised learning, you will usually solve one of two problems: classification or regression. The difference is in the type of labels: categorical in the first case and real numbers in the second.\n\nTo classify means simply to assign one of the labels from a predefined set. Binary classification is a special kind of classification, when you have only two labels (positive and negative). An example of a classification task is to assign _spam_ \/ _not-spam_ labels to letters. We will train our first classifier in the next chapter, and throughout this book we will apply different classifiers for many real-world tasks.\n\nRegression is the task of assigning a real number to a given case. For example, predicting a salary given employee characteristics. We will discuss regression in Chapter 6, _Linear Regression and Gradient Descent_ and Chapter 7, _Linear Classifier and Logistic Regression_ , in more detail.\n\nIf the task is to sort objects in some order (output a permutation, speaking combinatorial), and labels are not really real numbers but rather an order of objects, ranking learning is at hand. You see ranking algorithms in action when you open the Siri suggestions menu on iOS. Each app placed in the list there is done so according to its relevance for you.\n\nIf labels are complicated objects, like graphs or trees, neither classification nor regression will be of use. Structured prediction algorithms are the type of algorithms to tackle those problems. Parsing English sentences into syntactic trees is an example of this kind of task.\n\nRanking and structured learning are beyond the scope of this book because their use cases are not as common as classification or regression, but at least now you know what to Google search for when you need to.\n\n# Unsupervised learning\n\nIn unsupervised learning, you don't have the labels for the cases in your dataset. Types of tasks to solve with unsupervised learning are: clustering, anomaly detection, dimensionality reduction, and association rule learning.\n\nSometimes you don't have the labels for your data points but you still want to group them in some meaningful way. You may or may not know the exact number of groups. This is the setting where clustering algorithms are used. The most obvious example is clustering users into some groups, like students, parents, gamers, and so on. The important detail here is that a group's meaning is not predefined from the very beginning; you name it only after you've finished grouping your samples. Clustering also can be useful to extract additional features from the data as a preliminary step for supervised learning. We will discuss clustering in Chapter 4, _K-Means Clustering_.\n\n**Outlier** \/ **anomaly detection** algorithms are used when the goal is to find some anomalous patterns in the data, weird data points. This can be especially useful for automated fraud or intrusion detection. Outlier analysis is also an important detail of data cleansing.\n\n**Dimensionality reduction** is a way to distill data to the most informative and, at the same time, compact representation of it. The goal is to reduce a number of features without losing important information. It can be used as a preprocessing step before supervised learning or data visualization.\n\n**Association rule learning** looks for repeated patterns of user behavior and peculiar co-occurrences of items. An example from retail practice: if a customer buys milk, isn't it more probable that he will also buy cereal? If yes, then perhaps it's better to move shelves, with the cereals closer to the shelf with the milk. Having rules like this, owners of businesses can make informed decisions and adapt their services to customers' needs. In the context of software development, this can empower anticipatory design\u2014when the app seemingly knows what you want to do next and provides suggestions accordingly. In Chapter 5, _Association Rule Learning_ we will implement a priori one of the most well-known rule learning algorithms:\n\nFigure 1.2: Datasets for three types of learning: supervised, unsupervised, and semi-supervised\n\nLabeling data manually is usually a costly thing, especially if special qualification is required. Semi-supervised learning can help when only some of your samples are labeled and others are not (see the following diagram). It is a hybrid of supervised and unsupervised learning. At first, it looks for unlabeled instances, similar to the labeled ones in an unsupervised manner, and includes them in the training dataset. After this, the algorithm can be trained on this expanded dataset in a typical supervised manner.\n\n# Reinforcement learning\n\nReinforcement learning is special in the sense that it doesn't require a dataset (see the following diagram). Instead, it involves an agent who takes actions, changing the state of the environment. After each step, it gets a reward or punishment, depending on the state and previous actions. The goal is to obtain a maximum cumulative reward. It can be used to teach the computer to play video games or drive a car. If you think about it, reinforcement learning is the way our pets train us humans: by rewarding our actions with tail-wagging, or punishing with scratched furniture.\n\nOne of the central topics in reinforcement learning is the exploration-exploitation dilemma\u2014how to find a good balance between exploring new options and using what is already known:\n\nFigure 1.3: Reinforcement learning process\n\nTable 1.3: ML tasks:\n\n**Task** | **Output type** | **Problem example** | **Algorithms**  \n---|---|---|---  \n**Supervised learning**  \nRegression | Real numbers | Predict house prices, given its characteristics | Linear regression and polynomial regression  \nClassification | Categorical | Spam\/not-spam classification | KNN, Na\u00efve Bayes, logistic regression, decision trees, random forest, and SVM  \nRanking | Natural number (ordinal variable) | Sort search results per relevance | Ordinal regression  \nStructured prediction | Structures: trees, graphs, and so on | Part-of-speech tagging | Recurrent neural networks, and conditional random field  \n**Unsupervised learning**  \nClustering | Groups of objects | Build a tree of living organisms | Hierarchical clustering, _k_ -means, and GMM  \nDimensionality reduction | Compact representation of given features | Find most important components in brain activity | PCA, t-SNE, and LDA  \nOutlier\/anomaly detection | Objects that are out of pattern | Fraud detection | Local outlier factor  \nAssociation rule learning | Set of rules | Smart house intrusion detection | A priori  \n**Reinforcement learning**  \nControl learning | Policy with maximum expected return | Learn to play a video game | Q-learning\n\n# Mathematical optimization \u2013 how learning works\n\nThe magic behind the learning process is delivered by the branch of mathematics called **mathematical optimization**. Sometimes it's also somewhat misleading being referred to as mathematical programming; the term coined long before widespread computer programming and is not directly related to it. Optimization is the science of choosing the best option among available alternatives; for example, choosing the best ML model.\n\nMathematically speaking, ML models are functions. You as an engineer chose the function family depending on your preferences: linear models, trees, neural networks, support vector machines, and so on. Learning is a process of picking from the family the function which serves your goals the best. This notion of the best model is often defined by another function, the **loss function**. It estimates a goodness of the model according to some criteria; for instance, how good the model fits the data, how complex it is, and so on. You can think of the loss function as a judge at a competition whose role is to assess the models. The objective of the learning is to find such a model that delivers a minimum to the loss function (minimize the loss), so the whole learning process is formalized in mathematical terms as a task of function minimization.\n\nFunction minimum can be found in two ways: analytically (calculus) or numerically (iterative methods). In ML , we often go for the numerical optimization because the loss functions get too complex for analytical solutions.\n\nA nice interactive tutorial on numerical optimization can be found here: <http:\/\/www.benfrederickson.com\/numerical-optimization\/>.\n\nFrom the programmer's point of view, learning is an iterative process of adjusting model parameters until the optimal solution is found. In practice, after a number of iterations, the algorithm stops improving because it is stuck in a local optimum or has reached the global optimum (see the following diagram). If the algorithm always finds the local or global optimum, we say that it _converges_. On the other hand, if you see your algorithm oscillating more and more and never approaching a useful result, it diverges:\n\nFigure 1.4: Learner represented as a ball on a complex surface: it's possible for him to fall in a local minimum and never reach the global one\n\n# Mobile versus server-side ML\n\nMost Swift developers are writing their applications for iOS. Those among us who develop their Swift applications for macOS or server-side are in a lucky position regarding ML . They can use whatever libraries and tools they want, reckoning on powerful hardware and compatibility with interpretable languages. Most of the ML libraries and frameworks are developed with server-side (or at least powerful desktops) in mind. In this book, we talk mostly about iOS applications, and therefore most practical examples consider limitations of handheld devices.\n\nBut if mobile devices have limited capabilities, we can do all ML on the server-side, can't we? Why would anyone bother to do ML locally on mobile devices at all? There are at least three issues with client-server architecture:\n\n  * The client app will be fully functional only when it has an internet connection. This may not be a big problem in developed countries but this can limit your target audience significantly. Just imagine your translator app being non-functional during travel abroad.\n  * Additional time delay introduced by sending data to the server and getting a response. Who enjoys watching progress bars or, even worse, infinite spinners while your data is being uploaded, processed, and downloaded back again? What if you need those results immediately and without consuming your internet traffic? Client-server architecture makes it almost impossible for such applications of ML as real-time video and audio processing.\n  * Privacy concerns: any data you've uploaded to the internet is not yours anymore. In the age of total surveillance, how do you know that those funny selfies you've uploaded today to the cloud will not be used tomorrow to train face recognition, or for target-tracking algorithms for some interesting purposes, like killer drones? Many users don't like their personal information to be uploaded to some servers and possibly shared\/sold\/leaked to some third parties. Apple also argues for reducing data collection as much as possible.\n\nSome of the applications can be OK (can't be great, though) with those limitations, but most developers want their apps to be responsive, secure, and useful all the time. This is something only on-device ML can deliver.\n\nFor me, the most important argument is that we _can_ do ML without server-side. Hardware capabilities are increasing with each year and ML on mobile devices is a hot research field. Modern mobile devices are already powerful enough for many ML algorithms. Smartphones are the most personal and arguably the most important devices nowadays just because they are everywhere. Coding ML is fun and cool, so why should server-side developers have all the fun?\n\nAdditional bonuses that you get when implement ML on the mobile side are the free computation power (you are not paying for the electricity) and the unique marketing points (our app puts the power of AI inside of your pocket).\n\n# Understanding mobile platform limitations\n\nNow, if I have persuaded you to use ML on mobile devices, you should be aware of some limitations:\n\n  * Computation complexity restriction. The more you load your CPU, the faster your battery will die. It's easy to transform your iPhone into a compact heater with the help of some ML algorithms.\n  * Some models take a long time to train. On the server, you can let your neural networks train for weeks; but on a mobile device, even minutes are too long. iOS applications can run and process some data in background mode if they have some good reasons, like playing music. Unfortunately, ML is not on the list of good reasons, so most probably, you will not be able to run it in background mode.\n  * Some models take a long time to run. You should think in terms of frames per second and good user experience.\n  * Memory restrictions. Some models grow during the training process, while others remain a fixed size.\n  * Model size restrictions. Some trained models can take hundreds of megabytes or even gigabytes. But who wants to download your application from the App Store if it is so huge?\n  * Locally stored data is mostly restricted to different types of users' personal data, meaning that you will not be able to aggregate the data of different users and perform large-scale ML on mobile devices.\n  * Many open source ML libraries are built on top of interpretable languages, like Python, R, and MATLAB, or on top of the JVM, which makes them incompatible with iOS.\n\nThose are only the most obvious challenges. You'll see more as we start to develop real ML apps. But don't worry, there is a way to eat this elephant piece by piece. Efforts spent on it are paid off by a great user experience and users' love. Platform restrictions are not unique to mobile devices. Developers of autonomous devices (like drones), IoT developers, wearable device developers, and many others face the same problems and deal with them successfully.\n\nMany of these problems can be addressed by training the models on powerful hardware, and then deploying them to mobile devices. You can also choose a compromise with two models: a smaller one on a device for offline work, and a large one on the server. For offline work you can choose models with fast inference, then compress and optimize them for parallel execution; for instance, on GPU. We'll talk more about this in Chapter 12, _Optimizing Neural Networks for Mobile Devices_.\n\n# Summary\n\nIn this chapter, we learned about the main concepts in ML .\n\nWe discussed different definitions and subdomains of artificial intelligence, including ML . ML is the science and practice of extracting knowledge from data. We also explained the motivation behind ML . We had a brief overview of its application domains: digital signal processing, computer vision, and natural language processing.\n\nWe learned about the two core concepts in ML : the data, and the model. Your model is only as good as your data. A typical ML dataset consists of samples; each sample consists of features. There are many types of features and many techniques to extract useful information from the features. These techniques are known as feature engineering. For supervised learning tasks, dataset also includes label for each of the samples. We provided an overview of data collection and preprocessing.\n\nFinally, we learned about three types of common ML tasks: supervised, unsupervised, and reinforcement learning. In the next chapter, we're going to build our first ML application.\n\n# Bibliography\n\n  1. Good O. (July 29, 2015), _How Google Translate squeezes deep learning onto a phone_ , retrieved from Google Research Blog: <https:\/\/research.googleblog.com\/2015\/07\/how-google-translate-squeezes-deep.html>\n\n# Classification \u2013 Decision Tree Learning\n\nIn the previous chapter, we discussed different types of machine learning, including supervised classification tasks; in this chapter, we will build our first Swift application for this. We will discuss main components of machine learning development stack, and will also exercise in data generation, exploratory analysis, preprocessing, and models training and evaluation in Python. After this, we will transfer our model to Swift. We will also discuss a specific class of supervised learning algorithms\u2014decision tree learning and its extension: random forest.\n\nThe following topics are waiting for us in this chapter:\n\n  * Machine learning software development stack\n  * Python toolbox for machine learning: IPython, SciPy, scikit-learn\n  * Dataset generation and exploratory analysis\n  * Data preprocessing\n  * Decision tree learning and random forest\n  * Assessing the model performance using different performance metrics\n  * Underfitting and overfitting\n  * Exporting scikit-learn models to Core ML format\n  * Deploying trained models to iOS\n\n# Machine learning toolbox\n\nFor many years, the programming language of choice for machine learning was one of the following: Python, R, MATLAB, C++. This is not due to some specific language features, but because of the infrastructure around it: libraries and tools. Swift is a relatively young programming language, and anyone who chooses it as a primary tool for machine learning development should start from the very basic building blocks, and build his own tools and libraries. Recently, Apple became more open to third-party Python machine learning tools: Core ML can work with some of them.\n\nHere is a list of components that are needed for the successful machine learning research and development, and examples of popular libraries and tools of the type:\n\n  * **Linear algebra** : Machine learning developer needs data structures like vectors, matrices, and tensors with compact syntax and hardware-accelerated operations on them. Examples in other languages: NumPy, MATLAB, and R standard libraries, Torch.\n  * **Probability theory** : All kinds of random data generation: random numbers and collections of them; probability distributions; permutations; shuffling of collections, weighted sampling, and so on. Examples: NumPy, and R standard library.\n  * **Data input-output** : In machine learning, we are usually most interested in the parsing and saving data in the following formats: plain text, tabular files like CSV, databases like SQL, internet formats JSON, XML, HTML, and web scraping. There are also a lot of domain-specific formats.\n  * **Data wrangling** : Table-like data structures, data engineering tools: dataset cleaning, querying, splitting, merging, shuffling, and so on. Pandas, dplyr.\n  * **Data analysis\/statistic** : Descriptive statistic, hypotheses testing and all kinds of statistical stuff. R standard library, and a lot of CRAN packages.\n  * **Visualization** : Statistical data visualization (not pie charts): graph visualization, histograms, mosaic plots, heat maps, dendrograms, 3D-surfaces, spatial and multidimensional data visualization, interactive visualization, Matplotlib, Seaborn, Bokeh, ggplot2, ggmap, Graphviz, D3.js.\n  * **Symbolic computations** : Automatic differentiation: SymPy, Theano, Autograd.\n  * **Machine learning packages** : Machine learning algorithms and solvers. Scikit-learn, Keras, XGBoost, E1071, and caret.\n  * **Interactive prototyping environment** : Jupyter, R studio, MATLAB, and iTorch.\n\nThis is not referring to domain-specific tools, like NLP, or computer vision libraries.\n\nAs for summer 2017, I'm not aware of Swift alternatives of comparable quality and functionality to any of the mentioned tools. Also, none of these popular libraries are directly compatible with Swift, meaning you can't call Keras from your iOS Swift code. All this means that Swift cannot be the primary tool for machine learning research and development. Killing Python is not on Swift's agenda so far; however, to a different degree, there are some compatible libraries and tools, which using a wide scope of machine learning problems can be addressed in your Swift applications. In the following chapters, we're building our own tools, or introducing third-party tools as we need them. We are talking about machine learning libraries specifically in Chapter 10, _Natural Language Processing_. Still, for anyone who wants to work with machine learning, it's more than advisable to know well at least one from this list: Python, R, and MATLAB.\n\n# Prototyping the first machine learning app\n\nUsually, before implementing a machine learning application for mobile devices, you want to do a quick and dirty prototype just to check your ideas. This allows to save a lot of time when you realize that the model you initially thought works perfectly for your problem, in reality doesn't. The quickest way to do a prototype is to use Python or R tools listed in the previous section.\n\nPython is a general-purpose programming language with rich infrastructure and vibrant community. Its syntax is similar in many ways to Swift's one. Throughout this book, we'll use it for prototyping, and Swift for actual development.\n\nWhen you have tested your ideas and a model prototype works as you expect, you can start thinking about how to port it to an iOS. You have several options here:\n\nInference-only options:\n\n  * Check the Core ML, and a list of the Python libraries it supports. Maybe, you will be able to export your model in Core ML format, and run it on a device.\n  * Write the custom converter for your model if it is not supported by the Core ML.\n\nTraining and inference options:\n\n  * Write the algorithms from scratch. In this book, we are implementing a bunch of machine learning algorithms, so you'll see that it's not that hard. Still, this is the most time-consuming option, and the model's results may differ significantly.\n  * Check available iOS-compatible libraries (see chapter on Chapter 11, _Machine Learning Libraries_ ).\n\n# Tools\n\nHere is a list of tools that we're using in the following tutorial:\n\n  * **Homebrew** : This is a package manager for macOS. Official site: https:\/\/brew.sh\/.\n  * **Python** : This is a general-purpose programming language popular for machine learning and data science. Official site: https:\/\/www.python.org\/.\n  * pip: This is a Python package manager. Unlike CocoaPods, it installs libraries globally, and not in a per-project manner.\n  * **Virtualenv** : This is a tool for creating separate Python environments with different Python versions and library sets.\n  * **IPython** : This is an interactive Python REPL for scientific computations.\n  * **Jupyter** : This is a web-GUI for IPython. Official site: http:\/\/jupyter.org\/.\n  * **Graphviz** : This is an open source tool for graphs visualization. We're using it in this chapter to draw models' inner structures. Official site: http:\/\/www.graphviz.org\/.\n\nAnd, the Python packages are as follows:\n\n  * `scipy`: This is a Python-based ecosystem of open source software for mathematics, science, and engineering. Official site: https:\/\/www.scipy.org\/.\n  * `numpy`: This is a numerical library.\n  * `matplotlib`: This is a popular plotting library.\n  * `pydotplus`: This is a library for tree visualization, a counterpart of Graphviz.\n  * `scikit-learn`: This is a popular machine learning library. Official site: http:\/\/scikit-learn.org\/.\n  * `coremltools`: is an Apple package for saving scikit-learn models into Core ML format. Official site: <https:\/\/pypi.python.org\/pypi\/coremltools>.\n\n# Setting up a machine learning environment\n\nThere is a significant segmentation in the Python community due to an issue of back-compatibility between Python 2 and Python 3\u2014many active projects still use Python 2.7 (released in 2010), while many new tools are not backward-compatible with it, because they are based on the Python 3.x. Some tools have both versions. macOS is shipped with legacy Python 2.7.10 (released in 2015) pre-installed, while an up-to-date version at the moment of writing this book is Python 3.6.1. We will use the system's default Python throughout this book, if the opposite is not mentioned explicitly. The primary reason for this is that Core ML tools are compatible only with Python 2.7.x.\n\nThe following steps assume that you don't have other Python versions installed (like Anaconda, or through a Homebrew), except the system's default one. If you have other Python distributions installed, you likely know how to install required packages and create virtual environments.\n\nFirst, in the Terminal, go to the user's root:\n\n    **> cd ~**\n\nOn a Mac system, the user you use to log in by default has limited privileges by design for enhanced security measures. Using the `sudo` command allows you to perform tasks with additional privileges on a case-by-case basis. This process aids in simplifying the security features by avoiding accidentals actions.\n\npip is a Python package manager. Unlike the up-to-date version of Python, the system's one doesn't have it by default. Instead, it should have the old legacy package manager `easy_install`. Don't use it for anything except for pip installation; it will likely mess up your system. It requires sudo privileges to install things:\n\n    **> sudo easy_install pip**\n\nIf you have some version of pip preinstalled, you can upgrade it to the latest one with the following command:\n\n    **> pip install --upgrade pip**\n\nMany third-party programs are using the system's Python version, so to not interfere with them, it's safer to create the separate Python environment and install all dependencies that we need into it. The Virtualenv is a tool for isolated Python environment creation. It is also missing from the macOS Python, while present by default in all recent distributions starting with Python 3.3 and later. After successful installation of pip, we can use it to install `virtualenv`:\n\n    **> pip install -U virtualenv**\n\nThe `-U` option tells pip to install the package for the current user only.\n\nNever run `pip` with the `sudo`. Whenever you need the `sudo` to run it, you know that you're doing something wrong.\n\nTo create a virtual environment for the book, run:\n\n    **> cd ~**\n    **> virtualenv swift-ml-book**\n\nThis will create the `swift-ml-book` folder, and a separate copy of Python, pip, and other tools in it. To switch to this environment (activate the environment), run the following command:\n\n    **> source swift-ml-book\/bin\/activate**\n\nNow `swift-ml-book` prepends all your commands in the Terminal, so you know on which environment you are now. When you want to deactivate the Python 3 environment, run:\n\n    **> deactivate**\n\nFinally, we can install libraries; be sure that you've activated the environment:\n\n    **> pip install -U numpy scipy matplotlib ipython jupyter scikit-learn pydotplus coremltools**\n\nYou should see a long output to the command line; downloading and installing all dependencies may take a while. Eventually, you should see the message `Successfully installed ...`, and a list of installed packages. It will be much longer than the one that we've provided pip with, because it includes a bunch of transient dependencies.\n\nMost importantly, now you should have two new commands in your Terminal: `ipython`, and `jupyter notebook`. The first one runs interactive IPython REPL, and the second one runs a web-based GUI for IPython, where you can create notebooks\u2014interactive documents, similar to Swift playgrounds.\n\nAdditionally, we should install Graphviz\u2014an open source tool for graphs visualization. It can be downloaded from the official site, or installed using Homebrew:\n\n    **> brew install graphviz**\n\nIf you don't have Homebrew, install it. Installation instructions should look like the following, but you'd better check the official site (https:\/\/brew.sh\/) for the exact command:\n\n    **> ruby -e \"$(curl -fsSL https:\/\/raw.githubusercontent.com\/Homebrew\/install\/master\/install)\"**\n\n# IPython notebook crash course\n\nFeel free to skip this section if you're familiar with the Python and Jupyter notebooks.\n\nIPython notebook and its web-based GUI Jupyter are standard tools for data-driven machine learning development. Jupyter is also a handy tool for learning Python and its libraries. You can combine pieces of code with comments in markdown format. You can also execute pieces of code in place, chaining them one after another, and immediately seeing the results of computations. It also allows to embed interactive charts, tables, videos, and other multimedia objects inside the notebook. We will use Jupyter notebooks for writing quick prototypes of our models.\n\nTo create a new notebook, run in the Terminal:\n\n    **> jupyter notebook**\n\nYou will see output similar to this:\n\n    **[I 10:51:23.269 NotebookApp] Serving notebooks from local directory: ...**\n    **[I 10:51:23.269 NotebookApp] 0 active kernels**\n    **[I 10:51:23.270 NotebookApp] The Jupyter Notebook is running at: http:\/\/localhost:8888\/?token=3c073db5636e366fd750e661cc597652025fdbf41162c125**\n    **[I 10:51:23.270 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation).**\n\nNote those long URLs in the output: `http:\/\/localhost:8888\/token=3c073db5636e366fd750e661cc597652025fdbf41162c125`.\n\nCopy and paste this address to your browser to open Jupyter.\n\nWith Python 3, Jupyter automatically opens a new tab in your default browser's window, with the address `http:\/\/localhost:8888\/tree`.\n\nPress the New button, and choose Python 2 in a drop-down menu. This will open a new notebook in a new browser tab.\n\nTo stop IPython, you'll need to press _Ctrl_ \\+ _C_ in the Terminal, and enter `y` when prompted. Don't forget to save your changes in the notebook before quitting.\n\nLet's try something just to get the idea on how it works. In the top cell of the notebook, print `import this` and press _Shift_ \\+ _Enter_. You'll see `The Zen of Python`\u2014a short list of rules every Python programmer should conform to. We will also try to conform to them. The extended version of Python-style guidelines is known as **PEP 8** , and can be found here: python.org\/dev\/peps\/pep-0008\/.\n\nType into the new cell:\n\n    a = 2**32 \n    b = 64**(1\/2.) \n    a = a+b \n    a\n\nThen, press _Shift_ + _Enter_. This calculates 232 \\+ \u221a64, and stores the results into variable `a`. Unfamiliar operator `**` is a power, `a` and `b` are variables (no `let` or `var`). Typecasting between integer `1` and float `2` happens implicitly. Python is weak-typed, so you can assign float value of variable `b` to an integer variable `a`. Jupyter outputs the value of a last line in the cell. Also, note that variables `a` and `b` are available now in the next cells.\n\nIf you don't know Python, no worries\u2014it's a relatively simple language. For a crash course on Python, please visit: <https:\/\/learnxinyminutes.com\/docs\/python\/>.\n\nTo see how to add and format comments, place your cursor in the new cell, choose from the drop-down menu in the Instruments panel cell, type `Markdown`, and put some markdown snippet into the cell; for example, the following snippet is a simple text with MathJax-formatted formula and a picture:\n\n    # This is a sample text: \n    $$Formula = {Numerator over Denominator}$$ \n    ![]( https:\/\/imgs.xkcd.com\/comics\/conditional_risk.png) \n    > Sample text to demonstrate the few markdown feature available to easily create documents. [Packt Hyperlink](http:\/\/packtpub.com\/)\n\nYou'll get a nicely formatted MathJax formula, an image, and some formatted text. If you want to know more about markdown format, just Google for a markdown tutorial, or a cheat sheet.\n\nYou can also execute bash commands from the notebook; just prepend an exclamation mark to them:\n\n    In []: \n    ! ls\n\n    Out[]: \n    The content of your work folder goes here... \n\n# Time to practice\n\nIn the following sections, we'll dive into machine learning practice, to get a feeling of what it looks like. Just like in a theater play, in machine learning you have a list of characters and a list of acts.\n\nTwo main characters are:\n\n  * Dataset\n  * Model\n\nThree main acts are:\n\n  * Dataset preparation\n  * Model training\n  * Model evaluation\n\nWe'll go through all these acts, and by the end of the chapter we'll have our first trained model. First, we need to define a problem, and then we can start coding a prototype in Python. Our destination point is a working model in Swift. Don't take the problem itself too seriously, though, because as the first exercise, we're going to solve a fictional problem.\n\n# Machine learning for extra-terrestrial life explorers\n\nSwift is undoubtedly the programming language of the future. In the nearest years, we're expecting to see Swift being employed to program-intelligent scout robots that will explore alien planets and life forms on them. These robots should be able to recognize and classify aliens they will encounter. Let's build a model to distinguish between two alien species using their characteristic features.\n\nThe biosphere of the distant planet consists mainly of two species: night predators rabbosauruses, and peaceful, herbivorous platyhogs (see the following diagram). Roboscouts are equipped with sensors to measure only three features of each individual: length (in meters), color, and fluffiness.\n\nFigure 2.1: Objects of interest in our first machine learning task. Picture by Mykola Sosnovshchenko.\n\nThe full code of the Python part of this chapter can be found here: `ML_Intro.ipynb`.\n\n# Loading the dataset\n\nCreate and open a new IPython notebook. In the chapter's supplementary materials, you can see the file `extraterrestrials.csv`. Copy it to the same folder where you created your notebook. In the first cell of your notebook, execute the magical command:\n\n    In []: \n    %matplotlib inline\n\nThis is needed to see inline plots right in the notebook in the future.\n\nThe library we are using for datasets loading and manipulation is `pandas`. Let's import it, and load the `.csv` file:\n\n    In []: \n    import pandas as pd \n    df = pd.read_csv('extraterrestrials.csv', sep='t', encoding='utf-8', index_col=0)\n\nObject `df` is a data frame. This is a table-like data structured for efficient manipulations over the different data types. To see what's inside, execute:\n\n    In []: \n    df.head() \n    Out[]:  | **Length** | **Color** | **Fluffy** | **Label**\n\n---|---|---|---|---\n\n**0** | 27.545139 | Pink gold | True | Rabbosaurus\n\n**1** | 12.147357 | Pink gold | False | Platyhog\n\n**2** | 23.454173 | Light black | True | Rabbosaurus\n\n**3** | 29.956698 | Pink gold | True | Rabbosaurus\n\n**4** | 34.884065 | Light black | True | Rabbosaurus\n\nThis prints the first five rows of the table. The first three columns (length, color, and fluffy) are features, and the last one is the class label.\n\nHow many samples do we have in total? Run this code to find out:\n\n    In []: \n    len(df) \n    Out[]: \n    1000\n\nLooks like the most samples in the beginning are rabbosauruses. Let's fetch five samples at random to see if it holds true in other parts of the dataset:\n\n    In []: \n    df.sample(5) \n    Out[]:  | **Length** | **Color** | **Fluffy** | **Label**\n\n---|---|---|---|---\n\n**565** | 17.776481 | Purple polka dot | False | Platyhog\n\n**491** | 19.475358 | Light black | True | Rabbosaurus\n\n**230** | 15.453365 | Purple polka dot | False | Platyhog\n\n**511** | 17.408234 | Purple polka dot | True | Platyhog\n\n**875** | 24.105315 | Light black | True | Rabbosaurus\n\nWell, this isn't helpful, as it would be too tedious to analyze the table content in this way. We need some more advanced tools to perform descriptive statistics computations and data visualization.\n\n# Exploratory data analysis\n\nFirst, we want to see how many individuals of each class we have. This is important, because if the class distribution is very imbalanced (like 1 to 100, for example), we will have problems training our classification models. You can get data frame columns via the dot notation. For example, `df.label` will return you the label column as a new data frame. The data frame class has all kinds of useful methods for calculating the summary statistics. The `value_counts()` method returns the counts of each element type in the data frame:\n\n    In []: \n    df.label.value_counts() \n    Out[]: \n    platyhog       520 \n    rabbosaurus    480 \n    Name: label, dtype: int64\n\nThe class distribution looks okay for our purposes. Now let's explore the features.\n\nWe need to group our data by classes, and calculate feature statistics separately to see the difference between the creature classes. This can be done using the `groupby()` method. It takes the label of the column by which you want to group your data:\n\n    In []: \n    grouped = df.groupby('label')\n\nThe grouped data frame has all the same methods and column labels as the original data frame. Let's see the descriptive statistics of a length feature:\n\n    In []: \n    grouped.length.describe() \n    Out[]:\n\n**Label** | **Count** | **Mean** | **Std.** | **Min.** | **25%** | **50%** | **75%** | **Max.**\n\n---|---|---|---|---|---|---|---|---\n\n**Platyhog** | 520.0 | 19.894876 | 4.653044 | 4.164723 | 16.646311 | 20.168655 | 22.850191 | 32.779472\n\n**Rabbosaurus** | 480.0 | 29.984387 | 5.072308 | 16.027639 | 26.721621 | 29.956092 | 33.826660 | 47.857896\n\nWhat can we learn from this table? Platyhogs have a length with the mean of about 20 meters, and standard deviation of about 5. Rabbosauruses on average are 30 meters long, with a standard deviation of 5. The smallest platyhog is about 4 meters long, and the largest rabbosaurus is about 48 meters long. That's a lot, but less than the biggest Earth life forms (see Amphicoelias fragillimus, for example).\n\nColor distribution can be viewed using the familiar `value_counts()` method:\n\n    In []: \n    grouped.color.value_counts() \n    Out[]: \n    label        color            \n    platyhog     light black         195 \n                 purple polka-dot    174 \n                 pink gold           151 \n    rabbosaurus  light black         168 \n                 pink gold           156 \n                 space gray          156 \n    Name: color, dtype: int64\n\nWe can represent this in a more appealing form, using `unstack()` and `plot()` methods:\n\n    In []: \n    plot = grouped.color.value_counts().unstack().plot(kind='barh', stacked=True, figsize=[16,6], colormap='autumn') \n    Out[]:\n\n** **\n\nFigure 2.2: Color distribution\n\nLooks like purple polka dot is a strong predictor of a `platyhog` class. But if we see a space-gray individual, we can be sure we should run quickly.\n\nIn a similar manner, fluffiness distribution can be visualized using:\n\n    In []: \n    plot = grouped.fluffy.value_counts().unstack().plot(kind='barh', stacked=True, figsize=[16,6], colormap='winter') \n    Out[]:\n\nFigure 2.3: Fluffiness distribution\n\nRabbosauruses go in three colors: light black, pink gold, and space gray. 90% of them are fluffy (the remaining 10% are probably old and bald). Platyhogs, on the other hand, can be light black, pink gold, or purple polka dot. 30% of them are fluffy (mutants, maybe?).\n\nFor more complex data visualization, we need the `matplotlib` plotting library:\n\n    In []: \n    import matplotlib.pyplot as plt\n\nDrawing the histogram of length distribution:\n\n    In []: \n    plt.figure() \n    plt.hist(df[df.label == 'rabbosaurus'].length, bins=15, normed=True) \n    plt.hist(df[df.label == 'platyhog'].length, bins=15, normed=True) \n    plt.title(\"Length Distribution Histogram\") \n    plt.xlabel(\"Length\") \n    plt.ylabel(\"Frequency\") \n    fig = plt.gcf() \n    plt.show() \n    Out[]:\n\nFigure 2.4: Length distribution\n\nIn general, one can say that the platyhogs are smaller, but there is significant range of overlap approximately between 20 and 30 meters, where the length alone is not enough to discriminate between two classes.\n\n# Data preprocessing\n\nIn the following sections we will take a look at the different data processing techniques.\n\n# Converting categorical variables\n\nAs you already have noticed, a data frame can contain columns with the data of different types. To see which type has each column, we can check the `dtypes` attribute of the data frame. You can think about Python attributes as being similar to Swift properties:\n\n    In []: \n    df.dtypes \n    Out[]: \n    length    float64 \n    color      object \n    fluffy       bool \n    label      object \n    dtype: object\n\nWhile `length` and `fluffy` columns contain the expected datatypes, the types of `color` and `label` are less transparent. What are those objects? This means those columns can contain any type of the object. At the moment, we have strings in them, but what we really want them to be are categorical variables. In case you don't remember from the previous chapter, categorical variables are like Swift enums. Fortunately for us, data frame has handy methods for converting columns from one type to another:\n\n    In []: \n    df.color = df.color.astype('category') \n    df.label = df.label.astype('category')\n\nThat's it. Let's check:\n\n    In []: \n    df.dtypes \n    Out []: \n    length     float64 \n    color     category \n    fluffy        bool \n    label     category \n    dtype: object\n\n`color` and `label` are categories now. To see all colors in those categories, execute:\n\n    In []: \n    colors = df.color.cat.categories.get_values().astype('string') \n    colors \n    Out[]: \n    array(['light black', 'pink gold', 'purple polka-dot', 'space gray'], dtype='|S16')\n\nAs expected, we have four colors. `'|S16'` stands for strings of 16 characters in length.\n\n# Separating features from labels\n\nLet's separate our features from the labels, as we will feed them into the model separately:\n\n    In []: \n    features = df.loc[:,:'fluffy'] \n    labels = df.label\n\nThis horrible construction `df.loc[:,:'fluffy']` tells the data frame that we want all the rows (the first column), and the columns starting from the first, finishing with `'fluffy'`.\n\n# One-hot encoding\n\nMost of the machine learning algorithms can't work with the categorical variables, so usually we want to convert them to the one-hot vectors (statisticians prefer to call them **dummy variables** ). Let's convert first, and then I will explain what this is:\n\n    In []: \n    features = pd.get_dummies(features, columns = ['color']) \n    features.head() \n    Out[]:  | `length` | `fluffy` | `color_light black` | `color_pink gold` | `color_purple polka-dot` | `color_space gray`\n\n---|---|---|---|---|---|---\n\n0 | 27.545139 | True | 0 | 1 | 0 | 0\n\n1 | 12.147357 | False | 0 | 1 | 0 | 0\n\n2 | 23.454173 | True | 1 | 0 | 0 | 0\n\n3 | 29.956698 | True | 0 | 1 | 0 | 0\n\n4 | 34.884065 | True | 1 | 0 | 0 | 0\n\nSo now, instead of one column, `color`, we have four columns: `color_light black`, `color_pink gold`, `color_purple polka dot`, and `color_space gray`. The color of each sample is encoded as 1 in the corresponding column. Why do we need this if we could simply replace colors with the numbers from 1 to 4? Well, this is the problem: why to prefer 1 to 4 over the 4 to 1, or powers of 2, or prime numbers? These colors on their own don't carry any quantitative information associated to them. They can't be sorted from the largest to the smallest. If we introduce this information artificially, the machine learning algorithm may attempt to utilize that meaningless information, and we will end up with the classifier that sees regularities where there are none.\n\n# Splitting the data\n\nFinally, we want to split our data into training and test sets. We will train our classifier only on the training set, so it will never see the test set until we want to evaluate its performance. This is a very important step, because as we will see in the future, the quality of predictions on the test set can differ dramatically from the quality measured on the training set. Data splitting is an operation specific to machine learning tasks, so we will import scikit-learn (a machine learning package) and use some functions from it:\n\n    In []: \n    from sklearn.model_selection import train_test_split \n    X_train, X_test, y_train, y_test = train_test_split(features, labels, test_size=0.3, random_state=42) \n    X_train.shape, y_train.shape, X_test.shape, y_test.shape \n    Out[]: \n     ((700, 6), (700,), (300, 6), (300,))\n\nNow we have 700 training samples with 6 features each, and 300 test samples with the same number of features.\n\n# Decision trees everywhere\n\nThe algorithm that we're going to use for our first machine learning exercise is called a **decision tree classifier**. A decision tree is a set of rules that describe the process of decision making (see _figure 2.5_ for example).\n\nDecision trees are widely used outside the machine learning in different domains; for example, in business analysis. The popularity of decision trees is understandable: they are easy to interpret, and nice to visualize. For many years, they were built manually using the domain expert knowledge. Fortunately, now we have machine learning algorithms that can easily turn almost any labeled dataset into a decision tree.\n\n# Training the decision tree classifier\n\nLet's learn how to train the decision tree classifier as shown in the following code snippet:\n\n    In []: \n    from sklearn import tree \n    tree_model = tree.DecisionTreeClassifier(criterion='entropy', random_state=42) \n    tree_model = tree_model.fit(X_train, y_train) \n    tree_model \n    Out[]: \n    DecisionTreeClassifier(class_weight=None,  \n                criterion='entropy', max_depth=None, \n                max_features=None, max_leaf_nodes=None, \n                min_impurity_split=1e-07, min_samples_leaf=1, \n                min_samples_split=2, min_weight_fraction_leaf=0.0, \n                presort=False, random_state=42, splitter='best')\n\nThe most interesting for us are the class attributes of `DecisionTreeClassifier`:\n\n  * `criterion`: The way to estimate the best partition (see the _How decision tree learning works_ section).\n  * `max_depth`: Maximum tree depth.\n  * `max_features`: The maximum number of attributes to account in one split.\n  * `min_samples_leaf`: The minimum number of objects in the leaf; for example, if it is equal to `3`, then the tree will generate only those classification rules that are true for at least three objects.\n\nThese attributes are known as **hyperparameters**. They are different from model parameters: the former is something that users can tweak, and the latter is something that machine learning algorithm learns. In a decision tree, parameters are specific rules in its nodes. The tree hyperparameters must be adjusted depending on the input data, and this is usually done using cross-validation (stay tuned).\n\nDecision tree classifier documentation: http:\/\/scikit-learn.org\/stable\/modules\/tree.html.\n\nThe properties of the model, which are not adjusted (learned) by the model itself, but are available for the user's adjustments, are known as hyperparameters. In the case of the decision tree model, these hyperparameters are `class_weight`, `criterion`, `max_depth`, `max_features`, and so on. They are like knobs you can turn to adjust the model to your specific needs.\n\n# Tree visualization\n\nLet us take a look at the code to visualize a tree as follows:\n\n    In []: \n    labels = df.label.astype('category').cat.categories \n    labels = list(labels) \n    labels \n    Out[]: \n    [u'platyhog', u'rabbosaurus']\n\nDefine a variable to store all the names for the features:\n\n    In []: \n    feature_names = map(lambda x: x.encode('utf-8'), features.columns.get_values()) \n    feature_names \n    Out[]: \n    ['length', \n     'fluffy', \n     'color_light black', \n     'color_pink gold', \n     'color_purple polka-dot', \n     'color_space gray']\n\nThen, create the `graph` object using the `export_graphviz` function:\n\n    In []: \n    import pydotplus  \n    dot_data = tree.export_graphviz(tree_model, out_file=None,  \n                                    feature_names=feature_names,   \n                                     class_names=labels,   \n                                     filled=True, rounded=True,   \n                                     special_characters=True) \n    dot_data \n    Out[]: \n    u'digraph Tree {nnode [shape=box, style=\"filled, rounded\", color=\"black\", fontname=helvetica] ;nedge [fontname=helvetica] ;n0 [label=<length &le; 26.6917<br\/>entropy = 0.9971<br\/>samples = 700<br\/>value = [372, ... \n    In []: \n    graph = pydotplus.graph_from_dot_data(dot_data.encode('utf-8')) \n    graph.write_png('tree1.png') \n    Out[]: \n    True\n\nPut a markdown to the next cell to see the newly-created file as follows:\n\n    ![](tree1.png)\n\nFigure 2.5: Decision tree structure and a close-up of its fragment\n\nThe preceding diagram shows what our decision tree looks like. During the training, it grows upside-down. Data (features) travels through it from its root (top) to the leaves (bottom). To predict the label for a sample from our dataset using this classifier, we should start from the root, and move until we reach the leaf. In each node, one feature is compared to some value; for example, in the root node, the tree checks if the length is < 26.0261. If the condition is met, we move along the left branch; if not, along the right.\n\nLet's look closer at a part of the tree. In addition to the condition in each node, we have some useful information:\n\n  * Entropy value\n  * Number of samples in the training set which supports this node\n  * How many samples support each outcome\n  * The most likely outcome at this stage\n\n# Making predictions\n\nWe use the `predict` function to get outcome labels for two samples. The first one is light-black, fluffy creature, 24 meters long. The second one is purple polka dot, non-fluffy, and 34 meters long. If you already don't remember the meaning of each feature, consult the `feature_names` variable:\n\n    In []: \n    samples = [[24,1,0,1,0,0], [34,0,0,0,1,0]] \n    tree_model.predict(samples) \n    Out[]: \n    array([u'platyhog', u'rabbosaurus'], dtype=object)\n\nOur model predicted `platyhog` for the first sample, and `rabbosaurus` for the second one. A decision tree can also provide probabilistic output (how sure it is about the prediction):\n\n    In []: \n    tree_model.predict_proba(samples) \n    Out[]: \n    array([[ 1.,  0.], \n           [ 0.,  1.]])\n\nThe array contains two nested arrays, one for every prediction. Elements in the nested arrays are probabilities of the sample belonging to the corresponding class. This means that our model is 100% sure that the first sample belongs to the first class, and 100% sure that the second sample belongs to the second class.\n\nBut how sure can we be about these predictions? We have a whole set of different tools to evaluate the model's accuracy, and the simplest one is the built-in scoring functions.\n\n# Evaluating accuracy\n\nScore function calculates accuracy of the model using the data. Let's calculate the accuracy of our model on the training set:\n\n    In []: \n    tree_model.score(X_train, y_train) \n    Out[]: \n    1.0\n\nWow, looks like our model is 100% accurate. Isn't it a great result? Let's not hurry and check our model on held-out data. Evaluation on the test set is the golden standard of success in machine learning:\n\n    In []: \n    tree_model.score(X_test, y_test) \n    Out[]: \n    0.87666666666666671\n\nWorse now. What's just happened? Here, the first time we were faced with the problem of overfitting, when the model is trying to fit itself to every quirk in the data. Our model adjusted itself to the training data so much, that on the previously unseen data, it lacks the ability to generalize. As any real-world data contains noise and signal, we want our models to fit to the signal and to ignore the noise component. Overfitting is the most common problem in machine learning. It's common when datasets are too small, or models are too flexible. The opposite situation is called underfitting\u2014when the model is not able to fit the complex data well enough:\n\nFigure 2.6: Underfitting (right column) versus good fit (central column) versus overfitting (right column). Top row shows classification problem, bottom row shows regression problem.\n\nAn overfitting problem is familiar to anyone who looked at some item at the online store, and then was presented with targeted advertisement of the same item everywhere on the internet. This item most likely is not relevant anymore, but the machine learning algorithm already overfitted to the limited dataset, and now you have trinket rabbits (or whatever you've looked at on the e-store) on every page you open.\n\nIn any case, we must fight overfitting somehow. So, what can we do? The simplest solution is to make the model simpler and less flexible (or, speaking machine learning, to reduce model capacity).\n\n# Tuning hyperparameters\n\nThe simplest way to simplify the decision tree is to limit its depth. How deep is it now? You can see 20 splits, or 21 layers, in _Figure 2.5_. At the same time, we have only three features. There are six of them actually, if we are taking into account one-hot encoded categorical color. Let's limit the maximum depth of the tree aggressively to be comparable with the number of features. `tree_model` object has a `max_depth` property, and so we're setting it to be less than the number of features:\n\n    In []: \n    tree_model.max_depth = 4\n\nAfter these manipulations, we can retrain our model and reevaluate its accuracy:\n\n    In []: \n    tree_model = tree_model.fit(X_train, y_train) \n    tree_model.score(X_train, y_train) \n    Out[]: \n    0.90571428571428569\n\nNote that accuracy on training is now set less by about 6%. How about test set?\n\n    In []: \n    tree_model.score(X_test, y_test) \n    Out[]: \n    0.92000000000000004\n\nAccuracy on previously unseen data is now higher, by about 4%. This doesn't look like a great achievement, until you realize that it's an additional 40 correctly classified creatures from our initial set of 1,000. In modern machine learning contests, the final difference between 1st and 100th place can easily be about 1%.\n\nLet's draw a tree structure after pruning. Code for this visualization is the same as before:\n\nFigure 2.7: Tree structure after limiting its depth\n\n# Understanding model capacity trade-offs\n\nLet's train trees with different depths: starting from 1 split, and to maximal 23 splits:\n\n    In []: \n    train_losses = [] \n    test_losses = [] \n    for depth in xrange(1, 23): \n        tree_model.max_depth = depth \n        tree_model = tree_model.fit(X_train, y_train) \n        train_losses.append(1 - tree_model.score(X_train, y_train)) \n        test_losses.append(1 - tree_model.score(X_test, y_test)) \n    figure = plt.figure()  \n    plt.plot(train_losses, label=\"training loss\", linestyle='--') \n    plt.plot(test_losses, label=\"test loss\") \n    plt.legend(bbox_to_anchor=(0., 1.02, 1., .102), loc=3, ncol=2, mode=\"expand\", borderaxespad=0.) \n    Out[]:\n\nFigure 2.8: Training loss versus test loss, depending on the maximum tree depth\n\nOn the _x_ axis, we've plotted the tree depth, and on the _y_ axis, we've plotted the model's error. An interesting phenomenon that we're observing here is well familiar to any machine learning practitioner: as the model gets more complex, it gets more prone to overfitting. At first, as the model's capacity grows, both training and test loss (error) decreases, but then something strange happens: while error on the training set continues to go down, test error starts growing. This means that the model fits itself to the training examples so well, that it is not able to generalize well on unseen data anymore. That's why it's so important to have a held-out dataset, and perform your model validation on it. From the above plot, we can see that our more-or-less random choice of `max_depth=4` was lucky: test error at this point became even less than training error.\n\n# How decision tree learning works\n\nDecision tree learning is a supervised, non-parametric algorithm used for classification and regression.\n\n# Building a tree automatically from data\n\nThe _Twenty Questions_ game is a traditional game where one of the players is the answerer who chooses an object (or a famous person in some variants), not revealing what it is to the other participants. All the other players are trying to guess what the object is by asking questions like _Can I eat this?_ or I _s it a human?_ where answers can only be _yes_ or _no_.\n\nIf you have never heard about this game, refer to Wikipedia: https:\/\/en.wikipedia.org\/wiki\/Twenty_Questions.\n\nThis is essentially a tree learning algorithm. To win in a game, you should pose such questions that discriminate the most; for example, the question, _Is it alive?_ in the beginning of the game is clearly better than _Is it a cucumber?_. This ability to dissect the hypothesis space in an optimal way is formalized in the notion of information gain criterion.\n\n# Combinatorial entropy\n\nInformation gain criterion is based on the Shannon entropy notion. The Shannon entropy is a very important topic in the information theory, physics, and other domains. Mathematically, it is expressed as:\n\nWhere _i_ is a state of a system, _N_ is a total number of possible states, and _p i_ is a probability of the system being in the state _i_. Entropy describes the amount of uncertainty in the system. The more order you have in the system, the less entropy there is.\n\nFor the visual introduction to the information theory, check _Visual Information Theory_ by Christopher Olah at: http:\/\/colah.github.io\/posts\/2015-09-Visual-Information\/.\n\nIf you want to learn more about entropy, check the nice, interactive blog _Entropy Explained, With Sheep_ by Aatish Bhatia at: https:\/\/aatishb.com\/entropy\/.\n\nLet's show a simple example of how entropy can be useful for decision tree construction. For this, we'll simplify a task of alien creature classification, assuming that we can measure only one feature: body length. We have 10 individuals ( = platyhog and  = rabbosaurus) with the following body lengths:\n\n**True label** |   |   |   |   |   |   |   |   |   |\n\n---|---|---|---|---|---|---|---|---|---|---\n\n**Body length, meters** | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10\n\nIf we take one random individual from the group, it can be a platyhog with the probability of 0.6, or a rabbosaurus with the probability of 0.4. We have two states in this system for two outcomes. Let's calculate the entropy of it:\n\nSo, the amount of uncertainty in this dataset is 0.97. Is it a lot, or a little? We don't have anything yet to compare it with, so let's divide the set at the middle (> 5 meters), and calculate the entropy for both subsets:\n\n**True label** |   |   |   |   |\n\n|  |\n\n  |   |   |   |\n\n---|---|---|---|---|---|---|---|---|---|---|---\n\n**Body length** | 1 | 2 | 3 | 4 | 5\n\n|  |\n\n6 | 7 | 8 | 9 | 10\n\n_H_ and _H_ are now less than the original _H_. This demonstrates how you can reduce the entropy by splitting the dataset in the right place. This idea lies in the fundamentals of decision tree learning algorithms.\n\nWe can calculate how effectively we reduced the entropy by splitting the set using the **Information Gain** ( **IG** ) criterion:\n\n_Information Gain = Entropy(parent) - Weighted Sum of Entropy (Children)_ , or:\n\n_q_ is a number of groups after splitting, _N i_ is a count of elements in the i-th group, _N_ \u2014is the total count of elements before split. In our example, _q = 2_ , _N = 10_ , and _N 1 = N2 = 5_:\n\nThis means that asking the question _Is the body length greater than 5?_ gives us an information gain of _0.61_. Is it a lot, or a little? Let's compare it to the information loss of the split around length > 7:\n\nApparently, the choice of the middle point was lucky, because all other splits don't look promising. But you are free to check them if you want.\n\nThere is no sense to split the left part further, but we can continue splitting the right subset until entropy of each of its children will not be equal to zero (see _Figure 2.9_ ).\n\nSo, this is our decision tree, and a recursive algorithm for its building. But now comes an interesting question: how to know which split yields the maximal information gain? The simplest way is a greedy search: just check all possible variants.\n\nInformation gain is only one of the heuristics, there are more of them; for instance, in our scikit-learn decision tree learner, we used Gini impurity as a heuristic. According to the Michigan State University (<http:\/\/www.cse.msu.edu\/~cse802\/DecisionTrees.pdf>):\n\n\"Gini impurity is the expected error rate at node N if the category label is selected randomly from the class distribution present at N.\"\n\nCheck the documentation on the `criterion` property of `DecisionTreeClassifier` for more information about different heuristics available for tree learning in scikit-learn. In practice, Gini works very similarly to the information gain. A historical fact to dilute the theoretical exposition: Corrado Gini was an Italian statistician and the author of _The Scientific Basis of Fascism_ (1927):\n\nFigure 2.9: Building a decision tree. _H_ stands for entropy in each group. Picture by Mykola Sosnovshchenko.\n\n# Evaluating performance of the model with data\n\nThe ways to assess the quality of a model's predictions quantitatively are known as **metrics**. The simplest metric in classification is accuracy, a proportion of correctly classified cases. Accuracy metric can be misleading. Imagine that you have a training set with 1000 samples. 999 of them are of class A, and 1 of class B. Such a kind of dataset is called **imbalanced**. The baseline (the simplest) solution in this case would be to always predict class A. Accuracy of such a model would then be 0.999, which can be pretty impressive, but only if you don't know about the ratio of classes in the training set. Now imagine that class A corresponds to an outcome of healthy, and class B to cancer, in the medical diagnostic system. It's clear now that 0.999 accuracy is worth nothing, and totally misleading. Another thing to consider is that the cost of different errors can be different. What's worse: to diagnose a healthy person as ill, or an ill person as healthy? This leads to the notion of two types of error ( _Figure 2.10_ ):\n\n  * Type I error, also known as **false positive** : algorithm predicts **cancer** , while there is no cancer\n  * Type II error, also known as, **false negative** : algorithm predicts **no cancer** , while there is.\n\nFigure 2.9: Two types of errors represented as a Venn diagram\n\n# Precision, recall, and F1-score\n\nTo assess the quality of the algorithm considering the two types of error, accuracy metric is useless. That's why different metrics were proposed.\n\n**Precision** and **recall** are metrics used to evaluate a prediction's quality in information retrieval and binary classification. Precision is a proportion of true positives among all predicted positives. It shows how relevant results are. Recall, also known as **sensitivity** , is a proportion of true positives among all truly positive samples. For example, if the task is to distinguish cat photos from non-cat photos, precision is a fraction of correctly predicted cats to all predicted cats. Recall is a fraction of predicted cats to the total number of true cats.\n\nIf we denote the number of true positive cases as _T p_, and number of false positive cases as _F p_, then precision _P_ is calculated as:\n\nRecall _R_ is calculated as:\n\n ,\n\nWhere _F n_ is a number of false negative cases.\n\n_F1_ measure is calculated as:\n\nNow the same in Python:\n\n    In []: \n    import numpy as np \n    predictions = tree_model.predict(X_test) \n    predictions = np.array(map(lambda x: x == 'rabbosaurus', predictions), dtype='int') \n    true_labels = np.array(map(lambda x: x == 'rabbosaurus', y_test), dtype='int') \n    from sklearn.metrics import precision_score, recall_score, f1_score \n    precision_score(true_labels, predictions) \n    Out[]: \n    0.87096774193548387 \n    In []: \n    recall_score(true_labels, predictions) \n    Out[]: \n    0.88815789473684215 \n    In []: \n    f1_score(true_labels, predictions) \n    Out[]: \n    0.87947882736156346 \n\n# K-fold cross-validation\n\nThis method was invented and gained popularity in those days when the big date was not yet a problem, everyone had little data, but still needed to build reliable models. First thing we do is shuffle our dataset well, and then divide it randomly into several equal parts, say 10 (this is the _k_ in k-fold). We hold out the first part as a test set, and on the remaining nine parts we train the model. The trained model is then assessed on the test set that did not participate in the training as usual. Next, we hold out the second of 10 parts, and train the model on the remaining nine (including those previously served as a test set). We validate the new model again on the part that did not participate in the training. We continue this process until each of the 10 parts is in the role of the test set. The final quality metrics are determined by the averaging metrics from each of the 10 tests:\n\n    In []: \n    from sklearn.model_selection import cross_val_score \n    scores = cross_val_score(tree_model, features, df.label, cv=10) \n    np.mean(scores) \n    Out[]: \n    0.88300000000000001 \n    In []: \n    plot = plt.bar(range(1,11), scores) \n    Out[]:\n\nFigure 2.10: Cross-validation results\n\nFrom the preceding graph, you can see that the model's accuracy depends on how you split the data, but not much. By taking the average and variance of the cross-validation results, you can make a sense of how well your model can generalize on different data, and how stable it is.\n\n# Confusion matrix\n\nConfusion matrix helps to see what types of errors occur more often:\n\n    In []: \n    from sklearn.metrics import confusion_matrix \n    confusion_matrix(y_test, tree_model.predict(X_test)) \n    Out[]: \n    array([[128,  20], \n           [ 17, 135]])\n\nThis is how to read and interpret such matrices:\n\n| **Predicted labels**  \n---|---  \n**True labels** |  | **Platyhog** | **Rabbosaurus**  \n**Platyhog** | 128 | 20  \n**Rabbosaurus** | 17 | 135\n\nThe bigger the numbers on the matrix diagonally, the better.\n\n# Implementing first machine learning app in Swift\n\nYou can transfer your model from Python to Swift in two ways: transfer a trained model, or train a model from the ground up in Swift. The first option is easy in the case of decision trees, as a trained model can be expressed as a set of if-else conditions, which is trivial to code manually. Training the model from the ground up is required only in the situation where you want your app to learn in runtime. We will stick to the first approach in this example, but instead of coding rules manually, we will export the scikit-learn model for iOS using Core ML tools.\n\n# Introducing Core ML\n\nCore ML was first presented at Apple WWDC 2017. Defining Core ML as machine learning framework is not fair, because it lacks learning capabilities; it's rather a set of conversion scripts to plug the pre-trained model into your Apple applications. Still, it is an easy way for newcomers to start running their first models on iOS.\n\n# Core ML features\n\nHere is a list of Core ML features:\n\n  * `coremltools` Python package includes several converters for popular machine learning frameworks: scikit-learn, Keras, Caffe, LIBSVM, and XGBoost.\n  * Core ML framework allows running inference (making predictions) on a device. Scikit-learn converter also supports some data transformation and model pipelining.\n  * Hardware acceleration (Accelerate framework and Metal under the hood).\n  * Supports iOS, macOS, tvOS, and watchOS.\n  * Automatic code generation for OOP-style interoperability with Swift.\n\nThe biggest Core ML limitation is that it doesn't support models training.\n\n# Exporting the model for iOS\n\nIn our Jupyter notebook, execute the following code to export the model:\n\n    In []: \n    import coremltools as coreml \n    coreml_model = coreml.converters.sklearn.convert(tree_model, feature_names, 'label') \n    coreml_model.author = \"Author name goes here...\" \n    coreml_model.license = \"License type goes here ...\" \n    coreml_model.short_description = \"Decision tree classifier for extraterrestrials.\" \n    coreml_model.input_description['data'] = \"Extraterrestrials features\" \n    coreml_model.output_description['prob'] =  \"Probability of belonging to class.\" \n    coreml_model.save('DecisionTree.mlmodel')\n\nScikit-learn converter documentation: http:\/\/pythonhosted.org\/coremltools\/generated\/coremltools.converters.sklearn.convert.html#coremltools.converters.sklearn.convert\n\nThe code creates the `tree.mlmodel` file next to the Jupyter notebook file. This file can contain a single model, a model pipeline (several models chained one after another), or a list of scikit-learn models. According to the documentation, the scikit-learn converter supports the following types of machine learning models:\n\n  * Decision tree learning\n  * Tree ensembles\n  * Random forests\n  * Gradient boosting\n  * Linear and logistic regression (see Chapter 5, _Association Rule Learning_ )\n  * Support vector machines (several types)\n\nIt also supports the following data transformations:\n\n  * Normalizer\n  * Imputer\n  * Standard scaler\n  * DictVectorizer\n  * One-hot encoder\n\nNote that you can embed one-hot encoding as a part of pipeline, so you don't need to do it yourself in your Swift code. This is handy, because you don't need to keep track of the proper order of categorical variable levels.\n\nThe `.mlmodel` file can be one of three types: classifier, regressor, or a transformer, depending on the last model in the list, or a pipeline. It is important to understand that there is no direct correspondence between scikit-learn models (or other source framework) and Core ML models that run on a device. Because Core ML sources are closed, we don't know how it operates under the hood, and can't be sure that the model before and after the conversion will produce identical results. This means you need to validate the model after device deployment, to measure its performance and accuracy.\n\n# Ensemble learning random forest\n\nOne-sentence explanation for LOTR fans: if decision trees were Ents, the random forest would be an Entmoot. For everyone else, random forest algorithm works like this:\n\n  * Split data into random subsets of equal size, maybe with replacement\n  * On each of those subsets, build a decision tree, choosing for every split a random feature subset of fixed size\n  * To perform inference, perform a voting among the trees (classification), or average their predictions (regression)\n\nSuch tree ensembles are very popular in certain domains, because their prediction quality beats most other models.\n\nMost likely, this is not the model you want to train on a mobile device, due to the memory and time limitations, but you can still use it for inference thanks to Core ML. The workflow looks like this:\n\n  * Pre-train random forest in scikit-learn\n  * Export the model in the scikit-learn format\n  * Convert it to the Apple `mlmodel` format with the help of the `coremltool` Python package\n  * Import it in your iOS project using Core ML framework\n\nBy the way, if you look at the inner structure of the GameplayKit's tree learner in a debugger or playground, you'll see that it also uses random forest under the hood.\n\n# Training the random forest\n\nTraining the random forest model is not very different from training the decision tree:\n\n    In []: \n    from sklearn.ensemble import RandomForestClassifier \n    rf_model = RandomForestClassifier(criterion = 'entropy', random_state=42) \n    rf_model = rf_model.fit(X_train, y_train) \n    print(rf_model) \n    Out[]: \n    RandomForestClassifier(bootstrap=True, class_weight=None, criterion='entropy', \n                max_depth=None, max_features='auto', max_leaf_nodes=None, \n                min_impurity_split=1e-07, min_samples_leaf=1, \n                min_samples_split=2, min_weight_fraction_leaf=0.0, \n                n_estimators=10, n_jobs=1, oob_score=False, random_state=42, \n                verbose=0, warm_start=False)\n\nDocumentation at: <http:\/\/scikit-learn.org\/stable\/modules\/generated\/sklearn.ensemble.RandomForestClassifier.html>.\n\n# Random forest accuracy evaluation\n\nLoss on training data:\n\n    In []: \n    rf_model.score(X_train, y_train) \n    Out[]: \n    0.98999999999999999\n\nLoss on test data:\n\n    In []: \n    rf_model.score(X_test, y_test) \n    Out[]: \n    0.90333333333333332\n\nCross-validation:\n\n    In []:  \n    scores = cross_val_score(rf_model, features, df.label, cv=10) \n    np.mean(scores) \n    Out[]: \n    0.89700000000000002 \n    In []: \n    print(\"Accuracy: %0.2f (+\/- %0.2f)\" % (scores.mean(), scores.std() * 2)) \n    Accuracy: 0.90 (+\/- 0.06)\n\nPrecision and recall:\n\n    In []: \n    predictions = rf_model.predict(X_test) \n    predictions = np.array(map(lambda x: x == 'rabbosaurus', predictions), dtype='int') \n    true_labels = np.array(map(lambda x: x == 'rabbosaurus', y_test), dtype='int') \n    precision_score(true_labels, predictions) \n    Out[]: \n    0.9072847682119205 \n    In []: \n    recall_score(true_labels, predictions) \n    Out[]: \n    0.90131578947368418\n\n_F1-_ score:\n\n    In []: \n    f1_score(true_labels, predictions) \n    Out[]: \n    0.90429042904290435\n\nConfusion matrix:\n\n    In []: \n    confusion_matrix(y_test, rf_model.predict(X_test)) \n    Out[]: \n    array([[134,  14], \n           [ 15, 137]])\n\nYou export a random forest for the iOS in the same way you do for a decision tree.\n\n# Importing the Core ML model into an iOS project\n\nCreate a new iOS project and drag and drop the `DecisionTree.mlmodel` into a project tree in Xcode. Click on it to see a machine learning model navigator screen:\n\nFigure 2.11: Machine learning navigator screen\n\nOn this screen, you can find a familiar model description, model type (pipeline by some reason, in this case), the name of the Swift class that represents the model in the app, and lists of inputs and outputs. If you click on the small arrow next to the class name in the Model Class section, the autogenerated file `DecisionTree.swift` is opened. This reminds a Core Data framework, where you have autogenerated files for `NSMangedObject` subclasses. `DecisionTree.swift` contains three classes:\n\n  * `DecisionTreeInput`: `MLFeatureProvider`, contains the input features (six of them, all Double).\n  * `DecisionTreeOutput`: `MLFeatureProvider`, contains class label and class probability.\n  * `DecisionTree`: `NSObject`, the class of the model itself. It contains methods for initialization and making predictions.\n\nThe method `init(contentsOf: url)` allows to replace the model in runtime, but only if you preserve the input and output structure. For example, this is how the model is loaded from the file in the bundle:\n\n    let bundle = Bundle.main \n    let assetPath = bundle.url(forResource: \"DecisionTree\", withExtension:\"mlmodelc\") \n    let sklDecisionTree = DecisionTree(contentsOf: assetPath!)\n\nIn a same way, you can create a model with the content of a remote URL.\n\nDrag and drop the `RandomForest.ml` model to the project to also compare accuracy of the models on the iOS.\n\n# Evaluating performance of the model on iOS\n\nI'm not describing here a `.csv` parsing in Swift; if you are interested in the details, please see the supplementary materials. Assuming that you've successfully loaded the test data in the form of two arrays, `[Double]` for features and `[String]` for labels, have a go at the following code:\n\n    let (xMat, yVec) = loadCSVData()\n\nTo create a decision tree and evaluate it, try this:\n\n    let sklDecisionTree = DecisionTree()\n\n    let xSKLDecisionTree = xMat.map { (x: [Double]) -> DecisionTreeInput in \n        return DecisionTreeInput(length: x[0], \n                                 fluffy: x[1], \n                                 color_light_black: x[2], \n                                 color_pink_gold: x[3], \n                                 color_purple_polka_dot: x[4], \n                                 color_space_gray: x[5]) \n    }\n\n    let predictionsSKLTree = try! xSKLDecisionTree \n        .map(sklDecisionTree.prediction) \n        .map{ prediction in \n            return prediction.label == \"rabbosaurus\" ? 0 : 1 \n    }\n\n    let groundTruth = yVec.map{ $0 == \"rabbosaurus\" ? 0 : 1 }\n\n    let metricsSKLDecisionTree = evaluateAccuracy(yVecTest: groundTruth, predictions: predictionsSKLTree) \n    print(metricsSKLDecisionTree)\n\nTo create a random forest and evaluate it, trying using the following code:\n\n    let sklRandomForest = RandomForest()\n\n    let xSKLRandomForest = xMat.map { (x: [Double]) -> RandomForestInput in \n        return RandomForestInput(length: x[0], \n                                 fluffy: x[1], \n                                 color_light_black: x[2], \n                                 color_pink_gold: x[3], \n                                 color_purple_polka_dot: x[4], \n                                 color_space_gray: x[5]) \n    }\n\n    let predictionsSKLRandomForest = try! xSKLRandomForest.map(sklRandomForest.prediction).map{$0.label == \"rabbosaurus\" ? 0 : 1}\n\n    let metricsSKLRandomForest = evaluateAccuracy(yVecTest: groundTruth, predictions: predictionsSKLRandomForest) \n    print(metricsSKLRandomForest)\n\nThis is an example of how you can evaluate your model's prediction quality in the Swift application. The structure, that contains the results of evaluation is as follows:\n\n    struct Metrics: CustomStringConvertible { \n    let confusionMatrix: [[Int]] \n    let normalizedConfusionMatrix: [[Double]] \n    let accuracy: Double \n    let precision: Double \n    let recall: Double \n    let f1Score: Double\n\n    var description: String { \n        return \"\"\" \n        Confusion Matrix: \n        (confusionMatrix)\n\n        Normalized Confusion Matrix: \n        (normalizedConfusionMatrix)\n\n        Accuracy: (accuracy) \n        Precision: (precision) \n        Recall: (recall) \n        F1-score: (f1Score) \n        \"\"\" \n    } \n    }\n\nFor the function for quality assessment, here's the code:\n\n    func evaluateAccuracy(yVecTest: [Int], predictions: [Int]) -> Metrics { \n\n# Calculating the confusion matrix\n\nWe'll use a straightforward approach here to calculate the confusion matrix; however, this would not work for multiclass classification. Here, `p` stands for predicted value, and `t` is for ground truth:\n\n    let pairs: [(Int, Int)] = zip(predictions, yVecTest).map{ ($0.0, $0.1) } \n    var confusionMatrix = [[0,0], [0,0]] \n    for (p, t) in pairs { \n        switch (p, t) { \n        case (0, 0): \n            confusionMatrix[0][0] += 1 \n        case (0, _): \n            confusionMatrix[1][0] += 1 \n        case (_, 0): \n            confusionMatrix[0][1] += 1 \n        case (_, _): \n            confusionMatrix[1][1] += 1 \n        } \n    }\n\n    let totalCount = Double(yVecTest.count)\n\nNormalize the matrix by total count:\n\n    let normalizedConfusionMatrix = confusionMatrix.map{$0.map{Double($0)\/totalCount}}\n\nAs we already know, accuracy is a number of true predictions divided by the total number of cases.\n\nTo calculate accuracy, try using the following code:\n\n    let truePredictionsCount = pairs.filter{ $0.0 == $0.1 }.count \n    let accuracy = Double(truePredictionsCount) \/ totalCoun\n\nTo calculate true positive, false positive, and false negative counts, you can use the numbers from the confusion matrix, but let's do it the proper way:\n\n    let truePositive = Double(pairs.filter{ $0.0 == $0.1 && $0.0 == 0 }.count) \n    let falsePositive = Double(pairs.filter{ $0.0 != $0.1 && $0.0 == 0 }.count) \n    let falseNegative = Double(pairs.filter{ $0.0 != $0.1 && $0.0 == 1 }.count)\n\nTo calculate precision:\n\n    let precision = truePositive \/ (truePositive + falsePositive)\n\nTo calculate recall:\n\n    let recall = truePositive \/ (truePositive + falseNegative)\n\nTo calculate _F 1-_score:\n\n    let f1Score = 2 * precision * recall \/ (precision + recall)\n\n    return Metrics(confusionMatrix: confusionMatrix, normalizedConfusionMatrix: normalizedConfusionMatrix, accuracy: accuracy, precision: precision, recall: recall, f1Score: f1Score) \n    }\n\nHere is my result for the decision tree on iOS:\n\n    Confusion Matrix: \n    [[135, 17],  \n    [20, 128]]\n\n    Normalized Confusion Matrix: \n    [[0.45000000000000001, 0.056666666666666664],  \n    [0.066666666666666666, 0.42666666666666669]]\n\n    Accuracy: 0.876666666666667 \n    Precision: 0.870967741935484 \n    Recall: 0.888157894736842 \n    F1-score: 0.879478827361563\n\nAnd for the random forest:\n\n    Confusion Matrix: \n    [[138, 14],  \n    [18, 130]]\n\n    Normalized Confusion Matrix: \n    [[0.46000000000000002, 0.046666666666666669],  \n    [0.059999999999999998, 0.43333333333333335]]\n\n    Accuracy: 0.893333333333333 \n    Precision: 0.884615384615385 \n    Recall: 0.907894736842105 \n    F1-score: 0.896103896103896\n\nCongratulations! We've trained two machine learning algorithms, deployed them to the iOS, and evaluated their accuracy. Interesting that while decision tree metrics match perfectly, the random forest performance is slightly worse on Core ML. Don't forget to always validate your model after any type of conversion.\n\n# Decision tree learning pros and cons\n\nAdvantages:\n\n  * Easy to understand and interpret, perfect for visual representation. This is an example of a white box model, which closely mimics the human decision-making process.\n  * Can work with numerical and categorical features.\n  * Requires little data preprocessing: no need for one-hot encoding, dummy variables, and so on.\n  * Non-parametric model: no assumptions about the shape of data.\n  * Fast for inference.\n  * Feature selection happens automatically: unimportant features will not influence the result. The presence of features that depend on each other (multicollinearity) also doesn't affect the quality.\n\nDisadvantages:\n\n  * It tends to overfit. This usually can be mitigated in one of three ways: \n    * Limiting tree depth\n    * Setting the minimal number of objects in leaves\n    * Tree pruning by deleting unimportant splits moving from the leaves to the root\n  * It is unstable\u2014small changes in data can dramatically affect the structure of the tree and the final prediction.\n  * The problem with finding the globally optimal decision tree is NP-complete. That's why we use different heuristics and greedy search. Unfortunately, this approach doesn't guarantee learning the globally best tree, only locally optimal ones.\n  * Inflexible, in the sense that you can't incorporate a new data into them easily. If you obtained new labeled data, you should retrain the tree from scratch on the whole dataset. This makes decision trees a poor choice for any applications that require dynamic model adjustment.\n\n# Summary\n\nIn this chapter, we had our first experience of building a machine learning application, starting from the data and all the way over to the working iOS application. We went through several phases in this chapter:\n\n  * Exploratory data analysis using Jupyter, pandas, and Matplotlib\n  * Data preparation\u2014splitting, and handling categorical variables\n  * Model prototyping using scikit-learn\n  * Model tuning and evaluation\n  * Porting prototype for the mobile platform using Core ML\n  * Model validation on a mobile device\n\nThere are several machine learning topics that we've learned about in this chapter: model parameters vs. hyperparameters, overfitting vs. underfitting, evaluation metrics: cross-validation, accuracy, precision, recall, and _F_ _1-_ score. These are the basic things that will be recurring topics throughout this book.\n\nWe've become acquainted with two machine learning algorithms, namely decision trees and random forest, a type of model ensemble.\n\nIn the next chapter, we're going to continue exploring classification algorithms, and will learn about instance-based learning algorithms. We will also build an iOS app that can learn right on the device, this time not for alien classification, but for some real-world problem, I promise.\n\n# K-Nearest Neighbors Classifier\n\nThis chapter is devoted to an important class of machine learning algorithms, known as instance-based models. The name comes from the fact that they are built around the notion of similarity between instances (distance) and the geometrical intuition behind it. As a practical application of our newly learned skills, we will build an app that recognizes types of user movements based on the data from motion sensors and learns completely on device (no Python this time).\n\nThe algorithms that we are discussing and implementing in this chapter are **k-nearest neighbors** ( **KNN** ) and **dynamic time warping** ( **DTW** ).\n\nIn this chapter, we will cover the following topics:\n\n  * Choosing a distance metric\u2014Euclidean, edit distance, taxicab, and DTW\n  * Building a KNN multiclass classifier\n  * Geometrical intuition behind machine learning models\n  * Reasoning in high-dimensional spaces\n  * Choosing hyperparameters\n\n# Calculating the distance\n\nHow do we calculate a distance? Well, that depends on the kind of problem. In two-dimensional space, we used to calculate the distance between two points, ( _x_ 1, _y_ 1) and ( _x_ 2, _y_ 2), as  \u2014the **Euclidean distance**. But this is not how taxi drivers calculate distance because in the city you can't cut corners and go straight to your goal. So, they use (knowing it or not) another distance metric: **Manhattan distance** or **taxicab distance** , also known as _l_ 1-norm:  . This is the distance if we're only allowed to move along coordinate axes:\n\nFigure 3.1: The blue line represents the Euclidean distance, the red line represents the Manhattan distance. Map of Manhattan by OpenStreetMap\n\nJewish German mathematician Hermann Minkowski proposed a generalization of both Euclidean and Manhattan distances. Here is the formula for the Minkowski distance:\n\nwhere _p_ and _q_ are _n_ -dimensional vectors (or coordinates of points in _n_ -dimensional space if you wish). But what does _c_ stand for? It is an order of the Minkowsi distance: under the _c = 1_ , it gives an equation of Manhattan distance, and under _c = 2_ it gives Euclidean distance.\n\nVector operations, including the calculation of Manhattan and Euclidean distances, can be parallelized for efficiency. Apple's Accelerate framework provides APIs for fast vector and matrix computations.\n\nIn machine learning, we generalize the notion of distance to any kind of objects for which we can calculate how similar they are, using a function: distance metric. In this way, we can define the distance between two pieces of text, two pictures, or two audio signals. Let's take a look at two examples.\n\nWhen you deal with two pieces of text of equal length, you use **edit distance** ; for example, **Hamming distance** \u2014the minimum number of substitutions needed to transform one string into another. To calculate the edit distance, we use dynamic programming, an iterative approach where the problem is broken into small subproblems, and the result of each step is remembered for future computations. Edit distance is an important measure in applications that deal with text revisions; for example, in bioinformatics (see the following diagram):\n\nFigure 3.2: Four pieces of DNA from different species aligned together: modern human, neanderthal, gorilla, and cat. The Hamming edit distance from modern human to others is 1, 5, and 11 respectively.\n\nOften, we store different signals (audio, motion data, and so on) as arrays of numbers. How do we measure the similarity of such two arrays? We use the combination of Euclidean distance and edit distance, called DTW.\n\n# DTW\n\nDespite its Sci-Fi name, DTW has little to do with time travel, except for the fact that this technique was popular for speech recognition back in the 1980s. Imagine two signals as two springs oriented along the time axis. We place them next to each other on the table, and want to measure how similar (or how different... what's the same?) they are. One of them will serve as a template. And we start stretching and compressing another one, piece by piece, until it looks exactly as the first one (or the most similar). Then we account for how much effort we put into align two springs\u2014we sum up all tensions and stretches together, and get the DTW distance.\n\nDTW distance between two sound signals tells us how similar to each other they are. For example, having the record of an unknown voice command, we can compare it to voice commands in the database, and find the most similar one. DTW can be used not only with audio, but with many other types of signals. We will use it to calculate distance between signals from motion sensors:\n\nFigure 3.3: DTW alignment of two accelerometer signals. On the left: walking sample against another walking sample. On the right: brushing teeth against walking. The shorter the alignment is, the closer the two signals are to one another. Plots created using [1] and [2].\n\nLet's demonstrate this with a simple example. Say we have two arrays: _[5, 2, 1, 3]_ and _[10, 2, 4, 3]_. How do we calculate the distance between two arrays of length one: _[5]_ and [10]? You can use squared difference as a measure; for example, _(5 - 10) 2 = 25_. Okay, now let's extend one of them: _[5, 2]_ and _[10],_ and calculate the cumulative difference: | **[5]** | **[2]**\n\n---|---|---\n\n**[10]** | _25_ | _25 + (2-10) 2 = 89_\n\nLet's extend another array to have _[5, 2]_ and _[10, 2]_. Now, how to calculate the cumulative difference is not as clear as it was before, but let's assume that we are interested in the simplest way to transform one array into another (minimal distance, in other words): | **[5]** | **[2]**\n\n---|---|---\n\n**[10]** | _25_ | _89_\n\n**[2]** | _25 + (5-2) 2 = 34_ | _min (25, 89, 34) + (2-2) 2 = 25_\n\nBy extending arrays in such a way further, eventually we will get the following table: | **[5]** | **[2]** | **[1]** | **[3]**\n\n---|---|---|---|---\n\n**[10]** | _25_ | _89_ | _89 + (1-10) 2 = 170_ | _170 + (3-10) 2 = 219_\n\n**[2]** | _34_ | _25_ | _min (89, 170, 25)_\n\n_\\+ (1-2) 2 = 26_ | _min (170, 219, 26)_\n\n_\\+ (3-2) 2 = 27_\n\n**[4]** | _34+(5-4) 2=35_ | _min (34, 25, 35)_\n\n_\\+ (2-4) 2 = 29_ | _min (25, 26, 29)_\n\n_\\+ (1-4) 2 = 34_ | _min (26, 27, 34)_\n\n_\\+ (3-4) 2 = 27_\n\n**[3]** | _35+(5-3) 2=39_ | _min (35, 29, 39)_\n\n_\\+ (2-3) 2 = 30_ | _min (29, 34, 30)_\n\n_\\+ (1-3) 2 = 33_ | _min (34, 27, 33)_\n\n_\\+ (3-3) 2 = 27_\n\nThe bottom-right cell of the table contains the quantity we're interested in: DTW distance between two arrays, the measure of how hard it is to transform one array into another. We've just checked all the possible ways to transform arrays, and found the easiest of them (marked with a gray shading in the table). Movement along the diagonal of the table indicates the perfect match between arrays, while horizontal direction stands for deletion of the elements from the first array, and vertical movement indicates insertion into it (compare with _Figure 3.3_ ). The final array alignment looks like this:\n\n_[5, 2, 1, 3, -]_\n\n_[10, 2, -, 4, 3]_\n\nBy the way, DTW can be applied not only to arrays of single numbers. Replace squared difference with Euclidean or Manhattan distance, and you can compare trajectories in a three-dimensional space or taxi routes.\n\n# Implementing DTW in Swift\n\nThere are two versions of the algorithm (with locality constraint, and without it). We'll implement both.\n\nThe full source code for the application we are developing in this chapter can be found in the `MotionClassification` folder of supplementary materials.\n\nLet's define a DTW structure, and create a static function `distance` in it:\n\n    func distance(sVec: [Double], tVec: [Double]) -> Double {\n\nFirst, we're creating a distance matrix of size _(n+1_ x _m+1)_ _,_ and populating it with some values: the first cell of the matrix should be equal to zero, and the first row and the first column should be equal to a maximum double value. This is needed to handle border conditions in a proper way later. The first cell plays a role of initial value: initially, the distance is zero. All other cells are unimportant for now, as we'll overwrite their values later:\n\n    let n = sVec.count \n    let m = tVec.count \n    var dtwMat = [[Double]](repeating: [Double](repeating: Double.greatestFiniteMagnitude, count: m+1), count: n+1) \n    dtwMat[0][0] = 0\n\nAfter this, we iterate through both arrays from _1_ to _n_ and _1_ to _m_ , filling the distance matrix. At each position _[i, j]_ , we calculate the cost for the previous position _(i-1, j-1)_ as the squared difference between corresponding positions in the arrays: _(s i-1 \\- tj-1)2_:\n\n    for i in 1...n { \n      for j in 1...m { \n        let cost = pow(sVec[i-1] - tVec[j-1], 2) \n        let insertion = dtwMat[i-1][j] \n        let deletion = dtwMat[i][j-1] \n        let match = dtwMat[i-1][j-1] \n        let prevMin = min(insertion, deletion, match) \n        dtwMat[i][j] = cost + prevMin \n      } \n    }\n\nThe value we are now looking for is in the last cell of the matrix: _dtw[n, m]_. To make the result comparable between series with different lengths, we normalize it by the length of the longest series:\n\n        return dtwMat[n][m]\/Double(max(n, m))  \n    }\n\nThis gives us an average distance between two series.\n\nTo avoid warping the whole sequence to the small segment of its counterpart, locality constraint was introduced. It sets the upper limit to how many deletions\/insertions can be found in a row.\n\nAnd a version of the algorithm with locality constraint `w`:\n\n    func distance(sVec: [Double], tVec: [Double], w: Int) -> Double { \n        let n = sVec.count \n        let m = tVec.count \n        var dtwMat = [[Double]](repeating: [Double](repeating: Double.greatestFiniteMagnitude, count: m+1), count: n+1) \n        dtwMat[0][0] = 0 \n        let constraint = max(w, abs(n-m))\n\n        for i in 1...n { \n            for j in max(1, i-constraint)...min(m, i+constraint) { \n                let cost = pow(sVec[i-1] - tVec[j-1], 2) \n                let insertion = dtwMat[i-1][j] \n                let deletion = dtwMat[i][j-1] \n                let match = dtwMat[i-1][j-1] \n                dtwMat[i][j] = cost + min(insertion, deletion, match) \n            } \n        } \n        return dtwMat[n][m]\/Double(max(n, m)) \n    }\n\nLet's test our algorithm. The first two vectors are similar :\n\n    let aVec: [Double] = [1,2,3,4,5,6,7,6,5,4,3,2,1] \n    let bVec: [Double] = [2,3,4,5,7,7,6,5,4,3,2,1,0,-2]\n\n    let distance1 = DTW.distance(sVec: aVec, tVec: bVec) \n    let distance2 = DTW.distance(sVec: aVec, tVec: bVec, w: 3)\n\nThe result is about 0.857 in both cases.\n\nNow we have two very different vectors:\n\n    let cVec: [Double] = [1,2,3,4,5,6,7,6,5,4,3,2,1,0] \n    let dVec: [Double] = [30,2,2,0,1,1,1,14,44]\n\n    let distance3 = DTW.distance(sVec: cVec, tVec: dVec) \n    let distance4 = DTW.distance(sVec: cVec, tVec: dVec, w: 3)\n\nThe results are 216.571 and 218.286 correspondingly. Note that the distance with locality constraint is even bigger than without it.\n\nOur implementation of DTW is na\u00efve, and can be accelerated using parallel computing. To calculate the new row\/column in a distance matrix, you don't need to wait until the previous one is finished; you only need it to be filled one cell ahead of your row\/column. DTW can be effectively parallelized using GPU. See _Accelerating Dynamic Time Warping Subsequence Search with GPUs and FPGAs_ for more details [3].\n\n# Using instance-based models for classification and clustering\n\nInstance-based machine learning algorithms are usually easy to understand as they have some geometrical intuition behind them. They can be used to perform different kinds of tasks, including classification, regression, clustering, and anomaly detection.\n\nIt's easy to confuse classification and clustering at first. Just to remind you, classification is one of the many types of supervised learning. The task is to predict some discrete label from the set of features ( _Figure 3.4_ , left pane). Technically, classification goes in two types: binary (check _yes_ or _no_ ), and multiclass ( _yes_ \/ _no_ \/ _maybe_ \/ _I don't know_ \/ _can you repeat the question?_ ). But in practice, you can always build a multiclass classifier from several binary classifiers.\n\nOn the other hand, clustering is the task of unsupervised learning. This means that, unlike classification, it knows nothing about data labels, and works out clusters of similar samples in your data on its own. In the next chapter, we are going to discuss an instance-based clustering algorithm called _k_ -means (KNN), and in this chapter, we focus on applications of instance-based algorithm KNN to multiclass classification:\n\nFigure 3.4: Classification process (on the left) and clustering (on the right). Classification consists of two steps: training with the labelled data and inference with unlabeled data. Clustering groups samples according to their similarity.\n\n# People motion recognition using inertial sensors\n\nWouldn't it be awesome at the end of every day to see the statistics of it: how much time have you spent doing things you like, and how much time you've wasted? With this kind of report, you could make your time management decisions based on real data, not just a gut feeling. Wait, but there are a lot of time trackers out there on the App Store, right? Sure, but there is one problem with most of them: you have to fill them in manually, because they can't detect what are you doing at every moment. You can't teach them to recognize types of your activities. Fortunately, we can fix this using machine learning; specifically, **time series classification**.\n\nTime series is a special kind of dataset in which samples are arranged according to the time. Usually, time series are generated when samples are taken repeatedly after equal time intervals (sampling interval). In other words, the time series is a sequence of values measured at successive moments in time, after regular intervals, and describing a process unrolling in a time dimension.\n\nTime series data type is common in iOS applications: among examples are signals from inertial sensors, measurements from HealthKit, and any other data that has a clear time correspondence and sampled regularly. Some other types of data, such as application logs or records of user activity, can be reduced to a special type of time series: categorical time series, where categories are in place of numbers.\n\nThe motion recognition task is important in health monitoring and fitness applications, but can also have some unusual use cases. For example, the _Walk Me Up! Alarm Clock_ app makes you get out of your bed, because it doesn't allow you to snooze your alarm until you take a number of steps. It distinguishes real steps from attempts to cheat by shaking the device.\n\nThe Core Motion framework provides APIs to get a history of a user's movements or a real-time stream of data from motion sensors. It can also distinguish a limited set of movement types, but we're going to teach our app to recognize more types than the Core Motion can. With the growth in popularity of wearable accessories, the motion sensor became a very common source of data; however, the method described in this chapter is not specific to sensor data, so you can apply these algorithms to many other practical problems. That's the good thing about general-purpose machine learning algorithms: you can apply them to any kind of data, you only need to find an appropriate representation for the data.\n\n# Understanding the KNN algorithm\n\nTo recognize different types of motion activities, we will train the KNN classifier. The idea of the method is to find _k_ training samples closest to the sample with an unknown label, and predict the label as a most frequent class among those _k_. That's it:\n\nFigure 3.5: KNN classification algorithm. The new data point marked with ? gets classified based on the classes of its neighbors.\n\nNote how the choice of neighbor number affects the result of classification.\n\nIn fact, the algorithm is so simple, that it's tempting to formulate it in more complicated terms. Let's do it. The secret sauce of a KNN is a distance metric: function, which defines how close to each other two samples are. We have discussed several of them already: Euclidean, Manhattan, Minkowski, edit distance, and DTW. Following the terminology, samples are points in some _n_ -dimensional space, where _n_ equals to the number of features in each sample. This space is called **feature space** , and samples are distributed in it as clouds of points. Classification of an unknown data point happens in three steps:\n\n  1. Calculate distances from the point to all points in a training set\n  2. Choose the _k-_ closest neighbors to the unknown point\n  3. Perform a majority vote among them\n\nThe surface that separates one class of points from another class is known as a **decision boundary**. The KNN algorithm creates piecewise linear decision boundaries that can approximate a decision boundary of any complexity by adding more and more training samples:\n\nFigure 3.6: Voronoi cells graph shows the closest neighbor at each point with a color. Depending on the distance metric you choose, the graph looks quite different. From the left to the right: Manhattan ( _c = 1_ ), Euclidean ( _c = 2_ ), and Minkowski ( _c = 3_ ) distance metrics.\n\nAlgorithms similar to KNN are also known as **non-generalizing machine learning**. In Chapter 6, _Linear Regression and Gradient Descent_ , we will discuss a linear regression, an algorithm that constructs general representation of all data points\u2014a straight line, because it assumes that all data points lie along the line. Unlike linear regression, KNN makes no assumption about the underlying structure of the data, it just stores all the training samples. Both approaches have their advantages and downsides.\n\nYou may think that this algorithm is too simple to be used for anything but some toy tasks. But over the years, KNN has demonstrated to be successfully employed for a wide range of problems, such as handwriting recognition, and satellite photo classification. It's also worth noting that it's easy to turn this classification algorithm into regression\u2014you just need to replace categorical labels with the real numbers, and add an interpolation function.\n\nParametric versus non-parametric models  \nMany restrictions of the linear regressions come from the fact that it assumes that data is normally distributed. The class of statistical models which makes explicit assumptions about the statistical distribution underlying data is called **parametric models**.\n\nUnlike linear regression, KNN makes no assumptions about the distribution from which samples are generated. That's why we call them **non-parametric**. This is the right tool to choose in situations where data has unusual distribution, and the decision boundary is irregular.\n\n# Implementing KNN in Swift\n\nFast implementations of KNN and DTW can be found in many machine learning and DSP libraries, for example `lbimproved` and `matchbox` C++ libraries:\n\n  * github.com\/lemire\/lbimproved\n  * github.com\/hfink\/matchbox\n\nThe KNN classifier works with virtually any type of data since you define distance metric for your data points. That's why we define it as a generic structure parameterized with types for features and labels. Labels should conform to a `Hashable` protocol, as we're going to use them for dictionary keys:\n\n    struct kNN<X, Y> where Y: Hashable { ... }\n\nKNN has two hyperparameters: _k_ \u2014the number of neighbors `var k: Int`, and distance metric. We'll define it elsewhere, and pass during the initialization. Metric is a function, returning double distance for any two samples `x1` and `x2`:\n\n    var distanceMetric: (_ x1: X, _ x2: X) -> Double\n\nDuring the initialization, we just record the hyperparameters inside our structure. The definition of `init` looks like this:\n\n    init (k: Int, distanceMetric: @escaping (_ x1: X, _ x2: X) -> Double) { \n        self.k = k \n        self.distanceMetric = distanceMetric \n    }\n\nKNN stores all its training data points. We are using the array of pairs _(features, label)_ for this purposes:\n\n    private var data: [(X, Y)] = []\n\nAs usual with supervised learning models, we'll stick to the interface with two methods, `train` and `predict`, which reflect the two phases of a supervised algorithm's life. The `train` method in the case of KNN just saves the data points to use them later in the `predict` method:\n\n    mutating func train(X: [X], y: [Y]) { \n        data.append(contentsOf: zip(X, y)) \n    }\n\nThe `predict` method takes the data point and predicts the label for it:\n\n    func predict(x: X) -> Y? {  \n        assert(data.count > 0, \"Please, use method train() at first to provide training data.\") \n        assert(k > 0, \"Error, k must be greater then 0.\")\n\nFor this, we iterate through all samples in the training dataset, and compare them with the input sample `x`. We use _(distance, label)_ tuples to keep track of distances to each of the training samples. After this, we sort all the samples descending by distances, and take the (`prefix`) first `k` elements:\n\n        let tuples = data \n            .map { (distanceMetric(x, $0.0), $0.1) }  \n            .sorted { $0.0 < $1.0 }  \n            .prefix(upTo: k)\n\nThis implementation is not optimal, and can be improved by keeping track of only the best `k` samples at each step, but the goal of it is to demonstrate the simplest machine learning algorithm without diving into the complex data structures, and show that even such na\u00efve versions of it can perform well on complex tasks.\n\nNow we arrange majority voting among top `k` samples. We count the frequency of each label, and sort them from descending:\n\n    let countedSet = NSCountedSet(array: tuples.map{$0.1}) \n    let result = countedSet.allObjects.sorted { \n        countedSet.count(for: $0) > countedSet.count(for: $1) \n        }.first \n    return result as? Y \n    }\n\nThe `result` variable holds a predicted class label.\n\n# Recognizing human motion using KNN\n\nCore Motion is an iOS framework that provides an API for inertial sensors of mobile devices. It also recognizes some user motion types, and stores them to the HealthKit database.\n\nIf you are not familiar with Core Motion API, please check the framework reference: <https:\/\/developer.apple.com\/reference\/coremotion>.\n\nThe code for this example can be found in the `Code\/02DistanceBased\/ MotionClassification` folder of supplementary materials.\n\nAs per iOS 11 beta 2, the `CMMotionActivity` class includes the following types of motion:\n\n  * Stationary\n  * Walking\n  * Running\n  * Automotive\n  * Cycling\n\nEverything else falls into an unknown category or is recognized as one of the preceding. Core Motion doesn't provide a way to recognize custom motion types so we'll train our own classifier for this purpose. Unlike decision trees from the previous chapter, KNN will be trained on device end-to-end. It will also not be frozen inside Core ML because as we keep all the control on it, we'll be able to update it in the application runtime.\n\niOS devices have three types of motion sensors:\n\n  * **Gyroscope** : This measures device orientation in space\n  * **Accelerometer** : This measures device acceleration\n  * **Magnetometer or compass** : This measures magnetism\n\nThey also have a barometer to detect elevation and some other sensors, but they are less relevant for our purposes. We will use an accelerometer data stream to train our KNN classifier and predict different motion types, like shaking a phone or squatting.\n\nThe following listing shows how to get updates from the accelerometer:\n\n    let manager = CMMotionManager() \n    manager.accelerometerUpdateInterval = 0.1 \n    manager.startAccelerometerUpdates(to: OperationQueue.main) { (data: CMAccelerometerData?, error: Error?) in \n        if let acceleration = data?.acceleration { \n            print(acceleration.x, acceleration.y, acceleration.z) \n        } \n    }\n\nThe accelerometer APIs in Core Motion provide a time series of three-dimensional vectors, as shown in the following diagram:\n\nFigure 3.7: Core Motion coordinate system for accelerometer and gyroscope\n\nTo train our classifier, we need some labeled data. As we don't have a ready dataset and motion signals can be very different from person to person, we are going to allow the user to add new samples and improve the model. In the interface, the user selects the type of motion he wants to record, and presses the Record button, as shown in the next screenshot. The application samples 25 acceleration vectors, takes the magnitude of each vector, and feeds them with the label of the selected motion type into the KNN classifier. The user records as many samples as he wants.\n\n# Cold start problem\n\nA very common situation is when a machine learning system starts functioning in a new environment, where no information to pre-train is available. The situation is known as a **cold start**. Such a system requires a certain amount of time to collect enough training data, and start producing meaningful predictions. The problem often arises in the context of personalization and recommender systems.\n\nOne solution for this it is so-called **active learning** , where the system can actively seek new data that could improve its performance. Usually, this means that the system queries a user to label some data. For instance, the user can be asked to provide some labeled examples before the start of the system, or the system can ping him when it stumbles upon especially hard cases asking to label them manually. Active learning is a special case of semi-supervised learning.\n\nThe second component of active learning is estimating which samples are the most useful by associating weights to them. In the case of KNN, these can be the samples that the model is less confident about, for example, the samples for whom their neighbors' classes are divided almost equally or the samples that are far from all others (outliers).\n\nHowever, some researchers point out that active learning is built on flawed assumptions: the user is always available and willing to answer questions and he is always right in his\/her answers. This is also something worth keeping in mind when building an active learning solution.\n\nI guess when the Twitter app pings you at 4 AM with push notifications like _Take a look at this and 13 other Highlights_ , it just wants to update its small personalized binary classifier of _interesting_ \/ _not interesting_ content using active learning.\n\nFigure 3.8: App interface\n\nIn the classification phase, we feed unlabeled chunks of the same size into the classifier and get predictions which display to the user. We use DTW as a distance measure with locality constraint `3`. In my experiments, `k` as `1` gave the best results but you can experiment with other number of neighbors. I will show here only the machine learning part, without the data collection part and user interface.\n\nCreating the classifier:\n\n    classifier = kNN(k: 1, distanceMetric: DTW.distance(w: 3))\n\nTraining the classifier:\n\n    self.classifier.train(X: [magnitude(series3D: series)], y: [motionType])\n\nThe `magnitude()` function converts three-dimensional series into one-dimensional by calculating vector magnitude   to simplify the computations.\n\nMaking the predictions:\n\n    let motionType = self.classifier.predict(x: magnitude(series3D: series)) \n\n# Balanced dataset\n\nThe application allows you to record samples of different motion types. As you train the model, you may notice one interesting effect: to get accurate predictions, you need not only enough samples, but you also need the proportion of different classes in your dataset to be roughly equal. Think about it: if you have 100 samples of two classes (`walk` and `run`), and 99 of them belong to one class (`walk`), the classifier that delivers 99% accuracy may look like this:\n\n    func predict(x: [Double]) -> MotionType { \n       return .walk \n    }\n\nBut this is not what we want, obviously.\n\nThis observation lead us to the notion of the balanced data set; for most machine learning algorithms, you want the data set in which samples of different classes are represented equally frequently.\n\n# Choosing a good k\n\nIt is important to pick a proper value of hyperparameter _k_ , since it can improve a model's performance as well as degrade it when chosen incorrectly. One popular rule of thumb is to take a square root of the number of training samples. Many popular software packages use this heuristic as a default _k_ value. Unfortunately, this doesn't always work well, because of the differences in the data and distance metrics.\n\nThere is no mathematically-grounded way to come up with the optimal number of neighbors from the very beginning. The only option is to scan through a range of _k_ s, and choose the best one according to some performance metric. You can use any performance metric that we've already described in the previous chapter: accuracy, _F1_ , and so on. The cross-validation is especially useful when the data is scarce.\n\nIn fact, there is a variation of KNN, which doesn't require _k_ at all. The idea is to make the algorithm take the radius of a ball to search the neighbors within. The _k_ will be different for each point then, depending on the local density of points. This variation of the algorithm is known as **radius-based neighbor learning**. It suffers from the _n_ -ball volume problem (see next section), because the more features you have, the bigger the radius should be to catch at least one neighbor.\n\n# Reasoning in high-dimensional spaces\n\nWorking with feature spaces of high dimensions requires special mental precautions, since our intuition used to deal with three-dimensional space starts to fail. For example, let's look at one peculiar property of _n_ -dimensional spaces, known as an _n_ -ball volume problem. _N_ -ball is just a ball in _n_ -dimensional Euclidean space. If we plot the volume of such _n_ -ball ( _y_ axis) as a function of a number of dimensions ( _x_ axis), we'll see the following graph:\n\nFigure 3.9: Volume of _n_ -ball in _n_ -dimensional space\n\nNote that at the beginning the volume rises, until it reaches its peak in five-dimensional space, and then starts decreasing. What does it mean for our models? Specifically, for KNN, it means that starting from five features, the more features you have the greater should be the radius of the sphere centered on the point you're trying to classify to cover KNN.\n\nThe counter-intuitive phenomena that arise in a high-dimensional space are colloquially known as the **curse of dimensionality**. This includes a wide range of phenomena that can't be observed in the three-dimensional space we used to deal with. Pedro Domingos, in his _A Few Useful Things to Know about Machine Learning_ , provides some examples:\n\n\"In high dimensions, most of the mass of a multivariate Gaussian distribution is not near the mean, but in an increasingly distant shell around it; and most of the volume of a high-dimensional orange is in the skin, not the pulp. If a constant number of examples is distributed uniformly in a high-dimensional hypercube, beyond some dimensionality most examples are closer to a face of the hypercube than to their nearest neighbor. And if we approximate a hypersphere by inscribing it in a hypercube, in high dimensions almost all the volume of the hypercube is outside the hypersphere. This is bad news for machine learning, where shapes of one type are often approximated by shapes of another.\"\n\nSpeaking specifically of KNN, it treats all dimensions as equally important. This creates problems when some of the features are irrelevant, especially in high dimensions, because the noise introduced by these irrelevant features suppresses the signal comprised in the good features. In our example, we bypassed multidimensional problems by taking into account only the magnitude of each three-dimensional vector in our motion signals.\n\n# KNN pros\n\n  * It's simple to implement if you are not going for optimized versions which use advanced data structures.\n  * It's easy to understand and interpret. The algorithm is well studied theoretically, and much known about its mathematical properties in different settings.\n  * You can plug in any distance metric. This allows working with complex objects, like time series, graphs, geographical coordinates, and basically anything you can define distance metric for.\n  * Algorithms can be used for classification, ranking, regression (using neighbors average or weighted average), recommendations, and can even provide (a kind of) probabilistic output\u2014what proportion of neighbors voted for this class.\n  * It's easy to incorporate new data in the model or remove outdated data from it. This makes KNN a good choice for online learning (see Chapter 1, _Getting Started with Machine Learning_ ) systems.\n\n# KNN cons\n\n  * The algorithm is fast for training but slow for inference.\n  * You need to choose the best _k_ somehow (see _Choosing a good k_ section).\n  * With the small values of _k_ , the model can be badly affected by outliers; in other words, it's prone to overfitting.\n  * You need to choose a distance metric. For usual real value features, one can choose among many available options (see _Calculating the distance_ section) resulting in different closest neighbors. The metric used by default in many machine learning packages is the Euclidean distance; however, this choice is nothing more than a tradition and for many applications is not the optimal.\n  * Model size grows with the new data incorporated.\n  * What should we do if there are several identical samples with different labels? In this case, the result can be different depending on the order in which samples are stored.\n  * The model suffers from the curse of dimensionality.\n\n# Improving our solution\n\nThere are several directions in which we can proceed to improve our algorithm for motion recognition.\n\n# Probabilistic interpretation\n\nThe `CMMotionActivity` class provides a confidence level for each predicted motion type. We can also add this feature to our algorithm. Instead of returning one label, we can return the proportion of labels among neighbors.\n\n# More data sources\n\nWe've used only accelerometer, but we could use gyroscope and magnetometer also. This can be done in several ways: you can just merge three time series into one three-dimensional time series or you can train an ensemble of three independent classifiers.\n\nWe've also merged _x_ , _y_ , and _z_ of accelerometer into one magnitude value, but you can try to use them as separate time series. In this case, for three motion sensors, you'd have nine time series.\n\n# Smarter time series chunking\n\nWe split our time series into chunks of 25 elements length. This introduces delay when the motion type changes from one to another. This can also be fixed relatively easily by introducing sliding windows instead of chunks. With this approach, we don't need to wait for the new chunk to be delivered; we just record a frame or predict a new label every time when we get a new value from the motion sensor.\n\n# Hardware acceleration\n\nThe KNN algorithm is inherently parallel because to calculate the distance between two data points you don't need to know anything about other data points. This makes it a perfect candidate for GPU acceleration. DTW, as we've mentioned, can also be optimized for parallel execution.\n\n# Trees to speed up the inference\n\nAn array is not the only possible candidate for KNN's memory implementation. To make neighbors search faster, many implementations use special data structures such as a KD tree or a ball tree.\n\nCheck scikit-learn documentation if you're interested in more details: <http:\/\/scikit-learn.org\/stable\/modules\/neighbors.html#nearest-neighbor-algorithms>.\n\n# Utilizing state transitions\n\nTransitions between some motion types are more likely than between others: it's easy to imagine how a user can start walking after being still, but it's much harder to imagine how he could start running immediately after squatting. The popular way of modelling such probabilistic state changes is **hidden Markov model** ( **HMM** ), but that's a long story for some other time.\n\n# Summary\n\nIn this chapter, we implemented a working machine learning solution for motion data classification and trained it end-to-end on a device. The simplest of the instance-based models is the nearest neighbors classifier. You can use it to classify any type of data, the only tricky thing is to choose a suitable distance metric. For feature vectors (points in _n_ -dimensional space), many metrics have been invented, such as the Euclidean and Manhattan distances. For strings, editing distances are popular. For time series, we applied DTW.\n\nThe nearest neighbors method is a non-parametric model, which means that we can apply it without regard to statistical data distributions. Another advantage is that it is well suited for online learning and is easy to parallelize. Among the shortcomings is the curse of dimensionality and the algorithmic complexity of predictions (lazy learning).\n\nIn the next chapter, we're going to proceed with instance-based algorithms, this time focusing on the unsupervised clustering task.\n\n# Bibliography\n\n  1. Lichman, M. (2013), UCI Machine Learning Repository (<http:\/\/archive.ics.uci.edu\/ml>), Irvine, CA: University of California, School of Information and Computer Science, _Dataset for ADL Recognition with Wrist-worn Accelerometer Data Set_\n  2. Toni Giorgino (2009), _Computing and Visualizing Dynamic Time Warping Alignments in R: The dtw Package,_ Journal of Statistical Software, 31(7), 1-24, doi:10.18637\/jss.v031.i07\n\n  3. _Accelerating Dynamic Time Warping Subsequence Search with GPUs and FPGAs_ , Doruk Sart, Abdullah Mueen, Walid Najjar, Vit Niennattrakul, Eamonn Keogh, in the Proceedings of IEEE ICDM 2010. pp. 1001-1006 at: <http:\/\/alumni.cs.ucr.edu\/~mueen\/pdf\/icdm2010.pdf>\n  4. Domingos P. 2012, _A Few Useful Things to Know about Machine Learning_ , Communications of the ACM, October, 55(10), pp. 78-87\n\n# K-Means Clustering\n\nIn this chapter, we're going to switch our attention from supervised learning to unsupervised learning. The algorithms that we'll discuss and implement in this chapter are k-means and k-means++ clustering.\n\nIn this chapter, we will cover the following topics:\n\n  * Instance-based algorithm of k-means clustering\n  * The shortcomings of the k-means and how to fix them with the k-means++\n  * Where you can use k-means and where you shouldn't use it\n  * Application of clustering for signal quantization\n  * How to choose the number of clusters\n\n# Unsupervised learning\n\nUnsupervised learning is a way of making hidden patterns in data visible:\n\n  * Clustering finds groups or hierarchy of similar objects\n  * Unsupervised anomaly detection finds outliers (weird samples)\n  * Dimensionality reduction finds which details of data are the most important\n  * Factor analysis reveals the latent variables that influence the behavior of the observed variables\n  * Rule mining finds associations between different entities in the data\n\nAs usually, these tasks overlap pretty often, and many practical problems inhabit the neutral territory between supervised and unsupervised learning.\n\nWe will focus on clustering in this chapter and on rule mining in the next chapter. Others will remain mostly beyond the scope of this book, but in Chapter 10, _Natural Language Processing_ , we will nevertheless briefly discuss autoencoders; they can be used for both dimensionality reduction and anomaly detection.\n\nHere are some examples of real-world tasks where clustering would be your tool of choice:\n\n  * Cluster face photos by identity of a person depicted\n  * Find groups of customers for a targeted advertisement using the database of their transactions (market segmentation)\n  * Having a set of text documents, sort them into a folder according to the personal style (stylometry) of their author (authorship attribution) or according to their topics (topic modelling)\n  * Having DNA markers of relatives, build a phylogenetic or family tree (hierarchical clustering; clusters are nested in this case)\n\nNote, that these are clustering tasks only as long as groups\/categories\/clusters are not predefined in advance. As soon as you have predefined classes of objects, you would be better off with the classification algorithms.\n\nWhere might clustering be needed in the context of mobile development? Clustering pins on a map may look like the most natural idea. Having clusters of a user's locations, you can guess the location of his important locations, like house and workplace, for example. We will start from this and later discuss more complex applications of clustering. For now, we will concentrate on the classical clustering algorithm: _k_ -means.\n\n# K-means clustering\n\nThe name of this algorithm comes from the _k_ clusters into which the samples are divided, and the fact that each cluster is grouped around some mean value, a **centroid** of a cluster. This centroid serves as a prototype of a class. Each data point belongs to the cluster which centroid is the closest.\n\nThe algorithm was invented in 1957 at Bell Labs.\n\nIn this algorithm, each data point belongs to only one cluster. As a result of this algorithm, we get the feature space partitioned into Voronoi cells.\n\nBecause of the _k_ in its name, this algorithm is often confused with the KNN algorithm, but as we already have seen with _k_ -fold cross-validation, not all _k_ s are the same. You may wonder why machine learning people are so obsessed with this letter that they put it in every algorithm's name. I don't _k_ -now.\n\nFigure 4.1: Four different ways to cluster the same data using _k_ -means algorithm. Bald black dots are centroids of clusters. The samples are from the classical Iris dataset, plotted petals length against petal width.\n\nLet's define the algorithm's aim more formally. If _n_ is the number of your data points (samples, represented as real vectors of length _d_ ), then k-means algorithm splits them into _k_ sets (clusters, _k_ < _n_ ), such that within each cluster sum of distances from the points to the center (mean) is minimal. In other words, the objective of the algorithm is to find a set of clusters with the minimal WCSS:\n\nWhere:\n\n  * _k_ is a number of clusters\n  * _S_ _i_ clusters, _i_ =1, 2, ..., _k_ ,\n  * _x j_ is a sample (vector),\n  * _\u03bc i_ is a mean of samples in the cluster , in other words\u2014centroid of the cluster.\n\nFor the beginning we usually initialize centroids at random or with the values of some random samples from the dataset. Algorithm is iterative and each iteration consists of two steps:\n\n  1. Calculate the centroids for each cluster\n  2. Rreassign samples to clusters, according to the closest centroids\n\nAlgorithm ends, when after some iteration the coordinates of centroids haven't changed (convergence achieved) or after some predefined number of steps.\n\n# Implementing k-means in Swift\n\nSimilar to the KNN from the previous chapter, we'll have a structure to represent an algorithm and keep all its hyperparameters:\n\n    struct KMeans { \n      public let k: Int\n\nThe standard k-means algorithm was designed to be used only with Euclidean distance:\n\n    internal let distanceMetric = Euclidean.distance\n\nWe need several arrays to store different kinds of data during the clustering.\n\nStorage for samples:\n\n    internal var data: [[Double]] = []\n\nCoordinates of centroids:\n\n    public var centroids: [[Double]] = []\n\nAn array that matches each sample to its cluster. It should be of the same length as the data, and for every sample, it stores an index of centroid in the `centroids` array:\n\n    private(set) var clusters: [Int] = []\n\n**Within-cluster sum of squares** is a measure that we'll use later to assess the quality of the result:\n\n    internal var WCSS: Double = 0.0\n\nFor now, the only parameter that we pass on the initialization is the number of clusters:\n\n    public init (k: Int) { \n      self.k = k \n    }  \n    }\n\nUnlike KNN, k-means has only one method in its interface: `train(data:)`, which returns the results of clustering, the index of cluster each sample belongs to:\n\n    public mutating func train(data: [[Double]]) -> [Int] {\n\nBefore starting actual calculations, there are several inevitable ceremonies to perform.\n\nThe count of data points should be greater or equal to `k`, and the number of samples (`n`) should be greater than zero:\n\n    let n = data.count     \n    precondition(k <= n) \n    precondition(n > 0)\n\nCalculate the dimensionality of samples (number of features in each sample) and check that it is greater than zero:\n\n    let d = data.first!.count \n    precondition(d > 0)\n\nIf everything is fine, store the data:\n\n    self.data = data\n\nIf the number of clusters is equal to the number of data points, then we can just create a cluster for each data point and return the result:\n\n    if k == n { \n      centroids = data \n      clusters = Array<Int>(0..<k) \n      return clusters \n    }\n\nPopulating the `clusters` array with zeros:\n\n    clusters = [Int](repeating: 0, count: n)\n\nThe important detail of the k-means is an initial choice of centroids. The result of the algorithm may be significantly different, depending on the initialization step. Let's not overthink it for now, and go on with the random initialization:\n\n    chooseCentroidsAtRandom() \/\/ The function body will be described later.\n\nThe main part of the algorithm is a while loop, which breaks when the convergence conditions are met:\n\n    while true {\n\nInside those loops, the algorithm consists of two steps: update step, and assignment step.\n\n**Within-cluster sum of squares** ( **WCSS)** is an important performance measure for k-means. We set it to be zero on every iteration start and as it changes every time, we update the cluster centroids:\n\n    WCSS = 0.0 \n\n# Update step\n\nAfter the initial choice of centroids, we need to iterate over data points and update information about cluster assignment according to the distance to the closest centroid:\n\n    for (pointIndex, point) in data.enumerated() { \n      var minDistance = Double.infinity \n      for (clusterID, centroid) in centroids.enumerated() { \n        let distance = pow(distanceMetric(point, centroid), 2)\n\nRemember the newly calculated distance if it is less than the previously saved minimum distance for the corresponding data point:\n\n        if minDistance > distance { \n          clusters[pointIndex] = clusterID \n          minDistance = distance \n        } \n      }\n\nSave information about WCSS for the future:\n\n      WCSS += minDistance \n    } \n\n# Assignment step\n\nCalculate new centroids of clusters:\n\n    var centroidsCount = [Double](repeating: 0.0, count: k) \n    let rowStub = [Double](repeating: 0.0, count: d) \n    var centroidsCumulative = [[Double]](repeating: rowStub, count: k)\n\n    for (point, clusterID) in zip(data, clusters) { \n      centroidsCount[clusterID] += 1 \n      centroidsCumulative[clusterID] = vecAdd(centroidsCumulative[clusterID], point) \n    }\n\n    var newCentroids = centroidsCumulative \n    for (j, row) in centroidsCumulative.enumerated() { \n      for (i, element) in row.enumerated() { \n        let new = element\/centroidsCount[j] \n        assert(!new.isNaN) \n        newCentroids[j][i] = new \n      } \n    }\n\nAfter this, we have to check whether the new centroids are different from those previously calculated. If they are different, we perform another iteration of optimization, if not, we've reached the convergence and can break the loop:\n\n    var convergence = false \n    convergence = zip(centroids, newCentroids).map{$0.0 == $0.1}.reduce(true, and)  \n    \/\/ and(_: Bool, _:Bool) was added for convenience \n    if convergence { break } \n    centroids = newCentroids \n    }\n\n    return clusters \n    }\n\nDo you remember, we've skipped the cluster centroid's initialization implementation? So, here it goes:\n\n    internal mutating func chooseCentroidsAtRandom() { \n      let uniformWeights = [Double](repeating: 1.0, count: data.count) \n      let randomIndexesNoReplacement = Random.Weighted.indicesNoReplace(weights:uniformWeights, count: k)\n\n      var centroidID = 0 \n      for index in randomIndexesNoReplacement { \n        centroids.append(data[index]) \n        clusters[index] = centroidID \n        centroidID += 1 \n      } \n    }\n\nThis `Random.Weighted.indicesNoReplace(weights:uniformWeights, count: k)` looks mysterious, but it's just a utility function for random sampling with predefined weights from an array. It samples without replacement and returns an array of indices. In this case, all weights are equal so the probability of each element to be sampled is equal. Later, we'll change this to improve the quality of clustering and speed of convergence. I ported this function from the R standard library.\n\n# Clustering objects on a map\n\nWhere can we apply k-means in the context of mobile development? Clustering pins on a map may look like the most natural idea. Having the clusters of user locations, you can guess the location of the user's important locations like home and workplace, for example. We will implement pin clustering to visualize k-means, some of its unfortunate properties, and show why such an application of it may be not the best idea.\n\nYou can find a demo application under the `4_kmeans\/MapKMeans` folder of supplementary code. Everything interesting happens in the `ViewController.swift`. Clustering happens in the `clusterize()` method:\n\n    func clusterize() { \n      let k = Settings.k \n      colors = (0..<k).map{_ in Random.Uniform.randomColor()} \n      let data = savedAnnotations.map{ [Double]($0.coordinate) } \n      var kMeans = KMeans(k: k) \n      clusters = kMeans.train(data: data) \n      centroidAnnotations = kMeans.centroids \n        .map { CLLocationCoordinate2D(latitude: $0[0], longitude: $0[1]) } \n        .map { coordinate in \n          let annotation = MKPointAnnotation() \n          annotation.coordinate = coordinate \n          annotation.title = \"(coordinate)\" \n          return annotation \n        } \n    }\n\n`CLLocationCoordinate2D` gets converted to the `Double` in the following way:\n\n    extension Array where Element == Double { \n      init(_ coordinates: CLLocationCoordinate2D) { \n        self.init(arrayLiteral: Double(coordinates.latitude), Double(coordinates.longitude)) \n      } \n    }\n\nThe app shows the map and you can put in as many pins as you wish by tapping on it. Try to form several clusters and run the algorithm by pressing the circular arrow button several times. Note that the results of the algorithm are usually unstable: it can give different results every time. Check how the result differs when you chose another number of clusters.\n\nFigure 4.2: The results of the demo app work. Combine this with locations from the photo library metadata and you can guess where the user lives, works and spends his\/her free time.\n\nNow to the problems with using k-means with geospatial data. You may notice that it looks like algorithm treats horizontal and vertical space unequally important. That's because one degree of longitude is always about 111 km, but one degree of latitude has a different length depending on how far from the equator you are. Basically, you should not use geographical coordinates as features for Euclidean distance. More to the point, if you put some points on the border between Alaska and Russia, you would observe that KNN treats points in Russia as very different from points close to Alaska. This is due to the 180\u00b0 Meridian. Basically, we can say that using k-means for pin clustering was a mistake. It could work on some relatively small scale. for example, it could work fine on a city scale. Just don't use it on the world map scale. Maps are just ideal for vanilla k-means problem demonstrations.\n\nBut wait. Cut can't we replace the Euclidean distance with some other distance metric, like we did with KNN in the previous chapter? No, unfortunately not this time. Strictly speaking, k-means is not a distance-based algorithm. The objective of the algorithm is to minimize the within-cluster variance (or squared error). The formula of a variance happened to be identical to the formula of Euclidean distance. This makes it impossible to plug in to your custom distance metrics because the algorithm can stop converge. For pin clustering, it's better to choose some other algorithm, and for k-means it's better to choose some other application, and indeed we will. But first, let's discuss some problems that became apparent after our first experiment:\n\n  * <https:\/\/datascience.stackexchange.com\/questions\/761\/clustering-geo-location-coordinates-lat-long-pairs>\n  * <https:\/\/stats.stackexchange.com\/questions\/81481\/why-does-k-means-clustering-algorithm-use-only-euclidean-distance-metric>\n\n# Choosing the number of clusters\n\nIf you don't know in advance how many clusters you have, then how do you choose the optimal _k_? This is essentially an egg-and-chicken problem. Several approaches are popular and we'll discuss one of them: the elbow method.\n\nDo you remember those mysterious WCSS that we calculated on every iteration of k-means? This measure tells us how much points in every cluster are different from their centroid. We can calculate it for several different _k_ values and plot the result. It usually looks somewhat similar to the plot on the following graph:\n\nFigure 4.3: WCSS plotted against the number of clusters\n\nThis plot should remind you about the similar plots of loss functions from Chapter 3, _K-Nearest Neighbors Classifier_. It shows how well our model fits the data. The idea of the elbow method is to choose the _k_ value after which the result is not going to improve sharply anymore. The name comes from the similarity of the plot to an arm. We choose the point at the elbow, marked with the red line on the graph.\n\nFor more information refer to the following links:\n\n  * <https:\/\/en.wikipedia.org\/wiki\/Determining_the_number_of_clusters_in_a_data_set>\n  * <http:\/\/stackoverflow.com\/questions\/18042290\/implementing-the-elbow-method-for-finding-the-optimum-number-of-clusters-for-k-m>\n\n# K-means clustering \u2013 problems\n\nRefer to the following for more information about k-means and k-means++:\n\n  * <https:\/\/en.wikipedia.org\/wiki\/K-means_clustering>\n  * <https:\/\/en.wikipedia.org\/wiki\/K-means%2B%2B>\n\nK-means algorithm suffers from at least two shortcomings:\n\n  * The worst-case time complexity of the algorithm is super polynomial in the input size, meaning that it is not bounded above by any polynomial\n  * Standard algorithm can perform arbitrarily poor in comparison to the optimal clustering because it finds only an approximation of the real optimum\n\nTry it out yourself: put four pins on a map, as shown in the following image. After running clustering several times, you may notice that the algorithm often converges to the suboptimal solution:\n\nFigure 4.4: Optimal and non-optimal clustering results on the same dataset\n\n# K-means++\n\nAn improved algorithm was proposed in 2007. K-means++ addresses the problem of suboptimal clustering by introducing an additional step for a good centroids initialization.\n\nAn improved algorithm of initial centers selection looks like this:\n\n  1. Select randomly any data point to be the first center\n  2. For all other data points, calculate the distance to the first center _d_ ( _x_ )\n  3. Sample the next center from the weighted probability distribution, where the probability of each data point to become a next center is proportional to the square of distance _d_ ( _x_ )2\n  4. Until _k_ centers are chosen, repeat step 2 and step 3\n  5. Proceed with the standard k-means algorithm\n\nIn Swift, it looks like this:\n\n    internal mutating func chooseCentroids() { \n      let n = data.count\n\n      var minDistances = [Double](repeating: Double.infinity, count: n) \n      var centerIndices = [Int]()\n\n`clusterID` is an integer identifier of a cluster: the first cluster has identifier zero, the second has one, and so on:\n\n    for clusterID in 0 ..< k { \n      var pointIndex: Int \n      if clusterID == 0 {\n\nChoose the first centroid randomly from data points:\n\n    pointIndex = Random.Uniform.int(n) \n    } else {\n\nIn all other cases, choose center from the weighted distribution, proportionally to the squared distance to the closest centroid:\n\n    if let nextCenter = Random.Weighted.indicesNoReplace(weights: minDistances, count: 1).first { \n      pointIndex = nextCenter \n    } else { \n      fatalError() \n    } \n    } \n    centerIndices.append(pointIndex) \n    let center = data[pointIndex] \n    centroids.append(center)\n\nThe distance to the closest center is zero. Hence, the probability of sampling once again is also zero:\n\n    minDistances[pointIndex] = 0.0 \n    clusters[pointIndex] = clusterID\n\nAfter this, we have to perform one iteration of the assign step so that all points are assigned to the corresponding clusters when we proceed with the usual k-means algorithm.\n\nCalculate the distance from each of the data points to the centroid:\n\n    var nextI = (0, centerIndices.first ?? Int.max) \n    for (pointIndex, point) in data.enumerated() {\n\nSkip the data point if it was selected as a center already:\n\n    if pointIndex == nextI.1 {\n\nCheck if all centroids were attended:\n\n    if nextI.0 < clusterID { \n      let nextIndex = nextI.0+1 \n      nextI = (nextIndex, centerIndices[nextIndex]) \n    } \n    continue \n    }\n\nIf the data point is not selected as a center yet, calculate the distance from it to the last selected center:\n\n    let distance = pow(distanceMetric(point, center), 2)\n\nRemember the newly calculated distance if it is less than the minimum distance saved for the corresponding data point previously:\n\n    let currentMin = minDistances[pointIndex] \n    if currentMin > distance { \n      minDistances[pointIndex] = distance \n      clusters[pointIndex] = clusterID \n    } \n    } \n    } \n    }\n\nThat's it. Now don't forget to update the rest of the code to work with the ++ part:\n\n    public struct KMeans { \n      public enum InitializationMethod { \n        case random \n        case plusplus \n      } \n      ... \n      public var initialization: InitializationMethod = .plusplus \n      ... \n    }\n\n    public mutating func train(data: [[Double]]) -> [Int] { \n      ... \n      switch initialization { \n        case .random: \n        chooseCentroidsAtRandom() \n        case .plusplus: \n        chooseCentroids() \n      } \n      ... \n    } \n\n# Image segmentation using k-means\n\nThe k-means algorithm was invented in the field of digital signal processing and is still in common use in that field for signal quantization. For this task, it performs much better than for pin clustering. Let's look at an example on the following diagram. The picture can be segmented into meaningful parts using color space quantization. We choose the number of clusters, then run k-means on every pixel's RGB values, and find the cluster's centroids. Then we replace each pixel with the color of its corresponding centroid. This can be used in image editing for separating objects from the background or for lossy image compression. In Chapter 12, _Optimizing Neural Networks for Mobile Devices_ , we're going to use this approach for deep learning neural network compression:\n\nFigure 4.5: Image segmentation using k-means\n\nHere is a code sample in Objective-C++ using fast OpenCV implementation of k-means. You can find the whole iOS application in the folder `4_kmeans\/ImageSegmentation`:\n\n    - (cv::Mat)kMeansClustering:(cv::Mat)input withK:(int)k { \n      cv::cvtColor(input, input, CV_RGBA2RGB); \n      cv::Mat samples(input.rows * input.cols, 3, CV_32F);\n\n      for (int y = 0; y < input.rows; y++){ \n        for (int x = 0; x < input.cols; x++){ \n          for (int z = 0; z < 3; z++){ \n            samples.at<float>(y + x*input.rows, z) = input.at<cv::Vec3b>(y,x)[z]; \n          } \n        } \n      }\n\n      int clusterCount = k; \n      cv::Mat labels; \n      int attempts = 5; \n      cv::Mat centers; \n      kmeans(samples, clusterCount, labels, cv::TermCriteria(CV_TERMCRIT_ITER|CV_TERMCRIT_EPS, 100, 0.01), attempts, cv::KMEANS_PP_CENTERS, centers);\n\n      cv::Mat outputMatrix( input.rows, input.cols, input.type());\n\n      for (int y = 0; y < input.rows; y++) { \n        for (int x = 0; x < input.cols; x++) { \n          int cluster_idx = labels.at<int>(y + x*input.rows,0); \n          outputMatrix.at<cv::Vec3b>(y,x)[0] = centers.at<float>(cluster_idx, 0); \n          outputMatrix.at<cv::Vec3b>(y,x)[1] = centers.at<float>(cluster_idx, 1); \n          outputMatrix.at<cv::Vec3b>(y,x)[2] = centers.at<float>(cluster_idx, 2); \n        } \n      }\n\n      return outputMatrix; \n    } \n\n# Summary\n\nIn this chapter, we've discussed an important unsupervised learning task: clustering. The simplest clustering algorithm is k-means. It doesn't provide stable results and is computationally complex, but this can be improved using k-means++. The algorithm can be applied to any data for which Euclidean distance is a meaningful measure, but the best area to apply it is a signal quantization. For instance, we've used it for image segmentation. Many more clustering algorithms exist for different types of tasks.\n\nIn the next chapter, we're going to explore unsupervised learning more deeply. Specifically, we're going to talk about algorithms for finding association rules in data: association learning.\n\n# Association Rule Learning\n\nIn many practical applications data comes in the form of lists (ordered or unordered): grocery lists, playlists, visited locations or URLs, app logs, and so on. Sometimes those lists are generated as a byproduct of business processes, but they still contain potentially useful information and insights for process improvement. To extract some of that hidden knowledge, one can use a special kind of unsupervised learning algorithm\u2014association rule mining. In this chapter, we are going to build an app that can analyze your shopping lists to find out your preferences in the form of rules such as \" _If you've bought oatmeal and cornflakes, you also want to buy milk_.\" This can be used to create an adaptable user experience, for instance, contextual suggestions or reminders.\n\nIn this chapter, we will cover the following topics:\n\n  * Association rules\n  * Association measures\n  * Association rule mining algorithms\n  * Building an adaptable user experience\n\n# Seeing association rules\n\nThere are many situations where we're interested in patterns demonstrating the co-occurrence of some items. For example, marketers want to know which goods are often bought together, clinical personnel need to know symptoms associated with certain medical conditions, and in information security we want to know which activity patterns are associated with intrusion or fraud. All of these problems have a common structure: there are items (goods, symptoms, records in logs) organized in transactions (shopping list, medical case, user activity transaction). With this type of data, we can then analyze it to find association rules, such as _If the client bought a lemon and_ _some cookies, he is also likely to buy tea,_ or in more formal notation: (cookies, lemon \u2192 tea).\n\nWe will use pictograms throughout this chapter to facilitate the visual notation of item sets and rules: {    \u2192  }.\n\nThese rules allow us to make informed decisions, such as putting associated items on the same shelf, providing patients with the appropriate care, and alerting security staff if suspicious activity is suspected in a system. The unsupervised learning algorithms that find these rules are known as **association rule mining** or **association rule learning** algorithms. These are considered a type of unsupervised learning, as you do not need labeled data to generate a rule.\n\nAssociation rule learning is not the type of algorithm that one typically sees in an introductory-level book about machine learning. This is perhaps due to their relatively narrow use case. However, in the following sections, we will see how rule learning can become the engine of an adaptable user interface, and used in other important applications. After this, we hope you will agree that the power of these methods has been underestimated.\n\n# Defining data structures\n\nWhat we want to have by the end of this chapter is a rule learning algorithm called Apriori. We will learn about the algorithm details later; for now, we only want to define the data structures that we will work with throughout the chapter, along with some utility functions.\n\nThe generic structure for the algorithm is as follows:\n\n    public struct Apriori<Item: Hashable & Equatable> {\n\nIn the simplest case, the ordering of the items in the transaction doesn't matter, and neither does their number nor the associated timestamps. This means that we consider our item sets and transactions as mathematical or Swift sets:\n\n    public typealias ItemSet = Set<Item>\n\nThe parameter `I` is a type of item in your transactions. Next, we have to implement some structures for subsets and rules:\n\n    class Subsets: Sequence {  \n      var subsets: [ItemSet]  \n      init(_ set: ItemSet) {  \n        self.subsets = Array(set).combinations().map(Set.init)  \n      }  \n      func makeIterator() -> AnyIterator<ItemSet> {  \n        return AnyIterator { [weak self] in  \n        guard let `self` = self else {  \n          return nil  \n        }  \n        return self.subsets.popLast()  \n      }  \n    }  \n    public struct Rule {  \n      let ifPart: Set<I>  \n      let thenPart: Set<I>  \n    }\n\nApriori's structure variables are as follows:\n\n    public var elements: Set<Int> \n    public let transactions: ContiguousArray<ItemSet> \n    public let map: [I: Int] \n    public let invertedMap: [Int: I]\n\nSupports are stored here to prevent multiple computations:\n\n    public convenience init(transactions: [[I]]) { \n        self.init(transactions: transactions.map(Set<I>.init)) \n    }\n\n    public init(transactions: [Set<I>]) { \n        \/\/ delete\n\n        var indexedTransactions = [ItemSet]() \n        var counter = 0 \n        var map = [I: Int]() \n        var invertedMap = [Int: I]()\n\n        for transaction in transactions { \n            var indexedTransaction = ItemSet() \n            for item in transaction { \n                if let stored = map[item] { \n                    indexedTransaction.insert(stored) \n                } else { \n                    map[item] = counter \n                    invertedMap[counter] = item \n                    indexedTransaction.insert(counter) \n                    counter += 1 \n                } \n            } \n            indexedTransactions.append(indexedTransaction) \n        }\n\n        self.transactions = ContiguousArray(indexedTransactions) \n        self.elements = self.transactions.reduce(Set<Int>()) {$0.union($1)} \n        self.map = map \n        self.invertedMap = invertedMap\n\n        self.total = Double(self.transactions.count) \n    } \n\n# Using association measures to assess rules\n\nLook at these two rules:\n\n  * {Oatmeal, corn flakes \u2192 Milk}\n  * {Dog food, paperclips \u2192 Washing powder}\n\nIntuitively, the second rule looks more unlikely than the first one, doesn't it? How can we tell that for sure, though? In this case, we need some quantitative measures that will show us how likely each rule is. What we are looking for here are association measures, as we call them in machine learning and data mining. Rule mining algorithms revolve around this notion in a similar manner to how distance-based algorithms revolve around distance metrics. In this chapter, we're going to use four association measures: support, confidence, lift, and conviction (see _Table 5.1_ ).\n\nNote that these measures tell us nothing about how useful or interesting the rules are, but only quantify their probabilistic characteristics. A rule's usefulness and practicality can be hard to grasp mathematically and often requires human judgment in each case. As usual in statistics, interpreting analysis results is something left to the discretion of a domain expert or developer.\n\n# Supporting association measures\n\nLet's say that we have the following six shopping lists (six transactions) composed of only four items: a hot dog, tomatoes, tea, and cookies. This is our database:\n\n{    }\n\n{   }\n\n{   }\n\n{  }\n\n{  }\n\n{ }\n\nWe say that the item set {  } covers the transactions 1, 3, and 5, because the item set is a subset of each of those transactions. There are _2 n = 24 = 16_ possible item sets in our example, including the empty item set.\n\nThe support of the item set shows how often this set occurs as part of a transaction; in other words, what proportion of transactions is covered in this item set. For example:\n\nSupport for an empty item set is assumed to be equal to the number of transactions in the dataset ( _supp({}) = 6_ , in our case). If you represent all the item sets as a graph (see _Figure 5.1_ ), you may notice that support always decreases as the length of item sets grows. When mining for association rules, we are usually interested in the larger item sets that have support greater than a given threshold; for instance, for the support threshold 0.5, such item sets are {  }, { }, and { }. In other words, this means that each of the item sets cover at least half of all transactions.\n\nHere, we are extracting association measures to the separate structure extension for convenience:\n\n    public extension Apriori { \n      public mutating func support(_ set: ItemSet) -> Double {\n\nWe store support values that we've already calculated because they don't change during algorithm running, and we will be able to prevent the repetition of a costly operation. On the other hand, however, this solution increases the memory footprint:\n\n    if let stored = supports[set] { \n      return stored \n    }\n\n    let support = transactions.filter{set.isSubset(of: $0)}.count \n    let total = transactions.count\n\n    let result = Double(support)\/Double(total) \n    supports[set] = result\n\n    return result \n    }\n\nFigure 5.1: The graph of item sets for the transactions depicted in the top-left corner. The number of bold sides for each item set depicts the support value (for instance, a triangle means that support = 3). The width of the incoming edges for each node is proportional to its support. Note how the support monotonically decreases from the top (6) to the bottom (1) as the item set size grows\n\n# Confidence association measures\n\nThe confidence association measures shows how likely it is for an item to occur in a transaction including other items:\n\nNote that confidence can't be greater than _1_. The problem with this measure is that it doesn't take into account the general support of the item; if { } is common in the dataset, then it's likely that it will occur in the transaction independently of any associations.\n\nNote that:\n\nIn Swift:\n\n    public mutating func confidence(_ rule: Rule) -> Double { \n        return support(rule.ifPart.union(rule.thenPart))\/support(rule.ifPart) \n    } \n\n# Lift association measures\n\nLift is confidence normalized by the support of the item of interest, shown as follows:\n\nLift takes into consideration the support of both sets {  } and { }. The _lift_ > _1_ shows that items are associated positively, meaning that items after the arrow are likely to be bought if items before the arrow are present. The _lift_ < _1_ implies a negative association; in our case, this means that if the customer has taken already bought a hot dog and tomatoes it's unlikely he will add tea to the basket. _Lift_ = _1_ means no association at all.\n\nUnlike confidence:\n\nIn Swift:\n\n    public mutating func lift(_ rule: Rule) -> Double { \n        return support(rule.ifPart.union(rule.thenPart))\/support(rule.ifPart)\/support(rule.thenPart) \n    } \n\n# Conviction association measures\n\nConviction is a measure that helps to judge if the rule happened to be there by chance or not. It was introduced by Sergey Brin and coauthors in 1997 [1] as a replacement for confidence, which can't capture the direction of an association. Conviction is a comparison of the probability of _if_ appearing without _then,_ if they were dependent on the actual frequency of _if_ without _then:_\n\nIn the nominator, we have the expected frequency of item sets without { }. (In other words, how often the rule doesn't hold true.) In the denominator, we have an observed frequency of false predictions. In our example, it shows that the rule {  \u2192  } holds true approximately 67% more often ( _1.667_ as often) if the association between { } and { } was by chance.\n\n**Association measure** | **Formula** | **Range** | **Notes**\n\n---|---|---|---\n\nSupport | _supp(X) = P(X +)_\n\n_supp (X \u222a Y) = P(X + \u2229 Y+)_ | [0, 1] | How often does _X_ occur in the dataset? How often do _X_ and _Y_ occur in the dataset together?\n\nConfidence |\n\n  | [0, 1] | Given the presence of _X_ , what is the probability that _Y_ is also present?\n\nLift |   | [0, \u221e] | _1_ means independence between _X_ and _Y_ , because for independent events _P(A \u2229 B) = P(A)P(B)_.\n\nConviction |   | [0, \u221e] | _1_ means independence, \u221e means always true.\n\nTable 5.1: Common association measures\n\n_X_ and _Y_ stand for the item sets themselves, and they stand for the events of their presence in a transaction, so _P(X \\+ \u2229 Y+)_ denotes the probability that both _X_ and _Y_ are present in a transaction.\n\nFor the comprehensive list of association measures used in rule learning with explanations, formulas, and references, see _A Probabilistic Comparison of Commonly Used Interest Measures for Association Rules_ by Michael Hahsler at http:\/\/michael.hahsler.net\/research\/association_rules\/measures.html.\n\n# Decomposing the problem\n\nThe task of extracting all association rules with the given confidence and support from the dataset is non-trivial. Let's approach it by decomposing it into smaller subtasks:\n\n  * Find all item sets with the support above the given threshold\n  * Generate all possible rules from the item sets that have confidence above the given threshold\n\n# Generating all possible rules\n\nWe need a method to generate all possible combinations of elements of this array. Combinations are found via the binary representation of subsets, as shown in the following snippet:\n\n    public extension Array { \n    public func combinations() -> [[Element]] { \n            if isEmpty { return [] } \n            let numberOfSubsets = Int(pow(2, Double(count))) \n            var result = [[Element]]() \n            for i in 0..<numberOfSubsets { \n                var remainder = i \n                var index = 0 \n                var combination = [Element]() \n                while remainder > 0 { \n                    if remainder % 2 == 1 { \n                        combination.append(self[index]) \n                    } \n                    index += 1 \n                    remainder \/= 2 \n                } \n                result.append(combination) \n            } \n            return result \n        } \n    }\n\nThe following usage example:\n\n    let array = [1,2,3] \n    print(array.combinations())\n\nProduces:\n\n    **[[], [1], [2], [1, 2], [3], [1, 3], [2, 3], [1, 2, 3]]** \n\n# Finding frequent item sets\n\nThe first step of the algorithm that we implement is based on the support measure. This function returns a set of all item sets with support larger than `minSupport`:\n\n    func frequentItemSets(minSupport: Double) -> Set<ItemSet> {\n\n        var itemSets = Set<ItemSet>()\n\n        let emptyItemSet: ItemSet = ItemSet()\n\n        supporters[emptyItemSet] = Array(0 ..< transactions.count)\n\nHere we use the priority queue data structure to keep track of possible extensions.\n\nThere is no priority queue implementation in the Foundation or Swift standard libraries, and standard data structures are out of the scope of this book. We are using the open source implementation by David Kopec (MIT license): <https:\/\/github.com\/davecom\/SwiftPriorityQueue>.\n\nTo make it work with item sets we had to change the code a bit\u2014instead of being parameterized with the comparable types, it is now parameterized with types conforming to the equatable protocol:\n\n        var queue = PriorityQueue<ItemSet>(order: { (lh, rh) -> Bool in \n            lh.count > rh.count \n        }, startingValues: [emptyItemSet])\n\n        while let itemset = queue.pop() { \n            var isMax = true\n\n            for anExtension in allExtensions(itemset) { \n                if isAboveSupportThreshold(anExtension, extending: itemset, threshold: minSupport) { \n                    isMax = false \n                    queue.push(anExtension) \n                } \n            } \n            if isMax == true { \n                itemSets.insert(itemset) \n            } \n        }\n\n        return itemSets \n    }\n\nNote that this algorithm has one bad characteristic: it generates the same item sets multiple times. We'll return to this later.\n\n# The Apriori algorithm\n\nThe most famous algorithm for association rule learning is Apriori. It was proposed by Agrawal and Srikant in 1994. The input of the algorithm is a dataset of transactions where each transaction is a set of items. The output is a collection of association rules for which support and confidence are greater than some specified threshold. The name comes from the Latin phrase _a priori_ (literally, \"from what is before\") because of one smart observation behind the algorithm: _if the item set is infrequent, then we can be sure in advance that all its subsets are also infrequent._\n\nYou can implement Apriori with the following steps:\n\n  1. Count the support of all item sets of length 1, or calculate the frequency of every item in the dataset.\n  2. Drop the item sets that have support lower than the threshold.\n  3. Store all the remaining item sets.\n  4. Extend each stored item set by one element with all possible extensions. This step is known as candidate generation.\n  5. Calculate the support value of each candidate.\n  6. Drop all candidates below the threshold.\n  7. Drop all stored items from step 3 that have the same support as their extensions.\n  8. Add all the remaining candidates to storage.\n  9. Repeat steps 4 to step 8 until there are no more extensions with support greater than the threshold.\n\nThis is not a very efficient algorithm if you have a lot of data, but mobile applications are not recommended for use with big data anyway. This algorithm was influential in its time, and is also elegant and easy to understand today.\n\nIf you want to extract rules from your data as part of your server-side data processing pipeline, you may want to check Apriori's implementation in the `mlxtend` Python library: <http:\/\/rasbt.github.io\/mlxtend\/user_guide\/frequent_patterns\/apriori\/>.\n\nFigure 5.2: By excluding only one node we can reduce the number of possible rules twice. By excluding two nodes, we reduce the hypothesis space four times\n\n# Implementing Apriori in Swift\n\nWhat will follow is a simplified version of a code that can be found in supplementary materials. We will skip some of the less important parts here.\n\nThe main method, which returns association rules with the given support and confidence, is as follows:\n\n    public func associationRules(minSupport: Double, minConfidence: Double) -> [Rule] { \n        var rules = [Rule]() \n        let frequent = frequentItemSets(minSupport: minSupport)\n\n        for itemSet in frequent { \n            for (ifPart, thenPart) in nonOverlappingSubsetPairs(itemSet) { \n                if confidence(ifPart, thenPart) >= minConfidence { \n                    let rule = Rule(ifPart: convertIndexesToItems(ifPart), thenPart: convertIndexesToItems(thenPart)) \n                    rules.append(rule) \n                } \n            } \n        }\n\n        return rules \n    } \n    func nonOverlappingSubsetPairs(_ itemSet: ItemSet) -> [(ItemSet, ItemSet)] { \n        var result = [(ItemSet, ItemSet)]() \n        let ifParts = Subsets(itemSet) \n        for ifPart in ifParts { \n            let nonOverlapping = itemSet.subtracting(ifPart) \n            let thenParts = Subsets(nonOverlapping) \n            for thenPart in thenParts { \n                result.append((ifPart, thenPart)) \n            } \n        } \n        return result \n    } \n\n# Running Apriori\n\nAnd finally, this is how we use the algorithm with our toy example:\n\n    let transactions = [[\"![](assets\/0c8a58ea-7c29-4474-adf0-1e07d1976d70.png)\", \"![](assets\/03daed71-d8e7-478a-9bdf-08b84cb517a7.png)\", \"![](assets\/051951f4-7bc3-4fdf-884b-db711923f62a.png)\", \"![](assets\/2705c197-5814-4394-add9-50d4c19f03ae.png)\"], \n                        [\"![](assets\/0c8a58ea-7c29-4474-adf0-1e07d1976d70.png)\", \"![](assets\/03daed71-d8e7-478a-9bdf-08b84cb517a7.png)\", \"![](assets\/2705c197-5814-4394-add9-50d4c19f03ae.png)\"], \n                        [\"![](assets\/0c8a58ea-7c29-4474-adf0-1e07d1976d70.png)\", \"![](assets\/03daed71-d8e7-478a-9bdf-08b84cb517a7.png)\", \"![](assets\/051951f4-7bc3-4fdf-884b-db711923f62a.png)\"], \n                        [\"![](assets\/0c8a58ea-7c29-4474-adf0-1e07d1976d70.png)\", \"![](assets\/03daed71-d8e7-478a-9bdf-08b84cb517a7.png)\"], \n                        [\"![](assets\/051951f4-7bc3-4fdf-884b-db711923f62a.png)\", \"![](assets\/2705c197-5814-4394-add9-50d4c19f03ae.png)\"], \n                        [\"![](assets\/051951f4-7bc3-4fdf-884b-db711923f62a.png)\", \"![](assets\/2705c197-5814-4394-add9-50d4c19f03ae.png)\"], \n                        [\"![](assets\/0c8a58ea-7c29-4474-adf0-1e07d1976d70.png)\"] \n    ]\n\n    let apriori = Apriori<String>(transactions: transactions) \n    let rules = apriori.associationRules(minSupport: 0.3, minConfidence: 0.5) \n    for rule in rules { \n        print(rule) \n        print(\"Confidence: \", apriori.confidence(rule), \"Lift: \", apriori.lift(rule), \"Conviction: \", apriori.conviction(rule)) \n    }\n\nIt produces the following:\n\n    { ![](assets\/0c8a58ea-7c29-4474-adf0-1e07d1976d70.png) \u2192 ![](assets\/03daed71-d8e7-478a-9bdf-08b84cb517a7.png)} \n    Confidence:  0.8 Lift:  1.4 Conviction:  2.14285714285714 \n    { ![](assets\/03daed71-d8e7-478a-9bdf-08b84cb517a7.png) \u2192 ![](assets\/0c8a58ea-7c29-4474-adf0-1e07d1976d70.png)} \n    Confidence:  1.0 Lift:  1.4 Conviction:  inf \n    { ![](assets\/051951f4-7bc3-4fdf-884b-db711923f62a.png) \u2192 ![](assets\/2705c197-5814-4394-add9-50d4c19f03ae.png)} \n    Confidence:  0.75 Lift:  1.3125 Conviction:  1.71428571428571 \n    { ![](assets\/2705c197-5814-4394-add9-50d4c19f03ae.png) \u2192 ![](assets\/051951f4-7bc3-4fdf-884b-db711923f62a.png)} \n    Confidence:  0.75 Lift:  1.3125 Conviction:  1.71428571428571\n\nLet's analyze what's going on here. The second rule has the maximum confidence as well as conviction.\n\n# Running Apriori on real-world data\n\nIn this example, we collected real-world shopping lists from an apartment and composed a small, but nevertheless realistic, dataset. Let's see if we'll be able to extract any meaningful rules from it using our algorithm. Please note that this dataset is extremely small. For any production application of Apriori, you will need much larger datasets:\n\n    let transactions =  \n    [[\"Grapes\", \"Cheese\"], \n    [\"Cheese\", \"Milk\"], \n    [\"Apples\", \"Oranges\", \"Cheese\", \"Gingerbread\", \"Marshmallows\", \"Eggs\", \"Canned vegetables\"], \n    [\"Tea\", \"Apples\", \"Bagels\", \"Marshmallows\", \"Icecream\", \"Canned vegetables\"], \n    [\"Cheese\", \"Buckwheat\", \"Cookies\", \"Oatmeal\", \"Banana\", \"Butter\", \"Bread\", \"Apples\", \"Baby puree\"], \n    [\"Baby puree\", \"Cookies\"], \n    [\"Cookies\"], \n    [\"Chicken\", \"Grapes\", \"Pizza\", \"Cheese\", \"Marshmallows\", \"Cream\"], \n    [\"Potatoes\"], \n    [\"Chicken\"], \n    [\"\u0421ornflakes\", \"Cookies\", \"Oatmeal\"], \n    [\"Tea\"], \n    [\"Chicken\"], \n    [\"Chicken\", \"Eggs\", \"Cheese\", \"Oatmeal\", \"Bell pepper\", \"Bread\", \"Chocolate butter\", \"Buckwheat\", \"Tea\", \"Rice\", \"Corn\", \"\u0421ornflakes\", \"Juice\", \"Sugar\"], \n    [\"Bread\", \"Canned vegetables\"], \n    [\"Carrot\", \"Beetroot\", \"Apples\", \"Sugar\", \"Buckwheat\", \"Rice\", \"Pasta\", \"Salt\", \"Rice flour\", \"Dates\", \"Tea\", \"Butter\", \"Beef\", \"Cheese\", \"Eggs\", \"Bread\", \"Cookies\"] \n    ]\n\nAfter some experimentation with the threshold values, you can see that we ended up with 0.15 for support and 0.75 for confidence. This gives the 15 rules you can see in the following table:\n\n    let apriori = Apriori<String>(transactions: transactions) \n    let rules = apriori.associationRules(minSupport: 0.15, minConfidence: 0.75)\n\nThe resulting rules are sorted according to their lift:\n\n**Rules** | **Confidence** | **Lift** | **Conviction**\n\n---|---|---|---\n\n{Cheese, bread \u2192 Buckwheat} | 1 | 5.333333333 | \u221e\n\n{Buckwheat \u2192 Cheese, bread} | 1 | 5.333333333 | \u221e\n\n{Cheese, buckwheat \u2192 Bread} | 1 | 4 | \u221e\n\n{Buckwheat \u2192 Bread} | 1 | 4 | \u221e\n\n{Bread \u2192 Cheese, buckwheat} | 0.75 | 4 | 3.25\n\n{Bread \u2192 Buckwheat} | 0.75 | 4 | 3.25\n\n{Eggs \u2192 Cheese} | 1 | 2.285714286 | \u221e\n\n{Buckwheat, bread \u2192 Cheese} | 1 | 2.285714286 | \u221e\n\n{Buckwheat \u2192 Cheese} | 1 | 2.285714286 | \u221e\n\n{Bread \u2192 Cheese} | 0.75 | 1.714285714 | 2.25\n\n{Apples \u2192 Cheese} | 0.75 | 1.714285714 | 2.25\n\nTable 5.2: 15 rules\n\nBuckwheat is a type of cereal popular in Eastern Europe, Western Asia, and elsewhere. People usually eat buckwheat porridge with butter and bread (but not in Poland). In our case, however, it looks as though we prefer it with cheese rather than butter, which is not completely true. Seven out of the 11 rules recommend that we buy cheese, which is not surprising as it's the most common item across all transactions. The remaining four rules point at an association between bread and buckwheat, which is not a fluke because here in Ukraine many people consume these products together, and so the rules are valid. What is important here is that the algorithm is able to extract patterns that correspond to real-world phenomena: user preferences, cultural traditions, and so on.\n\n# The pros and cons of Apriori\n\nThe pros of Apriori are as follows:\n\n  * This is the most simple and easy-to-understand algorithm among association rule learning algorithms\n  * The resulting rules are intuitive and easy to communicate to an end user\n  * It doesn't require labeled data as it is fully unsupervised; as a result, you can use it in many different situations because unlabeled data is often more accessible\n  * Many extensions were proposed for different use cases based on this implementation\u2014for example, there are association learning algorithms that take into account the ordering of items, their number, and associated timestamps\n  * The algorithm is exhaustive, so it finds all the rules with the specified support and confidence\n\nThe cons of Apriori are as follows:\n\n  * If the dataset is small, the algorithm can find many false associations that happened simply by chance. You can address this issue by evaluating obtained rules on the held-out test data for the support, confidence, lift, and conviction values.\n  * As Agrawal and Srikant note at the end of their original paper, the algorithm doesn't take into account hierarchies of products or quantities of the items bought in a transaction. This additional information, while useful for market basket analysis, may be irrelevant in other rule mining domains and is out of the scope of this particular machine learning approach.\n  * The algorithm is computationally expensive, but there are many variants of Apriori that improve its algorithmic complexity.\n\n# Building an adaptable user experience\n\nHuman-computer interaction is never easy. Computers don't understand speech, sentiments, or body language. However, we are all used to communicating with our smart devices using not-so-smart buttons, drop downs, pickers, switches, checkboxes, sliders, and hundreds of other controls. They comply with a new kind of language that is commonly referred to as UI. Slowly but unavoidably, machine learning has made its way into all areas where computers interact directly with humans: voice input, handwriting input, lip reading, gesture recognition, body pose estimation, face emotion recognition, sentiment analysis, and so on. This may not be immediately obvious, but machine learning is the future of both UI and UX. Today, machine learning is already changing the way users interact with their devices. Machine learning-based solutions are likely to become widely-adopted in UIs because of their convenience. Furthermore, ranking, contextual suggestions, automatic translation, and personalization are also elements that most internet users have got used to. A nice example of how machine learning may fuel the UI is with the Facebook app, which runs on-device machine learning (even offline) to automatically sort posts in your timeline.\n\nIn the design community, such patterns of user interactions are commonly referred to as **anticipatory** or **algorithmic design** and are often described as **new trends** or **black magic**. Essentially, all the examples of anticipatory design that you may see in blogs or in presentations are cases of machine learning (just don't tell the design folks that). Machine learning can not only fuel big data analytics, but also small UI tweaks such as moving buttons on a screen or guessing what a user wants to do next and helping them with that. Such things when properly designed and tested can make your apps more enjoyable and easy to use. The main goal of such design patterns is to free the user from cognitive load when he or she uses the app. When you start using a new app, it's often like being in a new environment: you learn where different objects are placed, where to go to find whatever you want, and where short paths and pitfalls are located. By allowing the computer to also learn from the user, we can adjust the UX to make the user's learning curve steeper.\n\nLaura Busche explains this concept in her blog _What You Need To Know About Anticipatory Design_ at Smashing magazine particularly well:\n\n\"In psychology, we use the term cognitive load to describe the amount of mental effort being used in the working memory at any given moment in time. For everyone involved in user experience design, cognitive load is a crucial consideration. Are we doing everything in our power to relieve the strain caused by learning something new to use our product? How can we reduce the number of elements that our users need to worry about at any given time? Reducing cognitive load is one of the cornerstones of anticipatory design, as it helps create a more pleasurable experience by foreseeing our users' needs.\"\n\nThe main idea behind anticipatory design is the principle of having fewer choices \u2014 in other words, we reduce the number of options for the user in an intelligent way.\n\nIn practice, you can do this in plenty of different ways depending on your app; you can filter out irrelevant results, push the most likely options to the top of the list or increase their size, and so on.\n\nGoing back to the topic of this chapter, you can use association rule learning to analyze user activity in your app and narrow the space of possibilities for them. For example, in a photo editing app, a user applies a set of filters to their photo. When a user has chosen the first filter, rules can be used to predict which filter (or even which set of filters) are most likely to be applied next. Here, you can sort the candidates according to one of the association measures. In Chapter 7, _Linear Classifier and Logistic Regression_ , you will see another example of anticipatory design that will be based on supervised learning.\n\n# Summary\n\nIn this chapter, we explored association rule learning, which is a branch of unsupervised learning. We implemented the Apriori algorithm, which can be used to find patterns in the form of rules in different transactional datasets. Apriori's classical use case is market basket analysis. However, it is also important conceptually, because rule learning algorithms bridge the gap between classical artificial intelligence approaches (logical programming, concept learning, searching graphs, and so on) and logic-based machine learning (decision trees).\n\nIn the following chapter, we're going to return to supervised learning, but this time we will switch our attention from non-parametric models, such as KNN and k-means, to parametric linear models. We will also discuss linear regression and the gradient descent optimization method.\n\n# Bibliography\n\n  1. Sergey Brin, Rajeev Motwani, Jeffrey D. Ullman, and Shalom Tsur, _Dynamic itemset counting and implication rules for market basket data_ , in SIGMOD 1997, Proceedings ACM SIGMOD international conference on Management of data, pages 255-264, Tucson, Arizona, USA, May 1997\n  2. Rakesh Agrawal and Ramakrishnan Srikant, _Fast Algorithms for Mining Association Rules_ , Proceedings of the 20th international conference on very large databases, VLDB, pages 487-499, Santiago, Chile, September 1994 at: <http:\/\/www.vldb.org\/conf\/1994\/P487.PDF>\n\n# Linear Regression and Gradient Descent\n\nIn the previous chapters, we've implemented non-parametric models including kNN and _k_ -means and their applications to supervised classification and unsupervised clustering. In this chapter, we will proceed with the supervised learning by discussing algorithms for regression, this time focusing on the parametric models. **Linear regression** is the simple yet powerful tool for this kind of task. Linear regression was historically the first machine learning algorithm, so the math behind it is well developed, and you can find many books dedicated to this one topic exclusively. We will see when to use linear regression and when not to, how to analyze its errors, and how to interpret its results. As for the Swift part, we will get our feet wet with Apple's numerical libraries\u2014the **Accelerate framework**.\n\nLinear regression will serve as an example to explain an important mathematical optimization technique, **gradient descent**. This iterative algorithm will haunt us until the book's very end, because it is heavily used for training artificial neural networks.\n\nThe algorithms to be discussed and implemented in this chapter are:\n\n  * Simple linear regression for datasets with one feature\n  * Multiple linear regression for datasets with more than one feature\n  * Gradient descent algorithm\n  * Data normalization\n\n# Understanding the regression task\n\nRecall that the regression task is of a particular case of supervised learning, where real numbers take the place of labels. It is the primary difference from the classification, where all labels are categories. You can use regression analysis to study the interactions between two or more variables; for example, the way personal computer price depends on the computer's characteristics, such as a number of CPU cores and the type, memory size, video card characteristics, and storage type and size. In the context of regression, we usually call features _independent variables_ and labels _dependent variables_. In our example, independent variables are the computer's characteristics and the dependent variable is its price. Having a regression model, we can predict which machine is better to buy. Moreover, regression allows you to make educated guesses about the contribution of each feature to the final price. Could be an idea for the next viral app.\n\nRegression analysis is a subfield of statistics that investigates how dependent variable changes depend on changes of an independent variable. It also can be used for determining which independent variables are essential and which are not. In some cases, regression analysis can even be used to infer a causal relationship between variables.\n\nDifferent regression algorithms are implemented in several Swift libraries: AIToolbox, MLKit, multilinear-math, and YCML.\n\n# Introducing simple linear regression\n\nLinear regression is a kind of steampunk machine learning. It was invented in the time of Sherlock Holmes, long before the first electronic computer was invented and the term _machine learning_ was coined. The term _regression_ and its calculation algorithm was introduced by the English polymath Sir Francis Galton in 1886, in the publication named _Regression towards Mediocrity in Hereditary Stature_. Galton proposed the concept while performing research on how to create the perfect breed of people. The task of regression emerged from the need to predict the child's body parameters given the parent's body measurements. So nowadays, Sir Galton is mostly remembered as the father of eugenics rather than as an inventor of the first machine learning algorithm. Later in this chapter, we will follow the footsteps of Galton (but not too far), and employ the linear regression to predict some biological data. Linear regression often is the best choice of machine learning algorithm for fitness apps. You can model all kinds of simple dependencies using it: muscle growth depending on drills; weight loss depending on calorie intake; and so on:\n\nFigure 6.1: Linear regression terminology\n\nFrom the preceding diagram, you can find the following:\n\n  * **x** : Independent variable or feature\n  * **y** : Dependent variable or targets\n  *  : Predicted values of dependent variable\n  * **yi** : True value of dependent variable for a given data point\n  * ** i** : Predicted value of dependent variable for a given data point\n  * **a** : Slope\u2014rate of predicted changes for **y** scores for each unit increase in **x**\n  * **b** : **y** -intercept\u2014predicted value of **y** , when **x** is zero\n  * **\u03b5 1**: Residual (error) for a given data point\n\nThe idea of linear regression is quite simple. As we remember from the previous chapter, the model in the supervised learning is a mathematical function _f_ ( _x_ ), which predicts output label _y_ (height of sons, in Galton's research) for the input feature _x_ (height of parents). Perhaps, every reader keeps fond school memories about a simple formula for a straight line: _y = ax + b_ , where coefficient _a_ regulates the slope of a line, and the b term is a _y_ -intercept. Given that dependency between _x_ and _y_ is linear, we can suppose that the function underlying our dataset is _y i = axi \\+ b + \u03b5i_, where \u03b5 stands for errors (in measurements or of any other kind). The straight line plays the role of a model (or hypothesis function, _h_ ( _x_ )) in task. For now, let's focus on the situation when _y_ is determined by only one feature _x_. This kind of regression is called **simple linear regression**. To start predicting something, we need the parameters _a_ and _b_. In fact, the goal of the learning process is to choose the best parameters for our model, such that the line fits the dataset in the best possible way. In other words, the best parameters allow making the most accurate predictions. To tell accurate predictions apart from inaccurate, we use another function: loss (or what's the same, cost) function. Sir Francis Galton used the least squares method to estimate the model parameters.\n\nThe linear regression algorithm greatly relies on linear algebra. To implement it in Swift, we are using Accelerate framework. Don't forget to import it:\n\n    **import Accelerate**\n\nAccelerate framework **  \n**Accelerate contains low-level functions optimized for maximum performance on Apple hardware. vDSP sublibrary contains functions for vector operations and digital signal processing. We will go into the details of the Accelerate and other low-level numerical libraries inChapter 11, _Machine Learning Libraries_ , For now, all you need to know is that it's fast and it's low-level.\n\nFirst, let's create a class `SimpleLinearRegression`. It contains two double variables: model parameters `slope` ( _a_ ) and `intercept` ( _b_ ):\n\n    class SimpleLinearRegression { \n    var slope = 1.0 \n    var intercept = 0.0 \n    }\n\nThe main functionality of this class is to train the model and then make predictions using it. For this, we need to add the following methods:\n\n  * `predict()`, which goes in two forms: for one sample (takes feature value _x_ and returns prediction _h_ ( _x_ )) and for an array of samples (takes `Double` array of samples and returns `Double` array of predictions)\n  * `train()`, which takes the sample vector `xVec` and label vector `yVec` of equal length, and updates the parameters `slope` and `intercept`\n\nBoth `predict()` functions just call corresponding hypothesis functions, as shown in the following code. Later, we will add more functionality to them:\n\n    func predict(x: Double) -> Double { \n        return hypothesis(x: x) \n    }\n\nAnd for several samples:\n\n    func predict(xVec: [Double]) -> [Double] { \n        return hypothesis(xVec: xVec) \n    }\n\nNow let's add the hypothesis function   as follows:\n\n    func hypothesis(x: Double) -> Double { \n        return slope*x + intercept \n    }\n\nIn vectorized form, for several samples at once:\n\n    func hypothesis(xVec: [Double]) -> [Double] { \n        let count = UInt(xVec.count) \n        var scaledVec = [Double](repeating: 0.0, count: Int(count)) \n        vDSP_vsmulD(UnsafePointer(xVec), 1, &slope, &scaledVec, 1, count) \n        var resultVec = [Double](repeating: 0.0, count: Int(count)) \n        vDSP_vsaddD(UnsafePointer(scaledVec), 1, &intercept, &resultVec, 1, count) \n        return resultVec \n    }\n\nThe model can be trained as follows:\n\n    func train(xVec: [Double], yVec: [Double], learningRate: Double, maxSteps: Int) { \n            precondition(xVec.count == yVec.count) \n            precondition(maxSteps > 0)  \n    \/\/ The goal of training is to minimize cost function. \n    }\n\nLoss function **  \n**We've already given the loss function definition in,Chapter 1, _Getting Started with Machine Learning_ (see the _Mathematical optimization_ section), but here is the first time we actually implement real-value loss function in our code, so let's refresh what we know already.  \nIn machine learning, loss function (or cost function) maps model parameters onto a real-valued cost.\n\n# Fitting a regression line using the least squares method\n\nAs you remember from Chapter 1, _Getting Started with Machine Learning_ , for supervised learning we need two functions: the model and the loss function. We will use the least squares loss function to assess the quality of the model. The method was proposed by Carl Friedrich Gauss at the end of the 17th century. The essence of it is to minimize the distance between data points to the regression line. The difference (deviation) between the true value _y i_ and the predicted value _h_ ( _x i_) is called **residual** and denoted as _\u03b5 i_. Our loss function _J_ will be a **residual** **sum of squares** ( **RSS** ), modified just a bit. If there are _n_ samples of feature _x i_ and label _y i_, then the RSS can be calculated as:\n\nNote that all residuals are squared before summation to prevent residuals with the opposite sign from cancelling out each other. To make it independent of the size of dataset, we will divide the _RSS_ by _n_. Also, to make some future calculations simpler, we divide those by two.\n\nThe final loss function written in Swift is as follows:\n\n    func cost(trueVec: [Double], predictedVec: [Double]) -> Double { \n        let count = UInt(trueVec.count)\n\nLet's now calculate the squared Euclidean distance as follows:\n\n        var result = 0.0 \n        vDSP_distancesqD(UnsafePointer(trueVec), 1, UnsafePointer(predictedVec), 1, &result, 1)\n\nYou can normalize by vector length, as shown in the following code:\n\n        result \/= (2*Double(count)) \n        return result \n    }\n\nIt shows how well our hypothesis fits our data. Our goal is to minimize this function: changing the parameters _a_ and b finds the minimum of loss function:  . Using a bit of calculus (or looking at Wikipedia), one can show that _a_ and _b_ , which yield minimum value of loss function, can be calculated using the following formulas:\n\nIn these formulas, \u03c1 is a correlation coefficient, \u03c3\u2014a standard deviation, and \u03bc\u2014the mean.\n\n# Where to use GD and normal equation\n\nIf the goal is just to add linear regression to your app, at this point you are done; however, there is another more interesting way to obtain the same coefficients\u2014an optimization technique known as **gradient descent**. The gradient descent algorithm and its multiple descendants are used to find loss function minimums in many machine learning algorithms, including deep neural networks, where a direct analytical solution, like with linear regression is not possible. So, we'd better try it on a simple example like linear regression, so we're already familiar with it when we discuss more complex algorithms.\n\n# Using gradient descent for function minimization\n\nIf the machine learning algorithm is a car, then the optimization algorithm is its engine. For more information, refer to <http:\/\/www.pyimagesearch.com\/2016\/10\/10\/gradient-descent-with-python\/>.\n\nFrom your math classes, you should remember that geometrical interpretation is a derivative   of a function _f(x)_ is a slope at any given point ( _x)_ of the function. Now, if we have a function of two parameters _f(x, y)_ , we can't calculate just derivative as we did previously. Nevertheless, we can calculate partial derivatives:  . The vector composed of these partial derivatives is known as a gradient, and the corresponding operator is denoted by the Nabla symbol \u2207.\n\nThe gradient of a regression loss function _J(a, b)_ is a vector  . In a similar manner as a derivative is a slope of the curve at each point, the gradient is a map of heights with the vectors showing the steepest direction at each point of the map.\n\nThe loss function of the linear regression has the shape of a bowl (see _Figure 6.2_ ). Imagine a snail sitting on the edge of the bowl. It doesn't have good sight, it only can sense the direction of the surface slope. How can it reach the bottom of the bowl? The snail just needs to take small steps in the steepest direction, which is exactly against the direction that the gradient shows at each given point of the path.\n\nThe gradient descent algorithm for linear regression works in the following way:\n\n  1. Initialize _a_ and _b_ at random (or with some predefined values)\n  2. Take an _\u03b1_ (alpha) step in the direction opposite to where the gradient points\n  3. The coordinates of the point where you ended up are your new _a_ and _b_\n  4. Repeat from step 2 until convergence\n\nIn mathematical notation, it can be expressed as follows:\n\nFigure 6.2: Gradient descent trajectories for some hypothetical functions of two variables. This is a heights map depicted on the left and 3D-surface on the right\n\nNow let's implement the same algorithm in Swift. The gradient descent is an iterative algorithm, so we will utilize a loop and some stopping conditions: `maxSteps` (the maximum number of algorithm iterations), which will be checked for the convergence condition. The function explicitly takes input vectors _x_ and _y_ , a learning rate _\u03b1_ , and modifies the weights _a_ and _b_ implicitly:\n\n    func gradientDescent(xVec: [Double], yVec: [Double], \u03b1: Double, maxSteps: Int) { \n        for _ in 0 ..< maxSteps { \n            let (newSlope, newIntercept) = gradientDescentStep(xVec: xVec, yVec: yVec, \u03b1: \u03b1) \n            if (newSlope==slope && newIntercept==intercept) { break } \/\/ convergence \n            slope = newSlope \n            intercept = newIntercept \n        } \n    }\n\nNote that _a_ and _b_ (`slope` and `intercept`) should be updated simultaneously.\n\nFollowing is one step of the gradient descent:\n\n    \/\/ alpha is a learning rate \n    func gradientDescentStep(xVec: [Double], yVec: [Double], \u03b1: Double) -> (Double, Double) { \n        \/\/ Calculate hypothesis predictions. \n        let hVec = hypothesis(xVec: xVec) \n        \/\/ Calculate gradient with respect to parameters. \n        let slopeGradient = costGradient(trueVec: yVec, predictedVec: hVec, xVec: xVec) \n        let newSlope = slope + \u03b1*slopeGradient\n\n        let dummyVec = [Double](repeating: 1.0, count: xVec.count) \n        let interceptGradient = costGradient(trueVec: yVec, predictedVec: hVec, xVec: dummyVec) \n        let newIntercept = intercept + \u03b1*interceptGradient\n\n        return (newSlope, newIntercept) \n    }\n\nThe derivative of a cost function here is something that we simply derive using a pencil and paper:\n\n    \/\/ derivative of a cost function \n    func costGradient(trueVec: [Double], predictedVec: [Double], xVec: [Double]) -> Double { \n        let count = UInt(trueVec.count)\n\n        var diffVec = [Double](repeating: 0.0, count: Int(count)) \n        vDSP_vsubD(UnsafePointer(predictedVec), 1, UnsafePointer(trueVec), 1, &diffVec, 1, count)\n\n        var result = 0.0 \n        vDSP_dotprD(UnsafePointer(diffVec), 1, UnsafePointer(xVec), 1, &result, count)\n\n        \/\/ Normalize by vector length. \n        return result\/Double(count) \n    } \n    func gradientDescentStep(x: Vector<Double>,  \n    y: Vector<Double>, \u03b1: Double) -> (Double, Double) { \n        let new = Vector([b, a]) - \u03b1*cost_d(x: x, y: y)     \n        return (new[1], new[0]) \n    }\n\nDon't forget to update the `train` function:\n\n    func train(xVec: [Double], yVec: [Double], learningRate: Double, maxSteps: Int) { \n        gradientDescent(xVec: xVec, yVec: yVec, \u03b1: learningRate, maxSteps: maxSteps) \n    }\n\nFigure 6.3: The plot shows the graph of some loss function versus parameter w. The learning rate \u03b1: on the left it is too large and causes overshooting; on the right it is too small and leads to slow convergence and trapping in local minimum\n\n# Forecasting the future with simple linear regression\n\nWhile writing this book, I was using simple linear regression to estimate the approximate finish date. From time to time, I recorded the total number of pages I had written up to that moment, and then incorporated the data into linear regression. The number of pages here is a feature and the date is a label.\n\nLinear trend is a useful feature for any application that deals with some gradually progressing processes or tasks, especially in the combination with charts. Later in this chapter, we will learn how to build non-linear trend lines, namely polynomial.\n\nLet's make some forecasts:\n\n    let xVec: [Double] = [2,3,4,5] \n    let yVec: [Double] = [10,20,30,40]\n\n    let regression = SimpleLinearRegression() \n    regression.train(xVec: xVec, yVec: yVec, learningRate: 0.1, maxSteps: 31)\n\n    regression.slope \n    regression.intercept\n\n    regression.predict(x: 7) \n    regression.cost(trueVec: yVec, predictedVec: regression.predict(xVec: xVec)) \n\n# Feature scaling\n\nIf you have several features and their ranges differ significantly, many machine learning algorithms may have taught times with your data: the large feature may overwhelm the features with small absolute values. A standard way to deal with this obstacle is **feature scaling** (also known as **feature\/data normalization)**. There are several methods to perform it, but the two most common are rescaling and standardization. This is something you want to do as a preprocessing step before feeding your data into the learner.\n\nThe least squares method is almost the same as the Euclidean distance between two points. If we want to calculate how close two points are, we want each dimension to make an equal contribution to the result. In the case of the linear regression features, contributions depend on absolute values of each feature. That's why feature scaling is a must before linear regression. Later, we will meet similar technique batch normalization when we talk about deep learning neural networks.\n\nFeatures in the input data can have different ranges. For example: user age 0 to 120 years, user height 0 to 5 meters. Many loss functions would have a problem dealing with such data. Under the Euclidean distance feature with the large absolute values will suppress feature with the small absolute values. That's why usually, before passing the data into a machine learning algorithm, we want to normalize them. This is how we do it:\n\nFor scikit-learn, follow this link: <http:\/\/scikit-learn.org\/stable\/modules\/preprocessing.html>.\n\n# Feature standardization\n\nAn alternative approach is feature standardization:\n\nWhich of the two to use in your app is up to you, but be sure to use at least one of them.\n\nAs usual, we have an accelerate function for that:\n\n    func normalize(vec: [Double]) -> (normalizedVec: [Double], mean: Double, std: Double) { \n        let count = vec.count \n        var mean = 0.0 \n        var std = 0.0 \n        var normalizedVec = [Double](repeating: 0.0, count: count) \n        vDSP_normalizeD(UnsafePointer(vec), 1, &normalizedVec, 1, &mean, &std, UInt(count)) \n        return (normalizedVec, mean, std) \n    }\n\nNow we need to update the `train` method:\n\n    func train(xVec: [Double], yVec: [Double], learningRate: Double, maxSteps: Int) { \n        precondition(xVec.count == yVec.count) \n        precondition(maxSteps > 0) \n        if normalization { \n            let (normalizedXVec, xMean, xStd) = normalize(vec: xVec) \n            let (normalizedYVec, yMean, yStd) = normalize(vec: yVec)\n\n            \/\/ Save means and std-s for prediction phase. \n            self.xMean = xMean \n            self.xStd = xStd \n            self.yMean = yMean \n            self.yStd = yStd\n\n            gradientDescent(xVec: normalizedXVec, yVec: normalizedYVec, \u03b1: learningRate, maxSteps: maxSteps) \n        } else { \n            gradientDescent(xVec: xVec, yVec: yVec, \u03b1: learningRate, maxSteps: maxSteps) \n        } \n    }\n\nYou also have to update `predict` methods:\n\n    func predictOne(x: Double) -> Double { \n        if normalization { \n            return hypothesis(x: (x-xMean)\/xStd) * yStd + yMean \n        } else { \n            return hypothesis(x: x) \n        } \n    }\n\nFor a vectorized case it looks a bit more complicated, but essentially it does the same thing: shift-scale, unscale-unshift:\n\n    func predict(xVec: [Double]) -> [Double] { \n        if normalization { \n            let count = xVec.count \n            \/\/ Normalize \n            var centeredVec = [Double](repeating: 0.0, count: count) \n            var negMean = -xMean \n            vDSP_vsaddD(UnsafePointer(xVec), 1, &(negMean), &centeredVec, 1, UInt(count))\n\n            var scaledVec = [Double](repeating: 0.0, count: count) \n            vDSP_vsdivD(UnsafePointer(centeredVec), 1, &xStd, &scaledVec, 1, UInt(count))\n\n            \/\/ Predict \n            let hVec = hypothesis(xVec: scaledVec)\n\n            \/\/ Denormalize \n            var unScaledVec = [Double](repeating: 0.0, count: count) \n            vDSP_vsmulD(UnsafePointer(hVec), 1, &yStd, &unScaledVec, 1, UInt(count)) \n            var resultVec = [Double](repeating: 0.0, count: count) \n            vDSP_vsaddD(UnsafePointer(unScaledVec), 1, &yMean, &resultVec, 1, UInt(count)) \n            return resultVec \n        } else { \n            return hypothesis(xVec: xVec) \n        } \n    }\n\nLet's make some forecasts:\n\n    let xVec: [Double] = [2,3,4,5] \n    let yVec: [Double] = [10,20,30,40]\n\n    let regression = SimpleLinearRegression() \n    regression.normalization = true \n    regression.train(xVec: xVec, yVec: yVec, learningRate: 0.1, maxSteps: 31)\n\n    regression.slope \n    1.0 \n    regression.intercept \n    -1.970.... \n    regression.xMean \n    3.5 \n    regression.xStd \n    1.1180... \n    regression.yMean \n    25.0 \n    regression.yStd \n    11.18033987... \n    regression.predict(x: 7) \n    60.0 \n    regression.cost(trueVec: yVec, predictedVec: regression.predict(xVec: xVec)) \n    1.5777218... \n\n# Multiple linear regression\n\nIf we have a regression task on the dataset with multiple features, we can't use simple linear regression but we can apply its generalization: **multiple linear regression**. The formula to make a prediction now looks like this:\n\nIn this formula, _x i_ _T_ is a sample (feature vector) with _m_ features, and _w_ is a weights row vector of length _m_. The dependent variable yi is a scalar.\n\nThe task of loss minimization changes to be:\n\nIn this formula,   is the Euclidean norm (length of a vector):  . Note that this is the same as the Euclidean distance between vectors _wx_ and _y_.\n\nOne can also see multiple linear regression fitting as a solution of a system of linear equations, where each coefficient is a feature value, and each variable is a corresponding weight value:\n\nOr:\n\nThe problem here is that such a system may not have the exact solution, so we want to get a solution that, if not exact, nevertheless is optimal in some way.\n\n# Implementing multiple linear regression in Swift\n\nThe `MultipleLinearRegression` class contains a vector of weights, and staff for data normalization:\n\n    class MultipleLinearRegression { \n    public var weights: [Double]! \n    public init() {} \n    public var normalization = false \n    public var xMeanVec = [Double]() \n    public var xStdVec = [Double]() \n    public var yMean = 0.0 \n    public var yStd = 0.0 \n    ... \n    }\n\nHypothesis and prediction:\n\n    public func predict(xVec: [Double]) -> Double { \n    if normalization { \n        let input = xVec \n        let differenceVec = vecSubtract([1.0]+input, xMeanVec) \n        let normalizedInputVec = vecDivide(differenceVec, xStdVec)\n\n        let h = hypothesis(xVec: normalizedInputVec)\n\n        return h * yStd + yMean \n    } else { \n        return hypothesis(xVec: [1.0]+xVec) \n    } \n    }\n\n    private func hypothesis(xVec: [Double]) -> Double { \n    var result = 0.0 \n    vDSP_dotprD(xVec, 1, weights, 1, &result, vDSP_Length(xVec.count)) \n    return result \n    }\n\n    public func predict(xMat: [[Double]]) -> [Double] { \n    let rows = xMat.count \n    precondition(rows > 0) \n    let columns = xMat.first!.count \n    precondition(columns > 0)\n\n    if normalization { \n        let flattenedNormalizedX = xMat.map{ \n            return vecDivide(vecSubtract($0, xMeanVec), xStdVec) \n            }.reduce([], +)\n\n        \/\/ Add a column of ones in front of the matrix. \n        let basisExpanded = prepentColumnOfOnes(matrix: flattenedNormalizedX, rows: rows, columns: columns)\n\n        let hVec = hypothesis(xMatFlattened: basisExpanded) \n        let outputSize = hVec.count \n        let productVec = vecMultiply(hVec, [Double](repeating: yStd, count: outputSize)) \n        let outputVec = vecAdd(productVec, [Double](repeating: yMean, count: outputSize))\n\n        return outputVec \n    } else { \n        \/\/ Flatten and prepend a column of ones. \n        let flattened = xMat.map{[1.0]+$0}.reduce([], +) \n        return hypothesis(xMatFlattened: flattened) \n    } \n    }\n\n    private func hypothesis(xMatFlattened: [Double]) -> [Double] { \n    let matCount = xMatFlattened.count \n    let featureCount = weights.count \n    precondition(matCount > 0) \n    let sampleCount = matCount\/featureCount \n    precondition(sampleCount*featureCount == matCount) \n    let labelSize = 1 \n    let result = gemm(aMat: xMatFlattened, bMat: weights, rowsAC: sampleCount, colsBC: labelSize, colsA_rowsB: featureCount) \n    return result \n    }\n\nLeast squares cost function, almost the same as for simple regression:\n\n    public func cost(trueVec: [Double], predictedVec: [Double]) -> Double { \n    let count = trueVec.count \n    \/\/ Calculate squared Euclidean distance. \n    var result = 0.0 \n    vDSP_distancesqD(trueVec, 1, predictedVec, 1, &result, 1) \n    \/\/ Normalize by vector length. \n    result\/=(2*Double(count)) \n    return result \n    }  \n\n# Gradient descent for multiple linear regression\n\nThe gradient descent for multiple linear regression can be calculated as follows:\n\n     \/\/ derivative of a cost function \n        private func costGradient(trueVec: [Double], predictedVec: [Double], xMatFlattened: [Double]) -> [Double] { \n            let matCount = xMatFlattened.count \n            let featureCount = weights.count \n            precondition(matCount > 0) \n            precondition(Double(matCount).truncatingRemainder(dividingBy: Double(featureCount)) == 0) \n            let sampleCount = trueVec.count \n            precondition(sampleCount > 0) \n            precondition(sampleCount*featureCount == matCount) \n            let labelSize = 1\n\n            let diffVec = vecSubtract(predictedVec, trueVec)\n\n            \/\/ Normalize by vector length. \n            let scaleBy = 1\/Double(sampleCount) \n            let result = gemm(aMat: xMatFlattened, bMat: diffVec, rowsAC: featureCount, colsBC: labelSize, colsA_rowsB: sampleCount, transposeA: true, \u03b1: scaleBy)\n\n            return result \n        }\n\n        \/\/ alpha is a learning rate \n        private func gradientDescentStep(xMatFlattened: [Double], yVec: [Double], \u03b1: Double) -> [Double] {\n\n            \/\/ Calculate hypothesis predictions. \n            let hVec = hypothesis(xMatFlattened: xMatFlattened) \n            \/\/ Calculate gradient with respect to parameters. \n            let gradient = costGradient(trueVec: yVec, predictedVec: hVec, xMatFlattened: xMatFlattened) \n            let featureCount = gradient.count\n\n            \/\/ newWeights = weights - \u03b1*gradient \n            var alpha = \u03b1 \n            var scaledGradient = [Double](repeating: 0.0, count: featureCount) \n            vDSP_vsmulD(gradient, 1, &alpha, &scaledGradient, 1, vDSP_Length(featureCount))\n\n            let newWeights = vecSubtract(weights, scaledGradient)\n\n            return newWeights \n        }\n\n        private func gradientDescent(xMatFlattened: [Double], yVec: [Double], \u03b1: Double, maxSteps: Int) { \n            for _ in 0 ..< maxSteps { \n                let newWeights = gradientDescentStep(xMatFlattened: xMatFlattened, yVec: yVec, \u03b1: \u03b1) \n                if newWeights==weights { \n                    print(\"convergence\") \n                    break \n                } \/\/ convergence \n                weights = newWeights \n            } \n        } \n\n# Training multiple regression\n\nLet's now see how to train the multiple regression:\n\n     private func prepentColumnOfOnes(matrix: [Double], rows: Int, columns: Int) -> [Double] { \n    let weightsCount = columns+1\n\n    var withFirstDummyColumn = [Double](repeating: 1.0, count: rows * (columns+1)) \n    for row in 0..<rows { \n        for column in 1..<weightsCount { \n            withFirstDummyColumn[row*weightsCount + column] = matrix[row*columns + column-1] \n        } \n    } \n    return withFirstDummyColumn \n    }\n\n    public func train(xMat: [[Double]], yVec: [Double], learningRate: Double, maxSteps: Int) { \n    precondition(maxSteps > 0, \"The number of learning iterations should be grater then 0.\") \n    let sampleCount = xMat.count \n    precondition(sampleCount == yVec.count, \"The number of samples in xMat should be equal to the number of labels in yVec.\") \n    precondition(sampleCount > 0, \"xMat should contain at least one sample.\") \n    precondition(xMat.first!.count > 0, \"Samples should have at least one feature.\") \n    let featureCount = xMat.first!.count \n    let weightsCount = featureCount+1\n\n    weights = [Double](repeating: 1.0, count: weightsCount) \n    \/\/ Flatten and prepend a column of ones. \n    let flattenedXMat = xMat.reduce([], +)\n\n    if normalization { \n        let (normalizedXMat, xMeanVec, xStdVec) = matNormalize(matrix: flattenedXMat, rows: sampleCount, columns: featureCount) \n        let (normalizedYVec, yMean, yStd) = vecNormalize(vec: yVec)\n\n        \/\/ Save means and std-s for prediction phase. \n        self.xMeanVec = xMeanVec \n        self.xStdVec = xStdVec \n        self.yMean = yMean \n        self.yStd = yStd\n\n        \/\/ Add first column of ones to matrix \n        let designMatrix = prepentColumnOfOnes(matrix: normalizedXMat, rows: sampleCount, columns: featureCount)\n\n        gradientDescent(xMatFlattened: designMatrix, yVec: normalizedYVec, \u03b1: learningRate, maxSteps: maxSteps) \n    } else { \n        gradientDescent(xMatFlattened: flattenedXMat, yVec: yVec, \u03b1: learningRate, maxSteps: maxSteps) \n    } \n    } \n\n# Linear algebra operations\n\nNow we will take a look at how linear algebraic operations can be performed:\n\n     \/\/ Add two vectors. Equivalent to zip(a, b).map(+) \n    func vecAdd(_ a: [Double], _ b: [Double]) -> [Double] { \n        let count = a.count \n        assert(count == b.count, \"Vectors must be of equal length.\") \n        var c = [Double](repeating: 0.0, count: count) \n        vDSP_vaddD(a, 1, b, 1, &c, 1, vDSP_Length(count)) \n        return c \n    }\n\n    \/\/ Subtract vector b from vector a. Equivalent to zip(a, b).map(-) \n    func vecSubtract(_ a: [Double], _ b: [Double]) -> [Double] { \n        let count = a.count \n        assert(count == b.count, \"Vectors must be of equal length.\") \n        var c = [Double](repeating: 0.0, count: count) \n        vDSP_vsubD(b, 1, a, 1, &c, 1, vDSP_Length(count)) \n        return c \n    }\n\n    \/\/ Multiply two vectors elementwise. Equivalent to zip(a, b).map(*) \n    func vecMultiply(_ a: [Double], _ b: [Double]) -> [Double] { \n        let count = a.count \n        assert(count == b.count, \"Vectors must be of equal length.\") \n        var c = [Double](repeating: 0.0, count: count) \n        vDSP_vmulD(a, 1, b, 1, &c, 1, vDSP_Length(count)) \n        return c \n    }\n\n    \/\/ Divide vector a by vector b elementwise. Equivalent to zip(a, b).map(\/) \n    func vecDivide(_ a: [Double], _ b: [Double]) -> [Double] { \n        let count = a.count \n        assert(count == b.count, \"Vectors must be of equal length.\") \n        var c = [Double](repeating: 0.0, count: count) \n        \/\/ Note that parameters a and b are swapped. \n        vDSP_vdivD(b, 1, a, 1, &c, 1, vDSP_Length(count)) \n        return c \n    }\n\n    func vecNormalize(vec: [Double]) -> (normalizedVec: [Double], mean: Double, std: Double) { \n        let count = vec.count \n        var mean = 0.0 \n        var std = 0.0 \n        var normalizedVec = [Double](repeating: 0.0, count: count) \n        vDSP_normalizeD(vec, 1, &normalizedVec, 1, &mean, &std, vDSP_Length(count)) \n        return (normalizedVec, mean, std) \n    } \n    \/\/ C\u2190\u03b1AB + \u03b2C \n    \/\/ Pass flattened matrices in row-major order. \n    \/\/ rowsAC, colsBC, colsA_rowsB - Count of rows\/columns AFTER transpose. \n    func gemm(aMat: [Double], bMat: [Double], cMat: [Double]? = nil, \n              rowsAC: Int, colsBC: Int, colsA_rowsB: Int, \n              transposeA: Bool = false, transposeB: Bool = false, \n              \u03b1: Double = 1, \u03b2: Double = 0) -> [Double] { \n        var result = cMat ?? [Double](repeating: 0.0, count: rowsAC*colsBC)\n\n        \/\/ C\u2190\u03b1AB + \u03b2C \n        cblas_dgemm(CblasRowMajor, \/\/ Specifies row-major (C) or column-major (Fortran) data ordering. \n            transposeA ? CblasTrans : CblasNoTrans, \/\/ Specifies whether to transpose matrix A. \n            transposeB ? CblasTrans : CblasNoTrans, \/\/ Specifies whether to transpose matrix B. \n            Int32(rowsAC), \/\/ Number of rows in matrices A and C. \n            Int32(colsBC), \/\/ Number of columns in matrices B and C. \n            Int32(colsA_rowsB), \/\/ Number of columns in matrix A; number of rows in matrix B. \n            \u03b1, \/\/ \u03b1. \n            aMat, \/\/ Matrix A. \n            transposeA ? Int32(rowsAC) : Int32(colsA_rowsB), \/\/ The size of the first dimention of matrix A; if you are passing a matrix A[m][n], the value should be m. \n            bMat, \/\/ Matrix B. \n            transposeB ? Int32(colsA_rowsB) : Int32(colsBC), \/\/ The size of the first dimention of matrix B; if you are passing a matrix B[m][n], the value should be m. \n            \u03b2, \/\/ \u03b2. \n            &result, \/\/ Matrix C. \n            Int32(colsBC) \/\/ The size of the first dimention of matrix C; if you are passing a matrix C[m][n], the value should be m. \n        ) \n        return result \n    } \n\n# Feature-wise standardization\n\nFeature-wise standardization can be calculated as follows:\n\n    \/\/ Calculates mean for every matrix column. \n    func meanColumns(matrix: [Double], rows: Int, columns: Int) -> [Double] { \n        assert(matrix.count == rows*columns)\n\n        var resultVec = [Double](repeating: 0.0, count: columns)\n\n        matrix.withUnsafeBufferPointer{ inputBuffer in \n            resultVec.withUnsafeMutableBufferPointer{ outputBuffer in \n                let inputPointer = inputBuffer.baseAddress! \n                let outputPointer = outputBuffer.baseAddress! \n                for i in 0 ..< columns { \n                    vDSP_meanvD(inputPointer.advanced(by: i), columns, outputPointer.advanced(by: i), vDSP_Length(rows)) \n                } \n            } \n        } \n        return resultVec \n    }\n\n    \/\/ Calculates standard deviation for every matrix column. \n    func stdColumns(matrix: [Double], rows: Int, columns: Int) -> [Double] { \n        assert(matrix.count == rows*columns)\n\n        let meanVec = meanColumns(matrix: matrix, rows: rows, columns: columns)\n\n        var varianceVec = [Double](repeating: 0.0, count: columns) \n        var deviationsMat = [Double](repeating: 0.0, count: rows*columns)\n\n        \/\/ Calculating the variance for each column. \n        matrix.withUnsafeBufferPointer{ inputBuffer in \n            deviationsMat.withUnsafeMutableBufferPointer{ deviationsBuffer in \n                varianceVec.withUnsafeMutableBufferPointer{ outputBuffer in \n                    for i in 0 ..< columns { \n                        let inputPointer = inputBuffer.baseAddress!.advanced(by: i) \n                        let devPointer = deviationsBuffer.baseAddress!.advanced(by: i) \n                        let outputPointer = outputBuffer.baseAddress!.advanced(by: i)\n\n                        var mean = -meanVec[i] \n                        \/\/ Deviations of each column from its mean. \n                        vDSP_vsaddD(inputPointer, columns, &mean, devPointer, columns, vDSP_Length(rows)) \n                        \/\/ Squared deviations. \n                        vDSP_vsqD(devPointer, columns, devPointer, columns, vDSP_Length(rows)) \n                        \/\/ Sum for every column. Note, that parameters should be passed in a reverse order. \n                        vDSP_sveD(devPointer, columns, outputPointer, vDSP_Length(rows)) \n                    } \n                } \n            } \n        }\n\n        \/\/ -1 for Bessel's correction. \n        var devideBy = Double(rows) - 1 \n        vDSP_vsdivD(varianceVec, 1, &devideBy, &varianceVec, 1, vDSP_Length(columns))\n\n        \/\/ Calculating the standard deviation. \n        var length = Int32(columns) \n        var stdVec = varianceVec \n        vvsqrt(&stdVec, &varianceVec, &length)\n\n        return stdVec \n    }   \n    \/\/(x-\u03bc)\/\u03c3  \n    func matNormalize(matrix: [Double], rows: Int, columns: Int) -> (normalizedMat: [Double], meanVec: [Double], stdVec: [Double]) { \n        var meanVec = meanColumns(matrix: matrix, rows: rows, columns: columns) \n        var stdVec = stdColumns(matrix: matrix, rows: rows, columns: columns)\n\n        var result = [Double](repeating: 0.0, count: rows*columns)\n\n        matrix.withUnsafeBufferPointer{ inputBuffer in \n            result.withUnsafeMutableBufferPointer{ resultBuffer in \n                for i in 0 ..< columns { \n                    let inputPointer = inputBuffer.baseAddress!.advanced(by: i) \n                    let resultPointer = resultBuffer.baseAddress!.advanced(by: i)\n\n                    var mean = -meanVec[i] \n                    var std = stdVec[i] \n                    \/\/ Substract standard deviation. \n                    vDSP_vsaddD(inputPointer, columns, &mean, resultPointer, columns, vDSP_Length(rows)) \n                    \/\/ Devide by mean. \n                    vDSP_vsdivD(resultPointer, columns, &std, resultPointer, columns, vDSP_Length(rows)) \n                } \n            } \n        }\n\n        return (result, meanVec, stdVec) \n    }\n\n# Normal equation for multiple linear regression\n\nIf you want to implement regression in a production code, don't use matrix inversion operation explicitly. The problem with it is that it is very numerically inefficient. Instead, you can use one of the functions for solving a system of linear equations, which is essentially the same as finding regression coefficients. The good fit from the LAPACK package (part of the Accelerate framework) is the QR-decomposition function.\n\n# Understanding and overcoming the limitations of linear regression\n\nBefore building the predictive model, you should always perform an exploratory analysis. It will help you to select the right model by identifying the relationship and impact of features and samples. Linear regression has a whole bunch of preconditions and hidden assumptions. To get accurate results, you need to be sure that all those conditions are met and all assumptions are true:\n\n  * Linear regression assumes all features to be numerical variables. If you have categorical features, you cannot use linear regression. You need to be careful here, because often categorical variables are represented by numbers; for example, country code or E numbers of food additives (found on all food labels in the EU, for example E260 stands for acetic acid). In other words, linear regression can only be applied to quantities (the amount of something) and not to categories, ordered lists, scales, or numerical codes.\n  * Linear regression models linear relationships; that means that features should be linearly related to labels. Build a scatterplot to be sure that the data you have can be modeled using the line. My favorite quote in this context is Randall Munroe's famous quote from <https:\/\/xkcd.com\/1725\/>:\n\n\"I don't trust linear regressions when it's harder to guess the direction of the correlation from the scatter plot than to find new constellations on it.\"\n\nFigure 6.4: Datasaurus Dozen [1] shows how very different datasets can have very similar descriptive statistics. Note how linear regression parameters ( _a_ and _b_ ) vary insignificantly among these datasets. Linear regression is a bad model for non-linear data\n\n  * Linear regression is **s** ensitive to outliers (see _Figure 6.5_ ); in other words, it is an unstable algorithm\u2014it gives a bad result on a noisy data. How many outliers do you need to completely spoil the model? One outlier is enough. If you have outliers, be sure to check how they affect your model by building two regressions: one with outliers, and one without them. There are regression algorithms that were developed specifically for the use with the noisy data, such as robust regression, RANSAC, and its modifications:\n\nFigure 6.5: Anscombe's quartet [2] is commonly used to demonstrate linear regression limitations. The first graph shows the dataset, which can be modeled by linear regression. The second one shows nonlinearity in the data. The third and fourth show the instability of the algorithm: one outlier is enough to break the model completely\n\n  * Linear regression also has some requirements for the errors. Errors should be independent, homoscedastic, and normally distributed around the regression line (see _Figure 6.6_ ). The homoscedasticity requirement means that variance of errors remains the same over the entire dataset:\n\nFigure 6.6: Error distribution\n\nThere are three spells that break linear regression: multicollinearity (correlation between features), autocorrelation (correlation between samples), heteroscedasticity (error variance changes). Countermeasures are regularization, selecting the most important features by stepwise regression, forward selection, and backward elimination.\n\n# Fixing linear regression problems with regularization\n\nAs we've seen, one outlier is enough to break the least-squares regression. Such instability is a manifestation of overfitting problems. Methods that help prevent models from overfitting are generally referred to as **regularization** techniques. Usually, regularization is achieved by imposing additional constraints on the model. This can be an additional term in a loss function, noise injection, or something else. We've already implemented one such technique previously, in Chapter 3, _K-Nearest Neighbors Classifier_. Locality constraint _w_ in the DTW algorithm is essentially a way to regularize the result. In the case of linear regression, regularization imposes constraints on the weights vector values.\n\n# Ridge regression and Tikhonov regularization\n\nUnder the standard least squares method, the obtained regression coefficients can vary wildly. We can formulate the least squares regression as an optimization problem:\n\nWhat we have on the right here is just an RSS in a form of a scalar product. Tikhonov regularized least squares regression adds an additional penalty term\u2014squared _L_ 2 norm of weights vector:\n\nwhere _L_ 2 norm  and \u03bb is a scalar shrinkage parameter. It allows to control weights variance and keep it low. Similar to other hyperparameters, \u03bb needs to be defined separately, usually using held-out data or cross-validation. The larger it is, the smaller the regression coefficients (weights) will be.\n\nSuch an optimization problem has a closed-form solution, similar to the normal equation:\n\n_I_ is an identity matrix, where main diagonal elements are equal to _1_ , and all others equal to _0_.\n\nLinear regression regularized in such a way is known as **ridge regression**. One of its advantages is that it can be used even if features in a training data are highly correlated (multicollinearity). Unlike usual linear regression, ridge regression doesn't assume the normal distribution of errors. It reduces the absolute values of features but they don't reach zero, which means that this regression also performs badly if there are irrelevant features.\n\n# LASSO regression\n\nTo fix the problem with irrelevant features, one can replace the _L_ 2 norm with the _L_ 1 norm in a penalty term, and instead of penalizing squares of regression coefficients, penalize their absolute values:\n\nwhere _L_ 1 norm  . This is so-called **Least Absolute Shrinkage and Selection Operator** ( **LASSO** ) regression. Some of the weight coefficients under such a penalty can become exactly zero, which you can think of as a feature selection. If there are several highly correlated features in your dataset, LASSO picks one of them and sets all others to zero. This also means that LASSO often results in sparse weight vectors.\n\nThis type of regression also doesn't assume the normality of error distribution.\n\n# ElasticNet regression\n\nElasticNet regression is a combination of ridge and LASSO methods: add both penalty terms to the usual least squares loss function and you will get the ElasticNet regression. It also has two shrinkage parameters:\n\nIt is especially helpful when you have multiple correlated features. When two features are correlated, LASSO tends to choose one of them randomly, while ElasticNet keeps both. Similar to the ridge regression, ElasticNet is also more stable in many cases:\n\nFigure 6.7: Penalty terms in the space of the model parameters.\n\nRegularized linear regression is available in multiple machine learning packages, including Scikit-learn for integration with Core ML and AIToolbox for on-device training.\n\n# Summary\n\nIn this chapter, we've explored linear regression and gradient descent. Linear regression is a simple parametric model. It makes a certain assumption about data shape and error distribution. We were also acquainted with the Accelerate framework, a powerful hardware-accelerated framework from Apple for numerical computations.\n\nIn the next chapter, we'll continue by building different, more complex models on top of linear regression: polynomial regression, regularized regression, and logistic regression.\n\n# Bibliography\n\n  1. Justin Matejka, George Fitzmaurice (2017), _Same Stats, Different Graphs: Generating Datasets with Varied Appearance and Identical Statistics through Simulated Annealing_ , CHI 2017 Conference Proceedings: ACM SIGCHI Conference on Human Factors in Computing Systems\n  2. F. J. Anscombe, _Graphs in Statistical Analysis_ , The American Statistician, V-27 (1): 17-21 (1973), JSTOR 2682899\n\n# Linear Classifier and Logistic Regression\n\nIn the previous chapter, we added several useful supervised learning algorithms for regression tasks to our toolbox. Continuing with building on top of linear regression, in this chapter, we are going to build two classification algorithms: **linear classifier** and **logistic regression**. Both of them take familiar feature vectors as input, similar to multiple linear regression. The difference is in their output. The linear classifier will output true or false (binary classification) and logistic regression will provide the probability of some event happening.\n\nThe topics to discuss in this chapter are:\n\n  * Bias and variance\n  * Linear classifier\n  * Logistic regression\n\n# Revisiting the classification task\n\nWe already used and implemented some classification algorithms in the previous chapters: decision tree learning, random forest, and KNN are all well suited for solving this task. However, as Boromir used to say, \"\"One cannot simply walk into neural networks without knowing about logistic regression\"\". So, to remind you, classification is almost the same as regression, except that response variable _y_ is not a continuous (`float`) but takes values from some set of discrete values (`enum`). In this chapter, we're primarily concerned with the binary classification, where _y_ can be either true or false, one or zero, and belong to a positive or negative class.\n\nAlthough, if you think about this for a moment, it's not too hard to build a multiclass classifier from several binary classifiers by chaining them one after the other. In the classification domain, response variable _y_ is usually called a **label**.\n\n# Linear classifier\n\nLinear regression can be trivially adapted for binary classification: just predict a positive class for all regression outputs above some threshold and a negative class for everything below it. For example, in the following diagram, the threshold is 0.5. Everything with _x_ < 0.5 gets classified as a negative class and everything with _x_ > 0.5 as positive. The line that separates feature values of one class from another is called a **decision boundary**. With more than one feature, the decision boundary will not be a line but a hyperplane:\n\nFigure 7.1: Linear classifier\n\n# Logistic regression\n\nYou can come up with a lot of problems a linear classifier has. One of them is that many datasets just cannot be separated properly by a straight line:\n\nFigure 7.2: Linear separable (on the left) and nonlinear separable (on the right) data. The decision boundary as shown in the dashed line. Source: Mykola Sosnovshchenko.\n\nAnother problem is that a linear regression line can predict negative values or values greater than one for some samples even though we know for sure _y_ should be either zero or one. To fix it, we need some function that takes values from [-\u221e, +\u221e] and output values in the range from zero to one. One such function is a logistic function. Please refer to the following formula and graph:\n\nFigure 7.3: Logistic function\n\nRemember that in linear regression, we had our hypothesis function defined as a linear transformation (dot product of vectors):\n\nIn a logistic regression, we add a nonlinear logistic transformation like this:\n\nLogistic regression is used to estimate the probability of some event happening or not happening. In other words, it is a binary classification algorithm that outputs the probability of the sample belonging to one or another class. The typical example of logistic regression output looks like this: _There's a 0.95 chance the letter is spam, and 0.05 chance it's not_.\n\nThe output of the logistic regression is always in a range (0, 1). We still call this algorithm _regression_ , despite using it for classification, because it produces a continuous output; however, this is the closest we can get to a discrete output using differential functions. Why do we want them to be differential? Because we want to use our good friend gradient descent to learn the parameter vector _w_.\n\n# Implementing logistic regression in Swift\n\nThe most important differences of this implementation from multiple linear regression are the following:\n\n  * Normalization is required only for feature matrix _x_ , and not for the target vector _y_ , because the output has range (0, 1)\n  * The hypothesis is different\n  * The cost function looks different, but the cost gradient remains the same\n\nAgain, we'll need some accelerate functions:\n\n    import Accelerate\n\nThe logistic regression class definition looks similar to multiple linear regression:\n\n    public class LogisticRegression { \n    public var weights: [Double]!\n\n    public init(normalization: Bool) { \n        self.normalization = normalization \n    }\n\n    private(set) var normalization: Bool \n    private(set) var xMeanVec = [Double]() \n    private(set) var xStdVec = [Double]() \n\n# The prediction part of logistic regression\n\nThis is the code that implements hypotheses for one sample input and for a matrix of inputs:\n\n    public func predict(xVec: [Double]) -> Double { \n        if normalization { \n            let input = xVec \n            let differenceVec = vecSubtract(input, xMeanVec) \n            let normalizedInputVec = vecDivide(differenceVec, xStdVec)\n\n            let h = hypothesis(xVec: [1.0]+normalizedInputVec)\n\n            return h \n        } else { \n            return hypothesis(xVec: [1.0]+xVec) \n        } \n    }\n\n    private func hypothesis(xVec: [Double]) -> Double { \n        var result = 0.0 \n        vDSP_dotprD(xVec, 1, weights, 1, &result, vDSP_Length(xVec.count)) \n        return 1.0 \/ (1.0 + exp(-result)) \n    }\n\n    public func predict(xMat: [[Double]]) -> [Double] { \n        let rows = xMat.count \n        precondition(rows > 0) \n        let columns = xMat.first!.count \n        precondition(columns > 0)\n\n        if normalization { \n            let flattenedNormalizedX = xMat.map{ \n                return vecDivide(vecSubtract($0, xMeanVec), xStdVec) \n                }.reduce([], +)\n\n            \/\/ Add a column of ones in front of the matrix. \n            let basisExpanded = prependColumnOfOnes(matrix: flattenedNormalizedX, rows: rows, columns: columns) \n            let hVec = hypothesis(xMatFlattened: basisExpanded)\n\n            return hVec \n        } else { \n            \/\/ Flatten and prepend a column of ones. \n            let flattened = xMat.map{[1.0]+$0}.reduce([], +) \n            return hypothesis(xMatFlattened: flattened) \n        } \n    }\n\n    private func hypothesis(xMatFlattened: [Double]) -> [Double] { \n        let matCount = xMatFlattened.count \n        let featureCount = weights.count \n        precondition(matCount > 0) \n        let sampleCount = matCount\/featureCount \n        precondition(sampleCount*featureCount == matCount) \n        let labelSize = 1 \n        var result = gemm(aMat: xMatFlattened, bMat: weights, rowsAC: sampleCount, colsBC: labelSize, colsA_rowsB: featureCount)\n\n        \/\/ -h \n        vDSP_vnegD(result, 1, &result, 1, vDSP_Length(sampleCount))\n\n        \/\/ exp(-h) \n        \/\/ vForce function for double-precision exponent. \n        var outputLength = Int32(sampleCount) \n        vvexp(&result, result, &outputLength)\n\n        \/\/ 1.0 + exp(-h) \n        var one = 1.0 \n        vDSP_vsaddD(result, 1, &one, &result, 1, vDSP_Length(sampleCount))\n\n        \/\/ 1.0 \/ (1.0 + exp(-h)) \n        vDSP_svdivD(&one, result, 1, &result, 1, vDSP_Length(sampleCount))\n\n        return result \n    } \n\n# Training the logistic regression\n\nThe training part is also very similar to linear regression:\n\n    public func train(xMat: [[Double]], yVec: [Double], learningRate: Double, maxSteps: Int) { \n      precondition(maxSteps > 0, \"The number of learning iterations should be grater then 0.\") \n      let sampleCount = xMat.count \n      precondition(sampleCount == yVec.count, \"The number of samples in xMat should be equal to the number of labels in yVec.\") \n      precondition(sampleCount > 0, \"xMat should contain at least one sample.\") \n      precondition(xMat.first!.count > 0, \"Samples should have at least one feature.\") \n      let featureCount = xMat.first!.count \n      let weightsCount = featureCount+1\n\n      weights = [Double](repeating: 1.0, count: weightsCount)\n\n      if normalization {   \n       **\/\/ Flatten**  \n        let flattenedXMat = xMat.reduce([], +) \n        let (normalizedXMat, xMeanVec, xStdVec) = matNormalize(matrix: flattenedXMat, rows: sampleCount, columns: featureCount) \n        **\/\/ Save means and std-s for prediction phase**  \n        self.xMeanVec = xMeanVec \n        self.xStdVec = xStdVec \n        **\/\/ Add first column of ones to matrix**  \n        let designMatrix = prependColumnOfOnes(matrix: normalizedXMat, rows: sampleCount, columns: featureCount)\n\n        gradientDescent(xMatFlattened: designMatrix, yVec: yVec, \u03b1: learningRate, maxSteps: maxSteps) \n      } else {   \n       **\/\/ Flatten and prepend a column of ones**  \n        let flattenedXMat = xMat.map{[1.0]+$0}.reduce([], +) \n        gradientDescent(xMatFlattened: flattenedXMat, yVec: yVec, \u03b1: learningRate, maxSteps: maxSteps) \n      } \n    } \n\n# Cost function\n\nThe `cost` function is something we can use to assess the prediction quality:\n\n    \/\/ cost(y, h) = -sum(y.*log(h)+(1-y).*log(1-h))\/m \n    public func cost(trueVec: [Double], predictedVec: [Double]) -> Double { \n      let count = trueVec.count \n      \/\/ Calculate squared Euclidean distance. \n      var result = 0.0\n\n      var left = [Double](repeating: 0.0, count: count) \n      var right = [Double](repeating: 0.0, count: count)\n\n      \/\/ log(h) \n      var outputLength = Int32(count) \n      vvlog(&left, predictedVec, &outputLength)\n\n      \/\/ -y.*log(h) \n      left = vecMultiply(trueVec, left)\n\n      \/\/ 1-y \n      var minusOne = -1.0 \n      var oneMinusTrueVec = [Double](repeating: 0.0, count: count) \n      vDSP_vsaddD(trueVec, 1, &minusOne, &oneMinusTrueVec, 1, vDSP_Length(count)) \n      vDSP_vnegD(oneMinusTrueVec, 1, &oneMinusTrueVec, 1, vDSP_Length(count))\n\n      \/\/ 1-h \n      var oneMinusPredictedVec = [Double](repeating: 0.0, count: count) \n      vDSP_vsaddD(predictedVec, 1, &minusOne, &oneMinusPredictedVec, 1, vDSP_Length(count)) \n      vDSP_vnegD(oneMinusPredictedVec, 1, &oneMinusPredictedVec, 1, vDSP_Length(count))\n\n      \/\/ log(1-h) \n      vvlog(&right, oneMinusPredictedVec, &outputLength)\n\n      \/\/ (1-y).*log(1-h) \n      right = vecMultiply(oneMinusTrueVec, right)\n\n      \/\/ left+right \n      let sum = vecAdd(left, right)\n\n      \/\/ sum() \n      vDSP_sveD(sum, 1, &result, vDSP_Length(count))\n\n      \/\/ Normalize by vector length. \n      result\/=(Double(count)) \n      return -result \n    }\n\nThe derivative of a `cost` function is something we use to adjust weights to minimize the `cost` function itself:\n\n    \/\/ x'*sum(h-y) \n    private func costGradient(trueVec: [Double], predictedVec: [Double], xMatFlattened: [Double]) -> [Double] { \n      let matCount = xMatFlattened.count \n      let featureCount = weights.count \n      precondition(matCount > 0) \n      precondition(Double(matCount).truncatingRemainder(dividingBy: Double(featureCount)) == 0) \n      let sampleCount = trueVec.count \n      precondition(sampleCount > 0) \n      precondition(sampleCount*featureCount == matCount) \n      let labelSize = 1\n\n      let diffVec = vecSubtract(predictedVec, trueVec)\n\n      \/\/ Normalize by vector length. \n      let scaleBy = 1\/Double(sampleCount) \n      let result = gemm(aMat: xMatFlattened, bMat: diffVec, rowsAC: featureCount, colsBC: labelSize, colsA_rowsB: sampleCount, transposeA: true, \u03b1: scaleBy)\n\n      return result \n    }\n\n    \/\/ alpha is a learning rate \n    private func gradientDescentStep(xMatFlattened: [Double], yVec: [Double], \u03b1: Double) -> [Double] {\n\n      \/\/ Calculate hypothesis predictions. \n      let hVec = hypothesis(xMatFlattened: xMatFlattened) \n      \/\/ Calculate gradient with respect to parameters. \n      let gradient = costGradient(trueVec: yVec, predictedVec: hVec, xMatFlattened: xMatFlattened) \n      let featureCount = gradient.count\n\n      \/\/ newWeights = weights - \u03b1*gradient \n      var alpha = \u03b1 \n      var scaledGradient = [Double](repeating: 0.0, count: featureCount) \n      vDSP_vsmulD(gradient, 1, &alpha, &scaledGradient, 1, vDSP_Length(featureCount))\n\n      let newWeights = vecSubtract(weights, scaledGradient)\n\n      return newWeights \n    }\n\n    private func gradientDescent(xMatFlattened: [Double], yVec: [Double], \u03b1: Double, maxSteps: Int) { \n      for _ in 0 ..< maxSteps { \n        let newWeights = gradientDescentStep(xMatFlattened: xMatFlattened, yVec: yVec, \u03b1: \u03b1) \n        if newWeights==weights { \n          print(\"convergence\") \n          break \n        } \/\/ convergence \n        weights = newWeights \n      } \n    }  \n\n# Predicting user intents\n\nThe problem: Apple's default Clock app, if opened from the app switcher menu (the one you see when swiping from the bottom of the screen upward), always shows the Timer tab. I personally use this app mostly for one reason every day\u2014to set an alarm clock, which is in a different tab. By knowing the day of the week and time of the day, it's easy to make the app smarter (and less annoying) by opening the proper Alarm tab when needed and default tab otherwise. For this, we will need to collect historical records on what time we usually set an alarm on different days.\n\nLet's formulate the task more precisely:\n\n  * Input data: The day, hour, and minute when the user had opened the application\n  * Expected output: The probability that the user wants to set up an alarm\n\nThe task is of binary classification, which makes logistic regression a perfect candidate for the solution.\n\n# Handling dates\n\nThe straightforward way to transform dates and time into numerical features is by replacing them with integers. For example, days of the week (assuming that Sunday is the first day) can be encoded as numbers from 0 to 6, and hours as integers from 0 to 23:\n\n    Monday, 11:45 pm, alarm tab \u2192 [1, 23, 45, 1]  \n    Thursday, 1:15 am, alarm tab \u2192 [4, 1, 15, 1]  \n    Saturday, 10:55 am, timer tab \u2192 [6, 10, 55, 0]  \n    Tuesday, 5:30 pm, timer tab \u2192 [2, 17, 30, 0]\n\nTo explain why this is a bad approach, take a look at the following diagram. The samples 11:45 pm and 1:15 am are close to each other, but this will not be obvious to our model if we encode them in a straightforward way. We can fix this situation by projecting the day of the week ( _d_ ) together with the hour ( _h_ ) and minute ( _m_ ) on the circle:\n\n**Parameter** | **Formula**  \n---|---  \n`dow_sin` | ** **  \n`dow_cos` |    \n`time_sin` |    \n`time_cos` |\n\nThe result of this transformation can be seen in the following diagram:\n\nAfter the transformation, each sample in the dataset will contain four new features:\n\n**dow_sin** | **dow_cos** | **time_sin** | **time_cos** | **label**  \n---|---|---|---|---  \n`0.781831482` | `0.623489802` | `-0.065403129` | `0.997858` | `alert`  \n`-0.433883739` | `-0.900968868` | `0.321439465` | `0.946930129` | `alert`  \n`-0.781831482` | `0.623489802` | `0.279829014` | `-0.960049854` | `timer`  \n`0.974927912` | `-0.222520934` | `-0.991444861` | `-0.130526192` | `timer`\n\nNow these data points can be successfully separated by either the linear classifier or logistic regression.\n\n# Choosing the regression model for your problem\n\nBy now, you may feel overwhelmed by the number of models, regularization, and preprocessing techniques. No worries, there is a simple algorithm for choosing the model:\n\n  1. If your label is continuous\u2014linear regression\n  2. If your label is binary\u2014logistic regression\n  3. High dimensionality and multicollinearity\u2014regularization methods (lasso, ridge, and ElasticNet)\n\n# Bias-variance trade-off\n\nErrors in machine learning can be decomposed into two components: bias and variance. The difference between them is commonly explained using the shooting metaphor, as demonstrated in the following diagram. If you train a high-variance model on 10 different datasets, the results would be very different. If you train a high-bias model on 10 different datasets, you would get very similar results. In other words, high-bias models tend to underfit and high-variance models tend to overfit. Usually, the more parameters the model has the more it is prone to overfitting, but there are also differences between model classes: parametric models like linear and logistic regressions tend to be biased, while nonparametric models like KNN usually have a high variance:\n\nFigure 7.4: Two components of errors: bias and variance\n\n# Summary\n\nIn this chapter, we discussed how to turn linear regression into a classification algorithm. We also implemented logistic regression, an important classification algorithm.\n\nHaving gained an understanding of this will be of great use for us in the next chapter, where we will implement our first neural network.\n\n# Neural Networks\n\nJust a decade ago, artificial neural networks ( **NNs** ) were considered by most researchers as an unpromising branch of computer science. But as computational power grew, and efficient algorithms to train NNs on GPUs were found, the situation changed dramatically. The latest discoveries in the field have achieved unprecedented results, such as tracking objects in video; synthesizing realistic speech, paintings, and music, automatic translation from one language to another; and extracting meaning from text, images, and video. NNs were rebranded as _deep learning_ and they've set all kinds of records in computer vision and natural language processing, beating almost all other ML approaches over the last few years (2014-2018). Deep NNs caused a new machine learning boom, raising a wave of discussions and predictions about the artificial general intelligence forthcoming.\n\nNow, there are already so many NN types that it's hard to keep track of them: convolutional, recurrent, recursive, autoencoders, generative adversarial, binary, with memory, with attention, and so on. New architectures and applications continue to appear almost every week, thanks to the growing community of enthusiasts all around the world who experiment with NNs, applying them to all possible kinds of tasks.\n\nHere is a short list of what NNs are now doing more or less successfully:\n\n  * Coloring black and white photographs\n  * Drawing new Pokemon\n  * Writing scripts for advertisements\n  * Diagnosing cancer cells\n\nThanks to the breakthroughs in deep learning, we have come close to saying (though not yet loudly) that our computers can now fantasize, dream, and hallucinate. Today, researchers are working on NNs that can, on their own, design and train other NNs, write computer programs, help understand intracellular processes, and decipher forgotten scripts and the language of dolphins. Starting with this chapter, we will begin to plunge into deep learning.\n\nIn this chapter, we will cover the following topics:\n\n  * What are NNs, neurons, layers, and activation functions?\n  * What types of activation functions are there?\n  * How to train NNs: backpropagation, stochastic gradient descent\n  * What is deep learning?\n  * Which deep learning frameworks are best suited for iOS applications?\n  * Implementing a multilayer perceptron, and how to train it.\n\n# What are artificial NNs anyway?\n\nThe group of models that we call artificial NNs are universal approximation machines; in other words, the functions that can imitate the behavior of any other function of interest. Here, I mean functions in a more mathematical meaning, as opposed to computer science: functions that take a real-valued input vector and return a real-valued output vector. This definition holds true for feed-forward NNs, which we will be discussing in this chapter. In the following chapters, we'll see networks that map an input tensor (multidimensional array) to an output tensor, and also networks that take their own outputs as an input.\n\nWe can think of a NN as a graph and the neuron as a node in a directed acyclic graph. Each such node takes some input and produces some output. Modern NNs are only loosely inspired by the biological brain. If you want to know more about the biological prototype and its relation to NNs, check the _Seeing biological analogies_ section.\n\n# Building the neuron\n\nConsidering that a biological neuron has an astonishingly complex structure (see _Figure 8.1_ ), how do we approach modeling it in our programs? Actually, most of this complexity is, so to say, at the hardware level. We can abstract it out and think of the neuron as a node in a graph, which takes one or more inputs and produces some output (sometimes called _activation_ ).\n\nWait, but doesn't that sound like something familiar? Yes, you are right: an artificial neuron is just a mathematical function.\n\nThe most common way to model the neuron is by using the weighted sum of inputs with the non-linearity function _f_ :\n\nWhere _w_ is a weights vector, _x_ is an input vector, and _b_ is a bias term. The _y_ is a neuron's scalar output.\n\nFigure 8.1: A typical motor neuron of a vertebrate. Public domain diagram from Wikimedia Commons\n\nFigure 8.2: Artificial neuron diagram\n\nA typical artificial neuron processes input in the following three steps, as demonstrated in the preceding diagram ( _Figure 8.2_ ):\n\n  1. Take a weighted sum of inputs. Each neuron has a vector of weights of the same length as the number of inputs. Sometimes, one more weight is introduced as for bias term (always equal to one). The weighted sum of inputs is a dot product of the input vector and weight vector.\n  2. Pass the result through a non-linear function (synonyms: activation function, transfer function).\n  3. Pass the result of computations downstream to the next neurons.\n\nThe first step is a very familiar linear regression. If activation is a step function, this makes an individual neuron mathematically identical to a binary linear classifier. If you replace the step function with a logistic function, what you will get is a logistic regression. But now we call them neurons, and can assemble them into a network.\n\nThe learning of a neuron occurs when its input weight adjusts in such a way that the whole neuron produces better output. Again, this is the same as with a linear regression. To train a NN, we usually use a backpropagation algorithm, which is built on top of the familiar gradient descent.\n\n# Non-linearity function\n\nAn activation function maps the weighted input of a neuron into a real value to produce the neuron's output. Many of the NN's properties depend on the choice of activation function, including its ability to generalize, and the speed of the training process convergence. Usually, we want it to be differentiable, so we can optimize the whole network using the gradient descent. Most commonly used activation functions are non-linear: piecewise linear, or s-shaped (see _Table 8.1_ ). Nonlinear activation functions allow NNs to outperform other algorithms in many nontrivial tasks using only a few neurons. Oversimplifying, activation functions can be divided into two groups: step-like and rectifier-like (see _Figure 8.3_ ). Let's take a closer look at some examples:\n\n**Name** | **Formula** | **Derivative**\n\n---|---|---\n\nStep function |   |\n\nLogistic |   |\n\nHyperbolic tangent |   |\n\nReLU |   |\n\nLeaky ReLU |   |\n\nSoftplus |   |\n\nMaxout |   |\n\nTable 8.1: Commonly used activation functions\n\nFigure 8.3: Plots of the common activation functions: step-like in the left column, and rectifier-like in the right column\n\n# Step-like activation functions\n\nThe heaviside step function (also known as **unit step function** or **threshold function** ) outputs _0_ for all values less than _zero_ , and _1_ for everything else. This is a natural choice to model a biological neuron, which produces an electrical impulse, _1_ , or stays silent: _0_. Unfortunately, the function is not differentiable because of the discontinuity at _0_ , which makes it impossible to train such networks using gradient descent algorithm. Each individual neuron in such a network is a mathematical equivalent of a binary linear classifier, hence such networks are unable to perform well on nonlinear tasks.\n\nA **logistic (sigmoid) function** is a continuous approximation of a step function. The function squashes the input from range (-\u221e, +\u221e) to the range ( _0_ , _1_ ). It allows NNs to be trained using gradient descent, but it also has two problems:\n\n  * Because of the sigmoid's shape, NNs that use it are prone to a **vanishing gradient problem** , which will be explained later (see the _Vanishing gradient problem_ section).\n  * The output of the sigmoid is not zero-centered. This introduces undesirable zig-zagging behavior of the weights values during training, and the networks generally train slower.\n\nWith the sigmoid activation function, each neuron essentially performs the logistic regression.\n\n**Hyperbolic tangent** ( **tanh** ) is a scaled logistic function, so the shape of the function is very similar but the range of its output is ( _-1_ , _1_ ). This means that the tanh still suffers from the vanishing gradient, but at least its output is now zero-centered.\n\n# Rectifier-like activation functions\n\nA rectifier is a piecewise linear function, which you hardly ever meet outside of the NNs context. This class of function was designed specifically to mitigate the problems and limitations of traditional step-like activation functions. A rectifier applies a simple thresholding: `max(0, x)`. A neuron uses a rectifier is known as a **rectified linear unit** ( **ReLU** ).\n\nUnlike sigmoids, a rectifier doesn't saturate at the upper end. This helps the neuron to tell apart a poor prediction from a very poor prediction, and update weights accordingly even in such a difficult situation. ReLU is also very cheap computationally: unlike sigmoids, which require exponentials, ReLU can be implemented as a thresholding operation. It also has been shown that a network of ReLUs can converge up to six times faster than one using sigmoids, so ReLU quickly gained popularity in the deep learning community after its invention.\n\nReLU has its own drawbacks, so several modifications were proposed to fix them:\n\n  * **Leaky ReLU** : Instead of _0_ for all values, less than _0_ and this activation returns a tiny fraction of an input ( _0.01_ , for example). The size of the fraction is determined by the constant \u03b1. Presumably, this should prevent ReLU from saturation at the bottom end, but in practice it usually doesn't help much.\n  * **Randomized ReLU** : \u03b1 is random in some bounds. Randomization is a common way of NN regularization, which we will see later in this chapter.\n  * **Parametric ReLU (PReLU)** : \u03b1 is a trainable parameter, which is adjusted through a gradient descent.\n  * **Softplus** : An approximation of a ReLU using exponentials. A derivative of this function is a sigmoid.\n  * **Maxout unit** : Combines both ReLU and leaky ReLU in one expression. In this way, it allows the maxout unit to have all ReLU benefits, namely linearity, with no saturation, but doesn't have the dying ReLU problem. The downside here is that the maxout unit has double the number of parameters in comparison to ReLU, so it's computationally more expensive.\n\n# Building the network\n\nIndividual neurons can be organized in a network (see _Figure 8.4_ ), usually by joining several neurons in parallel in a layer and then stacking layers on top of each other. Such a network is known as a **feed-forward NN** or a **multilayer perceptron (MLP)**. The first layer is an input layer, the last layer is an output layer, and all inner layers are known as _hidden layers_. If each neuron of one layer is connected to the all neurons in the next layer, such a network is called a **fully-connected NN**.\n\nA fully-connected feed-forward multilayer perceptron with one type of activation (usually sigmoid) is a traditional (canonical) type of NN. It is mostly used for classification purposes. In the following chapters, we will discuss other types of NNs, but in this chapter we will stick to the MLP:\n\nFigure 8.4: Fully-connected feed-forward NN with five layers\n\n# Building a neural layer in Swift\n\nA fully-connected layer is easy to implement, because it can be expressed as two operations:\n\n  * A matrix multiplication between weights matrix _W_ and input vector _x._\n  * A point wise application of activation function _f_ :\n\nFigure 8.5: One layer in detail\n\nIn many frameworks, the two operations are separated so that matrix multiplication happens in the fully-connected layer and activation happens in the next nonlinearity layer. This is handy because in this way we can easily replace the weighted sum with convolution. In the next chapter, we will discuss convolutional NNs.\n\nBut for now, let's see how NNs can perform logical operations. One neuron is enough to model any logical gate, except XOR. This finding caused the first AI winter in the 1960s; however, XOR is trivial to a model having a network with two layers.\n\n# Using neurons to build logical functions\n\nAmong other obscured parts of iOS and macOS SDK, there is one interesting library called SIMD. It is an interface for direct access to vector instructions and vector types, which are mapped directly to the vector unit in the CPU, without the need to write an assembly code. You can reference vector and matrix types as well as linear algebra operators defined in this header right from your Swift code, starting from 2.0 version.\n\nThe **universal approximation** theorem states that a simple NN with one hidden layer can approximate a wide variety of continuous functions if proper weights are found. This is also commonly rephrased as NNs as universal function approximators. However, the theorem doesn't tell if it's possible to find such proper weights.\n\nTo get access to those goodies, you need to `import simd` in Swift files, or `#include <simd\/simd.h>` in C\/C++\/Objective-C files. GPU also has SIMD units in it, so you can import SIMD into your metal shader code as well.\n\nAs per iOS 10.3\/Xcode 8.2.1, some parts of C SIMD functionality are not available in the Swift version of it; for instance, logical and trigonometric operations. To see them, create Objective-C file, `#import <simd\/simd.h>` and click _command_ , click on the `simd.h` to go to the header files.\n\nThe part I like the most about the SIMD is that all vectors and matrices in it have size explicitly mentioned as a part of their type. For example, function `float 4()` returns the matrix of size 4 x 4. But it also makes SIMD inflexible because only matrices of sizes from 2 up to 4 are available.\n\nTake a look at the SIMD playground for some examples of SIMD usage:\n\n    let firstVector = float4(1.0, 2.0, 3.0, 4.0) \n    let secondVector = firstVector \n    let dotProduct = dot(firstVector, secondVector)\n\nThe result is as follows:\n\nFigure 8.6: A NN implementing XOR function\n\nTo illustrate that SIMD can be used for ML algorithms, let's implement a simple XOR NN in SIMD:\n\n    func xor(_ a: Bool, _ b: Bool) -> Bool { \n        let input = float2(Float(a), Float(b))\n\n        let weights1 = float2(1.0, 1.0) \n        let weights2 = float2(-1.0, -1.0)\n\n        let matrixOfWeights1 = float2x2([weights1, weights2]) \n        let weightedSums = input * matrixOfWeights1\n\n        let stepLayer = float2(0.5, -1.5)\n\n        let secondLayerOutput = step(weightedSums, edge: stepLayer)\n\n        let weights3 = float2(1.0, 1.0) \n        let outputStep: Float = 1.5\n\n        let weightedSum3 = reduce_add(secondLayerOutput * weights3)\n\n        let result = weightedSum3 > outputStep \n        return result \n    }\n\nThe good thing about SIMD is that it explicitly says to the CPU to calculate the dot product in one step, without looping over the vector but rather utilizing SIMD instructions.\n\n# Implementing layers in Swift\n\nThere are at least three options to consider when you want to implement a NN in Swift:\n\n  * Implement it in pure Swift (which may be useful mostly for the study purposes). A lot of implementations of different complexity and functionality can be found on the GitHub. It looks like every programmer at some stage of her\/his life starts to write a NN library in her\/his favourite programming language.\n  * Implement it using low-level acceleration libraries\u2014Metal Performance Shaders, or BNNS.\n  * Implement it using some general-purpose NN framework\u2014Keras, TensorFlow, PyTorch, and so on\u2014and then convert it to Core ML format.\n\nThe Metal Performance Shader library includes three types of activations for NNs: ReLU, sigmoid, and TanH (`MPSCNNNeuronReLU`, `MPSCNNNeuronSigmoid`, `MPSCNNNeuronTanH`). For more information refer to: https:\/\/developer.apple.com\/reference\/metalperformanceshaders.\n\n# Training the network\n\nThe most common way to train NNs these days is with a backward propagation of errors algorithm, or backpropagation (often _backprop_ for short). As we have seen already, individual neurons remind us of linear or logistic regression a lot, so it should not come as a surprise that backpropagation usually comes together with our old friend the gradient descent algorithm. NN training works in the following way:\n\n  * Forward pass\u2014input is presented to the layer and the transformations are applied to it layer by layer until the prediction is outputted on the last layer.\n  * Loss computation\u2014the prediction is compared to the ground truth, and an error value is calculated for each neuron of the output layer using the loss function _J_.\n  * The errors are then propagated backward (backpropagation), such that each neuron has an error associated to it, proportional to its contribution to the output.\n  * Weights ( _w)_ are updated using one step of gradient descent. The gradient of the loss function   is calculated for each neuron, with respect to its weights using the error value. Then the usual gradient descent step happens as in linear regression.\n\nBackpropagation is possible only if all the transformations in the forward propagation are differentiable (in the simplest case, dot products and activation functions) because it is essentially an application of a chain rule from calculus.\n\nFor further reading, go to https:\/\/en.wikipedia.org\/wiki\/back propagation.\n\n# Vanishing gradient problem\n\nThe sigmoid asymptotically approaches zero on the one end, and the _1_ on another end. On those tails, the derivative of a function is very small. This is bad news for the backpropagation algorithm because these almost-zero values are killing the signal when it propagates through the network back to update weights.\n\nThe problem with dead neurons: if you initialize network weights at random, sigmoidal neurons with large weights would be dead (almost not transmitting the signal) from the very beginning.\n\n# Seeing biological analogies\n\nEveryone has heard that artificial NNs mimic the way the brain works. This is actually far from the truth. What is true is that NNs as a field grew out of attempts to simulate how the brain works. The elementary unit of a brain is a neuron (a nerve cell). The human brain contains approximately 86 billion neurons. Neurons can generate electric potential (action potential) in its body. The neuron has branched projections of two types. One being short projections, known as dendrites (from Greek _\u03b4\u03b5v\u03b5\u03b4pov_ , tree). Usually, their function is to receive electrical impulses from other neurons. The other type is longer projections, known as axons (from Greek \u03b1\u03be\u03c9v, axis). Some neurons don't have axons, but no neurons have more than one. The function of the axon is to carry electrical impulses away from the neuron body to other cells.\n\nBy its axon, the neuron connects to the bodies (or to the dendrites) of other neurons and transmits electrical signals to them. Not all neurons transmit signals to other neurons; some of them excite muscles and glands.\n\nAxons on their ends have structures called synapses\u2014a connection to the cell body, or dendrite. To transmit a signal to the next neuron, the synapse emits chemical neuromediators (or rarely, electrical signals). There are about 1,014-1,015 synapses in the human brain. You can calculate yourself how much disk space is required to store such enormous amount of information. In our artificial brains, we have much fewer artificial neurons. Even the largest among the modern NNs are comparable rather to the brain of a jellyfish or a snail. However, number of neurons or synapses is not the whole story, as there are some animals whose brain contains more neurons than a human's. If you're interested in the topic, visit the Wikipedia page: <https:\/\/en.wikipedia.org\/wiki\/List_of_animals_by_number_of_neurons>.\n\nEven though the concept of NNs was borrowed from biology, one should have a powerful imagination to see how a biological prototype is similar to modern artificial NNs. That's why some researchers believe that some other names, like _computational graphs_ , are more appropriate. Biological terminology was once more popular in this domain, but now literally the only biological term that is widely used is _neuron_.\n\n# Basic neural network subroutines (BNNS)\n\nBNNS is a submodule of Accelerate, containing convolution NN primitives optimized for running inference on CPU. It was introduced in iOS 10 and macOS 10.12. Note that it contains only functions for inference, not for training.\n\nThe motivation behind this library was to provide unified API for common routines, such that app developers wouldn't need to re-implement convolutions and other primitives from scratch every time (which is hard, as we have seen already in the chapter on CNNs). In a typical CNN, most energy is spent in the convolution layers. Fully connected layers are more expensive computationally, but usually CNNs contain one or a few of them at the very end so convolutions still consume about 70% of energy. That's why it's important to have highly-optimized convolution layers. Unlike MPS, CNN is available on iOS, macOS, tvOS, and even watchOS. So, if you want to run deep learning on a TV set or your watch (just because you can), this is your tool of choice.\n\nSpeaking more seriously, BNNS is useful when you're implementing NNs for devices without Metal Performance Shaders support (older iOS devices and all macOS devices for now). In all other cases, you still want to use MPS CNNs to harness GPU massive parallelism.\n\nTo check availability of Metal features, look through <https:\/\/developer.apple.com\/metal\/availability\/>.\n\nBNNS contains three types of layers: convolution, pooling, and fully-connected layers, and several activations: identity, rectified linear, leaky rectified linear, sigmoid, tanh, scaled tanh, and abs.\n\n# BNNS example\n\nIn the following example, input images are of size 224 x 224 x 64 and output images are of size 222 x 222 x 96. The dimensionality of convolution weights is 3 x 3 x 64 x 96. That's 5.45 billion floating-point operations (gigaFloPS). In a whole MNIST recognition network, it's about 1-2 trillion operations per forward pass.\n\nBNNS is a part of Accelerate, so you need to import Accelerate to access the neural networks building blocks. The first thing you do is describing the input stack:\n\n    var inputStack = BNNSImageStackDescriptor(  \n      width: 224, height: 224, channels: 64,  \n      row_stride: 224, image_stride: 224*224,  \n      data_type: BNNSDataTypeFloat32,  \n      data_scale: 1.0, data_bias: 0.0)\n\nMost of the parameters are self-evident; `row_stride` is an increment to the next row in pixels, `image_stride` is similarly increment to the next channel in pixels, and `data_type` is a type of storage.\n\nThe output stack should look similar:\n\n    var outputStack = BNNSImageStackDescriptor(  \n      width: 1, height: 10, channels: 1,  \n      row_stride: 1, image_stride: 10,  \n      data_type: BNNSDataTypeFloat32,  \n      data_scale: 1.0, data_bias: 0.0)\n\nNow let's create a convolution layer. `BNNSConvolutionLayerParameters` contains the description of the convolution layer:\n\n    let activation = BNNSActivation(function: BNNSActivationFunctionIdentity, alpha: 0, beta: 0)\n\n    var convolutionParameters = BNNSConvolutionLayerParameters(  \n      x_stride: 1, y_stride: 1,  \n      x_padding: 0, y_padding: 0,  \n      k_width: 3, k_height: 3,  \n      in_channels: 64, out_channels: 96,  \n      weights: convolutionWeights,  \n      bias: convolutionBias,  \n      activation: activation)\n\n`k_width` and `k_height` are kernel width and height respectively.\n\nCreating the layer itself:\n\n    let convolutionLayer = BNNSFilterCreateConvolutionLayer(&inputStack, &outputStack, &convolutionParameters, nil)\n\n`nil` is for default `BNNSFilterParameters`.\n\nNow you can use the filter and destroy it when it's not needed anymore by calling `BNNSFilterDestroy(convolutionLayer)`.\n\nPooling layer:\n\n    \/\/ Describe pooling layer  \n    BNNSPoolingLayerParameters pool = {  \n        .k_width = 3,  \n    \/\/ kernel height  \n    \/\/ kernel width  \n    \/\/ X padding  \n    \/\/ Y padding  \n        .k_height = 3,  \n        .x_padding = 1,  \n        .y_padding = 1,  \n        .x_stride = 2,  \n        .y_stride = 2,  \n        .in_channels = 64,  \n        .out_channels = 64,  \n        .pooling_function = BNNSPoolingFunctionMax  \/\/ pooling function  \n    };  \n    \/\/ Create pooling layer filter  \n    BNNSFilter filter = BNNSFilterCreatePoolingLayer(  \n        &in_stack,     \/\/ BNNSImageStackDescriptor for input stack  \n        &out_stack,    \/\/ BNNSImageStackDescriptor for output stack  \n        &pool,         \/\/ BNNSPoolingLayerParameters  \n        NULL);         \/\/ BNNSFilterParameters (NULL = defaults)  \n    \/\/ Use the filter ...  \n    \/\/ Destroy filter  \n    BNNSFilterDestroy(filter);\n\n    \/\/ Describe input vector  \n    BNNSVectorDescriptor in_vec = {  \n        .size = 3000,  \n    \/\/ size  \n    \/\/ storage type  \n    };  \n    \/\/ Describe fully connected layer  \n    BNNSFullyConnectedLayerParameters full = {  \n        .in_size = 3000,  \n        .out_size = 20000,  \n        .weights = {  \n            .data_type = BNNSDataTypeFloat16,  \n            .data = weights  \n    \/\/ input vector size  \n    \/\/ output vector size  \n    \/\/ weights storage type  \n    \/\/ pointer to weights data  \n    } };  \n    \/\/ Create fully connected layer filter  \n    BNNSFilter filter = BNNSFilterCreateFullyConnectedLayer(  \n        &in_vec,       \/\/ BNNSVectorDescriptor for input vector  \n        &out_vec,      \/\/ BNNSVectorDescriptor for output vector  \n        &full,         \/\/ BNNSFullyConnectedLayerParameters  \n        NULL);         \/\/ BNNSFilterParameters (NULL = defaults)\/\/ Use the filter ...  \n    \/\/ Destroy filter  \n    BNNSFilterDestroy(filter);  \n    \/\/ Apply filter to one pair of (in,out)  \n    int status = BNNSFilterApply(filter,  \n                                 in,  \n    out);  \n    \/\/ BNNSFilter  \n    \/\/ pointer to input data  \n    \/\/ pointer to output data  \n    \/\/ Apply filter to N pairs of (in,out)  \n    int status = BNNSFilterApplyBatch(filter,  \n                                      20,  \n                                      in,  \n                                      3000,  \n                                      out,  \n                                      20000);  \n    \/\/ BNNSFilter  \n    \/\/ batch size (N)  \n    \/\/ pointer to input data  \n    \/\/ input stride (values)  \n    \/\/ pointer to output data  \n    \/\/ output stride (values)\n\n# Summary\n\nIn this chapter, we've become acquainted with artificial NNs and their main components. NNs are built from neurons that are usually organized in layers. A typical neuron performs a weighted sum of inputs and then applies a non-linear activation function on it to calculate its output. There are many different activation functions, but the most popular these days is ReLU and its modifications, due to their computational properties.\n\nNNs are usually trained using the backpropagation algorithm, built on top of stochastic gradient descent. Feed-forward NNs with several layers are also known as multilayer perceptrons. MLPs can be used for classification tasks.\n\nIn the next chapter, we'll continue to discuss NNs, but this time we'll focus on convolution NNs, which are especially popular in the computer vision domain.\n\n# Convolutional Neural Networks\n\nIn this chapter, we are discussing the **convolutional neural networks** ( **CNNs** ). At first we are going to discuss all components with examples in Swift just to develop an intuition about the algorithm and what is going on under the hood. However, in the real life you most likely will not develop CNN from scratch, because you will use some ready available and battle-tested deep learning framework.\n\nSo, in the second part of the chapter we will show a full development cycle of deep learning mobile application. We are going to take the photos of people's faces labeled with their emotions, train a CNN on a GPU workstation, and then integrate it into an iOS application using Keras, Vision, and Core ML frameworks.\n\nTo the end of this chapter you will have learned about:\n\n  * Affective computing\n  * Computer vision, its tasks, and its methods\n  * CNNs, their anatomy, and core concepts behind them\n  * Applications of CNNs in computer vision\n  * How to train CNNs using a GPU workstation and Keras\n  * Deep learning tricks: Regularization, data augmentation, and early stopping\n  * CNNs architectures\n  * How to convert a trained model to Core ML format for use in an iOS application\n  * How to detect faces and facial expressions in photos using CNNs and Vision framework\n\n# Understanding users emotions\n\nWhile voice input is undoubtedly a useful feature, we all well know how the actual meaning of the sentence can be opposite to the literal one, depending on the speaker's intonation, facial expression, and context. Try this simple sentence: _Oh, really?_ Depending on the conditions, this can mean: _I doubt_ , _I didn't know_ , _I'm impressed_ , _I don't care_ , _This is obvious_ , and so on. The problem is that speech is not the only mode of conversation for human beings, and that's why much research is focused these days on _teaching_ computers to understand (and also simulate) gestures, facial expressions, sentiments in a text, eye movements, sarcasm, and other affect manifestations. An interdisciplinary field that emerges around the question of emotional and compassionate AI is known as **affective computing**. It integrates knowledge from the computer and cognitive sciences, as well as psychology and robotics. The aim is the creation of computer systems that will adapt themselves to the user's emotional state, understand their mood, and simulate empathy. Back in 1995, in pre-smartphone epoch, the name of the field was coined by Rosalind Picard. In her technical report titled _Affective Computing_ [2], she predicts that it will be especially relevant in the context of wearable devices. Using facial expression recognition in this chapter, we're going to introduce elements of affective intelligence into our mobile app. This can be used in a context of language understanding, or in a multitude of other ways from emoji recommendations to smart photo sorting.\n\nPlease note that affective computing is similar to the sentiment analysis, but is a more broad term: the former is interested in all kind of affects, their detection and simulation, and the latter is mostly concerned with the polarity of the text piece (positive\/negative).\n\n# Introducing computer vision problems\n\nIn this book, we mentioned computer vision several times, but since this chapter is focused on this particular domain, we will look at it in more detail now. There are several practical tasks related to image and video processing, which are referred to as **computer vision domain**. While working on some computer vision task, it's important to know these names, to be able to find what you need in the vast ocean of computer vision publications:\n\n  * **Object recognition** : The same as classification. Assigning labels to the images. _This is a cat_. Age estimation. Facial expression recognition.\n  * **Object localization** : Finding frame of object in the image. _The cat is in this frame_.\n  * **Object detection** : Finding frames of objects in the image. _The cat is in this frame_.\n  * **Semantic segmentation** : Each point in the picture is assigned to one class. If the picture contains several cats, each cat's pixel would be assigned to the _cat_ class.\n  * **Instance segmentation** : Each point in the picture is assigned to one instance of class. If the picture contains several cats, each cat's pixel would be assigned to the separate segment of _cat_ class.\n  * **Pose estimation** : Determining the orientation of the object in the space.\n  * **Object tracking** : Analyzing the video to find the trajectory of the moving object.\n  * **Image segmentation** : Finding borders between different objects into an image. Background subtraction.\n  * 3D-scene restoration and depth estimation.\n  * Image search and retrieval.\n\nSome common computer vision tasks, like **optical character recognition** ( **OCR** ), consist of several steps; for example, _image segmentation \u2192 image recognition_ :\n\nFigure 9.1: Popular computer vision tasks. Top row: recognition, localization. Middle row: object detection, pose estimation. Bottom row: semantic segmentation and instance segmentation\n\nThese tasks are recognized as hard problems, because of the variable factors: different camera position, lighting, object's occlusion, intra-class variability, changes in object shape, and so on. Many familiar machine learning algorithms found their unexpected applications in computer vision. For example, we have already seen that _k_ -means can be used for image segmentation, and the extension of linear regression RANSAC for stitching photos into a panorama.\n\nCNNs\u2014historical background: **  \n**For many years, the progress in computer vision has been slow, arduous and involved a lot of domain expert knowledge, manual feature picking and model parameters tuning. Significant changes crept up unnoticed: In 2012, Alex Krizhevsky won the annual ImageNet image recognition competition, leaving the rest of the competitors far behind. For his classifier, he used then little known CNN (AlexNet architecture). What was even more surprising is the fact that CNNs were proposed at least as early as 1994, when Yan LeCunn published his LeNet5 architecture description for handwritten numbers recognition. But they were recognized impractical for most real-world tasks, because required almost eternity and tons of data to learn anything useful. The novelty was that Krizhevsky trained his network using graphics accelerators (GPUs) and not a CPU. Harnessing massive parallelism of these devices he reduced training time from weeks to hours. The result was sensational, and therefore the convolutional networks gained popularity very quickly in the community of researchers and practitioners.\n\nLet's take a closer look at this type of neural network.\n\n# Introducing convolutional neural networks\n\nCNNs, or ConvNets have gotten a lot of attention in the last few years, mainly due to their major successes in the domain of computer vision. They are at the core of most computer vision systems nowadays, including self-driving cars and large-scale photo classification systems.\n\nIn some sense, CNNs are very similar to multilayer perceptron, which we have discussed in the previous chapter. These networks also build from the layers, but unlike MLP, which usually has all layers similar to each other, CNNs usually include many layers of different types. And the most important type of the layer is (surprise, surprise) the convolutional layer. Modern CNNs can be really deep\u2014hundreds of different layers. Nevertheless, you can still see the whole network as one differentiable function that takes some input (usually raw values of image pixels), and produces some output (for example, class probabilities: 0.8 cat, 0.2 dog).\n\n# Pooling operation\n\nPooling or subsampling is a simple operation of input size decreasing ( _Figure 9.2_ ). If you have a black and white image, and you want to decrease its size, you can do it in the following way: chose a sliding window of size _n_ \u00d7 _m_ and stride _s_. Go through the image, applying sliding window and shifting on the _s_ pixels every time you want to move your window. At each position calculate an average (for average pooling) or maximum (for max pooling) and record this value into the destination matrix. Now, there are two common ways to handle borders of the image:\n\nFigure 9.2. Pooling operation. Grey window in the source image corresponds to the grey cell in the destination image\n\nThe pooling is used in the CNNs to reduce the size of the data, as it travels down the network.\n\n# Convolution operation\n\nConvolution is one of the most important operations in the image processing. Blurring, sharpening, edge detection, denoising, embossing and many other familiar operations in image editors are actually convolutions. It is similar to the pooling operation in some way, because it is also a sliding window operation, but instead of taking the average over the window, it performs element-wise multiplication by the kernel \u2013 matrix of size n \u00d7 n and sums the result. The result of the operation depends on the kernel (also known as **convolution filter** ) \u2013 a matrix, which is usually square, but not necessarily, see _Figure 9.3_. The notions of the stride and padding are the same as in the pooling case:\n\nFigure 9.3: Different convolution filters have different effects on the picture\n\nConvolution operation works in the following way (see the following diagram):\n\n  * The convolution kernel (filter) slides over the image from left to right, and from top to bottom\n  * At each position, we calculate an element-wise product of filter and the patch of the image, which is covered by a filter on this step\n  * The elements of the resulting matrix are summed up\n  * The result of the convolution is a matrix composed of the sums at each position of the filter:\n\nFigure 9.4: Convolution operation with 3 x 3 kernel, stride 1 and valid padding: : the source image is getting split into windows; each window is multiplied by filter elementwise; sum of the values into each of the windows\n\nThe algorithm looks simple at first glance, so you probably can come up with a Swift implementation similar to this one:\n\n    let input = ... \/\/source image, 2D array  \n    var output = ... \/\/destination image, 2D array  \n    for i in 0..<imageHeight {  \n         for j in 0..<imageWidth {  \n             var accumulator = 0;  \n             for ik in 0..<kernelHeight {  \n                 for jk in 0..<kernelWidth {  \n                     accumulator += kernel[ik][jk] *  \n                     input[i+ik-kernelHeight\/2][j+jk-kernelWidth\/2]  \n                 }   \n             }  \n             output[i][j] = accumulator;  \n         }  \n    }\n\nHowever, you really shouldn't implement your own convolution operation in production code. Here is a non-exhaustive list of problems that you'd have to deal with:\n\n  * Handle the edges of the image properly (see the following diagram)\n  * Handle stride\u2014how much the kernel shifts after each step\n  * Handle integer overflow properly\n  * Optimize for speed\n\nA good convolution implementation may require hundreds of lines of code (and even more). iOS SDK provides developers with several APIs that they can choose from, depending on their needs and goals:\n\n  * For image filtering: \n    * `CIImageFilter` from Core Image framework\n    * `vImage convolution` functions from accelerate framework, for instance `vImageConvolve_ARGB8888()`\n  * For running CNNs: \n    * **For CPU execution** : BNNS convolution layer: `BNNSFilterCreateConvolutionLayer`\n    * **For GPU execution** : `MPSCNNConvolution` class from Metal Performance Shaders framework\n\n`vImage functions` is a recommended way for the processing of large images, or for repeating the operation many times in a row. To filter middle-sized images, it's more efficient to stick with CoreImage. When you implement CNNs, you usually want to use either the Metal or accelerate framework. This is something we will discuss in much detail in the Chapter 11, _Machine Learning Libraries_.\n\nCheck `Convolution.playground` in the supplementary code to see how different convolution filters produce different effects. Try your own filters, and see what happens. We are using the `vImageConvolve_ARGB8888()` function from the accelerate framework to apply filters.\n\nThe following is an example:\n\n    import Accelerate\n\n    let error = vImageConvolve_ARGB8888(&input, &output, nil, 0, 0, kernel, kernelHeight, kernelWidth, Int32(divisor), nil, flags)\n\nFigure 9.5: Multichannel convolution\n\n# Convolutions in CNNs\n\nThe first layer applies the filters to the input image, and sends the resulting feature matrices downstream. The next layers apply their filters to the input they get, extracting features from a higher and higher level (see _Figure 9.6 - 9.8_ ). In the task of image classification (or _object recognition_ ), CNNs gradually adjust the filters during the training to extract useful features from the picture. In this case, useful features are different patterns, like eyes, beaks, wheels, and so on:\n\nFigure 9.6: Each CNN layer learns a set of convolutional kernels. Here you see outputs of the different filters from the 2nd convolutional layer in the 2nd block of VGG 16\n\nThe interesting thing is that convolution neural networks are used extensively outside of the computer vision domain; for instance, in natural language modeling and speech recognition. This is because of their capability to extract meaningful features from the raw data. There is no fundamental difference between extracting edges and corners from images, and syntactic features from texts.\n\n# Building the network\n\nWhen you first encounter a variety of architectures of CNNs, you feel overwhelmed by the abundance of new terms, different layers, and their hyperparameters. In fact, at the moment, only a few architectures have found broad application, and the number of designs suitable for mobile development is even smaller.\n\nThere are five basic types of layers plus an input layer, which usually does nothing except passing data forward:\n\n  * **Input layer** : The first layer in the neural network. It does nothing, only takes the input and passes it downstream.\n  * **Convolution layers** : Where convolutions happen\n  * **Fully** : Connected or dense layers\n  * **Nonlinearity layers** : These are layers which apply activation functions to the output of the previous layer: sigmoid, ReLU, tanh, softmax and so on.\n  * **Pooling layers** : Downsample their input.\n  * **Regularization layers** : layers to fight an overfitting.\n\nModern deep learning frameworks contain much more different types of layers for all needs, but these are the most commonly used. In the following sections, we will discuss each type of layer in details. As CNNs are normally not trained on the mobile devices, but used for predictions only, we do not go into the details of the backpropagation in CNNs here. However, if you are interested in knowing more details, the `NN Demo.playground` in the supplementary materials of this chapter contains pure Swift implementation of CNN and all its classical layers. It also contains an implementation of stochastic gradient descent algorithm and several of its variants. Please refer to this playground for more technical details on the layer's gears and training algorithm. The playground is a Swift port of `ConvNetJS` library by Andrej Karpathy. It was developed solely for the study purposes, to demonstrate how CNNs work inside, not for the use in the real applications, and it does not utilize GPU or CPU acceleration. The original JS library can be found here: <https:\/\/cs.stanford.edu\/people\/karpathy\/convnetjs\/>\n\n# Input layer\n\nThis is a dummy layer; it does nothing in both forward and backward passes. We only use it to define the input tensor size.\n\n# Convolutional layer\n\nIn CNNs convolutions happen in the special layers, called **convolutional layers**. Each layer has an array of convolutional filters, which also can be seen as one 3D convolutional filter with width, height, and number of channels (or depth). In the first convolutional layer, we usually want to have 3 channels corresponding to the RGB channels of the input images:\n\nFigure 9.7: The first convolutional layer takes a batch of images as its input and outputs a batch of feature maps. The results of each of the 16 red, green and blue filters are being summed up to obtain the final feature map\n\nThe output of a convolution layer is called a **feature map** , because it shows where specific features are located in the input image. Note, that only the first convolutional layer takes an image as its input, all subsequent layers take outputs of their predecessors (feature maps) as their inputs. Those feature maps are stored as tensors.\n\nTensor in the context of deep learning is a multidimensional array. Neural network parameters, for example convolutional filters, stored as tensors and all data travels through a deep neural network in the form of tensors. 0-dimensional tensor is a scalar, 1D is a vector, 2D is a matrix, 3D sometimes called **volume**.\n\nFigure 9.8: Each next layer in CNN extracts more abstract features, then the previous one. Example taken from the VGG-16 network.\n\nNote, that for your production apps, you usually don't want to write your own convolutional layers in Swift, because you want to utilize the power of the GPU, so you normally use the existing deep learning libraries (see Chapter 10, _Natural Language Processing_ ) or implement custom layers in Metal or Accelerate (see Chapter 11, _Machine Learning Libraries_ ).\n\n# Fully-connected layers\n\nFully-collected layer is like one layer of a multilayer perceptron from the previous chapter, but without activation function. You can imagine it as a matrix of weights multiplied by an input or as a layer of artificial neurons ( _Figure 9.9_ ):\n\nFigure 9.9: Two ways to represent a fully-connected layer: in a form of matrix-vector multiplication and in a form of a graph.\n\n# Nonlinearity layers\n\nThis are all kind of nonlinearities, that we've already discussed in the previous chapter: tanh, sigmoid, ReLU, and so on. You usually want to put them after a convolutional or a fully-connected layer.\n\nSoftmax is a generalization of a logistic function to vectors: while the logistic function `squashes` scalar values to be between 0 and 1, softmax squashes vectors so that its elements adds up to 1. In the statistics, probability of outcomes in discrete random distribution adds up to 1, so this function is really useful for the classification, where target variable is discrete.\n\n# Pooling layer\n\nPooling layer performs pooling operation. Put it after a convolutional layer, when you want to reduce the size of the tensor which will be passed to the next layer.\n\n# Regularization layers\n\nRegularization layers intended to fight overfitting and increase the speed of training. Among popular regularization layers are dropout and batch normalization layers, because both techniques were shown to be very useful in practice.\n\n# Dropout\n\nDropout is a common way of regularization for deep neural networks. The idea is to turn off random neurons in the previous layer with some predefined probability on each step of the training. The neurons which were turned off are not trained during this step, but will be restored on the next one with the original weights. This technique prevents overfitting because it does not allow to train all the neurons on all the data.\n\n# Batch normalization\n\nSmall changes in the layer parameters affects all the following layer inputs, and the effect gets amplified with each next layer. This is especially problematic for the deep networks.  \nThe distribution of inputs to each layer changes during training, because parameters of the previous layer are being adjusted. This problem is known as **internal covariate shift**.  \nBatch normalization technique was proposed in 2015 by Sergey Ioffe and Christian Szegedy from Google [1] to fix the problem. It allows normalizing layer inputs for each mini-batch as part of the network architecture. Batch normalization layer is usually inserted between the dot product and nonlinearity.\n\nThe benefits are as follows:\n\n  * You can use higher learning rate\n  * You can be less careful about weights initialization\n  * Works as regularization - no need for dropout\n  * Same model trains 14 times faster\n\nCovariate shift:\n\nThe common problem in machine learning systems formally known as **covariate shift** : when the model is being deployed to the production environment, it appears, that the data distribution in it is different from the distribution of the training data. The name comes from the covariates, which are basically the same as features. By analogy, the notion of internal covariate shift was introduced: when in the neural network the distribution of input data to each layer is not stable, but changes significantly after each step of the SGD.\n\nIf both input and output distributions changes, this is known as **dataset shift**.\n\n# Loss functions\n\nLoss function is a necessary part, because it is what we want to minimize during the training. You can find a few popular loss functions in the table:\n\n**Name** | **Formula** | **Usually used for**  \n---|---|---  \nMean squared error or L2-loss |   | Regression  \nMean absolute error or L1-loss |   | Regression  \nCategorical cross entropy |   |\n\nSoftmax multiclass classification\n\nWhere _y_ is a ground-truth vector and _\u0177_ is a vector of predictions of length _n_.\n\n# Training the network\n\n**Stochastic gradient descent** ( **SGD** ) is an effective way of training deep neural networks. SGD seeks such parameters _\u0398_ of the network, which minimize the loss function _\u2112_.\n\nWhere   is a training dataset.\n\nTraining happens in steps. At every step, we choose a subset of our training set of size m (mini-batch) and use it to approximate loss function gradient with respect to parameters _\u0398_ :\n\nMini-batch training advantages are as follows:\n\n  * Gradient of the loss function over a mini-batch is a better approximation of the gradient over the whole training set then calculated over only one sample\n  * Thanks to the GPU you can perform computations in parallel on every sample in the batch, which is faster, then processing them one-by-one\n\n# Training the CNN for facial expression recognition\n\nFor the demonstration of the CNNs we will implement a simple neural network for emotion recognition. We will use the dataset of face expressions `fer2013` from the ICML 2013 contest _Facial Expression Recognition Challenge_ [1].\n\nThe dataset can be downloaded from the kaggle site:\n\n<https:\/\/www.kaggle.com\/c\/challenges-in-representation-learning-facial-expression-recognition-challenge\/data>\n\nYou will be asked to register and accept the terms and conditions.\n\nThe archive `fer2013.tar.gz` contains `fer2013.csv` with the dataset itself and some supplementary information files. The `.csv` file contains 35,887 samples, of which 28,709 marked as training set, 3,589 as public test, and 3,589 private test. There are three columns in the table: emotion, pixels and usage. Every sample is a grayscale 48 \u00d7 48 pixels face photo in a form of pixel array. The faces were cropped in an automatic way, so there are some false-positives in the dataset (non-faces and cartoon faces). Each face is labeled as belonging to one of the 7 classes. The distribution of emotions in the dataset is as follows:\n\n**Class id** | 0 | 1 | 2 | 3 | 4 | 5 | 6  \n---|---|---|---|---|---|---|---  \n**Emotion** | Angry | Disgust | Fear | Happy  | Sad | Surprise | Neutral  \n**Count** | 4953 | 547 | 5121 | 8989 | 6077 | 4002 | 6198\n\n# Environment setup\n\nTo train the deep CNN, you will need a computer with a CUDA-compatible GPU. I used an Ubuntu 16.x machine with NVidia GTX980 GPU for model training, and a macOS machine to convert the model to Core ML format. If you don't have CUDA-compatible GPU, you can try to train the model on CPU; but be aware that this will take a lot of time. Also, the trained model for this chapter is available in the supplementary materials, so if you prefer not to contribute to the global warming by retraining the model from scratch, it's also possible.\n\nHere's a list of what should be installed on your system to train the network:\n\n  * Latest NVIDIA drivers\n  * CUDA 8.0\n  * cuDNN 5.1\n  * Python 2.7\n  * `tensorflow-gpu` (or TensorFlow for CPU-only mode)\n  * Keras\n  * Keras-viz\n  * Matplotlib, Pandas\n\nPlease, refer to the official sites for the installation instructions.\n\n# Deep learning frameworks\n\nThere are a plenty of deep learning toolkits and libraries for different kinds of platforms. For a long time, the three most popular of them were Theano (Python), Torch (Lua), and Caffe (C++). Somehow, Caffe became an industrial standard, while Theano and Torch were mostly used among researchers. I call these three libraries the first generation of deep learning frameworks. Most of the pre-trained neural networks that are available on the internet are still in Caffe format. They had their own problems, so the next generation of frameworks followed in several years. If the first generation was created mainly by efforts of individual researchers, the second generation was pushed by big IT companies. Today, apart from Apple, every internet giant has its own open source deep learning framework: Google has TensorFlow and Keras, Microsoft has CNTK, Facebook released Caffe 2, and Torch was reborn as PyTorch, thanks to Twitter and Facebook. Amazon has chosen MXNet as its deep learning framework of choice at AWS. Which one should you choose for your deep learning projects? At the moment, the best iOS support is provided by Caffe 2 and TensorFlow frameworks. With the release of Core ML we've also got an easy way to convert models trained in Caffe and Keras to Apple in `ml` model format. In this chapter, we're using Keras for our CNNs.\n\nA side note: Apple's Metal 2 also contains many primitives for building deep learning neural networks, but it's hard to call it a deep learning framework, most importantly because it doesn't support training neural networks.\n\n# Keras\n\nKeras is a popular Python package for building the deep learning neural networks. It has a user-friendly syntax. It's easy and fast to prototype and build your deep models in it. It started as a facade for the Theano symbolic computation library, but over time, it has also developed a TensorFlow backend, and so finally became a part of TensorFlow. So now, TensorFlow is a default backend, but you still have an option to switch back to Theano. There are also work-in-progress projects of MXNet and CNTK backends.\n\nKeras contains functions for pre-processing of most common data types: images, texts, and time series.\n\nCore ML supports convolution and recurrent neural networks built in Keras.\n\nOfficial website of Keras: <https:\/\/keras.io\/>\n\n# Loading the data\n\nAs usual, first we add some magic to display images inline in the Jupyter:\n\n    %matplotlib inline\n\nWe're using Pandas to handle our data:\n\n    import pandas\n\nPlease, visit the Kaggle site and download the dataset: <https:\/\/www.kaggle.com\/c\/challenges-in-representation-learning-facial-expression-recognition-challenge>\n\nLoad the dataset into the memory:\n\n    data = pandas.read_csv(\"fer2013\/fer2013.csv\")\n\nDataset consists of gray scale face photos encoded as pixel intensities. 48 x 48 gives 2304 pixels for each. Every image is marked according to the emotion on the face.\n\n    data.head() \n    emotion  pixels   Usage \n    0  0  70 80 82 72 58 58 60 63 54 58 60 48 89 115 121...  Training \n    1  0  151 150 147 155 148 133 111 140 170 174 182 15...  Training \n    2  2  231 212 156 164 174 138 161 173 182 200 106 38...  Training \n    3  4  24 32 36 30 32 23 19 20 30 41 21 22 32 34 21 1...  Training \n    4  6  4 0 0 0 0 0 0 0 0 0 0 0 3 15 23 28 48 50 58 84...  Training \n    How many faces of each class do we have?\n\n    data.emotion.value_counts() \n    3    8989 \n    6    6198 \n    4    6077 \n    2    5121 \n    0    4953 \n    5    4002 \n    1     547 \n    Name: emotion, dtype: int64\n\nHere 0=Angry, 1=Disgust, 2=Fear, 3=Happy, 4=Sad, 5=Surprise, and 6=Neutral.\n\nLet's remove `Disgust`, as we have too little samples for it:\n\n    data = data[data.emotion != 1] \n    data.loc[data.emotion > 1, \"emotion\"] -= 1 \n    data.emotion.value_counts() \n    2    8989 \n    5    6198 \n    3    6077 \n    1    5121 \n    0    4953 \n    4    4002 \n    Name: emotion, dtype: int64 \n    emotion_labels = [\"Angry\", \"Fear\", \"Happy\", \"Sad\", \"Surprise\", \"Neutral\"] \n    num_classes = 6\n\nThis is how samples are distributed among training and test. We'll be using training to train the model and everything else will go to test set:\n\n    data.Usage.value_counts() \n    Training       28273 \n    PrivateTest     3534 \n    PublicTest      3533 \n    Name: Usage, dtype: int64\n\nThe size of images and the number of channels (depth):\n\n    from math import sqrt \n    depth = 1 \n    height = int(sqrt(len(data.pixels[0].split()))) \n    width = int(height) \n    height \n    48\n\nLet's see some faces:\n\n    import numpy as np \n    import scipy.misc \n    from IPython.display import display \n    for i in xrange(0, 5): \n        array = np.mat(data.pixels[i]).reshape(48, 48) \n        image = scipy.misc.toimage(array, cmin=0.0) \n        display(image) \n        print(emotion_labels[data.emotion[i]])\n\n    \/\/Images are being shown in the notebook\n\nMany faces have ambiguous expressions, so our neural network will have a hard time classifying them. For example, the first face looks surprised or sad, rather than angry, and the second face doesn't look angry at all. Nevertheless, this is the dataset we have. For the real application, I would recommend collecting more samples of higher resolution, and then annotating them such that every photo is annotated several times by different independent annotators. Then, remove all photos that were annotated ambiguously.\n\n# Splitting the data\n\nDo not forget to split your data into training and test sets before training the model as shown in the following:\n\n    train_set = data[(data.Usage == 'Training')] \n    test_set = data[(data.Usage != 'Training')] \n    X_train = np.array(map(str.split, train_set.pixels), np.float32) \n    X_test = np.array(map(str.split, test_set.pixels), np.float32) \n    (X_train.shape, X_test.shape) \n    ((28273, 2304), (7067, 2304)) \n    48*48 \n    2304 \n    X_train = X_train.reshape(28273, 48, 48, 1) \n    X_test = X_test.reshape(7067, 48, 48, 1) \n    (X_train.shape, X_test.shape) \n    ((28273, 48, 48, 1), (7067, 48, 48, 1)) \n    num_train = X_train.shape[0] \n    num_test = X_test.shape[0] \n    (num_train, num_test) \n    (28273, 7067)\n\nConverting labels to categorical:\n\n    from keras.utils import np_utils # utilities for one-hot encoding of ground truth values \n    Using TensorFlow backend. \n    y_train = train_set.emotion \n    y_train = np_utils.to_categorical(y_train, num_classes) \n    y_test = test_set.emotion \n    y_test = np_utils.to_categorical(y_test, num_classes) \n\n# Data augmentation\n\nIn the deep learning applications, generally, the more data you have, the better. Deep neural networks usually have a lot of parameters, so on the small datasets they overfit easily. We can generate more training samples from the samples we already have by using the technique called **data augmentation**. The idea is to change samples at random. With the face photos, we could, for example, flip faces horizontally, shift them a bit, or add some rotations:\n\n    from keras.preprocessing.image import ImageDataGenerator  \n    datagen = ImageDataGenerator( \n        rotation_range=25, \n        width_shift_range=0.2, \n        height_shift_range=0.2, \n        horizontal_flip=True)\n\nCompute quantities required for featurewise normalization (std, mean, and principal components, if ZCA whitening is applied):\n\n     datagen.fit(X_train)  \n     batch_size = 32\n\nAt each iteration, we will consider 32 training examples at once, in other words, our batch size is 32. Let's see our images after augmentation:\n\n    from matplotlib import pyplot \n    for X_batch, y_batch in datagen.flow(X_train, y_train, batch_size=9):\n\nCreating a grid of 3 x 3 images:\n\n        for i in range(0, 9): \n            pyplot.axis('off') \n            pyplot.subplot(330 + 1 + i) \n            pyplot.imshow(X_batch[i].reshape(48, 48), cmap=pyplot.get_cmap('gray'))\n\nShowing the plot with images:\n\n        pyplot.axis('off') \n        pyplot.show() \n        break\n\n    <Images>\n\nGenerators that provide samples during training:\n\n    train_flow = datagen.flow(X_train, y_train, batch_size=batch_size) \n    test_flow = datagen.flow(X_test, y_test) \n\n# Creating the network\n\nKeras allows building the deep neural networks by adding new layers one by one. Note, that all layers should be familiar to you to this moment.\n\n    from keras.models import Sequential \n    from keras.layers import Activation, Dropout, Flatten, Dense, BatchNormalization, Conv2D, MaxPool2D \n    model = Sequential()\n\n    model.add(Conv2D(16, (3, 3), padding='same', activation='relu', input_shape=(height, width, depth))) \n    model.add(Conv2D(16, (3, 3), padding='same')) \n    model.add(BatchNormalization()) \n    model.add(Activation('relu')) \n    model.add(MaxPool2D((2,2)))\n\n    model.add(Conv2D(32, (3, 3), padding='same', activation='relu')) \n    model.add(Conv2D(32, (3, 3), padding='same')) \n    model.add(BatchNormalization()) \n    model.add(Activation('relu')) \n    model.add(MaxPool2D((2,2)))\n\n    model.add(Conv2D(64, (3, 3), padding='same', activation='relu')) \n    model.add(Conv2D(64, (3, 3), padding='same')) \n    model.add(BatchNormalization()) \n    model.add(Activation('relu')) \n    model.add(MaxPool2D((2,2)))\n\n    model.add(Flatten()) \n    model.add(Dense(128)) \n    model.add(BatchNormalization()) \n    model.add(Activation('relu')) \n    model.add(Dense(num_classes, activation='softmax')) \n    model.compile(loss='categorical_crossentropy',  \n                  optimizer='rmsprop', \n                  metrics=['accuracy'])\n\nThe list of layers can be accessed via the `layers` property of the `model` object:\n\n    model.layers \n    [<keras.layers.convolutional.Conv2D at 0x7f53b5d12fd0>, \n     <keras.layers.convolutional.Conv2D at 0x7f53b5ca2090>, \n     <keras.layers.normalization.BatchNormalization at 0x7f53b5ca2a10>, \n     <keras.layers.core.Activation at 0x7f53b5cbbe50>, \n     <keras.layers.pooling.MaxPooling2D at 0x7f53b5c68ed0>, \n     <keras.layers.convolutional.Conv2D at 0x7f53b5c68bd0>, \n     <keras.layers.convolutional.Conv2D at 0x7f53b5c8b310>, \n     <keras.layers.normalization.BatchNormalization at 0x7f53b5c3ad10>, \n     <keras.layers.core.Activation at 0x7f53b5c0e790>, \n     <keras.layers.pooling.MaxPooling2D at 0x7f53b5bd7c50>, \n     <keras.layers.convolutional.Conv2D at 0x7f53b5bbf990>, \n     <keras.layers.convolutional.Conv2D at 0x7f53b5bb1950>, \n     <keras.layers.normalization.BatchNormalization at 0x7f53b5b845d0>, \n     <keras.layers.core.Activation at 0x7f53b5b3f950>, \n     <keras.layers.pooling.MaxPooling2D at 0x7f53b5b05610>, \n     <keras.layers.core.Flatten at 0x7f53b5ae31d0>, \n     <keras.layers.core.Dense at 0x7f53b5af35d0>, \n     <keras.layers.normalization.BatchNormalization at 0x7f53b5ac6690>, \n     <keras.layers.core.Activation at 0x7f53b5a85750>, \n     <keras.layers.core.Dense at 0x7f53b5a2e910>] \n    model.summary() \n    _________________________________________________________________ \n    Layer (type)                 Output Shape              Param #    \n    ================================================================= \n    conv2d_1 (Conv2D)            (None, 48, 48, 16)        160        \n    _________________________________________________________________ \n    conv2d_2 (Conv2D)            (None, 48, 48, 16)        2320       \n    _________________________________________________________________ \n    batch_normalization_1 (Batch (None, 48, 48, 16)        64         \n    _________________________________________________________________ \n    activation_1 (Activation)    (None, 48, 48, 16)        0          \n    _________________________________________________________________ \n    max_pooling2d_1 (MaxPooling2 (None, 24, 24, 16)        0          \n    _________________________________________________________________ \n    conv2d_3 (Conv2D)            (None, 24, 24, 32)        4640       \n    _________________________________________________________________ \n    conv2d_4 (Conv2D)            (None, 24, 24, 32)        9248       \n    _________________________________________________________________ \n    batch_normalization_2 (Batch (None, 24, 24, 32)        128        \n    _________________________________________________________________ \n    activation_2 (Activation)    (None, 24, 24, 32)        0          \n    _________________________________________________________________ \n    max_pooling2d_2 (MaxPooling2 (None, 12, 12, 32)        0          \n    _________________________________________________________________ \n    conv2d_5 (Conv2D)            (None, 12, 12, 64)        18496      \n    _________________________________________________________________ \n    conv2d_6 (Conv2D)            (None, 12, 12, 64)        36928      \n    _________________________________________________________________ \n    batch_normalization_3 (Batch (None, 12, 12, 64)        256        \n    _________________________________________________________________ \n    activation_3 (Activation)    (None, 12, 12, 64)        0          \n    _________________________________________________________________ \n    max_pooling2d_3 (MaxPooling2 (None, 6, 6, 64)          0          \n    _________________________________________________________________ \n    flatten_1 (Flatten)          (None, 2304)              0          \n    _________________________________________________________________ \n    dense_1 (Dense)              (None, 128)               295040     \n    _________________________________________________________________ \n    batch_normalization_4 (Batch (None, 128)               512        \n    _________________________________________________________________ \n    activation_4 (Activation)    (None, 128)               0          \n    _________________________________________________________________ \n    dense_2 (Dense)              (None, 6)                 774        \n    ================================================================= \n    Total params: 368,566 \n    Trainable params: 368,086 \n    Non-trainable params: 480 \n    _________________________________________________________________ \n\n# Plotting the network structure\n\nPerhaps, the more convenient way to explore the structure of the network is to draw a picture. Let's do that:\n\n    from IPython.display import SVG \n    from keras.utils.vis_utils import model_to_dot\n\n    SVG(model_to_dot(model, show_shapes=True).create(prog='dot', format='svg'))\n\n    from IPython.display import Image \n    from keras.utils import plot_model \n    plot_model(model, show_shapes=True, show_layer_names=True, to_file='model.png')\n\nSee the _Figure 9.10_ for the result.\n\n# Training the network\n\nFirst, we have to define how long we want to train out network. One `epoch` is one full pass over the training set. The number of steps in the epoch depends on the batch size and the number of samples in the training set. Let's say we want to pass over the training set 100 times:\n\n    num_epochs = 100\n\nFit the model on batches with real-time data augmentation:\n\n    num_epochs = 100 # we iterate 200 times over the entire training set \n    history = model.fit_generator(train_flow, \n                        steps_per_epoch=len(X_train) \/ batch_size, \n                        epochs=num_epochs,  \n                        verbose=1,  \n                        validation_data=test_flow,  \n                        validation_steps=len(X_test) \/ batch_size) \n    Epoch 1\/100 \n    883\/883 [==============================] - 15s - loss: 1.7065 - acc: 0.2836 - val_loss: 1.8536 - val_acc: 0.1822 \n    Epoch 2\/100 \n    883\/883 [==============================] - 14s - loss: 1.4980 - acc: 0.4008 - val_loss: 1.5688 - val_acc: 0.3891 \n    ... \n    883\/883 [==============================] - 13s - loss: 0.9292 - acc: 0.6497 - val_loss: 1.1499 - val_acc: 0.5819 \n    Epoch 100\/100 \n    883\/883 [==============================] - 13s - loss: 0.9225 - acc: 0.6487 - val_loss: 1.0829 - val_acc: 0.6122\n\nIf the training goes fine, the loss values should decrease over the time as shown in the following image:\n\nFigure 9.10: Neural network structure\n\n# Plotting loss\n\nLoss values on training and validation sets allows to see, how our model improves over the time and decide when to stop training:\n\n    from matplotlib import pyplot as plt \n    history.history.keys() \n    ['acc', 'loss', 'val_acc', 'val_loss'] plt.plot(history.history['loss']) \n    plt.plot(history.history['val_loss']) \n    plt.title('model loss') \n    plt.ylabel('loss') \n    plt.xlabel('epoch') \n    plt.legend(['train', 'test'], loc='upper left') \n    plt.show()\n\nFigure 9.11: Loss on training and test sets over the training epochs\n\n# Making predictions\n\nFirst, let's prepare data to make predictions about the images:\n\n    array = np.mat(data.pixels[1]).reshape(48, 48) \n    image = scipy.misc.toimage(array, cmin=0.0) \n    display(image) \n    print(emotion_labels[data.emotion[1]])\n\n    <Image>\n\nLet us input an angry emotion image:\n\n    input_img = np.array(array).reshape(1,48,48,1)\n\nOkay, we have an angry face. Now let's make prediction and check if the network can recognize it correctly:\n\n    prediction = model.predict(input_img) \n    print(prediction) \n    [[ 0.05708674  0.35863262  0.03299783  0.17862292  0.00069717  0.37196276]] \n    emotion_labels[prediction.argmax()] \n    'Neutral'\n\nNote those array of 6 float numbers. These are probabilities of belonging to each class. In other words, the model predicts, that this face can be of an angry person only with the probability of 5%. The full table would look like this:\n\n**Angry** | **Fear** | **Happy** | **Sad** | **Surprise** | **Neutral**  \n---|---|---|---|---|---  \n0.05708674  | 0.35863262 | 0.03299783 | 0.17862292 | 0.00069717 | 0.37196276\n\n    for i in xrange(1, 100): \n        array = np.mat(data.pixels[i]).reshape(48, 48) \n        image = scipy.misc.toimage(array, cmin=0.0) \n        display(image) \n        print(emotion_labels[data.emotion[i]]) \n        input_img = np.array(array).reshape(1,48,48,1) \n        prediction = model.predict(input_img) \n        print(emotion_labels[prediction.argmax()])\n\nYou will obtain the following result:\n\n    Angry \n    Neutral\n\n    Fear \n    Sad\n\n    Sad \n    Sad\n\n    Neutral \n    Neutral\n\n    Fear \n    Sad\n\n    Sad \n    Sad\n\n    Happy \n    Happy\n\n    Happy \n    Happy\n\n    Fear \n    Fear\n\nEvaluating the trained model on the test set. The function reports the loss value and an accuracy as follows:\n\n    model.evaluate_generator(test_flow, steps=len(X_test) \/ batch_size) \n    [1.1285726155553546, 0.60696517426491459]\n\nSo, the final accuracy of our model is about 60%. Which is not that bad, considering how noisy is the dataset.\n\n# Saving the model in HDF5 format\n\nSaving the model is really easy as shown in the following:\n\n    model.save('Emotions.h5') \n\n# Converting to Core ML format\n\nThe easiest way to use pre-trained CNN on iOS is by converting it to the Core ML format:\n\n    from keras.models import load_model  \n    model = load_model('Emotions.h5')  \n    coreml_model = convert(model,   \n                           image_input_names = 'image',   \n                           class_labels = emotion_labels)  \n    ...  \n    coreml_model.save('Emotions.mlmodel')\n\n# Visualizing convolution filters\n\nDebugging CNNs is notoriously difficult. One of the ways to check if the convolutional layers learned anything meaningful is to visualize their outputs using `Keras-vis` package:\n\n    from vis.utils import utils \n    from vis.visualization import visualize_class_activation, get_num_filters\n\nWe have to convert grayscale images to `rgb` to use them with `keras-vis`:\n\n    def to_rgb(im): \n        # I think this will be slow \n        w, h = im.shape \n        ret = np.empty((w, h, 3), dtype=np.uint8) \n        ret[:, :, 0] = im \n        ret[:, :, 1] = im \n        ret[:, :, 2] = im \n        return ret\n\nNames of the layers we want to visualize (consult model structure for exact layer names):\n\n    layer_names = ['conv2d_1', 'conv2d_2',  \n                   'conv2d_3', 'conv2d_4',  \n                   'conv2d_5', 'conv2d_6']\n\n    layer_sizes = [(80, 20), (80, 20),  \n                   (80, 40), (80, 40),  \n                   (80, 80), (80, 80)]\n\n    stitched_figs = []\n\n    for (layer_name, layer_size) in zip(layer_names, layer_sizes): \n        layer_idx = [idx for idx, layer in enumerate(model.layers) if layer.name == layer_name][0]\n\nVisualizing all filters in this layer:\n\n    filters = np.arange(get_num_filters(model.layers[layer_idx]))\n\nGenerating input image for each filter as shown in the following. Here `text` field is used to overlay `filter_value` on top of the image:\n\n        vis_images = [] \n        for idx in filters: \n            img = visualize_class_activation(model, layer_idx, filter_indices=idx)  \n            vis_images.append(to_rgb(img.reshape(48,48)))\n\nGenerate stitched image palette with 8 cols as follows:\n\n        stitched = utils.stitch_images(vis_images, cols=8) \n        stitched_figs.append(stitched)\n\n        plt.figure(figsize = layer_size) \n        plt.axis('off') \n        plt.imshow(stitched, interpolation='nearest', aspect='auto') \n        plt.title(layer_name) \n        plt.savefig(layer_name+\"_filters.png\", bbox_inches='tight') \n        plt.show()\n\nFigure 9.12: Convolution filters in the last convolution layer of our network\n\n# Deploying CNN to iOS\n\nYou need to drag-and-drop the Core ML file generated in the previous section into your project to start working with the model.\n\nImports:\n\n    import Foundation  \n    import Vision  \n    import AVFoundation  \n    import UIKit\n\nAt first, let's define some data structures. An enumeration for possible classification results:\n\n    enum FaceExpressions: String {  \n     case angry = \"angry\"  \n     case anxious = \"anxious\"  \n     case neutral = \"neutral\"  \n     case happy = \"happy\"  \n     case sad = \"sad\"  \n    }\n\nAn enum for errors of the classifier:\n\n    enum ClassifierError: Error {  \n     case unableToResizeBuffer  \n     case noResults  \n    }\n\n`Classifier` is a wrapper singleton for Core ML model:\n\n    class Classifier {  \n     public static let shared = Classifier()\n\n     private let visionModel: VNCoreMLModel  \n     var visionRequests = [VNRequest]()  \n     var completion: ((_ label: [(FaceExpressions, Double)], _ error: Error?)->())?\n\n    private init() {  \n     guard let visionModel = try? VNCoreMLModel(for: Emotions().model) else {  \n     fatalError(\"Could not load model\")  \n     }\n\n     self.visionModel = visionModel\n\n     let classificationRequest = VNCoreMLRequest(model: visionModel, completionHandler: classificationResultHandler)  \n     classificationRequest.imageCropAndScaleOption = .centerCrop  \n     visionRequests = [classificationRequest]  \n     }\n\nFunction to run the network on inference:\n\n     public func classifyFace(image: CGImage, completion: @escaping (_ labels: [(FaceExpressions, Double)], _ error: Error?)->()) {  \n     self.completion = completion  \n     let imageRequestHandler = VNImageRequestHandler(cgImage: image, orientation: .up)  \n     do {  \n     try imageRequestHandler.perform(visionRequests)  \n     } catch {  \n     print(error)  \n     completion([], error)  \n     }  \n     }\n\nThis method will be called when the new classification result comes:\n\n    private func classificationResultHandler(request: VNRequest, error: Error?) {  \n     if let error = error {  \n     print(error.localizedDescription)  \n     self.completion?([], error)  \n     return  \n     }  \n     guard let results = request.results as? [VNClassificationObservation] else {  \n     print(\"No results\")  \n     self.completion?([], ClassifierError.noResults)  \n     return  \n     }\n\n     let sortedResults = results  \n     .sorted { $0.confidence > $1.confidence }  \n     .map{(FaceExpressions(rawValue:$0.identifier)!, Double($0.confidence))}\n\n    self.completion?(sortedResults, nil)  \n     print(sortedResults)  \n     }  \n    }\n\nWe omit the UI part of the application here, please refer to the demo app for the full code.\n\n# Summary\n\nIn this chapter, we built a deep learning CNN, and trained it using Keras to recognize facial expressions on photos. Then we ported it for the mobile application using Core ML. The model can work in real time. We've also become acquainted with the Apple Vision framework.\n\nCNNs are powerful tools that can be applied for many computer vision tasks, as well as for time-series prediction, natural language processing, and others. They are built around the concept of convolution\u2014a mathematical operation that can be used for defining many types of image transformations. CNNs learn convolution filters in the similar manner as usual neural networks learn weights using the same stochastic gradient descent. Convolution requires less computations than usual matrix multiplications, which is why they can be effectively used on mobile devices. Apart from convolutional layers, CNNs usually include other types of layers like pooling, fully-connected, nonlinearity, regularization, and so on. Over the years, researchers proposed many CNN architectures for different purposes. Some of them were designed specifically to run on mobile devices; for example, SquizeNet, and MobileNets.\n\nIn the next chapter, we're going to explore the amazing world of human natural language. We're also going to use neural networks to build several chatbots with different personalities.\n\n# Bibliography\n\n  1. _Challenges in Representation Learning: A report on three machine learning_ _contests_ , I Goodfellow, D Erhan, PL Carrier, A Courville, M Mirza, B Hamner, W Cukierski, Y Tang, DH Lee, Y Zhou, C Ramaiah, F Feng, R Li, X Wang, D Athanasakis, J Shawe-Taylor, M Milakov, J Park, R Ionescu, M Popescu, C Grozea, J Bergstra, J Xie, L Romaszko, B Xu, Z Chuang, and Y. Bengio. arXiv 2013. Site of the competition <http:\/\/deeplearning.net\/icml2013-workshop-competition>.\n  2. _Affective Computing_ , Rosalind Picard. MIT Technical Report #32, 1995 <http:\/\/affect.media.mit.edu\/pdfs\/95.picard.pdf>.\n  3. Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift. Sergey Ioffe, Christian Szegedy, 2015.\n\n# Natural Language Processing\n\nLanguage is an integral part of our daily life and a natural way of conveying ideas from person to person. But as easy it is for us to understand our native language, it is just as difficult for computers to process it. The internet changed the science of language forever because it allowed collecting huge volumes of text and audio records. The field of knowledge that arose at the intersection of linguistics, computer science, and machine learning was called **natural language processing** ( **NLP** ).\n\nIn this chapter, we will get acquainted with the basic concepts and applications of NLP, relevant in the context of mobile development. We will talk about the powerful tools provided by iOS and the macOS SDK for language processing. We also will learn about the theory of distributional semantics and vector representations of words as its embodiment. They will allow us to express the meaning of sentences in the computer's favorite format\u2014in the form of numbers. Based on vector representations, we will build a chatbot from scratch to play a _Word Association_ game.\n\nIn this chapter, we will cover the following topics:\n\n  * What is NLP?\n  * Python libraries\u2014NLTK and Gensim\n  * iOS NLP tools\u2014NSRegularExpression, NSDataDetector, NSLinguisticTagger, the Speech framework, and UIReferenceLibraryViewController\n  * macOS NLP tools\u2014LatentSemanticMapping\n  * How to use tokenizers, lemmatization, and part-of-speech tagging\n  * What are vector word representations?\n  * How to generate word embeddings\n  * How to use the Word2Vec model on iOS\n  * How to build a chatbot from scratch\n\n# NLP in the mobile development world\n\nUsually, NLP specialists deal with big amounts of raw text organized in **linguistics corpuses**. The algorithms in this domain are resource-consuming and often contain many hand-crafted heuristics. All this doesn't look like a good match for mobile applications, where each megabyte or frame per second is important. Despite these obstacles, NLP is widely used on mobile platforms, usually in tight integration with the server-side backend for heavy computations. Here is a list of some common NLP features that can be found in many mobile applications:\n\n  * Chatbots\n  * Spam filtering\n  * Automated translation\n  * Sentiment analysis\n  * Speech-to-text and text-to-speech\n  * Automatic spelling and grammar correction\n  * Automatic completion\n  * Keyboard suggestions\n\nUntil recently, all but the last two tasks were done on the server side, but as mobile computational power grows, more apps tend to do processing (at least partially) locally on the client. When we talk about NLP on a mobile device, in most cases, it is about processing private user information: messages, letters, notes, and similar texts. So the security issue here is particularly acute. By eliminating the server from our schema, we significantly reduce the risk of leaking user data. In this chapter, in addition to discussing common tricks and popular NLP tools, we will look at the solutions Apple provides in iOS SDK. Further, continuing our conversation on neural networks, we will teach six chatbots to play a _Word Association_ game. Each chatbot will have its individuality and will run on the device. Each of the models on average will not exceed 3 MB.\n\nFigure 10.1: Every chatbot in our application will have its own individuality\n\n# Word Association game\n\nMany of us may have played this game as kids. The rules are very simple:\n\n  * You say the word:\n\n    do while(true) {\n\n  * I say the first association to your word that came to my mind\n  * You give an association to my association:\n\n    }\n\nFor example, Dog \u2192 Cat \u2192 Pet \u2192 Toy \u2192 Baby \u2192 Girl \u2192 Wedding \u2192 Funeral. In the game, people reveal their life experience and way of thinking to each other; maybe that's why we could play it for hours as kids. Different people have different associations with the same word, and associations often head towards a completely unexpected direction. Psychologists have been studying associative series for more than a century, hoping to find in them the key to the mysteries of consciousness and the subconscious. Can you code a game AI to play like that? Perhaps you think you will need a manually composed database of associations. But what if you want your AI to have several personalities? Thanks to machine learning, this is definitely possible and you even don't need to compose the database manually. In the following screenshot, you can see the results of two games with two characters, Mark Twain and Benjamin Franklin:\n\nFigure 10.2: Playing a Word Association game with historical personalities\n\nIf you haven't got the idea of the game, visit Wikipedia to get a more thorough explanation of the game there: https:\/\/en.wikipedia.org\/wiki\/Word_Association.\n\n# Python NLP libraries\n\nThe two Python libraries that we're going to use in this chapter are **natural language toolkit** ( **NLTK** ) and Gensim. We will use the first one for text preprocessing and the second one for training or machine learning models. To install them, activate your Python virtual environment:\n\n    **> cd ~**\n    **> virtualenv swift-ml-book**\n\nAnd run `pip install`:\n\n    **> pip install -U nltk ****gensim**\n\nOfficial sites:\n\n  * NLTK, <http:\/\/www.nltk.org\/>\n\n  * Gensim, https:\/\/radimrehurek.com\/gensim\/\n\nOther popular libraries for NLP in Python:\n\n  * TextBlob, https:\/\/textblob.readthedocs.io\/en\/dev\/\n  * Stanford's CoreNLP, <https:\/\/stanfordnlp.github.io\/CoreNLP\/>\n  * SpaCy, https:\/\/spacy.io\/\n\n# Textual corpuses\n\nFor our NLP experiments, we need some reasonably big texts. I used the complete works of classical writers and statesmen from the Gutenberg project because they are in the public domain, but you can find your own texts and train models on them. If you want to use the same texts as I did, I included them in the supplementary material for this chapter under the `Corpuses` folder. There should be five of them: Benjamin Franklin, John Galsworthy, Mark Twain, William Shakespeare, and Winston Churchill. Create a new Jupyter notebook and load Mark Twain's corpus as one long string:\n\n    import zipfile \n    zip_ref = zipfile.ZipFile('Corpuses.zip', 'r') \n    zip_ref.extractall('') \n    zip_ref.close() \n    In [1]: \n    import codecs \n    In [2]: \n    one_long_string = \"\" \n    with codecs.open('Corpuses\/MarkTwain.txt', 'r', 'utf-8-sig') as text_file: \n        one_long_string = text_file.read() \n    In [3]: \n    one_long_string[99000:99900] \n    Out[3]: \n    u\"size, very elegantly wrought and dressed in the fancifulrncostumes of two centuries ago. The design was a history of something or somebody, but none of us were learned enough to read the story. The old father, reposing under a stone close by, dated 1686, might have told usrnif he could have risen. But he didn't.rnrnAs we came down through the town we encountered a squad of little donkeysrnready saddled for use. The saddles were peculiar, to say the least.rnThey consisted of a sort of saw-buck with a small mattress on it, andrnthis furniture covered about half the donkey. There were no stirrups,rnbut really such supports were not needed--to use such a saddle was thernnext thing to riding a dinner table--there was ample support clear out tornone's knee joints. A pack of ragged Portuguese muleteers crowded aroundrnus, offering their beasts at half a dollar an hour--more rascality to\" \n\n# Common NLP approaches and subtasks\n\nMost programmers are familiar with the simplest way of processing natural language: regular expressions. There are many regular expression implementations for different programming languages \u200b\u200bthat differ in small details. Because of these details, the same regular expression on various platforms can produce different results or not work at all. The two most popular standards are POSIX and Perl. The Foundation framework, however, contains its own version of regular expressions, based on the ICU C++ library. It is an extension of the POSIX standard for Unicode strings.\n\nWhy are we even talking about regular expressions here? Regular expressions are a great example of what NLP specialists call heuristics\u2014manually written rules, ad hoc solutions, and describing a complex structure in such a way that all exceptions and variations are taken into account. Sophisticated heuristics require deep domain expertise to build. Only when we are not able to capture all the complexity using heuristics will we go for machine learning. Heuristics are fragile and costly to build, but not necessarily something wrong; unlike machine learning, they are deterministic and easy to test.\n\nHeuristics and machine learning are the two arms that wield NLP. Big NLP tasks usually consist of smaller ones. To perform a grammar correction, you must split your text into sentences, split sentences into words, determine the parts of speech in those sentences, and so on. In our text corpus preprocessing, we will go through several such tasks: sentence tokenization, word tokenization, lemmatization, and stop words removal.\n\n# Tokenization\n\nTokens in linguistics are different from the authorization tokens were used to. They are linguistic units: words are tokens, numbers and punctuation marks are tokens, and sentences are tokens. In other words, they are discrete pieces of information or meaning. Tokenization is a process of splitting text into **lexical tokens**. Sentence tokenizers split texts into sentences, and word tokenizers split them further into separate words, punctuation marks, and so on. This task may seem simple ( **there is a regexp for that!** ), but this impression is deceptive. Here are a few problems to consider:\n\n  * How to tokenize words with a hyphen or an apostrophe, for example, _New York-based_ or _you're_?\n  * How to tokenize web addresses and emails, for example, `My_mail@examplewebsite.com`?\n  * What to do with emoji and kaomoji?  !\n  * What to do with languages \u200b\u200bin which gluing several words into one long word is the norm? An example is the German _siebenhundertsiebenundsiebzigtausendsiebenhundertsiebenundsiebzig_. This is the number 777,777 by the way.\n  * What to do with languages \u200b\u200bthat do not use spaces at all (Chinese and Thai)?\n\nFortunately, there are many tokenizer implementations for different languages, including the NLTK Python library and Apple `NSLinguisticTagger`:\n\n    In [4]: \n    from nltk import word_tokenize, sent_tokenize \n    In [5]: \n    sentences = sent_tokenize(one_long_string) \n    del(one_long_string) \n    In [6]: \n    sentences[200:205] \n    Out[6]: \n    [u'Ah, if I had only known then that he was only a common mortal, and thatrnhis mission had nothing more overpowering about it than the collecting ofrnseeds and uncommon yams and extraordinary cabbages and peculiar bullfrogsrnfor that poor, useless, innocent, mildewed old fossil the SmithsonianrnInstitute, I would have felt so much relieved.', \n     u'During that memorable month I basked in the happiness of being for oncernin my life drifting with the tide of a great popular movement.', \n     u'Everybodyrnwas going to Europe--I, too, was going to Europe.', \n     u'Everybody was going tornthe famous Paris Exposition--I, too, was going to the Paris Exposition.', \n     u'The steamship lines were carrying Americans out of the various ports ofrnthe country at the rate of four or five thousand a week in the aggregate.'] \n    In [7]: \n    tokenized_sentences = map(word_tokenize, sentences) \n    del(sentences) \n    In [8]: \n    print(tokenized_sentences[200:205]) \n    [[u'Ah', u',', u'if', u'I', u'had', u'only', u'known', u'then', u'that', u'he', u'was', u'only', u'a', u'common', u'mortal', u',', u'and', u'that', u'his', u'mission', u'had', u'nothing', u'more', u'overpowering', u'about', u'it', u'than', u'the', u'collecting', u'of', u'seeds', u'and', u'uncommon', u'yams', u'and', u'extraordinary', u'cabbages', u'and', u'peculiar', u'bullfrogs', u'for', u'that', u'poor', u',', u'useless', u',', u'innocent', u',', u'mildewed', u'old', u'fossil', u'the', u'Smithsonian', u'Institute', u',', u'I', u'would', u'have', u'felt', u'so', u'much', u'relieved', u'.'], [u'During', u'that', u'memorable', u'month', u'I', u'basked', u'in', u'the', u'happiness', u'of', u'being', u'for', u'once', u'in', u'my', u'life', u'drifting', u'with', u'the', u'tide', u'of', u'a', u'great', u'popular', u'movement', u'.'], [u'Everybody', u'was', u'going', u'to', u'Europe', u'--', u'I', u',', u'too', u',', u'was', u'going', u'to', u'Europe', u'.'], [u'Everybody', u'was', u'going', u'to', u'the', u'famous', u'Paris', u'Exposition', u'--', u'I', u',', u'too', u',', u'was', u'going', u'to', u'the', u'Paris', u'Exposition', u'.'], [u'The', u'steamship', u'lines', u'were', u'carrying', u'Americans', u'out', u'of', u'the', u'various', u'ports', u'of', u'the', u'country', u'at', u'the', u'rate', u'of', u'four', u'or', u'five', u'thousand', u'a', u'week', u'in', u'the', u'aggregate', u'.']] \n    In [9]: \n    from nltk import download \n    In [10]: \n    download('stopwords') \n    [nltk_data] Downloading package stopwords to \n    [nltk_data]     \/Users\/Oleksandr\/nltk_data... \n    [nltk_data]   Package stopwords is already up-to-date! \n    Out[10]: \n    True \n    In [11]: \n    from nltk.stem import WordNetLemmatizer \n    In [12]: \n    wordnet_lemmatizer = WordNetLemmatizer() \n    In [13]: \n    lemmatized_sentences = map(lambda sentence: map(wordnet_lemmatizer.lemmatize, sentence), tokenized_sentences) \n    In [14]: \n    print(lemmatized_sentences[200:205]) \n    [[u'Ah', u',', u'if', u'I', u'had', u'only', u'known', u'then', u'that', u'he', u'wa', u'only', u'a', u'common', u'mortal', u',', u'and', u'that', u'his', u'mission', u'had', u'nothing', u'more', u'overpowering', u'about', u'it', u'than', u'the', u'collecting', u'of', u'seed', u'and', u'uncommon', u'yam', u'and', u'extraordinary', u'cabbage', u'and', u'peculiar', u'bullfrog', u'for', u'that', u'poor', u',', u'useless', u',', u'innocent', u',', u'mildewed', u'old', u'fossil', u'the', u'Smithsonian', u'Institute', u',', u'I', u'would', u'have', u'felt', u'so', u'much', u'relieved', u'.'], [u'During', u'that', u'memorable', u'month', u'I', u'basked', u'in', u'the', u'happiness', u'of', u'being', u'for', u'once', u'in', u'my', u'life', u'drifting', u'with', u'the', u'tide', u'of', u'a', u'great', u'popular', u'movement', u'.'], [u'Everybody', u'wa', u'going', u'to', u'Europe', u'--', u'I', u',', u'too', u',', u'wa', u'going', u'to', u'Europe', u'.'], [u'Everybody', u'wa', u'going', u'to', u'the', u'famous', u'Paris', u'Exposition', u'--', u'I', u',', u'too', u',', u'wa', u'going', u'to', u'the', u'Paris', u'Exposition', u'.'], [u'The', u'steamship', u'line', u'were', u'carrying', u'Americans', u'out', u'of', u'the', u'various', u'port', u'of', u'the', u'country', u'at', u'the', u'rate', u'of', u'four', u'or', u'five', u'thousand', u'a', u'week', u'in', u'the', u'aggregate', u'.']] \n    In [15]: \n    del(tokenized_sentences) \n\n# Stemming\n\nStemming is the process of reducing words to their stems. The idea here is that related words can usually be reduced to a common stem.\n\nFor example: ( **whit** e, **whit** ening, **whit** ish, **whit** er) \u2192 **whit**.\n\nThis can be used, for instance, to expand user queries. But some cases can be tricky, consider the English _man_ and _men_ , the Irish _bhean_ = _woman_ and _mn\u00e0_ = _women_ , or the even more extreme English _am_ , _is_ , _are_ , _was_ , _were_ , and _been_. There are several popular stemmers for English.\n\n# Lemmatization\n\nThis is a more advanced approach than stemming. Instead of reducing words to stems, lemmatizers match every word to its _lemma_ , the form in a dictionary. This is especially useful for languages such as Polish, where one verb can easily have 220 different grammatical forms, mostly with different spellings: http:\/\/wsjp.pl\/do_druku.php?id_hasla=34745&id_znaczenia=0.\n\nThe problem here is homonyms.\n\n# Part-of-speech (POS) tagging\n\nNLTK uses a pre-trained machine learning model (averaged perceptron) for POS tagging. The task is especially hard for English because, unlike many other languages, the same word can play the role of different parts of speech depending on the context:\n\n    In [16]: \n    from nltk import download \n    In [17]: \n    download('averaged_perceptron_tagger') \n    [nltk_data] Downloading package averaged_perceptron_tagger to \n    [nltk_data]     \/Users\/Oleksandr\/nltk_data... \n    [nltk_data]   Package averaged_perceptron_tagger is already up-to- \n    [nltk_data]       date! \n    Out[17]: \n    True \n    In [18]: \n    from nltk import pos_tag, pos_tag_sents \n    In [19]: \n    pos_tag(word_tokenize('Cats, cat, Cat, and \"The Cats\"')) \n    Out[19]: \n    [('Cats', 'NNS'), \n     (',', ','), \n     ('cat', 'NN'), \n     (',', ','), \n     ('Cat', 'NNP'), \n     (',', ','), \n     ('and', 'CC'), \n     ('``', '``'), \n     ('The', 'DT'), \n     ('Cats', 'NNP'), \n     (\"''\", \"''\")] \n    In [20]: \n    pos_sentences = pos_tag_sents(lemmatized_sentences) \n    del(lemmatized_sentences) \n    In [21]: \n    print(pos_sentences[200:205]) \n    [[(u'Ah', 'NNP'), (u',', ','), (u'if', 'IN'), (u'I', 'PRP'), (u'had', 'VBD'), (u'only', 'RB'), (u'known', 'VBN'), (u'then', 'RB'), (u'that', 'IN'), (u'he', 'PRP'), (u'wa', 'VBZ'), (u'only', 'RB'), (u'a', 'DT'), (u'common', 'JJ'), (u'mortal', 'NN'), (u',', ','), (u'and', 'CC'), (u'that', 'IN'), (u'his', 'PRP$'), (u'mission', 'NN'), (u'had', 'VBD'), (u'nothing', 'NN'), (u'more', 'RBR'), (u'overpowering', 'VBG'), (u'about', 'IN'), (u'it', 'PRP'), (u'than', 'IN'), (u'the', 'DT'), (u'collecting', 'NN'), (u'of', 'IN'), (u'seed', 'NN'), (u'and', 'CC'), (u'uncommon', 'JJ'), (u'yam', 'NN'), (u'and', 'CC'), (u'extraordinary', 'JJ'), (u'cabbage', 'NN'), (u'and', 'CC'), (u'peculiar', 'JJ'), (u'bullfrog', 'NN'), (u'for', 'IN'), (u'that', 'DT'), (u'poor', 'JJ'), (u',', ','), (u'useless', 'JJ'), (u',', ','), (u'innocent', 'JJ'), (u',', ','), (u'mildewed', 'VBD'), (u'old', 'JJ'), (u'fossil', 'NN'), (u'the', 'DT'), (u'Smithsonian', 'NNP'), (u'Institute', 'NNP'), (u',', ','), (u'I', 'PRP'), (u'would', 'MD'), (u'have', 'VB'), (u'felt', 'VBN'), (u'so', 'RB'), (u'much', 'JJ'), (u'relieved', 'NN'), (u'.', '.')], [(u'During', 'IN'), (u'that', 'DT'), (u'memorable', 'JJ'), (u'month', 'NN'), (u'I', 'PRP'), (u'basked', 'VBD'), (u'in', 'IN'), (u'the', 'DT'), (u'happiness', 'NN'), (u'of', 'IN'), (u'being', 'VBG'), (u'for', 'IN'), (u'once', 'RB'), (u'in', 'IN'), (u'my', 'PRP$'), (u'life', 'NN'), (u'drifting', 'VBG'), (u'with', 'IN'), (u'the', 'DT'), (u'tide', 'NN'), (u'of', 'IN'), (u'a', 'DT'), (u'great', 'JJ'), (u'popular', 'JJ'), (u'movement', 'NN'), (u'.', '.')], [(u'Everybody', 'NN'), (u'wa', 'VBZ'), (u'going', 'VBG'), (u'to', 'TO'), (u'Europe', 'NNP'), (u'--', ':'), (u'I', 'PRP'), (u',', ','), (u'too', 'RB'), (u',', ','), (u'wa', 'VBZ'), (u'going', 'VBG'), (u'to', 'TO'), (u'Europe', 'NNP'), (u'.', '.')], [(u'Everybody', 'NN'), (u'wa', 'VBZ'), (u'going', 'VBG'), (u'to', 'TO'), (u'the', 'DT'), (u'famous', 'JJ'), (u'Paris', 'NNP'), (u'Exposition', 'NNP'), (u'--', ':'), (u'I', 'PRP'), (u',', ','), (u'too', 'RB'), (u',', ','), (u'wa', 'VBZ'), (u'going', 'VBG'), (u'to', 'TO'), (u'the', 'DT'), (u'Paris', 'NNP'), (u'Exposition', 'NNP'), (u'.', '.')], [(u'The', 'DT'), (u'steamship', 'NN'), (u'line', 'NN'), (u'were', 'VBD'), (u'carrying', 'VBG'), (u'Americans', 'NNPS'), (u'out', 'IN'), (u'of', 'IN'), (u'the', 'DT'), (u'various', 'JJ'), (u'port', 'NN'), (u'of', 'IN'), (u'the', 'DT'), (u'country', 'NN'), (u'at', 'IN'), (u'the', 'DT'), (u'rate', 'NN'), (u'of', 'IN'), (u'four', 'CD'), (u'or', 'CC'), (u'five', 'CD'), (u'thousand', 'NNS'), (u'a', 'DT'), (u'week', 'NN'), (u'in', 'IN'), (u'the', 'DT'), (u'aggregate', 'NN'), (u'.', '.')]] \n    In [22]: \n    download('tagsets') \n    [nltk_data] Downloading package tagsets to \n    [nltk_data]     \/Users\/Oleksandr\/nltk_data... \n    [nltk_data]   Package tagsets is already up-to-date! \n    Out[22]: \n    True \n    In [23]: \n    from nltk.help import upenn_tagset \n    In [24]: \n    upenn_tagset()\n\nYou can find the full list of tags in the notebook or in the NLTK documentation: <https:\/\/www.ling.upenn.edu\/courses\/Fall_2003\/ling001\/penn_treebank_pos.html>. Here I'm only recounting the POS important for our goal with some examples.\n\nAdjectives:\n\n    JJ: adjective or numeral, ordinal \n    third ill-mannered pre-war regrettable oiled calamitous first  \n    JJR: adjective, comparative \n    JJS: adjective, superlative\n\nNouns:\n\n    NN: noun, common, singular or mass \n    common-carrier cabbage knuckle-duster Casino afghan shed \n    NNP: noun, proper, singular \n    Conchita Escobar Kreisler Sawyer CTCA Shannon A.K.C. Liverpool \n    NNPS: noun, proper, plural \n    Americans Americas Anarcho-Syndicalists Andalusians Andes \n    NNS: noun, common, plural\n\nAdverbs:\n\n    RB: adverb \n    occasionally unabatingly maddeningly adventurously swiftly \n    RBR: adverb, comparative \n    RBS: adverb, superlative\n\nInterjections:\n\n    UH: interjection \n    Goodbye Wow Hey Oops amen huh uh anyways honey man baby hush\n\nVerbs:\n\n    VB: verb, base form \n    ask assemble assess assign assume avoid bake balkanize begin \n    VBD: verb, past tense \n    VBG: verb, present participle or gerund \n    VBN: verb, past participle \n    VBP: verb, present tense, not 3rd person singular \n    VBZ: verb, present tense, 3rd person singular \n\n# Named entity recognition (NER)\n\nNote the (u'Paris', 'NNP'), (u'Exposition', 'NNP'), (u'Americans, NNPS). NNP stands for proper noun, NNPS proper noun plural. We need to get rid of all capital letters from non-proper nouns and from all punctuation marks and numbers:\n\n    In [25]: \n    # tags_to_delete = ['$', \"''\", \"(\", \")\", \",\", \"--\", \".\", \":\", \"CC\"] \n    tags_to_not_lowercase = set(['NNP', 'NNPS']) \n    tags_to_preserve = set(['JJ', 'JJR', 'JJS', 'NN', 'NNP', 'NNPS', 'NNS', 'RB', 'RBR', 'RBS','UH', 'VB', 'VBD', 'VBG', 'VBN', 'VBP', 'VBZ']) \n    In [26]: \n    print(pos_sentences[203]) \n    [(u'Everybody', 'NN'), (u'wa', 'VBZ'), (u'going', 'VBG'), (u'to', 'TO'), (u'the', 'DT'), (u'famous', 'JJ'), (u'Paris', 'NNP'), (u'Exposition', 'NNP'), (u'--', ':'), (u'I', 'PRP'), (u',', ','), (u'too', 'RB'), (u',', ','), (u'wa', 'VBZ'), (u'going', 'VBG'), (u'to', 'TO'), (u'the', 'DT'), (u'Paris', 'NNP'), (u'Exposition', 'NNP'), (u'.', '.')] \n    In [27]: \n    def carefully_lowercase(words): \n        return [(word.lower(), pos) if pos not in tags_to_not_lowercase else (word, pos) \n                for (word, pos) in words] \n    In [28]: \n    def filter_meaningful(words): \n        return [word for (word, pos) in words if pos in tags_to_preserve] \n    In [29]: \n    res = map(carefully_lowercase, pos_sentences[203:205]) \n    print(res) \n    [[(u'everybody', 'NN'), (u'wa', 'VBZ'), (u'going', 'VBG'), (u'to', 'TO'), (u'the', 'DT'), (u'famous', 'JJ'), (u'Paris', 'NNP'), (u'Exposition', 'NNP'), (u'--', ':'), (u'i', 'PRP'), (u',', ','), (u'too', 'RB'), (u',', ','), (u'wa', 'VBZ'), (u'going', 'VBG'), (u'to', 'TO'), (u'the', 'DT'), (u'Paris', 'NNP'), (u'Exposition', 'NNP'), (u'.', '.')], [(u'the', 'DT'), (u'steamship', 'NN'), (u'line', 'NN'), (u'were', 'VBD'), (u'carrying', 'VBG'), (u'Americans', 'NNPS'), (u'out', 'IN'), (u'of', 'IN'), (u'the', 'DT'), (u'various', 'JJ'), (u'port', 'NN'), (u'of', 'IN'), (u'the', 'DT'), (u'country', 'NN'), (u'at', 'IN'), (u'the', 'DT'), (u'rate', 'NN'), (u'of', 'IN'), (u'four', 'CD'), (u'or', 'CC'), (u'five', 'CD'), (u'thousand', 'NNS'), (u'a', 'DT'), (u'week', 'NN'), (u'in', 'IN'), (u'the', 'DT'), (u'aggregate', 'NN'), (u'.', '.')]] \n    In [30]: \n    filtered = map(filter_meaningful, res) \n    del(res) \n    print(filtered) \n    [[u'everybody', u'wa', u'going', u'famous', u'Paris', u'Exposition', u'too', u'wa', u'going', u'Paris', u'Exposition'], [u'steamship', u'line', u'were', u'carrying', u'Americans', u'various', u'port', u'country', u'rate', u'thousand', u'week', u'aggregate']] \n    In [31]: \n    lowercased_pos_sentences = map(carefully_lowercase,  pos_sentences) \n    del(pos_sentences) \n\n# Removing stop words and punctuation\n\nStop words are all those words that don't add much information to the sentence. For example, the last sentence can be shortened to: _stop words don't add useful information sentence_. And despite the fact that it doesn't look like a proper English sentence, you'd likely understand the meaning if you heard it somewhere. That's why in many cases we can make our models simpler by simply ignoring these words. Stop words are usually the most common words in natural texts. For English, a list of them can be found in `nltk.corpus.stopwords`:\n\n    In [32]: \n    sentences_to_train_on = map(lambda words: [word for (word, pos) in words], lowercased_pos_sentences) \n    In [33]: \n    print(sentences_to_train_on[203:205]) \n    [[u'everybody', u'wa', u'going', u'to', u'the', u'famous', u'Paris', u'Exposition', u'--', u'i', u',', u'too', u',', u'wa', u'going', u'to', u'the', u'Paris', u'Exposition', u'.'], [u'the', u'steamship', u'line', u'were', u'carrying', u'Americans', u'out', u'of', u'the', u'various', u'port', u'of', u'the', u'country', u'at', u'the', u'rate', u'of', u'four', u'or', u'five', u'thousand', u'a', u'week', u'in', u'the', u'aggregate', u'.']] \n    In [34]: \n    import itertools \n    In [35]: \n    filtered = map(filter_meaningful, lowercased_pos_sentences) \n    flatten = list(itertools.chain(*filtered)) \n    words_to_keep = set(flatten) \n    In [36]: \n    del(filtered, flatten, lowercased_pos_sentences) \n    In [37]: \n    from nltk.corpus import stopwords \n    import string \n    In [38]: \n    stop_words = set(stopwords.words('english') + list(string.punctuation) + ['wa']) \n\n# Distributional semantics hypothesis\n\nIt's difficult to say what it means \"to understand meaning of a text\", but everyone will say that people can do this, and computers do not. Natural language understanding is one of the tough problems in Artificial Intelligence. How to capture the semantics of the sentence\n\nTraditionally there were two opposite approaches to the problem. The first one goes like this: start from the definitions of separate words, hard-code the relations between them, and write down the sentence structures. If you are persistent enough, hopefully you will end up with a complex model that will incorporate enough expert knowledge to parse some natural questions and produce meaningful answers. And then, you'll find out that for a new language, you need to start everything over.\n\nThat's why many researchers turned to the opposite approach: statistical methods. Here, we start from a big amount of textual data and allow the computer to figure out the meaning of the text. The hypothesis of **distributional semantics** assumes that the meaning of a specific word in a sentence is not defined by the word itself but rather by all contexts in which that word appears. Wikipedia gives a more formal wording:\n\n\"Linguistic items with similar distributions have similar meanings.\"\n\nNow hold on tight! In the following sections, we're going to discuss an algorithm that can blow your mind. When I first came across it, I spent nights experimenting with it. I was feeding different texts into it, starting from the movie reviews dataset and finishing with the New Testament in ancient Greek and the undeciphered Voynich manuscript. It felt just like some craze or magic. The algorithm was able to capture the meanings of the words and whole sentences from raw texts, even in long-dead languages. This was the first time for me that computers seemingly crossed the line between _crunching megabytes of texts_ and _understanding the meaning of text written by humans and for humans_.\n\n# Word vector representations\n\nDistributional semantics represents words as vectors in the space of senses. The vectors corresponding to the words with similar meanings should be close to each other in this space. How to build such vectors is not a simple question, however. The simplest approach to take is to start from one-hot vectors for the words, but then the vectors will be both sparse and giant, each one of the same length as the number of words in the vocabulary. That's why we use dimensionality reduction with autoencoder-like architecture.\n\n# Autoencoder neural networks\n\n**Autoencoder** is a neural network whose goal is to produce an output identical to an input. For example, if you pass a picture into it, it should return the same picture on the other end. This seems... not complicated! But the trick is the special architecture\u2014its inner layers have fewer neurons than input and output layers, usually with some extreme bottleneck in the middle. The layer before the bottleneck is called encoder and the layer after it is called **decoder network**. The encoder converts the input into some inner representation and the decoder then restores the data to its original form. During training, the network must figure out how to compress the input data most effectively and then un-compress it with the least possible information loss. This architecture can also be employed to train neural networks, which change input data in a way we want them to. For example, autoencoders have been successfully used to remove noise from images.\n\nAutoencoder neural networks are an example of so-called **representation learning**. It is something between supervised and unsupervised learning.\n\nFigure 10.3: Autoencoder architecture\n\nThe blue layers are the encoder part, the **yellow layer** in the middle is a bottleneck, and the **green layers** are the decoder. The network in the picture is a fully connected network (every neuron of one layer is connected to every neuron in the next layer); however, this is not the only option for autoencoders.\n\n# Word2Vec\n\nWord2Vec is an efficient algorithm for word embeddings generation based on neural networks. It was originally described by Mikolov et al. in _Distributed Representations of Words and Phrases and their Compositionality_ (2013). The original C implementation in the form of a command-line application is available at <https:\/\/code.google.com\/archive\/p\/word2vec\/>.\n\nFigure 10.4: Architecture of Word2Vec\n\nWord2Vec is often referred to as an instance of deep learning, but the architecture is actually quite shallow: only three layers in depth. This misconception is likely related to its wide adoption for enhancing productivity of deep networks in NLP. The Word2Vec architecture is similar to an autoencoder. The input of the neural network is a sufficiently big text corpus, and the output is a list of vectors (arrays of numbers), one vector for each word in the corpus. The algorithm uses the context of each word to encode in those vectors information about co-occurrences of words. As a result, the vectors have some peculiar properties; the vectors of words with similar meaning are also close to each other. While we can't mathematically calculate the precise distance between word meanings, we can calculate the similarity of two vectors without any problems. That's why algorithms that turn words into vectors are so important. For example, using the cosine similarity metric, we can find the words closest to cat:\n\n  * Bird: 0.760521\n  * Cow: 0.766533\n  * Dog: 0.831517\n  * Rat: 0.748557\n  * Blonde: 0.763721\n  * Pig: 0.751001\n  * Goat: 0.798104\n  * Hamster: 0.768635\n  * Bee: 0.774112\n  * Llama: 0.747295\n\nFigure 10.5: Examples of relationships between words that can be captured using Word2Vec\n\nYou can add, substract, and project vectors. Interestingly, these operations have some quite meaningful results, for example, _king-men+woman=queen_ , _dog-men+woman=cat_ and so on. By the way, as you can see from the last example, an algorithm captures all our stereotypes quite precisely.\n\nFigure 10.6: The distribution of words into vector spaces is similar among different languages\n\nWord2Vec can be used not only with natural text but also with any sequence of discrete states, where context matters: playlists, DNA, source codes, and so on.\n\n# Word2Vec in Gensim\n\nThere is no point in running Word2Vec on an iOS device: in the app, we need only the vectors it generates. For running Word2Vec, we will use the Python NLP package gensim. This library is popular for topic modeling and contains a fast Word2Vec implementation with a nice API. We don't want to load large corpuses of text on a mobile phone and don't want to train Word2vec on the iOS device, so we will learn a vector representation using the Gensim Python library. Then, we will do some preprocessing (remove everything except nouns) and plug this database into our iOS application:\n\n    In [39]: \n    import gensim \n    In [40]: \n    def trim_rule(word, count, min_count): \n        if word not in words_to_keep or word in stop_words: \n            return gensim.utils.RULE_DISCARD \n        else: \n            return gensim.utils.RULE_DEFAULT \n    In [41]: \n    model = gensim.models.Word2Vec(sentences_to_train_on, min_count=15, trim_rule=trim_rule) \n\n# Vector space properties\n\n\"The Hatter opened his eyes very wide on hearing this; but all he SAID was, 'Why is a raven like a writing-desk?'  \n'Come, we shall have some fun now!' thought Alice. 'I'm glad they've begun asking riddles. - I believe I can guess that,' she added aloud.  \n'Do you mean that you think you can find out the answer to it?' said the March Hare.\"\n\n\u2013 _Lewis Carroll_ , _Alice in a Wonderland_\n\n_Why is a raven like a writing desk?_ With the help of distributive semantic and vector word representations, finally we can help Alice to solve Hatter's riddle (in a mathematically precise way):\n\n    In [42]: \n    model.most_similar('house', topn=5) \n    Out[42]: \n    [(u'camp', 0.8188982009887695), \n     (u'cabin', 0.8176383972167969), \n     (u'town', 0.7998955845832825), \n     (u'room', 0.7963996529579163), \n     (u'street', 0.7951667308807373)] \n    In [43]: \n    model.most_similar('America', topn=5) \n    Out[43]: \n    [(u'India', 0.8678370714187622), \n     (u'Europe', 0.8501001596450806), \n     (u'number', 0.8464810848236084), \n     (u'member', 0.8352445363998413), \n     (u'date', 0.8332008123397827)] \n    In [44]: \n    model.most_similar('water', topn=5) \n    Out[44]: \n    [(u'bottom', 0.9041773676872253), \n     (u'sand', 0.9032160639762878), \n     (u'mud', 0.8798269033432007), \n     (u'level', 0.8781479597091675), \n     (u'rock', 0.8766734600067139)] \n    In [45]: \n    model.most_similar('money', topn=5) \n    Out[45]: \n    [(u'pay', 0.8744806051254272), \n     (u'sell', 0.8554744720458984), \n     (u'stock', 0.8477637767791748), \n     (u'bill', 0.8445131182670593), \n     (u'buy', 0.8271161913871765)] \n    In [46]: \n    model.most_similar('cat', topn=5) \n    Out[46]: \n    [(u'dog', 0.836624026298523), \n     (u'wear', 0.8159085512161255), \n     (u'cow', 0.7607206106185913), \n     (u'like', 0.7499277591705322), \n     (u'bird', 0.7386394739151001)] \n\n# iOS application\n\nTo use vectors in an iOS application, we must export them in a binary format:\n\n    In [47]: \n    model.wv.save_word2vec_format(fname='MarkTwain.bin', binary=True)\n\nThis binary contains words and their embedding vectors, all of the same length. The original implementation of Word2Vec was written in C, so I took it and adapted the code for our purpose\u2014to parse the binary file and find closest words to the one that we specify.\n\n# Chatbot anatomy\n\nMost chatbots look like reincarnations of console applications: you have a predefined set of commands and the bot produces an output for every command of yours. Someone even joked that Linux includes an awesome chatbot called **console**. But they don't always have to be that way. Let's see how we can make them more interesting. A typical chatbot consists of one or several input streams, a brain, and output streams. Inputs can be a keyboard, voice recognition, or set of predefined phrases. The brain is a sort of algorithm for transforming input into output. In our example, the brain will be based on word embeddings. Output streams also may be different, such as text, speech, search results (like Siri does), and so on.\n\n# Voice input\n\nThe code is as follows:\n\n    SFSpeechRecognizer \n    class func requestAuthorization(_ handler: @escaping (SFSpeechRecognizerAuthorizationStatus) -> Swift.Void)\n\n    import Speech\n\n    class VoiceRecognizer: NSObject, SFSpeechRecognizerDelegate { \n    static var shared = VoiceRecognizer()\n\n    private let speechRecognizer = SFSpeechRecognizer(locale: Locale(identifier: \"en-US\"))! \n    private var recognitionRequest: SFSpeechAudioBufferRecognitionRequest? \n    private var recognitionTask: SFSpeechRecognitionTask? \n    private let audioEngine = AVAudioEngine()\n\n    public var isListening: Bool { \n        return audioEngine.isRunning \n    }\n\n    public func stopListening() { \n        self.audioEngine.stop() \n        self.recognitionRequest?.endAudio() \n    }\n\n    public func startListening(gotResult: @escaping (String)->(), end: @escaping ()->()) { \n        speechRecognizer.delegate = self\n\nCancel the previous task if it's running:\n\n        if let recognitionTask = recognitionTask { \n            recognitionTask.cancel() \n            self.recognitionTask = nil \n        } \n        do { \n            let audioSession = AVAudioSession.sharedInstance() \n            try audioSession.setCategory(AVAudioSessionCategoryRecord) \n            try audioSession.setMode(AVAudioSessionModeMeasurement) \n            try audioSession.setActive(true, with: .notifyOthersOnDeactivation) \n        } catch { \n            print(error) \n        }\n\n        recognitionRequest = SFSpeechAudioBufferRecognitionRequest()\n\n        let inputNode = audioEngine.inputNode \n        guard let recognitionRequest = recognitionRequest else { fatalError(\"Unable to created a SFSpeechAudioBufferRecognitionRequest object\") }\n\nConfigure the request so that the results are returned before audio recording is finished:\n\n    recognitionRequest.shouldReportPartialResults = false\n\nCreate a recognition task. Store the recognition task as a property, to be able to cancel it if needed:\n\n        recognitionTask = speechRecognizer.recognitionTask(with: recognitionRequest) { [weak self] result, error in \n            guard let `self` = self else { return } \n            var isFinal = false\n\n            if let result = result { \n                let string = result.bestTranscription.formattedString \n                gotResult(string) \n                isFinal = result.isFinal \n            }\n\n            if error != nil || isFinal { \n                self.audioEngine.stop() \n                inputNode.removeTap(onBus: 0)\n\n                self.recognitionRequest = nil \n                self.recognitionTask = nil \n                end() \n            } \n        }\n\n        let recordingFormat = inputNode.outputFormat(forBus: 0) \n        inputNode.installTap(onBus: 0, bufferSize: 1024, format: recordingFormat) { (buffer: AVAudioPCMBuffer, when: AVAudioTime) in \n            self.recognitionRequest?.append(buffer) \n        }\n\n        audioEngine.prepare() \n        do { \n            try audioEngine.start() \n        } catch { \n            print(error) \n        } \n    } \n    } \n\n# NSLinguisticTagger and friends\n\n`NSLinguisticTagger` is an all-in-one class for language detection, tokenization, lemmatization, part-of-speech tagging, named entity recognition and so on. The API is in the traditions of Objective-C: You have to create an instance of the class with some options, then assign it with a string to analyze, and then iterate through tags that it had found using `enumerateTags()` method. For each tag, it returns `NSRange` object which is somewhat inconvenient to use in Swift, so we have to add some utility functions to convert them to Swift ranges:\n\n    extension String { \n    func range(from nsRange: NSRange) -> Range<String.Index>? { \n        guard \n            let from16 = utf16.index(utf16.startIndex, offsetBy: nsRange.location, limitedBy: utf16.endIndex), \n            let to16 = utf16.index(utf16.startIndex, offsetBy: nsRange.location + nsRange.length, limitedBy: utf16.endIndex), \n            let from = from16.samePosition(in: self), \n            let to = to16.samePosition(in: self) \n            else { return nil } \n        return from ..< to \n    } \n    }\n\n    struct NLPPreprocessor {\n\n    static func preprocess(inputString: String, errorCallback: (NLPPreprocessorError)->()) -> [String] {\n\n        let languageDetector = NSLinguisticTagger(tagSchemes: [.language], options: 0) \n        languageDetector.string = inputString \n        let language = languageDetector.dominantLanguage\n\n        if language != \"en\" { \n            errorCallback(.nonEnglishLanguage) \n            return [] \n        }\n\nThis is a workaround to make NSLinguisticTagger's lemmatizer work with short sentences:\n\n        let string = inputString + \". Hello, world!\"\n\n        let tagSchemes: [NSLinguisticTagScheme] = [.tokenType, .lemma, .lexicalClass]\n\n        let options = NSLinguisticTagger.Options.omitPunctuation.rawValue | NSLinguisticTagger.Options.omitWhitespace.rawValue \n        let tagger = NSLinguisticTagger(tagSchemes: NSLinguisticTagger.availableTagSchemes(forLanguage: \"en\"), options: Int(options)) \n        tagger.string = string \n        let range = NSRange(location: 0, length: string.utf16.count)\n\n        var resultTokens = [String?]() \n        let queryOptions = NSLinguisticTagger.Options(rawValue: options)\n\nUsing POS tagger to remove all word types that are not playable:\n\n        let posToPreserve: Set<NSLinguisticTag> = Set([.noun, .verb, .adjective, .adverb, .interjection, .idiom, .otherWord])\n\n        for scheme in tagSchemes { \n            var i = 0 \n            tagger.enumerateTags(in: range, scheme: scheme, options: queryOptions) \n            { (tag, range1, _, _) in \n                defer { i+=1 }\n\n                guard let tag = tag else { \n                    \/\/ Preserve total count of tokens. \n                    if scheme == .tokenType { resultTokens.append(nil) } \n                    return \n                }\n\n                switch scheme { \n                case .tokenType:\n\nSave only words while keeping the total count of tokens:\n\n                    let token = string.substring(with: string.range(from: range1)!)\n\n                    if tag == .word { \n                        resultTokens.append(token) \n                    } else { \n                        resultTokens.append(nil) \n                    } \n                case .lemma:\n\nIf a word has a lemma, save it:\n\n                    resultTokens[i] = tag.rawValue \n                case .lexicalClass: \n                    \/\/ Using POS tagger to remove all word types that are not playable. \n                    if !posToPreserve.contains(tag) { \n                        resultTokens[i] = nil \n                    } \n                default: \n                    break \n                } \n            } \n        }\n\nThis is again a workaround to make NSLinguisticTagger's lemmatizer work with short sentences:\n\n        var result = resultTokens.flatMap{$0} \n        print(result) \n        result.removeLast() \n        result.removeLast() \n        return result \n    } \n    } \n\n# Word2Vec on iOS\n\nThe original implementation is written in C, so I added a simple Objective-C wrapper:\n\n    @interface W2VDistance : NSObject\n\n    - (void)loadBinaryVectorFile:(NSURL * _Nonnull) fileURL \n                           error:(NSError *_Nullable* _Nullable) error;\n\n    - (NSDictionary <NSString *, NSNumber *>  * _Nullable)closestToWord:(NSString * _Nonnull) word \n                                                        numberOfClosest:(NSNumber * _Nullable) numberOfClosest;\n\n    - (NSDictionary <NSString *, NSNumber *>  * _Nullable)analogyToPhrase:(NSString * _Nonnull) phrase \n                                                          numberOfClosest:(NSNumber * _Nullable) numberOfClosest;\n\n    @end \n    private func getW2VAnalogy(sentence: String) -> String? { \n    guard let words = word2VecProvider?.analogy(toPhrase: sentence, numberOfClosest: 1)?.keys else { \n        return nil \n    } \n    return Array(words).last \n    }\n\n    private func getW2VWord(word: String) -> String? { \n    guard let words = word2VecProvider?.closest(toWord: word, numberOfClosest: 1)?.keys else { \n        return nil \n    } \n    return Array(words).last \n    } \n\n# Text-to-speech output\n\nThe code is as follows:\n\n    import Speech\n\n    class SpeechSynthesizer: NSObject, AVSpeechSynthesizerDelegate { \n    static var shared = SpeechSynthesizer()\n\n    private var synthesizer = AVSpeechSynthesizer() \n    var voice = AVSpeechSynthesisVoice(language: \"en-US\")\n\n    public func prepare() { \n        let dummyUtterance = AVSpeechUtterance(string: \" \") \n        dummyUtterance.voice = AVSpeechSynthesisVoice(language: \"en-US\")\n\n        synthesizer.speak(dummyUtterance) \n    }\n\n    public func speakAloud(word: String) { \n        if synthesizer.isSpeaking { \n            synthesizer.stopSpeaking(at: .immediate) \n        }\n\n        let utterance = AVSpeechUtterance(string: word) \n        utterance.rate = 0.4 \n        utterance.preUtteranceDelay = 0.1; \n        utterance.postUtteranceDelay = 0.1; \n        utterance.voice = self.voice\n\n        synthesizer.speak(utterance) \n    }\n\n    public func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didStart utterance: AVSpeechUtterance) {\n\n    }\n\n    public func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didFinish utterance: AVSpeechUtterance) {\n\n    }\n\n    public func speechSynthesizer(_ synthesizer: AVSpeechSynthesizer, didCancel utterance: AVSpeechUtterance) {\n\n    } \n    } \n\n# UIReferenceLibraryViewController\n\nThe code is as follows:\n\n     let hasDefinition = UIReferenceLibraryViewController.dictionaryHasDefinition(forTerm: term) \n    if hasDefinition { \n         let referenceController = UIReferenceLibraryViewController(term: term) \n    navigationController?.pushViewController(referenceController, animated: true) \n    }\n\nFigure 10.7: Reference library view controller user interface\n\n# Putting it all together\n\nThe code is as follows:\n\n    private func recognitionEnded() { \n    recordButton.isEnabled = true \n    recordButton.setTitle(\"Listen\", for: []) \n    let result = self.recognitionResult\n\n    let words: [String] \n    if allLowercase { \n        words = result.split(separator: \" \").map(String.init).map{$0.lowercased()} \n    } else { \n        words = NLPPreprocessor.preprocess(inputString: result) { error in \n            messages.append(result) \n            messages.append(\"This doesn't look like English.\") \n            reloadTable() \n        } \n    }\n\n    let wordCount = words.count\n\n    var stringToPassToW2V: String \n    var stringToShowInUI: String \n    switch wordCount { \n    case 1: \n        stringToPassToW2V = String(words.last!) \n        stringToShowInUI = String(words.last!) \n    case 2: \n        let wordPair = Array(words.suffix(2)) \n        stringToPassToW2V = \"(wordPair[0]) (wordPair[1])\" \n        stringToShowInUI = \"(wordPair[0]) - (wordPair[1])\" \n    case 3...: \n        let wordTriplet = Array(words.suffix(3)) \n        stringToPassToW2V = \"(wordTriplet[0]) (wordTriplet[1]) (wordTriplet[2])\" \n        stringToShowInUI = \"(wordTriplet[0]) - (wordTriplet[1]) + (wordTriplet[2])\" \n    default: \n        print(\"Warning: wrong number of input words.\") \n        return \n    } \n    print(stringToPassToW2V) \n    messages.append(stringToShowInUI) \n    reloadTable()\n\n    DispatchQueue.main.async() { [weak self] in \n        guard let `self` = self else { return } \n        var response: String? \n        if wordCount > 1 { \n            response = self.getW2VAnalogy(sentence: stringToPassToW2V)?.capitalized \n        } else { \n            response = self.getW2VWord(word: stringToPassToW2V)?.capitalized \n        }\n\n        if response?.isEmpty ?? true || response == \"``\" { \n            response = \"I don't know this word.\" \n        }\n\n    \/\/            SpeechSynthesizer.shared.speakAloud(word: response!) \n        self.messages.append(response!) \n        self.reloadTable()\n\n        print(response!) \n    } \n    }\n\n    private func gotNewWord(string: String) { \n    recognitionResult = string \n    } \n\n# Word2Vec friends and relatives\n\nGloVE, Lexvec FastText.\n\nOne popular alternative to word2vec is GloVe (Global Vectors).\n\nDoc2Vec - Efficient Vector Representation for Documents Through Corruption.\n\n<https:\/\/openreview.net\/pdf?id=B1Igu2ogg>\n\n<https:\/\/github.com\/mchen24\/iclr2017>\n\nBoth models learn geometrical encodings (vectors) of words from their co-occurrence information (how frequently they appear together in large text corpora). They differ in that word2vec is a \"predictive\" model, whereas GloVe is a \"count-based\" model. See this paper for more on the distinctions between these two approaches: <http:\/\/clic.cimec.unitn.it\/marco>\n\nPredictive models learn their vectors in order to improve their predictive ability of Loss(target word | context words; Vectors), that is, the loss of predicting the target words from the context words given the vector representations. In Word2Vec, this is cast as a feed-forward neural network and optimized as such using SGD, and so on.\n\nCount-based models learn their vectors by essentially doing dimensionality reduction on the co-occurrence counts matrix. They first construct a large matrix of (words x context) co-occurrence information, that is, for each \"word\" (the rows), you count how frequently we see this word in some \"context\" (the columns) in a large corpus. The number of \"contexts\" is of course large, since it is essentially combinatorial in size. So then they factorize this matrix to yield a lower-dimensional (word x features) matrix, where each row now yields a vector representation for each word. In general, this is done by minimizing a \"reconstruction loss\" which tries to find the lower-dimensional representations which can explain most of the variance in the high-dimensional data. In the specific case of GloVe, the counts matrix is preprocessed by normalizing the counts and log-smoothing them. This turns out to be A Good Thing in terms of the quality of the learned representations.\n\nHowever, as pointed out, when we control for all the training hyper-parameters, the embeddings generated using the two methods tend to perform very similarly in downstream NLP tasks. The additional benefits of GloVe over word2vec is that it is easier to parallelize the implementation which means it's easier to train over more data, which, with these models, is always A Good Thing.\n\nAnother related technique is latent semantic analysis. macOS SDK includes its implementation as Latent Semantic Mapping framework. LSM algorithm takes text documents (a lot of them), calculates term frequency vectors, and reduces the dimensionality of the obtained vector space. Having such space, you can then determine the topic of previously unseen documents, or calculate, how similar two documents are. Latent semantic mapping powers Junk Mail Filter, Parental Controls, Kanji Text Input, and Help in macOS. You can use it to improve such features as document search, sorting, filtering, classification and retrieval.\n\nmacOS includes command line tool lsm:\n\n<https:\/\/developer.apple.com\/legacy\/library\/documentation\/Darwin\/Reference\/ManPages\/man1\/lsm.1.html>\n\n<http:\/\/www.extinguishedscholar.com\/wpglob\/?p=297>\n\n<https:\/\/en.wikipedia.org\/wiki\/Latent_semantic_analysis>\n\nWatch WWDC 2011 session _Latent Semantic Mapping: Exposing the Meaning behind Words and Documents_ for more details of the API and algorithm as well as for general useful advices on doing machine learning:\n\n<https:\/\/developer.apple.com\/videos\/play\/wwdc2011\/136\/>\n\n# Where to go from here?\n\nWord embeddings are such an elegant idea that they immediately became an indispensable part of many applications in NLP and other domains. Here are several possible directions for your further exploration:\n\n  * You can easily transform the _Word Association_ game into a question-answer system by replacing vectors of words with vectors of sentences. The simplest way to get the sentence vectors is by adding all the word vectors together. Interestingly, such sentence vectors still keep the semantics, so you can use them to find similar sentences.\n  * Using clustering on embedding vectors, you can separate words, sentences, and documents into groups by similarity.\n  * As we have mentioned, Word2Vec vectors are popular as parts of the more complex NLP pipelines. For example, you can feed them into a neural network or some other machine learning algorithm. In this way, you can train a classifier for pieces of text, for example, to recognize text sentiments or topics.\n  * Word2Vec itself is just a compression algorithm; it doesn't know anything about languages or people. You can run it on anything similar to natural text and get equally good results: Code2Vec, Logs2Vec, Playlist2Vec, and so on.\n\n# Summary\n\nFor developing applications that can understand voice or text input, we use techniques from the natural language processing domain. We have just seen several widely used ways to preprocess texts: tokenization, stop words removal, stemming, lemmatization, POS tagging, and named entity recognition.\n\nWord embedding algorithms, and mainly Word2Vec, draw inspiration from the distributive semantics hypothesis, which states that the meaning of the word is defined by its context. Using an autoencoder-like neural network, we learn fixed-size vectors for each word in a text corpus. Effectively, this neural network captures the context of the word and encodes it in the corresponding vector. Then, using linear algebra operations with those vectors, we can discover different interesting relationships between words. For example, it allows us to find semantically close words (cosine similarity between vectors).\n\nIn the next section of the book, we are going to dig deeper into some practical questions of machine learning. We will start with an overview of existing iOS-compatible machine learning libraries.\n\n# Machine Learning Libraries\n\nThis chapter is an overview of existing iOS-compatible libraries for machine learning. We will look at important general-purpose machine learning libraries, frameworks, and APIs, as well as some domain-specific libraries.\n\nIn this chapter, we will cover the following topics:\n\n  * What third-party ML libraries and APIs are available for iOS developers\n  * Overview of some existing Swift-compatible ML libraries and their features\n  * How to use non-Swift libraries in Swift iOS project\n  * What low-level acceleration libraries are available for iOS\n\n# Machine learning and AI APIs\n\nWhen adding artificial intelligence to your application, it's not always necessary to write something from scratch, or even use a library. Many cloud providers offer data processing and analysis as a service. Almost all internet giants provide machine learning in the cloud in some form. In addition, plenty of smaller players are present in the market, and they often provide services of comparable quality at a similar cost. The greatest drawback of such small companies is that they tend to be rapidly acquired by major players. After that, their services are being merged with the services of the big companies (at best), or just being closed (at worst).\n\nThe range of such services is constantly growing and changing, so I do not see much point in a detailed consideration of these services in this book. I provide a list of services here that have not disappeared during the last year, and are unlikely to vanish in the coming year:\n\n  * **Amazon Machine Learning** provides general-purpose machine learning. Go to: <https:\/\/aws.amazon.com\/documentation\/machine-learning\/>.\n  * **Google Cloud Platform** provides general-purpose machine learning, computer vision, NLP, speech recognition, text translation, and so on. Go to: <https:\/\/cloud.google.com\/products\/machine-learning\/>.\n  * **IBM Watson** services include NLP, text to speech, speech to text, computer vision, and data analytics. Go to: http:\/\/www.ibm.com\/watson\/developercloud\/.\n  * **Microsoft Cognitive Services** include computer vision, speech recognition, NLP, search, bot framework, and others. Go to: https:\/\/www.microsoft.com\/cognitive-services\/en-us\/apis.\n  * **Microsoft Azure Machine Learning** is a cloud-based engine for training and deploying your own models. Go to: https:\/\/azure.microsoft.com\/en-us\/services\/machine-learning\/.\n  * **Wit.ai** (acquired by Facebook) provides speech recognition and intents understanding. Go to: https:\/\/wit.ai\/.\n\n# Libraries\n\nAs iOS developers, we are primarily interested in compatible high-performance libraries with a low memory footprint. Swift is a relatively young programming language, so the libraries for machine learning written in it are mostly amateur attempts. However, several more professional and actively growing Swift machine learning packages already exist.\n\nStill, it would be unwise to neglect libraries written in other iOS-compatible languages \u200b\u200bsuch as Objective-C, C, C ++, Lua, and JavaScript, because often they are much more mature and have a broad community. Several reliable cross-platform libraries are worth mentioning in this context.\n\nImporting C libraries into your Swift code is straightforward: you can read about C-Swift interoperability at Apple's Developer portal. Technically, C++ libraries are not compatible with Swift, but you can bridge them via Objective-C so that no C++ or Objective-C++ headers are visible to Swift. Fortunately, Objective C integrates with C++ smoothly.\n\nLua can be compiled as a standalone C library and can be included in the project. You can use JavaScript libraries with the help of the `CoreJavaScript` framework from iOS SDK.\n\n# General-purpose machine learning libraries\n\nIn the following comparison tables, I have included around twenty libraries for machine learning. I considered such characteristics as the language of implementation and interface, the availability and type of acceleration, license type, ongoing development status, and compatibility with popular package managers. Later in this chapter, we will look at the unique features of each library in more detail.\n\nTable 2.1: Comparison of general-purpose machine learning libraries for iOS (part 1):\n\n**Library** | **Language** | **Algorithms**\n\n---|---|---\n\nAIToolbox | Swift | LinReg, LogReg, GMM, MDP, SVM, NN, PCA, k-means, genetic algorithms, DL: LSTM, CNN.\n\nBrainCore | Swift | DL: FF, LSTM.\n\nCaffe, Caffe2, MXNet, TensorFlow, tiny-dnn | C++ | DL.\n\ndlib | C++ | Bayesian networks, SVMs, regressions, structured prediction, DL, clustering and other unsupervised, semi-supervised, reinforcement learning, feature selection.\n\nFANN | C | NNs.\n\nLearnKit | ObjC | Anomaly detection, collaborative filtering, decision trees, random forest, k-means, kNN, regressions, naive Bayes, NNs, PCA, SVMs.\n\nMLKit | Swift | Regressions, genetic algorithms, k-means, NN.\n\nmultilinear-math | Swift | Multilinear PCA, multilinear subspace learning, LinReg, LogReg, FF NN.\n\nOpenCV (`ml` module) | C++ | Normal Bayes, kNN, SVM, decision trees, boosting, gradient boosted trees, random trees, extremely randomized trees, expectation-maximization, NN, hierarchical clustering.\n\nShark | C++ | Supervised: Linear discriminant analysis, LinReg, SVMs, FF and recurrent NNs, radial basis function networks, regularization networks, Gaussian processes, kNN, decision trees, random forests.\n\nUnsupervised: PCA, RBM, hierarchical clustering, evolutionary algorithms.\n\nSwix | Swift | SVM, kNN, PCA.\n\nTorch | Lua | DL.\n\nYCML | ObjC | LinReg, SVM, extreme learning machines, forward selection, kernel process regression, binary RBM, feature learning, ranking, and others.\n\nTable 2.2: Comparison of general-purpose machine learning libraries for iOS (part 2):\n\n**Library** | **Acceleration** | **License** | **Development** | **Package Manager**\n\n---|---|---|---|---\n\nAIToolbox | Accelerate, Metal | Apache-2.0 | Active | -\n\nBrainCore | Metal | MIT | Inactive | CocoaPods, Carthage\n\nCaffe | -\/CUDA | BSD-2-Clause | Active | hunter\n\nCaffe2 | Metal\/NNPack | Custom, BSD-like | Active | CocoaPods\n\ndlib | - | Boost | Active | hunter\n\nFANN | - | LGPL | Inactive | CocoaPods\n\nLearnKit | Accelerate | MIT | Active | -\n\nMLKit | - | MIT | Active | Carthage\n\nmultilinear-math | Metal | Apache | Active | Swift Package Manager\n\nMXNet | -\/ CUDA, OpenMP, | Apache-2.0 | Active | -\n\nOpenCV (`ml` module) | Accelerate\/ CUDA, OpenCL, and others | BSD-3-Clause | Active | CocoaPods, hunter\n\nShark | - | GPL-3.0 | Inactive | CocoaPods\n\nSwix | Accelerate, OpenCV | MIT | Inactive | -\n\nTensorFlow | ?\/ CUDA | Apache-2.0 | Active | -\n\ntiny-dnn | -\/CUDA, OpenCL, OpenMP, and others | BSD-3-Clause | Active | hunter\n\nTorch | -\/ CUDA, OpenCL | BSD-3-Clause | Active | -\n\nYCML | Accelerate | GPL-3.0 | Active | -\n\nThe following is the list of abbreviations:\n\n  * **Convolution neural networks** ( **CNN** ) \n  * **Deep learning** ( **DL** ) \n  * **Feed-forward** ( **FF** ) \n  * **Gaussian mixture model** ( **GMM** ) \n  * **K-nearest neighbors** ( **KNN** ) \n  * **Linear regression** ( **LinReg** ) \n  * **Logistic regression** ( **LogReg** ) \n  * **Long short-term memory** ( **LSTM** ) \n  * **Markov decision process** ( **MDP** ) \n  * **Neural network** ( **NN** ) \n  * **Principal components analysis** ( **PCA** ) \n  * **Support vector machine** ( **SVM** )\n\nPackage managers\n\n  * Carthage: <https:\/\/github.com\/Carthage\/Carthage>\n  * CocoaPods: <https:\/\/cocoapods.org\/>\n  * Hunter: <https:\/\/github.com\/ruslo\/hunter>\n  * Swift Package Manager: https:\/\/swift.org\/package-manager\/\n\n# AIToolbox\n\nThe Swift library contains multiple machine learning models. They are compatible with both iOS and macOS. All models are implemented as separate classes with unified interfaces, so you can replace one model with another in your code with minimal effort. Some models support saving to `plist` files and loading from such files. The Accelerate framework is used throughout the library to boost the speed of calculations.\n\nFor regression tasks, you can choose between linear, nonlinear, and SVM regression. Linear regression supports regularization. The SVM model here is a port of the `libSVM` library initially written in C and can also be used for classification. Other classification algorithms include logistic regression and neural networks. Several types of nonlinearities for neural network layers are present (including convolutions). Your network can be a simple feedforward or recurrent (including LSTM) network, or can combine both types of layers. The Metal framework is used to accelerate neural networks. You can train your networks online or in batch mode.\n\nUnsupervised learning algorithms implemented in the AIToolbox are PCA, K-means, and Gaussian mixture model. MDP can be used for reinforcement learning. Other AI primitives and algorithms present in the library include graphs and trees, alpha-beta algorithms, genetic algorithms, and constraint propagation.\n\nThe library also provides handy classes for plotting descendants from UIView and NSView. There are several modes of plotting such as those to represent functions, classification or regression data, and classification areas. There are also classes for model validation or hyperparameter tuning, like k-fold validation.\n\nGitHub repository: https:\/\/github.com\/KevinCoble\/AIToolbox\n\n# BrainCore\n\nThis Swift library provides feedforward and recurrent neural networks. Several types of layers are present, including the inner product layer, the linear rectifier (ReLU) layer, the sigmoid layer, the RNN and LSTM layer and the L2 loss layer. BrainCore uses Metal acceleration for both training and inference stages. The definition of a new neural network looks very clear because of some pleasant syntactic sugar:\n\n    let net = Net.build { \n        [dataLayer1, dataLayer2] => lstmLayer  \n        lstmLayer =>> ipLayer1 => reluLayer1 => sinkLayer1 \n        lstmLayer =>> ipLayer2 => reluLayer2 => sinkLayer2  \n    }\n\nThe library can be used on both iOS and macOS. It uses the Upsurge mathematical library as a dependency. It is available via CocoaPods or Carthage.\n\nGitHub repository: https:\/\/github.com\/aleph7\/BrainCore\n\n# Caffe\n\nCaffe is one of the most popular deep learning frameworks. It is written in C++.\n\nFrom the official site:\n\n\"Caffe is a deep learning framework made with expression, speed, and modularity in mind. It is developed by the Berkeley Vision and Learning Center (BVLC) and by community contributors.\"\n\nThe library is primarily targeted at Linux and macOS X but unofficial Android, iOS, and Windows ports are also compatible. Caffe supports CUDA acceleration but can also be executed on the CPU alone. On iOS, it uses only CPU. The interfaces include C++, command line, Python, and MATLAB. Caffe provides recurrent and convolutional neural networks. To define a network, you need to describe its structure in a `config` file in a special format.\n\nModelZoo contains many pre-trained Caffe models. Unlike MXNet, Torch, and TensorFlow, it doesn't have automatic differentiation features.\n\nOfficial site: [http:\/\/caffe.berkeleyvision.org  \n](http:\/\/caffe.berkeleyvision.org)iOS port from the BrainCore author: https:\/\/github.com\/aleph7\/caffe\n\n# Caffe2\n\nCaffe2 is a mobile-first deep learning library from Facebook, and started as an attempt to refactor the original Caffe framework. It uses Metal for computation acceleration on the iOS and provides more flexibility than Caffe. For example, it includes recurrent neural networks.\n\nOfficial site: https:\/\/caffe2.ai\/\n\n# dlib\n\n`dlib` is a mature C++ machine learning library with a big community. It includes many advanced ML algorithms that are not present in any other iOS-compatible libraries. It also contains different useful additions, like metaprogramming, compression algorithms, and functions for digital signal and image processing. Porting `dlib` to iOS is relatively straightforward\u2014you need to delete UI and HTTP-related files and then you'll be able to compile it for iOS.\n\nOfficial site: http:\/\/dlib.net\/\n\n# FANN\n\nThe **FANN** (Fast Artificial Neural Network) library is a C implementation of multilayer neural networks. It includes different types of training (backpropagation, evolving topology) and different activation functions. Trained networks can be saved and loaded from file. It is well documented and has bindings to many programming languages. To connect it to your iOS project, use CocoaPods.\n\nOfficial site: http:\/\/leenissen.dk\/fann\/wp\/\n\n# LearnKit\n\nLearnKit is a Cocoa framework written in Objective-C for machine learning. It currently runs on top of the Accelerate framework on iOS and OS X. It supports a big variety of algorithms.\n\nGitHub repository: https:\/\/github.com\/mattrajca\/LearnKit\n\n# MLKit\n\nMLKit provides several regression algorithms: linear, polynomial and ridge regression (+ L2 regularization), lasso regression, k-means, genetic algorithms, and simple neural networks. It also includes classes for data splitting and k-fold model validation. MLKit uses the Upsurge mathematical library as a dependency.\n\nGitHub repository: https:\/\/github.com\/Somnibyte\/MLKit\n\n# Multilinear-math\n\nThe name refers to the tensor operations that this library provides. It also contains a set of machine learning and AI primitives. Its algorithms include principal component analysis, multilinear subspace learning algorithms for dimensionality reduction, linear and logistic regression, stochastic gradient descent, feedforward neural networks, sigmoid, ReLU, Softplus activation functions, and regularizations.\n\nIt also provides a Swift interface to the Accelerate framework and LAPACK, including vector and matrix operations, eigen decomposition, and SVD. On top of that, it implements the `MultidimensionData` protocol to work with multidimensional data.\n\nGitHub repository: https:\/\/github.com\/vincentherrmann\/multilinear-math\n\n# MXNet\n\nQuote from the official site:\n\n\"MXNet is a deep learning framework designed for both efficiency and flexibility.\"\n\nMXNet is compatible with Linux, macOS, Windows, Android, iOS, and JavaScript. Its interfaces include C++, Python, Julia, Matlab, JavaScript, Go, R, and Scala. MXNet supports automatic differentiation and acceleration using OpenMP and CUDA. MXNet is well documented and has a long list of tutorials and examples at its official site. The official site hosts its own Model Zoo with pre-trained neural networks. You can also convert pre-trained Caffe models using the `caffe_converter` tool.\n\nOfficial site: http:\/\/mxnet.readthedocs.org\/en\/latest\/\n\n# Shark\n\nShark is written in C++. It provides methods for linear and nonlinear optimization (evolutionary and gradient-based algorithms), SVMs and neural networks, regression algorithms, decision trees, random forests, and a wide range of unsupervised learning algorithms. An older version of Shark is available on CocoaPod.\n\nOfficial site: [http:\/\/image.diku.dk\/shark\/sphinx_pages\/build\/html\/index.html  \n](http:\/\/image.diku.dk\/shark\/sphinx_pages\/build\/html\/index.html)CocoaPods: https:\/\/cocoapods.org\/pods\/Shark-SDK\n\n# TensorFlow\n\nTensorFlow is a library for numerical computation from Google. It is widely used for deep learning and more traditional statistical learning. The library architecture is built around data flow graphs. The TensorFlow website states that:\n\n\"Nodes in the graph represent mathematical operations, while the graph edges represent the multidimensional data arrays (tensors) communicated between them. The flexible architecture allows you to deploy computation to one or more CPUs or GPUs in a desktop, server, or mobile device with a single API.\"\n\nThe library is so well documented that you can learn ML from scratch using only TensorFlow. The official site includes tons of tutorials, video courses, example apps for different platforms (including iOS), and pre-trained models. Two officially supported APIs are in Python and C++ (limited). It supports acceleration on CUDA GPUs and automatic differentiation. You can convert pre-trained Caffe models using the `caffe-tensorflow` tool.\n\nOfficial site: http:\/\/www.tensorflow.org\/\n\n# tiny-dnn\n\ntiny-dnn is a header-only convolutional neural network framework written in C++.\n\nThe following is an example of neural network creation from the official documentation:\n\n    network<sequential> net; \n    \/\/ add layers \n    net << conv<tan_h>(32, 32, 5, 1, 6)  \/\/ in:32x32x1, 5x5conv, 6fmaps \n        << ave_pool<tan_h>(28, 28, 6, 2) \/\/ in:28x28x6, 2x2pooling \n        << fc<tan_h>(14 * 14 * 6, 120)   \/\/ in:14x14x6, out:120 \n        << fc<identity>(120, 10);        \/\/ in:120,     out:10\n\nYou can train your models or you can convert pre-trained Caffe models using the `caffe_converter` tool. It supports a handful of acceleration types. OpenCV is a dependency for iOS. You can find an iOS example on tiny-dnn's GitHub repository.\n\nGitHub repository: https:\/\/github.com\/tiny-dnn\/tiny-dnn\n\n# Torch\n\nTorch is a framework for scientific computations with wide support for machine learning written in Lua. It is one of the most popular deep learning frameworks. Its supported platforms are Linux, macOS, Windows, Android, and iOS, and it also supports acceleration with CUDA and OpenCL (partially). There are numerous third-party packages that introduce additional capabilities to Torch. Autograd introduces automatic differentiation, `nn` package allows construct neural networks from simple building blocks, the `rnn` package provides recurrent neural networks, and iTorch provides interoperability with IPython Notebook. The Caffe model can be loaded using the `loadcaffe` package. Its library is well documented and easy to install.\n\nThe main problem for Swift developers will be the Lua language itself, because its paradigm is quite different from Swift's one; however, there are libraries that introduce strong typing and functional capabilities to Lua to make life less painful.\n\nOfficial site: [http:\/\/torch.ch\/  \n](http:\/\/torch.ch\/)Unofficial iOS port: https:\/\/github.com\/clementfarabet\/torch-ios\n\n# YCML\n\nYCML is a machine learning framework for Objective-C and Swift (macOS and iOS).\n\nFrom the documentation at official site: http:\/\/yconst.com\/software\/ycml\/:\n\n\"The following algorithms are currently available: Gradient Descent Backpropagation, Resilient Backpropagation (RProp), Extreme Learning Machines (ELM), Forward Selection using Orthogonal Least Squares (for RBF Net), also with the PRESS statistic, Binary Restricted Boltzmann Machines (CD & PCD, Untested!). YCML also contains some optimization algorithms as support for deriving predictive models, although they can be used for any kind of problem: Gradient Descent (Single-Objective, Unconstrained), RProp Gradient Descent (Single-Objective, Unconstrained), NSGA-II (Multi-Objective, Constrained).\"\n\n# Inference-only libraries\n\nWith the release of iOS 11, several popular machine learning frameworks became compatible with it on the model level. You can build and train your models with these frameworks, then export them in a framework-specific format and convert them to the CoreML format for future integration with your app. Such models have fixed parameters and can be used only for inference. As Xcode generates a separate Swift class for each of those models, there is no way to replace or update them in the runtime. The main usage area for such models are different pattern-recognizing applications, like _count your calories by taking a photo_. At the time of writing this book (iOS 11 Beta), CoreML is compatible with the following libraries and models:\n\n  * **Caffe 1.0** : Neural networks\n  * **Keras 1.2.2** : Neural networks\n  * **libSVM 3.22** : SVM\n  * **scikit-learn 0.18.1** : Tree ensembles, generalized linear models, SVM, feature engineering, and pipelines\n  * **XGBoost 0.6** : Gradient-boosted trees\n\nAll these libraries are long-standing industrial standards.\n\nCheck the official package documentation for the latest information: http:\/\/pythonhosted.org\/coremltools\/\n\nThere were also several attempts to implement deep inference on top of the Metal framework prior to Core ML's release:\n\n  * DeepLearningKit\n  * Espresso\n  * Forge\n  * Bender\n\nForge and Bender are still active, because they provide more flexible and clean API than MPS CNN Graph. They will likely become obsolete in the near future, as Apple keeps adding more and more features to Metal Performance Shaders.\n\n# Keras\n\nKeras is a popular Python package for building deep learning neural networks. It has a user-friendly syntax. It's easy and fast to prototype and build your deep models in. It started as a facade for the Theano symbolic computation library, but with time, it also got a TensorFlow backend and finally became a part of TensorFlow. So now, TensorFlow is a default backend, but you still have an option to switch back to Theano. There are also work-in-progress projects on MXNet and CNTK backends.\n\nKeras contains functions for pre-processing most common data types, which include images, texts, and time series.\n\nCoreML supports convolution and recurrent neural networks built in Keras.\n\nWebsite: <https:\/\/keras.io\/>\n\n# LibSVM\n\nClassification, regression, distribution estimation, and anomaly detection, for more information refer to: <https:\/\/www.csie.ntu.edu.tw\/~cjlin\/libsvm\/>.\n\n# Scikit-learn\n\nYou'll be familiar with this library if you have read the previous chapters. It contains a large set of general-purpose learners and data preprocessing methods. Its documentation is awesome.\n\nCoreML supports random forests, generalized linear models, and data built in scikit-learn. For more information refer to: <http:\/\/scikit-learn.org\/>.\n\n# XGBoost\n\nWhen I started writing this book, I didn't expect to write about this tool. Why? Because this is a heavy artillery of machine learning contests. XGBoost is a production standard in many areas, but it is very resource consuming during the training phase: all your gigabytes will be consumed in the blink of an eye. That's why it is mainly used on servers and clusters of servers for web-ranking and other heavy-lifting tasks. It is also a tool that's considered to be a silver bullet for winning a Kaggle machine learning competition (if it's not about computer vision). CoreML supports gradient-boosting decision trees trained in XGBoost. For more information refer to: <https:\/\/xgboost.readthedocs.io\/en\/latest\/>.\n\n# NLP libraries\n\nIn this section we will discuss the various NLP libraries:\n\n# Word2Vec\n\nThis is the original C implementation of the Word2Vec algorithm. It works on iOS, but consumes a significant amount of memory. It was released under the Apache 2.0 license.\n\nGoogle repository: https:\/\/code.google.com\/p\/word2vec\/\n\n# Twitter text\n\nParsing tweets is a common task in NLP. Tweets usually contain some unusual language (like usernames), they mention headers, hashtags, cashtags, and so on. Twitter provides an Objective-C API for tweet processing. This has nothing to do with machine learning per se, but it is still a useful tool for data preprocessing.\n\nGitHub repository: https:\/\/github.com\/twitter\/twitter-text\n\n# Speech recognition\n\nIn this section, we will discuss the frequently used libraries for speech recognition.\n\n# TLSphinx\n\nFrom the documentation:\n\n\"TLSphinx is a Swift wrapper around Pocketsphinx, a portable library, that allow an application to perform speech recognition without the audio ever leaving the device.\"\n\nIt was released under the MIT license.\n\nGitHub repository: https:\/\/github.com\/tryolabs\/TLSphinx\n\n# OpenEars\n\nThis free iOS library provides speech recognition and text-to-speech for Chinese, French, Spanish, English, Dutch, Italian, and German languages. The models were released under different licenses, with some of them being commercially-friendly. It has an Objective-C and Swift APIs. Paid plugins are available.\n\nOfficial site: http:\/\/www.politepix.com\/openears\/\n\n# Computer vision\n\nUnder this section let us look at few computer vision libraries in detail:\n\n# OpenCV\n\nOpenCV is a library of computer vision algorithms, image processing, and general-purpose numerical algorithms. It is implemented in C\/C++ but has interfaces for Python, Java, Ruby, Matlab, Lua, and other languages. It can be freely used for academic and commercial purposes because it is distributed under the BSD license.\n\nSince OpenCV 3.1, there is a DNN module and in OpenCV 3.3, the module has been promoted to `opencv_contrib`.\n\nOfficial site: <http:\/\/opencv.org>  \nAdditional OpenCV modules: <https:\/\/github.com\/opencv\/opencv_contrib>  \nDemo Swift app using OpenCV: https:\/\/github.com\/foundry\/OpenCVSwiftStitch\n\n# ccv\n\nccv is another C++ computer vision library for iOS, macOS, Android, Linux FreeBSD, and Windows. It is distributed under the BSD three-clause license.\n\nFrom the official site:\n\n\"One core concept of ccv development is application driven. Thus, ccv ends up implementing a handful state-of-art algorithms. It includes a close to state-of-the-art image classifier, a state-of-the-art frontal face detector, reasonable collection of object detectors for pedestrians and cars, a useful text detection algorithm, a long-term general object tracking algorithm, and the long-standing feature point extraction algorithm.\"\n\nOfficial site: http:\/\/libccv.org\/\n\n# OpenFace\n\nOpenFace is a state-of-the-art open source library that deals with faces. It includes algorithms for facial landmark detection, eye gaze estimation, head pose estimation, and facial action unit recognition.\n\nGitHub repository: [https:\/\/github.com\/TadasBaltrusaitis\/OpenFace  \n](https:\/\/github.com\/TadasBaltrusaitis\/OpenFace)Unofficial iOS port: https:\/\/github.com\/FaceAR\/OpenFaceIOS\n\n# Tesseract\n\nTesseract is an open source tool for **optical character recognition** ( **OCR** ) written in C++. It has bindings to many programming languages, including two for Objective-C. You can use it to train your own model or you can use one of the community-trained models (including Ancient Greek, Latin, Hebrew, Farsi, and Polish). The latest version of Tesseract uses the LSTM neural network for character recognition. The library is available under the Apache-2.0 license.\n\nGitHub repository: https:\/\/github.com\/tesseract-ocr\/tesseract\n\n# Low-level subroutine libraries\n\nSome libraries are not doing machine learning on their own, but provide important low-level primitives for it. An example of such a library is Apple BNNS. It is a part of the Accelerate framework which provides highly optimized subroutines for convolutional neural networks. We'll discuss it in much detail in Chapter 10, _Natural Language Processing_. In the following section, we'll list some third-party libraries of its kind.\n\n# Eigen\n\nEigen is a C++ template library implementing linear algebra primitives and related algorithms. It is under the LGPL3+ license. Many popular computationally-heavy projects (TensorFlow, for instance) rely on it for matrix and vector operations.\n\nOfficial site: eigen.tuxfamily.org\n\n# fmincg-c\n\nfmincg-c conjugates gradient implementation in C. It uses OpenCL to process algorithms faster. There are some examples written in Python.\n\nGitHub repository: https:\/\/github.com\/gautambhatrcb\/fmincg-c\n\n# IntuneFeatures\n\nAudio feature extraction. The IntuneFeatures framework contains code to generate features from audio files and feature labels from the respective MIDI files. It currently supports these features: Log-scale spectrum power estimate by bands, spectrum power flux, peak power, peak power flux, and peak locations.\n\nThe `CompileFeatures` command line app takes audio and MIDI files as input and generates HDF5 databases with the features and the labels. These HDF5 files can then be used to train a neural network for transcription or related tasks.\n\nGitHub repository: https:\/\/github.com\/venturemedia\/intune-features\n\n# SigmaSwiftStatistics\n\nThis library is a collection of functions that perform statistical calculations in Swift. It can be used in Swift apps for Apple devices and in open source Swift programs on other platforms.\n\nGitHub repository: https:\/\/github.com\/evgenyneu\/SigmaSwiftStatistics\n\n# STEM\n\nSTEM is a Swift Tensor library for machine learning in some way similar to Torch. It provides tensors, operations on them, random tensors generation, computational graphs, and optimization.\n\nGitHub repository: https:\/\/github.com\/abeschneider\/stem\n\n# Swix\n\nSwix is an attempt to implement the NumPy mathematical library in Swift. It wraps OpenCV for some machine learning algorithms and provides Swift API with the Accelerate framework.\n\nGitHub repository: https:\/\/github.com\/scottsievert\/swix\n\n# LibXtract\n\nLibXtract is a simple, portable, lightweight library of audio feature extraction functions. The purpose of the library is to provide a relatively exhaustive set of feature extraction primitives that are designed to be 'cascaded' to create extraction hierarchies.\n\nFor example, `variance`, `average deviation`, `skewness`, and `kurtosis` all require the `mean` of the input vector to be precomputed. However, rather than compute the `mean` and `inside` of each function, it is expected that the `mean` will be passed in as an argument. This means that if the user wishes to use all of these features, the mean is calculated only once, and then passed to any functions that require it.\n\nThis philosophy of **cascading** features is followed throughout the library; for example, with features that operate on the magnitude spectrum of a signal vector (for example, irregularity). The magnitude spectrum is not calculated `inside` the respective function; instead, a pointer to the first element in an array containing the magnitude spectrum is passed in as an argument.\n\nHopefully this not only makes the library more efficient when computing a large number of features, but also makes it more flexible because extraction functions can be combined arbitrarily (one can take the irregularity of the Mel Frequency Cepstral Coefficients, for example).\n\nA complete list of features can be found by viewing the header files, or reading the doxygen documentation, available with this package.\n\nGitHub repository: https:\/\/github.com\/jamiebullock\/LibXtract.\n\n# libLBFGS\n\nGitHub repository: <https:\/\/github.com\/chokkan\/liblbfgs>\n\nL-BFGS method of numerical optimization.\n\n# NNPACK\n\nNNPACK is an Acceleration package for neural networks on multi-core CPUs. Caffe2, tiny-dnn, and MXNet support NNPACK acceleration. Prisma uses this library in the mobile app to boost performance.\n\nGitHub repository: https:\/\/github.com\/Maratyszcza\/NNPACK.\n\n# Upsurge\n\nUpsurge is a SIMD-accelerated Swift library. It is a math utility for matrices, tensors, operators, and functions from the Accelerate framework, much like convolution. Matrix operations are its strength.\n\nGitHub repository: https:\/\/github.com\/aleph7\/Upsurge.\n\n# YCMatrix\n\nYCMatrix is a Matrix operations Objective-C library. It is essentially a wrapper around the Accelerate framework. YCML uses it for all computation accelerations.\n\nGitHub repository: https:\/\/github.com\/yconst\/YCMatrix.\n\n# Choosing a deep learning framework\n\nChoosing the correct deep learning framework is important to get the optimal speed and model size you desire. There are several things to consider\u2014overhead, added by the library, GPU acceleration, do you need training or inference only?, in which framework existing solutions were implemented.\n\nYou should understand that you don't always need GPU acceleration. Sometimes, SIMD\/Accelerate is more than enough to implement neural networks that do inference in real-time.\n\nSometimes, you have to consider whether the calculations are going to be done on the client side, on the server side, or if they will be balanced between both. Try to do benchmarks with an extreme number of records, and test them with different devices.\n\n# Summary\n\nIn this chapter, we learned about iOS compatible machine learning libraries and their features. We discussed five major deep learning frameworks: Caffe, TensorFlow, MXNet, and Torch, and so on. We also mentioned several smaller deep learning libraries and tools to convert deep learning models from one format to another. Among general purpose machine learning libraries, the most feature-rich and mature are AIToolbox, dlib, Shark, and YCML. NLP libraries for iOS are rare and restricted in their capabilities.\n\nIn addition to native iOS speech recognition and text-to-speech, there are several free and commercial libraries that provide the same functionality.\n\nIf you have some common computer vision tasks, you can find the appropriate algorithms in OpenCV or ccv libraries. OCR and all kinds of face-related tasks can also be performed using the open source toolchain. There are also several low-level libraries for linear algebra operations, tensors, and optimization that you can use to accelerate your ML algorithms.\n\n# Optimizing Neural Networks for Mobile Devices\n\nModern convolutional neural networks can be huge. For example, the pre-trained ResNet family network can be from 100 to 1,000 layers deep, and take from 138 MB to 0.5 GB in Torch data format. To deploy them to mobile or embedded devices can be problematic, especially if your app requires several models for different tasks. Also, CNNs are computationally heavy, and in some settings (for example, real-time video analysis) can drain device battery in no time. Actually, much faster than it took to write this chapter's intro. But why are they so big, and why do they consume so much energy? And how do we fix it without sacrificing accuracy?\n\nAs we've already discussed the speed optimization in the previous chapter, we are concentrating on the memory consumption in this chapter. We specifically focus on the deep learning neural networks, but we also give several general recommendations applicable to other kinds of machine learning models.\n\nIn this chapter, we will cover the following topics:\n\n  * Why compress the models?\n  * General recommendations for machine learning models compression\n  * Why deep neural networks are big\n  * What factors influence the size of a neural network?\n  * What parts of a neural network are the heaviest?\n  * Methods for model size reduction\u2014parameter number reduction, pruning, trained quantization, and Huffman coding\n  * Compact CNN architectures\n\n# Delivering perfect user experience\n\nAccording to the iTunes Connect Developer Guide, the total uncompressed size of the app should be less than 4 GB (as of December 15, 2017); however, this applies only to the binary itself, while asset files can take as much space as the disk capacity allows. There is also a limit on app size for the cellular download, as stated on the Apple Developer site (<https:\/\/developer.apple.com\/news\/?id=09192017b>):\n\n\"We've increased the cellular download limit from 100 MB to 150 MB, letting customers download more apps from the App Store over their cellular network.\"\n\nThe simple conclusion is that you'd better store you model parameters as on-demand resources, or download them from your server after the app is already installed; but this is only one half of the problem. The other half is that you really don't want your app to take a lot of space and consume tons of traffic, because this is a bad user experience.\n\nWe can attack the problem from several directions (from the easiest to the most complex):\n\n  * Use standard lossless compression algorithms\n  * Choose the compact architectures\n  * Prevent models from growing too big\n  * Use lossy compression techniques\u2014remove unimportant model parts\n\nThe first approach is only a half-measure, because you will still have to decompress your model in runtime. In the last case, we usually talk about reducing the number of the model's parameters, effectively reducing its capacity, and subsequently, accuracy.\n\n# Calculating the size of a convolutional neural network\n\nLet's take some well-known CNN, say VGG16, and see in detail how exactly the memory is being spent. You can print the summary of it using Keras:\n\n    from keras.applications import VGG16  \n    model = VGG16()  \n    print(model.summary())\n\nThe network consists of 13 2D-convolutional layers (with 3\u00d73 filters, stride 1 and pad 1) and 3 fully connected layers (\"Dense\"). Plus, there are an input layer, 5 max-pooling layers and a flatten layer, which do not hold parameters.\n\n**Layer** | **Output shape** | **Data memory** | **Parameters** | **Number of p** arameters\n\n---|---|---|---|---\n\nInputLayer | 224\u00d7224\u00d73 | 150528 | 0 | 0\n\nConv2D | 224\u00d7224\u00d764 | 3211264 | 3\u00d73\u00d73\u00d764+64 | 1792\n\nConv2D | 224\u00d7224\u00d764 | 3211264 | 3\u00d73\u00d764\u00d764+64 | 36928\n\nMaxPool2D | 112\u00d7112\u00d764 | 802816 | 0 | 0\n\nConv2D | 112\u00d7112\u00d7128 | 1605632 | 3\u00d73\u00d764\u00d7128+128 | 73856\n\nConv2D | 112\u00d7112\u00d7128 | 1605632 | 3\u00d73\u00d7128\u00d7128+128 | 147584\n\nMaxPool2D | 56\u00d756\u00d7128 | 401408 | 0 | 0\n\nConv2D | 56\u00d756\u00d7256 | 802816 | 3\u00d73\u00d7128\u00d7256+256 | 295168\n\nConv2D | 56\u00d756\u00d7256 | 802816 | 3\u00d73\u00d7256\u00d7256+256 | 590080\n\nConv2D | 56\u00d756\u00d7256 | 802816 | 3\u00d73\u00d7256\u00d7256+256 | 590080\n\nMaxPool2D | 28\u00d728\u00d7256 | 200704 | 0 | 0\n\nConv2D | 28\u00d728\u00d7512 | 401408 | 3\u00d73\u00d7256\u00d7512+512 | 1180160\n\nConv2D | 28\u00d728\u00d7512 | 401408 | 3\u00d73\u00d7512\u00d7512+512 | 2359808\n\nConv2D | 28\u00d728\u00d7512 | 401408 | 3\u00d73\u00d7512\u00d7512+512 | 2359808\n\nMaxPool2D | 14\u00d714\u00d7512 | 100352 | 0 | 0\n\nConv2D | 14\u00d714\u00d7512 | 100352 | 3\u00d73\u00d7512\u00d7512+512 | 2359808\n\nConv2D | 14\u00d714\u00d7512 | 100352 | 3\u00d73\u00d7512\u00d7512+512 | 2359808\n\nConv2D | 14\u00d714\u00d7512 | 100352 | 3\u00d73\u00d7512\u00d7512+512 | 2359808\n\nMaxPool2D | 7\u00d77\u00d7512 | 25088 | 0 | 0\n\nFlatten | 25088 | 0 | 0 | 0\n\nDense | 4096 | 4096 | 7\u00d77\u00d7512\u00d74096+4096 | 102764544\n\nDense | 4096 | 4096 | 4097\u00d74096 | 16781312\n\nDense | 1000 | 1000 | 4097\u00d71000 | 4097000\n\nTotal memory for data: Batch_size \u00d7 15,237,608 \u2248 15 M\n\n???Total memory: Batch_size \u00d7 24M 5; 4 bytes \u2248 93 MB\n\nReference:\n\n<http:\/\/cs231n.github.io\/convolutional-networks\/#case>  \n<https:\/\/datascience.stackexchange.com\/questions\/17286\/cnn-memory-consumption>\n\nTotal parameters: 138,357,544\u2248138M\n\n# Lossless compression\n\nA typical neural network contains a significant amount of redundant information. This enables us to apply both lossless and lossy compression to them, and often achieve fairly good results.\n\nHuffman encoding is a type of compression that is commonly referred to in research papers concerning CNN compression. You can also use Apple compression or Facebook `zstd` libraries, which deliver state-of-the-art compression. Apple compression contains four compression algorithms (three common and one Apple-specific):\n\n  * LZ4 is the fastest of the four.\n  * ZLIB is standard zip archiving.\n  * LZMA is slower but delivers the best compression.\n  * LZFSE is a bit faster and delivers slightly better compression than ZLIB. It is optimized for the Apple hardware to be energy efficient.\n\nHere is a code snippet for you to compress data using the LZFSE algorithm from the compression library, and decompress it back. You can find the full code in the `Compression.playground`:\n\n    import Compression \n    let data = ...\n\n`sourceSize` holds the size of the data before compression:\n\n    let sourceSize = data.count\n\nAllocating the buffer for the results of compression... we allocate it with the original (non-compressed) size:\n\n    let compressedBuffer = UnsafeMutablePointer<UInt8>.allocate(capacity: sourceSize)\n\n`compression_encode_buffer()` is the function used to compress your data. It takes the input and output buffers, their sizes, and the type of compression algorithm (`COMPRESSION_LZFSE`) and returns the size of compressed data:\n\n    var compressedSize: Int = 0 \n    data.withUnsafeBytes { (sourceBuffer: UnsafePointer<UInt8>) in \n    compressedSize = compression_encode_buffer(compressedBuffer, sourceSize, sourceBuffer, sourceSize, nil, COMPRESSION_LZFSE) \n    }\n\nThe `compressedSize` variable holds the size after compression.\n\nNow, to the decompression. Here's how to allocate a buffer of appropriate size for un-compressed data:\n\n    var uncompressedBuffer = UnsafeMutablePointer<UInt8>.allocate(capacity: sourceSize)\n\nAgain, the `compression_decode_buffer()` function returns the true size of the uncompressed data:\n\n    let uncompressedSize = compression_decode_buffer(uncompressedBuffer, sourceSize, compressedBuffer, compressedSize, nil, COMPRESSION_LZFSE)\n\nConverting the buffer to a normal data object:\n\n    let uncompressedData = Data(bytes: uncompressedBuffer, count: uncompressedSize)\n\n`uncompressedData.count` should be equal to the initial `sourceSize`.\n\nFor lossless compression to be effective, your network needs to have a lot of repetitive elements in its structure. This can be achieved using weights quantization of precision reduction (see the next section).\n\nApple lzfse compression library:\n\n  * https:\/\/github.com\/lzfse\/lzfse\n  * https:\/\/developer.apple.com\/reference\/compression\/data_compression\n\nFacebook zstd compression library:\n\n  * https:\/\/github.com\/facebook\/zstd\n  * https:\/\/github.com\/omniprog\/SwiftZSTD\n\n# Compact CNN architectures\n\nDuring the inference, the whole neural network should be loaded into the memory, so as mobile developers we are especially interested in the small architectures, which consume as little memory as possible. Small neural networks also allow to reduce the bandwidth consumption when downloaded from the network.\n\nSeveral architectures designed to reduce the size of convolutional neural networks have been proposed recently. We will discuss in brief several most known of them.\n\n# SqueezeNet\n\nThe architecture was proposed by Iandola et al. in 2017 for use in autonomous cars. As the baseline, researchers took the AlexNet architecture. This network takes 240 MB of memory, which is pretty much the equivalent of mobile devices. SqueezeNet has 50x fewer parameters, and achieves the same level of accuracy on the ImageNet dataset. Using additional compression, its size can be reduced to about 0.5 MB.\n\nSqueezeNet is built from the fire modules. The objective was to create a neural network with a small number of parameters, but preserving the competitive level of accuracy. It was done with the following approaches:\n\n  * Reduce the network size by replacing the 3 x 3 filters with 1 x 1 filters.  \nHere, by replacing the 3 x 3 filter with a 1 x 1 filter, we have an instant reduction in the number of parameters by 9x.\n  * Reduce the number of inputs for the remaining 3 x 3 filters. Here, the number of parameters are reduced by merely reducing the number of filters.\n  * Downsample late in the architecture for the convolution layers to have a larger activation map. To improve the classification accuracy, the authors of SqueezeNet decreased the stride with later convolution layers, and therefore created a larger activation\/feature map.\n\nThe original paper can be found here: https:\/\/arxiv.org\/abs\/1602.07360.\n\n# MobileNets\n\n**MobileNets** is a class of efficient CNNs targeted to mobile and embedded applications. It was proposed by the Google research team in _MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications_ , 2017. In comparison to the traditional CNN, it has much less parameters, and requires much less computation of the learning and prediction process. This makes it faster and lighter, preserving the predictions accuracy at the same time. The main innovation was the introduction of depth-wise separable convolutions: <http:\/\/machinethink.net\/blog\/googles-mobile-net-architecture-on-iphone\/>.\n\nThe original paper can be found here: <https:\/\/arxiv.org\/abs\/1704.04861>.\n\n# ShuffleNet\n\nShuffleNet architecture was proposed in 2017 by the research team from Face++ (Megvii Inc.). It is targeted on mobile devices with limited computation power (for example, 10 - 150 MFLOPs). In comparison to the classical CNN, ShuffleNet has less parameters and performs less computations, because it uses pointwise group convolution and channel shuffle; for instance, it works 13x faster than AlexNet. The accuracy remains the same: on ImageNet, it even performs slightly better (top-1 error metric) than MobileNet.\n\nThe original paper can be found here: <https:\/\/arxiv.org\/abs\/1707.01083>.\n\n# CondenseNet\n\nCondenseNet was proposed by Gao Huang, Shichen Liu, Laurens van der Maaten, and Kilian Q. Weinberger. It reaches unprecedented levels of efficiency by combining dense connectivity between layers with a mechanism to remove unused connections, and therefore enables reuse of features within the network. CondenseNet is believed to be much more efficient than the state-of-the-art compact convolutional networks, such as MobileNets and ShuffleNets.\n\nRefer to this: _CondenseNet: An Efficient DenseNet using Learned Group Convolutions_ , Gao Huang, Shichen Liu, Laurens van der Maaten, Kilian Q. Weinberger, November 25, 2017: <https:\/\/arxiv.org\/abs\/1711.09224>.\n\n# Preventing a neural network from growing big\n\nTo leverage cutting-edge deep learning networks on mobile platforms, it becomes extremely important to effectively tune the learning of a network such that we can do the most with the least resources. The implementation of the neural network for OCR by the Google Translate team is an interesting one to understand the few thumb rules to circumvent the network from growing too big.\n\nFollowing are excerpts from the press release from Google, found at: <https:\/\/translate.googleblog.com\/2015\/07\/how-google-translate-squeezes-deep.html>:\n\n\"We needed to develop a very small neural net, and put severe limits on how much we tried to teach it-in essence, put an upper bound on the density of information it handles. The challenge here was in creating the most effective training data. Since we're generating our own training data, we put a lot of effort into including just the right data and nothing more. For instance, we want to be able to recognize a letter with a small amount of rotation, but not too much. If we overdo the rotation, the neural network will use too much of its information density on unimportant things. So we put effort into making tools that would give us a fast iteration time and good visualizations. Inside of a few minutes, we can change the algorithms for generating training data, generate it, retrain, and visualize. From there we can look at what kind of letters are failing and why. At one point, we were warping our training data too much, and '$' started to be recognized as 'S'. We were able to quickly identify that and adjust the warping parameters to fix the problem. (Good, 2015)\"\n\nHere are the key takeaways from the above notes:\n\n  * Limit the learning capacity by limiting the variations within the training data.\n  * Effective training data can be created by augmenting only a small rotation in the images. Larger rotations would result in increased learning, and therefore increased size.\n  * Extensively leverage visualization to quickly fix incorrect results from the network.\n\nWhat general rules can we derive from these revelations?\n\n  * Put an upper bound for the model size. This will limit the capacity of your model.\n  * Create the most effective training data and make the task of you network as simple as possible. For example, if neural network is recognizing the characters in the photo, add to the dataset letters rotated just a bit, but don't make it to learn characters flipped upside-down or mirrored. If you are creating your dataset by data augmentation, put efforts in keeping the dataset clean, so the network is not learning anything except things it needs to know.\n  * Which characters you can neglect? For example, you can let the network recognize \"5\" and \"S\" as the same character and handle the problem on the level of the dictionary.\n  * Be sure to visualize, what is going on inside of your network, and in which places it has problems. What characters it confuses the most often?\n\n# Lossy compression\n\nAll lossy methods of compression involve a potential problem: when you lose part of the information from your model, you should check how it performs after this. Retraining on the compressed model will help to adapt the network to the new constraints.\n\nNetwork optimization techniques include:\n\n  * **Weight quantization** : Change computation precision. For example, the model can be trained in full precision (float32) and then compressed to int8. This improves the performance significantly.\n  * Weight pruning\n  * Weight decomposition\n  * Low rank approximation. Good approach for CPU.\n  * **Knowledge distillation** : Train a smaller model to predict an output of the bigger one.\n  * Dynamic memory allocation\n  * Layer and tensor fusion. The idea is to combine successive layers into one. This reduces the memory needed to store intermediate results.\n\nAt the moment, each of them has its own pros and cons, but no doubts, that more perfect techniques will be proposed in the closest future.\n\n  * **Kernel auto-tuning** : Optimizes execution time by choosing the best data layer and best parallel algorithms for the target Jetson, Tesla, or DrivePX GPU platform\n  * **Dynamic tensor memory** : Reduces memory footprint and improves memory re-use by allocating memory for each tensor only for the duration of its usage\n  * **Multi-stream execution** : Scales to multiple input streams, by processing them in parallel using the same model and weights\n\n# Optimizing for inference\n\nGet rid of elements of a graph that are only used for back propagation, and are useless for inference.\n\nFor example, batch normalization layers can be merged with the preceding convolution layers into one layer, because both convolution and batch normalization are linear operations.\n\n# Network pruning\n\nThe general idea behind this method is that not all weights in the neural network are equally important. So, we can reduce the size of the network by throwing out unimportant weights. Technically, this can be done in the following way:\n\n  1. Train a large network: \n    * Leverage any previously trained network, say, VGG16, and retrain only the fully connected layers\n  2. Rank the filters, or create a sparse network based on a criteria: \n    * We could rank each filter by using any feasible criteria (say, Taylor Criteria), and pruning the lowest-ranking filters, or alternatively, replace all the values less than a certain threshold, with zeros resulting in a sparse network\n  3. Fine tune and repeat: \n    * Perform several iterations of training on the sparse network\n\nThe tricky question here is how to decide which networks are not important enough. This can be solved in the same way as we usually choose our hyperparameters: check several thresholds, and compare the quality metrics of the resulting networks.\n\nBe sure to perform several rounds of training on the pruned model to allow it to fix the damage you caused.\n\n# Weights quantization\n\nTo 8 bits or less.\n\nWeights quantization allows to decrease the size of the model, but at expense of prediction accuracy. In any case, it requires the same amount of memory during the runtime. For quantization any general purpose clustering algorithm can be used, for example, k-means.\n\nStandard clustering algorithms applied to the weights. By this we replace all this floating-points numbers with a few bits, representing its cluster. 1 floating point per cluster. Retrain again.\n\n<https:\/\/petewarden.com\/2016\/05\/03\/how-to-quantize-neural-networks-with-tensorflow\/>\n\n<https:\/\/github.com\/tensorflow\/tensorflow\/tree\/master\/tensorflow\/contrib\/quantization>\n\n# Reducing precision\n\nAnother simple approach to reduce the size of the network is to directly convert weights from double\/float data type to another with lower memory size, or to a fixed precision. This (almost) doesn't affect quality of predictions, but allows to reduce the size of the model up to four times.\n\nReducing precision of the network is exclusively focused only after the training is complete. Previously, attempts to train a network with lower precision data types have been experimented, and the results indicated difficulties in handling the back propagation and gradients.\n\nOnce the network is trained, we could right away replace double by float, or even better, by fixed precision. For example, in the trained neural network, you have double weights like this:\n\n    0.954929658551372\n\nIt's highly unlikely that the neural network encodes something meaningful in all those numbers after the point. So, if you drop most of them, converting to float, nothing changes: 0.9549297. The neural networks are stable enough to deal with that kind of insignificant change. But even now, it looks like precision is too big. So, we can round it even more; for example, to 0.9550000. This will not decrease the size of a model in the memory, because weights are still float numbers; but it does reduce the size of your IPA binaries, because archiving can be more efficient. Also, compressed models will take less space on the disk.\n\n# Other approaches\n\nAnother popular approach to reduce the size of a neural network is via SVD. SVD is applied to pre-adjusted neural networks, and therefore reduces the number of parameters in the network. After reducing the number of parameters, an unconventional **backpropagation** algorithm is used to train the models restructured by SVD, which has lower time complexity than the conventional BP algorithm. Experimental results have shown almost 2x improved speed with zero loss in accuracy, and around 4x improvement with minor loss in accuracy.\n\nAdditional reading: you can explore several other approaches adopted by tech giants for the mobile platform:\n\n  * <https:\/\/handong1587.github.io\/deep_learning\/2015\/10\/09\/acceleration-model-compression.html>\n  * <https:\/\/research.googleblog.com\/2017\/02\/on-device-machine-intelligence.html>\n  * <https:\/\/www.slideshare.net\/embeddedvision\/tensorflow-enabling-mobile-and-embedded-machine-intelligence-a-presentation-from-google>\n\n# Facebook's approach in Caffe2\n\nDuring the developer conference, Facebook had recently announced their approach to render cutting-edge art work on images and videos on the phone, while effectively leveraging the computing resources with a highly mobile-optimized deep neural network. The overall approach can be studied using the following visual (<https:\/\/developers.facebook.com\/videos\/f8-2017\/delivering-real-time-ai-in-the-palm-of-your-hand\/>):\n\nFigure 12.1: Facebook pipeline for compressing neural networks\n\nFacebook used the following pipeline in their applications to achieve 50x size reduction, preserving the accuracy:\n\n  * Pruning\n  * Quantization\n  * Huffmann encoding, or standard general-purpose compression algorithms\n\n# Knowledge distillation\n\nKnowledge distillation\u2014you train your model to predict the logits of a more complex model. Use the large model's output as the ground truth to train the small model.\n\nAdditional reading:\n\n  * <https:\/\/arxiv.org\/abs\/1503.02531>\n  * <https:\/\/www.slideshare.net\/AlexanderKorbonits\/distilling-dark-knowledge-from-neural-networks>\n  * <https:\/\/github.com\/chengshengchan\/model_compression\/blob\/master\/teacher-student.py>\n\n# Tools\n\nThe following are the tool is used for lossy compression:\n\nTensorFlow compression tools\n\n# An example of the network compression\n\nYou can find suitable examples of the network compression at the following address:\n\n<https:\/\/github.com\/caffe2\/caffe2\/issues\/472>\n\n# Summary\n\nThere are several ways in which we can achieve appropriate size for deep neural network deployment on mobile platforms. So far, the most popular are choosing the compact architecture, and lossy compression: quantization, pruning, and others. Make sure to check your network's accuracy hasn't degraded after the compression was applied.\n\n# Bibliography\n\n  1. O. Good, _How Google Translate squeezes deep learning onto a phone_ , July 29, 2015: <https:\/\/research.googleblog.com\/2015\/07\/how-google-translate-squeezes-deep.html>\n  2. Y. LeCun, J. S. Denker, S. A. Solla, R. E. Howard, and L. D. Jackel. _Optimal Brain Damage_. In NIPS, volume 2, pages 598\u2013605, 1989\n\n# Best Practices\n\n\"The purpose of a storyteller is not to tell you how to think, but to give you questions to think upon.\"\n\n\u2013 _Brandon Sanderson_ , _The Way of Kings_\n\nImagine the field of AI as a huge national park. In previous chapters, we guided you along several exciting trails and showed you the most interesting sights for mobile developers. But there is still so much more that is unexplored. So, in this chapter, we want to provide you with a map of the common paths, from idea to production. We've outlined dangerous zones and left notes on solo hiking best practices! We also want to point out several interesting directions for your future exploration.\n\nIn this chapter, we will discuss the following topics:\n\n  * The path from idea to production\n  * Common pitfalls in machine learning projects also known as machine learning gremlins\n  * Machine learning best practices\n  * Recommended study resources\n\n# Mobile machine learning project life cycle\n\nWhen developing a mobile machine learning product, you typically go through several stages:\n\n  * Preparatory stage\n  * Prototype creation\n  * Porting to a mobile platform or deployment of the trained model\n  * Production\n\nDepending on your situation, your route may be shorter or longer; but usually, if you have skipped some stage, it just means that someone else did it for you. In the following explanation, we are omitting all the steps that are common to all kinds of mobile app projects and focusing only on the steps specific to machine learning.\n\n# Preparatory stage\n\nThis is the stage where you basically decide what you will do. There can be two possible outcomes for this stage: you have a plan on how to proceed, or you decide that you will not proceed:\n\nFigure 13.1: Preparatory stage map\n\n# Formulate the problem\n\nIf you can solve your problem without machine learning, don't use it. If the task can be solved with traditional programming techniques, congratulations! You don't need machine learning! Furthermore if your problem is of the kind where you can't allow errors, do not use machine learning.\n\nFor the start of your machine learning project, it is necessary to reduce a real-world problem to a machine learning task. Machine learning algorithms were developed by mathematicians and mostly tested on neat data in a controlled environment. You are fine if you can define your problem in terms of some existing machine learning approach: classification, regression, clustering, and so on. But to date, there are many problems that can't be easily adapted to the common machine learning blueprints. Among them are problems that require common sense reasoning and context understanding.\n\n# Define the constraints\n\nIt is easy to get lost in the variety of AI approaches. There is a set of constraints that will help you to focus and identify the optimal path to the solution. By answering the following questions to yourself, you will narrow the scope of your exploration significantly; or maybe, in some cases, you will conclude that the task is impossible (which is better to figure out at an earlier stage than a later one):\n\n  * What data can you use?\n  * What data should you not use?\n  * What should be the input and the output of your model?\n  * What is your desired accuracy or other measure of success? Remember that a machine learning algorithm will not be 100% accurate.\n  * Should the model be able to train on the target platform?\n  * How interpretable should your model be? Is it okay if your model is a black box?\n  * How much disk space and memory can your model consume in the training and inference stages? For example, the model shouldn't take more than 15 MB of disk space and more than 30 MB of RAM during inference.\n  * How fast should (or rather how slow can) the training and the inference be?\n  * What is your programming language and target platform?\n\n# Research the existing approaches\n\nThe key question here is, _how do other people solve similar problems?_\n\nCan you use native iOS SDK to solve your problem? For example, if you want to detect people's faces in photos, you don't need to train your own neural network or Haar cascade. Just use the Vision framework instead. In other words, do not reinvent the wheel. Look for ready solutions that can work on the device or server side. For the most common daily tasks, you will find something appropriate.\n\nExpanding the range of searches, perform a literature review. Even if you've not found ready solutions, you at least will get useful insights into the approaches and the domain specifics. The sites that will be handy at this stage are arXiv, Google Scholar, and GitHub. When you are done, you will have a clear understanding of classical and state-of-the-art approaches to the problem.\n\nEven if you do not find a good enough solution, you will probably find a baseline solution to compare your future models with.\n\n# Research the data\n\nIf you have not found an existing solution and want to train your own algorithm, you will need a dataset.\n\nHere, several scenarios are possible:\n\n  * You are feeling lucky, and you have found an existing dataset. The potential problem is that you may not be the only lucky person and your approach can be copied by others. There can also be licensing issues or other related problems.\n  * You collect or generate your own dataset.\n  * In the case of supervised learning, your dataset should be labeled. Hand-labeling is a laborious task, so it is often outsourced to some third-party services, such as Amazon Mechanical Turk.\n\nCalculate how much will it cost to label your data in terms of time and money.\n\nAnother important thing to mention is that you should have a clear understanding of how the data was collected. This is important because the way of collecting data for model training can be significantly different from the way the same data will be collected in your app, and this will influence the results of your model's work. For example, if all the faces in the dataset were collected using a professional camera with perfect lighting conditions on a white background, do not expect your face recognition model to perform equally well on a mobile phone when the user has a bright window behind his\/her back.\n\nThe question you should ask yourself is, \"If I were a machine learning algorithm, would I be able to perform well, having this data?\" If the data is insufficient, no algorithm will save the day. Remember that more data beats the better algorithm.\n\n# Make design choices\n\nWhen you have a clear understanding of the goals, constraints, competing solutions, and your data, you can start with defining the technical specifics for your future model. The following questions should be answered before you implement your model:\n\n  * Is this a supervised or unsupervised learning problem? Classification or clustering? Discriminative or generative model?\n  * What is the measurement of success? What is your baseline solution and what are your benchmarks? How do you select the best model? In other words, what is the set of metrics that defines the best model?\n  * What is your strategy of model quality evaluation? Accuracy, precision-recall, cross-validation, or something else? This depends mostly on what costs more in your application domain: false positives or false negatives. Choose the quality metrics and set clear goals; for instance, precision shouldn't be less than 80%.\n  * Can the model be trained once and then do inference on all the devices or do you need to train a separate model for each client?\n  * Is data from one user enough for your model to operate or do you need to aggregate data from many users? This question will help you realize whether you need to put your model on server side.\n  * What is more important for you: accuracy or interpretability? For a classification problem, in the first case, you may want to go with neural networks or ensembles; in the second, you may want to go with decision trees or Naive Bayes.\n  * Do you need a probabilistic estimate? And just yes\/no or 42% chance of yes and 58% chance of no?\n  * How do you clean your data? How do you choose good features?\n  * How do you split your data into training and test sets? 50\/50? 90\/10?\n  * Do you want your model to incorporate new data incrementally (online learning) or retrain the model on a whole bunch of data from time to time (batch learning)? Can your training data become outdated? How often does the environment in which your model operates change? Should it adapt or not?\n\n# Prototype creation\n\nIt is important to understand the difference between tools for prototyping and production, because very different requirements are imposed on the instruments during these two stages. Choosing the right tools for the right tasks will save you a lot of time.\n\nDuring the prototyping phase, you want to be able to test your hypotheses and conduct experiments quickly. That is why it is reasonable to choose a flexible programming language within the reach of the environment, such as Python or R. You also want to have tools for data visualization and model debugging. This is something that the Swift ecosystem is still weak at. The matters of model size, speed, and stability may be secondary during prototyping (which doesn't mean that you should put them at the back of your mind). But when you are preparing your solution for production, you see those problems face to face, and in most cases, you have to rely on native, highly optimized libraries. In the search for a universal solution, you risk ending up with tools that work equally badly for both prototyping and production.\n\nImplementing machine learning algorithms from scratch is a non-trivial task. Therefore, if it is possible, choose portable libraries (TensorFlow or OpenCV) or algorithms you know are already implemented for iOS. Otherwise, you will have to spend additional resources to reproduce algorithms written in Python on iOS:\n\nFigure 13.2: Prototype creation map\n\n# Data preprocessing\n\nStart from simple data preprocessing. Be aware that usually data preparation takes 80% of a project's time. Maintain a clean repository with your data and tidy your data up. Remember, **garbage in, garbage out** ( **GIGO** )!\n\nSplit the work into more or less independent chunks. Let's say you are writing an app for reading medical device indications via a phone camera to simplify nurses' work.\n\nWrite down separate chunks of work, their inputs, and outputs. In this way, you'll see which of them depend on each other and this also helps understand how to test each step. Table shows examples:\n\n**Serial number** |\n\n**Step** | **Input** | **Output**\n\n---|---|---|---\n\n1 | Device type recognition | Image | Device type\n\n2 | Device screen detection | Image | Vertices of a screen quadrilateral (points)\n\n3 | Perspective correction | Vertices of a quadrilateral, screen image | Screen image with the corrected perspective\n\n4 | Screen layout segmentation | Screen image, device type | Several images containing different elements of layout\n\n5 | Image preprocessing for OCR | Noisy images | Clean images\n\n6 | OCR | Images | Noisy text\n\n7 | Validation | Noisy text | Clean text\n\nThe data preprocessing pipeline should be documented. For example, if you are subtracting the mean and dividing your data by the standard deviation, when training your model, don't forget to write down exact values of the mean and standard deviation. This is a common problem with pre-trained neural networks on the internet. When authors forget to mention the preprocessing steps, the models became effectively useless.\n\nFor a classification task, dataset preprocessing usually includes engineering of informative features, class balancing, and missed values imputation. In the case of supervised learning, do not forget to split your dataset into three parts for the next stage: training set (most of your samples), test set, and validation set.\n\n# Model training, evaluation, and selection\n\nUsually, it is better to start from simple and classical models because sometimes the simplest model performs the best. But this is just a rule of thumb, not a law of nature.\n\nEvery machine learning algorithm embodies some assumptions or prior knowledge about the data: KNN assumes that similar examples are of the same class, linear regression assumes linear dependencies and normally distributed errors, many models assume independence or limited dependencies between features or samples, and so on. This helps them to generalize behind the training data successfully. All these assumptions work only because the samples are not distributed uniformly across the space of all possible inputs, and there is something we can call pattern in the data. The task of the machine learner engineer\/researcher is to know his data well enough to be able to make grounded assumptions about it. He chooses the algorithms based on those assumptions. What helps in practice is to ask yourself, \"If I were in the place of my algorithm, would I be able to generalize well with such features, number of samples, and assumptions?\" Think! What kind of knowledge do you have about the data? If you understand what precondition should be fulfilled for the sample to fall in one or another class (such as \"if it has four paws, then this is a cat, but if it has...\" then decision tree is your choice. If the similarity between instances is something that is quite well understood, then go for distance-based algorithms. If there is a lot of information about probabilistic dependencies in your data, try probabilistic graphical models.\n\nThe typical procedure for choosing the best model is as follows:\n\n  * Choose a set of models that look appropriate for your task. For example, for classification, this can be: KNN, logistic regression, decision tree, neural network, and so on.\n  * Use the training set to train your models and test set to validate their accuracy; adjust their hyperparameters (number of neighbors for KNN, number of decision tree splits, and number and types of layers for NN).\n  * When you have a set of trained models, choose the best among them using the validation set.\n\nOnce again, from the dataset perspective:\n\n  * **Training set** : To train all your models.\n  * **Test set** : To evaluate your models during the training phase, while you're still adjusting different hyperparameters.\n  * **Validation set** : To measure the ultimate accuracy. This one should be kept separate from the other two until the final choice between the set of models has been made.\n\nThe last one is important because you can overfit to both training and test sets by adjusting hyperparameters multiple times.\n\nWe do not recommend using ensembles of models on mobile devices as they usually take a lot of resources. Before you decide to go with one, check whether the same performance can or can't be reached if you put the same amount of effort in data collection, cleansing, and feature engineering.\n\nWork iteratively; try one set of algorithms and features, then another. Keep records about the results of each iteration. Set seeds for random number generators in order to be able to reproduce your own results later.\n\nAll your business problem formulation ultimately converges to one question, \"What loss function you are optimizing?\" It is important to remember that learning is the process of adjusting the model to the data in a way that minimizes the loss function. So, if the loss function was chosen carelessly, the results can be far from what your real goals are.\n\n# Field testing\n\nThis is an important stage because it reveals the biases you have in your training data, the user's perception of your product, and other potential pain points. Try to check the model in the most realistic scenarios and conditions. Suppose you are developing a voice assistant. How will it work:\n\n  * On a noisy street when the wind is blowing?\n  * When a child is crying in the background?\n  * When some music is playing?\n  * When the user is not a native speaker or is getting emotional or drunk? Well, those may be the users who need your assistance the most!\n  * In all of these cases together?\n\nIf your solution is security-related, how good is it in the event of an active attack by an adversary? How easy is it to unlock your touch ID with the help of an orange, or cheat your face recognition by presenting it with a photo?\n\nHaving all those observations, you then go back and update your dataset and models accordingly.\n\n# Porting or deployment for a mobile platform\n\nThe next logical step is deploying your solution on a mobile platform (or platforms). Here, you have several considerations:\n\n  * Model memory consumption\n  * Data memory consumption\n  * Training speed (if you need on-device training)\n  * Inference speed\n  * Disk space consumption\n  * Battery consumption\n\nYou can profile all of this using Xcode instruments.\n\nFor information on Swift code speed optimization check out this guide: _Writing High-Performance Swift Code_ , at: <https:\/\/github.com\/apple\/swift\/blob\/master\/docs\/OptimizationTips.rst>.\n\nIf your application includes several pre-trained models, for example, neural artistic style filters, you can use on-demand resources to store those models on the App Store and download them only when they are needed, not in the process of app installation. The On-Demand Resources Guide explains:\n\n\"On-demand resources are app contents that are hosted on the App Store and are separate from the related app bundle that you download. They enable smaller app bundles, faster downloads, and richer app content. The app requests sets of on-demand resources, and the operating system manages downloading and storage. ...  \nThe resources can be of any type supported by bundles except for executable code.\"\n\nAs of spring 2017, App Store allows you to store up to 20 GB of on-demand resources. You also can define which resources will be purged when the OS hits the limit of disk space.\n\nYou can find more details about this technology and how to adopt it in your application here: <https:\/\/developer.apple.com\/library\/content\/documentation\/FileManagement\/Conceptual\/On_Demand_Resources_Guide\/index.html>.\n\nIn the previous two chapters, we discussed questions of model acceleration and compression in more detail.\n\nIt is good to make sure in advance that your model is easily portable for mobile platforms. For example, suppose you've decided to train a model with one of the frameworks and convert it to a Core ML format for iOS deployment. Before training a complex neural network for a week on a GPU server, verify that untrained network with this architecture can be converted by `coremltools`. In this way, you will avoid disappointment later when you figure out that `coremltools` doesn't support one of the layers in your super-cool architecture. Actually, Core ML now supports custom layers, but do you really want to write one if you can replace it with something more traditional? You can call your solution portable only if porting costs much less than rewriting from scratch.\n\n# Production\n\nSome machine learning models require regular updates due to the changing nature of their environment; others do not. For example, language changes faster than human appearance, but fashions change even faster. In fraud detection systems, a constant arms race between defenders and attackers goes on, and both sides try to be creative. The problem of a changing environment is known as concept drift. The wrong word problem of the model getting irrelevant over time is known as model decay.\n\nHow can you tackle these problems? There are several possible ways:\n\n  * Periodically retrain your model\n  * Use online learning algorithms to incorporate new data and drop the old one: an example algorithm is KNN\n  * Use algorithms that allow you to weigh the importance of your data and assign highest importance to recent data\n\n# Best practices\n\nIn this section, we've collected some general ideas worth keeping in mind during the whole development process.\n\nIt's impossible to collect all important thoughts in one place, so here is a list of some really insightful guides from seasoned machine learning engineers on the best practices they recommend:\n\n  * _A Few Useful Things to Know about Machine Learning_ by Pedro Domingos, at: https:\/\/homes.cs.washington.edu\/~pedrod\/papers\/cacm12.pdf\n  * _Best Practices for Applying Deep Learning to Novel Applications_ by Leslie N. Smith, at: https:\/\/arxiv.org\/abs\/1704.01568\n  * _Rules of Machine Learning: Best Practices for ML Engineering_ by Martin Zinkevich, at: http:\/\/martin.zinkevich.org\/rules_of_ml\/rules_of_ml.pdf\n  * _Best Practices for Machine Learning Applications_ by Brett Wujek, Patrick Hall, and Funda G\u00fcne\u0219, at: https:\/\/support.sas.com\/resources\/papers\/proceedings16\/SAS2360-2016.pdf\n\n# Benchmarking\n\nWhen you are creating a model for solving a popular machine learning task, how do you know it is any better than anything else that has been invented by your predecessors? The answer is one word: benchmarks.\n\nThere are some well-known datasets that serve to compare accuracy across different models. For instance, for the task of large-scale visual object classification, a benchmark is the ImageNet dataset.\n\n# Privacy and differential privacy\n\nSurprisingly, in the last few years, most scientific papers where mobile devices and machine learning were mentioned together were not about computer vision or natural language processing. The topics discussed the most were information security and privacy. These two fields intersect in several scenarios:\n\n  * The attacker employs offensive machine learning as a part of his\/her toolkit. It can be used for discovery and analysis of vulnerabilities or for the attack itself. Examples are face or voice recognition for surveillance and finding data leaks in an improperly anonymized data.\n  * Defensive machine learning is used to protect against cyber attacks. It can be utilized for both threat detection and analysis. An example is fraud detection algorithms in banks and antivirus software.\n  * Adversarial machine learning is a setting when algorithm itself is under attack. Examples are search engine optimization (SEO) \u2013 tricking search engines and conversion rate optimization (CRO) \u2013 tricking spam filters.\n\nNow, if machine learning is used to maximize the spam emails open rate, it is clearly an adversarial setting; but both the offender and defender are armed with machine learning, so all three scenarios meet in one place.\n\nIn the context of mobile security, machine learning has been used for:\n\n  * User authentication based on different features: voice, face, gait, signature, and so on\n  * Side-channel attacks: speech recognition, key logging, and stealing passwords using only motion sensor data\n  * Manipulating with voice assistants using noises unintelligible for humans\n  * Tricking image classification algorithms into mislabeling one object as another\n  * Extracting all kinds of personal information from a user's photo library: documents, bar codes, NSFW photos, credit card info, and so on\n\nThe last example is especially troubling because in iOS, any app that has access to the photo library has an access to all of the user's photos, including those in the hidden folder. They can analyze it in any way without limitations. All this leads to the conclusion that at the moment, offensive machine learning on mobile devices prevails over defensive learning and it is restricted only by the attacker's imagination and battery consumption.\n\nOutside of the mobile development domain, machine learning is routinely used for surveillance, obtrusive targeted advertising, mining social media for personal information, and other ethically questionable practices. This is a problem that doesn't have technical solution. Like almost any other powerful tool, machine learning comes with responsibility. The computer can only optimize the objective function; the human is the one who chooses the function to optimize. Are you optimizing the revenue and number of items sold, or the quality of your products and well-being of your users?\n\nAt WWDC 2016, Apple officials brought up the topic of differential privacy in the context of machine learning. According to them, differential privacy is a major research topic and Apple is in the process of introducing differential privacy throughout the company's services. The idea here is to collect users' data but to add noise to it and aggregate it in such a way that information about any individual cannot be extracted.\n\nFor more information on Apple's approach, check out the differential privacy overview document at: <https:\/\/images.apple.com\/privacy\/docs\/Differential_Privacy_Overview.pdf> and the WWDC presentation at: <http:\/\/devstreaming.apple.com\/videos\/wwdc\/2016\/709tvxadw201avg5v7n\/709\/709_engineering_privacy_for_your_users.pdf>.\n\nAccording to Apple, iOS has a 200 MB dynamic cache of personal information to train models right on the iPhone. That personal information includes app usage data, interactions with other people, and keyboard and speech input; it never leaves the device. Because the data does not have to travel over the network in this case, this is a good example of how mobile machine learning can decrease the potential cyber attack surface area and improve a user's security.\n\nResearchers from Google also proposed a secure data aggregation protocol for machine learning. This was needed to implement the decentralized learning system\u2014small local models are being trained on mobile devices and then they send an update to the big central model, which aggregates the experience of all the small models.\n\nThis approach is known as federated learning. To learn more about it, check out the paper _Communication-Efficient Learning of Deep Networks from Decentralized Data_. H. Brendan McMahan et al, at: <https:\/\/arxiv.org\/abs\/1602.05629>.  \nAlso visit the Google research blog: <https:\/\/research.googleblog.com\/2017\/04\/federated-learning-collaborative.html>.\n\n# Debugging and visualization\n\nWhen our usual code has a bug, it either doesn't work or works in the wrong way. When ML code has a bug, it often continues working but just degrades in quality. Because machine learning algorithms can be extremely complex, good debugging and visualization tools are of extreme value. For TensorFlow for example, such a tool is TensorBoard, which allows exploring model graphs, weight distributions, loss charts, and so on.\n\nFor now, humanity has not invented a better way to understand data than to visualize it. Often, 10 minutes of writing code for visualization lead to more insights than hours of debugging on a console. As Prof. Ben Shneiderman from the University of Maryland once noted in his talk:\n\n\"Statistics without visualization should be illegal.\"\n\n# Documentation\n\nSurely, it's better when your tool is so simple; it doesn't require a manual. And we all know the deeply rooted tradition of self-documenting code in the Objective-C community. But in the machine learning domain, code without documentation is often useless. Even when there is a documentation, results often cannot be easily reproduced because some exact values of hyperparameters or other seemingly small details are not known.\n\nSo, what exactly should be documented in your machine learning-related code? Most importantly:\n\n  * Data sources\n  * Preprocessing steps\n  * Combinations of features\n  * Model hyperparameters\n  * All tricks of the trade\n  * Error messages\n  * Loss functions\n  * Experiments\n  * Model checkpoints\n  * Random number seeds\n  * Quality metrics\n\nDo not forget to put references to original research papers wherever appropriate. Try to avoid calling variables _a_ , _b_ , _c_ , _x_ , _y_ , _z_ , _w_ , \u03b1, \u03b2, \u03c1, \u03b8 and so on in your code if those names come only from some formula that is directly referenced in the comments nearby.\n\n# Machine learning gremlins\n\nBen Hamner, a data scientist at Kaggle, referred to common machine learning gotchas as ML gremlins.\n\nYou can watch Ben's original talk at: https:\/\/www.youtube.com\/watch?v=tleeC-KlsKA.\n\nI like the metaphor because it makes my brain think about evil characters rather than some vague, abstract concepts. In addition to the original gremlins presented by Ben, I want to add several of my own and also present a taxonomy of gremlins (see the following diagram). I employed this metaphor throughout this chapter to avoid boring issues and problems when discussing how to identify and neutralize those pests:\n\nFigure 13.3: The simplified taxonomy of machine learning problems\n\n# Data kobolds\n\nDealing with data is hard; that's why we call it data science and data mining! Many different things can go wrong at different stages. Ben mentions data insufficiency, data leakage, non-stationary distributions, poor data sampling and splitting, data quality, and poorly anonymized data. Let's add a few more.\n\n# Tough data\n\nYour data can be tough in a lot of ways: it can be sparse (in features or in target variable), it can contain outliers or missed values, or it can be high-dimensional or high-cardinal (for categorical features). Numerical features can be (and usually are) of different magnitude or suffer from multicollinearity. There is no bulletproof solution. Use force. Tidy your data up. The common techniques here are dimensionality reduction, missing values imputation, outlier detection, and statistical data normalization. Textbooks on statistics and data science will help you learn more on this topic.\n\n# Biased data\n\nThe Word2Vec algorithm (discussed in, Chapter 10, _Natural Language Processing_ ) is a good example of how easily cultural stereotypes and prejudices leak into machine learning models. For instance, vectors trained on the Google news corpus tell us that:\n\n_USA - Pizza + Russia = Vodka_\n\nWhile this may sound very funny for some people, this sounds equally offensive for many more. Is the algorithm biased? No, it is all in the dataset.\n\nAnother example of badly biased data was a web service based on a neural network that assessed a face's beauty by the photo. Apparently, all of the training data contained white faces, so the model was giving the lowest scores to all non-white faces. I truly believe that the developers had no bad intentions in training their model. They just did not pay enough attention to the variety of input data.\n\n# Batch effects\n\nUsually, if you have to label a big dataset manually, you split it into manageable batches. Several people can then work in parallel on different portions. The problem here is that each of those people will introduce a different amount of variability in his\/her batch. This is especially the case when subjective opinions are involved, such as \"Is this movie review slightly positive or rather neutral?\"\n\nBatch effect is also a common problem for datasets that were compiled from several different sources. In many cases, batch effects become apparent when you plot the data obtained from different sources separately.\n\n# Goblins of training\n\nIn addition to overfitting, in this category fall problems with resource consumption, model interpretability, hyperparameter tuning, and so on. Most of them we have already discussed elsewhere in this chapter and other chapters.\n\n# Product design ogres\n\nIn his talk, Ben mentions only one of this kind: solving the wrong business problem. But there are so many more!\n\n# Magical thinking\n\nI want to tell a story to illustrate the point. A friend of mine asked me to build a machine learning system for his startup because he believed it would solve some problem in his mobile app. I asked what data he has and he answered that they are planning to collect a lot of data from their users. They wanted to make highly personalized predictions for each of their users and right in time (precision, within minutes). \"Okay,\" I said, \"Imagine you have this kind of data about yourself, your wife, and your dog. Will it be useful to make correct predictions about me?\" \"No,\" he shook his head. \"Now imagine you've just started collecting information about me. How much time would it take to start making reasonable predictions?\" He looked disappointed. \"So is it just a statistic? I thought it would figure out somehow on its own.\" Fortunately or not, machine learning has nothing supernatural in it. It will not create a solution for you in a miraculous way out of nothing. What it can do for you is to get more value from less data. These basic facts are sometimes not obvious to non-technical people.\n\n# Cargo cult\n\nSomehow, it has happened that we live in a culture where technologies are a matter of fashion and objects of almost religious worship (think about tech evangelism, \"changing the world,\" and tabs versus space wars). AI is on the peak of its popularity now. We often say, \"Everybody does machine learning, so let us also build a neural network into our product and advertise it as artificial intelligence!\" Undoubtedly, machine learning is an excellent hammer, but not all things around are nails. As you probably know, any product gets better if you add Bluetooth to it. However, this rule does not hold true for machine learning. The author of this book believes that there were too many great services that became inconvenient and unpredictable when machine learning was added to them. Rephrasing the famous quote by Jamie Zawinski about regular expressions:\n\n\"Some people, when confronted with a problem, think \"I know, I'll use AI\". Now they have two problems.\"\n\n# Feedback loops\n\nAt a conference, a speaker talked about a new product that his company was developing. Airline sites, explained the speaker, change their ticket prices in a hardly predictable way depending on various indicators and using models known only to them. So the speaker and his colleagues collected the data on price trends from some airline sites and built a regression model. This would predict changes in ticket prices and provide users with advice on whether to buy a ticket for a flight at that time or wait to save money. One of the listeners (it wasn't me) raised a hand and asked: \"What will happen when the airlines learn about your site and update their models to take into account your forecast?\" The question took the speaker by surprise because this scenario was completely unforeseen by him. Setting aside the question of whether the airlines would actually take into account such a site, this is a good example of what is known in machine learning as a feedback loop. When your model's prediction affects the actual outcome, this can lead to one of two unwanted scenarios: self-fulfilling prophecy or self-negating prophecy.\n\nA simple example: your system predicts what news will be of interest to the user. The user reads them, and the system remembers that the user is interested in such information. In fact, the user opened it not so much because he is engaged but because you showed it to him (self-fulfilling prophecy). So he had no other choice except to read what was presented or to close an app. As a result, after a few cycles, the recommendations become monotonous and so dull that the user stops using your application. The problem here is that the training data gets polluted by the model's predictions and the model gradually degrades.\n\nHow to deal with feedback loops? There's no way. Just do not create them.\n\n# Uncanny valley effect\n\nThe term uncanny valley initially appeared in the context of robotics and described the feelings people experience from interacting with humanoid robots. Starting from 1970, Japanese and Korean companies have been producing androids, copying an appearance of a person up to the slightest detail. The androids usually were the copies of some visually attractive models. However, it was observed that such robots seem to cause rejection because they induce associations with corpses or mentally-impaired people. At the same time, robots that did not try to imitate a person's appearance evoked sympathy from observers. Later the concept was extended to the area of \u200b\u200b3D animation and video games, where uncanny valley was successfully employed to create scary characters:\n\nFigure 13.4: Uncanny valley effect. Picture by Mykola Sosnovshchenko.\n\nSome authors apply the concept of uncanny valley in the context of AI systems, such as recommender systems and voice assistants. Systems simulating human behavior that are not believable enough can induce emotional rejection by users. Why? Let's try to figure it out.\n\nInteractions between people are based on the ability to understand and predict each other's behavior. There are even specialized neurons in the brain (mirror neurons) that are responsible for this. You are greeting a person and hear the greetings in response, or you are telling a joke expecting that the listener will smile. If your companion does not respond to your greetings or reacts strangely to your jokes, you feel that something is wrong. Machine learning systems often behave insufficiently in this way. They are not predictable enough for human observers, which can cause a feeling of wrongness. For example, a news feed that is sorted by date or topic is similar to a room in which you know precisely where things are located. But if your news feed is sorted according to an unknown AI algorithm, it becomes similar to a mirage in the desert. There was a piece of news you were interested in, but now it has disappeared somewhere and you cannot find it, however hard you try.\n\nPredictability is the basis of good user experience. There can be predictable randomness and then users are aware that something happens by pure chance. But even then, our brain is trying to find some patterns in those random events. If you call it AI, the creepy concurrences become even creepier: \"Facebook AI algorithms recommend me the community \"Books on shamanism,\" because I am in the AI community.\" My favorite example of uncanny NLP in action is a blog named _Weird Duolingo Phrases_ that collects weird things that the app asks its user to translate.\n\nAlong the same lines, make sure that your personalized models do not produce a spooky user experience. If your app knows too much about the user, that may be a good reason to uninstall it.\n\n# Recommended learning resources\n\nIn this book, we've only scratched the surface of the immense body of knowledge behind the term machine learning. If you want to learn more, we highly recommend the following resources.\n\nThe main criteria for choosing the courses and books were clarity of presentation and a CS-oriented approach. Other criteria for the books were free online availability and open source code samples. All courses mentioned in this list are free (as of May 2017) and of introductory level.\n\n# Mathematical background\n\nThe handwritten comic-style lectures on Calculus by Robert Ghrist from the University of Pennsylvania can be found on YouTube or Coursera. This teaches single-variable calculus: Taylor series, Newton method. This should be your choice if you don't know how to take a derivative of a sigmoid function or which functions are differentiable. For more information refer to: https:\/\/www.math.upenn.edu\/~ghrist\/.\n\n_Coding The Matrix: Linear Algebra Through Computer Science Applications_ course and book by Philip N. Klein. Teaches linear algebra via Python examples and assignments: eigenvectors, eigenvalues, SVD, convolution, wavelet, and Fourier transform. For more information refer to: <http:\/\/codingthematrix.com\/>.\n\n_Immersive Linear Algebra_ by J. Str\u00f6m, K. \u00c5str\u00f6m, and T. Akenine-M\u00f6ller is an interactive online textbook found at: <http:\/\/immersivemath.com\/ila\/index.html>.\n\nOpen source textbooks, video lectures, and exercises on probability and statistics from OpenIntro. Also available as a Coursera course by Mine \u00c7etinkaya-Rundel. Probability, Bayesian statistics, probability distributions, conditional probability, inference, confidence level, chi-square, ANOVA, regression, coding assignments in R. For more information refer to: <https:\/\/www.openintro.org\/stat\/>.\n\n# Machine learning\n\n_The Analytics Edge_ course from MIT at edX. Teaches applied data analysis in the R programming language, including classification, clustering, and data visualization through a set of real-world cases. Model quality evaluation, sentiment analysis. For more information refer to: <https:\/\/www.edx.org\/course\/analytics-edge-mitx-15-071x-3>.\n\nNeural networks class\u2014Universit\u00e9 de Sherbrooke by Hugo Larochelle. Everything you want and don't want to know about neural networks. For more information refer to: <http:\/\/info.usherbrooke.ca\/hlarochelle\/neural_networks\/content.html>.\n\n_Deep Learning_ by Ian Goodfellow, Yoshua Bengio and Aaron Courville: This is a deep learning book. Available online for free at the book's site at <http:\/\/www.deeplearningbook.org\/>.\n\n_Programming Collective Intelligence_ by Toby Segaran. Code samples: <https:\/\/github.com\/ferronrsmith\/programming-collective-intelligence-code>.\n\n# Computer vision\n\nCS 6476: _Introduction to Computer Vision_ by Georgia Institute of Technology. Mathematically-light intro to computer vision with the coding assignment in MATLAB\/Octave. For more information refer to: <https:\/\/www.udacity.com\/course\/introduction-to-computer-vision--ud810>.\n\nCAP 5415: _Computer Vision_ course by University of Central Florida. For more information refer to: ><http:\/\/crcv.ucf.edu\/courses\/CAP5415\/Fall2014\/index.php>.\n\nA classic textbook by Richard Szeliski, _Computer Vision: Algorithms and Application_ , is freely available online: <http:\/\/szeliski.org\/Book\/drafts\/SzeliskiBook_20100903_draft.pdf>.\n\n_CS231n: Convolutional Neural Networks for Visual Recognition_ course from Stanford University. Introductory course on convolutional neural networks with coding assignments in Python. For more information refer to: <http:\/\/cs231n.stanford.edu\/index.html>.\n\n# NLP\n\nCS224n: _Natural Language Processing with Deep Learning_. Coding assignments in TensorFlow. Word vector representations, LSTM, GRU, neural machine translation. For more information refer to: <http:\/\/web.stanford.edu\/class\/cs224n\/>.\n\n# Summary\n\nThis was the final chapter of the book; so we discussed a machine learning app's life cycle, and common problems in AI projects and how to solve them. We also provided a list of good study material for further progress of our readers. We hope that you were not disappointed and wish you many successes in your own AI experiments!\n","meta":{"redpajama_set_name":"RedPajamaBook"}}
+{"text":" \nLiterature, Life, and Modernity\n\nCOLUMBIA THEMES IN PHILOSOPHY, SOCIAL CRITICISM, AND THE ARTS\nCOLUMBIA THEMES IN PHILOSOPHY, SOCIAL CRITICISM, AND THE ARTS\n\nLydia Goehr and Gregg M. Horowitz, Editors\n\nADVISORY BOARD\n\nJ. M. Bernstein | Eileen Gillooly  \n---|---  \nNo\u00ebl Carroll | Thomas S. Grey  \nT. J. Clark | Miriam Bratu Hansen  \nArthur C. Danto | Robert Hullot-Kentor  \nMartin Donougho | Michael Kelly  \nDavid Frisby | Richard Leppert  \nBoris Gasparov | Janet Wolff\n\nColumbia Themes in Philosophy, Social Criticism, and the Arts presents monographs, essay collections, and short books on philosophy and aesthetic theory. It aims to publish books that show the ability of the arts to stimulate critical reflection on modern and contemporary social, political, and cultural life. Art is not now, if it ever was, a realm of human activity independent of the complex realities of social organization and change, political authority and antagonism, cultural domination and resistance. The possibilities of critical thought embedded in the arts are most fruitfully expressed when addressed to readers across the various fields of social and humanistic inquiry. The idea of philosophy in the series' title ought to be understood, therefore, to embrace forms of discussion that begin where mere academic expertise exhausts itself, where the rules of social, political, and cultural practice are both affirmed and challenged, and where new thinking takes place. The series does not privilege any particular art, nor does it ask for the arts to be mutually isolated. The series encourages writing from the many fields of thoughtful and critical inquiry.\n\nLYDIA GOEHR AND DANIEL HERWITZ, eds.,\n\n_The Don Giovanni Moment: Essays on the Legacy of an Opera_\n\nROBERT HULLOT-KENTOR,\n\n_Things Beyond Resemblance: Collected Essays on Theodor W. Adorno_\n\nGIANNI VATTIMO,\n\n_Art's Claim to Truth_ , edited by Santiago Zabala, translated by Luca D'Isanto\n\nJOHN T. HAMILTON,\n\n_Music, Madness, and the Unworking of Language_\n\nSTEFAN JONSSON,\n\n_A Brief History of the Masses: Three Revolutions_\nLITERATURE, LIFE, AND MODERNITY\n\nRichard Eldridge\n\nCOLUMBIA UNIVERSITY PRESS   NEW YORK\nColumbia university Press\n\n_Publishers Since 1893_\n\nNew York Chichester, West Sussex\n\ncup.columbia.edu\n\nCopyright \u00a9 2008 Columbia University Press\n\nAll rights reserved\n\nE-ISBN 978-0-231-51552-8\n\nLibrary of Congress Cataloging-in-Publication Data\n\nEldridge, Richard Thomas, 1953\u2013\n\nLiterature, life, and modernity \/ Richard Eldridge.\n\np. cm.\u2014(Columbia themes in philosophy, social criticism, and the arts)\n\nIncludes bibliographical references and index.\n\nISBN 978-0-231-14454-4 (cloth: acid-free paper)\u2014\n\nISBN 978-0-231-51552-8 (e-book)\n\n1. Literature\u2014Philosophy. 2. European literature\u2014History and criticism.\n\n3. Literature and society. I. Title. II. Series.\n\nPN49.E43 2008\n\n801.3\u2014dc22 2008001172\n\nA Columbia University Press E-book.\n\nCUP would be pleased to hear about your reading experience with this e-book at cup-ebook@columbia.edu.\nNot for these I raise\n\nThe song of thanks and praise;\n\nBut for those obstinate questionings\n\nOf sense and outward things,\n\nFallings from us, vanishings;\n\nBlank misgivings of a Creature\n\nMoving about in worlds not realized,\n\nHigh instincts before which our mortal Nature\n\nDid tremble like a guilty Thing surprised:\n\n\u2014WILLIAM WORDSWORTH, \"Ode: Intimations of Immortality\n\nfrom Recollections of Early Childhood\" (1807)\n\nAh! As I listened with a heart forlorn,\n\nThe pulses of my being beat anew:\n\nAnd even as Life returns upon the drowned,\n\nLife's joy rekindling roused a throng of pains\u2014\n\n\u2014SAMUEL TAYLOR COLERIDGE, \"To William Wordsworth:\n\nComposed on the Night After His Recitation of a Poem\n\non the Growth of an Individual Mind\" (1807)\n\nThe poem of the mind in the act of finding\n\nWhat will suffice. It has not always had\n\nTo find: the scene was set; it repeated what\n\nWas in the script.\n\nThen the theatre was changed\n\nTo something else. Its past was a souvenir.\n\nIt has to be living, to learn the speech of the place.\n\nIt has to face the men of the time and to meet\n\nThe women of the time. It has to think about war\n\nAnd it has to find what will suffice.\n\n\u2014WALLACE STEVENS, \"Of Modern Poetry\" (1940)\nContents\n\nACKNOWLEDGMENTS\n\n1. Introduction: Subjectivity, Modernity, and the Uses of Literature\n\n2. Romanticism, Cartesianism, Humeanism, Byronism: Stoppard's _Arcadia_\n\n3. Romantic Subjectivity in Goethe and Wittgenstein\n\n4. Attention, Expressive Power, and Interest in Life: Wordsworth's \"Tintern Abbey\"\n\n5. The Ends of Literary Narrative: Rilke's \"Archaic Torso of Apollo\"\n\n6. \"New Centers of Reflection Are Continually Forming\": Benjamin, Sebald, and Modern Human Life in Time\n\nAppendix: William Wordsworth: \"Lines Composed a Few Miles Above Tintern Abbey\"\n\nNOTES\n\nINDEX\nAcknowledgments\n\nThis book has been fortunate in the many circumstances that occasioned it and in the comments and conversations that contributed to its development. I am grateful to many friends, readers, and interlocutors for their good will, insights, and friendship.\n\nIn conversations beginning around their editing of _The Literary Wittgenstein_ , Wolfgang Huemer and John Gibson prompted me to begin thinking and writing more directly about the nature of distinctively literary achievement than I had done in the past. On one or another occasion, either in Philadelphia or in Erfurt, Germany, I talked with John, Wolfgang, or both about almost everything written in the analytic philosophy of literature in the last ten or so years. These conversations were invaluable for the direction of my work. Wolfgang then invited me to spend a semester at the University of Erfurt, where I taught a seminar on the philosophy of literature with Wolfgang present. Wolfgang visited Swarthmore nine months later, where we taught the philosophy of literature together for six weeks and I had the pleasure of hearing his lectures. In both settings, we talked several times a week about recent professional philosophical work on cognition, morality, and the uses of literature, and reflections of these conversations run continuously through this book. I also found in Erfurt extraordinarily congenial colleagues in Alex Burri, Carsten Held, Christian Beyer, Winfried Franzen, and Jan-Hendrik Heinrichs, in addition to Wolfgang Huemer. I cannot imagine a happier and more supportive environment in which to do philosophical work. In addition, Wolfgang Huemer and Alex Burri organized a conference on literature and cognition in Erfurt during my visit. I was able there to present an early version of part of chapter 4 and to enjoy the presentations of and discussions with Bernard Harrison, John Gibson, Wolfgang Huemer, Luca Pocci, Catherine Elgin, Christiane Schildknecht, Alex Burri, Gottfried Gabriel, Peter Lamarque, and Joachim Schulte.\n\nThanks to Bettina Menke, I was able to present an early version of chapter 6 to the Faculty for General and Comparative Literature. Comments from Bettina Menke and Holt Meyer about the singularity of the exemplary work of literature and the nature of the literary experience have been much on my mind in subsequent writing and rewriting, and Bettina Menke helpfully corrected some points about Sebald. Reading Christoph Menke's account in _Die Souver\u00e4nit\u00e4t der Kunst_ of Kafka's literary achievement together with a subsequent conversation with him in Erfurt about this account also substantially shaped my thinking.\n\nA yet earlier version of chapter 6 was presented at a seminar on Philosophy and Literature, sponsored by the Philosophical Society of Finland and organized by Martha Nussbaum. There I was also fortunate to be able to talk about subject development with Jonathan Lear and about romanticism and modernity with Josef Fr\u00fcchtl. This earlier version subsequently appeared in _Visions of Value and Truth: Understanding Philosophy and Literature_ , ed. Floora Ruokonen and Laura Werner (Helsinki: The Philosophical Society of Finland, 2006), 13\u201329.\n\nA somewhat later version was then presented as a lecture to the philosophy department of Purchase College of the State University of New York, where during both a wonderful general discussion and dinner I learned from and was encouraged by Casey Haskins, Morris Kaplan, Jennie Uleman, and Frank Ferrell.\n\nNikolas Kompridis invited chapter 3 for his volume _Philosophical Romanticism_ (London: Routledge, 2006), 97\u2013112, and I was able to present an early draft of this chapter at a meeting of the International Conference on Romanticism, where I profited from conversations with Larry Peer, Diane Hoeveler, William Davis, Eugene Stelzig, and Joshua Wilner.\n\nFred Rush invited me to deliver a lecture on \"Romanticism and Tom Stoppard\" in a symposium for the inauguration of the DeBartolo Performing Arts Center at Notre Dame. This lecture subsequently became chapter 2, improved by the comments at Notre Dame of Fred Rush, Charles Larmore, and Neil Delaney. A slightly later version of this chapter was delivered as a lecture to the German Society for Aesthetics, to which I was invited by Josef Fr\u00fcchtl and where I received useful comments from Georg Bertram, David Lauer, and Ian Kaplow, in addition to Josef Fr\u00fcchtl. A portion of this chapter will be forthcoming in _Zeitschrift f\u00fcr Aesthetik und Allgemeine Kunstwissenschaft_ , vol. 1, 2008.\n\nRoss Wilson invited what became chapter 4 for a forthcoming volume on romantic conceptions of life. He provided acute comments on an early draft.\n\nA version of chapter 5 was presented as a Faculty Lecture at Swarthmore College, where I am grateful to Alfred Bloom, Peter Schmidt, and Robin Wagner-Pacifici for encouraging and useful remarks. Philip M. Weinstein subsequently read this chapter and provided written comments that not only improved it but also substantially prompted and contributed to some points in the introduction. A version of it has been published in _A Sense of the World_ , ed. John Gibson, Wolfgang Huemer, and Luca Pocci (London: Routledge, 2007), and in German in _Kunst denken_ , ed. Wolfgang Huemer and Alex Burri (Paderborn: Mentis, 2007).\n\nDuring the last year or so of the writing and revising of this manuscript, I was fortunate to have been editing _The Oxford Handbook of Philosophy and Literature_. Essays from contributors would arrive every three weeks or so for my comments. As a result, I was continually thinking about the arguments put forward by these contributors and comparing them with my own lines of thinking. Without exception, I was stimulated and encouraged by essay after essay. I am aware of specific turns of thought on my part being prompted by essays by Charles Altieri, J. M. Bernstein, Simona Bertacco, Anthony J. Cascardi, Ted Cohen, John Gibson, Bernard Harrison, Toril Moi, Kirk Pillow, Fred Rush, Susan Stewart, and Philip Weinstein.\n\nHannah Eldridge and Sarah Eldridge both attended the Erfurt conference on literature and cognition, and they have each talked regularly with me about Sebald, Rilke, Benjamin, Hegel, and romanticism, among many other topics. Each of them read drafts of sections of various chapters, and their responses were important to increasing my confidence in the coherence and direction of my argument.\n\nAdam Haslett read a demipenultimate and then a penultimate version of the introduction. His acute comments prompted a significant structural revision and helped me to tighten the development of the argument and to make it more accessible to more readers than it might otherwise have been.\n\nLydia Goehr and Gregg Horowitz each read a complete, penultimate draft of the book. They offered detailed and insightful comments, criticisms, and suggestions, all of which prompted significant revisions and additions, particularly to the introduction and to chapter 6, but also throughout the manuscript as a whole. Acute comments from readers for Columbia University Press led to further productive final revisions.\n\nJoan Vandegrift likewise read a complete, penultimate draft, and her comments led to some significant rearrangement of materials from chapter 5 to the introduction. In this, as in everything else, her ear and sense of development proved vital to the whole.\n1.\n\nIntroduction\n\n_Subjectivity, Modernity, and the Uses of Literature_\n\nThe term \"literature\" has a fairly wide range of reasonable uses. One can talk of the literature on _x_ \u2014ladybugs or chess or cello varnishes, as may be\u2014and mean only all or much of what has been written about a particular subject. In German-language scholarship, one often begins an essay with a _Literaturverzeichnis_ , a review of the most important prior work on a topic, whatever the topic might be. _Children's literature_ refers to books specially written for children to enjoy. These uses, however, are surely not what the Nobel or Booker Prize committees have in mind in awarding prizes for literature, nor are they what is suggested by the commendatory adjective \"literary,\" as in \"a literary person\" or \"of literary merit\"; nor do they figure in the senses of _English literature_ or _Francophone literature_ as names for disciplines of study within a university curriculum.\n\nEven where what is in view under the heading of the literary is some sort of value, there are nonetheless many different valuable experiences that literature affords. These include, among others, entertainment, consolation, the pleasures of archaic regression (reminiscent of being read to as a child), the acquisition of historical knowledge, the sharpening of personal or political hope, and absorption in and admiration of verbal virtuosity. It would be a mistake either to overlook or to underrate the considerable variety of uses that works rightly classed as literary can and do invite and support.\n\nDespite this variety of uses, however, there persists for many people a sense that literature, at least sometimes, has a central and distinctive way of mattering for human life. Such a sense supports the existence of special curricula of literary study, including the cultivation of habits of close reading, as these habits have developed with increasing, multiple specificities from the late eighteenth century to the present. Both literary scholars and many ordinary people now read novels, short stories, films, lyrics, plays, television programs, and advertising, among other things, with habits of attention to form, diction, imagery, ideology, materiality, use, and much more. Yet it is not always clear exactly what close reading discerns, nor is it clear why one should bother to read certain works closely, rather than doing the many other things there are to do in life. Somehow\u2014or so many people think\u2014a life would be less rich or less informed by sympathetic understanding without engagement with a specifically literary curriculum or with habits of close reading. But exactly how might this be so, if it is so at all? The undoubted existence of some persons of refined literary experience and sensibility but with little moral discernment or responsiveness shows that reading literature is not by itself sufficient to produce moral understanding. Conversations with actual people\u2014parents, friends, neighbors, and so on\u2014is in most cases far more important than reading in the shaping of character. Could reading some literary works nonetheless help to shape character in valuable ways? Perhaps, but then many literary works are fictions, so that there is no actual person with whom one can immediately sympathize, and some people who read literature intensively may use it mostly as a compensation for the pains of life or as a distraction from them. Might it not be easier and more reliable to learn sympathy by talking at length with wide varieties of people? Is reading literature only a shortcut for that? And if it is, then why not read memoirs, history, and journalism instead? Why all that fiction? And if sympathizing with actual persons is the aim ultimately in view, then what is the significance of the specific verbal densities and formal structures of exemplary literary texts? Why are many of them so difficult, and why, if at all, does that matter?\n\nIf there is any hope of articulating plausible and useful answers to these questions, then the hard question of what a human life as such is all about will have to be faced. If human lives have no common structure and directions of achievement but instead aim at only whatever individuals, in interaction with possibilities afforded by their cultural settings, arbitrarily undertake to pursue, then literature and literary curricula will have no distinctive places in the cultivation of human life as such, for there will be nothing significantly shared to cultivate.\n\nDuring the past two hundred to four hundred years in Europe and the Americas, common projects and senses of purpose were largely sufficiently established by common national language, culture, and material situation to make national literatures matter to many, at least within certain educated circles. However, in an era of increasingly global commodity markets, national and linguistic barriers become more permeable, competitive individualism and reactive fundamentalism increase, and the salience of national literatures as forms of reflection on common cultural life diminishes, as cultural life spreads out and diversifies. Common and overlapping projects give way to various forms of getting, spending, enjoying, and entertaining, supported by technological advancements, unless those should turn out to be self-defeating or subjectively undesirable. As a result, factionalism increasingly displaces any sense of a political commonwealth, and culture becomes often a matter more of multiple fluid and commodified styles than a stable source of significance. Yet there is little chance of simply returning to older sureties. In a modern social world with a highly complex division of labor and with the distinctive satisfactions that attach to different social roles, religious commitments of any kind may seem either pale, abstract, and empty (churches open to all shoppers) or tyrannical and self-consciously sectarian (closing the doors to all but the pure). Either competitive individualism or competitive factionalism comes to the fore, and chances of learning to live out a common humanity with more depth become increasingly attenuated. And yet it can seem in some moments of reading that certain literary works offer us some access to increased reflective depth without dogma or tyranny. How might this be so? Can this sense in any way be trusted? And might literature help to open up some senses of possible common purpose and some routes of possible mutual engagement, hesitantly and nondogmatically, without either denying or undertaking to rule over the complexities of modern social life?\n\nIn his valuable recent study _Why Does Literature Matter?_ , Frank Farrell takes up these questions, arguing that literary works function for human subjects as vehicles of partial and provisional recoveries of meaningfulness. Modern subjects, Farrell argues, suffer various kinds of loss in the courses of their developments. Thick, premodern social rituals are displaced in social space by economic transactions according to mysterious equivalences. The magical and metaphorical languages and ways of thinking that figure significantly in childhood experiences and in premodern cultures are displaced by analytical, grammatical, and scientific casts of mind and thought. Engagements with significances that are widely felt according to \"the way things are done\" are supplanted by individuals going opaquely about their mysterious \"private\" businesses. For these various losses of felt significances, \"the space of writing,\" Farrell argues, \"offers us a modest compensation.\"\n\nThe compensations that literature affords occur, according to Farrell, in various registers. Phenomenologically, experience itself, for example, of a landscape or of the face and bearing of another person, is recorded in and recovered through the literary text as meaningful, rather than being left either as a source of mere \"sensations\" or data or being submitted to automatic, preformed categorizations. Metaphors and other devices of figuration are used to achieve and express psychic investment in and attention to what is being presented. Metaphysically, mood as an overall style or color of engagement with a natural and social world is registered, and some engagements and moods are registered as more truthful than others. Psychologically, a childhood sense of self as being caught up, fuguelike, in mysterious, larger processes of development is posed against a too assured, too rounded sense of accomplished mastery of discourses and social roles. Archaic but not quite lost rituals and senses of place are recovered or refigured. And there are the compensations of \"style...as a staging of the psyche,\" with richer and more satisfying investments than are ready to hand in daily life. Overall, \"we seem to have the language of literature as a necessarily repeated, even obsessive, reworking of that transitional space\" between the prelinguistic and the linguistic, childhood and maturity, the premodern and the modern, the metaphorical and the literal. Its modest compensations for loss challenge punctual, individual hubris and open up routes of richer attention and engagement.\n\nFarrell has here located the function of literature against a compelling background story of how subject development is marked by the loss of various kinds of richness and intensity (experiential, premodern, ritual-archaic, fuguelike repetitive, etc.). His idea that literature works to recuperate these losses has much to recommend it. At the same time, however, it is possible to wonder how stable and assertational about human life the recuperations that literature offers finally are. Farrell himself calls them modest, as though to mark their difference from stable discovery of standing sources of sharable felt significance in life. When the recuperative and instructive powers of literature are emphasized, then both its powers to disrupt and its failures to arrive at conclusive doctrinal closure are underplayed.\n\nDavid Wellbery usefully registers literature's disruptive force and formal distinctiveness as he describes the \"problematic and uncertain representational, or perhaps epistemological status\" of certain poems and, by implication, of exemplary literature in general. Wellbery develops his conception of the literary as a site of formed disruption through commenting on Goethe's lyric \"Maifest\"\u2014a lyric that on a narrative-thematic level describes the achievement of bliss. The concluding couplet of \"Maifest\" represents and summarizes an outburst of bliss that is grounded in an experience of the gaze of the beloved. \"Und doch, welch Gl\u00fcck! Geliebt zu werden. \/ Und lieben, G\u00f6tter, welch ein Gl\u00fcck [And yet, what bliss!, to be loved, \/ And to love, you gods! What bliss!].\" Within the very structure of its formulation, however, this outburst reveals itself as artfully and rhetorically achieved, not simply the spontaneous, na\u00efve, and accessibly inimitable product of immediate passion. As Wellbery puts it, \"the chiastic structure 'Gl\u00fcck...-lieb \/\/ lieb-Gl\u00fcck' and the passive-active reversal of the verb... _constitute a structural emblem for the entire poem_.\" That is, the concluding lines (\"lieb-Gl\u00fcck\") embody an inversion, both thematic-semantic and phonological-formal, of material that has been used earlier in the poem, so that this \"closing formulation proves to be in many respects a recapitulation within a reduced format of the essential features of lines 25\u201330.\" Through this use of semantic and formal figuration, repetition, and condensation, the poem is marked as literary and achieves its end. It achieves aesthetic closure in historically specific ways, and it disrupts simpler communicative assertion of independent facts. It invites absorption in its artifices as much or more than proposing any recommendations for individual or social recuperation. Disruption and absorption in formal achievement significantly displace any moment of theoretical-instrumental instruction, individual or social. On the larger historical-thematic level, moreover, \"Maifest\" figures the gaze of an individual beloved rather than, say, either the presence of God or involvement in ritual as the source of bliss. Thus the poem is marked as a more or less modern work that, in using its theme, carries \"inadvertent traces and remainders of cultural production.\" Once upon a time, that is, things were otherwise: bliss was either figured as having other sources or was not so intensively pursued by subjects who were less inward and more clan-immersed and either epic-heroic or immiserated than the modern, individual speaking persona of \"Maifest.\"\n\nSo it is, always, with exemplary literature. The most successful writers use both thematic materials and devices of figuration that are in some measure historically specific. They use these materials and devices self-consciously to register and attend to a moment of crisis or loss in an individual, within a culture, or between cultures. They manage to represent this crisis fully, avoiding repression and clich\u00e9, and avoiding also resolution according to the terms of any philosophical or religious doctrine of value. Yet they manage to achieve, in and through the interaction of thematic materials with formal devices that mark the work as literary, densities and closures that compel their readers\u2014or those among their readers who share enough of their losses and crises\u2014to become absorbed in them, to follow their self-sustaining work, without taking away any formulable-assertible message about reality outside the work. Hence the terms of the modest compensations that literature offers are simultaneously thematic in relation to specific historical materials and formal-aesthetic-disruptive-autonomous. There is no single path, smooth and bright, for either the achievement of literary value or its transportation into the rest of life. The occasions of crisis and loss that provoke literary attention are too various for that, ungoverned by any superintending historical logic, and the use of figures as devices of attention is likewise both historically marked and bound to specific thematic historical materials. Yet somehow, nonetheless, exemplary writers come to terms in exemplary ways with a kind of permanent human immigrancy or fracturedness, with what Eric Santner has characterized as \"the _signifying stress_ at the core of creaturely life.\" Human beings in their courses of development are able sometimes to give voice to the situations of crisis and loss that mark their lives as subjects of and within culture, capable of awareness of their situations. They can attend to and work through the stresses, both individual and cultural, that mark their lives. But the work they accomplish is less the work of arriving at a doctrine than it is, in Heideggerian terms, the working of the work itself: its having its way of bringing together its thematic materials and figural-rhetorical devices to embody a fullness of attention coupled with a satisfaction in the forming of the work in which its readers may share (or may not).\n\nWithin modernity, the stresses that force themselves into consciousness\u2014stresses to which the work of art then responds\u2014come increasingly from the late eighteenth century on to involve conflict between the claims of the sensible (what we discern and attach ourselves to through embodied feeling) and the intelligible (what we discern and attach ourselves to via distantiation and the controlled measurement of what there is). Claims of intimacy, solidarity, and cathexis to daily routine jostle against claims to knowledge, objectivity, and clear-sightedness about what there \"at bottom\" \"really\" is. Feeling is itself internalized, by being cast as something \"subjective\" with measurable intensities and durations, and its claims to being a mode of responsive knowledge are challenged. Whatever any individual happens to like or dislike becomes a matter only of more or less measurable fact (perhaps as a revealed preference, perhaps something one can report about oneself); what emotion, feeling, and mood discern as worth responding to or being involved with fades in cognitive power. Our work, our intimate relations, and our political citizenship, among other things, become matters, at best, of private satisfactions, troubled by the fact or threat that the private satisfactions of tomorrow may displace them, as either the menus of options or one's own whims change. Stability, depth, and lived meaningfulness founder. As J. M. Bernstein puts it: \"The most profound challenge to the unity and unifying work of [modern] culture is the separation, diremption, gap, or abyss separating the sensible world we aspire to live in every day, the world of things known through sight and sound and touch and feel, from the exactitudes of scientific explanation.\" Unsupported by a sense that they are rooted in any accurate discernment of how things are, \"our moods do not believe in each other,\" and we drift, perhaps seeking medication to dull anxiety and depression.\n\nOne way to begin to address the problems of drift and of the disruption of cathexis is to see the modern work of art as occupying \"a strange place at the intersection of the axes of the actual and eternal,\" as J\u00fcrgen Habermas usefully characterizes Baudelaire's conception of the artwork. According to this conception, the authentic modern work of art \"is radically bound to the moment of its emergence; precisely because it consumes itself in actuality, it can bring the steady flow of trivialities to a standstill, break through normality, and satisfy for a moment the immortal longing for beauty\u2014a moment in which the eternal comes into fleeting contact with the actual.\"\n\nIt is, however, not so easy to say what the eternal's coming into fleeting contact with the actual amounts to. Baudelaire himself speaks of \"eternal and invariable...Beauty\" taking on an \"amusing, teasing, appetite-whetting coating\" from circumstantial actuality. Whatever the \"shining forth\" of the eternal within the coating of the actual may involve, however, it evidently does not involve accession on the part of the audience to any guiding doctrine or articulated sense of where beyond the work meaningfulness is to be found. Aesthetic absorption in the work overwhelms any moment of instruction. Where, as in the novel, more generalized reflections on meaningfulness sometimes appear, writers are continually forced to exercise powers of construction and of the making of meaning against the grain of an actuality that significantly involves the merely happenstantial. Fates experienced as meaningful\u2014certain exemplary marriages or deaths, say\u2014are as much the inventions of modern writers as they are found ready-made in modern life. As long as it avoids clich\u00e9 and sustains attention to life, the modern novel, along with modern art in general, suffers from what Georg Luk\u00e1cs calls a characteristic \"normative incompleteness\": it cannot say what is to be done. In Bernstein's similar perception, \"at its highest reach, [modern] art turns cultural melancholy into form.\" The work invites and sustains absorption in it, in the face of the pains of modern life, and within the work complexities and unresolved resistances come increasingly to displace meaningful closures.\n\nHistorically, modern and modernist literary texts present dramas of heroic individual resistance against decayed or opaque social formations. The forms of resistance may range from Quixote's comic fancies to Hamlet's tragic uncertainties to the compressed intensities of the lyrics of Goethe or Keats, among many others. Trauma and failure of fully stable and meaningful subject formation are registered in tragic losses, comic flights, or asides of lyric ecstasy. Sometimes a good enough resolution is found for a few, against the grain of the prevailing social order, though in chastened awareness of its presence, as in Jane Austen. Good enough resolutions become, perhaps, less available in more characteristically modernist as opposed to modern texts. More \"postmodern\" texts use devices of collage, juxtaposition, and intertextuality (satire and allusion, especially across genres and between popular and \"high\" culture) in order to emphasize the inabilities of cultures or individuals to settle on specific, clear, final narrative arcs. Positions, ideologies, cultures, and points of view collide with one another all but endlessly. It is impossible, however, to distinguish in sharp and absolute terms modern-modernist dramas of individual crisis (partially resolved or not) from postmodernist anarchic collage and juxtaposition. Where, for example, would one place _Tristram Shandy_? Is _Gravity's Rainbow_ not in part a drama of individual crisis? When they achieve exemplarity, literary texts present both dramas of crisis and moments of sheer contingency. Hence in either form\u2014relatively modern-modernist or relatively postmodern\u2014what modern literature knows is that no comprehensive resolution of crises within individual or social development is possible: some satisfaction must be found within the working of the work itself, as a kind of placeholder for what is never finally achieved. Human beings, at least within the orbit of a modern individualism that remains powerfully with us, persist as caught up in signifying stresses arising out of a sense of slippage of \"inner,\" passionate, embodied, archaic selfhood away from \"outer,\" articulated, social role and agency. In modernity, such slippage is inevitable, and the task of literature is more to figure its forms than to propose standing resolutions.\n\nThere are, in all likelihood, deep reasons for this kind of literary practice, deep reasons that suggest that literature's registerings of human finitude and the impossibility of specifically legislative moral knowledge are apt to human life as such. Human consciousness is marked by intentionality. That is, human beings not only represent their environments to themselves through perception, they are also aware of their own representings. They can \"intend\" objects that are not materially present (golden mountains, centaurs, time-travel machines, and the like) and they can imagine themselves perceiving and acting in counterfactual situations with much greater range, depth, and flexibility than can other animals.\n\nHence for human beings questions of correctness in judgment can arise explicitly. \"Am I,\" we are capable of asking, \"correct to judge that this is a stick or a weapon or a digging implement or firewood (or all four)? Or am I rather imagining a context of use that is either not ready to hand or not shared by others? Just what am I doing when I am judging that things are thus-and-so, and am I here and now right or not?\" Other sensate and conscious creatures do not display this kind of plasticity of attention, self-awareness, and engagement with questions of correctness.\n\nAs philosophers as different from one another as Aristotle, Hegel, Dewey, Wittgenstein, and Adorno have argued, this combination of plasticity of attention, self-awareness, and involvement with normativity is not a purely material phenomenon, even though it has a necessary material basis. This thought is further supported by the detailed ethnographic observations of childhood language learning carried out by Michael Tomasello. Instead of being wholly determined by biological-material processes alone, conceptual consciousness begins in training and mimicry, in learning within contexts of joint attention to see _this_ (this stick, this wooden object, this whatever it is) _as this or that_ (as a stick, a weapon, a digging implement, firewood, etc.) The emergence of self-awareness and the emergence of involvement with normativity are coeval with the emergence of conceptual consciousness. (\"Am I right to see this as that? How do others see it? Am I doing what is wanted of me in picking up this stick, this ball, this penny [as the child learns later to call them]?\") Is there a proof from contact with ultimate givens that conceptual consciousness, self-awareness, and involvement with normativity thus emerge? No. But try to explain its emergence either metaphysically or purely materially: all the familiar problems of the ineliminability of normativity and responsibility arise. (Do we live in order to represent or represent in order to live?)\n\nBecause, however, human beings are capable of plastic attention and face widely divergent and ever-changing problem situations, multiple patterns of engagement with objects under concepts are always available, and these patterns of engagement (conceptual repertoires) change and are contested. The dream of rooting perfect and unchallengeable conceptual consciousness, freed of all critical engagement with normativity, in ecstatic, intuitive contact with ultimate givens (Platonic forms, sempiternal atoms) is haunting but idle. Training in contexts of mutual attention that are open to contestation cannot be overleaped. Meaning\u2014what things are for in relation to contexts of use, what courses of action are fulfilling in what ways\u2014is not to be discovered in anything simply given in the absence of circuits of training and imitation. \"There will not be books in the running brooks until the dawn of hydro-semantics\"\u2014and hydro-semantics shows no sign of dawning.\n\nThe idea that we bear a continuing responsibility for and continuing anxieties about our lives as conceptually conscious subjects who are caught up in patterns of attention that are subject to contestation is originally and most powerfully formulated by Kant. Kant's philosophical anthropology is rooted in a sense of human life as having two aspects or dimensions. (Human reason has this peculiar, divided fate.) First, we are beings who possess apperceptive awareness or self-consciousness; that is, we are beings who are at least implicitly and potentially aware of our judgments and actions as our own. We further possess the power to become more explicitly aware of our judgments and actions as our own and to raise questions about their correctness: to submit them to critical reflection in the pursuit of greater reasonableness, fluency, stability of character, and human command. Second, we are finite beings who exist within nature and culture and who are unable to refer that existence to any ultimate grounding. Within both nature and culture, there is the possibility always of surprise, of a discovery of one's own itinerancy, and of being at this moment out of attunement with nature, culture, and oneself.\n\nThe fact that we possess both these senses of ourselves is brought powerfully into awareness by the experience of modernity. That Kant expresses both senses is what makes him, along with Descartes, a modern, even modernist, philosopher. Descartes proposes that \"a good man has no need to have read every book, nor to have carefully learned all that which is taught in the schools; it would even be a defect in his education were he to have devoted too much time to the study of letters.\" Thus he sets his face as a freethinking individual against the authority of culture as it stands, seeking a new form of the purely rational expression of purely individual rational powers in the practice of modern mathematical-experimental science. We must, as Kant will later put it, \"dare to know,\" against the grain of the culturally given.\n\nDescartes' confidence in the availability and value of this new form of practice is underwritten officially by his initial certainty of his own existence, coupled with his subsequent a priori arguments for the existence of God and for God's having made physical nature such that we can know it by doing the right kind of science. But one can also, if one listens closely, hear an undercurrent of anxiety in Descartes' formulations. \"'I am, I exist' is necessarily true each time that I pronounce it, or that I mentally conceive it.\" But what if I fail to do this, fail to pronounce my own existence, perhaps out of timidity in the exercise of my rational powers, or perhaps because I am more caught up than I suppose in the culture that does not embody fully human, rational life, or perhaps because nature in the end will not fully support the exercise of rational powers? Do I then fail to exist necessarily? \"I am, I exist, that is certain. But how often? Just when I think; for it might possibly be the case if I ceased entirely to think, that I should likewise cease altogether to exist.\" Coupled with the thoughts that the mass of humanity has more or less continuously failed to think fully or clearly and that nature is not intuitively or immediately knowable, this form of self-certainty is not exactly a recipe for confidence in life, even if the practices of modern science turn out to be comparatively fruitful and cognitively satisfying. The power to reflect on one's judgments and actions has here established a certain distance from ultimate grounding in either metaphysical givens or mere naturalness, no matter what assurances follow and no matter what the successes of modern science are. And is it clear that we either can or should forego wide-ranging reflectiveness? There is, in Stanley Cavell's terms, a kind of standing \"nextness of the self to the self.\" As Thoreau puts it, \"I am conscious of the presence and criticism of a part of me, which, as it were, is not a part of me, but a spectator, sharing no experience, but taking note of it, and that is no more I than it is you.\" Hence, in Cavell's gloss, the self bears, always, two attitudes toward itself, as it finds \"that it is the watchman or guardian of itself, and hence demands of itself transparence, settling, clearing, consistency; and that it is the workman, whose eye cannot see to the end of its labors, but whose answerability is endless for the constructions in which it houses itself.\" (If one can hear an inflection of class in the distinction between superintending watchman and laboring workman, one should also remember that, as in Hegel, it is the workman to whom any future belongs.) Our freely formed commitments are entangled in and yet outrun reflectiveness\u2014a sense of self that while perhaps always existentially given also becomes especially prominent in modernity, as possible directions of commitment multiply and sheer immersion in necessities of survival diminishes somewhat.\n\nKant then widens, sharpens, and literalizes a modern, Cartesian sense of the possibility and value of awakening through reflectiveness into new and better commitments, coupled with a sense of lingering anxieties and uncertainties. We have, always, according to Kant, apperceptive awareness of the possibility of noting and reflecting on our commitments as our own. While we are bound by the categorical imperative as a law of pure practical reason or reflective deliberation in abstraction from inclinations and desires (and what is that?), we know neither how or why this is so nor what the proper specific directions of response to our being so bound must be (even if certain prohibitions are clear). Nature \"reveals little, but very little\" of a path toward a kingdom of ends. \"Man must give [the] autocracy of the soul its full scope; otherwise he becomes a mere plaything of other forces and impressions which withstand his will, and a prey to the caprice of accident and circumstance.\" But how? And, especially, how over time, continuously, in relation to others and to the changing affordances of culture? We seem to bear, always, and especially in modernity, senses of ourselves as both capable of reflection and cast in courses of life we cannot wholly survey.\n\nTo the extent, then, that these senses of ourselves can be reconciled, that reconciliation will take the form of the expression in judgment and action of an increased (but not perfect) sense of reasonableness, fluency, character, and command mixed with a sense of finitude, apartness, and contingency. This reconciliation will always be partial and provisional. It is not circumscribable according to any fixed policy or order of conceptualization. (Claims to such a circumscription would transgress standing human finitude.) Instead, it is best conceived of as a kind of temporarily displayed power, roughly what Kant calls _M\u00fcndigkeit_ , or maturity. (\"Enlightenment is man's release from his self-incurred tutelage [ _Unm\u00fcndigkeit_ ].\") The implied metaphor in _M\u00fcndigkeit_ ( _Mund_ = mouth) of coming to speech or voice is apt. We are able to achieve, and we are to achieve, not final moral knowledge but rather a certain kind of more fluent, clearer, more formed, more focused, and more articulate stance or address to or in life. As Thomas Pfau remarks in commenting on the use of \"voice\" ( _Stimme_ ) and associated terms such as mood, attunement, determination, and agreement ( _Stimmung_ , _Bestimmung_ , _\u00dcbereinstimmung_ ) in the _Critique of Judgment_ , Kant's conception of voice is understandable as \"aiming to reconcile, however provisionally, the experience of a deeply significant interiority with an articulation of its social significance\" in a way that \"manifests a unique form of desire\"\u2014a desire for fluently expressive, reasonable self-command in judgment and action within social space: a desire for recognition, which desire does not admit of perfect satisfaction.\n\nOur efforts to move toward increased fluency, clarity, and command begin not simply in a grasp of abstract universals, not simply in the law-governed motions of physical particles, and not simply in our psychological hardwiring but also in and through following, imitating, and reacting to the subjectivities of others, as manifested in directions of gaze and interest. Aristotle captures this point in remarking that human beings, in contrast with other animals, are \"thoroughly mimetic and through mimesis take [their] first steps in understanding.\" Others use words of some generality and potential for use on further occasions; in doing so they manifest certain directions of gaze and interest. They manage their uses in virtue of having mastered prevailing routines well enough. But their masteries and the uses that flow from them remain ungrounded in any ultimate realities. Hence at least some uses are liable and likely to shift over time as routes of interest and feeling shift. To come to conceptualization through the mimesis of specific routes of usage, gaze, and interest is to be caught up in a stable enough but also pluralized and partially contested life of subjectivity in the world. There may be good reason to regard certain kinds as \"really instanced\" in nature. There is no good reason to suppose that water or tigers, say, are arbitrary human constructs, and this situation is unlikely to change. But this stable situation does not root the life of concepts \"in\" nature alone, independently of mimetic circuits.\n\nMastery and fluency within a life with concepts are hence to be understood not simply as a grasp of fixed archetypes, patterns, or _Bedeutungsk\u00f6rper_ that lie \"behind\" usage in a standing way, but rather as matters of a grasp of patterns together with an ability both to imitate and to redirect gaze and interest\u2014to respond anew to life. Only when we see that conceptualization involves all this can we arrive at a form of philosophical understanding that is not blind to the life of human subjectivity in its life with words. Adorno makes this point eloquently in recommending \"extinguishing the autarky of the concept,\" that is, recommending that we see both the possession and the very nature of concepts as bound up with stable enough but also sometimes contestable mimetic circuits rather than rooted in \"contact\" with absolute givens.\n\nA philosophy that... extinguishes the autarky of the concept strips the blindfold from our eyes.... Insight into the constitutive character of the nonconceptual [i.e., the deictic, sensuous, and mimetic] in the concept would end the compulsive identification which the concept brings unless halted by such reflection. Reflection upon its own meaning is the way out of the concept's seeming being-in-itself as a unit of meaning.\n\nOnly this recognition of the nature of the concept\u2014the beginning and partial continuance of the life of concepts in mimesis, where there are always residues, remainders, and other possibilities\u2014blocks philosophy from dehistoricized, potentially smug policymongering and permits a grasp of life. \"Disenchantment of the concept is the antidote of philosophy. It keeps it from growing rampant and becoming an absolute to itself.\"\n\nHere \"disenchantment of the concept\" and \"the antidote of philosophy\" might well be taken as significations for literature and its work. Literature foregrounds reconfigurative responsiveness to incidents and actions that take place in time over static depiction of the physically objectual, and it foregrounds figuration and the expression of attitude and emotion toward what is depicted over measurement and neutral classification. But literature is also a form of thinking that uses concepts in order to seek orientation in life under forms of emplotment and in order to work through perplexity. Conceptual identity thinking and mimesis, thought and emotion, recognition and pleasure in form, philosophy and literature\u2014the members of these pairs are all essentially interrelated, as human beings take their first steps in understanding (toward conceptualization) through mimetic responsiveness in practice and then continue to seek more fluent, stable orientation in their lives in time. Working against conceptual ossification and taking seriously perplexities and failures of orientation that demand address, literature undertakes to reconfigure patterns of mimesis so as to embody freer and fuller responsiveness, in order to form more whole and stable individuals, forms of culture, and conceptual repertoires. \"The trace of memory in mimesis, which every artwork seeks, is simultaneously always the anticipation of a condition beyond the diremption of the individual and the collective.\" \"Always the anticipation\"\u2014diremption, within individuals and between individual and collective, is never wholly overcome. There is no arrival at complete, detailed, specific understanding of shared, coherent institutions and practices and at satisfaction within them, no coming to fulfillment of any Hegelian idea of freedom. But greater fullness of orientation, resolution of perplexity, and clarity and adequacy of feeling remain possible, and literature remains, always (along with other forms of art, but with its own special verbal achievements and sense of temporality), a central form of the pursuit of these possibilities.\n\nPut somewhat more domestically, the thought is that literature helps us to engage anew\u2014more reasonably, with more wholeheartedness and fullness of attention and less incoherence\u2014with life. As Catherine Wilson puts it,\n\nA person may learn from a novel [or other work of literature]... if he is forced to revise or modify, e.g. his concept of \"reasonable action\" through recognition of an alternative as presented in the novel [or other work].... The term learning applies [here] primarily to a modification of a person's concepts, which is in turn capable of altering his thought or conduct, and not primarily to an increased disposition to utter factually correct statements or to display technical prowess.... The ability to go beyond what has actually been fed in in the teaching process stems... from a more fundamental\u2014and perhaps even radical\u2014alteration in the way in which he perceives [certain phenomena of life].\n\nTo these claims it needs to be added only that the occasion for literary writing and for responsive literary reading is typically perplexity in life or something not making emotional or narratable sense; that the modification of concepts involves also the modification of emotion, stance, and action; that the aim is increased fluency, clarity, coherence, and felt aptness of orientation within life in culture; and that the occasions for modification are endless.\n\nPressure is placed on our concepts, stances, and attitudes as they stand by perplexities\u2014in large cases by traumas\u2014that those concepts, stances, and attitudes do not readily accommodate. By taking up literary work as either a writer or a reader, one may respond fruitfully to such pressure in a variety of ways. Sometimes one may successfully work through a perplexity or trauma so as to arrive at a fuller, more emotionally and attitudinally apt stance and story about what is going on. Perplexities of emotional entanglement and of stance can sometimes be resolved in the achievement of a kind of more stable and apt calm, in a way that Spinoza describes in his _Ethics_. Or sometimes one can (also) become more actively engaged in and satisfied within the sheer activity of either making or following a literary form as an expression of alert and masterful subjectivity, as Charles Altieri has suggested, in tracking what he calls the particulars of rapture that writers sometimes achieve and in which readers sometimes share. Or sometimes one can (also) modify one's courses of action in life, so as to embody more fully both more resolved stances and more accomplished energies of form-making. Always the work of the formation and enactment of subjecthood and culture remains unfinished, remains to be done anew.\n\nIf that work remains always unfinished, one might nonetheless hope to elucidate it (rather than to master and explain it by reference to fixed externalities) by setting various exemplary pieces of that work in comparison and contrast with each other. Instead of subsuming all cases of the work of literature under a master universal or Platonic form, one might see some cases as forming what Wittgenstein calls a perspicuous representation ( _\u00fcbersichtliche Darstellung_ ), an arrangement of cases that enables \"just that understanding which consists in 'seeing connexions.'\" One can, or at least one can hope that one can, come to see how subjectivity begins its life within intersubjective, mimetic relations and thence seeks orientation within that life in a variety of ways via literary attention, with different achievements of composure, focus of attention, and deployment of energy. A particular arrangement of such cases will be, at least when it is successful, a \" _Darstellung der Darstellungen_ \"\u2014a figured, materially specific presentation of various figured, materially specific presentations of the life and work of subjectivity. Here, in the chapters that follow, a play, a novella (coupled with a philosophical self-interrogation), an extended lyric poem, an unconventional sonnet, and a long story are set in juxtaposition, together with various more generalizing materials, in the hope of constructing an elucidation of the workings of literature in relation to the lives of modern subjects.\n\nThis presentation of cases will neither erase all differences among them nor spare readers the critical work of comparing differences as well as similarities. Stoppard's recovery of the patterned ritual of dancing, for a viewing audience outside the action, is not the same as Goethe's letting go of Werther in order to continue his own life of writing. Wordsworth's ending in modest prayer and chastened hope is not the same as either Rilke's call for a turn or Sebald's witness and wonder. Notably, Stoppard and Sebald as contemporary writers seem less committed to hope and resolution, or are less able to give them articulate expression, than are Goethe and Wordsworth, as though the times were bleaker than they were one hundred or two hundred years ago. (Do our moods believe in each other less than they did around 1800? For what reasons? Are we to conclude that they are unable to believe in each other at all?) Yet there are also affinities among these cases. Relative calm and aesthetic closure are achieved, and life is seen through the work more steadily and whole, without denying complexity and conflict.\n\nSeeing that the reading and writing of literature cultivate a kind of reflective depth, a kind of complex seeing that is achieved through figure and form, may help us then to avoid reducing literature to \"anything that is written.\" It may further help us to find a way between the Scylla of didacticism and the Charybdis of formalism. Surely literature must \"say something\" about life. But surely, too, the way in which what is said matters, and literature produces less \"moral news\" than didacticism supposes. The suggestion, then, is that literature is a sort of formally significant attention to life, where what _shows_ in literary forms of attention and arrangement of materials is a continuing aspiration for expressive freedom and fulfillment, typically both shaped and frustrated, in part, in specific ways in specific cultural settings. We see ourselves as pursuers of expressive freedom in situ, under difficult conditions, and we so see ourselves in the protagonists and authorial personae whom we encounter (or create). By thus recognizing ourselves, we can become somewhat more reflectively deep about the contours of human life in time.\n\nIn a recent book on the philosophy of art, I claim that \"works of art [literary and otherwise] present a subject matter as a focus for thought and emotional attitude, distinctively fused to the imaginative exploration of material.\" This definition of art\u2014if that is the right word\u2014specifies criteria in Wittgenstein's sense for the use of the word \"art.\" It undertakes \"to elucidate and organize our linguistic and conceptual practice, in a situation in which we are confused by the varieties of artistic practice, by the varieties of things people say about them, and by the powerful but obscure character of our own responses,\" and yet where, still, something can be said about what we do.\n\nIn relation to literature, what this means is that literature as an art, when it is successful, has representational-thematic, expressive-attitudinal, and formal-material dimensions, all in interaction with one another. This thought is in the spirit of Aristotle's claim in the _Poetics_ that a successful tragic drama will be a presentation (mimesis) of an action with all of plot, character, thought, melody, diction, and spectacle. According to Aristotle, each of these parts of a successful tragic drama must be properly coordinated with the others. Too much spectacle and too little plot, for example, will yield in one way the episodic and in another way what are perhaps the excesses of Euripedean stagecraft or the special-effects movie. Too much plot and too little thought will yield a dramatic structure that lacks general thematic significance or that will fail to satisfy the requirement of presenting the universal in the particular. Too little melody, diction, and spectacle\u2014that is, too little concern for the embodiment of the presentation in just a certain set of words and stagings, both crafted and felt\u2014will likewise fail in presentational power or illumination, presenting instead only what is already known and distinctively clarifying nothing.\n\nMore abstractly, content and form both matter, and they matter in their specific ways of relating to one another. As Wolfgang Huemer has recently remarked, \"If we try to define what is particular about literary texts, we find that they put an emphasis not on _what_ is said, but on _how_ it is said; literary language makes itself manifest.... At least to some extent in literary texts language itself becomes the topic.\" An emphasis on _how_ what is said is said is especially prominent in the so-called L = A = N = G = U = A = G = E poets, such as Charles Bernstein and Clark Coolidge, with their radicalization of both the voice of lyric poetry and the symbolistic-imagism of objectivism. Yet even they, while foregrounding the sheer look, sound, and feel of words and assonances, produce texts that admit of some paraphrasability and readability. One can\u2014just barely\u2014say what they are about. Hence one should say instead that literary texts put emphasis not _only_ on what is said (represented and expressed) but _also_ on how it is said.\n\nSuch emphases are also often in play in so-called ordinary speech and writing, in contexts from journalism to conversation to criticism to history. But this just shows that, as Huemer goes on to remark, \"Literature... is not a niche phenomenon; it must not be viewed as an unnecessary but entertaining ornament, but rather as a practice central to our language without which we might not even be able to master a language as complex as ours in the first place.\" The special intensities literature achieves are not sideways to life, with their own special domains of objects known such as possible worlds or fictional objects; they are rather part of specially apt attention\u2014all at once representational-thematic, emotional-attitudinal, and craftlike-sensuous\u2014to ordinary life. These intensities are achieved through the controlled and aptly original use of devices that structure perception, thought, and feeling, including emplotment, metaphor, allegory, irony, hyperbole, understatement, and assonance, among many, many others.\n\nThrough the apt use of such literary devices, fullness of attention (ideational, emotional, and sensuous) is achieved rather than shirked. Clich\u00e9, or unthinking repetition of what is merely rote and stock, is the enemy of literary art. The satisfaction sought in literary art involves what Spinoza called the transformation of an inadequate idea of an affection into a more adequate idea. Through literary art one gains a better understanding of what is worth feeling and caring about in what ways, where this better understanding is grounded in what one does in fact feel and care about when one pays the fullest possible attention to the objects and quality of one's experience. By following the work of literary art, one may move into a structure of care, reflection, and investment in activity that is more stable and appropriate to the objects and events of human life. Such moves may also include involvement in the work itself, in its specifically formed patterns of attending, partly (but only partly) beyond the objects of attention. Literary art has its disruptive powers as well as its powers of focusing on phenomena of human life.\n\nTo be interested in literary art thus means being attentive to what William Rothman and Marian Keane have called \"the astonishing capacities for meaningfulness that [works of art] have discovered within the singular conditions of their medi[a].\" These capacities of meaningfulness involve in literature the achievement of fullness of attention to phenomena of life through the use of literary devices. Through such fullness of attention, a structure of care, reflection, and investment in activity is achieved, so that we lead more freely and fully the lives of persons or selves who take an interest in their worlds, rather than being buffeted about by experience received only passively and inchoately. Apart from the kinds of noticings, expressings, and respondings that art and literature can embody and support, our lives can become pale, conventionalized, anonymous, or, one might say, not deep, not so fully the lives of subjects. And beyond the attention to life it affords, there is also the astonishment of the work itself, that it has found a way to mean once again or anew.\n\nA significant corollary of this view about literature, attention, and fuller personhood is that writing and reading are understood as neighboring modes of activity, both of which involve the cultivation of attention. What we do when we read well is follow and identify with achievements of fuller attention as they are managed in situ by both writers, on the one hand, and protagonists in literary works, on the other. What I am suggesting is that the life of persons inherently involves the pursuit of a fuller and more stable structure of care, reflection, and investment in activity (despite or across tragic inhibitions) in densely textured ways in specific cultural settings.\n\nIn describing the various individual cases that I take up, I am sometimes led to formulate claims about how we respond to a particular line or image, or even more broadly about what we are like. I am aware that this usage of \"we\" is far from common in literary studies and that there are significant reasons for being suspicious of it. Perhaps, therefore, it will be of some help to say something explicitly about exactly what kind of enterprise is implied by this usage.\n\n\"We\" as I use it (and as it is typically used by philosophers of certain kinds) is meant to be improvisatory and invitational: to invite others to share in and test a thought for themselves. It is not meant to be a report on the results of research into what countable individuals have in fact said or thought or felt. Claims about how we respond, what we feel, and what we are like are, therefore, in a distinctive way vulnerable, naked, and exposed. This invitational (philosophical) usage of \"we\" traces back at least to Socrates, when he remarks in the _Republic_ that we can be (must be) \"ourselves both jury and advocates at once\" in considering what will count as more fulfilling conditions of human life. We may try, that is, to be clearer and more articulate about fundamental interests, in such a way that others may also share in both the process and the articulation of results\u2014a very tentative and vulnerable enterprise indeed, and not at all declamatory. This same usage of \"we\" appears in ordinary language philosophy when what we say is investigated. Astonishingly, such claims about what we say can sometimes command assent of some circumference, and they can do so for those within that circumference with an air of overwhelming naturalness, reasonableness, and rightness to the ear. (J. L. Austin was a master at articulating such claims.) When this happens, a community of articulate understanding and commitment discovers itself in and through the common acceptance of such claims.\n\nOnce upon a time, perhaps in the heyday of New Criticism, poetry was read in something like this spirit, that is, with attention to \"how we are moved\" by the poem, how \"we\" follow its sense, etc. Since the end of that heyday, it has been \"discovered\" that not everyone either interprets or responds to a given poem (to any poem) in the same way. This is certainly true. Even the claims about what we feel or say or respond to that achieve the widest circumference will fall short of universality. There is no perfect route for the mimetic enactment of subjecthood; there are too many contending ambitions and senses of self housed within persistent cultural antagonisms for everyone to respond alike. The energy would be drained out of cultural life were that uniformity of response _per impossible_ to come about. Difference neither should be nor can be so easily overleaped. As a result, however, of the \"discovery\" of diversity of response, the study of literature has become an increasingly sociologized enterprise. Cultural studies as a field arises out of the thought that we should study in a systematic, empirical, nondoctrinaire way who in actual fact says (and feels) what when. (Bourdieu is a paradigm of this study, and his empirical investigations of differences in cultural reception have been taken up in literary studies by figures such as John Barrell, Marjorie Levinson, Jerome McGann, and Edward Said, among many others.) That is, of course, something that can be done, and it can yield interesting and important results. It would by no means be an obvious advance to go back to the smugness and staleness of certain forms of New Criticism without having these other kinds of critical, cultural investigation also going on.\n\nIn the face of all this, why then might anyone still bother with the vain effort to articulate what we say, feel, or respond to, knowing that any such effort is doomed to partiality and so to a form of failure? The worry is that without such efforts we abandon ourselves to a modern, materialist, competitive, value-denigrating individualism that destroys all circuits of the mimesis of response and so destroys the very life of subjectivity as such. What is left without this effort is a culture of the competitively individualist seeking of the satisfaction of subjective preferences, without any sources of a commonwealth and without stability or depth of individual identity over time, but instead only pervasive cultural crassness, economic and political exploitation, and individual anomie. Hence it may be worthwhile, at least sometimes, to persist in the vain effort to form both communities of interpretation and evaluation and a more stable and fully invested life for individual subjectivity in and through the common acceptance of what we say. The effort to do this is a defining ambition for philosophy, literature, and criticism that it would be impoverishing to forego, however impossible it is to complete it. _Ich kann nicht anders_.\n\nIf we cannot productively engage in this work of the formation of deeper, subjectively fuller senses of self and of shared commitment, then we are, as subjects, lost, dead. But out of a fear of loss of culture and subjecthood in culture as it stands, one can sometimes do something. One can pay attention to what is perplexing in life (and in art) in the hope both of resolving one's emotional and attitudinal stance into something calmer and more stable and of mobilizing greater energies of commitment. One can write about one's perplexities, and one can read one's way through others' ways of encountering perplexity that are more articulate and more persuasive than one's own. And then one can, and must, wait, unable either to control the response of any audience or to form an audience of universal circumference. This is what certain writers (and other artists) have always somehow known how to do for us in detail, endlessly, with power, grace, and responsiveness to life, and in the furtherance of life.\n2.\n\nRomanticism, Cartesianism, Humeanism, Byronism\n\n_Stoppard's_ Arcadia\n\nWhat philosophy knows as the mind-body problem is also and perhaps more deeply a problem in our practical, cultural lives and in the self-images that are woven through them. It is hard to avoid thinking of ourselves as \"free subjectivities,\" capable of choice and responsiveness to reasons, who stand \"over against\" a physical nature in which objects are composed and events occur according to laws that make no reference to choices or reasons. But this makes it difficult to see how choice and responsiveness to reasons can be expressed within a \"mere\" nature that somehow \"houses\" our lives and practices. How, if at all, is free life according to reason and within the framework of the natural world possible? We shall scarcely be able to make progress on this question until we confront the cultural practices that embody and shape our images of nature and of ourselves.\n\nJohn Dewey makes the practical, cultural dimensions of the mind-body problem wonderfully clear in a long passage from _Art as Experience_ :\n\nWe inherit much from the cultures of the past. The influence of Greek science and philosophy, of Roman law, of religion having a Jewish source, upon our present institutions, beliefs and ways of thinking and feeling is too familiar to need more than mention. Into the operation of these factors two forces have been injected that are distinctly late in origin and that constitute the \"modern\" in the present epoch. These two forces are natural science and its application in industry and commerce through machinery and the use of non-human modes of energy....\n\nScience has brought with it a radically novel conception of physical nature and of our relation to it. This new conception stands as yet side by side with the conception of the world and man that is a heritage from the past, especially from that Christian tradition through which the typically European social imagination has been formed. The things of the physical world and the moral realm have fallen apart, while the Greek tradition and that of the medieval age held them in intimate union\u2014although a union accomplished by different means in the two periods. The opposition that now exists between the spiritual and ideal elements of our historic heritage and the structure of physical nature that is disclosed by science, is the ultimate source of the dualisms formulated by philosophy since Descartes and Locke. These formulations in turn reflect a conflict that is everywhere active in modern civilization. From one point of view the problem of recovering an organic place for art in civilization is like the problem of reorganizing our heritage from the past and the insights of our present knowledge into a coherent and integrated imaginative union.\n\nThe problem is so acute and widely influential that any solution that can be proposed is an anticipation that can at best be realized only by the course of events.... It is true that physical science strips its objects of the qualities that give the objects and scenes of ordinary experience all their poignancy and preciousness, leaving the world, as far as the scientific rendering of it is concerned, without the traits that have always constituted its immediate value. But the world of ordinary experience in which art operates, remains just what it was.\n\nAccording to this passage, there is, on the one hand, stuff or material itself indifferent to us and our aspirations, disenchanted (in Weber's famous phrase), and with its motions having no natural ends or purposes. This is the \"radically novel conception of physical nature\" that Dewey has in mind. At the very least, and metaphysics and epistemology to one side, it has served us well in many respects to think of nature in this way. Once we so conceive of nature, and then further carry out the appropriate investigations of the lawlike but nonpurposive behaviors of mere material things, then we can, sometimes, manipulate those things in order to satisfy desires, needs, and interests that we experience ourselves as just having. The modern scientific understanding of material nature lays the cognitive groundwork for practices and systems of, for example, medicine, transportation, communication, and industrial production that it would be difficult and undesirable to abandon.\n\nAnd there is, on the other hand, us, we with our purposes\u2014purposes that seem, in light of the disenchantment of things, ineluctably subjective, inner matters of groundless preference alone. If we should happen to be able to make use of material things to satisfy our preferences, great\u2014and likewise great if two or more people should happen to have overlapping preferences. Finally, just as a matter of political compromise to avoid violence that threatens to inhibit all preference satisfaction, it is very often best not to enforce preferences: let individuals with their preferences be who they are and let them trade with one another in free markets as they wish. For most of us, at least in the developed worlds, life without modern technologies and modern market systems of production and exchange would be both unthinkable and undesirable.\n\nYet, as Dewey suggests, this picture, however ineluctably built into our culture, of an inner, subjective mental life, with only subjective purposes, facing off against an outer, material, objective but meaningless nature is also not an entirely happy one. For one thing, this picture affords no basis for objective assessment of pursuits of subjective interest, that is, no basis for appeals to justice or fairness that might constrain rapacious or exploitative behavior. It may be that a free market works efficiently to maximize preference satisfaction among traders with relatively equal holdings and stocks of information but different preferences, and there is therefore good reason at least sometimes to think of free markets as fair. But if imbalances in holdings, power, or information grow too great, or military might intervenes, or free riding is possible, then this institutional arrangement is likely to prove unstable. Then the guns or lawsuits start. And what then? If there are only individuals who are competing with one another for the material resources to satisfy subjective desires, then it is likely in the end to be guns rather than lawsuits. Lawsuits and court verdicts may be construed as themselves covert forms of violence. Family life and citizenship are all too likely to decay into what Hegel calls \"particularity by itself, given free rein in every direction to satisfy its needs, accidental caprices, and subjective desires, [so that it] destroys itself and its substantive concept in this process of gratification.\" Anarchy, both social and personal, is loosed upon the world. Plato predicts explicitly that this will happen in a pluralist, subjective democracy that lacks any metaphysically founded conception of justice. There seems no longer to be any metaphysical standard for checking on what we do, and without one we seem likely to do just about anything, including a lot of fighting. Underlying this fighting, there is at least the risk that no one will really believe in the worth of a way of life. My preferences may seem to me to be just given and not to be of any worth to me or to anyone else. Why should I care about anything, I may worry? Subjective anomie, or what is generally now called depression, threatens us, and it is more or less endemic in modern industrial societies. And yet it would, again, be both difficult and undesirable to give up the benefits of modern science and its culture in order to revert to a more closed, traditionalist, metaphysically or religiously circumscribed way of life.\n\nHow, then, might we best think of ourselves and our place in nature so that we might both accept the benefits of modern science and democratic culture and yet avoid or at least curb their harms? This question has been raised for us by thinkers as various as Kant, Hegel, Marx, Dewey, and John Paul II, albeit that they each have quite different answers in view.\n\nOne particularly interesting suggestion for thinking about this problem comes from the eminent literary critic Northrop Frye. The suggestion arises out of a very broad sketch Frye offers of the history of Western thought and language. According to Frye, there are three successive historical stages of basic styles of thought and language. The first stage is the metaphorical-mythological stage, as people tell stories simultaneously about what we now call physical events and about the meanings of things. Science and religion, natural cosmology and creation theory, historical reporting and primeval storytelling are all not yet sharply distinguished from one another. One sees the dawn or the spring as the coming of a beneficent divine presence, or one sees a storm as the divine wrath of the sea itself, understood as both a physical something and a personality. Frye identifies this stage of thought and culture with pastoral and nomadic life generally and with roughly Homeric Greece and the early Hebrew tradition in particular.\n\nThe metaphorical-mythological stage is then superseded by a metonymic-intellectual stage. Allegory becomes a dominant form, as signs are taken to indicate a deeper order of reality in relation to which our ordinary experience is only a surface. Access to this reality is claimed by educated elites, who consequently lay further claim to the administration of daily life and general culture. Justifications for how things are to be done are propounded by these elites, on the basis of their expert knowledge of the deeper and fuller reality. According to Frye, one can see something of this stage of thought and language as early as the pre-Platonic Protagorean tradition in Greece. It figures in Plato's dialogues, and then in Christianity, which Nietzsche famously described as Platonism for the masses. In Christianity from at least the Augustinian period onward, the liturgies and sacraments are administered by expert priests, and the regulation of daily life is referred to the reality described in the Bible, read aright in Latin by an educated minority. Late-medieval Everyman plays participate in this form of language and thought, as the ordinary person's life is seen as an allegory of the sufferings and possibilities of resurrection that were disclosed by Jesus and that inform the lives of all of us.\n\nIn the early modern period, progressively from roughly 1550 or 1600 and into the present, this metonymic-intellectual system is succeeded by a demotic-scientific-manipulative system. Modern scientific knowledge is available to anyone who takes the trouble to educate himself. As Descartes once remarked, \"There are many things to do in life, and [a good man] has to direct that life in such a manner that the greater part of it shall remain to him for the performance of good actions, which his own reason ought to teach him, even supposing that he were to receive his lessons from it alone.\" The very idea that there are many things to do in life\u2014including at least discoveries to be made and technological devices to ameliorate our material situation to be invented\u2014as opposed to one central thing that is to be done, namely living according to the will of God, is itself revolutionary. The further idea that we can use our reason to figure out how to do the many things we might do is equally far-reaching. As these ideas are worked out in modern scientific culture, enormous benefits accrue, while at the same time our modern political and moral lives become pluralized and, potentially, evacuated of meaning, in being no longer referred to a larger reality that is either metaphorically or metonymically accessible.\n\nLike Kant and Hegel and Dewey, Frye, while accepting the benefits of modern scientific and technological culture, worries about this. He worries in particular that there is nothing any longer to hold us together within the terms of a common project. Without the ability to discern either metaphorically or metonymically possibilities and necessities of personal and cultural development that are latent within a larger reality itself, chaos threatens. We may fall into \"the subordination of everything creative to the expediencies and superstitions of authority... [or we may] fly apart into a chaos of mutually unintelligible elites, of which those nearest the center of society would soon take control. So atavistic a regression, in the present stage of technological development, might well wipe the human race off the planet.\" This passage is perhaps somewhat purple and apocalyptic, but the problem is clear. What, if anything, can any longer bring us together under a shared sense of common, objective possibilities of life and value? The old dispensations are dead, and for good reason, but a life lived without any objective dispensations threatens to be bleak, chaotic, and violent, or perhaps nasty, brutish, and short. The constructed institutions of the democratic state and the free market may, once again, intervene to moderate the problem. Social order and open trading are by no means insignificant institutional goods. But what is to prevent free riding and the domination of state and market institutions by the powerful?\n\nThis sense of a need for a new dispensation is the central _point de d\u00e9part_ for romantic thought and writing and for the thought of Kant, as he seeks to found a critical and constructive philosophy that avoids both traditionalist but baseless dogmatism and skeptical nomadism in life and in thought. Frye's own response to this need, building on Blake and on Blake's reading of the Bible, is to suggest that we can and should learn from the great poets and from the Bible to uncover and reactivate the myth of all mythologies: the tentative availability of a reconciled, pastoral, resurrected life. The idea is that we can, as it were, bypass the metonymic-intellectual stage of thought and regain contact with the metaphorical-mythological stage that remains present as a dim, underlying stratum of our lives. This is, Frye suggests, exactly what great poets and the writers of the great sacred texts, preeminently the Bible, do. Donald Marshall has elegantly summarized this strategy of recovery as it was pursued by Wordsworth:\n\nIn Wordsworth the synthetic, creative, and sympathetic power of imagination, nourished on a popular tradition of ballad and romance with roots in the great poetry pre-dating the Enlightenment, asserted itself against an instrumentalist reason, which in poetry took the form of a masquerade in the form of conscious and merely willed classicism. Wordsworth found the true source of imagination: in nature and particularly in the poet's experience of nature during childhood, when he was most open to its varied and spirited influence. The language in which this recollected experience was transformed into the guide of later life and feeling derived from the ordinary language of men, particularly rural men, whose lives preserved the great rhythms of pastoral and agricultural life, recorded in and mediated by the Bible, anonymous folk poetry, and related literary forms.\n\nIn _The Romantic Legacy_ , Charles Larmore has similarly argued that the romantic imagination functions to express and recover senses of community and of belonging to place, though he aptly notes also a contending sense of romantic irony, as the poet simultaneously feels apart from others in the possession of distinctive education and creative power.\n\nThis romantic sense of a recovery through imagination of a suppressed stratum of thought, language, and experience, so that we might once again feel ourselves to have a common situation and objective purposiveness, is a wonderful idea. But, as the careers and receptions of Blake and Wordsworth show, it will not be so easy to carry it out in a way that significantly influences public life. Those who pursue this strategy are all too likely to be dismissed as dreamers or balkanized as objects of mostly private, merely religioaesthetic reverence and reverie, at least in relation to serious questions of social policy that require fully worked out schemes for institutions. Can imagination, poetry, myth, and metaphor make high cognitive claims on us? On many or most of us? And what institutions will then serve? As Hegel noted in criticizing romanticism, a sense of subjective inwardness informs a good deal of romantic writing, as poets despite their best intentions for social effect withdraw into rehearsals of the progress of their own imaginations, as in Wordsworth ever withdrawing from work on _The Recluse_ to write _The Prelude_ instead. The thought that romantics withdraw from the world in order to find solace in nature has informed much of the reception of romanticism, in the sense in which Jerome McGann has criticized romantic _ism_ \u2014that is, the dominant teaching of romantic poetry within departments of literature up until, say, 1980\u2014for its subjectively cultic character. (McGann distinguishes between institutionalized romanticism and a tougher, stranger, more self-critical romantic writing.) When one then further takes into account a sense, inherited from Freud, of the anarchistic pressures placed by our sexual lives on both individual development and imaginative production, then the prospects for cultural restoration via romantic imagining grow even bleaker. And then there are the categories that are reinforced every day by an increasingly global commodity culture: subjective preference, taste, and want, which stand against objective production costs and processes. How is imaginative art to make a public claim on us in the midst of the domination of social life by these categories of thought and experience?\n\nThe address to our cultural situation that is offered us by romantic imagining has considerable pertinence and power. Given, however, the evident difficulties that attach to carrying out a romantic renovation of culture, it is worthwhile considering what other possibilities of general address to our cultural circumstances are on the books. Three further stances, each significantly different from romanticism, can be usefully distinguished.\n\nThe first is \"Relaxed Naturalism\" or \"Just Coping\" or, to give it a proper name, Humeanism, alluding to Hume's remark that we should acknowledge \"the whimsical condition of mankind, who must act and reason and believe; though they are not able, by their most diligent enquiry, to satisfy themselves concerning the foundation of these operations, or to remove the objections, which may be raised against them.\" One might, that is, think that there are properly or realistically no such things as an objective plight or an objective destiny to be recovered. There are, rather, just many people who want many different things, in a situation in which material resources for satisfying wants are simply moderately scarce but not altogether lacking. This is perhaps generally the situation in the North Atlantic democracies. Richard Rorty had the habit of claiming that these democracies offer us, as a merely contingent possibility that we have somehow invented or stumbled upon, a comparatively good enough way of life. Talk of achieving our destiny is to be rejected as pretty much amounting to the nostalgia of the priests. This Rortian view has considerable currency, at least for public life, against the more religious visions of Alasdair MacIntyre and Charles Taylor. Assignments and enforcements of human rights can be defended as matters of pragmatic compromise, given the practical necessities of at least some social cooperation. In Quine's elegant phrase, morality becomes a matter of \"birch rod and sugar plum.\" That is, there are certain behaviors that we more or less decide to reward and to punish, because rewarding and punishing these behaviors works well enough to keep us going and to enable us to satisfy some of our wants. We can drop all talk of renovation, destiny, and objective purpose. In political life, elbow room or negative liberty is good enough. Privately, a bit of Millian experimentalism in lifestyles is not really a bad thing. This is, perhaps, the dominant view of life in northern Europe and the \"Blue States\" of the United States nowadays. It is unlikely that it will pass away any time soon, and it is not at all clear that its passing would be desirable. It offers us a fair amount of independence from authoritarian comprehensive enforcements of social visions. This view urges us and even enables us one by one, or affinity group by affinity group, or as citizens who share at least some bits of history, just to do the best we can. There is at least a hint of this view in even John Dewey, alongside the strains of moral and cultural perfectionism in his work.\n\nThe difficulty of this view, already suggested, is that it leaves public life open to manipulation by powerful elites. It encourages free riding and an insidiously creeping social chaos and decline, as Plato and MacIntyre have argued. It makes it difficult to believe in one's way of life, so that social anomie and depression threaten. It has trouble figuring out what to do in real crisis situations, where there are not many relevant experiences and not many rules of thumb on which to draw.\n\nThe second view is Cartesianism. The cognitive and technological benefits of the Cartesian conception of disenchanted, material nature are manifest. But in addition to these cognitive and technological benefits, there is also an attractive moral, spiritual stance associated with this conception of nature and with the relation of mind to it. As Charles Taylor notes, Descartes furthers an\n\nethic of rational control that find[s] its sources in a sense of dignity and self-esteem [by] transpos[ing] _inward_ something of the spirit of the honour ethic.... Strength, firmness, resolution, control, these are the crucial qualities, a subset of the warrior-aristocratic virtues, but now internalized. They are not deployed in great deeds of military valor in public space, but rather in the inner domination of passion by thought.... Descartes constantly enjoins efficacious action for what we want [so that we may become \"masters and possessors of nature\"], alongside detachment from the outcome.\n\nAs Descartes puts it in the _Discourse_ ,\n\nMy third maxim was always to try especially to conquer myself rather than fortune, to change my desires rather than the order of the world; and generally to become accustomed to believing that there is nothing that is utterly within our power, except for our thoughts, so that, after having done our best regarding things external to us, everything that fails to bring us success, from our point of view, is absolutely impossible.\n\nBut satisfaction in correct thinking is virtually infinite. Thus Descartes argues that if we learn to follow correctly the proper principles for scientific investigation, then we can comport ourselves both with pride in our cognitive achievements and with humility, in acknowledgment of the limits of our finite understanding. Pride where knowledge and, sometimes, consequent technology are achievable, coupled with stoicism about our limits, leading to ataraxia or blessedness, is an available stance that has genuine charms. Descartes himself writes that we may, if we take up this stance, \"rival the gods in their happiness\" and experience \"intense satisfaction\" than which there is nothing \"sweeter or more innocent... in this life.\"\n\nIt would be folly to underestimate either the practical-technological or the moral-spiritual benefits of this stance. But it too faces problems. This stance does not point to any practices or styles of expressive action in politics, family life, or interpersonal relations generally. It is expressed directly only in cognitive practice, leaving everything else either to be ignored, to be coped with as a matter of convenience, or to be sorted out via the practical, ultimately market-structured adjustment of preferences. There is no distinctive worth or dignity attaching to any particular mode of interpersonal, familial, social, or political life. Thus Descartes remarks that, apart from the practice of natural science, \"the most useful course of action was to rule myself in accordance with those with whom I had to live,\" whether Persians, Chinese, or Frenchman. This policy kept Descartes free from the Inquisition, and it supports considerable broad-minded tolerance of what are ultimately the follies of one or another human group in interpersonal matters, where no well-founded rules are available. But it does not support the achievement of intimacy. It is hard to see how Descartes could tell the difference between getting along with a wife and getting along with the Chinese. It is, therefore, no accident that he never married, and the philosophical problem of other minds that arises in his work is itself perhaps a reflection of a pervasive sense of alienation from other human beings.\n\nThe third stance is Byronism. Byron's own literary and theoretical writings are less interesting systematically than those of any of Blakean-Wordsworthian romanticism, Humeanism, or Cartesianism. But there can nonetheless be little doubt that Byron both summed up and stands for a certain cultural stance, the stance that Bertrand Russell called Byronism as \"Titanic cosmic self-assertion,\" especially in matters sexual. Byron's own (let us call it) passionate and exuberant personal life expresses this stance, at least in part. And there has always been a well-motivated temptation to identify Byron with certain of his characters. Childe Harold, for example, is introduced to us as follows, in terms that seem to apply as well to Byron himself.\n\nWhileome in Albion's isle there dwelt a youth,\n\nWho ne in virtue's ways did take delight;\n\nBut spend his days in riot most uncouth,\n\nAnd vexed with mirth the drowsy ear of Night.\n\nAh, me! in sooth he was a shameless wight,\n\nSore given to revel and ungodly glee;\n\nFew earthly things found favor in his sight\n\nSave concubines and carnal companie,\n\nAnd flaunting wassailers of high and low degree.\n\nFour stanzas later, we learn that he is not much given to repentance.\n\nFor he through Sin's long labyrinth had run,\n\nNor made atonement when he did amiss.\n\nThe reasons for taking up a stance of passionate self-assertion, if reasons are in view, is that it is better to feel something, and in particular to feel one's own powers of command, than to feel nothing at all, albeit that the rate or variety or artistic imaginativeness of conquest may have to be increased in order to get the same effect in feeling, addiction being what it is. This is the stance that we can also see in Don Giovanni and in the seducer Johannes of Kierkegaard's _Either\/Or_. This is largely the popular conception of romanticism. As one fairly simple commentary for students puts it, \"In essence, Romanticism was, for a time, the triumph of feeling over thinking, the head over the heart.\" Romanticism so construed or, better, Byronism, does help to remind us of the felt character of our own inner lives. We can, and often do, feel intensely, without much prior reasoning or policy formation but nonetheless with a kind of imaginative involvement, blending anticipation, recollection, and fantasy, in a manner not present in the lives of other creatures. The power thus to feel with imaginative involvement is one, it seems, that we wish not to repudiate, even if we could. We give it free rein in adolescence, perhaps, or on Halloween, or for Carnival, or for Las Vegas weekends. The liability, of course, is that it is hard to see how to build a stable life out of the cultivation of the pursuit of this kind of intensity of feeling, as Faust is brought in the end to realize.\n\nSo we have these three stances\u2014Humeanism, Cartesianism, and Byronism, against which I have posed a more genuine romanticism, associated with Blake and Wordsworth. All three of these stances are lived\u2014Cartesianism for science and planning, Humeanism for buying and selling and for semistable social relations, Byronism for holidays. When they are thus lived together, in uneasy pragmatic compromise with each other, then what we have is the pragmatic liberalism that Gary Gutting has eloquently defended. This pragmatic compromise solution has a good claim to being our form of life, at least in the more or less well-off North Atlantic democracies and in the Blue States: naturalist rejection of comprehensive religious enforcements, Cartesianism for science, and Byronism on the side, all adopted because they seem to be what works best, pragmatically, in their particular spheres.\n\nThe question, then, is how stable this pragmatic compromise is. Or does it rather suffer from the liabilities of each stance taken individually, plus the added problem that each stance places pressure on the others? For example, Cartesianism may point toward the management of culture by so-called technical experts, rather than compromise, thus undermining democracy. Or Humeanism may undermine commitment to science as a vocation. Or Byronism may threaten to undermine just about anything. Just who are we, and where are we going?\n\nIn this situation, we might conjecture that the Wordsworthian romanticism first sketched offers us something of a middle way, since it accepts elements of Cartesianism, Humeanism, and Byronism but, unlike pragmatism, thinks of maintaining this acceptance as a continuing task. The characteristic Wordsworthian romantic writer\u2014Wordsworth\u2014unlike the Byronic romantic writer, is a continual scrutinizer of the terms of our current mixed settlement. Wordsworth is worried about the rise of a modern scientific culture in which a sense of value and meaning is lost. But, like Hume, he is unwilling, at least in his major writings, to accept comprehensive political enforcements of religious stances. He is too much of an individualist for that. He has a Byronic sense of the power and importance of his own imagination and his imaginative responses to events, but he seeks also to keep his imagination apt to the persons and events he encounters, where the marker of accuracy is that others can be brought to share in his imaginative responses, thus confirming them. Thus his poetic imagination courts not only excess and poetic glory but also depth of common response to common predicaments and possibilities of life. In the _Prelude_ , he undertakes to \"speak \/ A lasting inspiration,\" as he retraces his own fostering \"alike by beauty and by fear.\" The point of this rehearsal is not simply the particularities of his own life but further that within these particularities one can \"trace \/ Our Being's earthly progress,\" thus showing, as it were, the universal, or what is possible and valuable for us all, in the particular, that is, in the details of growing up in the Lake District, studying (or mostly partying) at Cambridge, traveling in France, and so on. The moral of this rehearsal is that \"these objects\"\u2014that is, the beautiful and the sublime\u2014should \"everlastingly affect the mind.\" The experience of the sublime awakens in us a felt sense of our own rational and expressive powers and dignity, so that we do not settle for Humean coping or trying to get what we already take ourselves to want, but instead seek to deploy and express our human powers originally. The experience of the beautiful connects us to the common, so that both Byronic excessiveness and Cartesian alienation are avoided. Throughout these rehearsals, Wordsworth continually questions his own progress in writing and avoids conclusive dogmatism. He wonders whether his tracing of his progress is really as exemplary as he hopes and whether his audience will receive him or repudiate him\u2014indeed, whether his audience exists at all. Since the poet has \"the task of _creating_ the taste by which he is to be enjoyed,\" it may not. In rehearsing his own history, Wordsworth has trouble finding the plot and its moral. We have in the _Prelude_ , he writes to Coleridge, \"Turned and returned with intricate delay.\" Yet the very ongoing effort to find the plot and to establish the importance of the experiences of both the sublime and the beautiful is itself the self-modifying way to balance Humeanism, Cartesianism, and Byronism against one another. This Wordsworthian practice of seeking expressive power in connection with the common is always crossed with self-questioning rather than dominated by a preformulated plan or conclusion. Engaging in this practice is what the best artists and literary writers, in particular situations, do.\n\nOne consequence of this conception of the seriousness of Wordsworthian romanticism is that there is no great romantic drama. The reason for this is that Wordsworthian romanticism lingers in the activity of accepting and working through conflicting commitments, as it accepts the attractiveness within consciousness of all of Humeanism, Cartesianism, and Byronism. It seeks to bring these stances into fuller and more coherent communication with one another within consciousness, so that more human, more expressive action can be achieved. It is no accident that Harold Bloom once described romantic poetry as the internalization of quest romance. The action of romantic drama is internal to consciousness itself. But this internalization of action then cuts against the possibility of presenting important dramatic conflicts between different characters, if these characters are themselves to be complex enough to participate in the movement of romantic consciousness. The only real candidates for great romantic dramas are Goethe's _Faust_ and Shakespeare's tragedies and comedies, so far as one finds confrontation between emergent, modern individualism and valuable, stable commitments to be central to them. But these dramas work so magnificently precisely because the claims on the individual of an existing external culture of honor, religion, nation, or clan as they stand are taken seriously against modern individualism, even if individualism turns out to be an irrepudiable force. Once its irrepudiability is fully accepted, then the drama is internalized, as individuals must sort out ever anew their standing conflicting commitments. (Hamlet is poised on the edge of accepting this irrepudiability, without yet being crassly individualistic.) Modern and modernist lyric and the modern novel can survive and flourish because they are able to focus on the interior life of protagonists in a way that modern drama, with its rejection of the artificiality of the extended soliloquy, cannot. The drama of modern life is largely that of individuals coming to terms\u2014or, increasingly, bleakly failing to come to terms, as in Beckett\u2014within their own consciousness or within restricted spheres of conversation, with how they are to stand in relation to the conflicting attractive possibilities afforded by, or latent within, a culture.\n\nBut what if one were to undertake to write a drama about the fact that modern life offers no ready way to blend naturalness (either Byronic-spontaneous or Humean-customary) with originality (either Cartesian-intellectual or Byronic-spontaneous)? (Beckett rejects ready blending, but by reducing his characters to the barely discursively percipient and his cultures to completely desiccated routines. He thus undervalues complexity, energy, and adaptive responsiveness in both individuals and cultures.) Could one _show_ Byronic types, Cartesian types, and Humean types somehow in interaction with one another, as types, while also intimating that these types represent aspects of us all, in our own divided commitments? What work could such a showing do? Could it point toward any kind of fuller acknowledgment of our complexities?\n\nThese are the questions that Tom Stoppard takes up in _Arcadia_. Stoppard accepts the structural necessity for drama of presenting conflict between characters, and he also rejects the extended Shakespearean soliloquy. We see his characters doing what they do, but we do not hear or overhear them in their internal movements of mind as they are pulled now toward Humeanism, now toward Cartesianism, now toward Byronism. Stoppard also refuses the great marriage plot, as in Jane Austen, within which plot two characters work out the possibility of a good enough life together, as they find their commitments and talents complemented in each other.\n\nThe action in _Arcadia_ takes place, instead, in a kind of public space that we witness from a privileged standpoint, able to watch without being ourselves watched. Stoppard's method of dramatic construction is juxtaposition. The characters are largely types, with Thomasina as a Cartesian figure (with a bit of Byronism struggling to get out); Septimus as a Byronic figure (overlaid with the surface Cartesianism of a Cambridge education); Hannah as a cooler mixture of Cartesian scholarship with modern Humean, tolerant whimsy at the follies of mankind; and Valentine as a contemporary Cartesian. Other characters are even closer to pure types in a way that makes for farce, as they are dominated by particular varieties of ambition and vanity. They think of themselves as something they are not: Bernard takes himself to be a scholar; Ezra takes himself to be a poet. The remaining characters are largely incidental to the action taking place between the principal, more rounded four (Thomasina, Septimus, Hannah, and Valentine) and the two figures of farce (Ezra and Bernard). Stoppard sets his characters as types within a space that we can witness, and he waits to see what happens. To some extent, his method of juxtaposition deliberately shirks the internal development of character and the working through of conflicting commitments in favor of the charms of farce.\n\nStoppard himself is quite aware of how his dramatic method of the juxtaposition of types works. As he remarked in an interview, \"I don't think _Arcadia_ says very much about these two sides of the human personality or temperament [that is, the Cartesian and the Byronic].... And yet it's firing all around the target, making a pattern around the target.\" He reports that his favorite line in modern English drama is \"I'm a man of no convictions\u2014at least I _think_ I am,\" from Christopher Hampton's _The Philanthropist_. He observes that he \"writes plays because writing dialogue is the only respectable way of contradicting yourself.\" He describes his objective as \"to perform a marriage between a play of ideas and a farce.... [This objective] represents two sides of my own personality, which can be described as seriousness compromised by my frivolity, or... frivolity redeemed by my seriousness.\" \"Happiness is equilibrium. Shift your weight,\" he remarks in his own voice, quoting his character Henry from _The Real Thing_. What all these remarks indicate is a shying\u2014perhaps as a result of the necessities of dramatic presentation, perhaps from overwhelming shyness expressed as wit, perhaps because of the sheer complexities of modern life\u2014from working through, from thinking. Juxtaposition, pattern, contradiction, equilibrium\u2014these trump internalization and the working through of thoughts, ideas, attitudes, and emotions.\n\nAnd yet, as J. L. Austin once wrote, \"there's the bit where you say it and the bit where you take it back.\" When _Arcadia_ opened in 1993 in London and in 1995 in New York, it was widely (though not without exception) praised as a breakthrough in character development and emotional expressiveness and as a move, in particular, beyond the consistently arch and dryly intellectual quality of his earlier work. As Tim Appelo put the point in his review in _The Nation_ , \"Unlike the spy-jive mac-guffins he juggles in _Hapgood_ , the mystery addressed in _Arcadia_ is one to which Stoppard is fully emotionally committed.\" There is something to this point, and it has mostly to do with the concluding scene, where the two worlds\u2014those of 1809 and of the present\u2014as it were overlap. Here is where, at last, we see not farce and wordplay and juxtaposition but development in character, consciousness, and relationship. First Thomasina and Septimus, and then Hannah and Gus, waltz. These parallel waltzes have the feel of a dream. They are surprising\u2014especially so in that, like dreams, they contravene time in occurring together. This gives them the feel of being somehow mythical or eternally recurring, something that ever haunts us. The waltzing together of these pairs, across time, has the fuguelike feeling of something half occurrent and half remembered.\n\nThe absence of dialogue during this waltzing is prepared by Valentine's earlier remark that he's given up on his analysis of the rise and fall of the grouse population on the Coverly estate because there's \"Too much noise. There's just too much bloody noise.\" (Too many unpredictable external factors induce deviations in the population that prevent any natural pattern from being evident.) This remark alerts us that we may, at least sometimes, find sense in silence rather than in speech. In this final waltzing, and in the overlapping of the two time periods, we find \"patterns making themselves out of nothing.\" We are left with a sense that the problems of human subjects struggling to express their emotional and intellectual subjectivities fully, originally, and with each other within settings of thermodynamically decaying nature and stale culture persist, making us above all interesting animals. But despite their persistence, a significant response to these problems may be, for a time, possible. Thomasina remarks to Septimus, \"there is another geometry which I am engaged in discovering by trial and error.\" Stoppard's juxtapositions work similarly, allowing a geometry or a set of shapes of human life, intelligence, and desire to show themselves.\n\nThe dreamlike, fuguelike feeling of the final waltzing is further reinforced by the fact that it is these pairs of characters who waltz, in particular by the fact that these waltzes are for each of them a kind of breakthrough. Septimus acknowledges Thomasina's just-about-adult sexuality, which has now come to expression along with her intelligence. In doing so, he further acknowledges his own depth of attraction to her as a person, to her embodied intelligence, thus overcoming his earlier libertinism in favor of something more like love. Hannah acknowledges Gus's pain and neediness and intelligence in his silence, thus lending depth of responsiveness to her own typical professional scholarly scrupulousness. Although she has earlier stuck to detachment, insisting to Chloe that \"I don't want a dancing partner, least of all Mr. Nightingale. I don't dance,\" she too is now able to dance, a bit awkwardly, when the right partner comes along at the right time. In inviting her to dance, Gus acknowledges her intelligence and passion together, in taking her as fit for dancing, thus acknowledging, too, that words and feelings can coexist in a single character: depth of feeling need not always engender muteness, and cleverness need not always suppress feeling. It is as though the two parts of the soul\u2014analytical intelligence and depth of feeling, Cartesianism and Byronism\u2014have at last been put together, at least for a moment, according to the logic of a dream, across time, and surrounded by music rather than parsed out in words. Is this final scene, blending disjoint times and moving to music, without words, an escape from actuality into form or a registering of human need and possibility? It is inescapable to ask this question, but it is not clear that it is necessary to answer it one way or the other.\n\nAccording to this scene and the logic that prepares it, the Wordsworthian practice of bringing the parts of the soul together, in pursuit of expressive fluency in thought and feeling and action, in relation to others and to what is common, both informs and haunts human life. But the work of this practice remains always in part unfinished. As Hannah remarks, \"It's wanting to know that makes us matter.... Better to struggle on knowing that failure is final.\" We will remain interesting animals, in pursuit of fullness of fluency and at-homeness as subjects that we will never quite achieve. The dancing of these pairs\u2014their real physical movement and intimacy, yet somehow outside historical time and \"for us\" as observing audience\u2014offers a clarifying catharsis of what is possible for us. But, in Marcuse's words, \"the reconciliation which the catharsis offers also preserves the irreconcileable.\" The space of dramatic art is not joined to the space we occupy as viewers and then as agents. The marking of this magical, dramatic space where some provisional reconciliation takes place as outside of time and as other to us testifies to what Adorno called our continuing \"suffering in an existence alien to the subject\"; the dancing that these pairs are able to achieve as human agents testifies \"to love for it as well.\" We both recognize ourselves in these characters and remain aware of their occupying a space of art that we can, would, and yet cannot occupy wholly in daily life. _Arcadia_ itself closes with this dancing and so is silent\u2014bleakly, pregnantly, undecidably\u2014about the rest of life.\n\nSerious writing must find some way to show that moments, perhaps even ones of considerable scope and duration, of good enough fluency and at-homeness are possible, if it is not to reduce us to ignorant and empty sites of mere coping with life. Yet it must also accept that such moments do not last forever, especially in light of modern complexities of desire and social life, and that there is no formula for either achieving or sustaining them. It must accept a constitutive incompleteness\u2014accept, that is, its own failure to track the achievement of any final happiness, if it is to be faithful to the lack of final happiness in human life. It must somehow avoid denying human finitude and temporality in complacent dogmatism while also succeeding in showing sometime achievements of expressive, embodied intelligence and the satisfaction of desire, as in dancing. This astonishing concluding scene in _Arcadia_ manages to blend skepticism and acknowledgment of finitude with the presentation of apt, fluent feeling and of gratitude for life. \"In the silence you don't know, you must go on, I can't go on, I'll go on.\"\n3.\n\nRomantic Subjectivity in Goethe and Wittgenstein\n\nAs a result of the fine work of Mark Rowe, Joachim Schulte, and Gordon Baker and Peter Hacker, it has now been evident for some time that there are deep affinities\u2014affinities in style and textual organization, in conceptions of elucidatory explanation via comparisons, and in a sense of subjectivity housed within nature\u2014between the Goethe of the _Farbenlehre_ and the Wittgenstein of _Philosophical Investigations_. Among the very deepest of these affinities is their shared sense of the limits of metaphysical explanation. The identification of simple elements is always relative to purposes and circumstances, never ultimate. Hence there is no single kind of ultimate explanation running from the nature and behavior of ultimate simples to the nature and behavior of complexes composed out of them. There are often useful explanations to be found of how the behaviors of complexes are determined by the behaviors of their parts, but this kind of explanation is one among many. Comparative descriptions of complexes\u2014whether of organisms, human practices, works of art, or chemical and physical structures\u2014are not to be supplanted in favor of ultimate metaphysical explanation.\n\nValuable and sound though these ideas are, they are, however, not the theme here. Instead, a different form of affinity between Goethe and Wittgenstein is more in view\u2014a substantive affinity in their senses of what it is to be a human subject. Methodological and general metaphysical affinities are taken for granted; beyond or behind them lie affinities of substance in their conceptions of human life.\n\nThomas Mann's remarks on _Die Leiden des jungen Werthers_ provide a useful starting point. \"It would,\" Mann observes,\n\nnot be a simple task to analyze the psychic state that determined the underpinning of European civilization at that time [1774, the date of publication of _Werther_ ].... A discontent with civilization, an emancipation of emotions, a gnawing yearning for a return to the natural and elemental, a shaking at the shackles of ossified culture, a revolt against convention and bourgeois confinement: everything converged to create a spirit that came up against the limitations of individuation itself, that allowed an effusive, boundless affirmation of life to take on the form of a death wish. Melancholy and discontent with the rhythmical monotony of life was the norm.\n\nIt is likewise not a simple task to say what may have made such melancholy and discontent the norm, at least in certain circles. Secularization, bringing with it a sense of lost meaningfulness as religious ritual became a smaller part of daily life, and modernization, bringing with it a market economy and new but very uncertain life chances, are surely part of the story. But secularization and modernization are themselves interwoven with deep, largely tacit self-understandings about what is worth doing, in ways that are difficult to disentangle. Charles Taylor, in his monumental survey of the making of the modern identity, describes what he calls \"three major facets of this identity: first, modern inwardness, the sense of ourselves as beings with inner depths; second, the affirmation of ordinary life which develops from the early modern period; third, the expressivist notion of nature as an inner moral source.\" These facets of identity that come to the fore in modernity are as much a part of a widely available human repertoire of identity as they are byproducts of something else. They are, according to Taylor, an inescapable part of our moral framework, a set of commitments that we cannot help but have, even where they also sit uneasily against one another, as the claims of the ordinary pull against the pursuit of original expressive power.\n\nOne important result of these commitments is an undecomposable intermingling of moral discovery with moral invention. We no longer think of ourselves as simply living out in one way or another basic human tendencies that are simply given. Rather, drawing on our reflective inwardness, on ordinary life, and on natural energies, we partly make ourselves what we are. As Taylor puts it, \"We find the sense of life through articulating it. And moderns have become acutely aware of how much sense being there for us depends on our own powers of expression. Discovering here depends on, is interwoven with, inventing.\" Finding is inseparable from founding.\n\nBut why should this occasion what Mann noted: discontent, yearning, shaking, revolt, melancholy, and a death wish? It is easy enough to see why a certain improvisatoriness and independence of mind might be valued. But how and why did our moral improvisations come to be freighted with all that? Here the answer has to do with a certain lack of both ground and closure to our moral efforts. Without fixed tendencies and _tele_ as starting points and endpoints, it becomes uncertain what moral progress and human achievement might look like\u2014even uncertain whether they are possible at all. Philippe Lacoue-Labarthe and Jean-Luc Nancy note the uncertainties that attach to our moral efforts in the wake of the felt absence of any fixed presentation of the self and its powers, as they describe the conception of the subject in Kant's moral theory.\n\nWithout oversimplifying or hardening the contours of a question that merits extended analysis, we cannot fail to note that this \"subject\" of morality can be defined only negatively, as a subject that is not the subject of knowledge (this knowledge suppressed \"to make room for belief\"), as a subject without _mathesis_ , even of itself. It is indeed posited as freedom, and freedom is the locus of \"self-consciousness.\" But this does not imply that there is any cognition\u2014or even consciousness\u2014of freedom.... [T]he question of [the moral subject's] unity, and thus of its very \"being subject,\" is brought to a pitch of high tension.\n\nLacking a fixed ground and definite _telos_ , efforts at articulating and enacting \"a sense of life\" come to be marked by a desperate intensity. Different subjects become variously lost within different ongoing projects of articulation, each maintaining its sense of its place and progress not through ratification by an audience, which is all too caught up in its own projects, but rather through a hysterical lingering in process. Articulation \"sets out to penetrate the essence of poiesy [poetic making], in which the [articulation] produces the truth of production itself... the truth of production _of itself_ , of autopoiesy.\" The manifold modern _Bildungsromanen_ and personal epics of coming to self-consciousness and assured social vocation, but specifically _Bildungsromanen_ and epics that have difficulty in reaching their own conclusions (other, perhaps, than by taking the artistic making of the very work in hand as the achieved _telos_ ), are evidence of the dominance of the project of autopoiesis in the modern moral imagination. Human moral self-imagination and achievement become a \"question of the _becoming_ present of the highest,\" not of its _being_ present.\n\nThe three inescapable parts of our moral framework that Taylor identifies\u2014inwardness, ordinary life, and nature as an expressive resource\u2014conspire in our experience with and against one another to inhibit the achievement of a stable sense of life. Either nature in the aspect of the sublime conspires with inwardness to resist the sways of ordinary life and conventionality, thus setting up the image of the chthonic genius as the exemplar of moral achievement, as in Nietzsche, or nature in the aspect of the beautiful conspires with ordinary life and conventionality, thus setting up an image of pastoralized domesticity as the exemplar of moral achievement, as in certain moments in Rousseau. Each image then stands in immediate criticism of the other, and no stable image of moral achievement persists.\n\nUnder such uncertainties and instabilities, it is all too plausible that one might not only become melancholic but come to wish for nothing more than surcease, even to regard the taking of one's own life as the only possible creative act with a fixed endpoint, as the only meaningful act. Or of course, more modestly, one might forego efforts to live according to a sense of life or to what is highest and assume instead an instrumentalist stance toward the things of life, seeking only modest satisfactions. This strategy is common in modernity, and it is surely honorable. But does it quite escape the silent melancholies, quiet desperations, and covert nihilisms about which Emerson and Thoreau and Nietzsche variously warned us?\n\nTo come now specifically to _Werther_ : Werther's own character is torn between the idealized images of chthonic originality, represented for him by the wild excesses of his own inner emotional life, and pastoralized domesticity, represented for him by the figure of Lotte, maternally feeding bread to her younger brothers and sisters. There are interesting historical specificities that surround the split in Werther's character\u2014and in Goethe's\u2014between these two ideals. In his monumental study of Goethe's development, Nicholas Boyle suggests that these ideals are posed, and posed as irresolvable, for Werther and for Goethe, by certain strains in eighteenth-century German culture. \"Werther's innermost life,\" he writes,\n\nis determined by a public mood; he lives out to the last, and inflicts on those around him, the loyalties which\u2014because they are literary, intellectual, in a sense imaginary loyalties, generated within the current media of communication\u2014most of his contemporaries take only half-seriously. His obsessions are not gratuitously idiosyncratic\u2014they belong to his real and socially determined character, not just to a pathologically self-absorbed consciousness.\n\nSpecifically, Boyle suggests that Leibnizianism, pietism, and sentimentalism offered images to Goethe of \"the self thirsting for its perfectly adequate object.\" This thirsting of the self for a confirming object took an especially inwardized turn in Germany, since it could not plausibly be welded to a project of political nation building. Autonomy or achieved selfhood had to be found within, and its principal marks were inner intensities of imagination, feeling, and devotion. Goethe's subjectivity, like Werther's, is dominated by \"his belief in binding moments of insight\" to be achieved fitfully against the sway of official and conventional culture. At the same time, however, Goethe also absorbed a certain political realism and social consciousness from the Storm and Stress movement. He had an awareness of individual character types, including his own, as specific social roles\u2014a novelist's sense (unlike anything in Werther himself) of social reality as narratable from multiple points of view. In _Werther_ , as Boyle characterizes it, \"the Sentimentalist content of the novel is in perfect but momentary balance with a Storm and Stress aesthetic which determines the manner of its presentation.\" Like Werther, Goethe in writing _Werther_ \"endeavored to find roles for himself to act out which both had some general moral or historical significance and could be filled by him with a sense of selfhood: roles which fused both a [social] character and [an intensely individual] consciousness.\" Inwardness and the pursuit of chthonic originality alone lead to empty solipsism; acceptance of oneself as a social type and conformity to convention alone lead to derivativeness and imaginative death. The task is to combine the pursuit of originality with acceptance of oneself as a social type. Unlike Werther, Goethe himself carried out this task through the act of writing about his innermost emotions and self-imaginations in a social setting. This act of writing gave him the opportunity both to cultivate his inner life and to achieve a certain realistic distance from it. For Werther, faced with the same task and torn between his hyperbolic idealizations of originality on the one hand and domesticity on the other, things do not go so well. The \"very impetus to self-destruction is being imposed on him by the German public mind\"\u2014itself faced with the problem of cultivating both autonomous selfhood and continuing sociality\u2014\"commerce with which he cannot avoid, or wish to avoid, if he is to express himself at all.\" The task of blending selfhood with social identity is unique neither to Werther, nor to Goethe, nor to the German public mind of the late eighteenth century. It is the fate of modern subjectivity as such either to face or to evade it. A sense of this problem as pressing and not to be evaded then arises with special intensity in late eighteenth-century Germany, in the wake of sentimentalism (itself a response to modernization and secularization) and in a hyperfractured political actuality.\n\nLike Boyle, Mann too characterizes Werther as \"the overrefined final product of the Christian-Pietist cult of the soul and of the emotions.\" What this means, above all, is a desire for singularity, specifically a desire to desire, intensely and infinitely. As Mann puts it, \"the desire to exchange that which is confining and conditional for that which is infinite and limitless is the fundamental character of Werther's nature, as it is of Faust's.... He is in love even before his love has an object.\" Even Lotte asks Werther, \"Why must you love me, me only, who belongs to another? I fear, I fear, that it is only the impossibility of possessing me that makes your desire for me so strong.\" Only a desire for the impossible can certify itself as genuinely singular and original, capable of confirming selfhood against the grain of conventionality.\n\nIn the grip of such a desire, impossibly seeking original selfhood both against the grain of all conventionality and yet blended with social identity, no one knows what to do. Our desires are original if and only if impossible, unrecognizable\u2014and they are recognizable and satisfiable if and only if they are mimes of the conventionalized desires of others. No wonder Werther observes that \"all learned teachers and tutors agree that children do not understand the cause of their desires; but no one likes to think that adults too wander about this earth like children, not knowing where they come from or where they are going, not acting in accord with genuine motives, but ruled like children by biscuits, sugarplums, and the rod\u2014and yet it seems to me so obvious\" (9). Werther cannot anywhere recognize, act on, and satisfy his own desire as his own.\n\nAs he then himself wanders the earth, impossibly seeking fully original selfhood blended with social identity, Werther alternates in his moments of attachment and identification between surrender to beautiful scenes of sociality, composure, convention, and pastoralized domesticity, on the one hand, and ecstatic abandonment to sublime scenes of wild creative energy, on the other. In neither moment is the attachment or abandonment either ordinary or in fact achieved; in both cases it is hyperbolized in Werther's imagination into something exceptional, and his hyperbolizing imagination blocks his actually doing anything.\n\nThe emblem in nature of the beautiful, of pastoralized domesticity, and of attachment, in Werther's imaginative perception, is the cozy valley of Wahlheim\u2014home's choice. \"It is,\" Werther writes early on,\n\ninterestingly situated on a hill, and by following one of the footpaths out of the village, you can have a view of the whole valley below you. A kindly woman keeps a small inn there, selling wine, beer, and coffee; and she is extremely cheerful and pleasant in spite of her age. The chief charm of this spot consists in two linden trees, spreading their enormous branches over the little green before the church, which is entirely surrounded by peasants' cottages, barns, and homesteads. Seldom have I seen a place so intimate and comfortable.\n\n(10)\n\nThe force and direction of Werther's idealization is evident in his litany of adjectives: \"kindly,\" \"small,\" \"cheerful,\" \"pleasant,\" \"little,\" \"intimate,\" and \"comfortable.\" Here he would\u2014originally and creatively\u2014surrender himself to a domesticated, given, human life in nature. But to desire to do this originally and creatively is to make one unable to do it, and Werther simply gazes on the scene until, as he thinks of himself, he reverts in thought to the idea of nature as also a source of iconoclastic creative energy.\n\nThe counterpart scene in which Werther imagines ecstatically abandoning himself to the sublime comes late in his correspondence, as things are not going well. On December 12, he writes:\n\nSometimes I am oppressed, not by apprehension or fear, but by an inexpressible inner fury which seems to tear up my heart and choke me. It's awful, awful. And then I wander about amid the horrors of the night, at this dreadful time of the year.\n\nYesterday evening it drove me outside. A rapid thaw had suddenly set in: I had been told that the river had risen, that the brooks had all overflowed their banks, and that the whole valley of Wahlheim was under water! I rushed out after eleven o'clock. A terrible sight. The furious torrents rolled from the mountains in the moonlight\u2014fields, trees, and hedges torn up, and the entire valley one deep lake agitated by the roaring wind! And when the moon shone forth, and tinged the black clouds, and the wild torrent at my feet foamed and resounded in this grand and frightening light, I was overcome by feelings of terror, and at the same time yearning. With arms extended, I looked down into the yawning abyss, and cried, \"Down! Down!\" For a moment I was lost in the intense delight of ending my sorrows and my sufferings by a plunge into that gulf! But then I felt rooted to the earth and incapable of ending my woes!\n\n(69\u201370)\n\nIf only he could give himself over to this energy in sublime nature, to this wild torrent, the problem of the satisfaction of impossible desire would at least be ended, if not solved. Werther's itinerary lets itself be read as a move from sometime attachment to the beautiful to complete domination by the sublime, ending in the realization that only this end is possible. In some earlier scenes of the perception of nature, Werther's awareness shifts abruptly and jarringly back and forth between a sense of the \"overflowing fullness\" of nature, before which he feels \"as if a god myself,\" and a sense of nature as \"an all-consuming, devouring monster\" (36\u201337). At this late moment in December, he remains in the condition he had earlier ascribed to humanity in general: \"we are as poor and limited as ever, and our soul still languishes for unattainable happiness\" (20). His death looms, but he does yet quite grasp it: \"My hour is not yet come: I feel it\" (70).\n\nWerther's relations to Lotte directly mirror his relations to nature. Both are dominated by his hyperbolizing imagination, as he sees her now as beautiful, now as sublime. When he first sees her, he finds\n\nsix children, from eleven to two years old... running about the room, surrounding a lovely girl of medium height, dressed in a simple white frock with pink ribbons. She was holding a loaf of dark bread in her hands, and was cutting slices for the little ones all round, in proportion to their age and appetite. She performed her task with such affection, and each child awaited his turn with outstretched hands and artlessly shouted his thanks.\n\n(15)\n\nEverything here is simple, cozy, natural, and artless, in forming a scene of mildness with which Werther would like to identify. But then he also dreams that \"I pressed her to me and covered with countless kisses those dear lips of hers which murmured words of love in response. Our eyes were one in the bliss of ecstasy\" (70). There is scarcely a better case than this of a fantasized, impossible specular moment.\n\nIn each case, Lotte is more a posited object of Werther's fevered imagination of himself in relation to her than she is seen by him as a being in her own right. She is an occasion for him to fantasize himself complete, both original and at home. Lotte here plays the same role as was played by the earlier object of his affections, whom Werther describes wholly in terms of her effect on him: \"I have felt that heart, that noble soul, in whose presence I seemed to be more than I really was, because I was all that I could be. God! Was there a single power in my soul that remained unused? And in her presence could I not develop fully that intense feeling with which my heart embraces Nature?\" (8). Here, as ever, the real object of Werther's consciousness is _my soul_ , _my heart_ , _my seeming to be more than I really was_. No wonder, then, that when he imagines that she loves him, Werther rhapsodizes in the same egocentric terms: \"That she loves me! How the idea exalts me in my own eyes! And... how I worship myself since she loves me!\" (27).\n\nWerther's self-claimed exceptionalism, his sense that, unlike in ordinary people, \"there lie dormant within me so many other qualities which wither unused, and which I must carefully conceal\" (8), leads him consistently to scorn ordinary life and the achievements of reciprocity, decency, and human relationship that are possible in it. In particular, he scorns, while also envying, Albert's staid conventionalism and decency. But he here finds Albert only to be typical of what most people are like. \"Most people,\" he writes, \"work the greater part of their time just for a living; and the little freedom which remains to them frightens them, so that they use every means of getting rid of it. Such is man's high calling!\" (8). In contrast, Werther seeks for himself a genuine high calling and exemplary, commanding achievement outside the framework of the ordinary. Not for him \"the gilded wretchedness, the boredom among the silly people who parade about in society here\" (44) at court, a world in which he stands as if \"before a puppet show and see[s] the little puppets move... completely occupied with etiquette and ceremony\" (45). Unable to mix with them, he argues that real love, constancy, and passion \"exists in its greatest purity among that class of people whom we call rude, uneducated\" (55), as he again hyperbolically idealizes a pastoralized ordinary life. Yet he is unable, with his dormant qualities he must carefully conceal lest he subject them to the risks of public scrutiny, to mix with ordinary people either.\n\nWork, too, is treated by Werther as something either stalely conventional and meaningless or idealized as salvific. On the one hand, \"the man who, purely for the sake of others, and without any passion or inner compulsion of his own, toils after wealth or dignity, or any other phantom, is simply a fool\" (28). On the other hand, \"Many a time I wish I were a common laborer, so that when I awake in the morning I might at least have one clear prospect, one pursuit, one hope, for the day which has dawned\" (37). In both cases, his attention is on the work as the vehicle of the exalted expression of his personality, not on the work itself and those who do it. Even when he imagines doing a small bit of work in first arriving at the court of Count C., his thoughts remain on himself and his superiority to others. \"But when, in spite of weakness and disappointments, we do our daily work in earnest, we shall find that with all false starts and compromises we make better headway than others who have wind and tide with them; and it gives one a real feeling of self to keep pace with others or outstrip them in the race\" (42).\n\nWerther's God is similarly exceptional\u2014a being whom he assumes either specifically listens to his pleas or specifically avoids them, without any mediating institutions or any involvements in the lives of others. On November 30, as he approaches his end, he addresses God directly and intimately, presuming to be his particular and special son.\n\nFather, Whom I know not\u2014Who were once wont to fill my soul, but Who now hidest Thy face from me\u2014call me back to Thee; be silent no longer! Thy silence cannot sustain a soul which thirsts after Thee. What man, what father, could be angry with a son for returning to him unexpectedly, for embracing him and exclaiming, \"Here I am again, my father! Forgive me if I have shortened my journey to return before the appointed time. The world is everywhere the same\u2014for labor and pain, pleasure and reward, but what does it all avail? I am happy only where thou art, and in thy presence I am content to suffer or enjoy.\" And Thou, Heavenly Father, wouldst Thou turn such a child from Thee?\n\n(64)\n\nHis address here is strikingly reminiscent of his earlier thoughts about Lotte, whom he similarly regards as his unique savior. \"I cannot pray except to her. My imagination sees nothing but her; nothing matters except what has to do with her\" (38).\n\nWhat does it all avail? Seeking absolute and perfect ratification of his exceptional personality and talents and perfect, autonomous selfhood joined to continuing sociality in a life of daily self-affirming divinity but finding only ordinary people and his own tortured thoughts and fantasies, Werther can in the end hit only on the strategy of giving it all up. The only freedom from continuing failure is death. \"We desire to surrender our whole being\" (20), and if partial, egocentric surrender to Lotte, to art, to nature, or to work is received and ratified by no one, ordinary as they all are, then genuine surrender must be complete, an escape from life itself. \"I have heard of a noble race of horses that instinctively bite open a vein when they are hot and exhausted by a long run, in order to breathe more freely. I am often tempted to open a vein, to gain everlasting liberty for myself\" (50). As he recalls almost kissing Lotte, \"And yet I want\u2014but it stands like a barrier before my soul\u2014this bliss\u2014and then die to expiate the sin! Is it sin?\" (62). In the end, \"The body was carried by workmen. No clergymen attended\" (87).\n\nGoethe himself, of course, did not commit suicide, despite the autobiographical character of the novel. Mann suggests that Goethe's willingness to go on living had to do with his sense of his identity as a writer. \"Goethe did not kill himself,\" Mann writes, \"because he had _Werther_ \u2014and quite a few other things\u2014to write. Werther has no other calling on this earth except his existential suffering, the tragic perspicacity for his imperfections, the Hamlet-like loathing of knowledge that suffocates him: thus he must perish.\" How did Goethe then come to have and to be aware of having another calling, one that made life for him worth living? As Mann suggests, the answer has to do with the very act of writing _Werther_ , as well as with the ongoing activity of writing for a public already begun with _Goetz von Berlichingen_. For Goethe, the act of writing in general, and of writing _Werther_ in particular, combined a kind of catharsis\u2014both a clarification and an unburdening\u2014of his emotional life with the achievement of a kind of distance or perspective on himself. He came through writing to achieve a sense of himself as having a social identity as a writer, so that the problem of wedding autonomous selfhood to continuing sociality did not for him go fully unsolved. It would be addressed again and again in the act of writing, from _Faust_ to the lyric poetry to _Elective Affinities_ , though with more maturity and never quite perhaps with the immediate cathartic intensity of address of _Werther_. Yet even in his maturity Goethe retained an intense subjectivity capable of responding to others as though they were vehicles of salvation for him. Mann notes that at the age of seventy-two he fell in love with the seventeen-year-old Ulrike Sophie von Levetzow. Though address and partial solution to the problem of subjectivity are possible, full solution is not.\n\nLudwig Wittgenstein's character strongly resembles those of Goethe and Werther. Both his personal and philosophical writings combine an intense wish for attachment to others and to activities as vehicles for the expression of the higher self he felt himself to have with an equally intense critical scrutiny of that wish. The subtitle of Ray Monk's biography, _The Duty of Genius_ , captures this feature of his character well. For the young Wittgenstein in particular, the realization and confirmation of genius was, in Monk's words, \"a Categorical Imperative,\" and the only alternative to failing to follow it was death: genius or suicide. \"Wittgenstein's recurring thoughts of suicide between 1903 and 1912, and the fact that these thoughts abated only after Russell's recognition of his genius, suggest that he accepted this imperative in all its terrifying severity.\"\n\nMonk traces Wittgenstein's submission to this imperative to his reading of Otto Weininger's _Sex and Character_ , published in 1903, the year of Weininger's own suicide. Brian McGuinness accepts this connection but goes further to read this imperative into the composition of the _Tractatus_ and to situate it in the context of Wittgenstein's family life and surrounding culture. McGuinness characterizes what he calls \"the final message of the _Tractatus_ \" as \"perhaps a clearer, a more concentrated view... would enable him to see the world aright. At any rate, if there was no real prospect of this: if he could not reach this insight, and if he could not get rid of his troubles by reconciling himself to the world, then his life was pointless.\" What made this question\u2014genius or suicide\u2014arise with special force in Wittgenstein's case, McGuinness argues, was not only the example and influence of Weininger or the general sickness of prewar Austrian culture but also and more deeply the influence on him of his father. The Wittgenstein family\n\nformed a sort of enclave, fortified against the corruption and inadequacy that surrounded it by severe and private moral standards, which, it seemed, some of them had not the temperament to match or meet. Ludwig's case... seems to have been that of a phenomenally strong assent and attachment to these standards, often at war not only with the normal human failings that became glaring in their light, but also with a particularly soft and affectionate nature.\n\nYet McGuinness immediately goes on to add that Wittgenstein himself \"was not one to see his problem as that of being unable to do what his father required,\" and he further comments that \"what we are describing here is no disease. As Tolstoy says: 'These questions are the simplest in the world. From the stupid child to the wisest old man, they are in the soul of every human being.'\" With the example of _Werther_ before us, we can see the problem of genius or suicide as forming a strong theme in German culture in its response to the yet more general problem in modernity of wedding autonomous selfhood to continuing sociality.\n\nWittgenstein's preoccupation with autonomy and with the realization and confirmation of genius against the grain of culture is pronounced in the remarks in his own voice published as _Vermischte Bemerkungen_ ( _Culture and Value_ ). \"It's a good thing,\" he writes, \"I don't allow myself to be influenced.\" As is typical in the post-Kantian, post-Goethean German tradition, the realization of genius is conceived of as a matter of letting something natural and divine come to the fore in one's thought and life, often under the prompting of nature itself. \"Just let nature speak and acknowledge only _one_ thing as higher than nature, but not what others may think\" (1e). \"Don't take the example of others as your guide, but nature!\" (41e).\n\nWhen one is thus guided, one's thinking and acting happen with significance, in and through one, rather than under one's personal control. \"One might say: art _shows_ us the miracles of nature. It is based on the _concept of the miracles of nature_. (The blossom, just opening out. What is _marvellous_ about it?) We say: 'Just look at it opening out!'\" (56e). It is just this kind of natural yet significant opening out of his own features of character that Wittgenstein anxiously hoped might inform his own thinking and writing.\n\nSchiller writes of a \"poetic mood.\" I think I know what he means, I believe I am familiar with it myself. It is a mood of receptivity to nature in which one's thought seems as vivid as nature itself.... I am not entirely convinced that what _I_ produce in such a mood is really worth anything. It may be that what gives my thoughts their lustre on these occasions is a light shining on them from behind. That they do not _themselves_ glow.\n\n(65e\u201366e)\n\nSomething hidden, powerful, and natural within oneself is to come to the fore, in a way that is not under one's egocentric control. One is to be swept along by one's genius into a natural-supernatural movement of thinking.\n\nYet talent can also be betrayed or misused, and so fail to confirm itself in its products. \"Talent is a spring from which fresh water is constantly flowing. But this spring loses its value if it is not used in the right way\" (10e). As a result, the most important thing is to come to think and write naturally, in faithfulness to one's talent and against the grain of culture. But the effort to do so takes place within the conventionalized space of personally controlled and discursive reflection, so that it is crossed by an anxious self-scrutiny. \"Am I thinking and writing as it were beyond myself, out of the depths of the natural?\" one egocentrically and discursively wonders, or Wittgenstein wonders, in just the sort of tragic obsessiveness about his own imperfections that Mann saw in Werther. \"Working in philosophy... is really more a working on oneself. On one's own interpretation. On one's way of seeing things. (And what one expects of them.)\" (16e). \"No one _can_ speak the truth; if he has still not mastered himself. He _cannot_ speak it;\u2014but not because he is not clever enough yet. The truth can be spoken only by someone who is already _at home_ in it; not by someone who still lives in falsehood and reaches out from falsehood towards truth on just one occasion\" (35e). All or nothing; natural-supernatural, nonconventionalized, poetic truth and expressiveness or imitative, derivative, nonexistence; genius or suicide.\n\nDomination by this imperative produces the same complex of attitudes toward work and toward religion that we find in Werther. On the one hand, Wittgenstein idealizes ordinary manual work as something beautiful and honest, more honest than intellectual chatter: \"what is ordinary is here filled with significance\" (52e), if the manual work is done with respect and integrity. It is no accident, but rather deeply part of his anxious self-scrutiny and his attitudes toward culture and value, that Wittgenstein so often urged others to take up this kind of work. On the other hand, \"Genius is what makes us forget skill\" (43e). It is beyond the ordinary. So how can one express genius within the framework of the ordinary? How can one write poetically\u2014originally and yet in a way that draws on the common and is accessible to others? How can one wed autonomous selfhood to continuing sociality? \"I think I summed up my attitude toward philosophy when I said: philosophy ought really to be written only as a _poetic composition_.... I was thereby revealing myself as someone who cannot quite do what he would like to be able to do\" (24e).\n\nReligious faithfulness offers a paradigm of significant expressiveness, but it is a paradigm that in its traditional, institutionalized form is dead for us, shot through with the conventionality that expressiveness is to overcome. \"What is good is also divine. Queer as it sounds, that sums up my ethics. Only something supernatural can express the Supernatural\" (3e). What is needed is \"a light from above\" that comes to the individual soul, not religious institutions and ordinary religious training. \"Is what I am doing really worth the effort? Yes, but only if a light shines on it from above.... And if the light from above is lacking, I can't in any case be more than clever\" (57e\u201358e). Religious belief cannot be something that is simply given and shared. It must rather be achieved through the dormant qualities of one's soul coming actively to take religious life as the vehicle of their expression, as providing the terms of deep significance. \"It strikes me that a religious belief could only be something like a passionate commitment to a system of reference. Hence, although it's a _belief_ , it's really a way of living, or a way of assessing life. It's passionately seizing hold of _this_ interpretation\" (64e).\n\nAbove all, what Wittgenstein wants from religion, from work, from the guidance of nature, from his genius, but can never quite find, is full-blooded and continuing significance in the face of mere conventionality and cleverness: a new life. \"A confession has to be part of your new life\" (183). And if not a new life, then death: genius or suicide, or suicide as the creative act of voluntarily removing oneself from a cycle of unending self-defeat. In 1946, in the middle of remarks about music, thought, Shakespeare, God, heroism, and the difficulty of philosophy, there occurs in _Culture and Value_ , in quotation marks, the very last words of Werther to his correspondent Wilhelm: \"Lebt wohl!\"\n\nThe intensities of Wittgenstein's character have been well documented in the biographical literature. Yet one might argue that these intensities have little to do with his actual philosophical thinking and writing, or at least with what he chose to have published. After all, as he also wrote in _Culture and Value_ , \"My ideal is a certain coolness. A temple providing a setting for the passions without meddling with them\" (2e). Yet it would be striking were his official philosophical writing to be wholly uninformed by the otherwise deepest preoccupations of his character. Even in this remark, he presents _a certain coolness_ as an ideal, not as something that he has actually achieved, and he did note that he was \"someone who cannot quite do what he would like to be able to do\" (24e). Is _Philosophical Investigations_ in any sense _about_ the problem of the realization of talent against the grain of but always in relation to the affordances of culture and the ordinary, _about_ the problem of wedding autonomous selfhood to continuing sociality? That _Philosophical Investigations_ is about this, in detail, line by line, as well as being about the nature of meaning, understanding, the will, and so on, and about this _by_ being about these latter topics, is a main line of argument of my _Leading a Human Life: Wittgenstein, Intentionality, and Romanticism_. I cannot recapitulate the whole of that argument here. But I will offer a few brief pointers to it.\n\nIn section 125 of _Philosophical Investigations_ , we find that \"das philosophische Problem... ist... die b\u00fcrgerliche Stellung des Widerspruchs, oder seine Stellung in der b\u00fcrgerlichen Welt\"; in English, and appropriately, that \"the philosophical problem... is... the civil status of a contradiction, or its status in civil life.\" \"Our entanglement in our rules is what we want to understand\" (\u00a7125). This entanglement \"throws light on our concept of _meaning_ something\" (\u00a7125). What is it to be entangled in rules in civil life, in ordinary life, in the ordinary, civil _b\u00fcrgerlichen_ world? Not to \"know one's way about\" (\u00a7123) is not to know how to engage with this world, not to know how to bring one's talents and selfhood to authentic, nonderivative, and yet ratifiable expression within it. To ask \"what does this knowledge [of how to go on in applying a rule] consist in?\" (\u00a7148) is to ask what there is in me\u2014what talent, what locus of understanding, what source of mastery\u2014that enables me to go on and _how_ to bring this talent, locus, or source to apt expression. Something must be there in me. I can do something, and we are not in using language either machines or other animals. But what is it? And do I bring whatever it is to expression aptly? How? I seem caught between an anxiety that the only routes of expression are those already laid down in surrounding practice, that I contribute nothing, that I am ordinary, and hence nonexistent: call this the anxiety of expressibility, and an anxiety that I cannot express that whatever-it-is in the ordinary, that I am alone, and mad: call this the anxiety of inexpressibility. To be able to mean something, to understand something: these are the results of the mysterious engagement of spontaneity in me, the source of originality, with the routes of expression that are given in practice, as though a seed in me\u2014but one I can never identify or cultivate deliberately\u2014grew in relation to its environment. \"Each morning you have to break through the dead rubble afresh so as to reach the living warm seed. A new word is like a fresh seed sewn on the ground of the discussion\" (2e). It may well be that there is for the language user \"a special experience\" of understanding, but this special experience cannot be grasped and deliberately deployed independently of engagement with the affordances of culture. \"For us it is the _circumstances_ under which he had such an experience that justify him in saying in such a case that he understands, that he knows how to go on\" (\u00a7155). The always mysterious interaction of circumstances, that is, of the affordances of culture in providing routes of expression, with the powers of selfhood is something to be accepted, not explained in either a scientific or intellectualistic theory.\n\nWittgenstein too, like Goethe but unlike Werther, did not commit suicide. His last words, famously, were \"Tell them I've had a wonderful life.\" Like Goethe, he achieved through the act of writing, repeatedly and day to day, a kind of catharsis, _some_ distance or perspective on his anxieties as a subject and some sense of himself as having a social identity as a writer. Hence there is some point to thinking that the second voice of the _Investigations_ (if there are only two)\u2014the voice that rebukes the tendency to seek scientific or intellectualist explanations of our cognitive abilities and that recalls us instead to the ways of the ordinary\u2014 _is_ Wittgenstein's more mature voice. At the same time, however, the first voice\u2014the voice of temptation and of intensities of perfect explanation and attunement\u2014is his too, a voice he cannot quite give up, much as Goethe in his maturity would not give up intensities of infatuation and would still also identify himself with such intensities in Edward in _Elective Affinities_. The mature voice of the ordinary, the voice of survival, comes to the fore and is allowed the last word within a section, but always in continuing critical engagement with the voice of perfectly grounded and explained attunement, the voice of temptation.\n\nWittgenstein knew all this about himself. In 1931, in one of the remarks of _Culture and Value_ , he wrote: \"The delight I take in my thoughts is delight in my own strange life. Is this joy of living?\" (22e). It is hard to tell. There is, once again, an all too present threat of narcissism in self-delight in thinking, in 1931 as in 1774. But it is also self-defeating simply to accommodate to the ordinary as it stands, eschewing thinking about better possibilities of life. By 1931, the idea that joy might be found in the activity of thinking one's own thoughts appears more a modernist or late romantic question in the face of increasing social fractures than as a first-generation romantic prayer and conjecture. If human being is the kind of being that can call its own being into question, that can think about life otherwise, exactly how it is to do this and with what prospects of fuller common life are, in the twentieth century, far from clear. To ask this question, and to write out this asking, again and again, is one powerful, anxious modernist face of the continued courting of responsiveness and responsibility, of the continued courting of the life of a subject.\n4.\n\nAttention, Expressive Power, and Interest in Life\n\n_Wordsworth's \"Tintern Abbey\"_\n\nI\n\nIn the first sentence of section 1 of _The Birth of Tragedy_ , Nietzsche urges us to think about art in relation to life in a new way: \"We shall have gained much for the science of aesthetics, once we perceive not merely by logical inference, but with the immediate certainty of vision, that the continuous development of art is bound up with the _Apollinian_ and _Dionysian_ duality\u2014just as procreation depends on the duality of the sexes, involving perpetual strife with only periodically intervening reconciliations.\" This claim forces us to ask two sets of questions.\n\n(1) What is the vision of the development of art as bound up with this duality opposed to? That is, what other way of looking at art in relation to life are we being asked to give up? And how is what we are to see anew\u2014the continuous development of art\u2014like the development of humanity through procreation? The answers to these questions must involve the thought that the development of art does not come to an end just as the development of the human species does not come to an end: both developments are continuing and embodied in essentially varying particulars. This fact makes otiose, then, the idea that the nature of art could be adequately and usefully described in a definition that specifies necessary and sufficient conditions to which the ideal work fully conforms, for no ideal, perfect work is possible, any more than a single ideal human being is. If no ideal work is possible, then there will always be imperfections in any particular work and questions about exactly how and how well it approximates the ideal. An ideal definition will be insufficient to settle such questions, and judgment and discernment will be called for. Just as women and men with their differences and peculiarities\u2014and not any perfect single human being\u2014produce further women and men with their differences and peculiarities, so works of art are produced by temporary couplings of two forces in us\u2014the Apollinian and the Dionysian\u2014that are never fully integrated and balanced to form a single perfect whole. To _see_ this fate\u2014a continuing failure to achieve the ideal and to overcome all difference, peculiarity, and opposition\u2014 _in_ every work of art is then to be weaned from the pursuit of a standing philosophical _logos_ or definition: _that_ pursuit functions only to deny the movement of life. When Plato, for example, assigned poetry \"the rank of _ancilla_ \" in relation to philosophy and its definitions, favoring only the \"enhanced Aesopian fable\" with a moral amenable to rational justification, he thereby shied away from both the genuine complex and disturbing powers of art and the genuine turbulence of life. With Plato, Nietzsche tells us, \"the Apollinian tendency has withdrawn into the cocoon of logical schematism\" and so given up on life, transformation, and development. Nietzsche, in contrast, is asking us to look the ongoing turbulence of life full in the face: to _see_ it _at work in_ every work of art and every human life.\n\n(2) But then how is a genuinely successful work of art possible at all? It must involve a creative coupling of the Apollinian and Dionysian forces or tendencies in us that figure in artistic making, yet success in this coupling is not assessable according to any fixed ideal or conceptual measuring stick. The various products and values that are, according to Nietzsche, typical of each of these tendencies can be roughly set out as follows:\n\nAPOLLINIAN | DIONYSIAN\n\n---|---\n\nDreams | Intoxications\n\nCreated supplement to life | Chaotic essence of life\n\nSculpture | Music\n\nForm-order | Passion, drive, content\n\nEnjoyable illusions | Fusion in \"feeling-with\"\n\nComposure-trance-absorption | _Ekstasis_\n\nCulture, civilization | Nature\n\nUpholding of _principium individuationis_ | Collapse of _principium individuationis_\n\nIn _The Will to Power_ , Nietzsche describes these tendencies further and repeats the analogy between artistic production and sexual reproduction.\n\nThe word \"Dionysian\" means: an urge to unity, a reaching out beyond personality, the everyday, society, reality, across the abyss of transitoriness: a passionate overflowing into darker, fuller, more floating states...\n\nThe word \"Apollinian\" means: the urge to perfect self-sufficiency, to the typical \"individual,\" to all that simplifies, distinguishes, makes strong, clear, unambiguous, typical: freedom under law.\n\nThe further development of art is necessarily tied to these two natural artistic powers as the further development of man is to that between the sexes. Plenitude of power and moderation.\n\nThese powers must then be jointly expressed in the successful work. This expression must involve something other than either simple fusion, which would leave these powers unrecognizable in their individuality and hence unexpressed, and simple juxtaposition, in which there would be no coupling, no productive interrelation. But then how is success in the joint expression of these powers possible at all?\n\nNietzsche offers an answer not in the form of a definition or principle but in the invocation of an example\u2014an example that remains central for him throughout his subsequent career\u2014at the end of the first paragraph of _The Birth of Tragedy_ : \"this coupling ultimately generated an equally Dionysian and Apollinian form of art\u2014Attic tragedy.\" So how, then, did Attic tragedy succeed in expressing these forces in an exemplary way, albeit one that is not successfully imitable according to a rule? The key to answering this question lies in seeing that the Attic Greeks courageously accepted the chaotic onwardness of meaningless, self-proliferating biohistorical life and then formed coherent, recognizable, individual lives anyway. \"The profound Hellene\"\u2014both certain central figures in Greek tragedies and the members of chorus and audience who see and respond to their actions\u2014\"uniquely susceptible to the tenderest and deepest suffering, comforts himself, having looked boldly right into the terrible destructiveness of so-called world history as well as the cruelty of nature.\"\n\nIn accepting clearly the destructiveness and cruelty of human life in nature and history, the Hellene resembles Hamlet and, in turn, us\u2014we for whom the consolations of a superintending logos or Providence story are gone.\n\nThe Dionysian man resembles Hamlet: both have once looked truly into the essence of things, and they have _gained knowledge_ , and nausea inhibits action; for their action could not change anything in the eternal nature of things; they feel it to be ridiculous or humiliating that they should be asked to set right a world that is out of joint.... An insight into the horrible truth outweighs any motive for action, both in Hamlet and the Dionysian man.\n\nFor one who bears such an insight, it is unclear what, if anything, is to be done, in any way that matters. \"Now no comfort avails any more; longing transcends a world after death, even the gods; existence is negated along with its glittering reflection in the gods or in an immortal beyond. Conscious of the truth he has once seen, man now sees everywhere only the horror or absurdity of existence.\"\n\nAnd yet, somehow, the Hellene nonetheless \"comforts himself.\" This comfort is achieved in two distinct ways. \"The comic\" enables \"the artistic discharge of the nausea of absurdity.\" That is, there is a kind of purging of nausea in the Dionysian self-abandonment of laughter. And, second, there is \"the sublime as the artistic taming of the horrible.\" This artistic taming happens through the setting up of a fiction within which it is possible for a life within the chaos of nature nonetheless to take on coherent form. \"The Greek built up the scaffolding of a fictitious _natural state_ and on it placed fictitious _natural beings_.\" The sublimity that attaches to these fictitious natural beings\u2014the protagonists of tragic drama\u2014is that they stand out in their coherence of personality, diction, thought, and action against the chaos of nature. Their lives have form. The imposition of artificial emplotment on and for these fictitious natural beings in this fictitious natural state requires the dramatist to \"dispense from the beginning with a painstaking portrayal of reality,\" with all its meaningless incidents. What is presented is rather an account of how a protagonist intelligibly moves toward his fate in his circumstances, with \"probability or necessity,\" as Aristotle says. The _hexis_ , or character, of the protagonist, with its one-sidedness ( _hamartia_ ) or excess of virtue that is ill-fit to the circumstances of action, intelligibly brings it about that a reversal ( _peripeteia_ ) occurs, accompanied with recognition ( _anagnoresis_ ) by the protagonist, the chorus, and the spectators of the intelligibility of the action. Yet though it is a fiction, this world in which protagonists coherently have characters and reach their fates intelligibly \"is no arbitrary world placed by whim between heaven and earth; rather it is a world with the same reality and credibility that Olympus with its inhabitants possessed for the believing Hellene.\" This world is set up \"for [the] chorus\" and for the audience whose responses it shapes and models. In this world, chorus and audience see that a character\u2014a protagonist with a _hexis_ or unified ensemble of powers of thought, reasoning, expression, and action\u2014can impress that _hexis_ on the world by expressing it in intelligible action, however ill-starred the outcome. In this way, the protagonist \"lives anyway\" for the chorus and audience, despite the meaningless of life \"in itself\" in nature and history \"in themselves.\" Antigone and Oedipus, and Hamlet and Lear, are figures of sublime accomplishment, Nietzsche is arguing, in standing out for us intelligibly from the chaos of life in the coherence and power of their thought, diction, and action. They have lived as subjects of their lives, experiences, and actions rather than as mere things, in a way that is both exemplary and comforting for us.\n\nII\n\nOn the surface, Wordsworth's tone is far more optimistic than Nietzsche's. The universe itself, he writes, \"moves with light and life informed, \/ Actual, divine, and true,\" and we may find Paradise to be \"A simple produce of the common day.\" Yet such displays of felt metaphysical confidence are never either self-standing or stable. They are surrounded by narratives that describe recurrent movements through despair and recovery, and they are strongly qualified by being cast in the subjunctive mood or as expressions of hopes about future reception. Wordsworth typically _conjectures at a moment_ both that he has so experienced nature and the human world in it and that others _may_ experience them similarly, thus sanctifying his prophetic authority in matters of culture and value. His major works conclude more typically with an expression of a hard won, prayerful hope that his vision will or may be taken up than with a confident pronouncement that that vision is true and proven. Put otherwise, his major poems are more records of experiences of thinking and feeling through which poetic identity and authority are _temporarily_ achieved than they are pieces of straight metaphysical philosophy.\n\nThe underlying problem that motivates Wordsworth's continual swervings among expression of feeling, metaphysical pronouncement, conjectures about reception, and recurrent hesitancy and doubt is that of achieving life as a fully responsive and responsible human subject. Wordsworth seeks both to become a locus of feeling coupled with apt understanding and to find that achievement of fullness of subjectivity certified by others. Absent such certification, the achievement itself is open to doubt. For Wordsworth, the very idea that he has lived or can live _as a subject_ is always threatening to falter, most memorably in the image of himself as one \"Unprofitably traveling toward the grave, \/ Like a false steward who hath much received \/ And renders nothing back\" that launches the _Prelude_ on its course of self-interrogation. When Wordsworth does at certain moments achieve a measure of confidence in his life and powers as a subject, he does so much more in the manner of the protagonists of tragic drama as Nietzsche understood them than in the manner of a theoretical philosopher. He manages, that is, recovery of himself _in time_ through achieving a stably and powerfully enough formed manner of thought, expression, and (writerly) action, at least for a moment, in the face of the chaos of life, rather than simply reverting to metaphysical pantheism or any other epistemically well-founded doctrine or doxa. Fullness of responsiveness and responsibility are won, essentially at a moment, plausibly and in an exemplary way, thus overcoming passivity, drift, and despair for a time.\n\nThat there is a threat to the existence of life as a subject\u2014a threat that dominates a great deal of contemporary life, but a threat that may be answered by the powers of poetry (not theoretical philosophy)\u2014is the chief argument of the \"Preface\" to _Lyrical Ballads_. This argument is inaugurated as Wordsworth announces that \"the principal object... proposed in these poems\" is \"above all, to make these incidents [and situations from common life] interesting,\" thus implying that common life is not interesting as it stands: we are dead to it and it to us. This implication is unpacked in the further thought that \"a multitude of causes, unknown to former times, are now acting with a combined force to blunt the discriminating powers of the mind and, unfitting it for all voluntary exertion, to reduce it to a state of almost savage torpor\" (449). Without the exercise of discriminating powers issuing in voluntary exertion\u2014without fullness of responsiveness and responsibility\u2014there is only more or less animal passivity in life, as one is buffeted about by circumstance, often compelled addictively to try to stop the pain of life by succumbing to a \"degrading thirst after outrageous stimulation\" (449). One fails to lead the life of an active subject moved by genuine interest.\n\nYet there is no hope that pure reasoning can save us. Wordsworth specifically eschews \"the selfish and foolish hope of _reasoning_ [the Reader] into approbation of these particular poems\" insofar as it is not possible \"to give a full account of the present state of public taste in this country, and to determine how far this taste is healthy or depraved\" (445). That is to say, there is no account vouchsafed to us by reason or by anything else that determines what any ideal human life must be like, which account could serve as a standing measure of present life and taste. For that, we would need to know \"in what manner language and the human mind act and re-act on each other\" (446) in general and how \"society\" in its \"revolutions\" might play out these interactions well or badly. Such knowledge of standing conditions that would determine what counts as an ideal human life, fitly expressing distinctively human powers, is unavailable. We are too finite for that, with our reflections on our condition too shaped by our specific, sectarian particulars of personal and social history and place. Embodiment and the exercise of intelligence within it has, always, its localities.\n\nTo undertake to make the scenes and incidents of common life interesting but without any external measure of interest is then to aim to do the work of animation from within the having of ordinary experience. As Stanley Cavell puts it,\n\nWhat the words \"make interesting\" say is that poetry is to make something happen\u2014in a certain way\u2014to the one to whom it speaks; something inside, if you like. That what is to happen to that one is that he or she is to become interested in something [is]... to perceive us as [at present] uninterested, in a condition of boredom, which [is regarded as], among other things, a sign of intellectual suicide.\n\nOne must begin _from_ the experience of common scenes and incidents, together with attendant thoughts and feelings. Then _within_ courses of thoughts and feelings that are often clich\u00e9d, inattentive, or unanimated, one must _discover_ or _uncover_ those that are aptly attentive to the subject matter, so that one becomes animated as a subject in dwelling in just these aptly attentive thoughts and feelings. In this way one might hope, without an external standard, to \"discover what is really important to men\" so that \"the understanding of the Reader must necessarily be in some degree enlightened, and his affections strengthened and purified\" (448). The poet here acts as a kind of bootstrapping device for the achievement of animation from within ordinary experience that is otherwise dead, unattended to, and insignificant for us. It is for this reason that \"the feeling therein developed gives importance to the action and situation, and not the action and situation to the feeling\" (448). The proper work of poetry is not simply the depictive presentation of a subject matter but rather the working through of feeling in relation to a subject, so that genuineness of feeling is achieved. The poet here arrives at the aptness and fullness of response that must animate the life of a subject, if the subject is to find anything interesting at all. The poet \"considers [man] as looking upon this complex scene of ideas and sensations, and finding every where objects that immediately excite in him sympathies which, from the necessities of his nature, are accompanied by an overbalance of enjoyment\" (455). Here the enjoyment is no simple wash of sensory pleasure; it is rather a lingering in feeling as apt to the object of attention. Even when the scene attended to is horrible, one may have the sense that here, apart from the ordinary rush of hectic and inattentive life in which we are mostly caught up, one is feeling and responding fully and aptly, as an active subject, not a thing.\n\nIn this work of the animation of the life of a subject, the use of the self is crucial. Wordsworth notes that the poet's \"own feelings are his stay and support; and, if he set them aside in one instance, he may be induced to repeat this act till his mind shall lose all confidence in itself and become utterly debilitated\" (461\u2013462). There is nothing to go on in beginning to aim at genuineness of feeling other than feeling as it already stands; nothing to go on to aim at fullness of life as a subject other than that life as it already exists, debilitated as it may be. One must find confidence in one's feelings from within them, even if they are at first cloudy and confused. Wordsworth tells us that he has \"at all times endeavoured to look steadily at [his] subject\" (450), where this effort at steadiness of looking includes a focus not only on the scene or incident at hand but also on himself as either debilitated or apt in his own course of feeling. The poet must ask himself: do my feeling and attention wander off into unsteadiness, absentmindedness, or unresponsive clich\u00e9, in relation either to the scene that initiates reflection or to the work of reflection on it? Am I, the poet asks, genuinely paying attention to the scene and to the work of reflection, in aptness of both thought and feeling? If, as may sometimes happen, the answer is yes, then the poet will be \"a man pleased with his own passions and volitions, and who rejoices more than other men in the spirit of life that is in him\" (453).\n\nThere is a considerable inherent risk of narcissism or of excessive self-satisfaction in taking upon oneself the role of the poet as a figure of exemplarity in thought and feeling. One might become _too_ pleased too quickly in one's aptness of thought and feeling. This inherent risk is an aspect of the internalization of quest romance, of finding oneself to bear a problem of having to find routes of significant action from within one's own resources rather than from culture as it stands. This problem becomes increasingly pressing and difficult throughout the development of modernity, as diversity increases and fewer feelings and commitments are shared, absent willed fundamentalism.\n\nFor Wordsworth, awareness of the risk of narcissism continually haunts the work of the stabilization of attention and the work of writing. Doubts about whether one is genuinely thinking and feeling _as a subject_ are inherent to the activity of seeking exemplarity in thought, feeling, and their expression. But if this risk is overcome and exemplarity is achieved, then the poet may arrive at \"truth... carried alive into the heart by passion; truth which is its own testimony, which gives competence and confidence to the tribunal to which it appeals, and receives them from the same tribunal\" (454). No other tribunal of aptness of feeling will serve. Instinct, tradition, fixities of form and craft, and proofs constructed by reason\u2014these are all either unavailable or impotent to sanction the work of the achievement of aptness and genuineness of feeling and its expression.\n\nWithout this work\u2014if we shy from it in anxiety, or under the conditions of modern economic life, or in simple distractedness of mind, or in reversion to what is comfortable enough as we all mostly do much of the time\u2014subjects do not exist _as subjects_ , as those who take _an interest in_ their own experiences. They fail to live according to \"the grand elementary principle of pleasure [in apt, genuine, and stable feeling], by which [man] knows, and feels, and lives, and moves\" (455). As Wordsworth notes, no advances in science will make this work irrelevant. \"If the labours of Men of science should ever create any material revolution, direct or indirect, in our condition, and in the impressions which we habitually receive, the Poet will sleep then no more than at present\" (456). The work of the animation of life for us as subjects, the work of finding felt significance in scenes and incidents of common life, will continue to be necessary, no matter how the scenes and incidents of life themselves may change.\n\nIn this conception then, \"the calling of poetry,\" as Cavell puts it, \"is to give the world back, to bring it back, as to life.\" The perception of the continuing need for this calling can be set out in a rough argument schema as follows:\n\n(1) A person lives as a subject in a world of significance if and only if that person lives with attentive wholeheartedness, felt interest, and commitment in relation to objects of common experience.\n\n(2) Mostly we do not live with attentive wholeheartedness, felt interest, and commitment in relation to objects of common experience.\n\nTherefore\n\n(3) Mostly we do not live as subjects in a world of significance.\n\nOne can of course reject the first premise. One might, in particular, wonder what sense can be made of the phrase \"lives as a subject.\" Isn't it enough for that just to live\u2014to be biologically alive\u2014and simply to be a subject, that is, simply, say, to speak a language and to be aware of oneself as speaking it? Why should attentive wholeheartedness, felt interest, and commitment in relation to objects of common experience matter ontologically, as it were?\n\nWhen, however, one is in the grip of the truth of the second premise, then the first premise seems all but inescapable, or at least Wordsworth in his thoughts about poetry registers a sense that it is for him inescapable. The issue is less naturalistically ontological than it is ethically ontological. We fail to exercise, or to exercise fully, defining powers that we possess and that we ought to exercise. Something is wrong with our present life, and that is just what premise 2 says in a specific way. When the thought that something in life is wrong is present for us, coupled with the thought that we can and should do better\u2014should do something to remedy that wrong\u2014then we are very close to accepting premise 1 as the expression of an ethical demand that (according to premise 2) we are failing to meet. The conclusion that we have arrived at a kind of ethical death-in-life is then itself all but inescapable. Mostly we do not live as subjects in a world of significance.\n\nPoetry then seeks to overcome this conclusion by undertaking to reanimate our wholeheartedness, interest, and commitment in our lives and world from within the broken, half-hearted feelings we already have. No acquisition simply of information about the world will serve, for what is sought is significance in feeling, not an addition to a collection of facts. Nor will any life-denying escape from the world, say to a heaven of Platonic forms, avail us in coming to terms with this life. Instead, what Cavell more or less terms getting the hang of a posture\u2014perhaps from reflection on feeling as it stands, perhaps also from picking up a precursor's routes of interest and expression\u2014is what is called for. \"You never know when someone will learn the posture, as for themselves, that will make sense of a field of movement, it may be writing, or dancing, or passing a ball, or sitting at a keyboard, or free associating. [A] sense of paradox expresses our not understanding how such learning happens.\"\n\nWhat is needed, then, is what Wordsworth cryptically calls the ability, possessed most typically by rural men, to \"communicate with the best objects\" (447). To communicate _with_ objects (including persons and events) is not to communicate _about_ them to others. It is rather to arrive at a communion or intimacy with them, or a finding in feeling that one shares with them a life of significance. Only through such communion in meaningfulness is the life of a subject stabilized in the exercise of human powers.\n\nDavid Wellbery has characterized this arrival at stabilization of the life of a subject as \"the specular moment\": \"a perfect (and wordless) reciprocity between two selves\" or between a subject and a scene, object, or incident experienced as self-like. Such a wordless reciprocity is required to lift one out of circuits of decayed conventionality, exemplified in uses of language that are thoughtless, inattentive, or unfelt. Despite their saving graces, such specular moments are, Wellbery suggests, both sociohistorically contingent and ultimately uncapturable. The need for and reversion to such moments arises typically or at least with special intensity in modernity, when other sources of stabilization are lacking, and typically or with special intensity for male subjects, caught up in routines of conventionalized work that they find meaningless and sealed off from an intimacy with nature that is stereotypically coded as feminine. Both women and nature are hence frequently forced to function as props for the male pursuit of the specular moment that is to stabilize anxious male selfhood. Even more troubling is the fact that specular moments themselves are transitory and subject to dispersion as soon as they become objects of explicit discursive awareness. To attend _to_ them and to try to articulate their significance is to destroy them. As Wellbery puts it, \"to render the specular moment in language is to submit it to an articulatory dismemberment and temporal deferral that fracture its essential unity.\"\n\nAs a result, there is no possibility of arriving at a specular moment that is lasting, that possesses explicit, articulated significance, and that is innocent of a self-centered use of its object. But then there is no stabilization of the life of a subject without such moments either. The best, therefore, that one can do in seeking to certify that one is a (more) fully human subject as a locus of apt feeling, attention, and reflective-discursive awareness is to move through or in and out of such moments. A narrative of such movements will trace, always, an itinerary of both achievement and loss. The subject is undermined by temporality, discursiveness, and self-centeredness in the very moment of arrival at an evanescent stability and power. Any teachings that may be derived from such a moment will be conjectural and subject to immediate doubt. Yet there is, again, no other route to the perfection and stabilization of felt responsiveness to life out of conditions of empty materiality and ossified conventionality in which distinctively human powers are mostly betrayed.\n\nWordsworth, somehow, knows all this. His moments of strongest self-stabilization as a distinctive and exemplary human subject, apt in responsiveness to life and in thus achieving the life of a subject, are at the same time immediately subjected to doubt, and his itinerary of self-constitution fails to reach a fully stable end. One can hear a distinctive, honest, Wordsworthian hesitation even in the moments of most forceful conclusion. In the _Prelude_ , Wordsworth announces that he and Coleridge will be\n\nRich in true happiness if allowed to be\n\nFaithful alike in forwarding a day\n\nOf firmer trust, joint labourers in the work\n\n(Should Providence such grace to us vouchsafe)\n\nOf their deliverance, surely yet to come.\n\n\" _If_ allowed to be faithful... _should_ Providence such grace to us vouchsafe\"\u2014it may not happen; the ways of the world as they stand may be too strong. Though the deliverance of humanity is \"surely yet to come,\" the \"surely\" hints at an effort here too at self-reassurance. It may not come: Wordsworth did not write \"Of their happy deliverance yet to come,\" which would scan about as well. In the \"Preface,\" Wordsworth tells us that when he thinks about the ability of his and Coleridge's poetic writing to have any significant effect under the present degraded conditions of life,\n\nI should be oppressed with no dishonourable melancholy, had I not a deep impression of certain inherent and indestructible powers of the human mind, and likewise of certain powers in the great and permanent objects that act upon it, which are equally inherent and indestructible; and were there not added to this impression a belief, that the time is approaching when the evil will be systematically opposed, by men of greater powers, and with far more distinguished success.\n\n(449)\n\nFor all the confidence that it expresses, this passage also says that melancholy may not be dishonorable. The only blocks against it are certain impressions, but where are they to be found, how lasting can they be, and what joint work will they support, under present conditions? What if the work of attention and feeling that these poems invite is not in fact taken up by others?\n\nYet Wordsworth does not embrace melancholy, nor does he simply acquiesce in present conditions. Instead he goes on in the endless and endlessly self-scrutinizing work of pursuing life-enabling specular moments, articulating them (and thus betraying them), and coming to terms with human life in time, with all its movements of both self-stabilization and self-undoing.\n\nDavid Miall has usefully called attention to the difference between typical locodescriptive poetry of the picturesque and Wordsworth's writing about his experience of nature. Where locodescriptive poetry focuses on what is seen, Wordsworth in contrast distinctively describes not so much what he sees as himself in the process of seeing. Miall develops this point by commenting on a fragment from Wordsworth's Alfoxden Notebook.\n\nTo gaze\n\nOn that green hill and on those scattered trees\n\nAnd feel a pleasant consciousness of life\n\nIn the [?impression] of that loveliness\n\nUntill the sweet sensation called the mind\n\nInto itself by image from without\n\nUnvisited: and all her reflex powers\n\nWrapp'd in a still dream forgetfulness\n\nI lived without the knowledge that I lived\n\nThen by those beauteous forms brought back again\n\nTo lose myself again as if my life\n\nDid ebb & flow with a strange mystery.\n\nAs Miall notes, traditional readings of Wordsworth would focus here only on the moment of restoration, on how \"the sweet sensation\" of the natural scene \"called the mind into itself.\" More recent New Historicist readings would argue that \"the vision of unmediated benefit from Nature that the poem famously provides is, in this view, only a screen on which Wordsworth projects his anxieties.\" But Miall calls our attention instead to the process, jointly of attention and of composition (of the poem and of the human subject), on display here. Wordsworth knows, and says, that his life \"did ebb & flow with a strange mystery.\" Moments of (recuperative) self-loss are crossed with moments of (discursive) self-awareness; the movement between these moments is all.\n\nPerhaps Stanley Cavell had something like this fact about Wordsworth (and Coleridge and Emerson and Thoreau) in mind when he remarked that \"Romantics are brave in noting the possibility... of what you might call death-in-life. My favorite romantics are the ones (I think the bravest ones) who do not attempt to escape these conditions by taking revenge on existence. But this means willing to continue to be born, to be natal, hence mortal.\" \"To continue to be born\" means here to eschew fundamentalisms involving the submission of the self to something apart from earthly life (Platonic forms or sacred texts, as may be) but also to eschew mere acquiescence and accommodation to a life of conventionalized getting and spending. Instead, movement both into and out of (Apollinian) moments of articulation and (Dionysian) moments of recuperative self-undoing is what is proper to the life of subject.\n\nIII\n\nNo matter what their theoretical desirability as both conditions of and contributions to life, whether such movements are possible\u2014whether romantic bravery is possible\u2014and what sort of closure or conclusiveness (without denying temporality) such movements might achieve are no small questions. That such movements are possible, and that measures of human closure and composure are available, within time and without denying life, are, I suggest, the central showings of \"Tintern Abbey.\" Rather than either a document that (only, merely) traces the saving influence of nature or that (only, merely) shows a consciousness always in anxiety about its reception and unable to compose itself, \"Tintern Abbey\" shows a consciousness achieving a measure of composure in time, without intellectual certainties. It points to and exemplifies a path between dogmatism and nomadism, intellectual and moral alike. It can help us to hear \"Tintern Abbey\"'s showings if we divide its progress into eight rough stages of subject matter (thought) and attitude.\n\n(1) \"Five years have past; five summers, with the length \/ Of five long winters!\" (1\u20132) [lines 1\u201322]. With these opening lines, the question of the meaning of life in time is raised. The apposition of the more subjective \"five summers, with the length of five long winters!\" (with subjectivity registered in both the felt succession of the seasons and in the sense that the winters have been long) to the more objective \"five years\" asks, already, what has the passing of these years meant in the life of a subject, in my life? The \"and again I hear\" that immediately follows introduces explicitly the I who is the subject of these reflections, an I that is wondering what its experience has meant. It has been suggested that the underlying subject and cause of this experience of questioning the meaning of some times of one's life are Wordsworth's guilt and anxiety over his evasion of the English military draft and over his affair and child with Annette Vallon. There may be some truth in this suggestion. The complex act of writing \"Tintern Abbey\" may well have occasioning circumstances that are rooted in the poet's past and that lie well outside the present scene alone. But whatever the occasioning circumstances may be, there is for us a question about whether the poem does any productive work in questioning the meaning of life in time. Any life will contain enough missteps and occasions for guilt and regret to prompt the raising of this question at some point. When one is in the grip of the thought that one is mostly living in half-heartedness and so failing to live as a full subject, then the question will be both natural and forward looking. What might I, or we, do in order to live with more wholeheartedness? Given the naturalness of this question at some point in any human life, it may repay our efforts if we attend to the work of the poem in its attempts to come to terms with it, without reversion to fundamentalism and without escapism. Perhaps these attempts are not even wholly successful. But we shall scarcely be able to see that before we engage with the work that the poem undertakes.\n\nIt has also been suggested that the move into the register of subjectivity, into the questioning of the meaning of a single subject's experience, more generally enacts a flight from the political. This suggestion too may well bear some truth. But whether it does depends in large measure on what sort of politics we have in mind. Here we should not shrink from engaging with \"Tintern Abbey\"'s effort to find or found its own politics, that is, its own vision of a fuller, more human life in practice and in time. That vision does not have the shape of urging us toward either electoral politics or class struggle. But it is surely in part a vision of a better, more human polis. The topic, after all, is what it would be to lead the life of a subject, with others, under present conditions and in time.\n\nTo say \"and again I hear these waters\" after \"five summers, with the length \/ Of five long winters\" is to raise the question of repetition. Are we fated to it? And what does it mean that one finds oneself again stopped or halted in a place, in a moment of reflection? That death is all but explicitly on the poet's mind as he raises the question of the significance of life is suggested in the thought that \"The day is come when I again repose \/ Here, under this dark sycamore\" (9\u201310). \"Repose\" is stronger, more suggestive of permanence, than \"recline,\" and \"dark\" suggests that one is in an enduring shade. To repose under this dark sycamore is almost to be under the ground. Though the poet goes on immediately to note that he views the present scene (and so is not dead, or not yet), nonetheless the thought that the day of a final repose will come is somewhere active in his consciousness of his place in relation to the scene before him. The thought of death is further echoed in the line \"The Hermit sits alone\" that, set apart by a line break, concludes this subject. With this Hermit, already a creature of the poet's imagination, not of perception alone, and with his isolation\u2014with his lack of audience and companions, and so perhaps his insignificance\u2014the poet may be taken to feel more than a little identification. With such apartness, what life? What doth life in time avail?\n\nConsciousness of temporality is then further registered, both semantically and in internal citation, in the two instances of \"Once again\" (4, 14) that introduce the sentences surrounding the thought about repose. What has happened in the five years before these once agains, and what, if anything, does what has happened mean? What has filled and what should fill the passing of time? In raising this question, the poet does not turn for either knowledge or salvation to any exterior entities. Instead, as Miall cannily argues, he observes himself observing, and he invites us into his own jointly perceptive and apperceptive processes of consciousness of the present scene. In particular, in commenting on the phrase \"hedgerows, hardly hedgerows\" (15), Miall notes that\n\nin the order of his phrases he recreates the _process_ of observation: conventional, or schematic expectation would first look for hedgerows and find them; yet, a second glance\u2014\"hardly hedgerows\"\u2014would show the hedges in fact to be running wild. These lines thus invite the reader to replicate Wordsworth's own process of observation, a feature of several other elements in the opening paragraph. An object (\"plots of cottage ground\"; \"pastoral farms\") is first named, as an objective component of the scene, or what is to be expected in such a location (perhaps what was remembered from 1793); but it is then qualified in ways that suggest a second more careful focus on the actual details before him.\n\nThe focus then is on how apperceptive awareness\u2014awareness of oneself as observing and thinking in relation to this scene\u2014may develop itself and on what assurances of possibilities of meaningfulness it may discover. This is, as Miall further notes, in strong contrast to typical locodescriptive poetry that takes as its central subject the scene itself rather than the poet's awareness of his processes of awareness of the scene.\n\n(2) \"Such, perhaps, \/ As have no slight or trivial influence...\" (31\u20132) [22\u201349]. Lines 22\u201349 have seemed to many readers to be one of two grand metaphysical centers of the poem. (The other is lines 85\u2013111.) Within these lines the poet all but asserts that he has seen \"into the life of things\" (49) in such a way that \"the heavy and the weary weight \/ Of all this unintelligible world, \/ Is lightened\" (39\u201341). Yet it is crucial that the poet does not in fact simply assert these claims. The passage begins with the claim that he has in the past at times remembered these forms, first experienced five years ago. His characterization of the feelings he has had as a result of these rememberings is strongly qualified. Some of them may be \"unremembered\" (31)\u2014an acknowledgment that the report of feelings had in the past may be as much present construction (perhaps driven by need and anxiety) as recovery of an actual past. Such feelings \"perhaps\" have \"no slight or trivial influence\" on our lives, but then perhaps their influence is only slight and trivial. To what \"little, nameless, unremembered acts \/ Of kindness and of love\" (34\u201335) have they led, and in what way? The poet arrives at no definite account in answer to this question, and his formulation concedes its relevance\u2014to him and to us. How can and should a life be constructed on the basis of feelings that may be unremembered and perhaps of slight influence? While there is a move _toward_ the redemption of life in time in the mention of these feelings and their influences, that move is considerably less than definite and conclusive. The gift of insight \"into the life of things\" provided by these feelings is also conjectural. \"Nor less, I trust, \/ To them I may have owed another gift, \/ Of aspect more sublime\" (35\u201337)\u2014\"trust\" not \"say,\" \"may\" not \"thus.\" Has this gift actually come, and if so, through what processes? With such gifts of such a provenance, what salvation? The very movement toward saving certainties remains a lingering in uncertainties. Apperceptive awareness of oneself as thinking and feeling (rather than full immersion in sense-experience without reflection) includes, always, the thought that one is in part constructing the experience, not simply taking in any salvific given. And with awareness of constructedness comes present doubt. Or, at any rate, all this\u2014perception plus conjecture plus intimately present doubt, in sharp contrast to simple intake and assertion\u2014lies within the process of this poetic subject's attendings.\n\n(3) \"If this \/ Be but a vain belief...\" (49\u201350) [lines 49\u201365]. Given the nature of the doubts expressed in the second section, \"vain\" has here the force of both \"empty\" and \"ego-centered\": perhaps the beliefs about the gifts presented by past feeling are empty _because_ merely constructed in fantasy by a needy subject. \"The fretful stir \/ Unprofitable, and the fever of the world, \/ Have hung upon the beatings of my heart\" (52\u201354). The feelings that result may be compensatory but vain reactions to this stir and fever; they may be empty, because untrustworthy to others, and so to oneself, in the face of the ways of the world. That there is a risk of escapism in turning to them is part of the poet's own movement of thought. Human powers are not exercised amid the stir unprofitable of the world\u2014mostly we do not lead the lives of subjects in a world of significance\u2014and if not there, then where might they be? The twice pronounced \"How oft(en)\" (50, 57) \"have I turned to thee, \/ O sylvan Wye\" (55\u201356) is _both_ an exclamation of gratitude to the Wye for being restoratively there to turn to _and_ a genuine question: how often have I forgotten the Wye? How often have I been myself caught up in the ways of the world? There is, after all, not much chance of a life with people apart from these ways. A life wholly apart would itself be the life of a Hermit, itself too a life of all but death, without reciprocity, intimacy, or recognition, though perhaps with some rough ontological power of persistence and survival, whatever the ways of the world may be. If \"the picture of the mind revives again\" (61) \"with pleasing thoughts \/ That in this moment there is life and food \/ For future years\" (63\u201365), there remains nonetheless a question about how much this food will be genuinely present and available. It has not, not always, been present in the past, as the world has had its ways, and the deliverances of feeling may, again, be vain. So \"here I stand\" (62). Is my life justified: before myself, before others, or before God? What is the meaning of life in time? What exercises of what human powers might lend significance to a life in time? There is survival, to be sure: \"here I stand\" again, after \"five years have passed,\" so that whatever death-in-life I have succumbed to, that succumbing has not been complete. There remain in me, at least latently, human powers of feeling and reflection that might issue in expressive action, and those powers are to some extent activated by the present scene: hence the thought that life and food for future years are to be found in them. Animation of and within the life of a subject is felt to be possible. But whither doth it point? A specular moment of recovery, submitted to apperceptive reflection and to expression in language, is immediately undone by these submissions. And they are unavoidable. Wordsworth at his strongest characteristically mixes a sense of survival and restoration with doubt and uncertainty, and this standing here with pleasing thoughts that may yet be vain and evanescent (\"how oft?\") shows Wordsworth at his strongest.\n\n(4) \"And so I dare to hope...\" (65) [lines 65\u201385]. Given his unresolved uncertainties, the poet's \"dare to hope\" is more apt than \"claim to know.\" Daring and hope indicate willed resolution backed by nothing more than the fact of survival as a subject who does _not_ know the best exercises of his human powers but has survived anyway. The object of the infinitive \"hope\" is not specified, unless we read the \"so\" as \"in this manner\" or \"thusly\" (rather than \"therefore\"), so that it refers back to the \"there is life and food \/ For future years\" of lines 64\u201365. \"To dare to hope\" is, again, far from a confident assertion of a truth. The poet dares to hope, uncertainly, that he stands, in a moment and with pleasing thoughts, but these deliverances of apperception are themselves less than apodictic. In this resolution to hope there may be some survival of a subject, but a subject whose substantial nature remains unknown to him as any kind of stable thing. Nor does he know what is commanded of him by his nature as a subject, what a fit life for a human subject would be. At best, the stance is that of book 2 of the _Prelude_ : \"I was left alone \/ Seeking the visible world, nor knowing why. \/ The props of my affections were removed, \/ And yet the building stood, as if sustained \/ By its own spirit!\" Neither parent nor nature nor the visible world is present to provide props or reassuring grounds for affections. Hence the affections\u2014the commitments and passions that motivate action\u2014may be called into question. Their source and their value remain mysterious. And the subject stands only \"as if sustained \/ By its own spirit!\" Given the predominant effort to find or found a better standing for the subject, one may well wonder how much sustenance, how much food for future years, is really to be found in this survival.\n\nEven the survival itself is immediately called into question. In the very moment of standing and daring to hope, the poet acknowledges that he is \"changed, no doubt, from what I was\" (66). There is no persistent substantial something to which the willed survival of the subject is referred. The subject has emerged within time through a fall or procession into discursive consciousness and apperceptive awareness. Somehow\u2014he knows not how\u2014he has come to be aware of himself as thinking, feeling, and judging, with less naturalness and automatism than attach to the life of a prereflective subject. He is no longer \"like a roe \/ [Who] bounded o'er the mountains, by the sides \/ Of the deep rivers, and the lonely streams \/ Wherever nature led\" (67\u201370).\n\nWith the fall out of naturalness in affection and activity comes a pressing need for reassurance or grounding. The mysterious joint onsets of discursive consciousness and apperceptive awareness bring a sense of distance from activity and of consequent anxiety. Before these onsets, one does not even seek \"the thing [one] loves\" (72), for there is only immediate activity without reflection and without seeking. After these onsets, one lives with desires ( _d\u00e9sir_ , _Begierde_ ), where the object of desire is explicitly conceptualized by and present to reflection, in contrast with the earlier, lost, more animal life of appetite or need ( _besoin_ , _Bed\u00fcrfnis_ ). (Compare the transition in book 2 of the _Republic_ from the first pastoral, natural, innocent but inhuman city of pigs to the second human city of luxuries and feverishness and competition.) Self-consciousness is desire in general. Before its emergence, the poet in his movements through the natural world \"had no need of a remoter charm, \/ By thought supplied, nor any interest \/ Unborrowed from the eye\" (81\u201383). But \"that time is past\" (83); once somehow fallen into discursive consciousness, apperceptive awareness, and desire, there is no possibility of any return to any more innocent state. The earlier condition of prereflective awareness that is prior to the emergence of self-consciousness and desire cannot even be described: \"I cannot paint \/ What then I was\" (75\u201376). The origins of discursive thought cannot be established: \"Hard task, vain hope, to analyse the mind, \/ If each most obvious and particular thought, \/ Not in a mystical and idle sense, \/ But in the words of Reason deeply weighed, \/ Hath no beginning.\" Yet the poet also immediately undertakes to say something about what it was like to live in appetite alone, immersed in naturalness and without any need for or access to the remoter charms supplied by thought. The only relevance of this admitted fiction, perhaps like the fiction in political theory of a state of nature, must be to provide some sense of a dim possibility of wholeness and healthy activity for a subject who remains caught up in partiality and the feverish ways of the world. As David Bromwich has noted, the odd description of the innocent, animal activities of boyhood as those of one who is \"more like a man \/ Flying from something that he dreads, than one \/ Who sought the thing he loved\" (69\u201371) makes most sense if it is taken to refer to Wordsworth's life from 1793 (five years past) until the present moment in 1798. Whatever occasioning circumstances we may suppose to lie behind these dreads and flights, it remains the case that this description suggests a life of a self-conscious being (of a man rather than boy) that is a life of dreads and flights and not at all a life of assurance in activity or of attentive wholeheartedness, felt interest, and commitment in relation to the objects of common experience. In this circumstance, without a recovery of a decipherable origin and without guiding assurances drawn from knowledge of the ultimate substance of the world and life, to dare to hope\u2014across changes and in the wake of dreads and flights\u2014must be at best an _act_ of resolution that remains haunted by internal uncertainties and instabilities.\n\n(5) \"For such loss, I would believe...\" (87) [lines 85\u2013111]. In this section of second resolution or standing, the subject again makes an effort to gather himself, to resolve his uncertainties about life through the use of his own powers, without appealing to exterior things. This effort is again simultaneously successful _en m\u00e9sure_ and haunted by uncertainties. \"I would believe\" registers both: \"I will believe (so far as I can)\" and \"I would like to believe (but can not, not quite wholly).\" In either case, this formulation is substantially weaker, less assertational, than \"I do believe.\" Verbs of agency that would express this belief are implied but not stated. The poet will not faint, mourn, or murmur, that is, will not give way to these at least partly passive and induced responses to life. He will or would in contrast more actively do something, but exactly what he would do is not specified. He has learned to adopt an attitude\u2014\"to look on nature\" in a certain way\u2014that should support a certain course of wholehearted interest, commitment, and activity. But what that course is remains unspecified, and learning to look on nature is itself not stabilized or grounded in learning theoretically _that_ nature is (really, ultimately) thus and so. The adoption of an attitude is not grounded in theoretical knowledge, nor can it be. This adoption must remain in the register of resolution, in the register of what one \"would believe.\"\n\nThe adoption of this attitude is supported, we are told, by the fact that \"I have felt \/ A presence that disturbs me with the joy \/ Of elevated thoughts\" (93\u201395). There is, however, no explanation of either the source of this feeling in ultimate things or its aptness to them. It is responsive to \"something far more deeply infused \/ Whose dwelling is the light of setting suns, \/ And the round ocean, and the living air, \/ And the blue sky, and in the mind of man\" (96\u201399). But this something is not named as an object of theoretical knowledge. At best it can be called an impelling \"motion and a spirit\" (100). Whatever joy felt responsiveness to this motion and spirit brings, it also brings disturbance, in the thought that one has not, not yet, lived fully in resonance with this spirit and its motions; perhaps one cannot so live. Metaphysical confidence is implied\u2014but only implied, not stated, and disturbance yielding to doubt is not banished.\n\nThe resolution that concludes this section is now explicitly in the mode of will, not knowledge. \"Therefore am I still\" (102) expresses a determined effort at self-stabilization, at lingering as a whole subject within the specular moment of meaningfulness. For at least this moment of feeling, the subject is stabilized enough in its responsive resoluteness to be a subject in resonance with a significant order and so capable of wholehearted interest, commitment, and activity. But it cannot last. The announcement that the subject is \"well pleased to recognize \/ In nature and the language of the sense, \/ The anchor of my purest thoughts, the nurse, \/ The guide, the guardian of my heart, and soul \/ Of all my moral being\" (107\u2013111) is honest to the moment of feeling and of self-resolution. But this announcement undoes itself in the very moment of its articulation. To say that one is \"well pleased\" is to hint that there is at least a danger that one is pleasing oneself by constructing this experience and its felt significance, where this construction may be driven by the needs of the subject rather than by how things are. Hence this construction may be both ego centered and empty. Wordsworth's special courage in the explorations of his movements of consciousness is to register this possibility and all the uncertainty it entails, rather than to deny it in reversion to dogma.\n\n(6) \"Nor perchance \/ If I were not thus taught\" (111\u2013112) [lines 111\u2013134]. And so the poet accepts the possibility of error. It may be that he has not been thus taught by nature, that his depth of feeling and his resolution to be a coherent, expressive, responsive subject may be vain constructions. The poet's response to this possibility again takes the form of a resolution crossed with an imperative. \"Nor... should I the more \/ Suffer my genial spirits to decay\" (111, 112\u2013113) were this the case. I will not suffer this, and it is best for me not to suffer this. Things would not go well (standing melancholy would be my lot) were I to do so. So I will not do so. I would not that it be so.\n\nThat Dorothy (\"thou my dearest Friend\" [115]) is present with him in this spot to confirm his feeling and to stabilize his resolution suggests to the poet some help with his plight. I should not suffer my genial spirits to decay because thou art with me. But exactly how does her presence help? It does not cancel the registers of resolution and imperative, does not transform these registers into confident assertion. Quite the contrary, it moves the poet's consciousness explicitly into the mode of prayer. \"Oh! Yet a little while \/ May I behold in thee what I was once, \/ My dear, dear Sister! And this prayer I make...\" 119\u2013121). \"May I behold,\" not \"shall I behold\"; let it be so _Deo volente_ , not I know that it is so. Nor is contact with an originary naturalness, putatively exemplified by Dorothy, so easily established. It is not clear that the poet in fact beholds that naturalness now: \"may I behold,\" not \"do I behold\" \"in thee what I was once.\" As John Barrell has argued, Wordsworth's desperate effort to achieve self-composure without turning to anything external to his experience\n\nrequires Dorothy to perform a double function in the ratification of his achievement of a transcendent subjectivity. First, he needs to believe that Dorothy will grow up and sober up, for by doing so she will naturalise and legitimate his own loss of immediate pleasure in nature. The transition she makes, from the language of the sense to that of intellect, will be an observable process, one which will recapitulate and historicise the transition Wordsworth has already made. But in the second place, the language of the sense, as presently employed by Dorothy, stands as a present and audible guarantee of the meanings of his own language of the intellect; it assures him of the secure foundation of his language in the language of the sense.\n\nBarrell goes on to argue, quite cogently, that the double function assigned to Dorothy is incoherent. She cannot both grow up into discursive consciousness and apperceptive awareness and remain immersed in naturalness. \"Dorothy can perform these two functions [repetition of growth into self-consciousness plus an anchoring persistence in naturalness], only if her potential for intellectual growth is acknowledged, but only if, also, that potential is never actualized.\" It makes no sense, however, to assign her a potential that is both actualized and never actualized.\n\nIn all this Barrell is quite correct. But the problem goes even deeper than the use of Dorothy, and the problem is, moreover, registered in the poem itself. The problem is that no transition from naturalness (appetite) to discursive subjectivity (desire) can be both historically accomplished _and_ naturalized, in such a way that the transition takes place (there is a fall into discursivity and into exteriority to nature) while a continuing saving resonance to pure naturalness is maintained. This is no more possible for William than it is for Dorothy. This fact accounts for the persistence until the end of the poem of the now explicit mode of prayer. What may happen, _Deo volente_ , and what would save one as a subject by bringing one continuously into a life of full meaningfulness, is something that can never be known to happen, something that cannot even be coherently imagined by us to happen. \"This prayer I make\" (121)\u2014that Dorothy fulfill this incoherent requirement, and that the poet himself has blended transcendent natural meaningfulness with finite subjectivity\u2014must remain a prayer, not an announcement of an accomplishment. When the poet makes this prayer \"Knowing that Nature never did betray \/ The heart that loves her\" (122\u2013123), the syntax undermines the claim to knowledge. \"Never did\" suggests an event of doubt, a fall into discursivity, apperceptive awareness, and exteriority to nature, a fall out of mere naturalness.\n\nIt is no accident then that the knowledge that is claimed modulates very quickly into the claim that nature \"can so inform \/ The mind that is within us\" (125\u2013126) that our cheerful faith in life is undisturbed. That this can or might be so is no guarantee that it will be so. Just what potentiality for informing our lives so as to produce cheerfulness is in fact actualized? When and how does this actualization of this potentiality take place? Exactly by what, when, and how might \"all \/ The dreary intercourse of daily life\" (130\u2013131) be transformed? These questions have no answers, at least not within this poet's movements of consciousness. Hence \"Our cheerful faith, that all which we behold \/ Is full of blessings\" (133\u2013134) remains a faith, perhaps all too vainly willed, rather than an article of knowledge.\n\n(7) \"Therefore let.... Thy memory be as a dwelling place\" (134, 141) [lines 134\u2013146]. The \"therefore let...\" of line 134 explicitly begins the prayer announced in line 121 (\"this prayer I make\"). It asks for Dorothy both now to persist in her naturalness or natural connection to nature and \"in after years\" (137) for her memory to be \"as a dwelling-place \/ For all sweet sounds and harmonies\" (141\u2013142). The prayer is that she should at least preserve in her maturer consciousness a sense of past connection to nature and its beneficent influences. Yet this prayer for Dorothy quickly modulates for the poet into thoughts about himself. In a question to himself that is not marked as a question but rather as a declarative interjection within his own consciousness, Wordsworth asks \"with what healing thoughts \/ Of tender joy wilt thou then remember me, \/ And these my exhortations\" (144\u2013146). The anxieties and doubts that haunt him are made explicit in this interrogative movement of his own consciousness. Will Dorothy, or anyone else for that matter, remember him, and if so, how (\"with what healing thoughts\") and why? What have his life in time and his just now occurring course of experience and reflection meant? Has he been a subject who has lived, fully and memorably, or not? No doctrine is available to ground any certainty of the achievement of the life of a subject. Only exhortations are possible; no declarations and no proofs can either ground the significance of a life or control the responses of any audience, even an audience as intimately present as Dorothy is in the scene. Prayer, not declaration, is the appropriate mode of acknowledgment of the ungroundedness of any claims to fullness of value on behalf of a human life in time.\n\n(8) \"Nor, perchance... wilt thou then forget\" (146, 149) [lines 146\u2013159]. Within the concluding prayer, the topic of death, and so of the significance of a life that is bounded by death, is raised explicitly. \"Nor, perchance\u2014\/ If I should be where I no more can hear thy voice\" (146\u2013148) implies, given Dorothy's role as instance of a saving audience in general, \"if I should be where no saving voice, no answering glance, neither Dorothy's nor anyone else's, is to be found.\" Within the awareness of death and the finite subjectivity bounded by it, one can declare very little, can say only \"nor perchance wilt thou then forget,\" not \"you will not forget.\" And yet a standing together, here and now, is possible, and that this standing together may be remembered is, at least here and now, enough to afford the poet some sense of stability and continuing selfhood. It is possible that the feeling or attitude borne by a subject so standing in this spot, the feeling or attitude of \"deeper zeal\" (154), may be appropriate to the life of a subject as such in this spot, and so it too may stand in another's memory. The poet confesses feelings (\"deeper zeal\" [154], \"to me... dear\" [158\u2013159]) and claims aptness and exemplarity both for them and for his confession of them. This claim and confession cannot be grounded in any argument. Perhaps either the feeling or the confession of feeling has been vainly constructed. Yet they may stand. Time will tell.\n\nIV\n\n\"Each individual that comes into the world is a new beginning; the universe itself is, as it were, taking a fresh start in him and trying to do something, even if on a small scale, that it has never done before.\" If John Dewey is right about this, then there is in the particularity of each person also a standing exteriority to the pure manifestation of the essence of human subjectivity in time and nature. Geoffrey Hartman has characterized Wordsworth as a poet as \"a ' limitour,' licensed to haunt only the borders of the country from which imagination comes and to which it seeks to return.\" The discursive subjectivity that is constructed through the work of the imagination in arriving at a point of view and that is bound up with apperceptive awareness and awareness of variously attentive others stands apart from the full, meaningful naturalness of subjectivity that it would wish to achieve by establishing an absolute, value- and stance-affording connection to nature and culture as such. \"The terror of discontinuity or separation enters... as soon as the imagination truly enters.\" As a result, Wordsworth's effort to find in nature a fully saving genius loci or natural source of transcendent meaningfulness founders. He can only wander from one moment to another of what is felt, almost, to be an accession to meaningfulness, but an accession that remains internally fragmented. Wordsworth's \"quest to localize his Idea of [A Saving] Nature in Nature fails.\" And so the pursuit of such moments goes on. Apart from such moments, there is the conventionalized, less meaningful life of ordinary subjectivity, caught up in circuits of antagonism (\"evil tongues, \/ Rash judgments, [and] the sneers of selfish men, \/ [And] greetings where no kindness is\" [128\u2013130]), a life that motivates, always, a wish for a fuller, more meaningful life of an exemplary, stabilized subject as such. The voice that arises out of the moments of almost accession to meaningfulness, almost achievement of exemplarity, \"is the voice of a man who has been separated from the hope he affirms.\" Hartman has characterized the progress of _The Prelude_ as \"no argument, but a vacillation between doubt and faith,\" and this halted progress is evident, too, in the waverings of prayerful hope and recurrent uncertainty in \"Tintern Abbey.\" Wordsworth's \"curious and never fully clarified restlessness\" is, Hartman suggests, \"the ultimate confession of his poetry.\"\n\nThough they afford no doctrine to guide us, perhaps such vacillations and confessions show us something (though only something) of the lives of human subjects as such. Wordsworth's expression of his internalized quest romance, showing arrival at a fuller assurance crossed with doubt, enacts a romance of subject and world that is both strongly subjectivized and aimed at increased life in the world. If the modality of aiming is that of reflection, rehearsal, and conjecture more than specific, actual worldly political praxis, that is perhaps in part because the world of England in 1798 does not yet readily admit specific, sustained practices of wholeheartedness in daily life. Hence Wordsworth's poetry of reflection, rehearsal, and conjecture functions in relation to that world as disturbance, provocation, and placeholder, not only or simply as a source of reassurance. Our world may differ from Wordsworth's politically, culturally, cognitively, and technologically in manifold ways, but it stands no less in need of the kind of reassurance mixed with disturbance that Wordsworth's poetry affords. Promises of things not seen that remain unfulfilled arise within the lives of discursive human subjects, and they function, sometimes, critically and productively in relation to those lives, for those who have ears to hear.\n5.\n\nThe Ends of Literary Narrative\n\n_Rilke's \"Archaic Torso of Apollo\"_\n\nThe claim of works of literature to represent truths about the world is, at best, peculiar. Most literary works are fictional. Authors spend their time and energies in thematizing, in developing attitudes toward subject matters, and in seeking formal power and coherence. There are no procedures in view that can arrive at results about matters independent of human subjects and attitudes. In contrast, a proof in mathematics ends by reaching its final line, where each line that is not an axiom is generated in explicit accord with a rule of inference that in principle anyone might follow. Reports of experimental results generated in a lab specify procedures that were followed in setting up equipment and carrying out tests. While they often also offer conjectural interpretations of results and suggestions for further work, they describe minimally a procedure that anyone might follow in order to achieve a like-enough result. Hence we can speak readily of objective evidence that a certain state of affairs can be produced thus and so. In statistical social science, one finds reports of results from questionnaires or other data about populations expressed in numerical terms. Under the assumption that a larger population will not be too different from a sample, one can draw conclusions about distributions of traits and tendencies of development. History undertakes to tell us what happened, and the claims of professional historians are supported with reference to primary sources, indicated in footnotes. In economics, one often finds abstract mathematical models that describe processes of income distribution or GNP growth, for example, that are supposed to occur underneath a confusing surface of extra variables that induce deviations from the model. Among these cognitive practices, literature is perhaps most like economics in giving a model of certain processes in the world. This is scant comfort, however, since whether the processes described by economic models really do occur, on the one hand, or are rather fairy tales invented by clever calculators, on the other, is itself a subject of more than a little dispute. Literary models, moreover, if that is what literary texts offer us, are in even worse shape, since they focus only on very small numbers of mostly made-up cases, and they lack even the potential of refinement through the incorporation of further data.\n\nInstead of focusing on literature as a form of cognitive work, then, we might think of works of literature as aiming at producing a certain sort of pleasure. If we further suppose that all pleasures are subjective and rankable only in terms of duration and intensity, then the point of literary works would be exhausted in their consumption, and there is not much more to be said than this. As Bentham notoriously remarked, all other things being equal, pushpin is as good as poetry.\n\nThis view is unsatisfying, however, in that the experience of reading a powerful literary work is not really much like the experience of eating an ice cream cone or wallowing in a warm bath. It takes some work to pay attention. It is not exactly fun at every instant. The pleasure, if that is the right word, seems not to have much to do with sensory processes but more with the work that the reader is doing. And surely writers are trying to do something that is both cognitively available to their audiences and cognitively significant. But then, again, works of literature do not offer us results that are much like those of mathematics, laboratory science, history, or statistical social science. So we are faced with a puzzle. We seem to learn something from reading literature, but we have trouble explaining exactly how or what we learn\u2014at least when we are in the grip of a certain picture of knowledge as the methodologically correct achievement of a fact-stating result.\n\nIt is easy to suggest that there must be a third way\u2014between the forms of knowledge that are available in other disciplines and mere, predominantly sensible pleasure\u2014in which literature is significant. We can see that this suggestion makes sense when we contrast art in general with scientific knowledge, on the one hand, and decoration and entertainment, on the other. Art is somehow in the middle here. If we are offered too many \"flat\" facts by a particular work, we are likely to find it merely journalistic and to want more pleasure. If we are offered too much pleasure, we are likely to find the work either decorative or an escapist guilty pleasure, like the novels of Ian Fleming or Dan Brown, say. We want, at least sometimes, to work harder and to learn more than that. But just how can we do this? The mere postulation of a third way does not yet answer this question.\n\nIn their valuable comprehensive survey _Truth, Fiction, and Literature_ , Peter Lamarque and Stein Haugom Olsen explore a number of ways of thinking about literature as a source of knowledge. Centrally, they consider the following three suggestions: (1) Literary works might help us to know \"what it is like\" to be (or to be in the situation of) a certain character, in the sense of \"subjective knowledge\" characterized by Thomas Nagel and worked out with regard to literature by Dorothy Walsh. Against this, Lamarque and Olsen object first that while we have experiences while reading, we mostly have our own experiences, not the experiences of Leda or Leopold Bloom, Yeats or Joyce. In particular, we mostly observe or imagine characters having experiences. And while we take an interest in this observation, we are not learning the felt qualia of, say, fried kidney for Leopold Bloom. Second, even if we did get some sense of what things are like for characters from reading literary fiction, it is strained, Lamarque and Olsen suggest, to describe what we get as learning something. There are no methods in view for accrediting or testing any knowledge claims, such as there are in the sciences, and much of what we might think we learn we must in fact already have known in order to understand what is going on: for example, that rape is a violent, terrifying, and world-altering experience. Or (2) literary works might enable us to enrich our store of concepts, or they might modify our sense of the application conditions of concepts we already possess, as Catherine Wilson and D. Z. Phillips have suggested. Against this suggestion, Lamarque and Olsen object that while some literary works might help us to deploy new concepts or to refigure the conditions of their application, this is by no means necessary for a work to have literary value. Second, and more sharply, they suggest that some authors sometimes explore the same concepts and conditions of application in different works, so that when one reads a second work, for example, a later play by Ibsen, one may not learn anything new. But the later work nonetheless has literary value; hence learning about concepts and their application conditions is not necessary for literary value (378\u2013386). Or (3) it might be that literary works help us to become better perceivers of the moral lives of persons and so better reasoners about what it is good or right to do when, as Martha Nussbaum and Hilary Putnam have suggested. Against this suggestion, Lamarque and Olsen object first that improvement in moral reasoning is by no means brought about by all successful literary works, and, second, that having or furthering the correct valuational stance is not a necessary condition for literary value: we can and do value as successful works with whose stances and points of view we disagree (386\u2013394).\n\nOne might suspect that there is something wrong here with Lamarque and Olsen's \"divide and dismiss\" strategy. Perhaps what we get from reading literature is some mixture of subjective knowledge, improvement of our conceptual capacities, and moral insight. Lamarque and Olsen themselves offer the positive suggestion that literature \"develops themes that are only vaguely felt or formulated in daily life and gives them a ' local habitation and a name'\" (452). \"Giving a name\" at least hints that some sort of cognitive achievement is on offer. Literary appreciation, they further remark, \"constitutes its own form of insight, its own kind of interpretation of thematic concepts\" (409). But this form of insight, they argue, is better construed as the cultivation of _understanding_ than as the acquisition of knowledge of true propositions. \"Literary works can contribute to the development and understanding of the deepest, most revered of a culture's conceptions without advancing propositions, statements, or hypotheses about them\" (22). \"We can imagine, ponder, entertain thoughts, or speculate about something without any commitment to the truth of our ruminations\" (11). Literary practice is best understood as an imaginative exploration of themes that is guided by the literary work, which undertakes \"to develop in depth, through subject and form, a theme which is in some sense central to human concerns\" (450).\n\nBut while this talk of understanding is a good start, it leaves us not so far beyond where we were before. Exactly what do we understand when we understand the theme of a literary work? How is this understanding related to, but different from, propositional knowledge either that the world is thus and so or that the author thought thus and so (itself just another fact about the world)? How is this understanding cultivated by the experience of the work itself? What does it mean to \"develop a theme _in depth_ \"? What, if anything, makes such development valuable in human life?\n\nJohn Gibson suggests that the important cognitive work of literature consists in \"bringing into full view our standards of representation [and] our linguistic criteria for what the world is.\" A literary work may show some phenomenon \"just as it is\" (61); for example, we may see the essence of racism in the figure of Iago (61\u201362). Shakespeare's presentation of Iago \"draws together at... a level of clarity and order everything we call racism\" (63), thus making the shape of our concept available to us for acknowledgment. This suggestion too is a useful start. But what it fails so far to explain is how we can fail to know the criteria of some of our concepts and, hence, why we need to explore and acknowledge them. Surely we need already to have some pretty clear command of the concept of racism in order to understand Iago's actions at all. What further dimensions of our concept, then, are subject to repression or forgetting, and how do the details of the presentation of Iago as a literary character activate these dimensions? What, exactly, is the cognitive import of having our concepts activated and somehow \"filled in\"?\n\nGibson further suggests that a general reason why we turn to works of literature is that we are able there to \"read the story of our shared form of life\" (50). This is the suggestion we must pursue if we are to have any hope of unpacking the jointly cognitive and emotional work of acknowledging and working through that reading literature makes available to us. So what is the story of our form of life? This enormous question is one that will have to be faced if we are to make any progress here.\n\nPart of that story is the playing out of a biologically engendered imperative to survive. We need to eat, sleep, protect ourselves, and procreate in order to survive as a species, and we are, so far, wired well enough for success in these endeavors. In the absence of extraordinary strength or speed, we have managed cope with our environments mostly through superior cunning. We are better at recognizing and manipulating more features of our environments than are members of other species. In particular, as concept-mongering creatures, we are able not only to see objects brutely, as it were, as members of kinds; we are also able to see them from a point of view, as this or that. For example, a stone may be recognized by us as a weapon, a piece of building material, or an implement for scratching or shaping. A fundamental part of learning language is developing this repertoire of seeing an object as something. We manage this achievement not simply through picking up on the individuals-just-sorted-into-natural-kinds that are present in our environment. Other animals do this as well, but they lack our conceptual repertoire. My dog responds to the sounds of squirrels but does not think of them as mammals or rodents or nut gatherers. We, however, manage feats like this by picking up not only on our environments brutely, but also by picking up on how other human subjects are interacting with our shared environment, by picking up on their points of view on things. Our having of a wide repertoire of concepts and application criteria, enabling manifold different responses to our environment, is not a matter only of matching inner idea or Platonic archetype or brain state with object. It is a matter of learning to see things within multiple and shifting contexts of engagement and use, a matter of catching on to a large number of things that are done or might be done, by others and by oneself, at once with objects and with words, within practical engagements. In coming to be masters of words that encode objects and events seen in one way or another, in relation to multiple contexts of engagement and possible response, we are neither machines nor the quasi-automatons of Wittgenstein's language game (2) in _Philosophical Investigations_. Rather we are creatures who have become capable of a life of plastic attention\u2014capable, that is, of culture.\n\nThe fact that we develop conceptual consciousness not only in relation to problems of biological survival but also in relation to cultural contexts of flexible attention and engagement brings with it certain distinctive burdens and possibilities. Not only is one trying to survive; one is also trying to play the game of attending according to concepts both with others and in competition with others to have one's own point of view and way of playing the game recognized. Concepts and words, for all that they register features of our environments there to be registered, are also, in their lives within cultural contexts of shifting attention and engagement, both stable enough to permit communication and sharing of a point of view on things and tolerant of new uses as new contexts of attention and interest develop. Hence coming to language and conceptual consciousness brings with it uncertainties about how to go on from where we are or where one is. Am I playing the game in the right way? Is my conceptual performance such that it can and should be taken up by others? Do I really know what I'm doing? What are evident and exemplary fluency and command in making moves with words? Just who do I think I am, and am I right about that?\n\nThese questions are such that they cannot and do not arise at every moment; comprehensive skepticism is not a genuinely available stance in life. But they are also such that they can always arise at some point, at least in modernity, where manifestly different ways of life develop and make themselves present in awareness. As the Kantian tradition emphasizes, a life with concepts is a life in which questions of judgment are always potentially in view, and the fact of continuing responsibility in and for conceptual performance is unavoidable. (It would be a mistake to see our concepts as self-standing representational \"entities\" existing outside of circuits of mimesis and training.) R. G. Collingwood tells the following wonderful story about what it is like to come to conceptual consciousness and language, thus becoming a subject of and in culture:\n\nA child throws its bonnet off its head and into the road with the exclamation \"Hattiaw.\" By comparison with the self-conscious cry discussed earlier in the present section, this represents a highly developed and sophisticated use of language. To begin with, consider the emotion involved. The child might remove its bonnet because it felt physically uncomfortable in it, hot or tickled or the like; but the satisfaction expressed by the cry of \"Hattiaw\" is not a merely psycho-physical pleasure like that of rubbing a fly off the nose. What is expressed is a sense of triumph, an emotion arising out of the possession of self-consciousness. The child is proving itself as good a man as its mother, who has previously taken its bonnet off with the words it is now imitating; better than its mother, because now she has put the bonnet on and wants it to stay on, so there is a conflict of wills in which the child feels himself victor.\n\nAs this example shows, even very early on in our life as possessors of conceptual consciousness and self-consciousness, we bear distinctive emotions and attitudes toward our situations. We are capable of accepting, working through, and expressing these emotions, with a resulting sense of a certain kind of triumph, when our point of view is recognized by others through our performances. We are capable also of sullenly shirking our emotions, avoiding them, or otherwise failing to express them, with a consequent sense of disappointment, frustration, and failure, and, sometimes, with a further wish to escape or reject the burdens of the responsibility for expression. When this happens, we then suffer or merely undergo our emotions, as we remain stuck in the state of having what Spinoza calls an inadequate idea of an affection: we don't know what is worth caring about; we take no delight in the investment of our energies in our performances, and confused, unexpressed feelings wash over us. Our actions are as much reactions as expressions of our selfhood. Philosophical skepticism and its intimate antagonist epistemological realism are both at bottom misbegotten intellectualized efforts to repudiate the situation and expressive possibilities of conceptual consciousness and self-consciousness by describing them away. (What Stanley Cavell calls the truth of skepticism is the fact that the skeptic, at least, registers a certain failure and disappointment that attach to these efforts.) More happily, however, there are also what Charles Altieri calls \"the kinds of satisfactions that are available for agents simply because of the qualities of consciousness they bring to what they are feeling.\" We can do something with these qualities of consciousness. As Wordsworth argues in the \"Preface to _Lyrical Ballads_ ,\" the poet, through thinking \"long and deeply\" in relation to our feelings, may uncover \"what is really important to men,\" with the result that, when this course of discovery is taken up and followed, \"the understanding of the Reader must necessarily be in some degree enlightened, and his affections strengthened and purified.\" Friedrich H\u00f6lderlin on infinite satisfaction and John Dewey on consummatory experience describe in quite similar terms the distinctive sorts of satisfactions open to us as human subjects. The achievement of further understanding coupled with strengthened and purified affections, with both understanding and affections then discharged in a dense, medium-specific performance of working through, in which a point of view is made manifest and recognition and like-mindedness are successfully solicited, is what I have elsewhere called the achievement of expressive freedom. It has, I think, some claim to be regarded as an immanent telos of human life, made both possible, partially, and valuable for us by our mysterious possession of conceptual consciousness and self-consciousness, developed and worked out in relation to public media of expression.\n\nIt is impossible to prove the correctness of this view according to the standards of proof that are held in place in the Cartesian tradition. (Those standards were specifically developed in order to block talk of the purposes of things.) But it remains nonetheless an articulation of what is going on in human life that may be unavoidable and illuminating. If it has any chance of being right, then Lamarque and Olsen are wrong when they remark that \"mostly, we simply do not meet the grand themes in trivial daily life\" (455). Yes and no. Yes, we do not meet them clearly formulated and perspicuously manifested there; there is too much muddle for that, and there are too many different circumstances in which lives are led for it to be just obvious that we are in pursuit of expressive freedom. But no, we do meet these themes there latently, to be acknowledged, as we come to see our lives as in part caught up in situ in the pursuit of expressive freedom, involving articulate clarity and wholeheartedness of interest in life.\n\nGreat writers, then, manage to achieve expressiveness: that is, to face up to and work through the emotions and attitudes that come with being a human subject, as those emotions and attitudes are given specific contours in specific situations. They make it manifest for themselves and for us how a specifically shaped emotion, mood, or feeling has been brought about in or by a situation and how, further, that emotion, mood, or feeling can be accepted as appropriate. As a result, the emotion, mood, or feeling is actively accepted, not passively suffered. Barbara Herrnstein Smith describes the achievement of poetic closure from the reader's point of view in just these terms:\n\nClosure occurs when the concluding portion of a poem creates in the reader a sense of appropriate cessation. It announces and justifies the absence of further development; it reinforces the feeling of finality, completion, and composure which we value in all works of art; and it gives ultimate unity and coherence to the reader's experience of the poem, by providing a point from which all the preceding elements may be received comprehensively and their relations grasped as part of a significant design.\n\nFor the reader, that is to say, the poem itself is experienced as coherent, closed, and designed, as its parts form a self-completing whole. This experience is a function of form, but not of form alone. It occurs in part because the poet has succeeded in making sense of experience and emotion, has succeeded in working them through to achieve acceptance and composure. As Herrnstein Smith notes, \"the experience of closure is the complex product of both formal and thematic elements\" (40). This means that the poet has found, formally, words and structures to thematize, connect, and accept experiences and emotions that were initially burdensome, troubling, exhilarating, or provocative. She goes on to note that many contemporary poems, beginning with Eliot and reaching a high point in Robert Lowell, exhibit increasingly \"dialectical-associative\" thematic structure. \"In much modern poetry,\" she remarks, \"the occasion for a poem is... likely to be the existence of an ultimately unresolvable process\" (247). There is what she calls a \"poetry of non-statement\" (254) that takes both subjective-lyrical stream-of-consciousness guises and objectivist-imagist-language play guises. The reason for this development is that we have grown, appropriately, skeptical of the availability and livability of \"they lived happily ever after.\" Nineteenth-century novels, as both Henry James and David Lodge have mordantly remarked, seem to end only with marriage, death, or an inheritance. In contrast, we have grown suspicious of the availability and value of these kinds of closure in life, which seems to us to be more complicated than that. But even in the contemporary poetry of antistatement, the shape and feeling of a particular instance of perplexity are expressively worked through, at least when things go well. The writer and the reader afterward come to know and accept exactly how there are complexities of situation and feeling. As Herrnstein Smith puts it, \"a poem allows us to know what we know, including our illusions and desires, by giving us the language in which to acknowledge it\" (154). Such an achievement of acknowledgment is available and important for us just insofar as we are human subjects who attempt to lead lives actively, with senses of meaning and of appropriate responsiveness to events, unlike Nietzsche's cows, who do little besides undergo their lives. Unlike other animals, we remember and anticipate incidents quite widely, together with an awareness of how incidents and things are seen by others from multiple points of view. And so we wonder: Who am I to see, remember, and anticipate things like this? To what extent are my point of view and emotions toward things apt and appropriate? Am I genuinely acting as a reasonable subject in seeing things and feeling as I do?\n\nIn the grip of a healthy empiricism, it is of course possible to find this talk of expressive freedom and of leading a life actively to be quite misplaced in relation to what is after all also a sheerly material situation. There is, again, nothing like a proof by Cartesian standards that expressive freedom is the immanent telos of human life. But what does it look like, according to this conception, when someone rejects it and denies that expressive freedom matters for us and that it is partially, but only partially, available to us through different actions in different settings? (It is possible to say anything.) The Humean-skeptical, Darwininan-naturalist insistence that we are nothing but natural beings who must simply cope with things and the Cartesian-Platonist insistence that absolute knowledge of our place in nature can guide us if we but somehow think aright both appear as hysteria-driven denials of what it is to be a finite, active being in time. \"You ask me,\" Nietzsche once wrote, \"which of the philosophers' traits are really idiosyncrasies? For example, their lack of historical sense, their hatred of the very idea of becoming, their Egypticism. They think that they show their _respect_ for a subject when they de-historicize it, _sub specie aeterni_ \u2014when they turn it into a mummy.\" To deny that our lives are caught up in becoming and in possibilities of the achievement of expressive freedom in part, but only in part, in relation to it can look like an attempt to deny or kill human life because it is too painful.\n\nYet as Nietzsche also remarked, it can also sometimes happen\u2014if and when we manage ourselves to work through and express our emotions in a dense, commanding performance, or if and when as readers if we follow and participate in the workings-through of others\u2014that we are left with the sense, at least for a time, \"that life is at the bottom of things, despite all the changes of appearances, indestructibly powerful and pleasurable.\" In a late notebook entry, Nietzsche describes the authentic state, a state that may either occur in life or be \"set up\" in art, as the state\n\nin which we put a _transfiguration and plenitude_ into things and work at shaping them until they reflect back to us our own plenitude and lust for life.... Art reminds us of states of animal vigour; it's on the one hand a surplus and overflow of flourishing corporeality into the world of images and wishes; on the other a rousing of the animal function through images and wishes of intensified life\u2014a heightening of the feeling of life, stimulus for it.\n\nWithin experience, a pattern can sometimes be discerned, partially and dimly, in our relations as subjects to things and events, and emotions, feelings, attitudes, and moods can be experienced and worked through as appropriate to that pattern. Discovery and exhilaration are mixed with a sense also of mystery and complexity in the face of a becoming, a life in time, that is not wholly masterable. For this reason, great endings, as Steven Winn remarks, \"define and disappoint, frustrate and gratify. They confer meaning and confirm the structure of what's come before\u2014in a movie, a sonata, a work of fiction. But they also kill off pleasure, snap us out of the dream, and clamp down order on experience that we, as citizens of the modern world, believe to be open-ended, ambiguous, and unresolved.\"\n\nThis experience of an ending is like what Aristotle describes as the catharsis\u2014at once the clarification and unburdening\u2014of an emotion in relation to a situation. But, as Frank Kermode notes, whereas \"for Aristotle the literary plot was analogous to the plot of the world in that both were eductions from the potency of matter,\" which eductions are presided over purposively by divine intelligence, for us the sense of plot in life proceeds at least in part from our own \"store of contrivances\" (40), as we are driven by \"a need to live by [a] pattern\" (109). We half believe in these patterns, as we experience our lives within them and experience possibilities of clarification of our situation. And yet we remain also aware of our own role as contrivers, aware of the lack of a presiding pattern that is everywhere evident in human life and aware also of our own failures to live in perfect freedom and infinite satisfaction, in the face of the mysterious complexities of becoming. And so we tell stories and attempt to work through our emotions in relation to the particulars of changing situations, so that we can, as Kermode puts it, both \"avoid the regress into [a] myth\" of presiding purposiveness and yet preserve the sense that \"the scene [of human significance] has not yet been finally and totally struck\" (43). Fictions that find plots, so as to work through emotions in relation to situations and experiences, remain for us both \"deeply distrusted,\" since they are only our contrivances, and \"humanly indispensable,\" since only these contrivances can give us the sense of leading a life meaningfully and actively. They offer us a way, even the way, to cope with both anxiety at a sense of the pervasive contingency of things and bad faith in fixed, master supernatural plots we can no longer trust (151). They are our means of coping with \"the tension or dissonance between paradigmatic form and contingent reality\" (133), between the sense that every life is a parable of each, with meaning to be found, and the sense that there are only material happenstance and subjective \"preference having.\"\n\nThat we may through producing and effectively receiving exemplary works of art come to a fuller, more animated, more ensouled life, beyond mere preference having, yet also as individuals free from submission to alien authority, is a major subject of Rainer Maria Rilke's _Neue Gedichte_ ( _New Poems_ , 1907; second part, 1908). The new poems in general and the Thing-poems in particular result significantly from Rilke's responses to the French sculptor Auguste Rodin. Rilke worked as Rodin's secretary in 1905 and 1906, he wrote a 1903 monograph and a 1907 lecture on Rodin, and the first volume of _Neue Gedichte_ is dedicated to him. Rodin's influence is evident in the _Neue Gedichte_ in two distinct ways. His sculptures provided for Rilke a model of an animated, all but living work of art. As Judith Ryan puts it, \"In Rodin's sculpture, the interplay of light between the various surfaces creates for Rilke the sense of something that continues to shape itself before the eye of the beholder,\" as though the sculpture were alive. Second, \"Rodin's ideal of persistent workmanship,\" his sense of the importance of patience and craft, makes itself felt in the tightly controlled formal structure of the poems. The effect of animation in the poems then arises from the way in which inspiration and living energy, ideals Rilke associated with Theodore de Banville, are submitted to formal control, in order to construct a living whole.\n\nIn a famous sonnet appearing as the first poem of _New Poems: Second Part_ , Rilke describes what it is like to come suddenly, from within one's middle situation, between dead, foreign materiality and perfect transcendence, to a sense of possibilities of fuller animation.\n\nARCHA\u00cfSCHER TORSO APOLLOS\n\nWir kannten nicht sein unerh\u00f6rtes Haupt,\n\ndarin die Augen\u00e4pfel reiften. Aber\n\nsein Torso gl\u00fcht noch wie ein Kandelaber,\n\nin dem sein Schauen, nur zur\u00fcckgeschraubt,\n\nsich h\u00e4lt und gl\u00e4nzt. Sonst k\u00f6nnte nicht der Bug\n\nder Brust dich blenden, und im leisen Drehen\n\nder Lenden k\u00f6nnte nicht ein L\u00e4cheln gehen\n\nzu jener Mitte, die die Zeugung trug.\n\nSonst st\u00fcnde dieser Stein entstellt und kurz\n\nunter der Schultern durchsichtigem Sturz\n\nund flimmerte nicht so wie Raubtierfelle;\n\nund br\u00e4chte nicht aus allen seinen R\u00e4ndern\n\naus wie ein Stern: denn da ist keine Stelle,\n\ndie dich nicht sieht. Du mu\u00dft dein Leben \u00e4ndern.\n\nARCHAIC TORSO OF APOLLO\n\nWe cannot know his legendary head\n\nwith eyes like ripening fruit. And yet his torso\n\nis still suffused with brilliance from inside,\n\nlike a lamp, in which his gaze, now turned to low,\n\ngleams in all its power. Otherwise\n\nthe curved breast could not dazzle you so, nor could\n\na smile run through the placid hips and thighs\n\nto that dark center where procreation flared.\n\nOtherwise this stone would seem defaced\n\nbeneath the translucent cascade of the shoulders\n\nand would not glisten like a wild beast's fur:\n\nwould not, from all the borders of itself,\n\nburst like a star: for here there is no place\n\nthat does not see you. You must change your life.\n\nThis poem describes not simply an object but preeminently an experience of an object, an experience had by or available to a \"we.\" The poem embodies an effort to formulate and to work through an experience of perplexity, both for the speaker and for other subject-viewers, to whom and for whom the poet may speak. The statue fragment is characterized above all in terms of its effect on the speaker-viewer, in its overwhelming presence to a viewing consciousness. Within that experience, the fragment presents itself as having an inside, felt as a source of expressive and sexual power that is brought to fullness of presence in its outer surface. The formed surface glows ( _gl\u00fcht_ ), gleams ( _gl\u00e4nzt_ ), blinds the viewer ( _dich blenden_ ), and bears a smile ( _ein L\u00e4cheln_ ) as a promise of responsive sexuality. Its parts are not detached or misplaced ( _entstellt_ ); instead the stone glistens ( _flimmerte_ ) in its translucent falling ( _durchsichtigem Sturz_ ), as though everywhere breaking out of its borders ( _br\u00e4chte aus allen seinen R\u00e4ndern_ ), as if seeing us from every part of itself. These verbs describe the presence of what is inner in what is outer, in a way that is typical in Rilke's mature work. As Ryan puts it, \"the exploration of the relation between inner and outer realities, an attempt to transfigure loss, and an understanding of the aesthetic as a phenomenon of displacement are all characteristic features... of Rilke's later writing.\" The speaker's gaze moves from (absent) head down the front of the torso, following the arc of the glowing surface from prowlike breast to curving loins to genitals. The image of a smile in the turn of the loins is drawn, Ryan notes, \"from an essay by Mallarm\u00e9 on Th\u00e9odore de Banville, whom Mallarm\u00e9 regards as the supreme lyric poet of his time.\" The image of the turned-back ( _zur\u00fcckgeschraubt_ , as in a gas jet turned back from flickering in order to provide steady illumination) candelabra is, Ryan suggests, \"a recollection of Mallarm\u00e9's sonnet 'Le Tombeau de Charles Baudelaire.'\" Together, these two images emphasize the presence of controlled, animating, procreative power showing itself in the statue's surface. The statue fragment is intensely expressively present, so that it serves as a standing rebuke to us, who fail to bring our own personality, intelligence, and expressive and sexual powers to full embodied expression but instead live at second hand, palely under conventions that lack full life for us. Hence the fragment rebukes us for failing to be what we dimly feel we might be and ought to be as possessors of unexpressed inner intelligence and power: more fully animate, more fitly ensouled.\n\nAnd yet the poem is itself a classical sonnet, with an octave rhyming abba cddc followed by a sestet rhyming eef gfg. In place of a classical turn or volta after the octave, however, there are two turns: in line 5, with a move into the subjunctive in order to clarify and deepen the initial sense of the fragment's glowing, and then in line 14, with the sudden and brutal ascription of quasi-agentive sight to the fragment, issuing in a rebuke to the viewer-subject, who falls under its gaze and judgment. This rebuke, felt by the viewer-subject and addressed first to himself and thence to us, his readers, is startling. But the eef gfg scheme of the sestet houses this rebuke in a structure of strong formal coherence, giving a sense of appropriateness and closure to the experience. Through the tightly controlled form and images, the rebuke is earned by the experience as it is registered in the poem itself. The poem itself, that is to say, strikes us, through its form and images, as a composed, animated, ensouled whole. As Ryan notes, in the first poem, \"Early Apollo,\" of _New Poems: First Part_ , Rilke himself equates the overwhelming effect of an Apollo statue with the overwhelming effect of poetry in general. \"There is nothing in the head, the speaker says, ' _was verhindern k\u00f6nnte, da\u00df der Glanz aller Gedichte uns fast t\u00f6dlich tr\u00e4fe_ ' (which could prevent us from being almost fatally wounded by the radiance of the poems).\" The poem both rebukes us, its readers, in the way that the fragment has rebuked the viewer-subject, and shows us concretely that the housing of expressive power in controlled surface is still possible and commanding for us, even after the loss of the older dispensations.\n\nFor the poet, and for us who follow and share in the poet's experience, first of the fragment and then of the poem itself as constructed yet as it were living object, it thus remains possible for experience to mean something, possible to have an adequate idea of an affection, with full investment in one's responses to things, at least in principle. The trouble is that we are mostly too half-hearted to take this possibility seriously, perhaps half-hearted because the late modern world is so thoroughgoingly complex and opaque. Forms of technological and social expertise multiply and diverge, so that it becomes harder to sustain a sense of how they do or might form a meaningful whole. They seem detached or misplaced ( _entstellt_ ), and coping and compromise seem inevitably to displace the pursuit of wholeheartedness. It may be a part of wisdom simply to accept much of this, much of the time. And yet: formal coherence blended with original power is possible and compelling. This poem exists, and its force can be felt, perhaps at some time by many and often by some: you must change your life.\n\nTo be sure, this poem is in a way a fiction. It does not report a material reality that is independent of subjectivity and discerned through practices of measurement. Rather it tells a story about an experience and its significance, where the terms of significance involve a sense of emplotment and possibility in human life that are not simply given in tradition or simple sense experience. That sense of emplotment and possibility is itself felt, both by the poet initially and subsequently by us who follow the poet's feeling and thinking, as shaped or contrived in human time, just as first the fragment and then the poem have been shaped or contrived: we, like the poet, must construct it. Yet this sense is also felt as inevitable, present, and altogether other than arbitrarily invented: it is commanded of us in our contrivings by something that makes itself manifest in the formal and thematic working through of experience. In this working through, both the emplotment of this experience and the relation of this particular emplotment to a larger emplotment of human life are both constructed and accepted as given, by the poet and by us. This half-constructed, half-given experience of emplotment does not, however, admit of being fully unpacked and generalized into a master plot of human life as such. Perhaps either the unanticipatedly other and new will inevitably disrupt any settled life policy; perhaps nature and passion are the always present unruly birthplaces of any civil order. No recipe for how one is to change one's life so as to achieve expressive power is on offer. Instead the speaker (by the statue) and we (first by the statue, insofar as we share the we-position of the speaker, and then by the poem) are stopped and reminded that something better, we know not what, at least not in specific detail, haunts and draws us. The initial we-position of the speaker is both reinforced, insofar as anyone may take the statue and the poem as admonishment and provocation, and suspended, as the force of any injunction is felt by an individual viewer or reader bereft of any directions about how to follow it objectively and with others. A \"we\" is posed but also suspended as expressive intensity is urged on us as individuals, and modern life remains a scene of both banality and unparameterized possibility.\n\nPerhaps we should not call what we get from deeply absorbing, cathartic, yet contingency-acknowledging workings through of experience knowledge. Even framing the issue about the role of literature in our lives in terms of knowledge as it is construed paradigmatically in the natural sciences expresses the theoretical philosopher's characteristic bad faith in wanting everything circumscribed and life guided by rationally obligatory rules. Yet we cannot live as human persons without this literature; what we get from it is a sense of life in a human reality that is, if marked by brute contingency, not everywhere dominated by it. Achievement of the free, powerful, and coherent ordering of the materials of one's life is possible\u2014if not wholly and continuously, nonetheless occasionally and exemplarily. A form of life that needs a highly charged modernist reminder of this possibility is a form of life whose energies are in danger of flagging and whose preoccupations are less than fulfilling as they stand. A form of life that contains such reminders and contains readers capable of responding to them is a form of life in which hope remains alive as hope. Rilke may be taken to have known that his concluding injunction would be experienced as brutal, as difficult and all but unreceivable, and hence to have known that it is a wager that his writing can find or forge its audience. Against Rilke, perhaps it is right that, often, we should refuse the wager, as complexity, compromise, and common decency make cowards of us all. The brutalities of art's claims on us in the modern world should not be flattened into comfortable niceties. Yet it is not clear either that they always are or should be refused. The rest is silence: I cannot live to hear the news from England.\n6.\n\n\"New Centers of Reflection Are Continually Forming\"\n\n_Benjamin, Sebald, and Modern Human Life in Time_\n\nI\n\nIn a poignant passage in the introduction to his _Lectures on Fine Art_ (1820), Hegel describes how, in his view, human subjects express themselves in the world through practical activity in order to recognize themselves.\n\nMan brings himself before himself by _practical_ activity, since he has the impulse, in whatever is directly given to him, in what is present to him externally, to produce himself and therein equally to recognize himself. This aim he achieves by altering external things whereon he impresses the seal of his inner being and in which he now finds again his own characteristics. Man does this in order, as a free subject, to strip the external world of its inflexible foreignness and to enjoy in the shape of things only an external realization of himself. Even a child's first impulse involves this practical alteration of external things; a boy throws stones into a river and marvels at the circles drawn in the water as an effect in which he gains an intuition of something that is his own doing.\n\nPoignant though this passage is, one may nonetheless wonder at its claims. Is it the first impulse of human subjects, as subjects, to seek recognition of themselves and of the reasonableness of their doings in relation to external things? How dominant is this impulse in comparison with other impulses? (Many problems of survival, of coping with life, and of satisfying one's wants seem to have little on the face of it to do with seeking self-recognition.) Worse yet, how far is it genuinely possible to win self-recognition or to gain \"an intuition of something that is [one's own] doing\"? Perhaps such intuitions are only relatively fleeting in the face of the chaos and press of life, and perhaps they are available also only to the few who have \"enough\"\u2014enough means, time, and training\u2014and who live in good enough societies in which to carry out their efforts at self-expression and self-recognition. Perhaps there are other things that many people care about more fundamentally than they care about gaining a stable intuition of something that is one's own free and reasonable doing.\n\nHegel concedes that the life of Spirit, which includes at least the lives of human subjects in historical time, is marked by \"tarrying with the negative\" and by \"death and devastation.\" Establishing or expressing anything requires negation and its working through, always; the life of Spirit is continually reforming itself in historical time. As Stephen Houlgate notes, \"Hegel's philosophy... contains within itself a principle of aesthetic and religious _resistance_ to its own 'totalizing claims.'\" Whatever their capacities for conceptually structured self-regulation and satisfaction within meaningful social roles, human beings remain also sensuous beings who stand in some need of feeling imaginatively that life makes (enough) sense. Moreover, as Terry Pinkard remarks, the shapes of our lives are not simply given, according to Hegel. Rather, \"we _come to be_ the kinds of agents we are,\" and so bear, at least potentially, at each moment \"a 'negative' stance toward ourselves\": we could become something we are, so far, not. This negative stance or open stance toward future possibility then \"inflicts a kind of 'wound,' a _Zerissenheit_ , a manner of being internally torn apart that demands healing,\" as we seek greater self-unity and more meaningful satisfaction. Or as Hegel himself puts it, \"in the spiritual nature of man duality and inner conflict burgeon, and in their contradiction he is tossed about.\"\n\nAnd yet, according to Hegel, philosophy, together with the life of the Spirit\u2014human life in time\u2014with which it is interwoven, \"proceeds to the cancellation\" of this opposition.\" Hegel argues that we are historically on the cusp of a time in which \" _basically everyone_ \" will be able \"to satisfy [his or her] knowledge and volition... within the actuality of the state,\" that is, within the framework of the modern democratic nation-state and its institutions of right. Basically all subjects will be able to see reflections of themselves in their doings, in ways that are simultaneously individual and yet stable and reasonable under modern institutions. Structural social revolution is and should be a thing of the past, and we may reasonably claim to live in large measure in reconciliation with actuality.\n\nAfter the horrors of the twentieth and twenty-first centuries, however, and with awareness of the persistence of problems of poverty and radical inequality, it scarcely requires much perceptiveness to wonder whether this is really possible. When one then considers further the insights of Marx and Althusser and of Nietzsche and Freud into social life and into standing pressures on individual psyches, then the prospects for a shared life of right\u2014a community of reciprocal respect and recognition among free subjects who freely lead lives that are meaningful and reasonable under shared social institutions\u2014seem dimmer yet. Even if a system of liberal civil rights is a relatively good idea for maintaining a degree of social peace and affording subjects a measure of liberty, no routes of direct political action promise to lead to a life of full freedom and right. Nor does theoretical-representational knowing as it is rigorously pursued in the experimental and mathematical sciences much point to solving problems of self-recognition and reciprocal recognition in worldly practice. We might then conclude that human subjects either do not or should not much care about the pursuit of recognition\u2014the pursuit, that is, of stable senses, maintained by others and by themselves, of their own doings as individually chosen, free, reasonable, and satisfying. Yet that conclusion then makes it hard to see how human subjects engage in either self-cultivation or the cultivation of their social relations at all. These forms of cultivation\u2014the sheer existences of cultures\u2014do seem to embody the pursuit of some sort of expressive power, coherence, and means for the development and recognition of stable, reasonable, and satisfying identity. And yet such pursuits seem always to founder, perhaps inevitably to founder to some extent.\n\nWalter Benjamin's writings on language, history, and culture offer us one way to think about how human beings live with the all but impossible task of the pursuit of recognition. Unlike Hegel, Benjamin begins from the thought that fullness of recognition\u2014what would amount to a paradise of free and meaningful life on earth\u2014is unavailable either on the basis of history as we have inherited it or through any particular specifiable efforts, political, cultural, or theoretical. In an essay written in 1916 but unpublished in his lifetime, Benjamin formulates the unavailability of utopia through or from present courses of culture by contrasting the fullness of the meaning of things to and for God with the standing failure to engage with things (and with each other) in fully meaningful ways that characterizes all human culture. He contrasts God's meaningful naming of things, bound up with understanding them in their proper places and interrelations, with what he calls the human overnaming of things, bound up with our taking them specifically this way or that, as objects of particular use, say, without living in full \"resonance\" with them. \"Things have no proper names except in God. For in his creative word God called them into being, calling them by their proper names. In the language of men, however, they are overnamed.... The naming word in the knowledge of man must fall short of the creative word of God.\" God, who made things freely and in accordance with fullness of understanding, knows and engages with things according to their proper natures and significances. The ways of knowing and engaging with things that is characteristic of human beings are contrasted with the way of God. No matter, then, what one thinks about the existence of God, the point must be that the ways of knowing and engaging with things that are characteristic of human beings are, all of them, one-sided and reflective of conflicts over what things are and how they are to be used. Culture\u2014which is the embodiment of ways of conceiving of things and of making use of them\u2014as it is made by human beings perpetuates one-sidedness and conflict rather than resolving them. Overnaming or living within particular mimetic circuits of conception and use is a form of overspecification that does not let things be in their full significance and prevents human beings from engaging with either materiality or with one another in ways that embody fullness of meaning. Overnaming is, therefore, \"the deepest reason for all melancholy and (from the point of view of the thing) for all deliberate muteness. Overnaming as the linguistic being of melancholy points to another curious relation of language: the overprecision that obtains in the tragic relationship between the languages of human speakers.\"\n\nTo say that relations between languages are tragic and characterized by overprecision is to say that some dimension of what is meant in a given language is somehow missed in its rendering in a second language, with tragic consequences. And that in turn must be because, to some extent, the ways of conceiving of and engaging with things opened up within a given language are not fully available for expression in a second language. All actual languages are marked by overprecise falling away from the fullness of meaning and attention embodied in the divine language. Given, further, that the boundaries of language identity may be as narrow as those of an individual speaker's idiolect, Benjamin's picture is of human subjects generally failing to some degree to understand how other human subjects \"take\" things and engage with them, hence failing to some degree to understand what other human subjects are doing, what might be reasonable and sensible.\n\nThis picture need by no means be taken as a Whorfian picture of subjects as somehow wholly sealed off from one another by boundaries of language and culture that are impermeable to translation. There is some good reason to think that some considerable success in translation and in understanding what others are up to is always possible in relation to any being whom we can recognize as a language-deploying and thinking subject at all. But conflict nonetheless remains. There are standing practical difficulties among subjects in undertaking fully to understand how other subjects \"take\" things, what they are up to in engaging with them, and, so, what a free, meaningful, and reasonable life together under common \"takings\" might be like. There is, for us, no standpoint available outside the partiality that attaches to any point of view that any of us might occupy as finite subjects, no way to see things \"whole\" and untainted by overnaming within the always partly particularized mimetic circuits that one has inherited and developed. Practical problems of human relationship persist. That is the thought captured in Benjamin's contrast between God's calling things by their proper names and human overnaming.\n\nOur ways of taking things conceptually and of engaging with them\u2014no matter how widely shared, pragmatically useful, and historically sensible they may be\u2014are not grounded on any grasp of ultimate, sempiternal reality. Something, some fullness of significance, is somehow missed in any way of taking and engaging with things (or with one another). And so, \"within all linguistic formation [and within all repertoires of culture] a conflict is waged between what is expressed and expressible and what is inexpressible and unexpressed.\" We are, always, failing to mean fully and transparently to others and to ourselves everything that might reasonably be meant in engaging with things and so failing also to live fully and transparently and meaningfully with others according to reason. Our languages and cultural practices ultimately express these facts, at least in certain moments in certain pockets of use. Experience is, therefore, not a matter only of simple classification of objects under concepts, without remainder, and it is not a matter of human subjects each fully knowing what they and each other are up to, according to articulated good reasons. Experience is rather a form of human life in which, sometimes, things happen unpredictably, coincidentally, and yet in such a way that unarticulated significances are displayed and felt. From within the repressed history of a language, something can make itself felt, and present overprecision can be disturbed. Such disturbances can also appear when one undertakes (but in some measure fails at) literary translation or through dislocations of one's ordinary habits of perception, as in travel.\n\nMichael Rosen usefully characterizes Benjamin's understanding of experience as including a sense of the existence of \"unseen affinities\": relations among things or significances of things that are somehow missed within our ordinary articulations of what we are up to (within our ordinary \"overnamings\") but that nonetheless sometimes make themselves manifest. As Rosen puts it, \"'unseen affinities' [such as the 'passion for roulette' in relation to 'the vogue for panoramas'], referring, as they do, to a subterranean level of awareness, are not such as, immediately and unambiguously, to strike the uninstructed observer; and yet it is their existence that provides Benjamin's concept of experience with its only possible verification.\" There are, that is to say, within experience, ways of taking things and ways of engaging with things that seem to disrupt ordinary, articulated, planned ways of thinking, living, and working, and that in disrupting them seem to show exactly how specifically partial those ordinary ways are, just insofar as they are bound up with specific mimetic circuits of overnaming. Attention to these moments of surprise and disruption functions, then, not only as a verification of Benjamin's concept of experience as always partial and fallen but also as a reminder of our finitude and of what we have failed to achieve in the way of fullness of significance within the predominant parts of daily cultural life. Since such moments are always possible\u2014since some surprises and disruptions always remain unrecuperated to transparent and reasonable social life, no matter how it develops\u2014there can be no Hegelian recipe in the face of their permanent presence within life for achieving full human freedom in cultural life according to reason. For this reason, Benjamin characterizes his own method as that of evidencing the disruptive rather than that of prescribing the normative, as that of showing rather than saying. \"Method of this work: literary montage. I have nothing to say\u2014only to show.\"\n\nII\n\nEarly reviewers and critics of W. G. Sebald's books have not been slow to notice the burden of sadness carried by his form of literary attention to life. Something in human life is not going well. Disruption, distraction, threat, and anxiety\u2014all modes of failing easily to settle in routines of given cultural life\u2014are all but omnipresent. Anthony Lane calls attention to Sebald's focusing on the happenstantial, on whatever disrupts smoothness of emplotment, in remarking that Sebald \"raised modesty to the brink of metaphysics.\" Susan Sontag finds in Sebald \"a mind in mourning\" somehow on behalf of us all, for experience failing to become fully and transparently meaningful. This mind expresses itself in a \"laconically evoked mental distress\" that embodies \"a mysterious surplus of pathos\" that is \"never solipsistic.\" Franz Loquai notes that Sebald's writing is marked by a sense that \"the actuality of experience is apparently not to be trusted\"; that is, things that actually and effectively happen or are to happen according to life plans that a traditional plot might track are continuously being disrupted by the unplanned, the surprising, and the sheerly contingent.\n\nFrank Farrell explains the existence of disruptions in part as a result of the fact that Sebald's narrator figure (a character sometimes called W. G. Sebald) is an emigrant, as are both many of the living figures whom he encounters and many of the historical figures upon whom he reflects. In addition to geographic emigration, both the narrator figure and many of those whom he encounters seem to suffer from a kind of developmental or psychic emigration. Childhood, with its parental figures superintending a round of daily rituals, has somehow been left behind, and adulthood seems to offer no relations and routines of comparable stability and sureness of significance. Sebald's world, as Farrell puts it, \"is unable to incarnate the present meanings of an ongoing life because of the need for a ritualized return of what can no longer come back.\"\n\nOne might be tempted to find either the Sebald figure or the characters whose wanderings are narrated uninteresting, on the ground that their senses of loss and their mournings are somehow pathological in being determined by a failure to form stable adult attachments into which most people need not and do not fall. But exactly how clear is it that we genuinely do better in a world dominated by getting and spending, commodity exchange, cultural slippage, and the fragmentation of work? Sebald's figures occupy \"a world of ruins and absences, with no features immanent to it that suggest any possibility of renewal.\" If we reject the relevance to us of this world and the figures within it, then we shall have somehow to sustain a sense of the existence immanently within our world of features that do suggest renewal, and that is not obviously so easy to do.\n\nMark R. McCulloh focuses similarly on Sebald's sense of outsiderliness in the face of cultural habits that seem to lack significance. \"What Sebald does is display openly, from the perspective of a wandering outsider who happens to have certain literary leanings, the very oddness of people, of history and its calamities, of the very predicament of being alive. Sebald's subject... is in the last analysis the unsettling strangeness of the familiar.\" In the face of this unsettling strangeness, the best that one can do, Sebald suggests, is to bear witness to it and to the traumas that somehow lie behind it. This witness, however, includes a sense, as in Benjamin, that traumas are intrinsic to historical life as such; no overcoming of trauma through either political revolution or cognition is available. Sebald himself as a critic characterized the work of the early twentieth-century German novelist Alfred D\u00f6blin as offering \"an exact illustration of the new concept of history, which is not based on the idea of progress, as the old bourgeois concept was, but on the notion of self-perpetuating catastrophe.\"\n\nEric L. Santner describes a \"'poetics of exposure' that would become the signature style and method of Sebald's fiction.\" Human subjects are exposed to damage, trauma, loss, and in general to failures to form stable and fulfilling attachments. The lives of human subjects hence seem to have something in common with certain other \"privileged materials and objects\" that recur in Sebald's fiction: \"dust, ash, moth, bones, flayed skin, silk.\" Disintegration, fragility, and decay are substantially more prominent than integration, construction, and progress, both for human subjects and for material things.\n\nSantner traces Sebald's poetics of exposure, sense of the fragility of human life, and feeling for the pervasiveness of trauma in history both to the conditions of subject or ego formation in general and to the specific shape of those conditions in late commodity society. He suggests that Sebald shares Benjamin's sense of our \"irreducible exposure to the violence of history.\" This exposure makes allegory (or at least the kind of deliberately overstylized, ritualized, \"bald,\" antieschatological, unparsable allegory characteristic of the seventeenth-century German Trauerspiel as Benjamin understood it) \"the symbolic mode proper to\" our experience, and it makes melancholy\u2014lingering in a persistent sense of damage, trauma, and loss\u2014the appropriate tonality for serious writing that would register the deepest tenor of human experience.\n\nSantner argues that the continuing presence of trauma in human life in such a way that only damaged, outsiderly, less than wholehearted subjects are formed is due initially to the basis of civilization in the renunciation of parricide, as Freud described that basis in _Totemism and Taboo_. There is originally, at least for adolescent male subjects, a primordial impulse to parricide, so that they may displace the primal father in the tribe and come to enjoy a position of unrivaled possession and enjoyment of its women. At some point, however, this parricidal impulse is renounced for the sake of social peace but therein also displaced and continued in a distorted form. \"The _renunciation_ of the parricidal impulse (along with the fantasy of absolute jouissance entailed by the yearned-for position of the primal father) can be fully sustained only by a _compulsion to enjoy_ that same impulse, though at the significant symbolic remove of ritual performance.\" Given this renunciation and remove, male subjects come to exist in a damaged state of both permanent excitation and dissatisfaction. \"The primal horde pattern [of actual slaying of the primal father] and its 'mythic violence' are in some sense both sustained and suspended _in the same stroke_ \" with the introduction of settled civilization, superego formation and the renunciations it entails, and the law. With their lives as subjects founded on renunciations overlying excitations and unsatisfied wishes for perfect enjoyment, (male) subjects are incapable of full and stable enjoyments, and they perpetuate violence and trauma through their rivalries and competitive pursuits of objects that are in the end always in some measure unsatisfying. Open murder is renounced, but rivalry is not.\n\nThe damage continually wrought on and by (male) subjects is then exacerbated by life in modern commodity society, which displays a \"paradoxical mixture of deadness and excitation, stuckness and agitation,\" a \"nihilistic vitality,\" and \"surplus excitation and agitation.\" (Compare Wordsworth on the Bartholomew Fair: a \"perpetual whirl of \/ Of trivial objects, melted and reduced \/ To one identity, by differences \/ That have no meaning, and no end\" and Hegel on Civil Society: \"particularity... indulging itself in all directions as it satisfies its needs, contingent arbitrariness, and subjective caprice [so that it] destroys itself and its substantial concept in the act of enjoyment... infinitely agitated and continually dependent on external contingency.\") Nor are nonmale subjects freed from the circuits of damage that result from surplus excitation and agitation. In a world that contains circuits of damage and in which commodity production, acquisition, and exchange have displaced stable social relations and rituals, it is difficult for anyone to grow up into accomplished subjectivity bound up in stable and meaningful commitments.\n\nIt is no accident, then, that Santner wonders, against the background of this story about the development and plights of contemporary subjectivity, whether there is any room, in either the contemporary world or in Sebald's conception of it, for \"a shift in subjective dispositions\" that might result in more meaningful life. The prospects are not on the face of it encouraging.\n\nSebald is utterly uninterested in what we might call the \"new age\" solutions to the dilemma, that is, the various therapies and techniques that proliferate throughout contemporary culture for reducing stress, enhancing well-being, and optimizing the pleasure\/reality principle\u2014in a word, for _soothing_ the agitations of creaturely life. The relevant question with respect to Sebald is whether his way of constructing our historical situation leaves open the possibility of an event, a radical shift of perspective whereby something genuinely new could emerge.\n\nSantner himself offers two interrelated suggestions in response to this question. Building on work by Jonathan Lear on the possibilities of achieving a good enough psychoanalytic cure, that is, of coming to be able, as Freud is supposed to have remarked, \"to work and to love,\" Santner proposes that it is sometimes possible \"to catch a lucky break in life,\" that is, to become able to \"appropriate the 'the possibilities for new possibilities' that are, as [Lear] puts it, ' breaking out all the time.'\" Building on late work by Jacques Lacan, Santner adds that the possibility of a lucky break in forming relations to other subjects may be significantly held open by the fact that women, though shaped in part by traumas attendant upon subject formation, are, in Lacan's term, \"Not all,\" \"not wholly determined by [the phallic function].\" As a result, there are modes of enjoyment and of investment in activity and in subject-subject relations open to women that are not so readily available to male subjects. They may have more diffuse enjoyments and investments in activity and in relationships that are not so obsessively marked by rivalry and the playing out of displaced aggression, and human subjects in general may hope that such enjoyments and investments may spread out more widely within both personal and social life.\n\nThe suggestion that the miracle of a lucky break might happen, specifically that it might happen through the agency of women in forming other modes of subject-subject relation, is by no means unimportant, and it perhaps captures well the sort of forward-looking adaptiveness in forming and maintaining relationships that figures in the personal and occupational affairs of relatively normal and happy people. When it comes to Sebald, however, this suggestion faces a number of difficulties. Though chance and coincidence abound, relatively happy, forward-looking characters capable of adaptation do not much figure in Sebald's fiction. Nor are women much present in his writings, and anything resembling a marriage plot is entirely absent. Nor does this suggestion by itself capture the work of Sebald's style and form of attention to life, which remain considerably more melancholic than celebratory of luck. Most important, this suggestion does not really address either the continuing dynamics of trauma that underlie subject formation within settled social life or the particular shape those dynamics have taken in advanced commodity society.\n\nSantner's second suggestion builds on Terry Eagleton's reading of Benjamin, who likewise addressed the problem of \"the possibility of an event.\" In response to the plights of subject development in contemporary life, we might, Eagleton proposes, either regress to an imaginary past, remain marooned in melancholia, or somehow, while remembering the traumatic, nonetheless \"re-channel desire from both past and present to the future: to detect in the decline of the aura the form of new social and libidinal relations, realizable by revolutionary practice.\" This suggestion, however, raises the questions of exactly how the rechanneling of desire is possible and whether Sebald's works present any plausible models for such a rechanneling. Just how and where are the forms of new social and libidinal relations to be detected? There is no hint of any turn toward revolutionary praxis in Sebald. Nor do any rechannelings of desire seem visibly present in the itineraries of Sebald's principal figures, unless, somehow, something like this, in a muted form, is accomplished within the consciousness of the narrator figure. This narrator figure manages somehow to go on with life, despite an omnipresent melancholy and consciousness of trauma, and it is possible that the possibility of going on is thereby somehow opened up to us as well. And here Santner suggests that there is in Sebald's writing \"the performance of acts of witnessing\" that express a \"love of neighbor\" that functions as \"the 'miraculous' opening of a social link\" first between the narrator figure and those whom he encounters and then, further, between us and like figures in our worlds. This love of neighbor in the form of witness offers us \"the resources for intervening in and supplementing the superego bind\" that haunts subject formation and that motivates the perpetuation of trauma.\n\nThe acts of witnessing Sebald carries out in and through his narrator figure are not simply a journalistic reporting of evident injury, loss, and suffering. Daily newspapers and local television news reports are already replete with such reportings, and they function within the commodity space of news reporting more to titillate, entertain, and numb their audiences than to mobilize fullness of attention. Instead, Santner argues, Sebald's attention to life takes the form of what Benjamin called \" _erstarrte Unruhe_ , petrified unrest.\" Like the figure of the halted traveler in Wordsworth, or like Rilke frozen before the commanding sculpture of Apollo's torso, the narrator figure finds himself stopped and plunged into reflection and feeling. Attention is suddenly held all but obsessively by something simultaneously strange and indecipherable yet altogether ordinary.\n\nThis combination of strangeness and indecipherability with ordinariness is the signature of the uncanny ( _das Unheimliche_ ) as Freud conceived of it. Sebald himself quotes and endorses Benjamin's remark that \"histrionic or fanatical stress on the mysterious side of the mysterious takes us no further; we penetrate the mystery only to the degree that we recognize it in the everyday world.\" Mark McCulloh observes that \"it is this restoration of a sense of the uncanny (as well as of the sublime) to everyday experience that accounts for much of Sebald's appeal.\" To see something\u2014something strange, indecipherable, and yet strangely familiar within the ordinary\u2014and then to dwell on this something in reflection\u2014not to explain it, but to follow and play out the sense of strangeness, familiarity, and significance\u2014is to reanimate one's sense of life as a human subject. It is to remind oneself, in detail, that one is capable of noticing and feeling and of sustaining attention to the strange phenomena of life in time. Such a reminder joins the narrator figure with those whom he encounters who have suffered traumas of which they are uncannily reminded and also with readers who are brought to their own sense of the presence of the traumatic and of the strangely familiar within their lives.\n\nThis form of attention to human life is a small thing. It is not an explicit praxis of economic or political-institutional life, it is neither the achievement of a happy marriage or partner-relation nor the story of one, and it is not therapy that immediately adjusts one to a workplace or to family life. But it is a small thing that might run through and renovate any of the politicoeconomic praxes, human relationships, or modes of work and family life one might take up. Without the reanimation of subject attention, there is only compulsive repetition, unthinking habit, surpluses of excitation and agitation, anomie, and dullness. With it there is a chance to lead a human life more actively, more expressively, and with more wholeheartedness of interest.\n\nIII\n\nIt is not easy to characterize stylistically, formally, technically, or linguistically the nature of the fullness of attention some writers achieve and in which we can participate by reading. Different works written in different historical circumstances and with different subject matters will have strikingly different ways formally to achieve fullness of attention. One way, however, to begin a characterization is to note the differences of fullness of literary attention from both theorizing or discursive-classificatory thinking that makes use of preexistent categories, on the one hand, and more or less instantaneous intuition or perception, on the other. This, in fact, is exactly how Benjamin, is his masterpiece essay \"The Concept of Criticism in German Romanticism,\" both explains and extends Friedrich Schlegel's conception of the achievement of literature as an art. Schlegel, Benjamin writes, found it\n\nnecessary... to seek a mediation between discursive thinking and intellectual intuition, since the one did not satisfy his imperative of intuitive comprehension, whereas the other failed to satisfy his systematic interests. He thus found himself... faced with the problem of combining the maximum systematic range of thought with the most extreme truncation of discursive thinking.... He searches for a noneidetic intuition of the system, and he finds this in language. (139\u2013140)\n\nThe system in question here is the system of achieved freedom or of a more fully human way of life, involving full and stable care, reflection, and investment in activity. To say that intellectual intuition fails to satisfy an interest in this system is to say that such a way of life is not simply there to be grasped in either an instantaneous intellectual intuition of the whole or in a moment of blinding perception of the actual. But more temporally extended discursive-theoretic characterization of such a way of life remains, for Schlegel, abstract, or something whose availability and worth we are unable to feel directly. So one needs not a theory of freedom, the right or the good, but a more intuitive yet also temporally extended comprehension, achieved in relation to feeling, of what is possible and valuable for us.\n\nSchlegel, then, described a certain use of language as a vehicle of attention in the successful literary work. The task of criticism is to follow and participate in the deployment of attention within the work. In so conceiving of criticism, Benjamin argues,\n\nSchlegel's concept of criticism achieve[d] freedom from heteronomous aesthetic doctrines, [and] it made this freedom possible in the first place by setting up for artworks a criterion other than the rule\u2014namely, the criterion of an immanent structure specific to the work itself. He did this not with the general concepts of harmony and organization which, in the case of Herder or Moritz, were incapable of establishing a criticism of art, but with a genuine theory of art... as a medium of reflection and of the work as a center of reflection. (155)\n\nThere are, that is to say, no a priori knowable forms, use of which is either necessary or sufficient for success in art. Nor is there any definite content that is required. As Benjamin remarks, \"the concept of measure is remote from Romanticism, which paid no heed to an a priori of content, something to be measured in art\" (184). Instead, \"the value of a work depends solely on whether it makes its immanent critique possible or not\" (159), that is, on whether it supports critical or readerly participation in its deployments of attention. Benjamin himself argues that this conception of the value of a work of art remains dominant for us today, even where it is contested by either staler classicisms, on the one hand, or by the vulgarizations of commodity valuation and psychobiographic cults of personality, on the other. Citing Flaubert and the Stefan George circle, Benjamin claims that \"the doctrine that art and its works are essentially neither appearances of beauty nor manifestations of immediately inspired emotions, but media of forms, resting in themselves, has not fallen into oblivion since the Romantics, at least not in the spirit of artistic development itself\" (177), where the forms in question are not those of classical rules or unities but forms of fuller attention.\n\nForm, then, does the work of reflection or of attention that blends thought with feeling.\n\nForm is the objective expression of the reflection proper to the work, the reflection that constitutes its essence. Form is the possibility of reflection in the work. It grounds the work a priori, therefore, as a principle of existence; it is through its form that the work of art is a living center of reflection. In the medium of reflection in art, new centers of reflection are continually forming.... The infinitude of art attains to reflection first of all only in such a center, as in a limiting value; that is, it attains to self-comprehension and therewith to comprehension generally. This limit-value is the form of presentation [ _Darstellungsweise_ ] of the individual work. On it rests the possibility of a relative unity and closure of the work in the medium of art, [even though] the work remains burdened with a moment of contingency. (156)\n\nThe relative unity and closure that Benjamin has in mind differ from the putatively absolute closure of a demonstrative argument, on the one hand, and the lack of closure that characterizes the merely incidental or episodic, on the other. Instead, the author's attention and interest are excited by an initiating scene or incident. In and through the act of writing, the writer imagines what might further happen, or what thoughts and feelings are in play in relation to the initiator, as well as how, exactly, to work out in words the presentation of initiator, consequents, and attendant thoughts and feelings. A material form of presentation is here achieved in relation to the initiator, as an immanent structure rather than as form imposed from without. A center of reflection is formed in the work in relation to the initiator. The relative closure and unity of the work are achieved when attention calms itself in a feeling of completeness, signaled in a sense that \"yes, it was all so.\" (Compare Herbert Marcuse's discussion of aesthetic unity, understood as culminating in the sense that \" _Es war doch so sch\u00f6n_ \"\u2014it was all so beautiful anyway; it made sense; the work has clarified the initiating scene, even in the absence of complete system of freedom, theory of value, or demonstrated moral.)\n\nCertain claims about the proper way of reading a literary work follow immediately from the conception of it as a material form of presentation of energies of attention, focused on an initiator and its consequents, in relation to further thoughts and feelings, in the face of the persistent onwardness of life. As Schlegel himself observed in his critique of Goethe's _Wilhelm Meister_ , \"it is fine and necessary to abandon oneself utterly to the impression a poetic work makes... and perhaps only in particular cases to confirm one's feeling through reflection and to raise it to the level of thought... and complete it. But it is no less necessary to be able to abstract from all that is particular, so that\u2014hovering\u2014one grasps the universal\" (as cited in Benjamin, 153). In reading, a certain abandonment to the literary text\u2014to its energies, forms, and movements of attention\u2014will be necessary, as one follows reflection in the process of forming itself. But it will also be necessary sometimes, intermittently, to stand back in one's own reflections, so as to balance the movement of thought one has followed against life itself, as one reflects on it, and against other courses of embodied reflection. Within these two movements of reading and with the work and its writer one can then, sometimes, for a time, come to say and feel, \"yes, it was all so.\"\n\nIV\n\nIn order to track and grasp more concretely this mode of conclusion and its manner of achievement through literary form, we may turn to a particular case, W. G. Sebald's long story \"Paul Bereyter,\" the second of his four long stories published together as _The Emigrants_. Like the other stories in this collection, \"Paul Bereyter\" focuses on its single titular character, in this case as that character is considered by a first-person narrator who had once been Paul's student in grammar school. It is a kind of muted elegy or meditation, opening with the lines: \"In January 1984, the news reached me from S that on the evening of the 30th of December, a week after his seventy-fourth birthday, Paul Bereyter, who had been my teacher at primary school, had put an end to his life. A short distance from S, where the railway track curves out of a willow copse into the open fields, he had lain himself down in front of a train\" (27\/41).\n\nThis first sentence already embodies Sebald's striking personal style, somewhat more natural in German than in English. It begins with a prepositional phrase, the main verb is in the passive voice, two time indications delay the appearance of the noun subject \"Paul Bereyter\" of the main dependent clause that gives the news, and again a relative clause stands between that subject-noun phrase and its verb phrase. The second sentence continues the interrupting focus on details, as two place indications, the second in explicatory apposition to the first and offering the perceptible details of the curving track and the willow copse, precede an independent clause in the past perfect. In the German, in fact, these two sentences are one long sentence, with the dramatic \"sich... vor den Zug legte\" coming only at the very end. In addition, the clauses in German are connected by explicatory conjunctions that the English omits: \" _thus_ a week after his 74th birthday\" (\"also eine Woche nach seinem 74. Geburtstag\") and \"in that he had...\" (\" _indem_ er sich\"). The railway line leads itself out in a curve (\"in einem Bogen... herausf\u00fchrt\") and then attains the open field (\"das offene Feld gewinnt\") almost as though it, too, were a character. These two sentences are accompanied by a somewhat blurry black-and-white photograph, showing in the foreground, where it occupies about half of the picture plane at its front edge, a single rail, with to the far right edge a dark companion rail curving along with it toward the right, away from trees on the left and toward a field. It is all, already, almost unbearably evocative and melancholy. The effect of the style and the photograph is one of delay _in_ details that, we may presume, have some significance in holding the narrator's attention, though this significance is not spelled out: rather, these details are to accumulate\u2014both in the narrator's consciousness and in the consciousness of we who follow his consciousness\u2014until they form a pattern whose significance can almost, but not quite, be explicated in a moral or secret key to the story. Instead of a moral or secret key, what the narrator gets, and what we get, is a more nearly unverbalizable sense of the pattern and of the pathos, and the beauty and fragility amid the pathos, that this life (Paul Bereyter's life) and human life (both the narrator's life and our lives) all embody.\n\nFollowing this opening sentence, the narrative plays out in this style through roughly nine further scenes, as the narrator attempts to come to terms with Paul Bereyter's life, more or less as follows.\n\n(1) The narrator notes that the obituary in the local newspaper from S fails to mention \"that Paul Bereyter had died of his own freewill\" ( _aus freien St\u00fccken_ ) (27\/42) and that, besides describing his dedication to his pupils, his inventiveness as a teacher, and his love of music, it \"added, with no further explanation [ _In einer weiter nicht erl\u00e4uterten Bemerkung_ ], that during the Third Reich Paul Bereyter had been prevented from practicing his chosen profession\" (27\/42). As a result of the manner of death and of \"this curiously unconnected, inconsequential statement\" (\"Diese g\u00e4nzlich unverbundene und unverbindliche Feststellung\") (27\/42), the narrator concerns himself more and more with Paul Bereyter (\"mich... immer h\u00e4ufiger mit Paul Bereyter besch\u00e4ftige\"), resolving to \"get beyond\" his own fond memories of him in order to find out more about his secret history.\n\n(2) The narrator returns to S, where he has been only occasionally since leaving school, in order to visit Paul's apartment and to talk with the villagers. He remembers how the students, like everyone else in the village, had spoken of their teacher simply as \"Paul\" (28\/43), and he imagines Paul lying on his balcony, skating in winter, and stretched out on the track (29\/44). Or, rather, \"I saw him on the airy balcony, his face vaulted over by the host of stars\" (\"Ich sah ihn liegen auf dem geschindelten Altan, seiner sommerlichen Schlafstatt, das Gesicht \u00fcberwolbt von den Heerz\u00fcgen der Gestirne\") (29\/44). These investigations and imaginings, however, do not bring him any closer to Paul, except in a few \"emotional moments that seemed presumptuous to me\" (\"in gewissen Ausuferungen des Gef\u00fchls [overflowings of feelings], wie sie mir unzul\u00e4ssig [inadmissible, forbidden] erscheinen\") (29\/45). As a result, he has now written down \"what I know of Paul Bereyter\" (\"zu deren Vermeidung ich jetzt aufgeschrieben habe, was ich von Paul Bereyter wei\u00df und im Verlauf meiner Erkundigen \u00fcber ihn in Erfahrung bringen konnte\") (29\/45).\n\n(3) The narrator describes his family's move from W to S, \"19 kilometers away\" (32\/45), and his joining Paul Bereyter's third class. He recalls his friendship with Fritz Binswanger, a slow boy who exactly shared the narrator's \"incorrigibly sloppy handwriting\" (\"unverbesserlich schweinisch Handschrift\") (31\/47). They study cockchafer beetles together and share lunches. Once they each receive a present of \"a white butterpear\" (32\/49). Fritz later became a chef of \"international renown\" (32\/49). The narrator and Fritz later meet in London, in 1984, \"in the reading room of the British Museum, where I was researching the history of Bering's Alaska expedition and Fritz was studying eighteenth-century French cookbooks\" (32\/49).\n\n(4) The narrator describes the layout of the classroom in S, with \"twenty-six desks screwed fast to the oiled floorboards\" (33\/50\u201351). A sketch accompanies this description. Paul's bearing and teaching style are described. He often stood not at the front but \"in one of the window bays towards the head of the room half facing the class and half turned to look out, his face at a slightly upturned angle with the sunlight glinting on his glasses; and from that position he would talk across to us\" (34\/52). He spoke \"in well-structured sentences\" \"without any touch of dialect but with a slight impediment of speech or timbre, as if the sound were coming not from the larynx but from somewhere near the heart\" (34\u201335\/52). Paul's freethinking in religion and his \"aversion to hypocrisy of any description\" (\"die Abneigung Pauls gegen alles Scheinheilige\") are described (36\/55). He did not attend church. Instead of using the prescribed text, he taught from a collection of stories, \"the _Rheinische Hausfreund_ \" (37\/56) that he \"had procured, I suspect at his own expense\" (37\/56\u201357). He spoke fluent French. He emphasized natural history, and he often took the class on visits to interesting sites: a brewery, a gunsmith's, a castle, and an abandoned coal mine. He played the clarinet and was strikingly good at whistling, favoring melodies that the narrator only later recognized as by Brahms and Bellini (41\/61). He once brought to the class a young conservatory violinist, and Paul was \"far from being able to hide the emotion that [the] playing produced in him [and] had to remove his glasses because his eyes had filled with tears\" (41\/62). Often \"he might stop or sit down somewhere, alone and apart from us all, as if he, who was always in good spirits and seemed so cheerful, was in fact desolation itself\" (\"die Untr\u00f6stlichkeit selber\") (42\/62). This teaching style, and these and other incidents, are strikingly close to what we know of Wittgenstein's career as a rural schoolteacher, and Wittgenstein is mentioned later in the text.\n\n(5) The narrator describes what he has learned about Paul from Lucy Landau, who now lives in the Villa Bonlieu in Yverdon, Switzerland, and who had arranged Paul's burial in S. Since his retirement from teaching in 1971, Paul had mostly lived in Yverdon. He and Lucy had met each other at Salin-les-Bains in the French Jura, where she had been reading Nabokov's autobiography on a park bench (43\/65). Some of Lucy's own childhood in Switzerland is recounted. Paul had explained to her in Salin-les-Bains that his \"condition\" and his \"claustrophobia\" had now made him unable to teach (43\/65). His condition included now seeing \"his pupils, although he had always felt affection for them (he stressed this), as contemptible and repulsive creatures [ _ver\u00e4chtliche und hassenswerte Kreaturen_ ], the very sight of whom had prompted an utterly groundless violence in him on more than one occasion\" that he had felt break out in him (43\u201344\/65\u201366). Paul, we are told Lucy said, \"was almost consumed by the loneliness within him [ _von seiner inneren Einsamkeit nahezu aufgefressenen_ ],\" though he was \"the most considerate and entertaining companion one could wish for\" (44\/66\u201367). In conversation, Paul \"had linked the bourgeois concept of Utopia and order... with the progressive destruction of natural life\" (45\/67). She herself, when gazing with Paul at Lake Geneva from the top of Montrond, \"had for the first time in her life... a sense of the contrarieties that are in our longings\" (\"die widerspr\u00fcchlichen Dimensionen unserer Sehnsucht\") (45\/68). Lucy explains to the narrator that Paul had earlier lived in France, from 1935 to 1939, and she gives the narrator an album of photographs and notes, kept by Paul, that covers \"almost the whole of [his] life\" (45\/68).\n\n(6) The narrator reports that \"since then I have returned to [the album] time and time again, because, looking at the pictures in it, it truly seemed to me, and still does, as if the dead were coming back, or as if we were on the point of joining them\" (46\/68\u201369). Various photographs from the album are reproduced in the narrator's story we are reading. The album and the photos tell of \"a happy childhood\" (46\/69) and years \"in a country boarding school\" (46\/69). Paul had submitted to the narrow-minded and morbidly Catholic demands of a teacher training school solely in order to be able to teach children (46\u201347\/69). In the summer following a year of probationary teaching in S in 1934\u20131935, Paul spent a good deal of time with Helen Hollaender from Vienna, \"an independent-spirited, clever woman\" whose \"waters ran deep\" and in which \"Paul liked to see his own reflection\" (48\/72), at least as Mme. Landau interprets the photographs to the narrator. In autumn 1935, Paul took up a teaching post \"in the remote village of W\" but was almost immediately dismissed \"because of the new laws\" (48\/72). Meanwhile, Helen had returned with her mother to Vienna, from where \"there could be little doubt that Helen and her mother had been deported, in one of those special trains that left Vienna at dawn, probably to Theresienstadt in the first instance\" (40\u201350\/73).\n\n(7) Paul too, Mme. Landau reports to the narrator, is one-quarter Jewish, the grandson of the Jewish merchant Amschel Bereyter from Gunzenhausen in Franconia, and the son of Theodor, who had trained in a department store in Nuremberg before opening his own shop in S (50\u201351\/75). \"In his childhood,\" Mme. Landau reports Paul to have said,\n\neverything in the emporium seemed far too high up for him, doubtless because he himself was small, but also because the shelves reached all the four metres up to the ceiling. The light in the emporium, coming through the small transom windows let into the tops of the display window backboards, was dim even on the brightest of days, and it must have seemed all the murkier to him as a child, Paul had said, as he moved on his tricycle, mostly on the lowest level, through the ravines between tables, boxes and counters, amidst a variety of smells\u2014mothballs and lily-of-the-valley soap were always the most pungent, while felted wool and loden cloth assailed the nose only in wet weather, herrings and linseed oil in hot. (51\/76)\n\nPaul's father Theo died of a heart attack on Palm Sunday, 1936, but perhaps also from \"the fury and fear that had been consuming him, ever since, precisely two years before his death, the Jewish families, resident in his home town of Gunzenhausen for generations, had been the target of violent attacks\" (53\/79). Even though it \"could not be 'Aryanized'\" officially, the shop had nevertheless to be sold for \"next to nothing,\" and Paul's mother Thela \"died within a few weeks\" (53\u201354\/79\u201380).\n\n(8) After it became \"no longer tenable\" (55\/81) for him as a German to serve as a tutor in France, Paul returned to Berlin in 1939 to work at an office job in a garage. A few months later he was called up, and he spent six years in the motorized artillery, serving on all three fronts. Under one photograph of himself from this period, Paul wrote that \"day by day, hour by hour, with every beat of the pulse, one lost more and more of one's qualities, became less comprehensible to oneself, increasingly abstract\" (56\/83). In 1945, \"a German to the marrow\" (\"von Grund auf\") (57\/84), Paul returned to S, \"which in fact he loathed... said Mme. Landau\" (57\/84), again to teach. He \"spent a lot of time gardening\" (57\/85). During this time, he read \"Altenberg, Trakl, Wittgenstein, Friedell, Hasenclever, Toller, Tucholsky, Klaus Mann, Ossietzky, Benjamin, Koestler, and Zweig: almost all of them writers who had taken their own lives or been close to doing so\" (58\/86). He \"copied out hundreds of pages\" into his notebooks, \"time and again... stories of suicides\" (58\/86), as though to convince himself \"that he belonged to the exiles and not to the people of S\" (59\/87\u201388). He retired in 1971 and thereafter lived principally in Yverdon, near Mme. Landau, where he devoted himself to his gardening and reading.\n\n(9) In 1982, Paul's vision once again began to deteriorate. In autumn 1983, he informed Lucy that he wished to give up his flat in S. \"Not long after Christmas\" (60\/89), they traveled to S together to settle affairs. \"No snow had fallen, there was no sign anywhere of any winter tourism.... On the third day a spell of mild _f\u00f6hn_ weather set in, quite unusual for the time of year. The pine forests were black on the mountainsides, the windows gleamed like lead, and the sky was so low and dark one expected ink to run out of it any moment\" (60\u201361\/89\u201390). While Mme. Landau was sleeping in the afternoon from a headache, Paul went out. Upon being informed of his death, she thought of the railway timetables and directories he had collected and of \"the M\u00e4rklin model railway he had laid\" (61\/91) out in his rooms. Hearing this, the narrator thinks of \"the stations, tracks, goods depots and signal boxes\" that he had as a child to copy from the blackboard in Paul's classroom (61\/91). Paul had told Mme. Landau of a summer holiday in his own childhood that he had spent watching trains pass \"from the mainland to the island and from the island to the mainland\" (62\/92). At that time, Paul's uncle had said he would \"end up on the railways\" (\"bei der Eisenbahn enden\") (63\/92). Though this struck her as \"darkly foreboding\" (\"er hatte auf mich die dunkle Wirkung eines Orakelspruchs\") (63\/93), Mme. Landau reported that \"the disquiet I experienced lasted only a very short time, and passed over me like the shadow of a bird in flight\" (\"ging \u00fcber mich hinweg wie der Schatten eines Vogels im Flug\") (63\/93).\n\nV\n\nThe action of this story, like the action of lyric, takes place entirely in memory, within the consciousness of the narrator and in the past tense. Also like lyric, the overall structural pattern of the story is out-in-out. That is, an initiating scene or incident in the world (here the news of Paul Bereyter's suicide and the subsequent newspaper account of it) prompts a course of memory, reflection, and further action, all of which are then recollected in the past tense of the narrative itself. The narrative then ends with a turn again out toward the world\u2014the image of Mme. Landau reporting the passing over her of her disquiet\u2014so as to let the world go its own way.\n\nThe recollective actions and attentions of the narrator model for us our attentions to details of our own lives, as we too seem sometimes to haunt the world, from within our reserves of loneliness, yet also seem sometimes to be bound up in things, without any clear sense of the forces or logic of loneliness and activity. As in lyric, we participate in the narrator's own recollections and attentions. In the case of the sixteen photographs and diagrams that appear within the story, we literally see what the narrator sees. The photographs and diagrams are chosen and placed by the author (who may or may not be the narrator), and it is not clear that they are in each case documentary in relation to the incidents of the story. Sometimes they seem documentary; sometimes they seem more general than that, to be deposited more for the sake of a mood or tonality that they evoke in relation to incidents than out of a documentary intention. This ambiguity heightens the sense of the uncanny and of \"unseen affinities\" between narrated events and some felt but scarcely verbalizable significances.\n\nSimilarly, the narrator himself, and we through the narrator, sometimes seem to see and feel what Paul or Lucy see and feel, especially, for example, in the long description of the details of Theo Bereyter's emporium, with its high windows, dim light, and pungent smells. It is easy for us here to recall the fuguelike actions of play (riding a tricycle down the aisles) and seemingly giant scale of objects of our own childhoods. And yet, stopped by the photographs, we seem sometimes to be thrown back on ourselves, seem to see only their mysteriously evocative black on white, detached from any narrative arc. With the narrator, and perhaps with anyone who reflects, we find ourselves left outside the plot, if there is a plot at all.\n\nA philosophical or theological theory might hope to describe the essence of the situation that we share, to some extent, with Paul, Lucy, and the narrator. A political or sexual history might hope to sort out and explain the relative influences on Paul, and on us, of Jewishness (or another religion) or of sexual longing of one or another shape, subsuming Paul and us under its generalizations. Such descriptions and generalizations might be apt. But something nonetheless would be missed in them: the intimate detail and density of consciousness and its movements in perception, as it finds itself now in this situation, now in that, struck by surprises and intensities that seem to resist full capture by either essential descriptions or subsumptive generalizations. A task of literature\u2014or at least of this kind of intensely lyrical and elegiac literature\u2014is to render some of these movements for our identification, thus enabling us, along with their narrators (and writers) to work them through, so as to be all at once ourselves, in our particular personalities, lonelinesses, and intensities of perception and recollection, and also in the world, able, in the end, to let it go its own way, with an appropriate sense of mystery and wonder at it, and at how one has been in it, but not, quite, ever altogether of it. The movement of working through is as important or more important than the events that are narrated. Sebald's literary technique heightens our awareness of this through his continual use of devices of interruption: narratively through shifts from the narrator's own investigations, to what Mme. Landau said, to what is actually in Paul's album; syntactically through the interjection of prepositional and appositional phrases, piling up details for perception, in between noun phrase subjects and verbs. These details invite reflections, both historical and philosophical; were there no such invitations, we would encounter only the incidental or episodic. But reflection is not allowed to settle into any definite metaphysical, sociohistorical, or psychoanalytic systems for interpretation. (One must, as Schlegel remarked, both abandon oneself to the poetry of the text and hover above it, seeking the universal. New centers of reflection are continually forming.) This kind of literature presents for our working through what this story has itself called \"a sense of the contrarieties that are in our longings\" (\"die widerspr\u00fcchlichen Dimensionen unserer Sehnsucht\") (45\/68), as we ourselves move through its details and in doing so reflect on them.\n\nThe story \"Paul Bereyter\" is headed by a motto, \"Manche Nebelflecken l\u00f6set kein Auge auf,\" well enough translated as \"There is mist that no eye can dispel\" (25\/39). In fact, however, this motto is a quotation from a quite special context: Jean Paul's _Vorschule der \u00c4sthetik_ , part 1, section 3, paragraph 14, entitled \"Instinct of Genius or the Matter of Genius\" (\"Instinkt des Genies oder genialer Stoff\"). The full passage runs as follows:\n\nMany godlike spirits have been impressed by destiny with a grotesque [ _unf\u00f6rmliche_ ] form, as Socrates had the body of a satyr; for time governs the form, but not the inner matter. Thus the poetic mirror with which Jakob B\u00f6hme rendered heaven and earth hung in a dark place; also in some places the glass lacks the foil. In this way the great Hamann is a deep heaven full of telescopic stars, and some nebula spots [ _Nebelflecken_ ] no eye can penetrate.\n\nThe immediate sense here is that Hamann's deep writings, writings that capture as it were the whole world and Hamann the man himself, have features that no one can understand. Discursive thought, seeking essential descriptions and subsumptive generalizations, will miss at least something of what they contain and present. These somethings can at best be looked on from a certain distance and with a certain awareness of one's own incomprehension. In the phrase \"in a dark place\" (\"in einem dunklen Orte\"), there is a further distant echo of Dante's \"in a dark wood\" in the opening lines of the _Inferno_ :\n\nMidway through this way of life we're bound upon,\n\nI woke to find myself in a dark wood,\n\nWhere the right road was wholly lost and gone.\n\nWe lead our lives in time and as finite subjects, where the relations between the form (the social shape) of a life and the internal matter (one's particular personality, feeling, and longing) remain always, to some extent, other than transparent. This lack of transparency shows itself especially in certain perplexing, initiating scenes and incidents and then in the emotionally modulating reflections that follow them, as one seeks in reflection more transparency, a better fit, or fuller attention\u2014until in the end life is allowed to go on, on its own. Literature\u2014some literature, this lyric literature of Sebald's\u2014knows this and makes it manifest for us. To see and feel this, and to see and feel it in detail, through perception and accompanying reflection, is to be, in a certain way, more fully seduced to life.\nAppendix\n\nLines Composed a Few Miles Above Tintern Abbey, on Revisiting the Banks of the Wye During a Tour, July 13, 1798\n\n_William Wordsworth_\n\nFive years have passed; five summers, with the length\n\n---\n\nOf five long winters! and again I hear\n\nThese waters, rolling from their mountain-springs\n\nWith a soft inland murmur.\u2014Once again\n\nDo I behold these steep and lofty cliffs,\n\n|  5\n\nThat on a wild secluded scene impress\n\nThoughts of more deep seclusion; and connect\n\nThe landscape with the quiet of the sky.\n\nThe day is come when I again repose\n\nHere, under this dark sycamore, and view\n\n|  10\n\nThese plots of cottage ground, these orchard tufts,\n\nWhich at this season, with their unripe fruits,\n\nAre clad in one green hue, and lose themselves\n\n'Mid groves and copses. Once again I see\n\nThese hedgerows, hardly hedgerows, little lines\n\n|  15\n\nOf sportive wood run wild; these pastoral farms,\n\nGreen to the very door; and wreaths of smoke\n\nSent up, in silence, from among the trees!\n\nWith some uncertain notice, as might seem\n\nOf vagrant dwellers in the houseless woods,\n\n|  20\n\nOr of some Hermit's cave, where by his fire\n\nThe Hermit sits alone.\n\nThese beauteous forms,\n\n|\n\nThrough a long absence, have not been to me\n\nAs is a landscape to a blind man's eye;\n\nBut oft, in lonely rooms, and 'mid the din\n\n|  25\n\nOf towns and cities, I have owed to them,\n\nIn hours of weariness, sensations sweet,\n\nFelt in the blood, and felt along the heart;\n\nAnd passing even into my purer mind,\n\nWith tranquil restoration\u2014feelings too\n\n|  30\n\nOf unremembered pleasure; such, perhaps,\n\nAs have no slight or trivial influence\n\nOn that best portion of a good man's life,\n\nHis little, nameless, unremembered, acts\n\nOf kindness and of love. Nor less, I trust,\n\n|  35\n\nTo them I may have owed another gift,\n\nOf aspect more sublime; that blessed mood,\n\nIn which the burthen of the mystery,\n\nIn which the heavy and the weary weight\n\nOf all this unintelligible world,\n\n|  40\n\nIs lightened:\u2014that serene and blessed mood,\n\nIn which the affections gently lead us on,\u2014\n\nUntil, the breath of this corporeal frame\n\nAnd even the motion of our human blood\n\nAlmost suspended, we are laid asleep\n\n|  45\n\nIn body, and become a living soul;\n\nWhile with an eye made quiet by the power\n\nOf harmony, and the deep power of joy,\n\nWe see into the life of things.\n\nIf this\n\nBe but a vain belief, yet, oh! how oft\u2014\n\n|  50\n\nIn darkness and amid the many shapes\n\nOf joyless daylight; when the fretful stir\n\nUnprofitable, and the fever of the world,\n\nHave hung upon the beatings of my heart\u2014\n\nHow oft, in spirit, have I turned to thee,\n\n|  55\n\nO sylvan Wye! thou wanderer through the woods,\n\nHow often has my spirit turned to thee!\n\nAnd now, with gleams of half-extinguished thought,\n\nWith many recognitions dim and faint,\n\nAnd somewhat of a sad perplexity,\n\n|  60\n\nThe picture of the mind revives again;\n\nWhile here I stand, not only with the sense\n\nOf present pleasure, but with pleasing thoughts\n\nThat in this moment there is life and food\n\nFor future years. And so I dare to hope,\n\n|  65\n\nThough changed, no doubt, from what I was when first\n\nI came among these hills; when like a roe\n\nI bounded o'er the mountains, by the sides\n\nOf the deep rivers, and the lonely streams,\n\nWherever nature led: more like a man\n\n|  70\n\nFlying from something that he dreads than one\n\nWho sought the thing he loved. For nature then\n\n(The coarser pleasures of my boyish days,\n\nAnd their glad animal movements all gone by)\n\nTo me was all in all.\u2014I cannot paint\n\n|  75\n\nWhat then I was. The sounding cataract\n\nHaunted me like a passion; the tall rock,\n\nThe mountain, and the deep and gloomy wood,\n\nTheir colours and their forms, were then to me\n\nAn appetite; a feeling and a love,\n\n|  80\n\nThat had no need of a remoter charm,\n\nBy thought supplied, nor any interest\n\nUnborrowed from the eye.\u2014That time is past,\n\nAnd all its aching joys are now no more,\n\nAnd all its dizzy raptures. Not for this\n\n|  85\n\nFaint I, nor mourn nor murmur; other gifts\n\nHave followed; for such loss, I would believe,\n\nAbundant recompense. For I have learned\n\nTo look on nature, not as in the hour\n\nOf thoughtless youth; but hearing oftentimes\n\n|  90\n\nThe still, sad music of humanity,\n\nNor harsh nor grating, though of ample power\n\nTo chasten and subdue. And I have felt\n\nA presence that disturbs me with the joy\n\nOf elevated thoughts; a sense sublime\n\n|  95\n\nOf something far more deeply interfused,\n\nWhose dwelling is the light of setting suns,\n\nAnd the round ocean and the living air,\n\nAnd the blue sky, and in the mind of man:\n\nA motion and a spirit, that impels\n\n|  100\n\nAll thinking things, all objects of all thought,\n\nAnd rolls through all things. Therefore am I still\n\nA lover of the meadows and the woods,\n\nAnd mountains; and of all that we behold\n\nFrom this green earth; of all the mighty world\n\n|  105\n\nOf eye, and ear\u2014both what they half create,\n\nAnd what perceive; well pleased to recognize\n\nIn nature and the language of the sense\n\nThe anchor of my purest thoughts, the nurse,\n\nThe guide, the guardian of my heart, and soul\n\n|  110\n\nOf all my moral being.\n\nNor perchance,\n\nIf I were not thus taught, should I the more\n\nSuffer my genial spirits to decay:\n\nFor thou art with me here upon the banks\n\nOf this fair river; thou my dearest Friend,\n\n|  115\n\nMy dear, dear Friend; and in thy voice I catch\n\nThe language of my former heart, and read\n\nMy former pleasures in the shooting lights\n\nOf thy wild eyes. Oh! yet a little while\n\nMay I behold in thee what I was once,\n\n|  120\n\nMy dear, dear Sister! and this prayer I make,\n\nKnowing that Nature never did betray\n\nThe heart that loved her; 'tis her privilege,\n\nThrough all the years of this our life, to lead\n\nFrom joy to joy: for she can so inform\n\n|  125\n\nThe mind that is within us, so impress\n\nWith quietness and beauty, and so feed\n\nWith lofty thoughts, that neither evil tongues,\n\nRash judgments, nor the sneers of selfish men,\n\nNor greetings where no kindness is, nor all\n\n|  130\n\nThe dreary intercourse of daily life,\n\nShall e'er prevail against us, or disturb\n\nOur cheerful faith, that all which we behold\n\nIs full of blessings. Therefore let the moon\n\nShine on thee in thy solitary walk;\n\n|  135\n\nAnd let the misty mountain winds be free\n\nTo blow against thee: and, in after years,\n\nWhen these wild ecstasies shall be matured\n\nInto a sober pleasure; when thy mind\n\nShall be a mansion for all lovely forms,\n\n|  140\n\nThy memory be as a dwelling-place\n\nFor all sweet sounds and harmonies; oh! then,\n\nIf solitude, or fear, or pain, or grief\n\nShould be thy portion, with what healing thoughts\n\nOf tender joy wilt thou remember me,\n\n|  145\n\nAnd these my exhortations! Nor, perchance\u2014\n\nIf I should be where I no more can hear\n\nThy voice, nor catch from thy wild eyes these gleams\n\nOf past existence\u2014wilt thou then forget\n\nThat on the banks of this delightful stream\n\n|  150\n\nWe stood together; and that I, so long\n\nA worshipper of Nature, hither came\n\nUnwearied in that service; rather say\n\nWith warmer love\u2014oh! with far deeper zeal\n\nOf holier love. Nor wilt thou then forget,\n\n|  155\n\nThat after many wanderings, many years\n\nOf absence, these steep woods and lofty cliffs,\n\nAnd this green pastoral landscape, were to me\n\nMore dear, both for themselves and for thy sake!\nNotes\n\n_1. Introduction: Subjectivity, Modernity, and the Uses of Literature_\n\n. If one is interested in fixing the extension of \"literature\" in such a way that all plausible cases are covered, one will end up with something like the radically disjunctive definition offered by Robert Stecker: \"A work _w_ is a work of literature if and only if _w_ is produced in a linguistic medium, and, (1) _w_ is a novel, short story, tale, drama, or poem, and the writer of _w_ intended that it possess aesthetic, cognitive or interpretation-centered value, and the work is written with sufficient technical skill for it to be possible to take that intention seriously, or (2) _w_ possesses aesthetic, cognitive, or interpretation-centered value to a significant degree, or (3) _w_ falls under a predecessor concept to our concept of literature and was written while the predecessor concept held sway, or (4) _w_ belongs to the work of a great writer\" (\"What is Literature?\" _Revue Internationale de Philosophie_ 50 [1996]: 694). This may be correct enough, but it offers little illumination about how literature achieves aesthetic, cognitive, or interpretation-centered value or about how the achievement of such values matters for human life.\n\n. This is roughly the view of Wayne Booth in _The Company We Keep: An Ethics of Fiction_ (Berkeley: University of California Press, 1988). Books are like friends who help us to widen the range of points of view we can occupy. While this suggestion has some truth to it, it underrates the formal, cognitive, and affective intensities of literary structure and distinctively literary craft.\n\n. Bill Readings usefully surveys the rise and decline of the humanities within the modern university from Humboldt to the present as central disciplines devoted to teaching national literatures in _The University in Ruins_ (Cambridge, Mass.: Harvard University Press, 1997).\n\n. Frank Ferrell, _Why Does Literature Matter?_ (Ithaca, N.Y.: Cornell University Press, 2004). See 213 ff. for a summary of the \"modern\" developments listed in the remainder of this paragraph.\n\n. Ibid., 201.\n\n. Farrell summarizes the registers of recovery that literary works offer us on ibid., 9\u201319.\n\n. Ibid., 187.\n\n. Ibid., 72.\n\n. David E. Wellbery, _The Specular Moment: Goethe's Lyric and Early Romanticism_ (Stanford, Calif.: Stanford University Press, 1996), 48.\n\n. The complete text of \"Maifest\" in German and in English translation appears on 28\u201329 of ibid.\n\n. Ibid., 49.\n\n. Ibid.\n\n. Ibid., 3.\n\n. Eric L. Santner, _On Creaturely Life_ (Chicago: University of Chicago Press, 2006), 33.\n\n. J. M. Bernstein, _Against Voluptuous Bodies: Late Modernism and the Meaning of Painting_ (Stanford, Calif.: Stanford University Press, 2006), 45.\n\n. Ralph Waldo Emerson, \"Circles,\" in _Emerson: Essays and Lectures_ (New York: Viking Press, 1983), 406.\n\n. J\u00fcrgen Habermas, _The Philosophical Discourse of Modernity_ , trans. Frederick G. Lawrence (Cambridge, Mass.: The MIT Press, 1987), 8.\n\n. Ibid., 9.\n\n. Charles Baudelaire, \"The Painter of Modern Life,\" in Baudelaire, _Selected Writings on Art and Artists_ , trans. P. E. Charvet (Harmondsworth: Penguin, 1972), 392, cited in Habermas, _The Philosophical Discourse of Modernity_ , 9.\n\n. Bernstein, _Against Voluptuous Bodies_ , 11.\n\n. Philip Weinstein powerfully and usefully surveys the registering of trauma and resistance to overly stable Bildungsroman plots on the parts of Kafka, Proust, and Faulkner in his _Unknowing: The Work of Modernist Fiction_ (Ithaca, N.Y.: Cornell University Press, 2005).\n\n. Tzachi Zamir powerfully traces various forms of this slippage as they are expressed in the careers of the protagonists of the major Shakespearean tragedies in his _Double Vision: Moral Philosophy and Shakespearean Drama_ (Princeton, N.J.: Princeton University Press, 2007).\n\n. See Michael Tomasello, _The Cultural Origins of Human Cognition_ (Cambridge, Mass.: Harvard University Press, 2001), and _Constructing a Language: A Usage-Based Theory of Language Acquisition_ (Cambridge, Mass.: Harvard University Press, 2005). For a summary of this work that brings it into connection with Wittgenstein's _Philosophical Investigations_ , see Richard Eldridge, \"Wittgenstein on Aspect-Seeing, the Nature of Discursive Consciousness, and the Experience of Agency,\" in _Seeing Wittgenstein Anew: New Essays on Aspect-Seeing_ , ed. William Day and Victor Krebs (Cambridge: Cambridge University Press), forthcoming.\n\n. This is the main argument, as I read it, of Wittgenstein's _Philosophical Investigations_. See Eldridge, _Leading a Human Life: Wittgenstein, Intentionality, and Romanticism_ (Chicago: University of Chicago Press, 1997).\n\n. J. L. Austin, \"Truth,\" in Austin, _Philosophical Papers_ , 2nd ed. (Oxford: Oxford University Press, 1970), 126 n. 1.\n\n. Ren\u00e9 Descartes, _The Search After Truth by the Light of Nature_ , trans. E. S. Haldane and G. R. T. Ross, in _The Philosophical Works of Descartes_ , ed. E. S. Haldane and G. R. T. Ross (Cambridge: Cambridge University Press, 1911), 1:305.\n\n. Kant, \"What Is Enlightenment?\" trans. Lewis White Beck, in Kant, _On History_ , ed. Lewis White Beck (Indianapolis, Ind.: Bobbs-Merrill, 1963), 3 n. 1.\n\n. Here I follow Stanley Cavell's reading, itself developed in response to Emerson's readings of Descartes and Kant, of Descartes' account in the _Meditations_ of his knowledge of his own existence. See Cavell, \"Being Odd, Getting Even (Descartes, Emerson, Poe)\" in Cavell, _In Quest of the Ordinary: Lines of Skepticism and Romanticism_ (Chicago: University of Chicago Press, 1988), 105\u2013130.\n\n. Descartes, _Meditations on First Philosophy_ , in _The Philosophical Works of Descartes_ , 1:150.\n\n. Ibid., 1:151\u2013152.\n\n. Cavell, _The Senses of Walden: An Expanded Edition_ (San Francisco: North Point Press, 1981), 107.\n\n. Henry David Thoreau, _Walden_ , ed. J. Lyndon Shanley (Princeton, N.J.: Princeton University Press, 2004), 159.\n\n. Cavell, _The Senses of Walden_ , 107\u2013108.\n\n. Kant, \"Idea for a Universal History from a Cosmopolitan Point of View,\" in Kant, _On History_ , 22.\n\n. Kant, _Lectures on Ethics_ , trans. Louis Infield (Gloucester: Peter Smith, 1978), 140.\n\n. Kant, \"What Is Enlightenment?\" 3.\n\n. Thomas Pfau, \"The Voice of Critique: Aesthetic Cognition After Kant,\" part 1, in _Romantic Circles Praxis Series_ , available online at <http:\/\/www.rc.umd\/edu\/praxis\/philosophy\/pfau1\/tp1.html>.\n\n. Aristotle, _Poetics_ , trans. Stephen Halliwell (London: Gerald Duckworth, 1987), excerpted in _The Philosophy of Art: Readings Ancient and Modern_ , ed. Alex Neill and Aaron Ridly (New York: McGraw-Hill, 1995), 490. A slightly different wording appears in Aristotle, _Poetics_ , trans. Stephen Halliwell, in Aristotle, _Poetics_ , together with Longinus, _On the Sublime_ , and Demetrius, _On Style_ (Cambridge, Mass.: Harvard University Press, 1995), 37.\n\n. Adorno, _Negative Dialectics_ , trans. E. B. Ashton (New York: Continuum, 1973), 12; emphasis and interjection added.\n\n. Ibid., 13.\n\n. Adorno, _Aesthetic Theory_ , trans. Robert Hullot-Kentor (Minneapolis: University of Minnesota Press, 1997), 131.\n\n. Catherine Wilson, \"Literature and Knowledge,\" _Philosophy_ 58 (1983). Reprinted in _Philosophy of Literature_ , ed. Eileen John and Dominic McIver Lopes (Oxford: Blackwell, 2004), 327.\n\n. Charles Altieri, _The Particulars of Rapture_ (Ithaca, N.Y.: Cornell University Press, 2003).\n\n. Ludwig Wittgenstein, _Philosophical Investigations_ , 2nd. ed., trans. G. E. M. Anscombe (New York: The Macmillan Company [1953], 1958), \u00a7122, 49e. See also Richard Eldridge, \"Hypotheses, Criterial Claims, and Perspicuous Representations: Wittgenstein's 'Remarks on Frazer's _The Golden Bough_ ,'\" in Eldridge, _The Persistence of Romanticism_ (Cambridge: Cambridge University Press, 2001), 127\u2013144.\n\n. See chapter 5, note 9.\n\n. Richard Eldridge, _An Introduction to the Philosophy of Art_ (Cambridge: Cambridge University Press, 2003), 259 and passim.\n\n. Ibid., 260.\n\n. Wolfgang Huemer, \"Introduction: Wittgenstein, Language, Philosophy of Literature,\" in _The Literary Wittgenstein_ , ed. John Gibson and Wolfgang Huemer (London: Routledge, 2004), 5.\n\n. Ibid., 6\u20137.\n\n. See Spinoza, _Ethics_ , in Spinoza, _Selections_ , ed. John Wild (New York: Charles Scribner's Sons, 1930), part 5, propositions 3\u201310, pp. 369\u2013377. Compare also both R. G. Collingwood on artistic expression, _The Principles of Art_ (Oxford: Oxford University Press, 1938), 282\u2013283; and William Wordsworth, \"Preface to Preface to _Lyrical Ballads_ ,\" in _Selected Poems and Prefaces_ , ed. Jack Stillinger (Boston: Houghton Mifflin Company, 1965), 448, on how a poet may uncover \"what is really important to men\" through thinking \"long and deeply\" in relation to our feelings.\n\n. William Rothman and Marian Keane, _Reading Cavell's_ The World Viewed (Detroit, Mich.: Wayne State University Press, 2000), 19.\n\n. A wonderful essay on these parallel identifications is Ted Cohen's \"Identifying with Metaphor: Metaphors of Personal Identification,\" _The Journal of Aesthetics and Art Criticism_ 57, no. 4 (1999): 399\u2013409, on identifying with Lily Bart, Jake Gittis, and Marlowe, as well as Shakespeare, Mozart, and Conrad.\n\n. Plato, _Republic_ , trans. G. M. A. Grube, revised C. D. C. Reeve (Indianapolis, Ind.: Hackett Publishing Company, 1992), 348b, p. 24.\n\n_2. Romanticism, Cartesianism, Humeanism, Byronism: Stoppard's_ Arcadia\n\n. John Dewey, _Art as Experience_ (New York: Perigee, 1980 [1934]), 337\u2013338.\n\n. G. W. F. Hegel, _Philosophy of Right_ , trans. T. M. Knox (Oxford: Oxford University Press, 1967), \u00a7185, p. 123.\n\n. See Plato, _Republic_ , trans. G. M. A. Grube (Indianapolis, Ind.: Hackett Publishing Company, 1992), book 7, 557a\u2013563e, pp. 227\u2013234.\n\n. Northrop Frye, _The Great Code: The Bible and Literature_ (New York: Harcourt Brace Jovanovich, 1983), 5\u201314.\n\n. Rene Descartes, preface to _The Search After Truth by the Light of Nature_ , in _The Philosophical Works of Descartes_ , trans. Elizabeth S. Haldane and G. R. T. Ross (Cambridge: Cambridge University Press, 1931), 1:305.\n\n. Frye, _The Great Code_ , 52.\n\n. Donald G. Marshall, \"Foreword: Wordsworth and Post-Enlightenment Culture,\" in Geoffrey H. Hartman, _The Unremarkable Wordsworth_ (Minneapolis: University of Minnesota Press, 1987), vii.\n\n. Charles Larmore, _The Romantic Legacy_ (New York: Columbia University Press, 1996).\n\n. See, for example, G. W. F. Hegel, _Hegel's Introduction to Aesthetics_ , trans. T. M. Knox (Oxford: Oxford University Press, 1979), 81.\n\n. See Jerome J. McGann, _The Romantic Ideology_ (Chicago: University of Chicago Press, 1983).\n\n. David Hume, _An Enquiry Concerning Human Understanding_ , ed. Eric Steinberg (Indianapolis, Ind.: Hackett Publishing Company, 1977), section 12, p. 111.\n\n. W. V. O. Quine, \"On the Nature of Moral Values,\" in Quine, _Theories and Things_ (Cambridge, Mass.: Harvard University Press, 1981), 61.\n\n. Charles Taylor, _The Sources of the Self_ (Cambridge, Mass.: Harvard University Press, 1989), 151, 154.\n\n. Descartes, _Discourse on Method_ , in _Discourse on Method and Meditations_ , trans. Donald A. Cress (Indianapolis, Ind.: Hackett Publishing Company, 1980), 14.\n\n. Ibid., 14, 15.\n\n. Ibid., 13.\n\n. Bertrand Russell, _A History of Western Philosophy_ (New York: Simon and Schuster, 1945), 747.\n\n. Byron, George Gordon, Baron, \"Childe Harold's Pilgrimage\" in _Byron's Poetry_ , ed. Frank D. McConnell (New York: W. W. Norton, 1978), Canto the First, II, p. 26.\n\n. Ibid., Canto the First, V, p. 26.\n\n. _Tom Stoppard's Arcadia: A Study Guide from Gale's Drama for Students_ , vol. 5, chap. 2, e-text PDF document (Farmington Hills, Mich.: The Gale Group, 2002), DOI 10.1223\/GALFSDSF0000074, p. 48.\n\n. See Gary Gutting, _Pragmatic Liberalism and the Critique of Modernity_ (Cambridge: Cambridge University Press, 1999).\n\n. William Wordsworth, _The Prelude_ [1850], in Wordsworth, _Selected Poems and Prefaces_ , ed. Jack Stillinger (Boston: Houghton Mifflin Company, 1965), XIV, 446\u2013447, p. 366.\n\n. Ibid., I, 302, p. 199.\n\n. Ibid., II, 233\u2013234, p. 212.\n\n. Wordsworth, \"The Sublime and the Beautiful,\" in _The Prose Works of William Wordsworth_ , ed. W. J. B. Owen and J. W. Smyser (Oxford: Clarendon, 1974), 2:349\u2013360, at 2:357.\n\n. Wordsworth, \"Essay, Supplementary to The Preface\" [1815], in Wordsworth, _Selected Poems and Prefaces_ , 471\u2013481, at 477.\n\n. Wordsworth, _The Prelude_ , IX, 8, p. 304.\n\n. Harold Bloom, \"The Internalization of Quest Romance,\" in _Romanticism and Consciousness_ , ed. H. Bloom (New York: W. W. Norton, 1970), 3\u201324.\n\n. Tom Stoppard, in Mel Gussow, _Conversations with Tom Stoppard_ (New York: Grove Press, 1995), 91.\n\n. Stoppard, cited in _Tom Stoppard's Arcadia: A Study Guide from Gale's Drama for Students_ , 56.\n\n. Stoppard, _Conversations with Tom Stoppard_ , 3.\n\n. Ibid., 14.\n\n. Ibid., 74.\n\n. J. L. Austin, _Sense and Sensibilia_ (Oxford: Oxford University Press, 1962), 2.\n\n. See the unfavorable review by John Simon, cited in _Tom Stoppard's Arcadia: A Study Guide from Gale's Drama for Students_ , 54.\n\n. See the reviews cited in _Tom Stoppard's Arcadia: A Study Guide from Gale's Drama for Students_ , 34, 54\u201355, 56, 59.\n\n. Tim Appelo, \"Review of _Arcadia_ ,\" _The Nation_ , Jan. 5, 1995, reprinted in Tom _Stoppard's Arcadia: A Study Guide from Gale's Drama for Students_ , 54A.\n\n. Stoppard, _Arcadia_ (London: Faber and Faber, 1993), II, 5, p. 62.\n\n. Ibid., II, 7, p. 76.\n\n. Ibid., II, 7, p. 84.\n\n. Ibid., I, 2, p. 33.\n\n. Ibid., II, 7, pp. 75, 76.\n\n. Herbert Marcuse, _The Aesthetic Dimension_ (Boston: Beacon Press, 1978), 60.\n\n. Theodor W. Adorno, \"On Lyric Poetry and Society,\" in Adorno, _Notes to Literature_ , trans. S. W. Nicholsen (New York: Columbia University Press, 1991), 1:41.\n\n. Samuel Beckett, _The Unnamable_ , in Beckett, _Three Novels: Molloy, Malone Dies, The Unnamable_ (New York: Grove Press, 1995), 414.\n\n_3. Romantic Subjectivity in Goethe and Wittgenstein_\n\n. See M. W. Rowe, \"Goethe and Wittgenstein,\" _Philosophy_ 66 (1991); Joachim Schulte, \"Chor und Gesetz: Zur 'Morphologischen Methode' bei Goethe and Wittgenstein,\" _Grazer Philosophische Studien_ 21 (1984); and G. P. Baker and P. M. S. Hacker, _Wittgenstein: Understanding and Meaning_ (Oxford: Basil Blackwell, 1980), 537\u2013540. I summarize and comment on this work in my _Leading a Human Life: Wittgenstein, Intentionality, and Romanticism_ (Chicago: University of Chicago Press, 1997), 177\u2013181.\n\n. Thomas Mann, \"On Goethe's _Werther_ ,\" trans. Elizabeth Corra, in _The Sufferings of Young Werther and Elective Affinities_ , ed. Victor Lange (New York: Continuum, 1990), 2.\n\n. Charles Taylor, _Sources of the Self: The Making of the Modern Identity_ (Cambridge, Mass.: Harvard University Press, 1989), x.\n\n. Ibid., 18.\n\n. Philippe Lacoue-Labarthe and Jean-Luc Nancy, _The Literary Absolute: The Theory of Literature in German Romanticism_ , trans. Philip Barnard and Cheryl Lester (Albany: State University of New York Press, 1988), 31.\n\n. Ibid., 12. The subject term of Lacoue-Labarthe and Nancy's clauses is \"literary production,\" not \"articulation,\" but with the migration of human self-production toward the literary, in a mix of discovery and invention, the latter, more general term makes their characterizations appropriate to human moral efforts in general.\n\n. Rodolphe Gasch\u00e9, foreword to Friedrich Schlegel, _Philosophical Fragments_ , trans. Peter Firchow (Minneapolis: University of Minnesota Press, 1991), xix.\n\n. Nicholas Boyle, _Goethe: The Poet and the Age_ , vol. 1.: _The Poetry of Desire_ (Oxford: Clarendon Press, 1991), 176.\n\n. Ibid., 124.\n\n. Ibid., 110.\n\n. Ibid., 177.\n\n. Ibid., 162.\n\n. Ibid., 176.\n\n. Mann, \"On Goethe's _Werther_ ,\" 9.\n\n. Ibid., 9, 10.\n\n. Johann Wolfgang von Goethe, _The Sorrows of Young Werther_ , in _Goethe: The Collected Works_ , ed. David E. Wellbery, trans. Victor Lange (Princeton, N.J.: Princeton University Press, 1994), 11:72. Subsequent references to _Werther_ will all be to this edition and will be indicated in the text by page number.\n\n. Mann, \"On Goethe's _Werther_ ,\" 8.\n\n. Ray Monk, _Ludwig Wittgenstein: The Duty of Genius_ (New York: The Free Press, 1990), 25.\n\n. Brian McGuinness, _Wittgenstein, A Life: Young Ludwig 1889\u20131921_ (Berkeley: University of California Press, 1988), 156.\n\n. Ibid., 50.\n\n. Ibid.\n\n. Ibid., 156, citing Tolstoy, \"A Confession.\"\n\n. Ludwig Wittgenstein, _Culture and Value_ , trans. Peter Winch (Oxford: Basil Blackwell, 1980), 1e. Subsequent references to this work will be given by page numbers in parentheses.\n\n. This remark is about \"an ordinary conventional figure\" at the end of Schubert's \"Death and the Maiden,\" but it captures well Wittgenstein's attitude toward the manual work he repeatedly urged on others.\n\n. Goethe, _Die Leiden des jungen Werther_ (Stuttgart: Reclam, 1948), 147. Werther actually adds one more line to Wilhelm: \"wir sehen uns wieder und freudiger\" (147). But \"Lebt wohl!\" alone is what lives in the memory of his readers as his valedictory to life, particularly since his last diary entry, addressed to Lotte, concludes \"Es schl\u00e4gt zw\u00f6lfe. So sei es denn!\u2014Lotte! Lotte, lebe wohl! lebe wohl!\" (150).\n\n. Wittgenstein, _Philosophical Investigations_ , 3rd ed., trans. G. E. M. Anscombe (New York: Macmillan, 1958), \u00a7125, pp. 50, 50e. Subsequent references to this work will be given in the text by section number.\n\n. Wittgenstein, cited in Norman Malcolm, _Ludwig Wittgenstein: A Memoir_ (Oxford: Oxford University Press, 1958), 100.\n\n_4. Attention, Expressive Power, and Interest in Life: Wordsworth's \"Tintern Abbey\"_\n\n. Nietzsche, _The Birth of Tragedy and The Case of Wagner_ , trans. Walter Kaufmann (New York: Random House, 1967), \u00a71, 33.\n\n. While some may argue that Jesus is an ideal human being, he is at best ideal as a person and personification, not as 5'2\", eyes of blue.\n\n. Nietzsche, _The Birth of Tragedy_ , \u00a714, 91.\n\n. Nietzsche, _The Will to Power_ , trans. Walter Kaufmann (New York: Random House, 1968), 1050.\n\n. Nietzsche, _The Birth of Tragedy_ , \u00a71, 33.\n\n. Ibid., \u00a77, 59.\n\n. Ibid., \u00a77, 60.\n\n. Ibid.\n\n. Ibid.\n\n. Ibid., \u00a77, 58.\n\n. Ibid.\n\n. Aristotle, _Poetics_ , trans. Richard Janko (Indianapolis, Ind.: Hackett Publishing Company, 1987), 51b1, p. 12.\n\n. For a defense of this reading of Aristotle's account of the nature of the _hamartia_ or \"tragic flaw\" as an excess of virtue ill-suited to the circumstances of action, see Richard Eldridge, \"How Can Tragedy Matter for Us?\" in Eldridge, _The Persistence of Romanticism_ (Cambridge: Cambridge University Press, 2001), 146\u2013164.\n\n. Nietzsche, _The Birth of Tragedy_ , \u00a77, 58.\n\n. Ibid.\n\n. William Wordsworth, _The Prelude_ [1850], in Wordsworth, _Selected Poems and Prefaces_ , ed. Jack Stillinger (Boston: Houghton Mifflin Company, 1965), book 14, ll. 161\u2013162, p. 360.\n\n. Wordsworth, \"From _The Recluse_ \" [Prospectus], in _Selected Poems and Prefaces_ , l. 808, p. 46.\n\n. On Wordsworth's conjecturalism, see Eldridge, \"Internal Transcendentalism: Wordsworth and 'A New Condition of Philosophy,'\" in Eldridge, _The Persistence of Romanticism_ , 102\u2013123.\n\n. Wordsworth, _The Prelude_ , 1, ll. 267\u2013269, p. 199.\n\n. Wordsworth, \"Preface to the Second Edition of _Lyrical Ballads_ ,\" in Wordsworth, _Selected Poems and Prefaces_ , 446, 447. Subsequent references to the \"Preface\" will be given by page number in the text.\n\n. Stanley Cavell, \"The Philosopher in American Life,\" in Cavell, _In Quest of the Ordinary_ (Chicago: The University of Chicago Press,), 7.\n\n. Cavell, \"Emerson, Coleridge, Kant,\" in Cavell, _In Quest of the Ordinary_ , 36. See also Cavell, \"Texts of Recovery,\" 52\u201353.\n\n. Cavell, \"Being Odd, Getting Even,\" in Cavell, _In Quest of the Ordinary_ , 115\u2013116.\n\n. Cavell notes the interest of the formulation \"communicate with\" and its difference from \"communicate about\" in \"Texts of Recovery,\" 71\u201372.\n\n. David E. Wellbery, _The Specular Moment: Goethe's Early Lyric and the Beginnings of Romanticism_ (Stanford, Calif.: Stanford University Press, 1996), 39.\n\n. See ibid., 11.\n\n. Ibid., 55.\n\n. Wordsworth, _The Prelude_ , 14, ll. 439\u2013443, p. 366.\n\n. Wordsworth, _Alfoxden Notebook_ , 21v, in _The Ruined Cottage and the Pedlar_ , ed. James Butler (Ithaca, N.Y.: Cornell University Press, 1979), 125.\n\n. David S. Miall, \"Locating Wordsworth: 'Tintern Abbey' and the Community with Nature,\" _Romanticism on the Net_ 20 (November 2000), available online at <http:\/\/www.erudit.org\/revue\/ron\/2000\/v\/n20\/005949ar.html>, p. 1.\n\n. Cavell, \"Postscript B: Poe's Perversity and the Imp(ulse) of Skepticism,\" in Cavell, _In Quest of the Ordinary_ , 143.\n\n. Wordsworth, \"Lines Composed a Few Miles Above Tintern Abbey, On Revisiting the Banks of the Wye During a Tour, July 13, 1798,\" in Wordsworth, _Selected Poems and Prefaces_ , 108\u2013111. References to the poem will be to this edition and will be given by line number in the text.\n\n. David Bromwich, \"The French Revolution and 'Tintern Abbey,'\" _Raritan_ 10, no. 3 (Winter 1991): 1\u201323.\n\n. This suggestion is made most notably by Marjorie Levinson in \"Insight and Oversight: Reading 'Tintern Abbey,'\" in Levinson, _Wordsworth's Great Period Poems_ (Cambridge: Cambridge University Press, 1986), 14\u201357, esp. 37: \"the primary poetic action [of 'Tintern Abbey'] is the suppression of the social\" in favor of a 'fiercely private vision.'\" See also Jerome McGann, _The Romantic Ideology_ (Chicago: University of Chicago Press, 1983), 85\u201388.\n\n. Miall, \"Locating Wordsworth,\" 3.\n\n. Wordsworth, _The Prelude_ , 2, ll. 277\u2013281, p. 213.\n\n. For further discussion of this famous sentence from chapter 4 of Hegel's _Phenomenology of Spirit_ , trans. A. V. Miller (Oxford: Clarendon Press, 1977), para. 167, p. 105, see Eldridge, _Leading a Human Life: Wittgenstein, Intentionality, and Romanticism_ (Chicago: University of Chicago Press, 1997), 27\u201332.\n\n. Wordsworth, _The Prelude_ , 2, ll. 228\u2013232, p. 212.\n\n. Bromwich, \"The French Revolution and 'Tintern Abbey,'\" 8.\n\n. John Barrell, _Poetry, Language, and Politics_ (Manchester: Manchester University Press, 1988), 162.\n\n. Ibid.\n\n. I seem to recall having learned this interpretation of \"never did\" from reading Geoffrey Hartman, but I cannot now locate the reference.\n\n. John Dewey, \"Construction and Criticism,\" in _Later Works_ (Carbondale, Ind.: The Center for Dewey Studies, 1988), 5:125\u2013146. I thank Nikolas Kompridis for directing my attention to this remark.\n\n. Geoffrey H. Hartman, _Wordsworth's Poetry 1787\u20131814_ , 2nd. ed. (New Haven, Conn.: Yale University Press, 1971), xv.\n\n. Ibid., 190.\n\n. Ibid., 104.\n\n. Ibid.\n\n. Ibid., 218.\n\n. Ibid., 38.\n\n_5. The Ends of Literary Narrative: Rilke's \"Archaic Torso of Apollo\"_\n\n. Peter Lamarque and Stein Haugom Olsen, _Truth, Fiction, and Literature_ (Oxford: The Clarendon Press, 1996), 369\u2013378. Subsequent references to this work will be given in the main text by page numbers in parentheses.\n\n. Lamarque and Olsen are quoting Theseus, in Shakespeare, _A Midsummer Night's Dream_ , act 5, scene 1.\n\n. John Gibson, \"Reality and the Language of Fiction,\" in _Writing the Austrian Traditions: Themes in Philosophy and Literature_ , ed. Wolfgang Huemer and Marc-Oliver Schuster (Toronto: University of Toronto Press, 2003), 63. Subsequent references to this work will be given in the main text by page numbers in parentheses.\n\n. Gibson takes up these questions in much greater detail, in ways that fill in a story about human life in ways I find congenial, in his _Fiction and the Weave of Life_ (Oxford University Press, 2008) and \"Literature and Knowledge,\" in _Oxford Handbook of Philosophy and Literature_ , ed. Richard Eldridge (Oxford University Press, forthcoming). Both these pieces develop a version of the working-through conception that I am urging.\n\n. The cognitive developmental psychologist Michael Tomasello has recently developed a rich account of language learning as depending essentially on intention-reading in his _The Cultural Origins of Human Cognition_ (Cambridge, Mass.: Harvard University Press, 1999) and _Constructing a Language: A Usage-Based Theory of Language Acquisition_ (Cambridge, Mass.: Harvard University Press, 2003). His account builds in part on Wittgenstein's work on seeing-as in part 2 of _Philosophical Investigations_ , 3rd. ed., trans. G. E. M. Anscombe (New York: Macmillan, 1958). I survey the affinities between the views of Tomasello and Wittgenstein in \"Wittgenstein on Aspect-Seeing, the Nature of Discursive Consciousness, and the Experience of Agency,\" in _Seeing Wittgenstein Anew: New Essays on Aspect-Seeing_ , ed. William Day and Victor Krebs (Cambridge: Cambridge University Press, forthcoming). R. G. Collingwood treats language learning and concept learning in similar terms, as a matter of learning by interacting with others and how to attend to aspects, in Collingwood, _The Principles of Art_ (Oxford: The Clarendon Press, 1938), esp. 239\u2013241.\n\n. I take the idea that language _must_ be both _stable_ in providing us with ways of thinking of things that we use internally and unhesitatingly and _tolerant_ of new usages from Stanley Cavell, _The Claim of Reason_ (New York: Oxford University Press, 1979), 185\u2013186.\n\n. I take up the essential \"immigrancy\" involved in our inheritance of language and development of conceptual consciousness in \"Cavell and H\u00f6lderlin on Human Immigrancy,\" in Eldridge, _The Persistence of Romanticism_ (Cambridge: Cambridge University Press, 2001), 229\u2013245.\n\n. Collingwood, _The Principles of Art_ , 239. Collingwood is almost surely thinking here also of Freud's account of the development of the ego in and through plays of mutual attention and contestation. See Freud on the _fortda_ game in Freud, _Beyond the Pleasure Principle_ , trans. James Strachey (New York: Bantam Books, 1961).\n\n. See Spinoza, Benedict de, _Ethics_ , in Spinoza, _Selections_ , ed. John Wild (New York: Charles Scribner's Sons, 1930), part 5, propositions 3\u201310, pp. 369\u2013377.\n\n. Cavell develops his account of the truth of skepticism in various major writings, including \"Knowing and Acknowledging\" (1969), _The Claim of Reason_ (1979), \"Being Odd, Getting Even\" (1986), and _Conditions Handsome and Unhandsome_ (1990). For an overview of Cavell's thoughts about skepticism, see Richard Eldridge, \"'A Continuing Task': Cavell and the Truth of Skepticism,\" in Eldridge, _The Persistence of Romanticism_ , 189\u2013204.\n\n. Charles Altieri, _The Particulars of Rapture: An Aesthetics of the Affects_ (Ithaca, N.Y.: Cornell University Press, 2003), 107.\n\n. William Wordsworth, \"Preface to _Lyrical Ballads_ ,\" in _Selected Poems and Prefaces_ , ed. Jack Stillinger (Boston: Houghton Mifflin Company, 1965), 448.\n\n. See Friedrich H\u00f6lderlin, \"On Religion,\" in H\u00f6lderlin, _Essays and Letters on Theory_ , trans. and ed. Thomas Pfau (Albany: State University of New York Press, 1988), 90\u201391.\n\n. John Dewey, _Art as Experience_ (New York: Penguin, 1980), esp. chapter 3, \"Having an Experience,\" and 17\u201319.\n\n. See Richard Eldridge, _Leading a Human Life: Wittgenstein, Intentionality, and Romanticism_ (Chicago: University of Chicago Press, 1997), 6\u20137; _The Persistence of Romanticism_ , 19\u201320, 55\u201357, 158\u2013163, 235; and _An Introduction to the Philosophy of Art_ (Cambridge: Cambridge University Press, 2003), 7\u201312, 262.\n\n. Barbara Herrnstein Smith, _Poetic Closure: A Study of How Poems End_ (Chicago: University of Chicago Press, 1968), 36. Subsequent references to this work will be given in the main text by page numbers in parentheses.\n\n. See Friedrich Nietzsche, _The Use and Abuse of History_ , trans. Adrian Collins (Indianapolis, Ind.: The Bobbs-Merrill Company, 1949), 5.\n\n. Friedrich Nietzsche, \"'Reason' in Philosophy,\" trans. Walter Kaufmann, in Nietzsche, _The Twilight of the Idols_ , excerpted in _The Portable Nietzsche_ , ed. Walter Kaufmann (New York: The Viking Press, 1954), 479.\n\n. Friedrich Nietzsche, _The Birth of Tragedy and The Case of Wagner_ , trans. Walter Kaufmann (New York: Random House, 1967), 59. In _The Aesthetic Dimension_ , Marcuse argues that \"aesthetic affirmation\" in life that is not a matter of escapist fantasizing must include a sense of the ontologically \"irreconcilable\" and that it is expressed aptly in the last words of the \"Song of the Tower Warden\" in Goethe's _Faust_ : \"Es war doch so sch\u00f6n.\" Herbert Marcuse, _The Aesthetic Dimension_ , trans. Herbert Marcuse and Erica Sherover (Boston: Beacon Press, 1978), 59. See also the concluding discussion of gratitude as a response to the experience of the truth of skepticism in Eldridge, _Leading a Human Life_ , 286\u2013290.\n\n. Friedrich Nietzsche, _Writings from the Late Notebook_ , ed. R\u00fcdiger Bittner (Cambridge: Cambridge University Press, 2003), 159\u2013160.\n\n. Steven Winn, \"Endings Are a Catharsis,\" _San Francisco Chronicle_ , January 1, 2005. Available online at <http:\/\/sfgate.com\/cgi-bin\/article.cgi?file=\/chronicle\/archive\/2005\/01\/01\/DDG7VAJAL81.DTL>.\n\n. Frank Kermode, _The Sense of an Ending: Studies in the Theory of Fiction, with a New Epilogue_ (Oxford: Oxford University Press, 2000), 138. Subsequent references to this work will be given in the main text by page numbers in parentheses.\n\n. Judith Ryan, _Rilke, Modernism, and Poetic Tradition_ (Cambridge: Cambridge University Press, 1999), 82.\n\n. Ibid., 89.\n\n. Ibid., 98.\n\n. Ibid., 83\u201384.\n\n. Rainer Maria Rilke, _The Selected Poetry of Rainer Maria Rilke_ , ed. and trans. Stephen Mitchell (New York: Random House, 1982), 60\u201361.\n\n. Ryan, _Rilke, Modernism, and Poetic Tradition_ , 36.\n\n. Ibid., 83.\n\n. Ibid., 84.\n\n. Ibid., 86.\n\n_6. \"New Centers of Reflection Are Continually Forming\": Benjamin, Sebald, and Modern Human Life in Time_\n\n. The quotation that forms the title of this chapter is from Walter Benjamin, \"The Concept of Criticism in German Romanticism,\" trans. David Lachterman, Howard Eiland, and Ian Balfour, in Benjamin, _Selected Writings, Volume 1: 1913\u20131926_ , ed. Marcus Bullock and Michael W. Jennings (Cambridge, Mass.: Harvard University Press, 1996), 156.\n\n. G. W. F. Hegel, _Aesthetics: Lectures on Fine Art_ , trans. T. M. Knox (Oxford: Clarendon Press, 1975), 31.\n\n. Hegel, _Phenomenology of Spirit_ , trans. A. V. Miller (Oxford: Clarendon Press, 1977), para. 32, p. 19.\n\n. An attentive reader for Columbia University Press urged this formulation on me.\n\n. Stephen Houlgate, \"Introduction: An Overview of Hegel's Aesthetics,\" in _Hegel and the Arts_ , ed. Stephen Houlgate (Evanston, Ill.: Northwestern University Press, 2007), xxv.\n\n. Terry Pinkard, \"Symbolic, Classical, and Romantic Art,\" in _Hegel and the Arts_ , 5.\n\n. G. W. F. Hegel, _Lectures on Fine Art_ , trans. T. M. Knox (Oxford: The Clarendon Press, 1975), 1:97.\n\n. Ibid., 98.\n\n. Hegel, _Elements of the Philosophy of Right_ , trans. H. B. Nisbet (Cambridge: Cambridge University Press, 1991), 14.\n\n. In emphasizing the \"good-enough\" reconciliation theme in Hegel, I have been influenced by Michael O. Hardimon, _Hegel's Social Philosophy: The Project of Reconciliation_ (Cambridge: Cambridge University Press, 1994).\n\n. Walter Benjamin, \"On Language as Such,\" trans. Rodney Livingstone, in Benjamin, _Selected Writings: Volume 1, 1913\u20131926_ , 73, 70.\n\n. Ibid., 73.\n\n. See Donald Davidson's classic \"On the Very Idea of a Conceptual Scheme\" and \"The Method of Truth in Metaphysics,\" both in Davidson, _Inquiries Into Truth and Interpretation_ (Oxford: Clarendon Press, 1984), for an argument to this effect. For a commentary on the powers but also on the limits of this argument, specifically how it leaves specifically problems of practical engagement between subjects unaddressed, see Alasdair MacIntyre, \"Relativism, Power, and Philosophy,\" in _After Philosophy: End or Transformation?_ , ed. Kenneth Baynes, James Bohman, and Thomas McCarthy (Cambridge, Mass.: The MIT Press, 1987), and Richard Eldridge, \"Metaphysics and the Interpretation of Persons: Davidson on Thinking and Conceptual Schemes,\" _Synthese_ 66, no. 3 (March 1986): 477\u2013503.\n\n. Benjamin, \"On Language as Such,\" 66.\n\n. Michael Rosen, \"Benjamin, Adorno, and the Decline of the Aura,\" in _The Cambridge Companion to Critical Theory_ , ed. Fred Rush (Cambridge: Cambridge University Press, 2004), 46.\n\n. Benjamin, _The Arcades Project_ , trans. H. Eiland and K. McLaughlin (Cambridge, Mass.: Harvard University Press, 1999), 460.\n\n. Anthony Lane, \"Higher Ground: Adventures in Fact and Fiction from W. G. Sebald,\" _The New Yorker_ , May 29, 2000.\n\n. Susan Sontag, _Where the Stress Falls_ (New York: Farrar, Straus and Giroux, 2001), 41.\n\n. Ibid., 46, 48, 46.\n\n. Franz Loquai, \"Vom Beinhaus der Geschichte ins wiedergefundene Paradies: Zu Werk und Poetik W. G. Sebalds,\" in _Sebald. Lekt\u00fcren_ , ed. Marcel Atze and Franz Loquai (Eggingen: Edition Isele, 2005), 244. My translation.\n\n. See Farrell, _Why Does Literature Matter?_ , 199.\n\n. Ibid., 197.\n\n. Ibid., 200.\n\n. Mark R. McCulloh, _Understanding W. G. Sebald_ (Columbia: University of South Carolina Press, 2003), 19.\n\n. W. G. Sebald, _Der Mythus der Zerst\u00f6rung im Werk D\u00f6blins_ (Stuttgart: Klett, 1980), 58; cited in McCulloh, _Understanding W. G. Sebald_ , 148. McCulloh's translation.\n\n. Eric L. Santner, _On Creaturely Life_ (Chicago: University of Chicago Press, 2006), 49.\n\n. Ibid., 114, n. 20.\n\n. Ibid., 20.\n\n. Ibid.\n\n. Ibid., 71.\n\n. Ibid., 74.\n\n. Ibid., 81, 84; in the latter passage Santner is drawing on Alenka Zupancic, _The Shortest Shadow: Nietzsche's Philosophy of the Two_ (Cambridge, Mass.: The MIT Press, 2003), 49.\n\n. Wordsworth, _The Prelude_ (1850), book 7, ll. 725\u2013728, p. 288.\n\n. Hegel, _Elements of the Philosophy of Right_ , para. 185, p. 222.\n\n. Santner, _On Creaturely Life_ , 134.\n\n. Ibid., 133.\n\n. Ibid., 136, citing Jonathan Lear, _Happiness, Death, and the Remainder of Life_ (Cambridge, Mass.: Harvard University Press, 2000), 129.\n\n. Ibid., 203.\n\n. Terry Eagleton, _Walter Benjamin, or Towards a Revolutionary Criticism_ (London: Verson, 1985), 42; cited in Santner, _On Creaturely Life_ , 134, n. 54.\n\n. Santner, _On Creaturely Life_ , 75.\n\n. Ibid.\n\n. Ibid., 81.\n\n. Walter Benjamin, \"Surrealism,\" trans. Rodney Livingstone, in Benjamin, _Selected Writings: Volume 2, 1927\u20131934_ , ed. Michael W. Jennings, Howard Eiland, and Gary Smith (Cambridge, Mass.: Harvard University Press, 1999), 216. Sebald cites this passage in _Die Beschreibung des Ungl\u00fccks: Zur \u00f6sterreichischen Literatur von Stifter bis Handke_ (Frankfurt am Main: Fisscher, 1994), 132. McCulloh, _Understanding W. G. Sebald_ , 155, n. 11, notes this citation.\n\n. McCulloh, _Understanding W. G. Sebald_ , 3.\n\n. Walter Benjamin, \"The Concept of Criticism in German Romanticism,\" in Benjamin, _Selected Writings, Volume 1: 1913\u20131926_ , 116\u2013200. All references to this work will be given by page number in the text.\n\n. In the opening lines of his _The Origin of German Tragic Drama_ , Benjamin remarks that \"it is characteristic of philosophical writing that at every turn it must confront the question of representation [ _Darstellung_ ] anew.\" On the significance of this remark, see Azade Seyhan, _Representation and Its Discontents: The Critical Legacy of German Romanticism_ (Berkeley: University of California Press, 1992); and Martha B. Helfer, _The Retreat of Representation: The Concept of Darstellung in German Critical Discourse_ (Albany: State University of New York Press, 1996).\n\n. Herbert Marcuse, _The Aesthetic Dimension_ , trans. Herbert Marcuse and Erica Sherover (Boston: Beacon Press, 1978), 59.\n\n. See also the slightly different translation in Friedrich Schlegel, \"On Goethe's _Meister_ ,\" trans. Joyce Crick, in _Classical and Romantic German Aesthetics_ , ed. J. M. Bernstein (Cambridge: Cambridge University Press, 2003), 273.\n\n. W. G. Sebald, \"Paul Bereyter,\" in _The Emigrants_ , trans. Michael Hulse (London: Vintage, 2002), 25\u201363; W. G. Sebald, \"Paul Bereyter,\" in _Die Ausgewanderten: Vier lange Erz\u00e4hlungen_ (Frankfurt am Main: Fischer, 1994), 39\u201393. Citations to these works will be given in the text by English page number followed by German page number.\n\n. \"Manchem g\u00f6ttliche Gem\u00fcte wird vom Schicksal eine unf\u00f6rmliche Form aufgedrungen, wie dem Sokrates der Satyr-Leib; denn \u00fcber die Form, nicht \u00fcber den innern Stoff regiert die Zeit. So hing der poetische Spiegel, womit Jakob B\u00f6hme Himmel und Erde wiedergibt, in einem dunklen Orte; auch mangelt dem Glase an einigen Stellen die Folie. So ist der gro\u00dfe Hamann ein tiefer Himmel voll teleskopischer Sterne, und manche Nebelflecken l\u00f6set kein Auge auf.\" Jean Paul, _Vorschule der \u00c4sthetik_ , in _S\u00e4mtliche Werke_ , vol. 3 (Paris, 1836\u20131837); English translation in Jean Paul Richter, _Horn of Oberon_ , trans. Margaret R. Hale, (Detroit, Mich.: Wayne State University Press, 1973), 41\u201342.\n\n. Dante, _The Divine Comedy I: Hell_ , trans. Dorothy Sayers (Harmondsworth: Penguin, 1949), canto 1, ll. 1\u20133, p. 71.\nIndex\n\nabsorption\n\nAdorno, T. W.\n\nAlthusser, Louis\n\nAltieri, Charles\n\nAppelo, Tim\n\nAristotle\n\nAusten, Jane\n\nAustin, J. L.\n\nautopoiesis\n\nBaker, Gordon\n\nBanville, Theodore de\n\nBarrell, John\n\nBaudelaire, Charles\n\nBeckett, Samuel\n\nBenjamin, Walter\n\nBentham, Jeremy\n\nBernstein, Charles\n\nBernstein, J. M.\n\nBlake, William\n\nBloom, Harold\n\nBooth, Wayne\n\nBourdieu, Pierre\n\nBoyle, Nicholas\n\nBromwich, David\n\nBrown, Dan\n\nByron, Lord (Baron George Gordon)\n\nByronism\n\nCartesianism\n\ncatharsis\n\nCavell, Stanley\n\nchildhood\n\nclose reading\n\nclosure, aesthetic\n\nCohen, Ted\n\nCollingwood, R. G.\n\nconsciousness, conceptual\n\nCoolidge, Clark\n\ncurricula, literary\n\nDante (Alighieri)\n\nDavidson, Donald\n\ndepression\n\nDescartes, Ren\u00e9\n\ndesire\n\nDewey, John\n\ndidacticism\n\ndisruption, aesthetic\n\nD\u00f6blin, Alfred\n\nDon Giovanni\n\nDon Quixote\n\nEagleton, Terry\n\nEldridge, Richard\n\nEliot, T. S.\n\nEmerson, R. W.\n\nfactionalism\n\nFarrell, Frank\n\nFlaubert, Gustave\n\nFleming, Ian\n\nformalism\n\nfreedom, expressive\n\nFreud, Sigmund\n\nFrye, Northrop\n\nGasch\u00e9, Rodolphe\n\ngenius\n\nGeorge, Stefan\n\nGibson, John\n\nglobalization\n\nGoethe, J. W.; \"Maifest\"; _The Sorrows of Young Werther_\n\n_Gravity's Rainbow_\n\nGutting, Gary\n\nHabermas, J\u00fcrgen\n\nHacker, Peter\n\nHamann, J. G.\n\nHamlet\n\nHampton, Christopher\n\nHardimon, Michael\n\nHartman, Geoffrey\n\nHegel, G. W. F.; on romanticism\n\nHeidegger, Martin\n\nHelfer, Martha B.\n\nH\u00f6lderlin, Friedrich\n\nHoulgate, Stephen\n\nHuemer, Wolfgang\n\nHume, David\n\nHumeanism\n\nindividualism\n\nJames, Henry\n\nJohn Paul II\n\njustice\n\nKant, I.\n\nKeane, Marian\n\nKeats, John\n\nKermode, Frank\n\nKierkegaard, S\u00f8ren\n\nLacan, Jacques\n\nLacoue-Labarthe, Philippe\n\nLamarque, Peter\n\nLane, Anthony\n\nLarmore, Charles\n\nLear, Jonathan\n\nLevinson, Marjorie\n\nLocke, John\n\nLodge, David\n\nLoquai, Franz\n\nLowell, Robert\n\nLuk\u00e1cs, Georg\n\nMacIntyre, Alasdair\n\nMallarm\u00e9, Stephane\n\nMann, Thomas\n\nMarcuse, Herbert\n\nMarshall, Donald\n\nMarx, Karl\n\nMcCulloh, Mark R.\n\nMcGann, Jerome\n\nMcGuinness, Brian\n\nMiall, David\n\nMill, J. S.\n\nmodernism\n\nmodernity\n\nMonk, Ray\n\nNagel, Thomas\n\nNancy, Jean-Luc\n\nNietzsche, Friedrich; _The Birth of Tragedy_\n\nNussbaum, Martha\n\nOlsen, Stein Haugom\n\noriginality\n\nPfau, Thomas\n\nPhillips, D. Z.\n\nPinkard, Terry\n\nPlato\n\npostmodernism\n\nPutnam, Hilary\n\nReadings, Bill\n\nrepresentation, perspicuous\n\nRichter, Jean Paul\n\nRilke, Rainer Maria\n\nRodin, Auguste\n\nromanticism\n\nRorty, Richard\n\nRosen, Michael\n\nRothman, William\n\nRousseau, Jean-Jacques\n\nRowe, Mark\n\nRyan, Judith\n\nQuine, W. V. O.\n\nSaid, Edward\n\nSantner, Eric\n\nSchlegel, Friedrich\n\nSchulte, Joachim\n\nSebald, W. G.; \"Paul Bereyter\"\n\nsentimentalism\n\nSeyhan, Azade\n\nShakespeare, William\n\nSimon, John\n\nSmith, Barbara Herrnstein\n\nSontag, Susan\n\nSpinoza, Benedict de\n\nStecker, Robert\n\nStoppard, Tom\n\nTaylor, Charles\n\nThoreau, Henry David\n\nTomasello, Michael\n\n_Tristram Shandy_\n\nvalue, literary\n\nWalsh, Dorothy\n\nWeinstein, Philip\n\nWellbery, David\n\nWhorf, Benjamin\n\nWilson, Catherine\n\nWinn, Steven\n\nWittgenstein, Ludwig\n\nWordsworth, William; \"Preface\" to _Lyrical Ballads_ ; _The Prelude_ ; \"Tintern Abbey\"\n\nworking through\n\nZamir, Tachi\n\nZupancic, Alenka\n","meta":{"redpajama_set_name":"RedPajamaBook"}}
+{"text":"\n\nE-text prepared by Andrea Ball, Christine Bell, & Marc D'Hooghe\n(http:\/\/www.freeliterature.org) from page images generously made available\nby the Google Books Library Project (http:\/\/books.google.com\/)\n\n\n\nNote: Project Gutenberg also has the other two volumes of\n      this book.\n      Volume I:  See http:\/\/www.gutenberg.org\/ebooks\/36289\n      Volume II: See http:\/\/www.gutenberg.org\/ebooks\/36290\n\n\n      Images of the original pages are available through\n      the the Google Books Library Project. See\n      http:\/\/books.google.com\/books?id=aRgGAAAAQAAJ&oe=UTF-8\n\n\n\n\n\nORMOND;\n\nOr,\n\nThe Secret Witness.\n\nby\n\nB. C. BROWN,\n\nAuthor of Wieland, or Transformation.\n\nIn Three Volumes.\n\nVOL. III.\n\n\n\"Saepe intereunt aliis meditantes necem.\"\n\n                                  PHAEDRUS\n\n\"Those who plot the destruction of others, very often fall,\nthemselves the victims.\"\n\n\n\n\n\n\n\nPhiladelphia  Printed,\nLondon, Re-Printed for Henry Colburn,\nEnglish and Foreign Public Library,\nConduit-Street, Bond-Street.\n1811\n\n\n       *       *       *       *       *\n\n\nTO THE RIGHT HONOURABLE\n\nLADY CASTLEREAGH,\n\nTHESE VOLUMES\n\nare respectfully inscribed,\n\nby her Ladyship's\n\nmost obedient, and humble Servant,\n\nHENRY COLBURN.\n\n\n\n\nCHAPTER I.\n\n\n\"My father, in proportion as he grew old and rich, became weary of\nAleppo. His natal soil, had it been the haunt of Calmucks or Bedouins,\nhis fancy would have transformed into Paradise. No wonder that the\nequitable aristocracy and the peaceful husbandmen of Ragusa should be\nendeared to his heart by comparison with Egyptian plagues and Turkish\ntyranny. Besides, he lived for his children as well as himself. Their\neducation and future lot required him to seek a permanent home.\n\n\"He embarked, with his wife and offspring, at Scanderoon. No immediate\nconveyance to Ragusa offering, the appearance of the plague in Syria\ninduced him to hasten his departure. He entered a French vessel for\nMarseilles. After being three days at sea, one of the crew was seized by\nthe fatal disease which had depopulated all the towns upon the coast.\nThe voyage was made with more than usual despatch; but, before we\nreached our port, my mother and half the crew perished. My father died\nin the Lazaretto, more through grief than disease.\n\n\"My brother and I were children and helpless. My father's fortune was on\nboard this vessel, and was left by his death to the mercy of the\ncaptain. This man was honest, and consigned us and our property to the\nmerchant with whom he dealt. Happily for us, our protector was childless\nand of scrupulous integrity. We henceforth became his adopted children.\nMy brother's education and my own were conducted on the justest\nprinciples.\n\n\"At the end of four years, our protector found it expedient to make a\nvoyage to Cayenne. His brother was an extensive proprietor in that\ncolony, but his sudden death made way for the succession of our friend.\nTo establish his claims, his presence was necessary on the spot. He was\nlittle qualified for arduous enterprises, and his age demanded repose;\nbut, his own acquisitions having been small, and being desirous of\nleaving us in possession of competence, he cheerfully embarked.\n\n\"Meanwhile, my brother was placed at a celebrated seminary in the Pays\nde Vaud, and I was sent to a sister who resided at Verona. I was at this\ntime fourteen years old,--one year younger than my brother, whom, since\nthat period, I have neither heard of nor seen.\n\n\"I was now a woman, and qualified to judge and act for myself. The\ncharacter of my new friend was austere and devout, and there were so\nmany incongenial points between us that but little tranquillity was\nenjoyed under her control. The priest who discharged the office of her\nconfessor thought proper to entertain views with regard to me, grossly\ninconsistent with the sanctity of his profession. He was a man of\nprofound dissimulation and masterly address. His efforts, however, were\nrepelled with disdain. My security against his attempts lay in the\nuncouthness and deformity which nature had bestowed upon his person and\nvisage, rather than in the firmness of my own principles.\n\n\"The courtship of Father Bartoli, the austerities of Madame Roselli, the\ndisgustful or insipid occupations to which I was condemned, made me\nimpatiently wish for a change; but my father (so I will call him) had\ndecreed that I should remain under his sister's guardianship till his\nreturn from Guiana. When this would happen was uncertain. Events\nunforeseen might protract it for years, but it could not arrive in less\nthan a twelvemonth.\n\n\"I was incessantly preyed upon by discontent. My solitude was loathsome.\nI panted after liberty and friendship, and the want of these were not\nrecompensed by luxury and quiet, and by the instructions in useful\nscience which I received from Bartoli, who, though detested as a\nhypocrite and lover, was venerable as a scholar. He would fain have been\nan Abelard, but it was not his fate to meet with an Eloisa.\n\n\"Two years passed away in this durance. My miseries were exquisite. I am\nalmost at a loss to account for the unhappiness of that time, for,\nlooking back upon it, I perceive that an equal period could not have\nbeen spent with more benefit. For the sake of being near me, Bartoli\nimportunately offered his instructions. He had nothing to communicate\nbut metaphysics and geometry. These were little to my taste, but I could\nnot keep him at a distance. I had no other alternative than to endure\nhim as a lover or a teacher. His passion for science was at least equal\nto that which ho entertained for me, and both these passions combined to\nmake him a sedulous instructor. He was a disciple of the newest\ndoctrines respecting matter and mind. He denied the impenetrability of\nthe first, and the immateriality of the second. These he endeavoured to\ninculcate upon me, as well as to subvert my religious tenets, because he\ndelighted, like all men, in transfusing his opinions, and because he\nregarded my piety as the only obstacle to his designs. He succeeded in\ndissolving the spell of ignorance, but not in producing that kind of\nacquiescence he wished. He had, in this respect, to struggle not only\nwith my principles, but my weakness. He might have overcome every\nobstacle but my abhorrence of deformity and age. To cure me of this\naversion was beyond his power. My servitude grew daily more painful. I\ngrew tired of chasing a comet to its aphelion, and of untying the knot\nof an infinite series. A change in my condition became indispensable to\nmy very existence. Languor and sadness, and unwillingness to eat or to\nmove, were at last my perpetual attendants!\n\n\"Madame Roselli was alarmed at my condition. The sources of my\ninquietude were incomprehensible to her. The truth was, that I scarcely\nunderstood them myself, and my endeavours to explain them to my friend\nmerely instilled into her an opinion that I was either lunatic or\ndeceitful. She complained and admonished; but my disinclination to my\nusual employments would not be conquered, and my health rapidly\ndeclined. A physician, who was called, confessed that my case was beyond\nhis power to understand, but recommended, as a sort of desperate\nexpedient, a change of scene. A succession and variety of objects might\npossibly contribute to my cure.\n\n\"At this time there arrived, at Verona, Lady D'Arcy,--an Englishwoman\nof fortune and rank, and a strenuous Catholic. Her husband had lately\ndied; and, in order to divert her grief, as well as to gratify her\ncuriosity in viewing the great seat of her religion, she had come to\nItaly. Intercourse took place between her and Madame Roselli. By this\nmeans she gained a knowledge of my person and condition, and kindly\noffered to take me under her protection. She meant to traverse every\npart of Italy, and was willing that I should accompany her in all her\nwanderings.\n\n\"This offer was gratefully accepted, in spite of the artifices and\nremonstrances of Bartoli. My companion speedily contracted for me the\naffection of a mother. She was without kindred of her own religion,\nhaving acquired her faith, not by inheritance, but conversion. She\ndesired to abjure her native country, and to bind herself, by every\nsocial tie, to a people who adhered to the same faith. Me she promised\nto adopt as her daughter, provided her first impressions in my favour\nwere not belied by my future deportment.\n\n\"My principles were opposite to hers; but habit, an aversion to\ndisplease my friend, my passion for knowledge, which my new condition\nenabled me to gratify, all combined to make me a deceiver. But my\nimposture was merely of a negative kind; I deceived her rather by\nforbearance to contradict, and by acting as she acted, than by open\nassent and zealous concurrence. My new state was, on this account, not\ndevoid of inconvenience. The general deportment and sentiments of Lady\nD'Arcy testified a vigorous and pure mind. New avenues to knowledge, by\nconverse with mankind and with books, and by the survey of new scenes,\nwere open for my use. Gratitude and veneration attached me to my friend,\nand made the task of pleasing her, by a seeming conformity of\nsentiments, less irksome.\n\n\"During this interval, no tidings were received by his sister, at\nVerona, respecting the fate of Sebastian Roselli. The supposition of\nhis death was too plausible not to be adopted. What influence this\ndisaster possessed over my brother's destiny, I know not. The generosity\nof Lady D'Arcy hindered me from experiencing any disadvantage from this\ncircumstance. Fortune seemed to have decreed that I should not be\nreduced to the condition of an orphan.\n\n\"At an age and in a situation like mine, I could not remain long\nunacquainted with love. My abode at Rome introduced me to the knowledge\nof a youth from England, who had every property which I regarded as\nworthy of esteem. He was a kinsman of--Lady D'Arcy, and as such admitted\nat her house on the most familiar footing. His patrimony was extremely\nslender, but was in his own possession. He had no intention of\nincreasing it by any professional pursuit, but was contented with the\nfrugal provision it afforded. He proposed no other end of his existence\nthan the acquisition of virtue and knowledge.\n\n\"The property of Lady D'Arcy was subject to her own disposal, but, on\nthe failure of a testament, this youth was, in legal succession, the\nnext heir. He was well acquainted with her temper and views, but, in the\nmidst of urbanity and gentleness, studied none of those concealments of\nopinion which would have secured him her favour. That he was not of her\nown faith was an insuperable, but the only, obstacle to the admission of\nhis claims.\n\n\"If conformity of age and opinions, and the mutual fascination of love,\nbe a suitable basis for marriage, Wentworth and I were destined for each\nother. Mutual disclosure added sanctity to our affection; but, the\nhappiness of Lady D'Arcy being made to depend upon the dissolution of\nour compact, the heroism of Wentworth made him hasten to dissolve it. As\nsoon as she discovered our attachment, she displayed symptoms of the\ndeepest anguish. In addition to religious motives, her fondness for me\nforbade her to exist but in my society and in the belief of the purity\nof my faith. The contention, on my part, was vehement between the\nregards due to her felicity and to my own. Had Wentworth left me the\npower to decide, my decision would doubtless have evinced the frailty of\nmy fortitude and the strength of my passion; but, having informed me\nfully of the reasons of his conduct, he precipitately retired from Rome.\nHe left me no means of tracing his footsteps and of assailing his\nweakness by expostulation and entreaty.\n\n\"Lady D'Arcy was no less eager to abandon a spot where her happiness had\nbeen so imminently endangered. Our next residence was Palermo. I will\nnot dwell upon the sensations produced by this disappointment in me. I\nreview them with astonishment and self-compassion. If I thought it\npossible for me to sink again into imbecility so ignominious, I should\nbe disposed to kill myself.\n\n\"There was no end to vows of fondness and tokens of gratitude in Lady\nD'Arcy. Her future life should be devoted to compensate me for this\nsacrifice. Nothing could console her in that single state in which she\nintended to live, but the consolations of my fellowship. Her conduct\ncoincided for some time with these professions, and my anguish was\nallayed by the contemplation of the happiness conferred upon one whom I\nrevered.\n\n\"My friend could not be charged with dissimulation and artifice. Her\ncharacter had been mistaken by herself as well as by me. Devout\naffections seemed to have filled her heart, to the exclusion of any\nobject besides myself. She cherished with romantic tenderness the memory\nof her husband, and imagined that a single state was indispensably\nenjoined upon her by religious duty. This persuasion, however, was\nsubverted by the arts of a Spanish cavalier, young, opulent, and\nromantic as herself in devotion. An event like this might, indeed, have\nbeen easily predicted, by those who reflected that the lady was still in\nthe bloom of life, ardent in her temper, and bewitching in her manners.\n\n\"The fondness she had lavished upon me was now, in some degree,\ntransferred to a new object; but I still received the treatment due to a\nbeloved daughter. She was solicitous as ever to promote my\ngratification, and a diminution of kindness would not have been\nsuspected by those who had not witnessed the excesses of her former\npassion. Her marriage with the Spaniard removed the obstacle to union\nwith Wentworth. This man, however, had set himself beyond the reach of\nmy inquiries. Had there been the shadow of a clue afforded me, I should\ncertainly have sought him to the ends of the world.\n\n\"I continued to reside with my friend, and accompanied her and her\nhusband to Spain. Antonio de Leyva was a man of probity. His mind was\nenlightened by knowledge and his actions dictated by humanity. Though\nbut little older than myself, and young enough to be the son of his\nspouse, his deportment to me was a model of rectitude and delicacy. I\nspent a year in Spain, partly in the mountains of Castile and partly at\nSegovia. New manners and a new language occupied my attention for a\ntime; but these, losing their novelty, lost their power to please. I\nbetook myself to books, to beguile the tediousness and diversify the\ntenor of my life.\n\n\"This would not have long availed; but I was relieved from new\nrepinings, by the appointment of Antonio de Leyva to a diplomatic office\nat Vienna. Thither we accordingly repaired. A coincidence of\ncircumstances had led me wide from the path of ambition and study\nusually allotted to my sex and age. From the computation of eclipses, I\nnow betook myself to the study of man. My proficiency, when I allowed it\nto be seen, attracted great attention. Instead of adulation and\ngallantry, I was engaged in watching the conduct of states and revolving\nthe theories of politicians.\n\n\"Superficial observers were either incredulous with regard to my\ncharacter, or connected a stupid wonder with their belief. My\nattainments and habits they did not see to be perfectly consonant with\nthe principles of human nature. They unavoidably flowed from the illicit\nattachment of Bartoli, and the erring magnanimity of Wentworth. Aversion\nto the priest was the grand inciter of my former studies; the love of\nWentworth, whom I hoped once more to meet, made me labour to exclude the\nimportunities of others, and to qualify myself for securing his\naffections.\n\n\"Since our parting in Italy, Wentworth had traversed Syria and Egypt,\nand arrived some months after me at Vienna. He was on the point of\nleaving the city, when accident informed me of his being there. An\ninterview was effected, and, our former sentiments respecting each other\nhaving undergone no change, we were united. Madame de Leyva reluctantly\nconcurred with our wishes, and, at parting, forced upon me a\nconsiderable sum of money.\n\n\"Wentworth's was a character not frequently met with in the world. He\nwas a political enthusiast, who esteemed nothing more graceful or\nglorious than to die for the liberties of mankind. He had traversed\nGreece with an imagination full of the exploits of ancient times, and\nderived, from contemplating Thermopylae and Marathon, an enthusiasm that\nbordered upon frenzy.\n\n\"It was now the third year of the Revolutionary War in America, and,\nprevious to our meeting at Vienna, he had formed the resolution of\nrepairing thither and tendering his service to the Congress as a\nvolunteer. Our marriage made no change in his plans. My soul was\nengrossed by two passions,--a wild spirit of adventure, and a boundless\ndevotion to him. I vowed to accompany him in every danger, to vie with\nhim in military ardour, to combat and to die by his side.\n\n\"I delighted to assume the male dress, to acquire skill at the sword,\nand dexterity in every boisterous exercise. The timidity that commonly\nattends women gradually vanished. I felt as if imbued by a soul that was\na stranger to the sexual distinction. We embarked at Brest, in a frigate\ndestined for St. Domingo. A desperate conflict with an English ship in\nthe Bay of Biscay was my first introduction to a scene of tumult and\ndanger of whose true nature I had formed no previous conception. At\nfirst I was spiritless and full of dismay. Experience, however,\ngradually reconciled me to the life that I had chosen.\n\n\"A fortunate shot, by dismasting the enemy, allowed us to prosecute our\nvoyage unmolested. At Cape Francois we found a ship which transported\nus, after various perils, to Richmond, in Virginia. I will not carry you\nthrough the adventures of four years. You, sitting all your life in\npeaceful corners, can scarcely imagine that variety of hardship and\nturmoil which attends the female who lives in a camp.\n\n\"Few would sustain these hardships with better grace than I did. I could\nseldom be prevailed on to remain at a distance, and inactive, when my\nhusband was in battle, and more than once rescued him from death by the\nseasonable destruction of his adversary.\n\n\"At the repulse of the Americans at Germantown, Wentworth was wounded\nand taken prisoner. I obtained permission to attend his sick-bed and\nsupply that care without which he would assuredly have died. Being\nimperfectly recovered, he was sent to England and subjected to a\nrigorous imprisonment. Milder treatment might have permitted his\ncomplete restoration to health; but, as it was, he died.\n\n\"His kindred were noble, and rich, and powerful; but it was difficult to\nmake them acquainted with Wentworth's situation. Their assistance, when\ndemanded, was readily afforded; but it came too late to prevent his\ndeath. Me they snatched from my voluntary prison, and employed every\nfriendly art to efface from my mind the images of recent calamity.\n\n\"Wentworth's singularities of conduct and opinion had estranged him at\nan early age from his family. They felt little regret at his fate, but\nevery motive concurred to secure their affection and succour to me. My\ncharacter was known to many officers, returned from America, whose\nreport, joined with the influence of my conversation, rendered me an\nobject to be gazed at by thousands. Strange vicissitude! Now immersed in\nthe infection of a military hospital, the sport of a wayward fortune,\nstruggling with cold and hunger, with negligence and contumely. A month\nafter, passing into scenes of gayety and luxury, exhibited at operas and\nmasquerades, made the theme of inquiry and encomium at every place of\nresort, and caressed by the most illustrious among the votaries of\nscience and the advocates of the American cause.\n\n\"Here I again met Madame de Leyva. This woman was perpetually assuming\nnew forms. She was a sincere convert to the Catholic religion, but she\nwas open to every new impression. She was the dupe of every powerful\nreasoner, and assumed with equal facility the most opposite shapes. She\nhad again reverted to the Protestant religion, and, governed by a\nheadlong zeal in whatever cause she engaged, she had sacrificed her\nhusband and child to a new conviction.\n\n\"The instrument of this change was a man who passed, at that time, for a\nFrenchman. He was young, accomplished, and addressful, but was not\nsuspected of having been prompted by illicit views, or of having seduced\nthe lady from allegiance to her husband as well as to her God. De Leyva,\nhowever, who was sincere in his religion as well as his love, was hasty\nto avenge this injury, and, in a contest with the Frenchman, was killed.\nHis wife adopted at once her ancient religion and country, and was once\nmore an Englishwoman.\n\n\"At our meeting her affection for me seemed to be revived, and the most\npassionate entreaties were used to detain me in England. My previous\narrangements would not suffer it. I foresaw restraints and\ninconveniences from the violence and caprice of her passions, and\nintended henceforth to keep my liberty inviolate by any species of\nengagement, either of friendship or marriage. My habits were French, and\nI proposed henceforward to take up my abode at Paris. Since his voyage\nto Guiana, I had heard no tidings of Sebastian Roselli. This man's image\nwas cherished with filial emotions, and I conceived that the sight of\nhim would amply reward a longer journey than from London to Marseilles.\n\n\"Beyond my hopes, I found him in his ancient abode. The voyage, and a\nresidence of three years at Cayenne, had been beneficial to his\nappearance and health. He greeted me with paternal tenderness, and\nadmitted me to a full participation of his fortune, which the sale of\nhis American property had greatly enhanced. He was a stranger to the\nfate of my brother. On his return home he had gone to Switzerland, with\na view of ascertaining his destiny. The youth, a few months after his\narrival at Lausanne, had eloped with a companion, and had hitherto\neluded all Roselli's searches and inquiries. My father was easily\nprevailed upon to transfer his residence from Provence to Paris.\"\n\nHere Martinette paused, and, marking the clock, \"It is time,\" resumed\nshe, \"to begone. Are you not weary of my tale? On the day I entered\nFrance, I entered the twenty-third year of my age, so that my promise of\ndetailing my youthful adventures is fulfilled. I must away. Till we meet\nagain, farewell.\"\n\n\n\n\nCHAPTER II.\n\n\nSuch was the wild series of Martinette's adventures. Each incident\nfastened on the memory of Constantia, and gave birth to numberless\nreflections. Her prospect of mankind seemed to be enlarged, on a sudden,\nto double its ancient dimensions. Ormond's narratives had carried her\nbeyond the Mississippi, and into the deserts of Siberia. He had\nrecounted the perils of a Russian war, and painted the manners of\nMongols and Naudowessies. Her new friend had led her back to the\ncivilized world and portrayed the other half of the species. Men, in\ntheir two forms of savage and refined, had been scrutinized by these\nobservers; and what was wanting in the delineations of the one was\nliberally supplied by the other.\n\nEleven years in the life of Martinette was unrelated. Her conversation\nsuggested the opinion that this interval had been spent in France. It\nwas obvious to suppose that a woman thus fearless and sagacious had not\nbeen inactive at a period like the present, which called forth talents\nand courage without distinction of sex, and had been particularly\ndistinguished by female enterprise and heroism. Her name easily led to\nthe suspicion of concurrence with the subverters of monarchy, and of\nparticipation in their fall. Her flight from the merciless tribunals of\nthe faction that now reigned would explain present appearances.\n\nMartinette brought to their next interview an air of uncommon\nexultation. On this being remarked, she communicated the tidings of the\nfall of the sanguinary tyranny of Robespierre. Her eyes sparkled, and\nevery feature was pregnant with delight, while she unfolded, with her\naccustomed energy, the particulars of this tremendous revolution. The\nblood which it occasioned to flow was mentioned without any symptoms of\ndisgust or horror.\n\nConstantia ventured to ask if this incident was likely to influence her\nown condition.\n\n\"Yes. It will open the way for my return.\"\n\n\"Then you think of returning to a scene of so much danger?\"\n\n\"Danger, my girl? It is my element. I am an adorer of liberty, and\nliberty without peril can never exist.\"\n\n\"But so much bloodshed and injustice! Does not your heart shrink from\nthe view of a scene of massacre and tumult, such as Paris has lately\nexhibited and will probably continue to exhibit?\"\n\n\"Thou talkest, Constantia, in a way scarcely worthy of thy good sense.\nHave I not been three years in a camp? What are bleeding wounds and\nmangled corpses, when accustomed to the daily sight of them for years?\nAm I not a lover of liberty? and must I not exult in the fall of\ntyrants, and regret only that my hand had no share in their\ndestruction?\"\n\n\"But a woman--how can the heart of woman be inured to the shedding of\nblood?\"\n\n\"Have women, I beseech thee, no capacity to reason and infer? Are they\nless open than men to the influence of habit? My hand never faltered\nwhen liberty demanded the victim. If thou wert with me at Paris, I could\nshow thee a fusil of two barrels, which is precious beyond any other\nrelic, merely because it enabled me to kill thirteen officers at\nJemappe. Two of these were emigrant nobles, whom I knew and loved before\nthe Revolution, but the cause they had since espoused cancelled their\nclaims to mercy.\"\n\n\"What!\" said the startled Constantia; \"have you fought in the ranks?\"\n\n\"Certainly. Hundreds of my sex have done the same. Some were impelled by\nthe enthusiasm of love, and some by a mere passion for war; some by the\ncontagion of example; and some--with whom I myself must be ranked--by a\ngenerous devotion to liberty. Brunswick and Saxe-Coburg had to contend\nwith whole regiments of women,--regiments they would have formed, if\nthey had been collected into separate bodies.\n\n\"I will tell thee a secret. Thou wouldst never have seen Martinette de\nBeauvais, if Brunswick had deferred one day longer his orders for\nretreating into Germany.\"\n\n\"How so?\"\n\n\"She would have died by her own hand.\"\n\n\"What could lead to such an outrage?\"\n\n\"The love of liberty.\"\n\n\"I cannot comprehend how that love should prompt you to suicide.\"\n\n\"I will tell thee. The plan was formed, and could not miscarry. A woman\nwas to play the part of a banished Royalist, was to repair to the\nPrussian camp, and to gain admission to the general. This would have\neasily been granted to a female and an ex-noble. There she was to\nassassinate the enemy of her country, and to attest her magnanimity by\nslaughtering herself. I was weak enough to regret the ignominious\nretreat of the Prussians, because it precluded the necessity of such a\nsacrifice.\"\n\nThis was related with accents and looks that sufficiently attested its\ntruth. Constantia shuddered, and drew back, to contemplate more\ndeliberately the features of her guest. Hitherto she had read in them\nnothing that bespoke the desperate courage of a martyr and the deep\ndesigning of an assassin. The image which her mind had reflected from\nthe deportment of this woman was changed. The likeness which she had,\nfeigned to herself was no longer seen. She felt that antipathy was\npreparing to displace love. These sentiments, however, she concealed,\nand suffered the conversation to proceed.\n\nTheir discourse now turned upon the exploits of several women who\nmingled in the tumults of the capital and in the armies on the\nfrontiers. Instances were mentioned of ferocity in some, and magnanimity\nin others, which almost surpassed belief. Constantia listened greedily,\nthough not with approbation, and acquired, at every sentence, new desire\nto be acquainted with the personal history of Martinette. On mentioning\nthis wish, her friend said that she endeavoured to amuse her exile by\ncomposing her own memoirs, and that, on her next visit, she would bring\nwith her the volume, which she would suffer Constantia to read.\n\nA separation of a week elapsed. She felt some impatience for the renewal\nof their intercourse, and for the perusal of the volume that had been\nmentioned. One evening Sarah Baxter, whom Constantia had placed in her\nown occasional service, entered the room with marks of great joy and\nsurprise, and informed her that she at length had discovered Miss\nMonrose. From her abrupt and prolix account, it appeared that Sarah had\novertaken Miss Monrose in the street, and, guided by her own curiosity,\nas well as by the wish to gratify her mistress, she had followed the\nstranger. To her utter astonishment, the lady had paused at Mr. Dudley's\ndoor, with a seeming resolution to enter it, but presently resumed her\nway. Instead of pursuing her steps farther, Sarah had stopped to\ncommunicate this intelligence to Constantia. Having delivered her news,\nshe hastened away, but, returning, in a moment, with a countenance of\nnew surprise, she informed her mistress that on leaving the house she\nhad met Miss Monrose at the door, on the point of entering. She added\nthat the stranger had inquired for Constantia, and was now waiting\nbelow.\n\nConstantia took no time to reflect upon an incident so unexpected and so\nstrange, but proceeded forthwith to the parlour. Martinette only was\nthere. It did not instantly occur to her that this lady and Mademoiselle\nMonrose might possibly be the same. The inquiries she made speedily\nremoved her doubts, and it now appeared that the woman about whose\ndestiny she had formed so many conjectures and fostered so much anxiety\nwas no other than the daughter of Roselli.\n\nHaving readily answered her questions, Martinette inquired, in her turn,\ninto the motives of her friend's curiosity. These were explained by a\nsuccinct account of the transactions to which the deceased Baxter had\nbeen a witness. Constantia concluded with mentioning her own reflections\non the tale, and intimating her wish to be informed how Martinette had\nextricated herself from a situation so calamitous.\n\n\"Is there any room for wonder on that head?\" replied the guest. \"It was\nabsurd to stay longer in the house. Having finished the interment of\nRoselli, (soldier-fashion,) for he was the man who suffered his foolish\nregrets to destroy him, I forsook the house. Roselli was by no means\npoor, but he could not consent to live at ease, or to live at all, while\nhis country endured such horrible oppressions, and when so many of his\nfriends had perished. I complied with his humour, because it could not\nbe changed, and I revered him too much to desert him.\"\n\n\"But whither,\" said Constantia, \"could you seek shelter at a time like\nthat? The city was desolate, and a wandering female could scarcely be\nreceived under any roof. All inhabited houses were closed at that hour,\nand the fear of infection would have shut them against you if they had\nnot been already so.\"\n\n\"Hast thou forgotten that there were at that time at least ten thousand\nFrench in this city, fugitives from Marat and from St. Domingo? That\nthey lived in utter fearlessness of the reigning disease,--sung and\nloitered in the public walks, and prattled at their doors, with all\ntheir customary unconcern? Supposest thou that there were none among\nthese who would receive a countrywoman, even if her name had not been\nMartinette de Beauvais? Thy fancy has depicted strange things; but\nbelieve me that, without a farthing and without a name, I should not\nhave incurred the slightest inconvenience. The death of Roselli I\nforesaw, because it was gradual in its approach, and was sought by him\nas a good. My grief, therefore, was exhausted before it came, and I\nrejoiced at his death, because it was the close of all his sorrows. The\nrueful pictures of my distress and weakness which were given by Baxter\nexisted only in his own fancy.\"\n\nMartinette pleaded an engagement, and took her leave, professing to have\ncome merely to leave with her the promised manuscript. This interview,\nthough short, was productive of many reflections on the deceitfulness of\nappearances, and on the variety of maxims by which the conduct of human\nbeings is regulated. She was accustomed to impart all her thoughts and\nrelate every new incident to her father. With this view she now hied to\nhis apartment. This hour it was her custom, when disengaged, always to\nspend with him.\n\nShe found Mr. Dudley busy in revolving a scheme which various\ncircumstances had suggested and gradually conducted to maturity. No\nperiod of his life had been equally delightful with that portion of his\nyouth which he had spent in Italy. The climate, the language, the\nmanners of the people, and the sources of intellectual gratification in\npainting and music, were congenial to his taste. He had reluctantly\nforsaken these enchanting seats, at the summons of his father, but, on\nhis return to his native country, had encountered nothing but ignominy\nand pain. Poverty and blindness had beset his path, and it seemed as if\nit were impossible to fly too far from the scene of his disasters. His\nmisfortunes could not be concealed from others, and every thing around\nhim seemed to renew the memory of all that he had suffered. All the\nevents of his youth served to entice him to Italy, while all the\nincidents of his subsequent life concurred to render disgustful his\npresent abode.\n\nHis daughter's happiness was not to be forgotten. This he imagined would\nbe eminently promoted by the scheme. It would open to her new avenues to\nknowledge. It would snatch her from the odious pursuit of Ormond, and,\nby a variety of objects and adventures, efface from her mind any\nimpression which his dangerous artifices might have made upon it.\n\nThis project was now communicated to Constantia. Every argument adapted\nto influence her choice was employed. He justly conceived that the only\nobstacle to her adoption of it related to Ormond. He expatiated on the\ndubious character of this man, the wildness of his schemes, and the\nmagnitude of his errors. What could be expected from a man, half of\nwhose life had been spent at the head of a band of Cossacks, spreading\ndevastation in the regions of the Danube, and supporting by flagitious\nintrigues the tyranny of Catharine, and the other half in traversing\ninhospitable countries, and extinguishing what remained of clemency and\njustice by intercourse with savages?\n\nIt was admitted that his energies were great, but misdirected, and that\nto restore them to the guidance of truth was not in itself impossible;\nbut it was so with relation to any power that she possessed. Conformity\nwould flow from their marriage, but this conformity was not to be\nexpected from him. It was not his custom to abjure any of his doctrines\nor recede from any of his claims. She knew likewise the conditions of\ntheir union. She must go with him to some corner of the world where his\nboasted system was established. What was the road to it he had carefully\nconcealed, but it was evident that it lay beyond the precincts of\ncivilized existence.\n\nWhatever were her ultimate decision, it was at least proper to delay it.\nSix years were yet wanting of that period at which only she formerly\nconsidered marriage as proper. To all the general motives for deferring\nher choice, the conduct of Ormond superadded the weightiest. Their\ncorrespondence might continue, but her residence in Europe and converse\nwith mankind might enlighten her judgement and qualify her for a more\nrational decision.\n\nConstantia was not uninfluenced by these reasonings. Instead of\nreluctantly admitting them, she somewhat wondered that they had not been\nsuggested by her own reflections. Her imagination anticipated her\nentrance on that mighty scene with emotions little less than rapturous.\nHer studies had conferred a thousand ideal charms on a theatre where\nScipio and Caesar had performed their parts. Her wishes were no less\nimportunate to gaze upon the Alps and Pyrenees, and to vivify and\nchasten the images collected from books, by comparing them with their\nreal prototypes.\n\nNo social ties existed to hold her to America. Her only kinsman and\nfriend would be the companion of her journeys. This project was likewise\nrecommended by advantages of which she only was qualified to judge.\nSophia Westwyn had embarked, four years previous to this date, for\nEngland, in company with an English lady and her husband. The\narrangements that were made forbade either of the friends to hope for a\nfuture meeting. Yet now, by virtue of this project, this meeting seemed\nno longer to be hopeless.\n\nThis burst of new ideas and now hopes on the mind of Constantia took\nplace in the course of a single hour. No change in her external\nsituation had been wrought, and yet her mind had undergone the most\nsignal revolution. Tho novelty as well as greatness of the prospect kept\nher in a state of elevation and awe, more ravishing than any she had\never experienced. Anticipations of intercourse with nature in her most\naugust forms, with men in diversified states of society, with the\nposterity of Greeks and Romans, and with the actors that were now upon\nthe stage, and, above all, with the being whom absence and the want of\nother attachments had, in some sort, contributed to deify, made this\nnight pass away upon the wings of transport.\n\nThe hesitation which existed on parting with her father speedily gave\nplace to an ardour impatient of the least delay. She saw no impediments\nto the immediate commencement of the voyage. To delay it a month, or\neven a week, seemed to be unprofitable tardiness. In this ferment of her\nthoughts, she was neither able nor willing to sleep. In arranging the\nmeans of departure and anticipating the events that would successively\narise, there was abundant food for contemplation.\n\nShe marked the first dawnings of the day, and rose. She felt reluctance\nto break upon her father's morning slumbers, but considered that her\nmotives were extremely urgent, and that the pleasure afforded him by her\nzealous approbation of his scheme would amply compensate him for this\nunseasonable intrusion on his rest. She hastened therefore to his\nchamber. She entered with blithesome steps, and softly drew aside the\ncurtain.\n\n\n\n\nCHAPTER III.\n\n\nUnhappy Constantia! At the moment when thy dearest hopes had budded\nafresh, when the clouds of insecurity and disquiet had retired from thy\nvision, wast thou assailed by the great subverter of human schemes. Thou\nsawest nothing in futurity but an eternal variation and succession of\ndelights. Thou wast hastening to forget dangers and sorrows which thou\nfondly imaginedst were never to return. This day was to be the outset of\na new career; existence was henceforth to be embellished with enjoyments\nhitherto scarcely within the reach of hope.\n\nAlas! thy predictions of calamity seldom failed to be verified. Not so\nthy prognostics of pleasure. These, though fortified by every\ncalculation of contingencies, were edifices grounded upon nothing. Thy\nlife was a struggle with malignant destiny,--a contest for happiness in\nwhich thou wast fated to be overcome.\n\nShe stooped to kiss the venerable cheek of her father, and, by\nwhispering, to break his slumber. Her eye was no sooner fixed upon his\ncountenance, than she started back and shrieked. She had no power to\nforbear. Her outcries were piercing and vehement. They ceased only with\nthe cessation of breath. She sunk upon a chair in a state partaking more\nof death than of life, mechanically prompted to give vent to her agonies\nin shrieks, but incapable of uttering a sound.\n\nThe alarm called her servants to the spot. They beheld her dumb, wildly\ngazing, and gesticulating in a way that indicated frenzy. She made no\nresistance to their efforts, but permitted them to carry her back to her\nown chamber. Sarah called upon her to speak, and to explain the cause\nof these appearances; but the shock which she had endured seemed to have\nirretrievably destroyed her powers of utterance.\n\nThe terrors of the affectionate Sarah were increased. She kneeled by the\nbedside of her mistress, and, with streaming eyes, besought the unhappy\nlady to compose herself. Perhaps the sight of weeping in another\npossessed a sympathetic influence, or nature had made provision for this\nsalutary change. However that be, a torrent of tears now came to her\nsuccour, and rescued her from a paroxysm of insanity which its longer\ncontinuance might have set beyond the reach of cure.\n\nMeanwhile, a glance at his master's countenance made Fabian fully\nacquainted with the nature of the scene. The ghastly visage of Mr.\nDudley showed that he was dead, and that he had died in some terrific\nand mysterious manner. As soon as this faithful servant recovered from\nsurprise, the first expedient which his ingenuity suggested was to fly\nwith tidings of this event to Mr. Melbourne. That gentleman instantly\nobeyed the summons. With the power of weeping, Constantia recovered the\npower of reflection. This, for a time, served her only as a medium of\nanguish. Melbourne mingled his tears with hers, and endeavoured, by\nsuitable remonstrances, to revive her fortitude.\n\nThe filial passion is perhaps instinctive to man; but its energy is\nmodified by various circumstances. Every event in the life of Constantia\ncontributed to heighten this passion beyond customary bounds. In the\nhabit of perpetual attendance on her father, of deriving from him her\nknowledge, and sharing with him the hourly fruits of observation and\nreflection, his existence seemed blended with her own. There was no\nother whose concurrence and council she could claim, with whom a\ndomestic and uninterrupted alliance could be maintained. The only bond\nof consanguinity was loosened, the only prop of friendship was taken\naway.\n\nOthers, perhaps, would have observed that her father's existence had\nbeen merely a source of obstruction and perplexity; that she had\nhitherto acted by her own wisdom, and would find, hereafter, less\ndifficulty in her choice of schemes, and fewer impediments to the\nexecution. These reflections occurred not to her. This disaster had\nincreased, to an insupportable degree, the vacancy and dreariness of her\nexistence. The face she was habituated to behold had disappeared\nforever; the voice whose mild and affecting tones had so long been\nfamiliar to her ears was hushed into eternal silence. The felicity to\nwhich she clung was ravished away; nothing remained to hinder her from\nsinking into utter despair.\n\nThe first transports of grief having subsided, a source of consolation\nseemed to be opened in the belief that her father had only changed one\nform of being for another; that he still lived to be the guardian of her\npeace and honour, to enter the recesses of her thought, to forewarn her\nof evil and invite her to good. She grasped at these images with\neagerness, and fostered them as the only solaces of her calamity. They\nwere not adapted to inspire her with cheerfulness, but they sublimed her\nsensations, and added an inexplicable fascination to sorrow.\n\nIt was unavoidable sometimes to reflect upon the nature of that death\nwhich had occurred. Tokens were sufficiently apparent that outward\nviolence had been the cause. Who could be the performer of so black a\ndeed, by what motives he was guided, were topics of fruitless\nconjecture. She mused upon this subject, not from the thirst of\nvengeance, but from a mournful curiosity. Had the perpetrator stood\nbefore her and challenged retribution, she would not have lifted a\nfinger to accuse or to punish. The evil already endured left her no\npower to concert and execute projects for extending that evil to others.\nHer mind was unnerved, and recoiled with loathing from considerations of\nabstract justice, or political utility, when they prompted to the\nprosecution of the murderer.\n\nMelbourne was actuated by different views, but on this subject he was\npainfully bewildered. Mr. Dudley's deportment to his servants and\nneighbours was gentle and humane. He had no dealings with the\ntrafficking or labouring part of mankind. The fund which supplied his\ncravings of necessity or habit was his daughter's. His recreations and\nemployments were harmless and lonely. The evil purpose was limited to\nhis death, for his chamber was exactly in the same state in which\nnegligent security had left it. No midnight footstep or voice, no\nunbarred door or lifted window, afforded tokens of the presence or\ntraces of the entrance or flight of the assassin.\n\nThe meditations of Constantia, however, could not fail in some of their\ncircuities to encounter the image of Craig. His agency in the\nimpoverishment of her father, and in the scheme by which she had like to\nhave been loaded with the penalties of forgery, was of an impervious and\nunprecedented kind. Motives were unveiled by time, in some degree\naccounting for his treacherous proceeding; but there was room to suppose\nan inborn propensity to mischief. Was he not the author of this new\nevil? His motives and his means were equally inscrutable, but their\ninscrutability might flow from her own defects in discernment and\nknowledge, and time might supply her defects in this as in former\ninstances.\n\nThese images were casual. The causes of the evil were seldom\ncontemplated. Her mind was rarely at liberty to wander from reflection\non her irremediable loss. Frequently, when confused by distressful\nrecollections, she would detect herself going to her father's chamber.\nOften his well-known accents would ring in her ears, and the momentary\nimpulse would be to answer his calls. Her reluctance to sit down to her\nmeals without her usual companion could scarcely be surmounted.\n\nIn this state of mind, the image of the only friend who survived, or\nwhose destiny, at least, was doubtful, occurred to her. She sunk into\nfits of deeper abstraction and dissolved away in tears of more agonizing\ntenderness. A week after her father's interment, she shut herself up in\nher chamber, to torment herself with fruitless remembrances. The name of\nSophia Westwyn was pronounced, and the ditty that solemnized their\nparting was sung. Now, more than formerly, she became sensible of the\nloss of that portrait which had been deposited in the hands of M'Crea as\na pledge. As soon as her change of fortune had supplied her with the\nmeans of redeeming it, she hastened to M'Crea for that end. To her\nunspeakable disappointment, he was absent from the city; he had taken a\nlong journey, and the exact period of his return could not be\nascertained. His clerks refused to deliver the picture, or even, by\nsearching, to discover whether it was still in their master's\npossession. This application had frequently and lately been repeated,\nbut without success; M'Crea had not yet returned, and his family were\nequally in the dark as to the day on which his return might be expected.\n\nShe determined, on this occasion, to renew her visit. Her incessant\ndisappointments had almost extinguished hope, and she made inquiries at\nhis door, with a faltering accent and sinking heart. These emotions were\nchanged into surprise and delight, when answer was made that he had just\narrived. She was instantly conducted into his presence.\n\nThe countenance of M'Crea easily denoted that his visitant was by no\nmeans acceptable. There was a mixture of embarrassment and sullenness in\nhis air, which was far from being diminished when the purpose of this\nvisit was explained. Constantia reminded him of the offer and acceptance\nof this pledge, and of the conditions with which the transaction was\naccompanied.\n\nHe acknowledged, with some hesitation, that a promise had been given to\nretain the pledge until it were in her power to redeem it; but the long\ndelay, the urgency of his own wants, and particularly the ill treatment\nwhich he conceived himself to have suffered in the transaction\nrespecting the forged note, had, in his own opinion, absolved him from\nthis promise. He had therefore sold the picture to a goldsmith, for as\nmuch as the gold about it was worth.\n\nThis information produced, in the heart of Constantia, a contest between\nindignation and sorrow, that for a time debarred her from speech. She\nstifled the anger that was, at length, rising to her lips, and calmly\ninquired to whom the picture had been sold.\n\nM'Crea answered that for his part he had little dealings in gold and\nsilver, but every thing of that kind which fell to his share he\ntransacted with Mr. D----. This person was one of the most eminent of\nhis profession. His character and place of abode were universally\nknown. Tho only expedient that remained was to apply to him, and to\nascertain, forthwith, the destiny of the picture. It was too probable\nthat, when separated from its case, the portrait was thrown away or\ndestroyed, as a mere encumbrance, but the truth was too momentous to be\nmade the sport of mere probability. She left the house of M'Crea, and\nhastened to that of the goldsmith.\n\nThe circumstance was easily recalled to his remembrance. It was true\nthat such a picture had been offered for sale, and that he had purchased\nit. The workmanship was curious, and he felt unwilling to destroy it. He\ntherefore hung it up in his shop and indulged the hope that a purchaser\nwould some time be attracted by the mere beauty of the toy.\n\nConstantia's hopes were revived by these tidings, and she earnestly\ninquired if it were still in his possession.\n\n\"No. A young gentleman had entered his shop some months before: the\npicture had caught his fancy, and he had given a price which the artist\nowned he should not have demanded, had he not been encouraged by the\neagerness which the gentleman betrayed to possess it.\"\n\n\"Who was this gentleman? Had there been any previous acquaintance\nbetween them? What was his name, his profession, and where was he to be\nfound?\"\n\n\"Really,\" the goldsmith answered, \"he was ignorant respecting all those\nparticulars. Previously to this purchase, the gentleman had sometimes\nvisited his shop; but he did not recollect to have since seen him. He\nwas unacquainted with his name and his residence.\"\n\n\"What appeared to be his motives for purchasing this picture?\"\n\n\"The customer appeared highly pleased with it. Pleasure, rather than\nsurprise, seemed to be produced by the sight of it. If I were permitted\nto judge,\" continued the artist, \"I should imagine that the young man\nwas acquainted with the original. To say the truth, I hinted as much at\nthe time, and I did not see that he discouraged the supposition. Indeed,\nI cannot conceive how the picture could otherwise have gained any value\nin his eyes.\"\n\nThis only heightened the eagerness of Constantia to trace the footsteps\nof the youth. It was obvious to suppose some communication or connection\nbetween her friend and this purchaser. She repeated her inquiries, and\nthe goldsmith, after some consideration, said, \"Why, on second thoughts,\nI seem to have some notion of having seen a figure like that of my\ncustomer go into a lodging-house in Front Street, some time before I met\nwith him at my shop.\"\n\nThe situation of this house being satisfactorily described, and the\nartist being able to afford her no further information, except as to\nstature and guise, she took her leave. There were two motives impelling\nher to prosecute her search after this person,--the desire of regaining\nthis portrait and of procuring tidings of her friend. Involved as she\nwas in ignorance, it was impossible to conjecture how far this incident\nwould be subservient to these inestimable purposes. To procure an\ninterview with this stranger was the first measure which prudence\nsuggested.\n\nShe knew not his name or his person. He was once seen entering a\nlodging-house. Thither she must immediately repair; but how to introduce\nherself, how to describe the person of whom she was in search, she knew\nnot. She was beset with embarrassments and difficulties. While her\nattention was entangled by these, she proceeded unconsciously on her\nway, and stopped not until she reached the mansion that had been\ndescribed. Here she paused to collect her thoughts.\n\nShe found no relief in deliberation. Every moment added to her\nperplexity and indecision. Irresistibly impelled by her wishes, she at\nlength, in a mood that partook of desperate, advanced to the door and\nknocked. The summons was immediately obeyed by a woman of decent\nappearance. A pause ensued, which Constantia at length terminated by a\nrequest to see the mistress of the house.\n\nThe lady courteously answered that she was the person, and immediately\nushered her visitant into an apartment. Constantia being seated, the\nlady waited for the disclosure of her message. To prolong the silence\nwas only to multiply embarrassments. She reverted to the state of her\nfeelings, and saw that they flowed from inconsistency and folly. One\nvigorous effort was sufficient to restore her to composure and\nself-command.\n\nShe began with apologizing for a visit unpreceded by an introduction.\nThe object of her inquiries was a person with whom it was of the utmost\nmoment that she should procure a meeting, but whom, by an unfortunate\nconcurrence of circumstances, she was unable to describe by the usual\nincidents of name and profession. Her knowledge was confined to his\nexternal appearance, and to the probability of his being an inmate of\nthis house at the beginning of the year. She then proceeded to describe\nhis person and dress.\n\n\"It is true,\" said the lady; \"such a one as you describe has boarded in\nthis house. His name was Martynne. I have good reason to remember him,\nfor he lived with me three months, and then left the country without\npaying for his board.\"\n\n\"He has gone, then?\" said Constantia, greatly discouraged by these\ntidings.\n\n\"Yes. He was a man of specious manners and loud pretensions. He came\nfrom England, bringing with him forged recommendatory letters, and,\nafter passing from one end of the country to the other, contracting\ndebts which he never paid and making bargains which he never fulfilled,\nhe suddenly disappeared. It is likely that he has returned to Europe.\"\n\n\"Had he no kindred, no friends, no companions?\"\n\n\"He found none here. He made pretences to alliances in England, which\nbetter information has, I believe, since shown to be false.\"\n\nThis was the sum of the information procurable from this source.\nConstantia was unable to conceal her chagrin. These symptoms were\nobserved by the lady, whose curiosity was awakened in turn. Questions\nwere obliquely started, inviting Constantia to a disclosure of her\nthoughts. No advantage would arise from confidence, and the guest, after\na few minutes of abstraction and silence, rose to take her leave.\n\nDuring this conference, some one appeared to be negligently sporting\nwith the keys of a harpsichord, in the next apartment. The notes were\ntoo irregular and faint to make a forcible impression on the ear. In the\npresent state of her mind, Constantia was merely conscious of the sound,\nin the intervals of conversation. Having arisen from her seat, her\nanxiety to obtain some information that might lead to the point she\nwished made her again pause. She endeavoured to invent some new\ninterrogatory better suited to her purpose than those which had already\nbeen employed. A silence on both sides ensued.\n\nDuring this interval, the unseen musician suddenly refrained from\nrambling, and glided into notes of some refinement and complexity. The\ncadence was aerial; but a thunderbolt, falling at her feet, would not\nhave communicated a more visible shock to the senses of Constantia. A\nglance that denoted a tumult of soul bordering on distraction was now\nfixed upon the door that led into the room from whence the harmony\nproceeded. Instantly the cadence was revived, and some accompanying\nvoice was heard to warble,--\n\n     \"Ah! far beyond this world of woes\n     We meet to part,--to part no more.\"\n\nJoy and grief, in their sudden onset and their violent extremes,\napproach so nearly in their influence on human beings as scarce to be\ndistinguished. Constantia's frame was still enfeebled by her recent\ndistresses. The torrent of emotion was too abrupt and too vehement. Her\nfaculties were overwhelmed, and she sunk upon the floor motionless and\nwithout sense, but not till she had faintly articulated,--\n\n\"My God! My God! This is a joy unmerited and too great.\"\n\n\n\n\nCHAPTER IV.\n\n\nI must be forgiven if I now introduce myself on the stage. Sophia\nWestwyn is the friend of Constantia, and the writer of this narrative.\nSo far as my fate was connected with that of my friend, it is worthy to\nbe known. That connection has constituted the joy and misery of my\nexistence, and has prompted me to undertake this task.\n\nI assume no merit from the desire of knowledge and superiority to\ntemptation. There is little of which I can boast; but that little I\nderived, instrumentally, from Constantia. Poor as my attainments are, it\nis to her that I am indebted for them all. Life itself was the gift of\nher father, but my virtue and felicity are her gifts. That I am neither\nindigent nor profligate, flows from her bounty.\n\nI am not unaware of the divine superintendence,--of the claims upon my\ngratitude and service which pertain to my God. I know that all physical\nand moral agents are merely instrumental to the purpose that he wills;\nbut, though the great Author of being and felicity must not be\nforgotten, it is neither possible nor just to overlook the claims upon\nour love with which our fellow-beings are invested.\n\nThe supreme love does not absorb, but chastens and enforces, all\nsubordinate affections. In proportion to the rectitude of my perceptions\nand the ardour of my piety, must I clearly discern and fervently love\nthe excellence discovered in my fellow-beings, and industriously promote\ntheir improvement and felicity.\n\nFrom my infancy to my seventeenth year, I lived in the house of Mr.\nDudley. On the day of my birth I was deserted by my mother. Her temper\nwas more akin to that of tigress than woman. Yet that is unjust; for\nbeasts cherish their offspring. No natures but human are capable of that\ndepravity which makes insensible to the claims of innocence and\nhelplessness.\n\nBut let me not recall her to memory. Have I not enough of sorrow? Yet to\nomit my causes of disquiet, the unprecedented forlornness of my\ncondition, and the persecutions of an unnatural parent, would be to\nleave my character a problem, and the sources of my love of Miss Dudley\nunexplored. Yet I must not dwell upon that complication of iniquities,\nthat savage ferocity and unextinguishable hatred of me, which\ncharacterized my unhappy mother.\n\nI was not safe under the protection of Mr. Dudley, nor happy in the\ncaresses of his daughter. My mother asserted the privilege of that\nrelation: she laboured for years to obtain the control of my person and\nactions, to snatch me from a peaceful and chaste asylum, and detain me\nin her own house, where, indeed, I should not have been in want of\nraiment and food; but where--\n\nO my mother! Let me not dishonour thy name! Yet it is not in my power to\nenhance thy infamy. Thy crimes, unequalled as they were, were perhaps\nexpiated by thy penitence. Thy offences are too well known; but perhaps\nthey who witnessed thy freaks of intoxication, thy defiance of public\nshame, the enormity of thy pollutions, the infatuation that made thee\nglory in the pursuit of a loathsome and detestable trade, may be\nstrangers to the remorse and the abstinence which accompanied the close\nof thy ignominious life.\n\nFor ten years was my peace incessantly molested by the menaces or\nmachinations of my mother. The longer she meditated my destruction, the\nmore tenacious of her purpose and indefatigable in her efforts she\nbecame. That my mind was harassed with perpetual alarms was not enough.\nThe fame and tranquillity of Mr. Dudley and his daughter were hourly\nassailed. My mother resigned herself to the impulses of malignity and\nrage. Headlong passions, and a vigorous though perverted understanding,\nwere hers. Hence, her stratagems to undermine the reputation of my\nprotector, and to bereave him of domestic comfort, were subtle and\nprofound. Had she not herself been careless of that good which she\nendeavoured to wrest from others, her artifices could scarcely have been\nfrustrated.\n\nIn proportion to the hazard which accrued to my protector and friend,\nthe more ardent their zeal in my defence and their affection for my\nperson became. They watched over me with ineffable solicitude. At all\nhours and in every occupation, I was the companion of Constantia. All my\nwants were supplied in the same proportion as hers. The tenderness of\nMr. Dudley seemed equally divided between us. I partook of his\ninstructions, and the means of every intellectual and personal\ngratification were lavished upon me.\n\nThe speed of my mother's career in infamy was at length slackened. She\nleft New York, which had long been the theatre of her vices. Actuated by\na now caprice, she determined to travel through the Southern States.\nEarly indulgence was the cause of her ruin, but her parents had given\nher the embellishments of a fashionable education. She delighted to\nassume all parts, and personate the most opposite characters. She now\nresolved to carry a new name, and the mask of virtue, into scenes\nhitherto unvisited.\n\nShe journeyed as far as Charleston. Here she met an inexperienced youth,\nlately arrived from England, and in possession of an ample fortune. Her\nspeciousness and artifices seduced him into a precipitate marriage. Her\ntrue character, however, could not be long concealed by herself, and her\nvices had been too conspicuous for her long to escape recognition. Her\nhusband was infatuated by her blandishments. To abandon her, or to\ncontemplate her depravity with unconcern, were equally beyond his power.\nRomantic in his sentiments, his fortitude was unequal to his\ndisappointments, and he speedily sunk into the grave. By a similar\nrefinement in generosity, he bequeathed to her his property.\n\nWith this accession of wealth, she returned to her ancient abode. The\nmask lately worn seemed preparing to be thrown aside, and her profligate\nhabits to be resumed with more eagerness than ever; but an unexpected\nand total revolution was effected, by the exhortations of a Methodist\ndivine. Her heart seemed, on a sudden, to be remoulded, her vices and\nthe abettors of them were abjured, she shut out the intrusions of\nsociety, and prepared to expiate, by the rigours of abstinence and the\nbitterness of tears, the offences of her past life.\n\nIn this, as in her former career, she was unacquainted with restraint\nand moderation. Her remorses gained strength in proportion as she\ncherished them. She brooded over the images of her guilt, till the\npossibility of forgiveness and remission disappeared. Her treatment of\nher daughter and her husband constituted the chief source of her\ntorment. Her awakened conscience refused her a momentary respite from\nits persecutions. Her thoughts became, by rapid degrees, tempestuous and\ngloomy, and it was at length evident that her condition was maniacal.\n\nIn this state, she was to me an object, no longer of terror, but\ncompassion. She was surrounded by hirelings, devoid of personal\nattachment, and anxious only to convert her misfortunes to their own\nadvantage. This evil it was my duty to obviate. My presence, for a time,\nonly enhanced the vehemence of her malady; but at length it was only by\nmy attendance and soothing that she was diverted from the fellest\npurposes. Shocking execrations and outrages, resolutions and efforts to\ndestroy herself and those around her, were sure to take place in my\nabsence. The moment I appeared before her, her fury abated, her\ngesticulations were becalmed, and her voice exerted only in incoherent\nand pathetic lamentations.\n\nThese scenes, though so different from those which I had formerly been\ncondemned to witness, were scarcely less excruciating. The friendship of\nConstantia Dudley was my only consolation. She took up her abode with\nme, and shared with me every disgustful and perilous office which my\nmother's insanity prescribed.\n\nOf this consolation, however, it was my fate to be bereaved. My mother's\nstate was deplorable, and no remedy hitherto employed was efficacious. A\nvoyage to England was conceived likely to benefit, by change of\ntemperature and scenes, and by the opportunity it would afford of trying\nthe superior skill of English physicians. This scheme, after various\nstruggles on my part, was adopted. It was detestable to my imagination,\nbecause it severed me from that friend in whose existence mine was\ninvolved, and without whose participation knowledge lost its attractions\nand society became a torment.\n\nThe prescriptions of my duty could not be disguised or disobeyed, and we\nparted. A mutual engagement was formed to record every sentiment and\nrelate every event that happened in the life of either, and no\nopportunity of communicating information was to be omitted. This\nengagement was punctually performed on my part. I sought out every\nmethod of conveyance to my friend, and took infinite pains to procure\ntidings from her; but all were ineffectual.\n\nMy mother's malady declined, but was succeeded by a pulmonary disease,\nwhich threatened her speedy destruction. By the restoration of her\nunderstanding, the purpose of her voyage was obtained, and my impatience\nto return, which the inexplicable and ominous silence of my friend daily\nincreased, prompted me to exert all my powers of persuasion to induce\nher to revisit America.\n\nMy mother's frenzy was a salutary crisis in her moral history. She\nlooked back upon her past conduct with unspeakable loathing, but this\nretrospect only invigorated her devotion and her virtue; but the thought\nof returning to the scene of her unhappiness and infamy could not be\nendured. Besides, life, in her eyes, possessed considerable attractions,\nand her physicians flattered her with recovery from her present disease,\nif she would change the atmosphere of England for that of Languedoc and\nNaples.\n\nI followed her with murmurs and reluctance. To desert her in her present\ncritical state would have been inhuman. My mother's aversions and\nattachments, habits and views, were dissonant with my own. Conformity of\nsentiments and impressions of maternal tenderness did not exist to bind\nus to each other. My attendance was assiduous, but it was the sense of\nduty that rendered my attendance a supportable task.\n\nHer decay was eminently gradual. No time seemed to diminish her appetite\nfor novelty and change. During three years we traversed every part of\nFrance, Switzerland, and Italy. I could not but attend to surrounding\nscenes, and mark the progress of the mighty revolution, whose effects,\nlike agitation in a fluid, gradually spread from Paris, the centre, over\nthe face of the neighbouring kingdoms; but there passed not a day or an\nhour in which the image of Constantia was not recalled, in which the\nmost pungent regrets were not felt at the inexplicable silence which had\nbeen observed by her, and the most vehement longings indulged to return\nto my native country. My exertions to ascertain her condition by\nindirect means, by interrogating natives of America with whom I chanced\nto meet, were unwearied, but, for a long period, ineffectual.\n\nDuring this pilgrimage, Rome was thrice visited. My mother's\nindisposition was hastening to a crisis, and she formed the resolution\nof closing her life at the bottom of Vesuvius. We stopped, for the sake\nof a few days' repose, at Rome. On the morning after our arrival, I\naccompanied some friends to view the public edifices. Casting my eyes\nover the vast and ruinous interior of the Coliseum, my attention was\nfixed by the figure of a young man whom, after a moment's pause, I\nrecollected to have seen in the streets of New York. At a distance from\nhome, mere community of country is no inconsiderable bond of affection.\nThe social spirit prompts us to cling even to inanimate objects, when\nthey remind us of ancient fellowships and juvenile attachments.\n\nA servant was despatched to summon this stranger, who recognised a\ncountrywoman with a pleasure equal to that which I had received. On\nnearer view, this person, whose name was Courtland, did not belie my\nfavourable prepossessions. Our intercourse was soon established on a\nfooting of confidence and intimacy.\n\nThe destiny of Constantia was always uppermost in my thoughts. This\nperson's acquaintance was originally sought chiefly in the hope of\nobtaining from him some information respecting my friend. On inquiry, I\ndiscovered that he had left his native city seven months after me.\nHaving tasked his recollection and compared a number of facts, the name\nof Dudley at length recurred to him. He had casually heard the history\nof Craig's imposture and its consequences. These were now related as\ncircumstantially as a memory occupied by subsequent incidents enabled\nhim. The tale had been told to him, in a domestic circle which he was\naccustomed to frequent, by the person who purchased Mr. Dudley's lute\nand restored it to its previous owner on the conditions formerly\nmentioned.\n\nThis tale filled me with anguish and doubt. My impatience to search out\nthis unfortunate girl, and share with her her sorrows or relieve them,\nwas anew excited by this mournful intelligence. That Constantia Dudley\nwas reduced to beggary was too abhorrent to my feelings to receive\ncredit; yet the sale of her father's property, comprising even his\nfurniture and clothing, seemed to prove that she had fallen even to this\ndepth. This enabled me in some degree to account for her silence. Her\ngenerous spirit would induce her to conceal misfortunes from her friend\nwhich no communication would alleviate. It was possible that she had\nselected some new abode, and that, in consequence, the letters I had\nwritten, and which amounted to volumes, had never reached her hands.\n\nMy mother's state would not suffer me to obey the impulse of my heart.\nHer frame was verging towards dissolution. Courtland's engagements\nallowed him to accompany us to Naples, and here the long series of my\nmother's pilgrimages closed in death. Her obsequies were no sooner\nperformed, than I determined to set out on my long-projected voyage. My\nmother's property, which, in consequence of her decease, devolved upon\nme, was not inconsiderable. There is scarcely any good so dear to a\nrational being as competence. I was not unacquainted with its benefits,\nbut this acquisition was valuable to mo chiefly as it enabled me to\nreunite my fate to that of Constantia.\n\nCourtland was my countryman and friend. He was destitute of fortune, and\nhad been led to Europe partly by the spirit of adventure, and partly on\na mercantile project. He had made sale of his property on advantageous\nterms, in the ports of France, and resolved to consume the produce in\nexamining this scene of heroic exploits and memorable revolutions. His\nslender stock, though frugally and even parsimoniously administered, was\nnearly exhausted; and, at the time of our meeting at Rome, he was making\nreluctant preparations to return.\n\nSufficient opportunity was afforded us, in an unrestrained and domestic\nintercourse of three months, which succeeded our Roman interview, to\ngain a knowledge of each other. There was that conformity of tastes and\nviews between us which could scarcely fail, at an age and in a situation\nlike ours, to give birth to tenderness. My resolution to hasten to\nAmerica was peculiarly unwelcome to my friend. He had offered to be my\ncompanion, but this offer my regard to his interest obliged me to\ndecline; but I was willing to compensate him for this denial, as well as\nto gratify my own heart, by an immediate marriage.\n\nSo long a residence in England and Italy had given birth to friendships\nand connections of the dearest kind. I had no view but to spend my life\nwith Courtland, in the midst of my maternal kindred, who were English. A\nvoyage to America and reunion with Constantia were previously\nindispensable; but I hoped that my friend might be prevailed upon, and\nthat her disconnected situation would permit her to return with me to\nEurope. If this end could not be accomplished, it was my inflexible\npurpose to live and die with her. Suitably to this arrangement,\nCourtland was to repair to London, and wait patiently till I should be\nable to rejoin him there, or to summon him to meet me in America.\n\nA week after my mother's death, I became a wife, and embarked the next\nday, at Naples, in a Ragusan ship, destined for New York. The voyage was\ntempestuous and tedious. The vessel was necessitated to make a short\nstay at Toulon. The state of that city, however, then in possession of\nthe English and besieged by the revolutionary forces, was adverse to\ncommercial views. Happily, we resumed our voyage on the day previous to\nthat on which the place was evacuated by the British. Our seasonable\ndeparture rescued us from witnessing a scene of horrors of which the\nhistory of former wars furnishes us with few examples.\n\nA cold and boisterous navigation awaited us. My palpitations and\ninquietudes augmented as we approached the American coast. I shall not\nforget the sensations which I experienced on the sight of the Beacon at\nSandy Hook. It was first seen at midnight, in a stormy and beclouded\natmosphere, emerging from the waves, whose fluctuation allowed it, for\nsome time, to be visible only by fits. This token of approaching land\naffected me as much as if I had reached the threshold of my friend's\ndwelling.\n\nAt length we entered the port, and I viewed, with high-raised but\ninexplicable feelings, objects with which I had been from infancy\nfamiliar. The flagstaff erected on the Battery recalled to my\nimagination the pleasures of the evening and morning walks which I had\ntaken on that spot with the lost Constantia. The dream was fondly\ncherished, that the figure which I saw loitering along the terrace was\nhers.\n\nOn disembarking, I gazed at every female passenger, in hope that it was\nshe whom I sought. An absence of three years had obliterated from my\nmemory none of the images which attended me on my departure.\n\n\n\n\nCHAPTER V.\n\n\nAfter a night of repose rather than of sleep, I began the search after\nmy friend. I went to the house which the Dudleys formerly inhabited, and\nwhich had been the asylum of my infancy. It was now occupied by\nstrangers, by whom no account could be given of its former tenants. I\nobtained directions to the owner of the house. He was equally unable to\nsatisfy my curiosity. The purchase had been made at a public sale, and\nterms had been settled, not with Dudley, but with the sheriff.\n\nIt is needless to say that the history of Craig's imposture and its\nconsequences were confirmed by every one who resided at that period in\nNew York. The Dudleys were well remembered, and their disappearance,\nimmediately after their fall, had been generally noticed; but whither\nthey had retired was a problem which no one was able to solve.\n\nThis evasion was strange. By what motives the Dudleys were induced to\nchange their ancient abode could be vaguely guessed. My friend's\ngrandfather was a native of the West Indies. Descendants of the same\nstock still resided in Tobago. They might be affluent, and to them it\nwas possible that Mr. Dudley, in this change of fortune, had betaken\nhimself for relief. This was a mournful expedient, since it would raise\na barrier between my friend and myself scarcely to be surmounted.\n\nConstantia's mother was stolen by Mr. Dudley from a convent at Amiens.\nThere were no affinities, therefore, to draw them to France. Her\ngrandmother was a native of Baltimore, of a family of some note, by name\nRidgeley. This family might still exist, and have either afforded an\nasylum to the Dudleys, or, at least, be apprized of their destiny. It\nwas obvious to conclude that they no longer existed within the precincts\nof New York. A journey to Baltimore was the next expedient.\n\nThis journey was made in the depth of winter, and by the speediest\nconveyance. I made no more than a day's sojourn in Philadelphia. The\nepidemic by which that city had been lately ravaged, I had not heard of\ntill my arrival in America. Its devastations were then painted to my\nfancy in the most formidable colours. A few months only had elapsed\nsince its extinction, and I expected to see numerous marks of misery and\ndepopulation.\n\nTo my no small surprise, however, no vestiges of this calamity were to\nbe discerned. All houses were open, all streets thronged, and all faces\nthoughtless or busy. The arts and the amusements of life seemed as\nsedulously cultivated as ever. Little did I then think what had been,\nand what at that moment was, the condition of my friend. I stopped for\nthe sake of respite from fatigue, and did not, therefore, pass much time\nin the streets. Perhaps, had I walked seasonably abroad, we might have\nencountered each other, and thus have saved ourselves from a thousand\nanxieties.\n\nAt Baltimore I made myself known, without the formality of introduction,\nto the Ridgeleys. They acknowledged their relationship to Mr. Dudley,\nbut professed absolute ignorance of his fate. Indirect intercourse only\nhad been maintained, formerly, by Dudley with his mother's kindred. They\nhad heard of his misfortune a twelvemonth after it happened; but what\nmeasures had been subsequently pursued, their kinsman had not thought\nproper to inform them.\n\nThe failure of this expedient almost bereft me of hope. Neither my own\nimagination nor the Ridgeleys could suggest any new mode by which my\npurpose was likely to be accomplished. To leave America without\nobtaining the end of my visit could not be thought of without agony; and\nyet the continuance of my stay promised me no relief from my\nuncertainties.\n\nOn this theme I ruminated without ceasing. I recalled every conversation\nand incident of former times, and sought in them a clue by which my\npresent conjectures might be guided. One night, immersed alone in my\nchamber, my thoughts were thus employed. My train of meditation was, on\nthis occasion, new. From the review of particulars from which no\nsatisfaction had hitherto been gained, I passed to a vague and\ncomprehensive retrospect.\n\nMr. Dudley's early life, his profession of a painter, his zeal in this\npursuit, and his reluctance to quit it, were remembered. Would he not\nrevert to this profession when other means of subsistence were gone? It\nis true, similar obstacles with those which had formerly occasioned his\nresort to a different path existed at present, and no painter of his\nname was to be found in Philadelphia, Baltimore, or New York. But would\nit not occur to him, that the patronage denied to his skill by the\nfrugal and unpolished habits of his countrymen might, with more\nprobability of success, be sought from the opulence and luxury of\nLondon? Nay, had he not once affirmed, in my hearing, that, if he ever\nwere reduced to poverty, this was the method he would pursue?\n\nThis conjecture was too bewitching to be easily dismissed. Every new\nreflection augmented its force. I was suddenly raised by it from the\ndeepest melancholy to the region of lofty and gay hopes. Happiness, of\nwhich I had begun to imagine myself irretrievably bereft, seemed once\nmore to approach within my reach. Constantia would not only be found,\nbut be met in the midst of those comforts which her father's skill could\nnot fail to procure, and on that very stage where I most desired to\nencounter her. Mr. Dudley had many friends and associates of his youth\nin London. Filial duty had repelled their importunities to fix his abode\nin Europe, when summoned home by his father. On his father's death these\nsolicitations had been renewed, but were disregarded for reasons which\nhe, afterwards, himself confessed were fallacious. That they would a\nthird time be preferred, and would regulate his conduct, seemed to me\nincontestable.\n\nI regarded with wonder and deep regret the infatuation that had\nhitherto excluded these images from my understanding and my memory. How\nmany dangers and toils had I endured since my embarkation at Naples, to\nthe present moment! How many lingering minutes had I told since my first\ninterview with Courtland! All were owing to my own stupidity. Had my\npresent thoughts been seasonably suggested, I might long since have been\nrestored to the embraces of my friend, without the necessity of an\nhour's separation from my husband.\n\nThese were evils to be repaired as far as it was possible. Nothing now\nremained but to procure a passage to Europe. For this end diligent\ninquiries were immediately set on foot. A vessel was found, which, in a\nfew weeks, would set out upon the voyage. Having bespoken a conveyance,\nit was incumbent on me to sustain with patience the unwelcome delay.\n\nMeanwhile, my mind, delivered from the dejection and perplexities that\nlately haunted it, was capable of some attention to surrounding objects.\nI marked the peculiarities of manners and language in my new abode, and\nstudied the effects which a political and religious system so opposite\nto that with which I had conversed in Italy and Switzerland had\nproduced. I found that the difference between Europe and America lay\nchiefly in this:--that, in the former, all things tended to extremes,\nwhereas, in the latter, all things tended to the same level. Genius, and\nvirtue, and happiness, on these shores, were distinguished by a sort of\nmediocrity. Conditions were less unequal, and men were strangers to the\nheights of enjoyment and the depths of misery to which the inhabitants\nof Europe are accustomed.\n\nI received friendly notice and hospitable treatment from the Ridgeleys.\nThese people were mercantile and plodding in their habits. I found in\ntheir social circle little exercise for the sympathies of my heart, and\nwillingly accepted their aid to enlarge the sphere of my observation.\n\nAbout a week before my intended embarkation, and when suitable\npreparation had been made for that event, a lady arrived in town, who\nwas cousin to my Constantia. She had frequently been mentioned in\nfavourable terms in my hearing. She had passed her life in a rural\nabode with her father, who cultivated his own domain, lying forty miles\nfrom Baltimore.\n\nOn an offer being made to introduce us to each other, I consented to\nknow one whose chief recommendation in my eyes consisted in her affinity\nto Constantia Dudley. I found an artless and attractive female,\nunpolished and undepraved by much intercourse with mankind. At first\nsight, I was powerfully struck by the resemblance of her features to\nthose of my friend, which sufficiently denoted their connection with a\ncommon stock.\n\nThe first interview afforded mutual satisfaction. On our second meeting,\ndiscourse insensibly led to the mention of Miss Dudley, and of the\ndesign which had brought me to America. She was deeply affected by the\nearnestness with which I expatiated on her cousin's merits, and by the\nproofs which my conduct had given of unlimited attachment.\n\nI dwelt immediately on the measures which I had hitherto ineffectually\npursued to trace her footsteps, and detailed the grounds of my present\nbelief that we should meet in London. During this recital, my companion\nsighed and wept. When I finished my tale, her tears, instead of ceasing,\nflowed with new vehemence. This appearance excited some surprise, and I\nventured to ask the cause of her grief.\n\n\"Alas!\" she replied, \"I am personally a stranger to my cousin, but her\ncharacter has been amply displayed to me by one who knew her well. I\nweep to think how much she has suffered. How much excellence we have\nlost!\"\n\n\"Nay,\" said I, \"all her sufferings will, I hope, be compensated, and I\nby no means consider her as lost. If my search in London be\nunsuccessful, then shall I indeed despair.\"\n\n\"Despair, then, already,\" said my sobbing companion, \"for your search\nwill be unsuccessful. How I feel for your disappointment! but it cannot\nbe known too soon. My cousin is dead!\"\n\nThese tidings were communicated with tokens of sincerity and sorrow that\nleft me no room to doubt that they were believed by the relater. My own\nemotions were suspended till interrogations had obtained a knowledge of\nher reasons for crediting this fatal event, and till she had explained\nthe time and manner of her death. A friend of Miss Ridgeley's father had\nwitnessed the devastations of the yellow fever in Philadelphia. He was\napprized of the relationship that subsisted between his friend and the\nDudleys. He gave a minute and circumstantial account of the arts of\nCraig. He mentioned the removal of my friends to Philadelphia, their\nobscure and indigent life, and, finally, their falling victims to the\npestilence.\n\nHe related the means by which he became apprized of their fate, and drew\na picture of their death, surpassing all that imagination can conceive\nof shocking and deplorable. The quarter where they lived was nearly\ndesolate. Their house was shut up, and, for a time, imagined to be\nuninhabited. Some suspicions being awakened in those who superintended\nthe burial of the dead, the house was entered, and the father and child\ndiscovered to be dead. The former was stretched upon his wretched\npallet, while the daughter was found on the floor of the lower room, in\na state that denoted the sufferance not only of disease, but of famine.\n\nThis tale was false. Subsequent discoveries proved this to be a\ndetestable artifice of Craig, who, stimulated by incurable habits, had\ninvented these disasters, for the purpose of enhancing the opinion of\nhis humanity and of furthering his views on the fortune and daughter of\nMr. Ridgeley.\n\nIts falsehood, however, I had as yet no means of ascertaining. I\nreceived it as true, and at once dismissed all my claims upon futurity.\nAll hope of happiness, in this mutable and sublunary scene, was fled.\nNothing remained but to join my friend in a world where woes are at an\nend and virtue finds recompense. \"Surely,\" said I, \"there will some time\nbe a close to calamity and discord. To those whose lives have been\nblameless, but harassed by inquietudes to which not their own but the\nerrors of others have given birth, a fortress will hereafter be\nassigned unassailable by change, impregnable to sorrow.\n\n\"O my ill-fated Constantia! I will live to cherish thy remembrance, and\nto emulate thy virtue. I will endure the privation of thy friendship and\nthe vicissitudes that shall befall me, and draw my consolation and\ncourage from the foresight of no distant close to this terrestrial\nscene, and of ultimate and everlasting union with thee.\"\n\nThis consideration, though it kept me from confusion and despair, could\nnot, but with the healing aid of time, render me tranquil or strenuous.\nMy strength was unequal to the struggle of my passions. The ship in\nwhich I engaged to embark could not wait for my restoration to health,\nand I was left behind.\n\nMary Ridgeley was artless and affectionate. She saw that her society was\ndearer to me than that of any other, and was therefore seldom willing to\nleave my chamber. Her presence, less on her own account than by reason\nof her personal resemblance and her affinity by birth to Constantia, was\na powerful solace.\n\nI had nothing to detain me longer in America. I was anxious to change my\npresent lonely state, for the communion of those friends in England, and\nthe performance of those duties, which were left to me. I was informed\nthat a British packet would shortly sail from New York. My frame was\nsunk into greater weakness than I had felt at any former period; and I\nconceived that to return to New York by water was more commodious than\nto perform the journey by land.\n\nThis arrangement was likewise destined to be disappointed. One morning I\nvisited, according to my custom, Mary Ridgeley. I found her in a temper\nsomewhat inclined to gayety. She rallied me, with great archness, on the\ncare with which I had concealed from her a tender engagement into which\nI had lately entered.\n\nI supposed myself to comprehend her allusion, and therefore answered\nthat accident, rather than design, had made me silent on the subject of\nmarriage. She had hitherto known me by no appellation but Sophia\nCourtland. I had thought it needless to inform her that I was indebted\nfor my name to my husband, Courtland being his name.\n\n\"All that,\" said my friend, \"I know already. And so you sagely think\nthat my knowledge goes no further than that? We are not bound to love\nour husbands longer than their lives. There is no crime, I believe, in\nreferring the living to the dead; and most heartily do congratulate you\non your present choice.\"\n\n\"What mean you? I confess, your discourse surpasses my comprehension.\"\n\nAt that moment the bell at the door rung a loud peal. Miss Ridgeley\nhastened down at this signal, saying, with much significance,--\n\n\"I am a poor hand at solving a riddle. Here comes one who, if I mistake\nnot, will find no difficulty in clearing up your doubts.\"\n\nPresently she came up, and said, with a smile of still greater archness,\n\"Here is a young gentleman, a friend of mine, to whom I must have the\npleasure of introducing you. He has come for the special purpose of\nsolving my riddle.\" I attended her to the parlour without hesitation.\n\nShe presented me, with great formality, to a youth, whose appearance did\nnot greatly prepossess me in favour of his judgement. He approached me\nwith an air supercilious and ceremonious; but the moment he caught a\nglance at my face, he shrunk back, visibly confounded and embarrassed. A\npause ensued, in which Miss Ridgeley had opportunity to detect the error\ninto which she had been led by the vanity of this young man.\n\n\"How now, Mr. Martynne!\" said my friend, in a tone of ridicule; \"is it\npossible you do not know the lady who is the queen of your affections,\nthe tender and indulgent fair one whose portrait you carry in your\nbosom, and whose image you daily and nightly bedew with your tears and\nkisses?\"\n\nMr. Martynne's confusion, instead of being subdued by his struggle, only\ngrew more conspicuous; and, after a few incoherent speeches and\napologies, during which he carefully avoided encountering my eyes, he\nhastily departed.\n\nI applied to my friend, with great earnestness, for an explanation of\nthis scene. It seems that, in the course of conversation with him on the\npreceding day, he had suffered a portrait which hung at his breast to\ncatch Miss Ridgeley's eye. On her betraying a desire to inspect it more\nnearly, he readily produced it. My image had been too well copied by the\nartist not to be instantly recognised.\n\nShe concealed her knowledge of the original, and, by questions well\nadapted to the purpose, easily drew from him confessions that this was\nthe portrait of his mistress. He let fall sundry innuendoes and\nsurmises, tending to impress her with a notion of the rank, fortune, and\nintellectual accomplishments of the nymph, and particularly of the\ndoting fondness and measureless confidence with which she regarded him.\n\nHer imperfect knowledge of my situation left her in some doubt as to the\ntruth of these pretensions, and she was willing to ascertain the truth\nby bringing about an interview. To guard against evasions and artifice\nin the lover, she carefully concealed from him her knowledge of the\noriginal, and merely pretended that a friend of hers was far more\nbeautiful than her whom this picture represented. She added, that she\nexpected a visit from her friend the next morning, and was willing, by\nshowing her to Mr. Martynne, to convince him how much he was mistaken in\nsupposing the perfections of his mistress unrivalled.\n\n\n\n\nCHAPTER VI.\n\n\nMartynne, while ho expressed his confidence that the experiment would\nonly confirm his triumph, readily assented to the proposal, and the\ninterview above described took place, accordingly, the next morning. Had\nhe not been taken by surprise, it is likely the address of a man who\npossessed no contemptible powers would have extricated him from some of\nhis embarrassment.\n\nThat my portrait should be in the possession of one whom I had never\nbefore seen, and whose character and manners entitled him to no respect,\nwas a source of some surprise. This mode of multiplying faces is\nextremely prevalent in this age, and was eminently characteristic of\nthose with whom I had associated in different parts of Europe. The\nnature of my thoughts had modified my features into an expression which\nmy friends were pleased to consider as a model for those who desired to\npersonify the genius of suffering and resignation.\n\nHence, among those whose religion permitted their devotion to a picture\nof a female, the symbols of their chosen deity were added to features\nand shape that resembled mine. My own caprice, as well as that of\nothers, always dictated a symbolical, and, in every new instance, a\ndifferent accompaniment of this kind. Hence was offered the means of\ntracing the history of that picture which Martynne possessed.\n\nIt had been accurately examined by Miss Ridgeley, and her description of\nthe frame in which it was placed instantly informed me that it was the\nsame which, at our parting, I left in the possession of Constantia. My\nfriend and myself were desirous of employing the skill of a Saxon\npainter, by name Eckstein. Each of us were drawn by him, she with the\ncincture of Venus, and I with the crescent of Dian. This symbol was\nstill conspicuous on the brow of that image which Miss Ridgeley had\nexamined, and served to identify the original proprietor.\n\nThis circumstance tended to confirm my fears that Constantia was dead,\nsince that she would part with this picture during her life was not to\nbe believed. It was of little moment to discover how it came into the\nhands of the present possessor. Those who carried her remains to the\ngrave had probably torn it from her neck and afterwards disposed of it\nfor money.\n\nBy whatever means, honest or illicit, it had been acquired by Martynne,\nit was proper that it should be restored to me. It was valuable to me,\nbecause it had been the property of one whom I loved, and it might prove\nhighly injurious to my fame and my happiness, as the tool of this man's\nvanity and the attestor of his falsehood. I therefore wrote him a\nletter, acquainting him with my reasons for desiring the repossession of\nthis picture, and offering a price for it at least double its value as a\nmere article of traffic. Martynne accepted the terms. He transmitted the\npicture, and with it a note, apologizing for the artifice of which he\nhad been guilty, and mentioning, in order to justify his acceptance of\nthe price which I had offered, that he had lately purchased it for an\nequal sum, of a goldsmith in Philadelphia.\n\nThis information suggested a new reflection. Constantia had engaged to\npreserve, for the use of her friend, copious and accurate memorials of\nher life. Copies of these were, on suitable occasions, to be transmitted\nto me during my residence abroad. These I had never received, but it was\nhighly probable that her punctuality, in the performance of the first\npart of her engagement, had been equal to my own.\n\nWhat, I asked, had become of these precious memorials? In the wreck of\nher property were these irretrievably engulfed? It was not probable that\nthey had been wantonly destroyed. They had fallen, perhaps, into hands\ncareless or unconscious of their value, or still lay, unknown and\nneglected, at the bottom of some closet or chest. Their recovery might\nbe effected by vehement exertions, or by some miraculous accident.\nSuitable inquiries, carried on among those who were active in those\nscenes of calamity, might afford some clue by which the fate of the\nDudleys, and the disposition of their property, might come into fuller\nlight. These inquiries could be made only in Philadelphia, and thither,\nfor that purpose, I now resolved to repair. There was still an interval\nof some weeks before the departure of the packet in which I proposed to\nembark.\n\nHaving returned to the capital, I devoted all my zeal to my darling\nproject. My efforts, however, were without success. Those who\nadministered charity and succour during that memorable season, and who\nsurvived, could remove none of my doubts, nor answer any of my\ninquiries. Innumerable tales, equally disastrous with those which Miss\nRidgeley had heard, were related; but, for a considerable period, none\nof their circumstances were sufficiently accordant with the history of\nthe Dudleys.\n\nIt is worthy of remark, in how many ways, and by what complexity of\nmotives, human curiosity is awakened and knowledge obtained. By its\nconnection with my darling purpose, every event in the history of this\nmemorable pest was earnestly sought and deeply pondered. The powerful\nconsiderations which governed me made me slight those punctilious\nimpediments which, in other circumstances, would have debarred me from\nintercourse with the immediate actors and observers. I found none who\nwere unwilling to expatiate on this topic, or to communicate the\nknowledge they possessed. Their details were copious in particulars and\nvivid in minuteness. They exhibited the state of manners, the\ndiversified effects of evil or heroic passions, and the endless forms\nwhich sickness and poverty assume in the obscure recesses of a\ncommercial and populous city.\n\nSome of these details are too precious to be lost. It is above all\nthings necessary that we should be thoroughly acquainted with the\ncondition of our fellow-beings. Justice and compassion are the fruit of\nknowledge. The misery that overspreads so large a part of mankind exists\nchiefly because those who are able to relieve it do not know that it\nexists. Forcibly to paint the evil, seldom fails to excite the virtue of\nthe spectator and seduce him into wishes, at least, if not into\nexertions, of beneficence.\n\nThe circumstances in which I was placed were, perhaps, wholly singular.\nHence, the knowledge I obtained was more comprehensive and authentic\nthan was possessed by any one, even of the immediate actors or\nsufferers. This knowledge will not be useless to myself or to the world.\nThe motives which dictated the present narrative will hinder me from\nrelinquishing the pen till my fund of observation and experience be\nexhausted. Meanwhile, let me resume the thread of my tale.\n\nThe period allowed me before my departure was nearly expired, and my\npurpose seemed to be as far from its accomplishment as ever. One evening\nI visited a lady who was the widow of a physician whose disinterested\nexertions had cost him his life. She dwelt with pathetic earnestness on\nthe particulars of her own distress, and listened with deep attention to\nthe inquiries and doubts which I had laid before her.\n\nAfter a pause of consideration, she said that an incident like that\nrelated by me she had previously heard from one of her friends, whose\nname she mentioned. This person was one of those whose office consisted\nin searching out the sufferers, and affording them unsought and\nunsolicited relief. She was offering to introduce me to this person,\nwhen he entered the apartment.\n\nAfter the usual compliments, my friend led the conversation as I wished.\nBetween Mr. Thompson's tale and that related to Miss Ridgeley there was\nan obvious resemblance. The sufferers resided in an obscure alley. They\nhad shut themselves up from all intercourse with their neighbours, and\nhad died, neglected and unknown. Mr. Thompson was vested with the\nsuperintendence of this district, and had passed the house frequently\nwithout suspicion of its being tenanted.\n\nHe was at length informed, by one of those who conducted a hearse, that\nhe had seen the window in the upper story of this house lifted and a\nfemale show herself. It was night, and the hearseman chanced to be\npassing the door. He immediately supposed that the person stood in need\nof his services, and stopped.\n\nThis procedure was comprehended by the person at the window, who,\nleaning out, addressed him in a broken and feeble voice. She asked him\nwhy he had not taken a different route, and upbraided him for inhumanity\nin leading his noisy vehicle past her door. She wanted repose, but the\nceaseless rumbling of his wheels would not allow her the sweet respite\nof a moment.\n\nThis invective was singular, and uttered in a voice which united the\nutmost degree of earnestness with a feebleness that rendered it almost\ninarticulate. The man was at a loss for a suitable answer. His pause\nonly increased the impatience of the person at the window, who called\nupon him, in a still more anxious tone, to proceed, and entreated him to\navoid this alley for the future.\n\nHe answered that he must come whenever the occasion called him; that\nthree persons now lay dead in this alley, and that he must be\nexpeditious in their removal; but that he would return as seldom and\nmake as little noise as possible.\n\nHe was interrupted by new exclamations and upbraidings. These terminated\nin a burst of tears, and assertions that God and man were her\nenemies,--that they were determined to destroy her; but she trusted that\nthe time would come when their own experience would avenge her wrongs,\nand teach them some compassion for the misery of others. Saying this,\nshe shut the window with violence, and retired from it, sobbing with a\nvehemence that could be distinctly overheard by him in the street.\n\nHe paused for some time, listening when this passion should cease. The\nhabitation was slight, and he imagined that he heard her traversing the\nfloor. While he stayed, she continued to vent her anguish in\nexclamations and sighs and passionate weeping. It did not appear that\nany other person was within.\n\nMr. Thompson, being next day informed of these incidents, endeavoured to\nenter the house; but his signals, though loud and frequently repeated,\nbeing unnoticed, he was obliged to gain admission by violence. An old\nman, and a female lovely in the midst of emaciation and decay, were\ndiscovered without signs of life. The death of the latter appeared to\nhave been very recent.\n\nIn examining the house, no traces of other inhabitants were to be found.\nNothing serviceable as food was discovered, but the remnants of mouldy\nbread scattered on a table. No information could be gathered from\nneighbours respecting the condition and name of these unfortunate\npeople. They had taken possession of this house during the rage of this\nmalady, and refrained from all communication with their neighbours.\n\nThere was too much resemblance between this and the story formerly\nheard, not to produce the belief that they related to the same persons.\nAll that remained was to obtain directions to the proprietor of this\ndwelling, and exact from him all that he knew respecting his tenants.\n\nI found in him a man of worth and affability. He readily related, that a\nman applied to him for the use of this house, and that the application\nwas received. At the beginning of the pestilence, a numerous family\ninhabited this tenement, but had died in rapid succession. This new\napplicant was the first to apprize him of this circumstance, and\nappeared extremely anxious to enter on immediate possession.\n\nIt was intimated to him that danger would arise from the pestilential\ncondition of the house. Unless cleansed and purified, disease would be\nunavoidably contracted. The inconvenience and hazard this applicant was\nwilling to encounter, and, at length, hinted that no alternative was\nallowed him by his present landlord but to lie in the street or to\nprocure some other abode.\n\n\"What was the external appearance of this person?\"\n\n\"He was infirm, past the middle age, of melancholy aspect and indigent\ngarb. A year had since elapsed, and more characteristic particulars had\nnot been remarked, or were forgotten. The name had been mentioned, but,\nin the midst of more recent and momentous transactions, had vanished\nfrom remembrance. Dudley, or Dolby, or Hadley, seemed to approach more\nnearly than any other sounds.\"\n\nPermission to inspect the house was readily granted. It had remained,\nsince that period, unoccupied. The furniture and goods were scanty and\nwretched, and he did not care to endanger his safety by meddling with\nthem. He believed that they had not been removed or touched.\n\nI was insensible of any hazard which attended my visit, and, with the\nguidance of a servant, who felt as little apprehension as myself,\nhastened to the spot. I found nothing but tables and chairs. Clothing\nwas nowhere to be seen. An earthen pot, without handle, and broken,\nstood upon the kitchen-hearth. No other implement or vessel for the\npreparation of food appeared.\n\nThese forlorn appearances were accounted for by the servant, by\nsupposing the house to have been long since rifled of every thing worth\nthe trouble of removal, by the villains who occupied the neighbouring\nhouses,--this alley, it seems, being noted for the profligacy of its\ninhabitants.\n\nWhen I reflected that a wretched hovel like this had been, probably, the\nlast retreat of the Dudleys, when I painted their sufferings, of which\nthe numberless tales of distress of which I had lately been an auditor\nenabled me to form an adequate conception, I felt as if to lie down and\nexpire on the very spot where Constantia had fallen was the only\nsacrifice to friendship which time had left to me.\n\nFrom this house I wandered to the field where the dead had been,\npromiscuously and by hundreds, interred. I counted the long series of\ngraves, which were closely ranged, and, being recently levelled,\nexhibited the appearance of a harrowed field. Methought I could have\ngiven thousands to know in what spot the body of my friend lay, that I\nmight moisten the sacred earth with my tears. Boards hastily nailed\ntogether formed the best receptacle which the exigencies of the time\ncould grant to the dead. Many corpses were thrown into a single\nexcavation, and all distinctions founded on merit and rank were\nobliterated. The father and child had been placed in the same cart and\nthrown into the same hole.\n\nDespairing, by any longer stay in the city, to effect my purpose, and\nthe period of my embarkation being near, I prepared to resume my\njourney. I should have set out the next day, but, a family with whom I\nhad made acquaintance expecting to proceed to New York within a week, I\nconsented to be their companion, and, for that end, to delay my\ndeparture.\n\nMeanwhile, I shut myself up in my apartment, and pursued avocations that\nwere adapted to the melancholy tenor of my thoughts. The day preceding\nthat appointed for my journey arrived. It was necessary to complete my\narrangements with the family with whom I was to travel, and to settle\nwith the lady whose apartments I occupied.\n\nOn how slender threads does our destiny hang! Had not a momentary\nimpulse tempted me to sing my favourite ditty to the harpsichord, to\nbeguile the short interval during which my hostess was conversing with\nher visitor in the next apartment, I should have speeded to New York,\nhave embarked for Europe, and been eternally severed from my friend,\nwhom I believed to have died in frenzy and beggary, but who was alive\nand affluent, and who sought me with a diligence scarcely inferior to my\nown. We imagined ourselves severed from each other by death or by\nimpassable seas; but, at the moment when our hopes had sunk to the\nlowest ebb, a mysterious destiny conducted our footsteps to the same\nspot.\n\nI heard a murmuring exclamation; I heard my hostess call, in a voice of\nterror, for help; I rushed into the room; I saw one stretched on the\nfloor, in the attitude of death; I sprung forward and fixed my eyes upon\nher countenance; I clasped my hands and articulated, \"Constantia!\"\n\nShe speedily recovered from her swoon. Her eyes opened; she moved, she\nspoke. Still methought it was an illusion of the senses that created the\nphantom. I could not bear to withdraw my eyes from her countenance. If\nthey wandered for a moment, I fell into doubt and perplexity, and again\nfixed them upon her, to assure myself of her existence.\n\nThe succeeding three days were spent in a state of dizziness and\nintoxication. The ordinary functions of nature were disturbed. The\nappetite for sleep and for food were confounded and lost amidst the\nimpetuosities of a master-passion. To look and to talk to each other\nafforded enchanting occupation for every moment. I would not part from\nher side, but eat and slept, walked and mused and read, with my arm\nlocked in hers, and with her breath fanning my cheek.\n\nI have indeed much to learn. Sophia Courtland has never been wise. Her\naffections disdain the cold dictates of discretion, and spurn at every\nlimit that contending duties and mixed obligations prescribe.\n\nAnd yet, O precious inebriation of the heart! O pre-eminent love! what\npleasure of reason or of sense can stand in competition with those\nattendant upon thee? Whether thou hiest to the fanes of a benevolent\ndeity, or layest all thy homage at the feet of one who most visibly\nresembles the perfections of our Maker, surely thy sanction is divine,\nthy boon is happiness!\n\n\n\n\nCHAPTER VII.\n\n\nThe tumults of curiosity and pleasure did not speedily subside. The\nstory of each other's wanderings was told with endless amplification and\nminuteness. Henceforth, the stream of our existence was to mix; we were\nto act and to think in common; casual witnesses and written testimony\nshould become superfluous. Eyes and ears were to be eternally employed\nupon the conduct of each other; death, when it should come, was not to\nbe deplored, because it was an unavoidable and brief privation to her\nthat should survive. Being, under any modification, is dear; but that\nstate to which death is a passage is all-desirable to virtue and\nall-compensating to grief.\n\nMeanwhile, precedent events were made the themes of endless\nconversation. Every incident and passion in the course of four years was\nrevived and exhibited. The name of Ormond was, of course, frequently\nrepeated by my friend. His features and deportment were described; her\nmeditations and resolutions, with regard to him, fully disclosed. My\ncounsel was asked, in what manner it became her to act.\n\nI could not but harbour aversion to a scheme which should tend to sever\nme from Constantia, or to give me a competitor in her affections.\nBesides this, the properties of Ormond were of too mysterious a nature\nto make him worthy of acceptance. Little more was known concerning him\nthan what he himself had disclosed to the Dudleys, but this knowledge\nwould suffice to invalidate his claims.\n\nHe had dwelt, in his conversations with Constantia, sparingly on his own\nconcerns. Yet he did not hide from her that he had been left in early\nyouth to his own guidance; that he had embraced, when almost a child,\nthe trade of arms; that he had found service and promotion in the armies\nof Potemkin and Romanzow; that he had executed secret and diplomatic\nfunctions at Constantinople and Berlin; that in the latter city he had\nmet with schemers and reasoners who aimed at the new-modelling of the\nworld, and the subversion of all that has hitherto been conceived\nelementary and fundamental in the constitution of man and of government;\nthat some of those reformers had secretly united to break down the\nmilitary and monarchical fabric of German policy; that others, more\nwisely, had devoted their secret efforts, not to overturn, but to build;\nthat, for this end, they embraced an exploring and colonizing project;\nthat he had allied himself to these, and for the promotion of their\nprojects had spent six years of his life in journeys by sea and land, in\ntracts unfrequented till then by any European.\n\nWhat were the moral or political maxims which this adventurous and\nvisionary sect had adopted, and what was the seat of their new-born\nempire,--whether on the shore of an _austral_ continent, or in the heart\nof desert America,--he carefully concealed. These were exhibited or\nhidden, or shifted, according to his purpose. Not to reveal too much,\nand not to tire curiosity or overtask belief, was his daily labour. He\ntalked of alliance with the family whose name he bore, and who had lost\ntheir honours and estates by the Hanoverian succession to the crown of\nEngland.\n\nI had seen too much of innovation and imposture, in, France and Italy,\nnot to regard a man like this with aversion and fear. The mind of my\nfriend was wavering and unsuspicious. She had lived at a distance from\nscenes where principles are hourly put to the test of experiment; where\nall extremes of fortitude and pusillanimity are accustomed to meet;\nwhere recluse virtue and speculative heroism gives place, as if by\nmagic, to the last excesses of debauchery and wickedness; where pillage\nand murder are engrafted on systems of all-embracing and self-oblivious\nbenevolence, and the good of mankind is professed to be pursued with\nbonds of association and covenants of secrecy. Hence, my friend had\ndecided without the sanction of experience, had allowed herself to\nwander into untried paths, and had hearkened to positions pregnant with\ndestruction and ignominy.\n\nIt was not difficult to exhibit in their true light the enormous errors\nof this man, and the danger of prolonging their intercourse. Her assent\nto accompany me to England was readily obtained. Too much despatch could\nnot be used; but the disposal of her property must first take place.\nThis was necessarily productive of some delay.\n\nI had been made, contrary to inclination, expert in the management of\nall affairs relative to property. My mother's lunacy, subsequent\ndisease, and death, had imposed upon me obligations and cares little\nsuitable to my sex and age. They could not be eluded or transferred to\nothers; and, by degrees, experience enlarged my knowledge and\nfamiliarized my tasks.\n\nIt was agreed that I should visit and inspect my friend's estate in\nJersey, while she remained in her present abode, to put an end to the\nviews and expectations of Ormond, and to make preparation for her\nvoyage. We were reconciled to a temporary separation by the necessity\nthat prescribed it.\n\nDuring our residence together, the mind of Constantia was kept in\nperpetual ferment. The second day after my departure, the turbulence of\nher feelings began to subside, and she found herself at leisure to\npursue those measures which her present situation prescribed.\n\nThe time prefixed by Ormond for the termination of his absence had\nnearly arrived. Her resolutions respecting this man, lately formed, now\noccurred to her. Her heart drooped as she revolved the necessity of\ndisuniting their fates; but that this disunion was proper could not\nadmit of doubt. How information of her present views might be most\nsatisfactorily imparted to him, was a question not instantly decided.\nShe reflected on the impetuosity of his character, and conceived that\nher intentions might be most conveniently unfolded in a letter. This\nletter she immediately sat down to write. Just then the door opened, and\nOrmond entered the apartment.\n\nShe was somewhat, and for a moment, startled by this abrupt and\nunlooked-for entrance. Yet she greeted him with pleasure. Her greeting\nwas received with coldness. A second glance at his countenance informed\nher that his mind was somewhat discomposed.\n\nFolding his hands on his breast, ho stalked to the window and looked up\nat the moon. Presently he withdrew his gaze from this object, and fixed\nit upon Constantia. He spoke, but his words were produced by a kind of\neffort.\n\n\"Fit emblem,\" he exclaimed, \"of human versatility! One impediment is\ngone. I hoped it was the only one. But no! the removal of that merely\nmade room for another. Let this be removed. Well, fate will interplace a\nthird. All our toils will thus be frustrated, and the ruin will finally\nredound upon our heads.\" There he stopped.\n\nThis strain could not be interpreted by Constantia. She smiled, and,\nwithout noticing his incoherences, proceeded to inquire into his\nadventures during their separation. He listened to her, but his eyes,\nfixed upon hers, and his solemnity of aspect, were immovable. When she\npaused, he seated himself close to her, and, grasping her hand with a\nvehemence that almost pained her, said,--\n\n\"Look at me; steadfastly. Can you read my thoughts? Can your discernment\nreach the bounds of my knowledge and the bottom of my purposes? Catch\nyou not a view of the monsters that are starting into birth _here_?\"\n(and he put his left hand to his forehead.) \"But you cannot. Should I\npaint them to you verbally, you would call me jester or deceiver. What\npity that you have not instruments for piercing into thoughts!\"\n\n\"I presume,\" said Constantia, affecting cheerfulness which she did not\nfeel, \"such instruments would be useless to me. You never scruple to say\nwhat you think. Your designs are no sooner conceived than they are\nexpressed. All you know, all you wish, and all you purpose, are known\nto others as soon as to yourself. No scruples of decorum, no foresight\nof consequences, are obstacles in your way.\"\n\n\"True,\" replied he; \"all obstacles are trampled under foot but one.\"\n\n\"What is the insuperable one?\"\n\n\"Incredulity in him that hears. I must not say what will not be\ncredited. I must not relate feats and avow schemes, when my hearer will\nsay, 'Those feats were never performed; these schemes are not yours.' I\ncare not if the truth of my tenets and the practicability of my purposes\nbe denied. Still, I will openly maintain them; but when my assertions\nwill themselves be disbelieved, when it is denied that I adopt the creed\nand project the plans which I affirm to be adopted and projected by me,\nit is needless to affirm.\n\n\"To-morrow I mean to ascertain the height of the lunar mountains by\ntravelling to the top of them. Then I will station myself in the track\nof the last comet, and wait till its circumvolution suffers me to leap\nupon it; then, by walking on its surface, I will ascertain whether it be\nhot enough to burn my soles. Do you believe that this can be done?\"\n\n\"No.\"\n\n\"Do you believe, in consequence of my assertion, that I design to do\nthis, and that, in my apprehension, it is easy to be done?\"\n\n\"Not unless I previously believe you to be lunatic.\"\n\n\"Then why should I assert my purposes? Why speak, when the hearer will\ninfer nothing from my speech but that I am either lunatic or liar?\"\n\n\"In that predicament, silence is best.\"\n\n\"In that predicament I now stand. I am not going to unfold myself. Just\nnow, I pitied thee for want of eyes. 'Twas a foolish compassion. Thou\nart happy, because thou seest not an inch before thee or behind.\" Here\nhe was for a moment buried in thought; then, breaking from his reverie,\nhe said, \"So your father is dead?\"\n\n\"True,\" said Constantia, endeavouring to suppress her rising emotions;\n\"he is no more. It is so recent an event that I imagined you a stranger\nto it.\"\n\n\"False imagination! Thinkest thou I would refrain from knowing what so\nnearly concerns us both? Perhaps your opinion of my ignorance extends\nbeyond this. Perhaps I know not your fruitless search for a picture.\nPerhaps I neither followed you nor led you to a being called Sophia\nCourtland. I was not present at the meeting. I am unapprized of the\neffects of your romantic passion for each other. I did not witness the\nrapturous effusions and inexorable counsels of the newcomer. I know not\nthe contents of the letter which you are preparing to write.\"\n\nAs he spoke this, the accents of Ormond gradually augmented in\nvehemence. His countenance bespoke a deepening inquietude and growing\npassion. He stopped at the mention of the letter, because his voice was\noverpowered by emotion. This pause afforded room for the astonishment of\nConstantia. Her interviews and conversations with me took place at\nseasons of general repose, when all doors were fast and avenues shut, in\nthe midst of silence, and in the bosom of retirement. The theme of our\ndiscourse was, commonly, too sacred for any ears but our own;\ndisclosures were of too intimate and delicate a nature for any but a\nfemale audience; they were too injurious to the fame and peace of Ormond\nfor him to be admitted to partake of them: yet his words implied a full\nacquaintance with recent events, and with purposes and deliberations\nshrouded, as we imagined, in impenetrable secrecy.\n\nAs soon as Constantia recovered from the confusion of these thoughts,\nshe eagerly questioned him:--\"What do you know? How do you know what has\nhappened, or what is intended?\"\n\n\"Poor Constantia!\" he exclaimed, in a tone bitter and sarcastic. \"How\nhopeless is thy ignorance! To enlighten thee is past my power. What\ndo I know? Every thing. Not a tittle has escaped me. Thy letter is\nsuperfluous; I know its contents before they are written. I was\nto be told that a soldier and a traveller, a man who refused his\nfaith to dreams, and his homage to shadows, merited only scorn and\nforgetfulness. That thy affections and person were due to another; that\nintercourse between us was henceforth to cease; that preparation was\nmaking for a voyage to Britain, and that Ormond was to walk to his grave\nalone!\"\n\nIn spite of harsh tones and inflexible features, these words were\naccompanied with somewhat that betrayed a mind full of discord and\nagony. Constantia's astonishment was mingled with dejection. The\ndiscovery of a passion deeper and less curable than she suspected--the\nperception of embarrassments and difficulties in the path which she had\nchosen, that had not previously occurred to her--threw her mind into\nanxious suspense.\n\nThe measures she had previously concerted were still approved. To part\nfrom Ormond was enjoined by every dictate of discretion and duty. An\nexplanation of her motives and views could not take place more\nseasonably than at present. Every consideration of justice to herself\nand humanity to Ormond made it desirable that this interview should be\nthe last. By inexplicable means, he had gained a knowledge of her\nintentions. It was expedient, therefore, to state them with clearness\nand force. In what words this was to be done, was the subject of\nmomentary deliberation.\n\nHer thoughts were discerned, and her speech anticipated, by her\ncompanion:--\"Why droopest thou, and why thus silent, Constantia? The\nsecret of thy fate will never be detected. Till thy destiny be finished,\nit will not be the topic of a single fear. But not for thyself, but me,\nart thou concerned. Thou dreadest, yet determinest, to confirm my\npredictions of thy voyage to Europe and thy severance from me.\n\n\"Dismiss thy inquietudes on that score. What misery thy scorn and thy\nrejection are able to inflict is inflicted already. Thy decision was\nknown to me as soon as it was formed. Thy motives were known. Not an\nargument or plea of thy counsellor, not a syllable of her invective, not\na sound of her persuasive rhetoric, escaped my hearing. I know thy\ndecree to be immutable. As my doubts, so my wishes have taken their\nflight. Perhaps, in the depth of thy ignorance, it was supposed that I\nshould struggle to reverse thy purpose by menaces or supplications; that\nI should boast of the cruelty with which I should avenge an imaginary\nwrong upon myself. No. All is very well. Go. Not a whisper of objection\nor reluctance shalt thou hear from me.\"\n\n\"If I could think,\" said Constantia, with tremulous hesitation, \"that\nyou part from me without anger; that you see the rectitude of my\nproceeding--\"\n\n\"Anger! Rectitude! I pr'ythee, peace. I know thou art going.--I know\nthat all objection to thy purpose would be vain. Thinkest thou that thy\nstay, undictated by love, the mere fruit of compassion, would afford me\npleasure or crown my wishes? No. I am not so dastardly a wretch. There\nwas something in thy power to bestow, but thy will accords not with thy\npower. I merit not the boon, and thou refusest it. I am content.\"\n\nHere Ormond fixed more significant eyes upon her. \"Poor Constantia!\" he\ncontinued. \"Shall I warn thee of the danger that awaits thee? For what\nend? To elude it is impossible. It will come, and thou, perhaps, wilt be\nunhappy. Foresight that enables not to shun, only precreates, the evil.\n\n\"Come it will. Though future, it knows not the empire of contingency. An\ninexorable and immutable decree enjoins it. Perhaps it is thy nature to\nmeet with calmness what cannot be shunned. Perhaps, when it is past, thy\nreason will perceive its irrevocable nature, and restore thee to peace.\nSuch is the conduct of the wise; but such, I fear, the education of\nConstantia Dudley will debar her from pursuing.\n\n\"Fain would I regard it as the test of thy wisdom. I look upon thy past\nlife. All the forms of genuine adversity have beset thy youth. Poverty,\ndisease, servile labour, a criminal and hapless parent, have been evils\nwhich thou hast not ungracefully sustained. An absent friend and\nmurdered father were added to thy list of woes, and here thy courage was\ndeficient. Thy soul was proof against substantial misery, but sunk into\nhelpless cowardice at the sight of phantoms.\n\n\"One more disaster remains. To call it by its true name would be useless\nor pernicious. Useless, because thou wouldst pronounce its occurrence\nimpossible; pernicious, because, if its possibility were granted, the\nomen would distract thee with fear. How shall I describe it? Is it loss\nof fame? No. The deed will be unwitnessed by a human creature. Thy\nreputation will be spotless, for nothing will be done by thee unsuitable\nto the tenor of thy past life. Calumny will not be heard to whisper. All\nthat know thee will be lavish of their eulogies as ever. Their eulogies\nwill be as justly merited. Of this merit thou wilt entertain as just and\nas adequate conceptions as now.\n\n\"It is no repetition of the evils thou hast already endured; it is\nneither drudgery, nor sickness, nor privation of friends. Strange\nperverseness of human reason! It is an evil; it will be thought upon\nwith agony; it will close up all the sources of pleasurable\nrecollection; it will exterminate hope; it will endear oblivion, and\npush thee into an untimely grave. Yet to grasp it is impossible. The\nmoment we inspect it nearly, it vanishes. Thy claims to human\napprobation and divine applause will be undiminished and unaltered by\nit. The testimony of approving conscience will have lost none of its\nexplicitness and energy. Yet thou wilt feed upon sighs; thy tears will\nflow without remission; thou wilt grow enamoured of death, and perhaps\nwilt anticipate the stroke of disease.\n\n\"Yet perhaps my prediction is groundless as my knowledge. Perhaps thy\ndiscernment will avail to make thee wise and happy. Perhaps thou wilt\nperceive thy privilege of sympathetic and intellectual activity to be\nuntouched. Heaven grant the non-fulfilment of my prophecy, thy\ndisenthralment from error, and the perpetuation of thy happiness.\"\n\nSaying this, Ormond withdrew. His words were always accompanied with\ngestures and looks and tones that fastened the attention of the hearer;\nbut the terms of his present discourse afforded, independently of\ngesticulation and utterance, sufficient motives to attention and\nremembrance. He was gone, but his image was contemplated by Constantia;\nhis words still rung in her ears.\n\nThe letter she designed to compose was rendered, by this interview,\nunnecessary. Meanings of which she and her friend alone were conscious\nwere discovered by Ormond, through some other medium than words; yet\nthat was impossible. A being unendowed with preternatural attributes\ncould gain the information which this man possessed, only by the\nexertion of his senses.\n\nAll human precautions had been used to baffle the attempts of any secret\nwitness. She recalled to mind the circumstances in which conversations\nwith her friend had taken place. All had been retirement, secrecy, and\nsilence. The hours usually dedicated to sleep had been devoted to this\nbetter purpose. Much had been said, in a voice low and scarcely louder\nthan a whisper. To have overheard it at the distance of a few feet was\napparently impossible.\n\nTheir conversations had not been recorded by her. It could not be\nbelieved that this had been done by Sophia Courtland. Had Ormond and her\nfriend met during the interval that had elapsed between her separation\nfrom the latter and her meeting with the former? Human events are\nconjoined by links imperceptible to keenest eyes. Of Ormond's means of\ninformation she was wholly unapprized. Perhaps accident would some time\nunfold them. One thing was incontestable:--that her schemes and her\nreasons for adopting them were known to him.\n\nWhat unforeseen effects had that knowledge produced! In what ambiguous\nterms had he couched his prognostics of some mighty evil that awaited\nher! He had given a terrible but contradictory description of her\ndestiny. An event was to happen, akin to no calamity which she had\nalready endured, disconnected with all which the imagination of man is\naccustomed to deprecate, capable of urging her to suicide, and yet of a\nkind which left it undecided whether she would regard it with\nindifference.\n\nWhat reliance should she place upon prophetic incoherences thus wild?\nWhat precautions should she take against a danger thus inscrutable and\nimminent?\n\n\n\n\nCHAPTER VIII.\n\n\nThese incidents and reflections were speedily transmitted to me. I had\nalways believed the character and machinations of Ormond to be worthy of\ncaution and fear. His means of information I did not pretend, and\nthought it useless, to investigate. We cannot hide our actions and\nthoughts from one of powerful sagacity, whom the detection sufficiently\ninterests to make him use all the methods of detection in his power. The\nstudy of concealment is, in all cases, fruitless or hurtful. All that\nduty enjoins is to design and to execute nothing which may not be\napproved by a divine and omniscient Observer. Human scrutiny is neither\nto be solicited nor shunned. Human approbation or censure can never be\nexempt from injustice, because our limited perceptions debar us from a\nthorough knowledge of any actions and motives but our own.\n\nOn reviewing what had passed between Constantia and me, I recollected\nnothing incompatible with purity and rectitude. That Ormond was apprized\nof all that had passed, I by no means inferred from the tenor of his\nconversation with Constantia; nor, if this had been incontestably\nproved, should I have experienced any trepidation or anxiety on that\naccount.\n\nHis obscure and indirect menaces of evil were of more importance. His\ndiscourse on this topic seemed susceptible only of two constructions.\nEither he intended some fatal mischief, and was willing to torment her\nby fears, while he concealed from her the nature of her danger, that he\nmight hinder her from guarding her safety by suitable precautions; or,\nbeing hopeless of rendering her propitious to his wishes, his malice was\nsatisfied with leaving her a legacy of apprehension and doubt.\nConstantia's unacquaintance with the doctrines of that school in which\nOrmond was probably instructed led her to regard the conduct of this man\nwith more curiosity and wonder than fear. She saw nothing but a\ndisposition to sport with her ignorance and bewilder her with doubts.\n\nI do not believe myself destitute of courage. Rightly to estimate the\ndanger and encounter it with firmness are worthy of a rational being;\nbut to place our security in thoughtlessness and blindness is only less\nignoble than cowardice. I could not forget the proofs of violence which\naccompanied the death of Mr. Dudley. I could not overlook, in the recent\nconversation with Constantia, Ormond's allusion to her murdered father.\nIt was possible that the nature of this death had been accidentally\nimparted to him; but it was likewise possible that his was the knowledge\nof one who performed the act.\n\nThe enormity of this deed appeared by no means incongruous with the\nsentiments of Ormond. Human life is momentous or trivial in our eyes,\naccording to the course which our habits and opinions have taken.\nPassion greedily accepts, and habit readily offers, the sacrifice of\nanother's life, and reason obeys the impulse of education and desire.\n\nA youth of eighteen, a volunteer in a Russian army encamped in\nBessarabia, made prey of a Tartar girl, found in the field of a recent\nbattle. Conducting her to his quarters, he met a friend, who, on some\npretence, claimed the victim. From angry words they betook themselves to\nswords. A combat ensued, in which the first claimant ran his antagonist\nthrough the body. He then bore his prize unmolested away, and, having\nexercised brutality of one kind upon the helpless victim, stabbed her to\nthe heart, as an offering to the _manes_ of Sarsefield, the friend whom\nhe had slain. Next morning, willing more signally to expiate his guilt,\nhe rushed alone upon a troop of Turkish foragers, and brought away five\nheads, suspended, by their gory locks, to his horse's mane. These he\ncast upon the grave of Sarsefield, and conceived himself fully to have\nexpiated yesterday's offence. In reward for his prowess, the general\ngave him a commission in the Cossack troops. This youth was Ormond; and\nsuch is a specimen of his exploits during a military career of eight\nyears, in a warfare the most savage and implacable, and, at the same\ntime, the most iniquitous and wanton, which history records.\n\nWith passions and habits like these, the life of another was a trifling\nsacrifice to vengeance or impatience. How Mr. Dudley had excited the\nresentment of Ormond, by what means the assassin had accomplished his\nintention without awakening alarm or incurring suspicion, it was not for\nme to discover. The inextricability of human events, the imperviousness\nof cunning, and the obduracy of malice, I had frequent occasions to\nremark.\n\nI did not labour to vanquish the security of my friend. As to\nprecautions, they were useless. There was no fortress, guarded by\nbarriers of stone and iron and watched by sentinels that never slept, to\nwhich she might retire from his stratagems. If there were such a\nretreat, it would scarcely avail her against a foe circumspect and\nsubtle as Ormond.\n\nI pondered on the condition of my friend. I reviewed the incidents of\nher life. I compared her lot with that of others. I could not but\ndiscover a sort of incurable malignity in her fate. I felt as if it were\ndenied to her to enjoy a long life or permanent tranquillity. I asked\nmyself what she had done, entitling her to this incessant persecution.\nImpatience and murmuring took place of sorrow and fear in my heart. When\nI reflected that all human agency was merely subservient to a divine\npurpose, I fell into fits of accusation and impiety.\n\nThis injustice was transient, and soberer views convinced me that every\nscheme, comprising the whole, must be productive of partial and\ntemporary evil. The sufferings of Constantia were limited to a moment;\nthey were the unavoidable appendages of terrestrial existence; they\nformed the only avenue to wisdom, and the only claim to uninterrupted\nfruition and eternal repose in an after-scene.\n\nThe course of my reflections, and the issue to which they led, were\nunforeseen by myself. Fondly as I doted upon this woman, methought I\ncould resign her to the grave without a murmur or a tear. While my\nthoughts were calmed by resignation, and my fancy occupied with nothing\nbut the briefness of that space and evanescence of that time which\nsevers the living from the dead, I contemplated, almost with\ncomplacency, a violent or untimely close to her existence.\n\nThis loftiness of mind could not always be accomplished or constantly\nmaintained. One effect of my fears was to hasten my departure to Europe.\nThere existed no impediment but the want of a suitable conveyance. In\nthe first packet that should leave America, it was determined to secure\na passage. Mr. Melbourne consented to take charge of Constantia's\nproperty, and, after the sale of it, to transmit to her the money that\nshould thence arise.\n\nMeanwhile, I was anxious that Constantia should leave her present abode\nand join me in New York. She willingly adopted this arrangement, but\nconceived it necessary to spend a few days at her house in Jersey. She\ncould reach the latter place without much deviation from the straight\nroad, and she was desirous of resurveying a spot where many of her\ninfantile days had been spent.\n\nThis house and domain I have already mentioned to have once belonged to\nMr. Dudley. It was selected with the judgement and adorned with the taste\nof a disciple of the schools of Florence and Vicenza. In his view,\ncultivation was subservient to the picturesque, and a mansion was\nerected, eminent for nothing but chastity of ornaments and simplicity of\nstructure. The massive parts were of stone; the outer surfaces were\nsmooth, snow-white, and diversified by apertures and cornices, in which\na cement uncommonly tenacious was wrought into proportions the most\ncorrect and forms the most graceful. The floors, walls, and ceilings,\nconsisted of a still more exquisitely-tempered substance, and were\npainted by Mr. Dudley's own hand. All appendages of this building, as\nseats, tables, and cabinets, were modelled by the owner's particular\ndirection, and in a manner scrupulously classical.\n\nHe had scarcely entered on the enjoyment of this splendid possession,\nwhen it was ravished away. No privation was endured with more impatience\nthan this; but, happily, it was purchased by one who left Mr. Dudley's\narrangements unmolested, and who shortly after conveyed it entire to\nOrmond. By him it was finally appropriated to the use of Helena Cleves,\nand now, by a singular contexture of events, it had reverted to those\nhands in which the death of the original proprietor, if no other change\nhad been made in his condition, would have left it. The farm still\nremained in the tenure of a German emigrant, who held it partly on\ncondition of preserving the garden and mansion in safety and in perfect\norder.\n\nThis retreat was now revisited by Constantia, after an interval of four\nyears. Autumn had made some progress, but the aspect of nature was, so\nto speak, more significant than at any other season. She was agreeably\naccommodated under the tenant's roof, and found a nameless pleasure in\ntraversing spaces in which every object prompted an endless train of\nrecollections.\n\nHer sensations were not foreseen. They led to a state of mind\ninconsistent, in some degree, with the projects adopted in obedience to\nthe suggestions of a friend. Every thing in this scene had been created\nand modelled by the genius of her father. It was a kind of fane,\nsanctified by his imaginary presence.\n\nTo consign the fruits of his industry and invention to foreign and\nunsparing hands seemed a kind of sacrilege, for which she almost feared\nthat the dead would rise to upbraid her. Those images which bind us to\nour natal soil, to the abode of our innocent and careless youth, were\nrecalled to her fancy by the scenes which she now beheld. These were\nenforced by considerations of the dangers which attended her voyage from\nstorms and from enemies, and from the tendency to revolution and war\nwhich seemed to actuate all the nations of Europe. Her native country\nwas by no means exempt from similar tendencies, but these evils were\nless imminent, and its manners and government, in their present\nmodifications, were unspeakably more favourable to the dignity and\nimprovement of the human race than those which prevailed in any part of\nthe ancient world.\n\nMy solicitations and my obligation to repair to England overweighed her\nobjections, but her new reflections led her to form new determinations\nwith regard to this part of her property. She concluded to retain\npossession, and hoped that some future event would allow her to return\nto this favourite spot without forfeiture of my society. An abode of\nsome years in Europe would more eminently qualify her for the enjoyment\nof retirement and safety in her native country. The time that should\nelapse before her embarkation, she was desirous of passing among the\nshades of this romantic retreat.\n\nI was by no means reconciled to this proceeding. I loved my friend too\nwell to endure any needless separation without repining. In addition to\nthis, the image of Ormond haunted my thoughts, and gave birth to\nincessant but indefinable fears. I believed that her safety would very\nlittle depend upon the nature of her abode, or the number or\nwatchfulness of her companions. My nearness to her person would\nfrustrate no stratagem, nor promote any other end than my own\nentanglement in the same fold. Still, that I was not apprized each hour\nof her condition, that her state was lonely and sequestered, were\nsources of disquiet, the obvious remedy to which was her coming to New\nYork. Preparations for departure were assigned to me, and these required\nmy continuance in the city.\n\nOnce a week, Laffert, her tenant, visited, for purposes of traffic, the\ncity. He was the medium of our correspondence. To him I intrusted a\nletter, in which my dissatisfaction at her absence, and the causes which\ngave it birth, were freely confessed.\n\nThe confidence of safety seldom deserted my friend. Since her mysterious\nconversation with Ormond, he had utterly vanished. Previous to that\ninterview, his visits or his letters were incessant and punctual; but\nsince, no token was given that he existed. Two months had elapsed. He\ngave her no reason to expect a cessation of intercourse. He had parted\nfrom her with his usual abruptness and informality. She did not conceive\nit incumbent on her to search him out, but she would not have been\ndispleased with an opportunity to discuss with him more fully the\nmotives of her conduct. This opportunity had been hitherto denied.\n\nHer occupations in her present retreat were, for the most part, dictated\nby caprice or by chance. The mildness of autumn permitted her to ramble,\nduring the day, from one rock and one grove to another. There was a\nluxury in musing, and in the sensations which the scenery and silence\nproduced, which, in consequence of her long estrangement from them, were\naccompanied with all the attractions of novelty, and from which she\nwould not consent to withdraw.\n\nIn the evening she usually retired to the mansion, and shut herself up\nin that apartment which, in the original structure of the house, had\nbeen designed for study, and no part of whose furniture had been removed\nor displaced. It was a kind of closet on the second floor, illuminated\nby a spacious window, through which a landscape of uncommon amplitude\nand beauty was presented to the view. Here the pleasures of the day were\nrevived, by recalling and enumerating them in letters to her friend. She\nalways quitted this recess with reluctance, and seldom till the night\nwas half spent.\n\nOne evening she retired hither when the sun had just dipped beneath the\nhorizon. Her implements of writing were prepared; but, before the pen\nwas assumed, her eyes rested for a moment on the variegated hues which\nwere poured out upon the western sky and upon the scene of intermingled\nwaters, copses, and fields. The view comprised a part of the road which\nled to this dwelling. It was partially and distantly seen, and the\npassage of horses or men was betokened chiefly by the dust which was\nraised by their footsteps.\n\nA token of this kind now caught her attention. It fixed her eye chiefly\nby the picturesque effect produced by interposing its obscurity between\nher and the splendours which the sun had left. Presently she gained a\nfaint view of a man and horse. This circumstance laid no claim to\nattention, and she was withdrawing her eye, when the traveller's\nstopping and dismounting at the gate made her renew her scrutiny. This\nwas reinforced by something in the figure and movements of the horseman\nwhich reminded her of Ormond.\n\nShe started from her seat with some degree of palpitation. Whence this\narose, whether from fear or from joy, or from intermixed emotions, it\nwould not be easy to ascertain. Having entered the gate, the visitant,\nremounting his horse, set the animal on full speed. Every moment brought\nhim nearer, and added to her first belief. He stopped not till he\nreached the mansion. The person of Ormond was distinctly recognised.\n\nAn interview at this dusky and lonely hour, in circumstances so abrupt\nand unexpected, could not fail to surprise, and, in some degree, to\nalarm. The substance of his last conversation was recalled. The evils\nwhich were darkly and ambiguously predicted thronged to her memory. It\nseemed as if the present moment was to be, in some way, decisive of her\nfate. This visit she did not hesitate to suppose designed for her, but\nsomewhat uncommonly momentous must have prompted him to take so long a\njourney.\n\nThe rooms on the lower floor were dark, the windows and doors being\nfastened. She had entered the house by the principal door, and this was\nthe only one at present unlocked. The room in which she sat was over the\nhall, and the massive door beneath could not be opened without noisy\nsignals. The question that occurred to her, by what means Ormond would\ngain admittance to her presence, she supposed would be instantly\ndecided. She listened to hear his footsteps on the pavement, or the\ncreaking of hinges. The silence, however, continued profound as before.\n\nAfter a minute's pause, she approached the window more nearly and\nendeavoured to gain a view of the space before the house. She saw\nnothing but the horse, whose bridle was thrown over his neck, and who\nwas left at liberty to pick up what scanty herbage the lawn afforded to\nhis hunger. The rider had disappeared.\n\nIt now occurred to her that this visit had a purpose different from that\nwhich she at first conjectured. It was easily conceived that Ormond was\nunacquainted with her residence at this spot. The knowledge could only\nbe imparted to him by indirect or illicit means. That these means had\nbeen employed by him, she was by no means authorized to infer from the\nsilence and distance he had lately maintained. But if an interview with\nher were not the purpose of his coming, how should she interpret it?\n\n\n\n\nCHAPTER IX.\n\n\nWhile occupied with these reflections, the light hastily disappeared,\nand darkness, rendered, by a cloudy atmosphere, uncommonly intense,\nsucceeded. She had the means of lighting a lamp that hung against the\nwall, but had been too much immersed in thought to notice the deepening\nof the gloom. Recovering from her reverie, she looked around her with\nsome degree of trepidation, and prepared to strike a spark that would\nenable her to light her lamp.\n\nShe had hitherto indulged an habitual indifference to danger. Now the\npresence of Ormond, the unknown purpose that led him hither, and the\ndefencelessness of her condition, inspired her with apprehensions to\nwhich she had hitherto been a stranger. She had been accustomed to pass\nmany nocturnal hours in this closet. Till now, nothing had occurred that\nmade her enter it with circumspection or continue in it with reluctance.\n\nHer sensations were no longer tranquil. Each minute that she spent in\nthis recess appeared to multiply her hazards. To linger here appeared to\nher the height of culpable temerity. She hastily resolved to return to\nthe farmer's dwelling, and, on the morrow, to repair to New York. For\nthis end she was desirous to produce a light. The materials were at\nhand.\n\nShe lifted her hand to strike the flint, when her ear caught a sound\nwhich betokened the opening of the door that led into the next\napartment. Her motion was suspended, and she listened as well as a\nthrobbing heart would permit. That Ormond's was the hand that opened,\nwas the first suggestion of her fears. The motives of this unseasonable\nentrance could not be reconciled with her safety. He had given no\nwarning of his approach, and the door was opened with tardiness and\nseeming caution.\n\nSounds continued, of which no distinct conception could be obtained, or\nthe cause that produced them assigned. The floors of every apartment\nbeing composed, like the walls and ceiling, of cement, footsteps were\nrendered almost undistinguishable. It was plain, however, that some one\napproached her own door.\n\nThe panic and confusion that now invaded her was owing to surprise, and\nto the singularity of her situation. The mansion was desolate and\nlonely. It was night. She was immersed in darkness. She had not the\nmeans, and was unaccustomed to the office, of repelling personal\ninjuries. What injuries she had reason to dread, who was the agent, and\nwhat were his motives, were subjects Of vague and incoherent meditation.\n\nMeanwhile, low and imperfect sounds, that had in them more of inanimate\nthan human, assailed her ear. Presently they ceased. An inexplicable\nfear deterred her from calling. Light would have exercised a friendly\ninfluence. This it was in her power to produce, but not without motion\nand noise; and these, by occasioning the discovery of her being in the\ncloset, might possibly enhance her danger.\n\nConceptions like these were unworthy of the mind of Constantia. An\ninterval of silence succeeded, interrupted only by the whistling of the\nblast without. It was sufficient for the restoration of her courage. She\nblushed at the cowardice which had trembled at a sound. She considered\nthat Ormond might, indeed, be near, but that he was probably unconscious\nof her situation. His coming was not with the circumspection of an\nenemy. He might be acquainted with the place of her retreat, and had\ncome to obtain an interview, with no clandestine or mysterious purposes.\nThe noises she had heard had, doubtless, proceeded from the next\napartment, but might be produced by some harmless or vagrant creature.\n\nThese considerations restored her tranquillity. They enabled her,\ndeliberately, to create a light, but they did not dissuade her from\nleaving the house. Omens of evil seemed to be connected with this\nsolitary and darksome abode. Besides, Ormond had unquestionably entered\nupon this scene It could not be doubted that she was the object of his\nvisit. The farm-house was a place of meeting more suitable and safe than\nany other. Thither, therefore, she determined immediately to return.\n\nThe closet had but one door, and this led into the chamber where the\nsounds had arisen. Through this chamber, therefore, she was obliged to\npass, in order to reach the staircase, which terminated in the hall\nbelow.\n\nBearing the light in her left hand, she withdrew the bolt of the door\nand opened. In spite of courageous efforts, she opened with\nunwillingness, and shuddered to throw a glance forward or advance a step\ninto the room. This was not needed, to reveal to her the cause of her\nlate disturbance. Her eye instantly lighted on the body of a man,\nsupine, motionless, stretched on the floor, close to the door through\nwhich she was about to pass.\n\nA spectacle like this was qualified to startle her. She shrunk back, and\nfixed a more steadfast eye upon the prostrate person. There was no mark\nof blood or of wounds, but there was something in the attitude more\nsignificant of death than of sleep. His face rested on the floor, and\nhis ragged locks concealed what part of his visage was not hidden by his\nposture. His garb was characterized by fashionable elegance, but was\npolluted with dust.\n\nThe image that first occurred to her was that of Ormond. This instantly\ngave place to another, which was familiar to her apprehension. It was at\nfirst too indistinctly seen to suggest a name. She continued to gaze and\nto be lost in fearful astonishment. Was this the person whose entrance\nhad been overheard, and who had dragged himself hither to die at her\ndoor? Yet, in that case, would not groans and expiring efforts have\ntestified his condition and invoked her succour? Was he not brought\nhither in the arms of his assassin? She mused upon the possible motives\nthat induced some one thus to act, and upon the connection that might\nsubsist between her destiny and that of the dead.\n\nHer meditations, however fruitless in other respects, could not fail to\nshow her the propriety of hastening from this spot. To scrutinize the\nform or face of the dead was a task to which her courage was unequal.\nSuitably accompanied and guarded, she would not scruple to return and\nascertain, by the most sedulous examination, the cause of this ominous\nevent.\n\nShe stepped over the breathless corpse, and hurried to the staircase. It\nbecame her to maintain the command of her muscles and joints, and to\nproceed without faltering or hesitation. Scarcely had she reached the\nentrance of the hall, when, casting anxious looks forward, she beheld a\nhuman figure. No scrutiny was requisite to inform her that this was\nOrmond.\n\nShe stopped. He approached her with looks and gestures placid but\nsolemn. There was nothing in his countenance rugged or malignant. On the\ncontrary, there were tokens of compassion.\n\n\"So,\" said he, \"I expected to meet you. Alight, gleaming from the\nwindow, marked you out. This and Laffert's directions have guided me.\"\n\n\"What,\" said Constantia, with discomposure in her accent, \"was your\nmotive for seeking me?\"\n\n\"Have you forgotten,\" said Ormond, \"what passed at our last interview?\nThe evil that I then predicted is at hand. Perhaps you were incredulous;\nyou accounted me a madman or deceiver; now I am come to witness the\nfulfilment of my words and the completion of your destiny. To rescue you\nI have not come: that is not within the compass of human powers.\n\n\"Poor Constantia,\" he continued, in tones that manifested genuine\nsympathy, \"look upon thyself as lost. The toils that beset thee are\ninextricable. Summon up thy patience to endure the evil. Now will the\nlast and heaviest trial betide thy fortitude. I could weep for thee, if\nmy manly nature would permit. This is the scene of thy calamity, and\nthis the hour.\"\n\nThese words were adapted to excite curiosity mingled with terror.\nOrmond's deportment was of an unexampled tenor, as well as that evil\nwhich he had so ambiguously predicted. He offered no protection from\ndanger, and yet gave no proof of being himself an agent or auxiliary.\nAfter a minute's pause, Constantia, recovering a firm tone, said,--\n\n\"Mr. Ormond, your recent deportment but ill accords with your\nprofessions of sincerity and plain dealing. What your purpose is, or\nwhether you have any purpose, I am at a loss to conjecture. Whether you\nmost deserve censure or ridicule, is a point which you afford me not the\nmeans of deciding, and to which, unless on your own account, I am\nindifferent. If you are willing to be more explicit, or if there be any\ntopic on which you wish further to converse, I will not refuse your\ncompany to Laffert's dwelling. Longer to remain here would be indiscreet\nand absurd.\"\n\nSo saying, she motioned towards the door. Ormond was passive, and seemed\nindisposed to prevent her departure, till she laid her hand upon the\nlock. He then, without moving from his place, exclaimed,--\n\n\"Stay! Must this meeting, which fate ordains to be the last, be so\nshort? Must a time and place so suitable for what remains to be said and\ndone be neglected or misused? No. You charge me with duplicity, and deem\nmy conduct either ridiculous or criminal. I have stated my reasons for\nconcealment, but these have failed to convince you. Well, here is now an\nend to doubt. All ambiguities are preparing to vanish.\"\n\nWhen Ormond began to speak, Constantia paused to hearken to him. His\nvehemence was not of that nature which threatened to obstruct her\npassage. It was by entreaty that he apparently endeavoured to detain her\nsteps, and not by violence. Hence arose her patience to listen. He\ncontinued:--\n\n\"Constantia! thy father is dead. Art thou not desirous of detecting the\nauthor of his fate? Will it afford thee no consolation to know that the\ndeed is punished? Wilt thou suffer me to drag the murderer to thy feet?\nThy justice will be gratified by this sacrifice. Somewhat will be due to\nhim who avenged thy wrong in the blood of the perpetrator. What sayest\nthou? Grant me thy permission, and in a moment I will drag him hither.\"\n\nThese words called up the image of the person whose corpse she had\nlately seen. It was readily conceived that to him Ormond alluded; but\nthis was the assassin of her father, and his crime had been detected and\npunished by Ormond! These images had no other effect than to urge her\ndeparture: she again applied her hand to the lock, and said,--\n\n\"This scene must not be prolonged. My father's death I desire not to\nhear explained or to see revenged, but whatever information you are\nwilling or able to communicate must be deferred.\"\n\n\"Nay,\" interrupted Ormond, with augmented vehemence, \"art thou equally\ndevoid of curiosity and justice? Thinkest thou that the enmity which\nbereft thy father of life will not seek thy own? There are evils which I\ncannot prevent thee from enduring, but there are, likewise, ills which\nmy counsel will enable thee and thy friend to shun. Save me from\nwitnessing thy death. Thy father's destiny is sealed; all that remained\nwas to punish his assassin; but thou and thy Sophia still live. Why\nshould ye perish by a like stroke?\"\n\nThis intimation was sufficient to arrest the steps of Constantia. She\nwithdrew her hand from the door, and fixed eyes of the deepest anxiety\non Ormond:--\"What mean you? How am I to understand--\"\n\n\"Ah!\" said Ormond, \"I see thou wilt consent to stay. Thy detention shall\nnot be long. Remain where thou art during one moment,--merely while I\ndrag hither thy enemy and show thee a visage which thou wilt not be slow\nto recognise.\" Saying this, he hastily ascended the staircase, and\nquickly passed beyond her sight.\n\nDeportment thus mysterious could not fail of bewildering her thoughts.\nThere was somewhat in the looks and accents of Ormond, different from\nformer appearances; tokens of a hidden purpose and a smothered meaning\nwere perceptible,--a mixture of the inoffensive and the lawless, which,\nadded to the loneliness and silence that encompassed her, produced a\nfaltering emotion. Her curiosity was overpowered by her fear, and the\nresolution was suddenly conceived of seizing this opportunity to escape.\n\nA third time she put her hand to the lock and attempted to open. The\neffort was ineffectual. The door that was accustomed to obey the\ngentlest touch was now immovable. She had lately unlocked and passed\nthrough it. Her eager inspection convinced her that the principal bolt\nwas still withdrawn, but a small one was now perceived, of whose\nexistence she had not been apprized, and over which her key had no\npower.\n\nNow did she first harbour a fear that was intelligible in its dictates.\nNow did she first perceive herself sinking in the toils of some lurking\nenemy. Hope whispered that this foe was not Ormond. His conduct had\nbespoken no willingness to put constraint upon her steps. He talked not\nas if he was aware of this obstruction, and yet his seeming acquiescence\nmight have flowed from a knowledge that she had no power to remove\nbeyond his reach.\n\nHe warned her of danger to her life, of which he was her self-appointed\nrescuer. His counsel was to arm her with sufficient caution; the peril\nthat awaited her was imminent; this was the time and place of its\noccurrence, and here she was compelled to remain, till the power that\nfastened would condescend to loose the door. There were other avenues to\nthe hall. These were accustomed to be locked; but Ormond had found\naccess, and, if all continued fast, it was incontestable that he was the\nauthor of this new impediment.\n\nThe other avenues were hastily examined. All were bolted and locked. The\nfirst impulse led her to call for help from without; but the mansion was\ndistant from Laffert's habitation. This spot was wholly unfrequented. No\npassenger was likely to be stationed where her call could be heard.\nBesides, this forcible detention might operate for a short time, and be\nattended with no mischievous consequences. Whatever was to come, it was\nher duty to collect her courage and encounter it.\n\nTho steps of Ormond above now gave tokens of his approach. Vigilant\nobservance of this man was all that her situation permitted. A vehement\neffort restored her to some degree of composure. Her stifled\npalpitations allowed her steadfastly to notice him as he now descended\nthe stairs, bearing a lifeless body in his arms. \"There!\" said he, as he\ncast it at her feet; \"whose countenance is that? Who would imagine that\nfeatures like those belonged to an assassin and impostor?\"\n\nClosed eyelids and fallen muscles could not hide from her lineaments so\noften seen. She shrunk back and exclaimed, \"Thomas Craig!\"\n\nA pause succeeded, in which she alternately gazed at the countenance of\nthis unfortunate wretch and at Ormond. At length, the latter\nexclaimed,--\n\n\"Well, my girl, hast thou examined him? Dost thou recognise a friend or\nan enemy?\"\n\n\"I know him well: but how came this? What purpose brought him hither?\nWho was the author of his fate?\"\n\n\"Have I not already told thee that Ormond was his own avenger and thine?\nTo thee and to me he has been a robber. To him thy father is indebted\nfor the loss not only of property but life. Did crimes like these merit\na less punishment? And what recompense is due to him whose vigilance\npursued him hither and made him pay for his offences with his blood?\nWhat benefit have I received at thy hand to authorize me, for thy sake,\nto take away his life?\"\n\n\"No benefit received from me,\" said Constantia, \"would justify such an\nact. I should have abhorred myself for annexing to my benefits so bloody\na condition. It calls for no gratitude or recompense. Its suitable\nattendant is remorse. That he is a thief, I know but too well; that my\nfather died by his hand is incredible. No motives or means--\"\n\n\"Why so?\" interrupted Ormond. \"Does not sleep seal up the senses? Cannot\nclosets be unlocked at midnight? Cannot adjoining houses communicate by\ndoors? Cannot these doors be hidden from suspicion by a sheet of\ncanvas?\"\n\nThese words were of startling and abundant import. They reminded her of\ncircumstances in her father's chamber, which sufficiently explained the\nmeans by which his life was assailed. The closet, and its canvas-covered\nwall; the adjoining house untenanted and shut up--but this house, though\nunoccupied, belonged to Ormond. From the inferences which flowed hence,\nher attention was withdrawn by her companion, who continued:--\n\n\"Do these means imply the interposal of a miracle? His motives? What\nscruples can be expected from a man inured from infancy to cunning and\npillage? Will he abstain from murder when urged by excruciating poverty,\nby menaces of persecution, by terror of expiring on the gallows?\"\n\nTumultuous suspicions were now awakened in the mind of Constantia. Her\nfaltering voice scarcely allowed her to ask, \"How know _you_ that Craig\nwas thus guilty?--that these were his incitements and means?\"\n\nOrmond's solemnity now gave place to a tone of sarcasm and looks of\nexultation:--\"Poor Constantia! Thou art still pestered with incredulity\nand doubts! My veracity is still in question! My knowledge, girl, is\ninfallible. That these were his means of access I cannot be ignorant,\nfor I pointed them out. He was urged by these motives, for they were\nstated and enforced by me. His was the deed, for I stood beside him when\nit was done.\"\n\nThese, indeed, were terms that stood in no need of further explanation.\nThe veil that shrouded this formidable being was lifted high enough to\nmake him be regarded with inexplicable horror. What his future acts\nshould be, how his omens of ill were to be solved, were still involved\nin uncertainty.\n\nIn the midst of fears for her own safety, by which Constantia was now\nassailed, the image of her father was revived; keen regret and vehement\nupbraiding were conjured up.\n\n\"Craig, then, was the instrument, and yours the instigation, that\ndestroyed my father! In what had he offended you? What cause had he\ngiven for resentment?\"\n\n\"Cause!\" replied he, with impetuous accents. \"Resentment! None. My\nmotive was benevolent; my deed conferred a benefit. I gave him sight and\ntook away his life, from motives equally wise. Know you not that Ormond\nwas fool enough to set value on the affections of a woman? These were\nsought with preposterous anxiety and endless labour. Among other\nfacilitators of his purpose, he summoned gratitude to his aid. To\nsnatch you from poverty, to restore his sight to your father, were\nexpected to operate as incentives to love.\n\n\"But here I was the dupe of error. A thousand prejudices stood in my\nway. These, provided our intercourse were not obstructed, I hoped to\nsubdue. The rage of innovation seized your father: this, blended with a\nmortal antipathy to me, made him labour to seduce you from the bosom of\nyour peaceful country; to make you enter on a boisterous sea; to visit\nlands where all is havoc and hostility; to snatch you from the influence\nof my arguments.\n\n\"This new obstacle I was bound to remove. While revolving the means,\nchance and his evil destiny threw Craig in my way. I soon convinced him\nthat his reputation and his life were in my hands. His retention of\nthese depended upon my will, on the performance of conditions which I\nprescribed.\n\n\"My happiness and yours depended on your concurrence with my wishes.\nYour father's life was an obstacle to your concurrence. For killing him,\ntherefore, I may claim your gratitude. His death was a due and\ndisinterested offering at the altar of your felicity and mine.\n\n\"My deed was not injurious to him. At his age, death, whose coming at\nsome period is inevitable, could not be distant. To make it unforeseen\nand brief, and void of pain,--to preclude the torments of a lingering\nmalady, a slow and visible descent to the grave,--was the dictate of\nbeneficence. But of what value was a continuance of his life? Either you\nwould have gone with him to Europe or have stayed at home with me. In\nthe first case, his life would have been rapidly consumed by perils and\ncares. In the second, separation from you, and union with me,--a being\nso detestable,--would equally have poisoned his existence.\n\n\"Craig's cowardice and crimes made him a pliant and commodious tool. I\npointed out the way. The unsuspected door which led into the closet of\nyour father's chamber was made, by my direction, during the life of\nHelena. By this avenue I was wont to post myself where all your\nconversations could be overheard. By this avenue an entrance and\nretreat were afforded to the agent of my newest purpose.\n\n\"Fool that I was! I solaced myself with the belief that all impediments\nwere now smoothed, when a new enemy appeared. My folly lasted as long as\nmy hope. I saw that to gain your affections, fortified by antiquated\nscruples and obsequious to the guidance of this new monitor, was\nimpossible. It is not my way to toil after that which is beyond my\nreach. If the greater good be inaccessible, I learn to be contented with\nthe less.\n\n\"I have served you with successless sedulity. I have set an engine in\nact to obliterate an obstacle to your felicity, and lay your father at\nrest. Under my guidance, this engine was productive only of good.\nGoverned by itself or by another, it will only work you harm. I have,\ntherefore, hastened to destroy it. Lo! it is now before you motionless\nand impotent.\n\n\"For this complexity of benefit I look for no reward. I am not tired of\nwell-doing. Having ceased to labour for an unattainable good, I have\ncome hither to possess myself of all that I now crave, and by the same\ndeed to afford you an illustrious opportunity to signalize your wisdom\nand your fortitude.\"\n\nDuring this speech, the mind of Constantia became more deeply pervaded\nwith dread of some overhanging but incomprehensible evil. The strongest\nimpulse was to gain a safe asylum, at a distance from this spot and from\nthe presence of this extraordinary being. This impulse was followed by\nthe recollection that her liberty was taken away, that egress from the\nhall was denied her, and that this restriction might be part of some\nconspiracy of Ormond against her life.\n\nSecurity from danger like this would be, in the first place, sought, by\none of Constantia's sex and opinions, in flight. This had been rendered,\nby some fatal chance or by the precautions of her foe, impracticable.\nStratagem or force was all that remained to elude or disarm her\nadversary. For the contrivance and execution of fraud, all the habits of\nher life and all the maxims of her education had conspired to unfit her.\nHer force of muscles would avail her nothing against the superior\nenergy of Ormond.\n\nShe remembered that to inflict death was no iniquitous exertion of\nself-defence, and that the penknife which she held in her hand was\ncapable of this service. She had used it to remove any lurking\nobstruction in the wards of her key, supposing, for a time, this to be\nthe cause of her failing to withdraw the bolt of the door. This resource\nwas, indeed, scarcely less disastrous and deplorable than any fate from\nwhich it could rescue her. Some uncertainty still involved the\nintentions of Ormond. As soon as he paused, she spoke:--\n\n\"How am I to understand this prelude? Let me know the full extent of my\ndanger,--why it is that I am hindered from leaving this house, and why\nthis interview was sought.\"\n\n\"Ah, Constantia, this, indeed, is merely a prelude to a scene that is to\nterminate my influence over thy fate. When this is past I have sworn to\npart with thee forever. Art thou still dubious of my purpose? Art thou\nnot a woman? And have I not entreated for thy love and been rejected?\n\n\"Canst thou imagine that I aim at thy life? My avowals of love were\nsincere; my passion was vehement and undisguised. It gave dignity and\nvalue to a gift in thy power, as a woman, to bestow. This has been\ndenied. That gift has lost none of its value in my eyes. What thou\nrefusest to bestow it is in my power to extort. I came for that end.\nWhen this end is accomplished, I will restore thee to liberty.\"\n\nThese words were accompanied by looks that rendered all explanation of\ntheir meaning useless. The evil reserved for her, hitherto obscured by\nhalf-disclosed and contradictory attributes, was now sufficiently\napparent. The truth in this respect unveiled itself with the rapidity\nand brightness of an electrical flash.\n\nShe was silent. She cast her eyes at the windows and doors. Escape\nthrough them was hopeless. She looked at those lineaments of Ormond\nwhich evinced his disdain of supplication and inexorable passions. She\nfelt that entreaty and argument would be vain; that all appeals to his\ncompassion and benevolence would counteract her purpose, since, in the\nunexampled conformation of this man's mind, these principles were made\nsubservient to his most flagitious designs. Considerations of justice\nand pity were made, by a fatal perverseness of reasoning, champions and\nbulwarks of his most atrocious mistakes.\n\nThe last extremes of opposition, the most violent expedients for\ndefence, would be justified by being indispensable. To find safety for\nher honour, even in the blood of an assailant, was the prescription of\nduty. Tho equity of this species of defence was not, in the present\nconfusion of her mind, a subject of momentary doubt.\n\nTo forewarn him of her desperate purpose would be to furnish him with\nmeans of counteraction. Her weapon would easily be wrested from her\nfeeble hand. Ineffectual opposition would only precipitate her evil\ndestiny. A rage, contented with nothing less than her life, might be\nawakened in his bosom. But was not this to be desired? Death, untimely\nand violent, was better than the loss of honour.\n\nThis thought led to a new series of reflections. She involuntarily\nshrunk from the act of killing: but would her efforts to destroy her\nadversary be effectual? Would not his strength and dexterity easily\nrepel or elude them? Her power in this respect was questionable, but her\npower was undeniably sufficient to a different end. The instrument which\ncould not rescue her from this injury by the destruction of another\nmight save her from it by her own destruction.\n\nThese thoughts rapidly occurred; but the resolution to which they led\nwas scarcely formed, when Ormond advanced towards her. She recoiled a\nfew steps, and, showing the knife which she held, said,--\n\n\"Ormond! Beware! Know that my unalterable resolution is to die\nuninjured. I have the means in my power. Stop where you are; one step\nmore, and I plunge this knife into my heart. I know that to contend with\nyour strength or your reason would be vain. To turn this weapon against\nyou I should not fear, if I were sure of success; but to that I will\nnot trust. To save a greater good by the sacrifice of life is in my\npower, and that sacrifice shall be made.\"\n\n\"Poor Constantia!\" replied Ormond, in a tone of contempt; \"so thou\npreferrest thy imaginary honour to life! To escape this injury without a\nname or substance, without connection with the past or future, without\ncontamination of thy purity or thraldom of thy will, thou wilt kill\nthyself; put an end to thy activity in virtue's cause; rob thy friend of\nher solace, the world of thy beneficence, thyself of being and pleasure?\n\n\"I shall be grieved for the fatal issue of my experiment; I shall mourn\nover thy martyrdom to the most opprobrious and contemptible of all\nerrors: but that thou shouldst undergo the trial is decreed. There is\nstill an interval of hope that thy cowardice is counterfeited, or that\nit will give place to wisdom and courage.\n\n\"Whatever thou intendest by way of prevention or cure, it behooves thee\nto employ with steadfastness. Die with the guilt of suicide and the\nbrand of cowardice upon thy memory, or live with thy claims to felicity\nand approbation undiminished. Choose which thou wilt. Thy decision is of\nmoment to thyself, but of none to me. Living or dead, the prize that I\nhave in view shall be mine.\"\n\n\n\n\nCHAPTER X.\n\n\nIt will be requisite to withdraw your attention from this scene for a\nmoment, and fix it on myself. My impatience of my friend's delay, for\nsome days preceding this disastrous interview, became continually more\npainful. As the time of our departure approached, my dread of some\nmisfortune or impediment increased. Ormond's disappearance from the\nscene contributed but little to my consolation. To wrap his purposes in\nmystery, to place himself at seeming distance, was the usual artifice of\nsuch as he,--was necessary to the maturing of his project and the\nhopeless entanglement of his victim. I saw no means of placing the\nsafety of my friend beyond his reach. Between different methods of\nprocedure, there was, however, room for choice. Her present abode was\nmore hazardous than an abode in the city. To be alone argued a state\nmore defenceless and perilous than to be attended by me.\n\nI wrote her an urgent admonition to return. My remonstrances were\ncouched in such terms as, in my own opinion, laid her under the\nnecessity of immediate compliance. The letter was despatched by the\nusual messenger, and for some hours I solaced myself with the prospect\nof a speedy meeting.\n\nThese thoughts gave place to doubt and apprehension. I began to distrust\nthe efficacy of my arguments, and to invent a thousand reasons, inducing\nher, in defiance of my rhetoric, at least to protract her absence. These\nreasons I had not previously conceived, and had not, therefore,\nattempted, in my letter, to invalidate their force. This omission was\npossible to be supplied in a second epistle; but, meanwhile, time would\nbe lost, and my new arguments might, like the old, fail to convince\nher. At least, the tongue was a much more versatile and powerful\nadvocate than the pen; and, by hastening to her habitation, I might\neither compel her to return with me, or ward off danger by my presence,\nor share it with her. I finally resolved to join her by the speediest\nconveyance.\n\nThis resolution was suggested by the meditations of a sleepless night. I\nrose with the dawn, and sought out the means of transporting myself,\nwith most celerity, to the abode of my friend. A stage-boat, accustomed\ntwice a day to cross New York Bay to Staten Island, was prevailed upon,\nby liberal offers, to set out upon the voyage at the dawn of day. The\nsky was gloomy, and the air boisterous and unsettled. The wind, suddenly\nbecoming tempestuous and adverse, rendered the voyage at once tedious\nand full of peril. A voyage of nine miles was not effected in less than\neight hours and without imminent and hairbreadth danger of being\ndrowned.\n\nFifteen miles of the journey remained to be performed by land. A\ncarriage, with the utmost difficulty, was procured, but lank horses and\na crazy vehicle were but little in unison with my impatience. We reached\nnot Amboy ferry till some hours after nightfall. I was rowed across the\nSound, and proceeded to accomplish the remainder of my journey--about\nthree miles--on foot.\n\nI was actuated to this speed by indefinite but powerful motives. The\nbelief that my speedy arrival was essential to the rescue of my friend\nfrom some inexplicable injury haunted me with ceaseless importunity. On\nno account would I have consented to postpone this precipitate\nexpedition till the morrow.\n\nI at length arrived at Dudley's farm-house. The inhabitants were struck\nwith wonder at the sight of me. My clothes were stained by the water by\nwhich every passenger was copiously sprinkled during our boisterous\nnavigation, and soiled by dust; my frame was almost overpowered by\nfatigue and abstinence.\n\nTo my anxious inquiries respecting my friend, they told me that her\nevenings were usually spent at the mansion, where it was probable she\nwas now to be found. They were not apprized of any inconvenience or\ndanger that betided her. It was her custom sometimes to prolong her\nabsence till midnight.\n\nI could not applaud the discretion nor censure the temerity of this\nproceeding. My mind was harassed by unintelligible omens and\nself-confuted fears. To obviate the danger and to banish my inquietudes\nwas my first duty. For this end I hastened to the mansion. Having passed\nthe intervening hillocks and copses, I gained a view of the front of the\nbuilding. My heart suddenly sunk, on observing that no apartment--not\neven that in which I knew it was her custom to sit at these unseasonable\nhours--was illuminated. A gleam from the window of the study I should\nhave regarded as an argument at once of her presence and her safety.\n\nI approached the house with misgiving and faltering steps. The gate\nleading into a spacious court was open. A sound on one side attracted my\nattention. In the present state of my thoughts, any near or unexplained\nsound sufficed to startle me. Looking towards the quarter whence my\npanic was excited, I espied, through the dusk, a horse grazing, with his\nbridle thrown over his neck.\n\nThis appearance was a new source of perplexity and alarm. The inference\nwas unavoidable that a visitant was here. Who that visitant was, and how\nhe was now employed, was a subject of eager but fruitless curiosity.\nWithin and around the mansion, all was buried in the deepest repose. I\nnow approached the principal door, and, looking through the keyhole,\nperceived a lamp, standing on the lowest step of the staircase. It shed\na pale light over the lofty ceiling and marble balustrades. No face or\nmovement of a human being was perceptible.\n\nThese tokens assured me that some one was within: they also accounted\nfor the non-appearance of light at the window above. I withdrew my eye\nfrom this avenue, and was preparing to knock loudly for admission, when\nmy attention was awakened by some one who advanced to the door from the\ninside and seemed busily engaged in unlocking. I started back and waited\nwith impatience till the door should open and the person issue forth.\n\nPresently I heard a voice within exclaim, in accents of mingled terror\nand grief, \"Oh, what--what will become of me? Shall I never be released\nfrom this detested prison?\"\n\nThe voice was that of Constantia. It penetrated to my heart like an\nicebolt. I once more darted a glance through the crevice. A figure, with\ndifficulty recognised to be that of my friend, now appeared in sight.\nHer hands were clasped on her breast, her eyes wildly fixed upon the\nceiling and streaming with tears, and her hair unbound and falling\nconfusedly over her bosom and neck.\n\nMy sensations scarcely permitted me to call, \"Constantia! For Heaven's\nsake, what has happened to you? Open the door, I beseech you.\"\n\n\"What voice is that? Sophia Courtland! O my friend! I am imprisoned!\nSome demon has barred the door, beyond my power to unfasten. Ah, why\ncomest thou so late? Thy succour would have somewhat profited if sooner\ngiven; but now, the lost Constantia--\" Here her voice sunk into\nconvulsive sobs.\n\nIn the midst of my own despair, on perceiving the fulfilment of my\napprehensions, and what I regarded as the fatal execution of some\nproject of Ormond, I was not insensible to the suggestions of prudence.\nI entreated my friend to retain her courage, while I flew to Laffert's\nand returned with suitable assistance to burst open the door.\n\nThe people of the farm-house readily obeyed my summons. Accompanied by\nthree men of powerful sinews, sons and servants of the farmer, I\nreturned with the utmost expedition to the mansion. The lamp still\nremained in its former place, but our loudest calls were unanswered. The\nsilence was uninterrupted and profound.\n\nThe door yielded to strenuous and repeated efforts, and I rushed into\nthe hall. The first object that met my sight was my friend, stretched\nupon the floor, pale and motionless, supine, and with all the tokens of\ndeath.\n\nFrom this object my attention was speedily attracted by two figures,\nbreathless and supine like that of Constantia. One of them was Ormond. A\nsmile of disdain still sat upon his features. The wound by which he fell\nwas secret, and was scarcely betrayed by the effusion of a drop of\nblood. The face of the third victim was familiar to my early days. It\nwas that of the impostor whose artifice had torn from Mr. Dudley his\npeace and fortune.\n\nAn explication of this scene was hopeless. By what disastrous and\ninscrutable fate a place like this became the scene of such complicated\nhavoc, to whom Craig was indebted for his death, what evil had been\nmeditated or inflicted by Ormond, and by what means his project had\narrived at this bloody consummation, were topics of wild and fearful\nconjecture.\n\nBut my friend--the first impulse of my fears was to regard her as dead.\nHope and a closer observation outrooted, or, at least, suspended, this\nopinion. One of the men lifted her in his arms. No trace of blood or\nmark of fatal violence was discoverable, and the effusion of cold water\nrestored her, though slowly, to life.\n\nTo withdraw her from this spectacle of death was my first care. She\nsuffered herself to be led to the farm-house. She was carried to her\nchamber. For a time she appeared incapable of recollection. She grasped\nmy hand, as I sat by her bedside, but scarcely gave any other tokens of\nlife.\n\nFrom this state of inactivity she gradually recovered. I was actuated by\na thousand forebodings, but refrained from molesting her by\ninterrogation or condolence. I watched by her side in silence, but was\neager to collect from her own lips an account of this mysterious\ntransaction.\n\nAt length she opened her eyes, and appeared to recollect her present\nsituation, and the events which led to it. I inquired into her\ncondition, and asked if there were any thing in my power to procure or\nperform for her.\n\n\"Oh, my friend,\" she answered, \"what have I done, what have I suffered,\nwithin the last dreadful hour! The remembrance, though insupportable,\nwill never leave me. You can do nothing for my relief. All I claim is\nyour compassion and your sympathy.\"\n\n\"I hope,\" said I, \"that nothing has happened to load you with guilt or\nwith shame?\"\n\n\"Alas! I know not. My deed was scarcely the fruit of intention. It was\nsuggested by a momentary frenzy. I saw no other means of escaping from\nvileness and pollution. I was menaced with an evil worse than death. I\nforebore till my strength was almost subdued: the lapse of another\nmoment would have placed me beyond hope.\n\n\"My stroke was desperate and at random. It answered my purpose too well.\nHe cast at me a look of terrible upbraiding, but spoke not. His heart\nwas pierced, and he sunk, as if struck by lightning, at my feet. O much\nerring and unhappy Ormond! That thou shouldst thus untimely perish! That\nI should be thy executioner!\"\n\nThese words sufficiently explained the scene that I had witnessed. The\nviolence of Ormond had been repulsed by equal violence. His foul\nattempts had been prevented by his death. Not to deplore the necessity\nwhich had produced this act was impossible; but, since this necessity\nexisted, it was surely not a deed to be thought upon with lasting\nhorror, or to be allowed to generate remorse.\n\nIn consequence of this catastrophe, arduous duties had devolved upon me.\nThe people that surrounded me were powerless with terror. Their\nignorance and cowardice left them at a loss how to act in this\nemergency. They besought my direction, and willingly performed whatever\nI thought proper to enjoin upon them.\n\nNo deliberation was necessary to acquaint me with my duty. Laffert was\ndespatched to the nearest magistrate with a letter, in which his\nimmediate presence was entreated and these transactions were briefly\nexplained. Early the next day the formalities of justice, in the\ninspection of the bodies and the examination of witnesses, were\nexecuted. It would be needless to dwell on the particulars of this\ncatastrophe. A sufficient explanation has been given of the causes that\nled to it. They were such as exempted my friend from legal\nanimadversion. Her act was prompted by motives which every scheme of\njurisprudence known in the world not only exculpates, but applauds. To\nstate these motives before a tribunal hastily formed and exercising its\nfunctions on the spot was a task not to be avoided, though infinitely\npainful. Remonstrances the most urgent and pathetic could scarcely\nconquer her reluctance.\n\nThis task, however, was easy, in comparison with that which remained. To\nrestore health and equanimity to my friend; to repel the erroneous\naccusations of her conscience; to hinder her from musing, with eternal\nanguish, upon this catastrophe; to lay the spirit of secret upbraiding\nby which she was incessantly tormented, which bereft her of repose,\nempoisoned all her enjoyments, and menaced not only the subversion of\nher peace but the speedy destruction of her life, became my next\nemployment.\n\nMy counsels and remonstrances were not wholly inefficacious. They\nafforded me the prospect of her ultimate restoration to tranquillity.\nMeanwhile, I called to my aid the influence of time and of a change of\nscene. I hastened to embark with her for Europe. Our voyage was\ntempestuous and dangerous, but storms and perils at length gave way to\nsecurity and repose.\n\nBefore our voyage was commenced, I endeavoured to procure tidings of the\ntrue condition and designs of Ormond. My information extended no further\nthan that he had put his American property into the hands of Mr.\nMelbourne, and was preparing to embark for France. Courtland, who has\nsince been at Paris, and who, while there, became confidentially\nacquainted with Martinette de Beauvais, has communicated facts of an\nunexpected nature.\n\nAt the period of Ormond's return to Philadelphia, at which his last\ninterview with Constantia in that city took place, he visited\nMartinette. He avowed himself to be her brother, and supported his\npretensions by relating the incidents of his early life. A separation at\nthe age of fifteen, and which had lasted for the same number of years,\nmay be supposed to have considerably changed the countenance and figure\nshe had formerly known. His relationship was chiefly proved by the\nenumeration of incidents of which her brother only could be apprized.\n\nHe possessed a minute acquaintance with her own adventures, but\nconcealed from her the means by which he had procured the knowledge. He\nhad rarely and imperfectly alluded to his own opinions and projects, and\nhad maintained an invariable silence on the subject of his connection\nwith Constantia and Helena. Being informed of her intention to return to\nFrance, he readily complied with her request to accompany her in this\nvoyage. His intentions in this respect were frustrated by the dreadful\ncatastrophe that has been just related. Respecting this event,\nMartinette had collected only vague and perplexing information.\nCourtland, though able to remove her doubts, thought proper to withhold\nfrom her the knowledge he possessed.\n\nSince her arrival in England, the life of my friend has experienced\nlittle variation. Of her personal deportment and domestic habits you\nhave been a witness. These, therefore, it would be needless for me to\nexhibit. It is sufficient to have related events which the recentness of\nyour intercourse with her hindered you from knowing but by means of some\nformal narrative like the present. She and her friend only were able to\nimpart to you the knowledge which you have so anxiously sought. In\nconsideration of your merits and of your attachment to my friend, I have\nconsented to devote my leisure to this task.\n\nIt is now finished; and I have only to add my wishes that the perusal of\nthis tale may afford you as much instruction as the contemplation of the\nsufferings and vicissitudes of Constantia Dudley has afforded to me.\nFarewell.\n\nTHE END.\n\n\n\n***","meta":{"redpajama_set_name":"RedPajamaBook"}}