`Then others for the breath of words respect,
         Me for my dumb thoughts, speaking in effect'.
                                                                           Shakespeare, Sonnet 85.

The first text I passively consumed - whatever the damage done to my juvenile psyche by Morally Improving Fairy Tales - was The Odyssey. This was read to us during Latin classes that burnt with somnolence and sunlight. It was 1966. Then there were the sprightly ballads of Causley and Auden; there was Brecht; there were the lyrics of W.S. Gilbert, the perpetual autumn of Keats. I languished disastrously under a drunken moon, and largely avoided games. Then came Shakespeare, even more unsuitable reading, and an extraordinary woman whom I'll call Phyllida Creech. Mrs Creech was one of the leading lights of Scarborough's Writers' Circle. I was invited to tea, a spurious adolescent who at the time (1974) was writing lines like `... their heads on cold stone pillows laid./But these same heads groan mouthingly in sleep...'[etc.].Thirty years on, and I quote from memory. I was probably trying to impress a girl. If there was a clumsy inversion, I half in love with it and easeful death became. If there was an unnecessary adjective or adverb, I would use it gratefully, with mercenary abandon. But these same heads groan mouthingly... `Mouthingly'? I guess my sixteen-year-old self meant that people groaned while moving their mouths. No doubt I was shattered by this insight. At the time, too good for the trade, I rewrote nothing.

Mrs Creech had been published somewhere in the gardening column of The Scarborough Mercury. She also had a smart line in scones, and kept a semi-bald parrot to which she'd taught the opening line of The Waste Land. The parrot was nearly as old as she was.

`Serendipity, dear boy,' she fluted, through a mouthful of crumbs. ` Apriw is de crullish mumf,' cried the parrot from under a cloth. `Yes, dear Simon,' she continued, `yes. Never, of course, pay to have one's work published. If it is of good enough quality, then Editors' - I heard the capitalisation at five paces - `then Editors will pay one for the privilege of bringing out one's work. But never forget serendipity.' She burped gamely. Her 1920s dress rearranged itself dimly on the sofa. `Sera-who?' I wondered whether I'd read him. ` Apriw is de crullish mumf,' cried the parrot.

Coleridge once famously described poetic faith as `That willing suspension of disbelief...' Keats defined the poetic imagination, whatever that may be, as negative capability. I had no faith in anything except feeble ventriloquism, and my literary imagination was summoned by an eighty-year- old lady, a spoonful of raspberry jam, and The Waste Land as rendered by avian distress. These dismal conjunctions, whose aspiration was the gardening column of The Scarborough Mercury, help to account for the mimicries and ambitions that pass, on a good day, for the mess of aural shadows that is my unconscious.

I've often pondered the fact that poets take the trouble to redraft. Translation I can begin to understand, where fidelity to an original is bought at a high price in burnt oil, but one's own originals? Why bother if, as the nostra of this post-critical age suggest, any text is as good as any other text? What price literary worth, and who is willing to pay it?

Soon after Mrs Creech there was Robin Skelton, whose The Practice of Poetry (London: Heinemann, 1971) I borrowed surreptitiously from Scarborough Public Library. Feeling like a milk-shaken, but not entirely stirred, version of Dylan Thomas as I rolled my first cigarette in a coffee bar that was mercifully free of parrots, I read quickly. Twice. I can scarcely bear to re-read Skelton's text now - it's too full of precious analogues between poetry and priesthood - but one chapter was devoted to re-writing, and gave examples of drafts from Thomas Kinsella's `Mirror in February'. As I read through Kinsella's re-drafts, all spidery handwriting and what looked like a difficult, honest effort to batter the poem into integrity and type, I became intrigued by the process through which poems become themselves, and impressed by the sheer work of attaining a finished, or better, an abandoned artefact out of the remorseless structures and fecundity of the English language. Perhaps Kinsella, too, was good enough to be published in The Scarborough Mercury.

It says something about how little, or how far, one develops that I'm intrigued by the same process now - and by the same tone: the dry, down-turning mouth, the arid and disappointed diction. It's a northern register, formed from a wry realism aware of evanescence, the presence of death. Those born under the northern moon successively rewrite the book of vanishings. `Not young and not renewable, but man,' Kinsella had written - a magnificently laconic closing line. I'd recognise the same tone and register in Larkin, in early R.S. Thomas. They were powerful enough to spawn a slew of imitations.

Kinsella's poem had become strong. Its metrics were sinews, its rhyme scheme flexible enough to accommodate normative diction, the ambit of its line weighing with a simultaneous mixture of exploration and decision. And this miracle had been achieved by choice, not arrived at by ventriloquism, ambition and accident. The choices, as Kinsella's successive drafts made clear, gave the abandoned text a clarity and integrity it would otherwise have lacked. I looked at my own non-serendipitous offerings and decided to burn them. The intolerable wrestle with words and meanings, rhythms and disasters had begun. `Apriw is de crullish mumf,' cried the parrot, sotto voce.

Not long afterwards I went up to Newcastle to read English. In the intervening years I had read everything. Ginsberg, Ferlinghetti, Lowell; Leopardi; Dante; the Daily Mail; Richard Brautigan; Joyce, Woolf; Hermann Hesse; Trout and Salmon; Goethe, Rilke; and Mayfair (particularly the Problem Pages). I had heard again - and like most of my generation, inevitably - the laconic, demotic brilliance of Auden: `About suffering they were never wrong, the Old Masters...' And I had started to read in the little magazines, poring over issues of Stand in the public library and wondering why so many of its published poems seemed to be written, no doubt on silver cigarette foil, by Rumanian revolutionaries whose names felt like biting on granite. I think I probably decided to study English because I wanted to be an ex-Rumanian revolutionary, or possibly a Rumanian ex-revolutionary, and grow a face as seamed and celebratory as Auden's. I broke out into non-filtered Sobranies, drank black coffee and cocktails, and was a pale, myopic mess.

The accident was waiting to happen. It duly did.

The accident was called Sweet's Anglo-Saxon Reader, whose eighth edition (Oxford: Clarendon Press, 1908) I'd picked up at a book-sale. I was entranced by the Anglo- Saxon alphabet, and completely mystified as to why these spiky, diacritic-ridden texts could function as verse. I began to translate The Battle of Maldon (AD 993):

Þær com flowende	flod æfter ebban,
lucon lagustreamas;	to lang hit him þuhte,
hwænne hi togædere	garas bæron...

`Flood after ebb-tide came flowing there,
locked lake-streams. It seemed too long to them all
before they could carry spears to each other...'

Gradually some verse-specific principles came clear. The long line of the original divided into two half-lines, and the half-lines were linked by predictable patterns of alliteration. The chief stressed syllables in each half-line had something in common: they were formed by syllables whose vowels were long, or were short, but closed by a consonant. There were anomalies. There were stressed positions spanning two syllables. (Later I would learn to call this resolution.) There were half-lines that didn't alliterate predictably. I had begun to work, to study, to read with pain for pleasure. Since I was lazy, and impatient with footnotes, I doubted I would ever make a scholar, but I got hold of a tweed jacket and a pair of horn-rims, and began to read Middle English lyrics alongside the Old English corpus. I even bought a recorder, and started composing simple melodic lines to accompany Middle English texts such as `He cam also stille...', or `An hendy hap Ichab ihent...' I wonder how anyone could have put up with me.

To their credit, not many people did, but the University of Newcastle had a writing group whose Creative Fellows were successively Fleur Adcock and Anne Stevenson. I inflicted myself for three years, and was rarely thrown out, largely because I paid for drinks. My verse, of course, was impeccable and above criticism, but the other members of the group - among them Gavin Smith, Margaret Whittle, Andy Bell - were tireless, hectic re-drafters, and very fine writers. Together with Fleur and Anne, they taught me again that the abandoned artefact was earned, and rarely given. It was cut from the muscles of language, however they were singing with pain, however the criticism may have hurt. I began to translate, to re-draft, and, though I pretended to ignore it, I accepted my colleagues' criticism because I admired their talents.

I learned the most valuable lesson of the entire drafting process: if it hasn't earned its form as a poem, if that precious structure could be put in any other way, then what you've written isn't a poem, and certainly not a poem worth publishing. This has been the linguistic litmus-test I've tried to apply to everything I've subsequently written. It allows me to proceed with less than twenty per cent of what I otherwise might have published. I have abandoned eighty per cent more.

In a spirit of gratitude, I'd like to reproduce one poem by Andy Bell:

Sinister Painting
In memory of Robert Lowell

They carried your casket to the sea.
I watched from high ground
their progress, their bootprints
in the sand; saw the wind
lay their bright neckerchiefs across their shoulders -
silken blue
and crimson splashes and the gulls
who tore at the eyes of the wind.

They were back by nightfall
carrying lobsters and saying things
like `Well, that's that.'
We tuned the drums and the crochetty fiddles
and gathered round the barrel
with our mugs, heard the women
coming up through the trees, their laughter
like tambourines in the dark.

The colours of calamity were in the sky
but I wasn't afraid
having looked up before and seen that before.
                                                           (from Spirit Level, privately printed, 1980)

Andy, and Gavin Smith, worked hard on line-breaks, typically ending the line so that it was co-extensive with a clause, however that clause would be subsequently post-modified (`their laughter/like tambourines...'; `the gulls/who tore at the eyes...'). Both wrote largely if not entirely in non-metrical forms. They taught me that though non-metrical forms might be `free' there are principles - of equivalence, parallelism, phrasing, juxtaposition and process - underlying apparently the free-est of verse. I often wished that my non-metrical writing aspired to that rigour, but at the same time the myopic student was beginning to acquire rudimentary metrical skills, and experimenting with them as fastidiously as if he were flicking a piece of ash from the sleeve of a fine-bone jacket bought for a shilling at a local jumble sale.

It was a piece of accident - serendipity? - that my first poems were published around the same time. They weren't entirely pieces of ventriloquism. `Apriw is de crullish mumf,' cried a parrot from its grave.

Publication is perilous. Any writer who's prepared to pay the price for literary worth has drawers-ful of rejection slips. They're often instructive. Several editors (though with hindsight they probably hoped I would just shut up and go away) sent back helpful or encouraging comments on my submissions, and those comments kept me writing, abandoning, trying again.

I can't let the invocation of editors pass without summoning the name of Michael Schmidt, to whom Anne Stevenson wrote me a letter of introduction in 1983. Michael's encouragement, and his invitation to write for PNR, sustained a chaotic spirit, taught me to work to deadlines, helped me to read, introduced me to work I would never otherwise have touched, and finessed some of my work into print. The reader will realise how much I owe him (typically, I receive no additional fees for this endorsement) when I confess he once told me bluntly that one of my poems was `simply flirting with nonsense'. And of course, it was.

I've been somewhat autobiographical not because I wish to emphasise the link between the life and the work, but because there are principles of re-drafting to recognise through the muddle of imitations, the fierce, often disillusioned experiments with voices.

First, there's a practical point. Since 1982 I've drafted material longhand, in A4 exercise books. Though I write everything else straight into the laptop, I've found that drafting verse in a workbook saves shuffling through random pieces of paper, or the uncertain business of opening computer files. In a longhand workbook, the rhythmic germ can be found again merely by skipping back a page or two; the thematic impetus can be kept from leaf to leaf. If the margins are wide, annotations can be made alongside doodles. I'm too old to change these habits. Below is a poem from a 1982 workbook which was subsequently published on BBC Radio 3 and in the first collection, Time Signatures (Carcanet, 1993):

Draft fragment, originally titled `Surbiton at Christmas'

In winter it is always four o'clock and Thursday
and the culverts in flood.
You need whisky in your blood-
stream to get you going,
say, in your early morning cup of tea...

This wouldn't do: the fussy line-break on the compound (`blood-/stream...'); the clumsy demotic (`say'). The next draft was some improvement:

In winter it is always four o'clock and Thursday
and the culverts in flood.
Arrowheads of birds on sky
eagerly, eagerly [replaced by: urgently, urgently] fly the passing weather by.
The allotments are mud.

And [inserted: every day] at four o'clock on Thursday
a neighbour calls
bringing mince-tarts (hot)
and a new [bawling] brat in a [blue] carry-cot.
Light falls.

The adjectives in the penultimate line were inserted on two grounds, alliteration and rhythm. This text was eventually printed, although by then I'd modified the title (to `Suburb in December', since Surbiton was too localised), and tinkered with the final line, which became `Light fails', eschewing the full-chime calls and falls, which added nothing. The half-rhyme was itself a kind of failure, and it seemed appropriate to end the poem on something almost finished but permanently incomplete: form as thematic mimesis.

This poem was written quickly, and unlike many pieces, seems to have been drafted from the theme embodied in its opening. There's a stress-timed rhythmic structure, where longer lines of (usually) five main stresses alternate with two-stressed lines. I suppose what I was working towards was rhythmic tension in the structure, aligned with an equivalence between the stanzas - a preference I'd learned from Newcastle days. At the same time, the tensions of the rhythmic structure, that sequence of deferred expectations, were recapitulated in the poem's closure.

A second principle of re-drafting recognisable through the long experience of failure is the importance of translations. In 1982, for example, I was clearly trying to translate the Old English poem The Seafarer, aware that whatever I was trying to do was in direct competition with Ezra Pound's wonderful re-creation of the same poem. My selfconscious workbook disaster, with its even more self-conscious title, The Vaiger [voyager], is given below:

Drafts [abandoned]

Often the one-treader waits in winter, bitter-minded, by
sea's anger, often each dawn complains his care to the
pleated winds that pester his exile. By
tide's torrent he tarries.... [etc.]

... No one now living knows that pledging;
the quick I knew once have quittance taken,
long leave-taking. Loving comrades
have knell under night-fall, no more vaunt
in the fray's vanguard. All is vanquished,
swept into silence. That is my story... [etc.]

In the first excerpt, I've retained the workbook lineation, where the text is set as prose. In the second excerpt, the original lineation suggests I had a clearer idea of the relevant form.

Even at the time I knew I was flogging an Anglo-Saxon dead horse, but the principles of these turgid lines could be put to the service of something more modest. I used a modification of them - aurally imagining the metrical structure of Old English, but without the inflectional endings which, in the original language, often act as metrical fillers - in a short poem about tying artificial fishing-flies which again appears in Time Signatures:

Fly Tying [opening lines]

There are the pleasures of the vice:
trying the temper of the hook,
running the silk in touching turns
down to the tail, a short stratagem
of pheasant tippet whipped well in;
then the body of dubbed wool....

A third principle is simply that of work. Poetic exercises may not be morally improving, but, as Eliot pointed out, they help to keep a writer rhythmically, metrically, and thematically alert, even if the results will almost always be abandoned. So, as a matter of practice, I write and reject frequently. I'm always astonished by poets who claim to have written half a dozen deathless pieces in a day; it's all I can do to manage a dozen in a year. And these days, perhaps even fewer. Or none.

Exercises sometimes contain surprises. I once had the unenviable task of invigilating in Manchester one heat- stricken afternoon where nothing moved except nerves. Since invigilators were enjoined to do nothing except `actively invigilate', I wondered how I could appear to be invigilating (actively) while doing something completely other. I decided to set myself an exercise. `Compose a poem in iambic dimeter.' Now iambic dimeter is often a jokey form, and can't be sustained (cf. Robin Skelton's `It's hard to write/ this kind of verse./ It makes you bite/ your nails and curse...'). `So' (continued my rubric) `make your poem long, and make it serious.'

I wandered the rows of sweating students, appearing to invigilate (actively). Lines came, went; and then a line stuck: `Down all each day'. A line of iambic dimeter, yes, but was there more? There was: `to make you know'. Where was this leading? `What could I say/ to think of you?'

Three hours later I had two stanzas committed to memory. I jotted them down on a piece of yellow manuscript paper (it had been a music exam), went home, and copied them into the workbook.

What interests me isn't the finished piece. It's the fact that I'd been experimenting with aspects of this form for several years beforehand, as the following abandoned drafts testify.

The first is from a poem called `The Language Exam', which was written, then lost, in 1987. It's drafted in the bob- and-wheel metre of Gawain and the Green Knight, where the wheel is composed in syllabics, the final two lines being quadri-syllabic. In another abandoned draft, there's an Audenesque attempt to say something unsuitable about love-and-gardens, where some rhythmic fragments are cast in iambic dimeter. These had clearly been speaking to me for a considerable time before I had anything like the occasion, or the ability, to finish the piece that was eventually called `Chosen' (Not Only I, Carcanet, 1996), whose opening is given in the third excerpt below:

Drafts [abandoned, 1987]

... I imagine the moment they met the questions,
the placid panic, the complacent fear
in the sweltering exam-room sodden with old nerves,
the tired traffic of the town outside
through wide-open windows withering them - and
          that fly
          crawling the glass
          it couldn't pass:
          as they, so I.

Drafts [abandoned, 1987-8]

For all I know
it could be true:
I know astringency,
the brain-machinery
working to disabuse
the soiled bruise
you would call love.
Is it enough
to make a life?
For all I know.
What happens now [etc.]

Chosen [opening, from Not Only I, 1996]

Down all each day
to make you know
what could I say
to think of you?
And what could I
ask in reply,
except for choice
to hear your voice
come close?

One significant thing about this poem was that it told a truth of which I'd never been conscious. Since then, I've been uncomfortably aware that poems, at least poems worth writing, may contain subconscious truths to which we have otherwise little access. I've also noticed that accessing these truths - they are mythic truths, but myth itself is invention about truth - tends to happen when the conscious mind is otherwise unoccupied, or distracted, or weakened. 'Flu is poetically potent, as is the edge of sleep, and the hangover. I've sometimes wondered if this is why poets have a fatally unjustified reputation as drinkers: it isn't the vodka, it's the brain-sick, compelling and structured violence of what the vodka leaves behind.

At some level, then, poems are magic riddles - but alas, there's no magic without hard work, just as there's no philosophy without food. `Grub first; then ethics.'

A fourth principle of re-drafting is the process of listening. As I grow older I know one of the black arts of writing is to listen to what the poem intends. A poem can't be willed or forced any more than can a waterfall. The act of listening may involve hours, days, years. You're intriguing with the overhearings of solitude, and the solitude often repeats itself in rhythmic fragments that carry sufficient thematic impetus to allow poems to form around them.

I was conscious of this when, over eight years between 1993 and 2001, I drafted Mass, a long `poem for voices' that finally appeared as part of the third collection, The Country of Perhaps (Carcanet, 2002).

The poem began as what appeared to be a single piece that would stand towards the end of Time Signatures. That wasn't what the poem wanted. It announced another poem, a piece which subsequently became the Kyrie, which presaged another piece, the consequential Gloria. Much later, I would write a Sanctus, a Stabat Mater, an Agnus Dei and a Benedictus. Thematically, the piece unravels the implications of the very first poem, which wouldn't allow itself to stand alone. Rhythmically, each of the sections of the Mass has a distinct but binding cadence.

What I was attempting to do was acknowledge, and then explore, the fact that rhythmic fragments may contain embedded thematic structures, and to find a poetic equivalent for the musical settings of the Eucharist with which I was familiar. This meant adopting a contrapuntal texture for the whole piece, so that the voices would follow each other, section by section, and finally, speak to each other inevitably in the final rhythmic blocks of the poem, which I heard as the concluding bars of an extended fugue:

From the final section of `Mass'

You know and I know
no mind has mass
without the pattern of god-likeness in us.

And this is love, called crucify
before it can be called
`you can be called
the resurrection and the life'.

Without the pattern of god-likeness in us,
out of the vagrancy of days and choice,
out of betrayal, out of malice,
after the Word has been heard unheard,
still
                                still
you know and I know
no mind has mass.
Who calls it sacrament?
You call it waste.
So who eats Christ's body
with its bitter taste

without the pattern of god-likeness in us
without the pattern of god-likeness
without the pattern

without release?

After the Word has been heard unheard

only
            only
from mercy, goodness
is time's weight and place.
Grant us its peace.

It's a long way from the gardening column of The Scarborough Mercury. I'm also reminded that the Mass took eight years to complete, or if you prefer, to abandon. It was an act of powerful listening, and of difficult patience.

Last, drafting may throw up (I use the phrase advisedly) marginalia, doodles, fragments that are light-hearted, whimsical, or in bad taste. In one of the workbooks, there's a piece - I can't call it a poem - called `366 Nights in a False Wig'. Another of these gobs of mischief is a casual reply to someone who'd invoked Hughes (`It takes a lunchtimeful of booze/ To like Ted Hughes...'). Throughout, there are doodled limericks and nonsense. Clearly the Sweet clone, the pedant, the drunk and the parrot all had a fondness for light verse.

One of these effusions overtook me at Athens airport at five in the morning:

If love is the true art of miming,
And limericks needs accurate rhyming,
Then why do both tend
To run in the end
Into rhetoric, lies, and very bad timing?

This process has begun to betray me, probably out of laziness, into listening for the `found poem'. One such piece I lifted without shame from the pages of a Greek phrase- book:

Translation exercises [workbook, unpublished fragment]

Help, help.
Is there a tobogganing run?
Someone has fallen into the sea.

What time is the next picture showing?
Please stop the bus, my child is feeling sick.
When will the policeman come?

My blood group is A positive.
Can we rent a motorboat?
The towels have run out.

I need to see a dentist (urgently).
Is there a reduction for a group?
The room is too hot.

I have two children.
Do you give lessons?
The toilet won't flush.

A fuse has blown.
What is this?
Can I rent some equipment?

I have two children.
Is there a paddling pool?
It is too expensive.

Please give me an injection.
I have two children.
Let me off here, please.

Drafting, then, seems to be an effortful intrigue with the overhearing of solitude. It's work, choice, accident and trouble, the inchoate becoming the only-occasionally- coherent. Though I've never forgotten serendipity, I have lost my ventriloquism but found a voice, however cracked. But in the middle of the night, as another rhythmic fragment starts to demand attention, I can still hear - beyond the threshold of hearing, almost as if it's in another and quieter country, and among the snows of yesteryear - I can still hear a bird rustling in a cage.

And then, far from speaking in effect...
Then it starts screaming.

The present essay was first delivered - in a very different version - in December 1998 as a talk, part of the Re Writing the Poem series (1998-2001), The Poetry Centre, University of Manchester.