talks‎ > ‎

Working the airloom: on writing historical fiction

As I pack up my notes after five years' work on an historical novel, the question I'm asking myself is what was I thinking? If history's a pack of tricks we play on the dead, what's historical fiction? And why me? Why have I just spent four years turning history into fiction?

The reason, it happens, is eighteenth-century language and the accident of its freshness.

Three hundred years ago, the English language owed its freshness to the innocent eye of British empiricism. Today, eighteenth-century English owes its freshness to its innocence of Romanticism and Freud. In the sentence Miri's husband was shouting in his sleep, not words that she could recognize but simple, blurting fanfares of distress (Jim Crace, Quarantine, 1997) the metaphor is given pride of (last) place because what matters is Miri's husband's unconscious anguish. This is a twentieth-century first sentence. The sentence An author ought to consider himself, not as a gentleman who gives a private or eleemosynary treat, but rather as one who keeps a public ordinary, at which all persons are welcome for their money (Henry Fielding, Tom Jones, 1749), with its spelled-out correlation of author, gentleman, and innkeeper and the wit that finally settles on that unmetaphorical word money, is an eighteenth-century sentence. Today we think of eighteenth-century prose as Latinate, euphemistic, prone to grandiloquence. But in its time it felt unpacked and stripped down, and it can still feel unpacked and stripped down today. Then as now, its freshness resides in its freedom from an assumption from the past: then, from the Medieval and Renaissance assumption of an analogical universe; now, from nineteenth- and twentieth-century assumptions of the importance of formative experiences and unconscious life.

I'm thinking of eighteenth-century language like Defoe's, Richardson 's, or Sterne's. Language that puts on display the backtracks and detours of thought: But that which I was too vain of was my ruin, or rather my vanity was the cause of it (Daniel Defoe, Moll Flanders, 1722).  Language like Swift's, or Berkeley's. Language of concision and elision, telescoped by the immediacy of speech. Language in which vivid metaphor and evocative modification are not a priority, and when metaphor and modification do occur, they're explained, perfunctory, pale, or operate by placement. For cant and vision are to the ear and the eye the same that tickling is to the touch (Jonathan Swift, A Tale of a Tub, 1704). Language whose sentences at their most periodic or ironic, or both, are still efficiently denotative: The people of Rome, viewing with a secret pleasure the humiliation of the aristocracy, demanded only bread and public shows, and were supplied with both by the liberal hand of Augustus (Edward Gibbon, Decline and Fall of the Roman Empire, 1776-88). Language like the Brontes': Mr. Earnshaw snatched up the culprit directly and conveyed him to his chamber, where, doubtless, he administered a rough remedy to cool the fit of passion, for he reappeared red and breathless (Emily Brontë, Wuthering Heights, 1847). Language snatched up like a culprit and conveyed direct to the page. In other words, a close, naked, natural way of speaking: the way eighteenth-century British writers thought of themselves as writing when they wrote best, in their resolve (in the words of the Royal Society historian Thomas Sprat) to "return back to the primitive purity, and shortness, when men deliver'd so many things, almost in an equal number of words."

Put this way, the eighteenth-century project sounds as nutty as Swift knew it was. But if nutty then, pseudo now. Why not be satisfied with close, naked, and natural in contemporary English? By avoiding Romanticism and Freud, coming up with a style closer to the writer George Saunders', say, than the writer George Berkeley's?

Because eighteenth-century language is capable of more than closeness, nakedness, and naturalness. It also has a straightahead capacity for homeliness, strangeness, intensity, and verisimitude. The straightaheadness comes out of the Enlightenment stance of direct engagement with the world, out of the unruffled way eighteenth-century authors set down ordinary, everyday substantives. The birds peck the berries or the corn, and fly away to the groves where they sit in seeming happiness on the branches, and waste their lives in tuning one unvaried series of sounds (Samuel Johnson, Rasselas, 1759). This is also the homeliness of eighteenth-century language: here, the statedness of the set-up for a person/songbird comparison; the proliferation of simple nouns; with only three modifiers; the efficiency of the verbs: the sense, in all, of an orderly arrangement of things in unagonized moral space. As if the world were a home and morality the good housekeeping of clear and honest perception.

For eighteenth-century readers, what was sufficiently strange about Johnson's sentence to make it feel new was its show of innocent confidence in the value of natural observation. But what's really innocent is the eye of the perceiver. Because if innocent perception + clear language + rational moral principles = communicable truth, then subjectivity and Nature are redeemed together. And for modern readers as well, demoralized by the daily slog through the Media Blur, the Lies of Politics and Business, and the Aporia of Universal Relativism, our real lives lived elsewhere, in our secret moments, in our unconscious, and in our dreams, what is new and strange is to come upon an author who greets the world with unapologetic, rustic lucidity.

So much for homely and strange. As for intensity, this, when it comes to the novel, is an issue for me, and was probably the main reason I was looking to eighteenth-century language. Being too long-haul a read for its own good, the novel, while it generally welcomes homeliness, resists the intensity of literary language, as it does strangeness not rooted in simple-to-imagine characters or a simple-to-imagine world, as too exhausting for the reader. Poetry can be intense and strange, short stories can be intense and strange, but how do you make a novel intense and strange without tiring everybody out? One way—the eighteenth-century way—is by attention to things in their plenitude. In eighteenth-century language, even the more abstract substantives—like that Brontëan "rough remedy"—have a thinglike status. And this thingliness of the language is, in addition, a wonderfully efficient tool for generating versimilitude out of fantasy. Look at Robinson Crusoe, look at Wuthering Heights . (And just to be sure, how domestic these two narratives are.) This is what most interests me about the novel: a fantasy presented as if it were a likely account of the likeliest course of events you could hope to come across.  

So that's what I wanted and why: close, naked, natural language conducive to

homeliness, strangeness, intensity, and verisimilitude—language capable of displaying these seven virtues without convolution, or complication. Not much to ask, really, when all seven are available in one compendious, lately (if you ask me) new-all-over-again 280-year-old package.

Now all I needed was eighteenth-century characters and an eighteenth-century story.

A work of short fiction can originate in a voice. The dramatic tension within the voice can generate a pair of characters whose conflict will make a story. But language alone may not be enough when it comes to starting something as long, and longer in the writing, as a novel. You may be able to hear the voices in your head, but you may not be able to write them until they have something to tell.

After six or seven miserable starts, in the spring of 1998 I took another look at notes I'd made ten or eleven years before when I read Roy Porter's Faber Book of Madness, where he provides excerpts from an 1810 case history entitled Illustrations of Madness by the then apothecary of Bethlem Hospital, John Haslam. Haslam's Illustrations of Madness concerns a lunatic in his care named James Tilly Matthews. The Illustrations is the first book-length account in English of a paranoid system. I'd made the notes because two things struck me about the excerpts. First, the homely strangeness of Matthews' system. Second, the homely verisimilitude of Haslam's account of it. Matthews believed himself to be—like all lunatics, politicians, nobility, and royalty—a victim of secret gangs of French agents operating "influencing engines." These engines were magnetic-assailment machines he called Air Looms. Here is Haslam on one of the gang's favourite methods of assault, called Kiteing:

Kiteing.This is a very singular and distressing mode of assailment, and much practised by the gang. As boys raise a kite in the air, so these wretches, by means of the air-loom and magnetic impregnations, contrive to lift into the brain some particular idea, which floats and undulates in the intellect for hours together; and how much soever the person assailed may wish to direct his mind to other objects, and banish the idea forced upon him, he finds himself unable; as the idea which they have kited keeps waving in his mind, and fixes his attention to the exclusion of other thoughts. He is, during the whole time, conscious that the kited idea is extraneous, and does not belong to the train of his own cogitations.

James Matthews' Engraving of the Air Loom,

I liked the cruel solicitude of "very singular and distressing." I appreciated Haslam's care with the kite/idea analogy. I liked the way the image of the "air loom" inspires him to accentuate, Swift-style, the concrete in the metaphorical: "impregnations," "lift into the brain," "the train of his cogitations." I loved "floats and undulates."

Matthews was a madman in for political reasons. Attempting to prevent the outbreak of war with France after the Revolution, he'd set himself up as an emissary, travelling back and forth four times between London and Paris . Despite his efforts, war was declared, and he spent three years in French prisons before returning to England . But a year later—in 1796—he stood up in the public gallery of the House and charged the Government with treason. That got him put in Bethlem Hospital .

There the policy was to discharge patients after a year, unless they were judged a danger to themselves or others. But whatever exactly Matthews had done as a runner for the British government and the Girondins, it was enough to have him held there seventeen years, during fourteen of which his wife tried every means, including a writ of habeas corpus, to win his release. But since he wasn't legitimately in, it wasn't legitimately possible to get him out.

(Image: Bethlam Hospital, from London Wall. For more images, see

As for Haslam, he knew Matthews was a danger to no one, did not know why he had him, but as chief on-site medical officer, had no choice (if he wanted to keep his job) but to defend an unjust incarceration. Besides this, his medical judgment concerning Matthews having counted for nothing in the affair, he'd come out of it humiliated. The Illustrations, so lucidly, so empathically, so lovingly written, had, ironically, the purpose of discrediting Matthews, of saying to the public, Look, this is what he actually believes. The man is obviously a hopeless lunatic. Why shouldn't we have him?

So here, by a great stroke of luck, I had not only three excellent characters—James Tilly Matthews; his wife, nameless in the accounts but christened by me Margaret; and John Haslam—not only three excellent characters but my story, in fact the entire arc of my story. A superb, relatively unknown story. All I had to do was make a novel out of it.

Judging from the story and my own predilections, it would be as much a psychological novel with an historical setting, as an "historical novel" of the Sir Walter Scott variety, concerned with historical events and their human effects—though it would end up being that too. For contemporary models, I looked to Peter Ackroyd, Hilary Mantel; T. C. Boyle's Water Music, Andrew Millar's Ingenious Pain, Charles Palliser's The Quincunx, Iain Sinclair's White Chappell, Scarlet Tracings, and two years later to Emma Donahue's Slammerkin and Sheri Holman's The Dress Lodger. Mainly I was looking at their language, and to some extent their characters and the structure of their narratives.

What I didn't consciously appreciate at the time, but have since, is how much contemporary historical fiction—and more recent examples would be Michel Faber's The Crimson Petal and the White, Philip Hensher's The Mulberry Empire, and Sarah Waters' Fingersmith—how much is pastiche, more or less overt, as likely to be inspired by historical authors and styles as by historical figures and events. Waters' Fingersmith, for example, is said to be a subversion of Wilkie Collins' The Woman in White. In The Mulberry Empire, Hensher holds up flash cards of Austen, Dickens, Surtees, Thackeray, George Eliot, etc. that invite literary readers to play Name the Author. At some point early in the 1980s a new age of pastiche seems to have dawned, mainly of Victorian novels, probably inspired by reprints of Victorian porn such as The Pearl and The Romance of Lust, which have kept Grove Press solvent and served as bedside boomer reading for thirty years. Most of these novels are romps; their subject matter is what John Sutherland calls "low-life, high filth," but their provenance in fiction is arguably purer than that of fictionalized history. And what more appropriately postmodern strategy for flaunting and defying the embarrassment of pseudo than high-kicking pastiche?

Two novels I particularly enjoyed and learned from. One was Beryl Bainbridge's Master Georgie (1998), set in nineteenth-century Liverpool and amidst the Crimean War. But listen to this: I was twelve years old the first time Master Georgie ordered me to stand stock still and not blink. Is this a nineteenth-century first sentence? Master Georgie reminded me of something I'd once heard about J.M. Synge: he made up the West Ireland dialect of his plays. What was needed was a bespoke poetics. Not imitation, not parody, not cautious syntax-mapping, but to make something new that would give the impression of another time. Eighteenth-century language wasn't fresh for me for the same reasons it was fresh for the eighteenth century. There was no reason to pretend I'd be writing for an audience already dead more than a century and a half. For five years in the 1970s I had spent five and a half days a week in the British Library reading eighteenth-century literature, mostly poetry but also quite a lot of scientific, philosophical, and theological nonfiction. For the next twenty years I taught the primary eighteenth-century literary texts and many of the secondary. That would be the linguistic repository out of which I would shape, with my left hand, a language suitable to the task. There would be no mapping, no conscious imitation, no pastiche, but instead a style transmuted by eighteenth-century reading and more immediately inspired by specific historical voices, specific historical events.

Another thing: My story would concern three unusually intelligent and capable people who had risen by their own achievements to a precariously middle-class status: Matthews, before his madness, had come out of Spitalfields, where his mother was a widowed Huguenot silkweaver, to operate his own tea broking firm in Leadenhall Street . Of his wife, virtually nothing is known except that she took over the business when he was in Bethlem and spent fourteen years fighting the system to get him out. As for Haslam, though as an apothecary and mad-doctor, he occupied a low, dubious rank in the medical profession, he was responsible for the daily running of the greatest and most visible public mental hospital in the kingdom if not the Western World, but a hospital by this time a doddering hell-house in its last days at the Moorfields site. My point is, my characters were neither low nor high nor securely middle. Their status—their identities, their lives—were in every way unstable. Until Dickens, such characters don't show up much in literary fiction, which tends to bifurcate high and low. These characters promised an openness of possibility for how they would think, write, and talk.

As narrators, each would be a writer: Margaret in a diary, for her own consolation in the night, and in letters to her husband after the hospital governors stopped her visits; Matthews in undeliverable letters to her but also as part of his larger project of a log of Bethlem abuses that would eventually lead to public condemnation of Bethlem at Moorfields and Haslam's dismissal; and Haslam more retrospectively, in a spirit of setting the record straight in the wake of the collapse of his reputation, career and fortune.

The other novel that impressed me, even more than Master Georgie, was Peter Carey's Jack Maggs (1997): As the two footmen carried the remnants of the pudding down the staircase to the kitchen, Jack Maggs asked if Mr. Oates was in the habit of calling on Mr. Buckle, but what with the clatter of their heels and the rattle of the china, the question seemed to go unheard. Jack Maggs is energetic, witty, speedy, intensely thingly, homely, unaffected, realistic, and vivid. It's a remarkably accomplished evocation of the early nineteenth century by the virtuoso skill of its language. It sets the bar high.

But it wasn't the height of the bar that had me stalled for months on my first page.

Although by the time I started writing, or trying to, I had my language, characters, story, and a shuffle of usable plot ideas, I couldn't enter in. I wanted a third-person narrator like Carey's, but he shows no strain, I was showing nothing but. Who was my narrator? What kind of voice does a narrator have when his sole reason for existing is to translate the Eighteenth Century for the Twenty-first? Artificial, convoluted, and tedious, that's what kind of voice.

Before I found my story in Porter, I'd been clinging to the idea of some kind of clever interweaving of historical and present time. In other words, my anxiety about the problem of turning history into fiction was pushing me toward the common artistic dodge of foregrounding my technique for addressing it. This was also why my dual-time, time-travel contraptions were so geeky. This is why my narrator had no life.

Of course, there was a simple solution, which of course I resisted as long as I could: the first-person voice. What finally pushed me to it was three things: First, seeing Bainbridge do it with such grace in Master Georgie, where she uses three first-person narrators in six sections. Second, the realization that the first person would also solve another problem, which I'd been trying not to think about, though I knew the third-person was only going to compound it. This was the historical-fiction writer's dilemma of what to do about what kinds of chairs people sat on and what sorts of carriages they travelled in. If your narrator's primary function is to communicate a world, don't be surprised if he buries your characters and story in petticoats and spittoons. A thingly language is the stuff of a workable, workaday style. Logorrhea descriptitis is something else. The beauty of first-person narrators, particularly if they're still alive to the events they're recounting, is they notice their surroundings only when they have dramatic occasion to. This is a good thing.

The third reason I knuckled down to the first-person was that by this time I had a feeling, misconceived or not, that writing in character was something I could do, which is to say without thinking about it. Really, it's a form of acting: very, slow, acting. I'd tried it with eighteenth-century characters twice before, in two short stories, the 1986 one a little less stiff than the 1977 one. So this would be my way in: not through a mediating narrator, or a clever time-travel construct, but good old dramatic role-playing.

I had resisted the first person because I knew it would mean not one but three eighteenth-century voices, each recognizably different from the other two, and each required to negotiate with discretion the distance of two centuries while neither sounding not eighteenth-century enough nor too eighteenth-century and thus creating obstacles for contemporary readers for whom the stylings of Atwood and Carver are the quintessence of the natural. As with pastiche, a strategy more in the spirit of my anxiety, and of the late Twentieth Century, would have been to make a thing of the artifice. But the greater challenge, as Atwood and Carver always knew, is to make the artifice seem natural. Except, that way there's no avoiding displaying not your cleverness but the limits of your language skills.

But once I made the decision, my hope grew that the three voices, both in themselves and in their interaction, would be what mainly conveyed the eighteenth-centuriness of my story. The closeness, nakedness, naturalness, etc. of their language, along with the eighteenth-centuriness of their characters as evoked by their language and their concerns, would do what no amount of description of gaslamps and hairstyles could: create the sense of a former age. In this I'd be assisted by the eighteenth-century hunger for referring to things but not especially for describing them—and impulse that comes later, thanks to Scott and a cultural mood of historical self-consciousness, which resulted, it's been said, in the Victorians' writing historical fiction about the present. Instead of that, I'd work on my readers with the air loom of language and that way take them back in time.

Now, one thing about Bedlam is that its reputation has been preceding it for 600years. The connotations are in place even as the reader picks up the book, particularly if, as mine would be, it's a book called Bedlam. It had taken readers' reactions to my novel The Healer to teach me that my subject matter there was intractably gothic, that you don't get to write realistically about spiritual experience in the novel—an adamantly secular and social form, it took me writing The Healer to learn—without people thinking you're telling them a supernatural tale. With Bedlam too, the material was intractably gothic—horror, suffering, and alternate realities—but this time I knew it before I started writing. This time too the primary subject was not spiritual experience and its social ramifications but love vs. power, a subject infinitely friendlier to a psychological/historical novel. Given half a millennium of reader expectations about Bedlam, I had two choices: either over the top or proceed with restraint, letting reader preconceptions do the heavy lifting. My choice was the latter: not the Grand Guignol but the most accurate evocation I could manage of the reality of the historical place. I wanted people to spend time in the old Moorfields building, not for titillation sake but to gain some sense of what, behind the image, it actually might have been like and how it actually came to be as it was.

Put this way, how naive, how out-to-lunch unfashionable the project sounds. more in the spirit of today is what Sheri Holman has the narrator of The Dress Lodger (2000) say of a nineteenth-century Sunderland barman who sweetens his wine with lead, He knows he stands a chance of poisoning half his clientele, but most drink beer, so he doesn't sweat it. Since it's all a verbal game, why not have fun with it? and while you're at it, let your reader see how much fun you're having? On the one hand, this is not a bad option at a time when history is viewed as essentially unrecoverable and reality beyond the signage is more signage at a further remove.

On the other, my own experience is that reality has a way of impressing itself upon the mind with a degree and subtlety of complexity that semiotics ultimately fails to respect. And just as a mind that holds perception as a priority will have a different quality to it than a mind that priorizes thought, a fictional text that owes its provenance to historical texts will have a different quality to it than one whose provenance is fiction. And while the historical does seem to introduce an element of adulteration, or bastardization, or phoniness, to the endeavour—especially when not only the setting but the characters and the events are historical—the thing is this: it's all a confection anyway, an air loom emanation, that much we know, so why not have fun with it another way and try for versimilitude? Why labour five years on a text that struts its own technique when the message that literary texts are verbal artifacts is no longer the news it was over thirty years ago now, when John Fowles put an atomic bomb in The French Lieutenant's Woman? If I want to advertise my devices, I can always give a lecture and put it on the Web.

Still, if like Sheri Holman I could have come up with a modern idiom capable of evoking eighteenth-century London in a way I found satisfying, I might have made the same choice she did. The problem was, anachronism—whether of diction or syntax—really bothers me. I do believe the spirit of an age is inscribed in its language. Sloppy imitations of eighteenth- and nineteenth-century literature pain me, whereas skilful ones, such as Carey's Jack Maggs, give me tremendous pleasure. This is why in Bedlam I've tried hard to do James Murray and Eric Partridge proud: no anachronisms. No sweating it. My strict rule has been only diction that appears in an earlier example of usage in the OED, or when Partridge allows a piece of slang to be eighteenth- or, for the later sections of the book, early nineteeth-century. So kow-tow (earliest recorded use as a verb, 1826) was out, cab (short for cabriolet, earliest use 1827) was out, air-gun (Chambers, 1753) was in, peepers (not spectacles but slang for eyes, 1700) was in, small hours (Dickens, 1836-37) was out, hard-favoured (ugly, 1513) was in, invertebrate (as a noun, 1826) was out, knacky (ingenious, 1710) was in; turnabout (1833) was out, dirty puzzle (slut, 1680-1830) was in, bogus (U.S., 1852) was out, molly (sodomite, Ned Ward 1709) was in, shortfall (1895) was definitely out. The gang would not practise telepathy (1882) but mind-sight (1587). My surgeon Crowther wouldn't be drunk as a rolling fart (c. 1860), but he would swill like tinker (late 17th C). Haslam won't want Margaret Matthews looking smart as a thrupence (c. 1887) but smart as a carrot (1780).

But why observe finicky rules when not one reader in a thousand can spot an 1830s word in an 1810 sentence and maybe three in 10,000 are going to worry about it? Well, don't reviews of historical novels traditionally end with a list of things the author "got wrong"? Didn't Annie Proulx when she reviewed Guy Vanderhaeghe's The Last Crossing in the Globe and Mail take a moment from her rave to point out that cigarettes weren't smoked until 18-such-and-such? And here's another, more general good reason: formal constraints are helps to improve the art. If an otherwise perfect piece of 1820s slang won't do for a 1797 letter, then it might just inspire you to find something earlier as good that will. This means ransacking Partridge but also rummaging deeper in your own eighteenth-century word-hoard, in the process of which all kinds of other things will turn up. Like ancient pottery surfacing behind a plough, language fragments will rise to mind that eventually you'll find a place for and so your text be made a little more "likely."

I mean likely within the parameters indicated by Jane Austen when she advises her niece concerning a story the girl has written:

I have scratched out Sir Tho: from walking with the other Men to the stables &c the very day after his breaking his arm—for though I find your Papa did walk out immediately after his arm was set, I think it can be so little usual as to appear unnatural in a book.

The appearance of the natural is the morality of fiction. The attraction of historical narrative for a fiction writer is at the same time healthy and potential for disaster. Historians, like journalists, lacking the necessary information, inclination, imagination, or all three, tend to stop short of getting at the "inside," human-experience dimension of the story that fiction is designed to explore. Fiction writers will ever be drawn into the breach. But more than this, the attraction for fiction writers of true narratives is that, while not necessarily complex in themselves, they're often studded with circumstantial, puzzling, and suggestive details that have the power to inspire more complex, subtle, lifelike, and plausible fictions than imagination is normally capable of on its own. For example, after the failure of her writ of habeas corpus, Mrs. Matthews sailed with her children for Jamaica. That's what I would call an anomalous, suggestive detail. Were this my own fiction, it's a plot development I wouldn't have come up with in ten years. But it certainly does have a way of making a story about marital devotion pitted against insurmountable forces more complex, credible, and interesting.   

But the fact remains that the modus operandi of fiction being imagination, it can't afford too much of what really happened, because the truth is too strange, too implausible. Whoever it was said truth is stranger than fiction was a student of the obvious. While a certain amount of the strangeness of truth is generally good for fiction, for suggesting the anomalous texture of the real, import too much, slip it in merely "because it happened," because you know someone who walked out the day after he broke his arm, or simply fail to assimilate what you do import, and your reader will think either, Hell, that's not very likely, is it? Or, What's with all these chunks of undigested fact? Nobody wants to know what the author had for breakfast.

Like linguistic precision, a policy of historical accuracy in fiction spurs the author to serendipity-friendly industry, minimizes the chance of reader distraction by perceived error, and can, if carefully used, contribute to verisimilitude. The primary issue is not referential or historical truth per se but plausibility, which has everything to do with the integrity of the illusion and nothing to do with accuracy for its own sake. Melville can make all kinds of errors of whaling fact. They'll only ever be secondary, chinks in the armour of authorial authority. Moby Dick being art, what matters is the formal coherence of the text. In historical fiction an authorial bias toward the referential of history results in the reader's beginning to feel a weight—and weight is the word—of language, of research, of ball gowns and furniture, of historic events, or of (God forbid) ideas, because the referential, if assumed to have priority, has a way of swelling to a significance out of proportion to, or not in key with, that of the story. 

In the writing of Bedlam, the first-person did help me to go light enough on furniture and events that my editor was asking for more on what London looked like in 1800. While some readers, she pointed out, will have heard of the French Revolution, few will know when it was, what it was about, or that it was followed by war with England. Putting in such information later—the Gunter Grass method—was easy enough, because as Dickens and Emily Brontë had taught me, you don't really need much. Up to a point, the less the better. There is not a single set-piece description in Wuthering Heights. Because the thing is, if you're in it mainly for the referential, unless your ambition is to be James Michener or Tom Clancy, you're better off, from an aesthetic, though not necessarily a financial point of view, though probably that too, to come clean and write history.

Still, my Bedlam remains a composite, a concoction. My John Haslam does not sound quite like the historical John Haslam, who tends to be wordier, more ponderous, though some of his inspired phrasings have floated and undulated intact through many drafts. For the most part my James Tilly Matthews is more lucid than the man one imagines from the passages contributed by his historical counterpart to Haslam's Illustrations and from the two extant letters he wrote to Lord Liverpool calling his Lordship a murderous traitor—though I'm pleased to report that, to my ear, my character could be the author of the sane if desperate letters Matthews wrote during his house arrest and imprisonment in France, letters I discovered in the Paris foreign archives only after the novel was mostly finished. In any case, mental illness is not particularly dramatic—what will the mad not say?—and Matthews' raving seems to have waxed and waned according to the degree of stress he was under. My Margaret Matthews has been cut, by necessity, from whole cloth, and I can only trust she sounds like a faithful, resourceful eighteenth-century wife of a mad tea broker, who like her husband was nobody's fool.

As to whether Bedlam works, well . . . . It's too easy to offer promising reflections on a literary text, even reflections of the most accurate and formally sophisticated kind (which these are not), without coming close to saying anything useful about whether or not the thing is engaging, or even readable. Such commentary exists in a parallel universe, in a state of eternal dislocation from the actual experience of the words on the page. A commentary such as this may point to that experience, but it's not it, only draws inspiration from the insights that went into creating it. Just as the literary text draws inspiration from where it points, while remaining, if it works, ineffably what it is.

Or so every writer will ever vainly hope.

A version of "Working the Air Loom: On Writing Eighteenth-century Fiction" was delivered as the F.M. Salter Lecture on Language at the University of Alberta in March 2003.