In the grip of Possession

How do ideas take shape in an artist's mind? Why are some thoughts nurt ured and others cast aside? Here, A S Byatt, a speaker in the conference on art istic decision-making at the Tate tomorrow, looks back on the long gestation of her Booker Prize-winning novel, Possession

The beginning of Possession, and the first choice, was most unusually for me the title. I thought of it in the British Library, watching that great Coleridge scholar Kathleen Coburn circumambulating the catalogue. I thought: "She has given all her life to his thoughts", and then I thought: "She has mediated his thoughts to me." And then I thought: "Does he possess her, or does she possess him? There could be a novel called Possession about the relations between living and dead minds." This must have been in the late Sixties. It was the time of the nouveau roman, of the novel as "text". When I first recognise a thought as the germ of a story, I form a shape, or file, in a corner of my mind, to which I add things that seem to belong to it - quotations, observations. At that stage this Gestalt is more like the plan for a painting than a novel. It has colour and texture, though I have to think hard to call these to mind. The ur Gestalt of Possession was a grey cloudy web to do with the ghostliness and connectedness of the original idea. I think it was also to do with the nouveau roman, which I still visualise in that form. I imagined my text as a web of scholarly quotations and parodies through which the poems and writings of the dead should loom at the reader, to be surmised and guessed at.

The next decisive choices came in the 1980s when I was teaching Browning and George Eliot, and also lecturing on Henry James and his father, Henry James senior, a leading Swedenborgian. I had had the idea that the word possession involved both the daemonic and the economic. Reading the Browning letters made me see that "possession" had a primary sexual connotation, too. I made a decision: there should be two couples, man and woman, one alive and one dead. The novel would concern the relations between these two pairs.

My grey cobwebby palimpsest changed colour. It took on a lurid black, shot with crimson and scarlet, colours of passion. I was teaching that great novel, The Bostonians, with its world of "witches, wizards, mediums, and spirit-rappers and roaring radicals" to a generation of students involved in the politics of gender, who disliked James's tragicomic treatment of lesbian passion. It occurred to me that in the world of 19th-century spiritualism and feminism, possession had both its meanings at once.

So there was a need for the 19th-century woman to be a lesbian, or to be thought to be a lesbian, and the 20th-century woman scholar to be a feminist. What George Eliot's letters added to this texture of texts was the sense I always have that her real passionate self is splendidly absent from the letters kept by the people who kept them.

Her love-letters, unlike those of the Brownings, were buried with her. It is the luck of an unusually devoted marriage between poets once separated that we have the Browning letters. There have been serious proposals to dig up George Eliot. There is a gothic plot, I thought, of violence and skulduggery. The Gestalt got more lurid, purple, black, vermilion, with flying white forms.

I half-knew that the form of my novel should be a parody of every possible form, popular and "high culture", when I was asked to review Umberto Eco's Reflections on the Name of the Rose, which combines medieval theology, church history, gleefully bloodthirsty horrors, reflections on the form of the novel, with a hero who is an avatar or precursor of Sherlock Holmes. What entranced me about Eco's Reflections was his pleasure ("I wanted to murder a monk") and his technical reflections on the fact that detective stories and melodramas had to be written backwards. If you want to burn down a library quickly and irretrievably, you must make it burnable when you invent its architecture.

I had been thinking a lot about the pleasure principle in art. Art does not exist for politics or for instruction - it exists primarily for pleasure, or it is nothing. It can do the other things if it gives pleasure, as Coleridge knew. And the pleasure of fiction is narrative discovery, as it was easy to say about television serials and detective stories, but not, in those days, about serious novels.

So my novel should be a parody, not of Sherlock Holmes, but of the Margery Allingham detective stories I grew up on. It should learn from my childhood obsession, Georgette Heyer, to be a romance, and it could learn simultaneously from Hawthorne, Henry James's predecessor, that a historical romance is not realist, and desires to "connect a bygone time with the very present that is flitting away from us".

I added things: it should be an epistolary novel, which meant writing letters the scholars should find; it should contain early narrative forms (Victorian women writers wrote fairy-tales) and late ones (bits of biographies and critical "accounts" of what was going on). The Gestalt in my mind changed colour and form and became delicious, green and gold, the colours of Tennyson illustrations in my mind as a child, of dream landscapes, of childhood imaginings of a world brighter, more jewel-like than thisone.

There was a huge problem. I knew modern forms were parodic - not only Eco but the criticism of Malcolm Bradbury had been pointing that out - parodic not in a sneering or mocking way, but as "rewriting" or "representing" the past. The structural necessityof my new form was that the poems of my two poets, the most important thing about them in my own view, should be in this no-longer ghostly text. And I am not a poet, and novelists who write poems usually come to grief. Robertson Davies, the Canadian novelist, had written a novel with a parodic libretto made up of the poems of Thomas Lovell Beddoes. I said to the poet DJ Enright I was contemplating using the early poems of Pound that look as though they could be by Browning, "Nonsense," he said. "Write your own."

So I tried. My mind has been full since childhood of the rhythms of Tennyson and Browning, Rossetti and Keats. I read and reread Emily Dickinson, whose harsher and more sceptical voice I found more exciting than Christina Rossetti's meekness. I wanted a fierce female voice. And I found I was possessed - it was actually quite frightening. The 19th-century poems that were not 19th-century poems wrote themselves. They fitted into the metaphorical structure of my novel, but were not mine, as my prose is mine.

There is one further late choice I should like to mention. There are three passages in the 19th-century narrative which are recounted by a Victorian "omniscient" third-person narrator. These three include the epilogue, and tell what might be thought of as the most important, beautiful and terrible moments in the lives of the Victorian characters.

I still receive angry letters from time to time from all over the world, saying these passages are a mistake - that I have cleverly told the story of the past through documents, diaries, letters and poems, and that I am breaking my own convention incompetently. But my decision was very deliberate.

It was partly polemical, for two reasons. I do believe that biographies are a kind of shadow-play, and that what really mattered is likely to elude the piecers-together of lives. (Doris Lessing endorses this view, mischievously, at the start of her autobiography.) I also believe the third-person narrator has been much maligned in the recent past - it does not pretend to "God" - simply the narrative voice, which knows what it does know. And I wanted to show that such a voice can bring the reader nearer the passions and the thoughts of characters, without any obligation to admire the cleverness of the novelist. There is a nice irony about this - the writer and reader share what critics and scholars can't discover.

And the Gestalt now? A green and gold and blue balloon, far away, untouchable. A writer can't think about novels that have gone away. The Gestalt of the one I am writing, about the 1960s, is a jagged harlequin pattern of coloured fragments and smoking bonfires. And there is something weak about the narrative line, or tension, connecting these, that I'm trying to deal with.

n A S Byatt speaks tomorrow at `Decisions', a day event at the Tate Gallery, launching the London Consortium's radical new PhD in humanities. Artists, writers and film-makers will reflect on decisions they have made.

n Course prospectuses are available from Administrator, London Consortium, British Film Institute, 21 Stephen St, London W1. The Consortium is an amalgam of the BFI, Tate, Architectural Association and Birkbeck College, London University