Is poetry dangerous? The question hardly sounds worth answering, until you recall with a start that there have been times or places when it was. In the China of the Gang of Four, according to Kang Zhengguo in his Confessions, poetry was "the most dangerous career you could possibly choose" - or so his father, a man of letters, told him. It was like working as a circus acrobat without a net.

Poetry today is widely seen as harmless for other reasons: not because of free laws and institutions but because great poems, at first encounter, look triumphantly achieved and effortless. A pebble is smooth because of the slow action of tides, and a Shakespeare sonnet or an ode by John Keats is seldom seen as the fruit of a process of forethought and revision. "A line may take us hours, maybe," said W.B. Yeats in Adam's Curse, and the curse God laid on Adam was work. But a great poem, at first meeting, does not feel or look like a place where anything laborious ever happened.

That makes poetry tough to teach. Philip Larkin, who was a university librarian as well as a poet, used to insist that poems are to be read rather than studied; but since they are studied it matters how it is done. In a 1979 talk entitled "A Neglected Responsibility" he called on British libraries to acquire and preserve poetic manuscripts, hopeful that a corrected draft might persuade the young that a poem is the end of a deliberative process rather than a spontaneous act. A recent Poet Laureate, Andrew Motion, echoing Larkin, has told how an edition of Wilfred Owen's poems with a facsimile of a corrected draft of the sonnet Anthem for Doomed Youth changed his life as a teenager by vividly showing a poet in the act of revision. It was a salutary lesson, but the problem runs wider and deeper.

Poems are not nowadays seen as dangerous because their truth-content is little regarded. It is not widely believed that it matters whether a poem is true or false: it is only a poem. That is because many confuse a sense of truth with the notion that it would involve the assumption that only true propositions are worth considering, though nobody makes that assumption in reading prose or in listening to people talk. We often listen eagerly to people getting things wrong, in public and in private life, and delight in stories about the U-turns of party leaders. Somehow that does not work with a page of verse. Dante's views, or Milton's, of divine providence, or Shakespeare's of the divine right of monarchs, are widely dismissed as little better than embarrassing, and I suspect poetic meaning is not taken seriously because there is a silent assumption that to be interesting it would have to be believed.

ÖйúAƬ

There are cheering counter-instances of worldly success. A.E. Housman's A Shropshire Lad appeared in 1896, and in the last decades of his life, which ended in 1936, it is said to have sold some 16,000 copies a year. T.S. Eliot and Dylan Thomas did not have to wait long for public acclaim. W.H. Auden once rejoiced to hear that a prostitute in a prison near his home in Greenwich Village had quoted his poem First Things First as she marched to the communal bathhouse: "Thousands have lived without love, not one without water." That is the sort of thing poets like to hear.

But do they? Poetry in recent generations has dramatically shrunk. Explanations have been offered for the shrinkage, and one of the earliest and best was Macaulay's in his 1825 essay on Milton. It was the rise of science, he believed, that made it inescapable, and the triumph of the Enlightenment. "The vocabulary of an enlightened society is philosophical," he wrote sadly, contemplating the rise of the natural sciences, and Milton could only justify the ways of God in a great epic because language in the 17th century was still in a rude state; Lucretius in an earlier millennium could expound a theory of atoms in verse because prose barely rivalled verse as an expository tool. Such achievements now seem out of reach. It is rather as if a poet today were to write an epic about the global economy, the cosmic Big Bang or the crisis in the Middle East, and it is enough to try to imagine such a thing to realise its stark impossibility. Robert Bridges, laureate to George V, tried something like it in the 1920s with The Testament of Beauty, which deals with (among other things) the moral life under the shadow of evolutionary theories - an act of spectacular daring dedicated to the King and without successors. But there are no great philosophical poems now, no epics, almost no verse tales.

ÖйúAƬ

There may after all be a connection between preserving poetic manuscripts and reviving the status of the art. Manuscripts have been called the engine room of the creative act, and an age that values the self-expressive might easily interest itself in that. Or perhaps too easily. The catch is that self-expression is widely seen as spontaneous, and the shock of discovering it is not can be severe and even fatal. I remember expounding Dylan Thomas' Fern Hill to a highly intelligent sixth form, explaining how assonance differs from rhyme, among other technicalities, and what difference that difference makes. The reception was ambiguous. "The trouble is," one bright teenager said, "it takes all the spontaneity out of it." The task now, and a formidable one, is to persuade people there is no spontaneity to take. Writing verse is not like playing with a rubber duck, and the myth of poetic spontaneity only promotes contempt. No one is likely to respect an utterance of any kind - a law of physics, an oration, a public policy - that is not born of reflection. "It must puzzle us to know what thinking is", Lionel Trilling once said, "if Shakespeare and Dante did not do it." The myth of poetic spontaneity implies a refusal to look hard at what great poets have painfully achieved. Nor is revision confined to drafts. Much that is spoken and written was revised in the mind before it was spoken or written at all.

Revision, what is more, can be practised on others. Wordsworth remarks in a letter of 1843 that he had spent a lifetime revising great poems as an exercise. None of the exercises has survived, but any attentive reader of Shakespeare or Milton might think of improvements as he read. Revision is a friendly and loving act. Eliot gratefully heeded John Hayward's advice to improve the Four Quartets, and though students can be shocked and startled to be asked to improve a famous poem they seem to settle happily to the task.

Nor do more recent poets mind. When Stephen Spender was teaching at the University of London in the 1960s he was so captivated by a class for aspiring poets that he thought of submitting his own poems, and did. "We could all do with it," I recall his saying, with enthusiasm. "Hard deeds, the body's pains," said John Donne in his third satire, mindful of the sheer sweat of composition: "Hard knowledge too/The mind's endeavours reach." The poet, he believed, has a moral duty to get it right: "Keep the truth which thou hast found." You discover what you mean progressively, by saying and resaying, and with much sweat and a little luck you end with something that looks good enough to have been written by someone else.

Revision can disimprove, and a poet can bother to the point of being bothersome. Auden's publisher used to tell how hard it was to choke a new edition out of him when he was endlessly intent on revising; Wordsworth spent half a lifetime rewriting The Prelude without improving it; Milton dictated Paradise Lost when blind, but the surviving manuscript of Book One is corrected and some of the corrections may be his. Perhaps he had it read back to him, with interruptions. Shakespeare seldom even hints at creative process, which he would have thought no business of any reader or audience; but his 77th sonnet recommends keeping a commonplace book laid out with page headings - a practice usual enough in his age and one he may have used himself:

Look what thy memory cannot contain

Commit to these waste blanks, and thou shalt find

Those children nursed, delivered from thy brain,

To take a new acquaintance of thy mind.

Prose writers did it too, and John Aubrey tells how Hobbes scribbled in a notebook on the quays of Paris during the Civil War when he was writing Leviathan. It is advice to profit from. Anyone who does it knows what Shakespeare meant by new acquaintance, and only those who do.

Second thoughts permeate authorship, even writing a letter. In The Rape of Lucrece Shakespeare has his heroine writing to her husband to tell him Tarquin has raped her. Dear Collatinus ... But the task is inherently delicate, and she gets it wrong:

What wit sets down is blotted straight with will.

This is too curious good, this blunt and ill.

One knows the feeling. Lucrece's problem is every author's problem, and you do not have to be raped to know about it. Spontaneity does not come into it, if only because (as Samuel Butler once said) everything written has a sayee as well as a sayer - whether spouse, friend, publisher or solitary reader. Wilfred Owen's corrected sonnet was a road to Damascus for one future poet, and weak drafts are nothing to be ashamed of. Second thoughts, like third or fourth, are commonly better. You can only do it at all by doing it badly, and a poem as it emerges is about as spontaneous as a Ming vase. "It gave me a devil of a lot of trouble to get into verse the poems that I am going to read", Yeats remarked in 1932, introducing some of his poems, "and that is why I will not read them as if they were prose."

ÖйúAƬ

The news has hardly travelled, however, and beginners can agonise needlessly about an inadequacy that is not there. Part of the problem is a fear of discussing technicalities. Fantasies about creativity in adolescence see it as a single act, and talk about metre and syntax is easily felt to be alien. That is a monumental misapprehension. Poets talk about metre endlessly - "if it's not money it's metre" - and the English tradition is pre-eminent in poets who are critics and critics who are poets. They include Ben Jonson, Dryden, Samuel Johnson, Coleridge and Eliot. If we murder to dissect, which I doubt, the long record of poet-critics suggests otherwise. Some people are too easily alarmed. Charles Lamb tells how he was so shocked at seeing a corrected Milton manuscript in a Cambridge college that he swore never again to enter a poet's workshop, which seems odd. After all, he had written poems himself. Johnson, a more experienced poet, dissects till the cows come home, and in the Lives of the Poets he commends Pope's practice of putting down first thoughts in the first words he thought of, later to "amplify, decorate, rectify and refine". Creativity is an exciting word, but it is a wildly misleading one if it encourages anyone to think a poet is like God in the Book of Genesis. The God of Scripture was not a reviser, but his creatures are.

ÖйúAƬ

Another failure is to know poems only in silence. Other cultures speak them aloud. Years ago, on a visit to Iran, I sat on the floor of a private home watching a game of dominoes played by squatting figures in a language I did not understand. Then one man straightened and began an incantation; his audience heard it with satisfaction, considered it in silence and resumed the game. It was a poem by Hafiz, I was told, a contemporary of Chaucer who died in 1389. That is a scene hard to imagine in any English-speaking country, apart from limericks, for we resent the sound of verse. As a social interdiction I reluctantly observe it, most of the time. But poets put things better than other people - that is what they are for - and it is a pity not to use them.

In teaching, oddly enough, it works the other way. To know a poem by heart is a show-stopper in a lecture, and Auden used to raise his eyes to heaven as he spoke, perhaps to make you wonder where it all came from. It excited awe, which is hard to understand. It is easy, after all, to learn a poem and often hard to forget one. In his letters Macaulay tells how he once thought his way through the whole of Paradise Lost one dark and sleepless night on the open deck of a ship crossing the Irish Sea: "I have never enjoyed it so much." That is no doubt a mark too high to shoot at, but it shows what memory can do, and any poem is likely to be better in the head or on the tongue than on the page. The late Graham Hough once found a class unresponsive to a poem by Robert Browning, so he read it to them, with startling results: "If you're going to make it sound like that ... ". They had never heard a poem, apparently, and knew nothing of the heartbeat verse can have. Rhythm can inspire. Housman once told G.M. Young it was the beat of Samuel Johnson's Short Song of Congratulation that prompted him to write A Shropshire Lad:

Wealth, Sir John, was made to wander,

Let it wander as it will.

So a poet can begin with a beat that echoes in his head, much as composers do.

A monopoly of silent reading is the more surprising when you consider that everybody has songs in the head, from the national anthem to Tin Pan Alley tunes, and everybody knows they are there to sing. So the educational neglect of metre must be an affectation. It is easy to pretend you do not care, difficult not to care. Popular ballads are heavily crafted, after all, and need to be. "I love you" in 32 bars is the classic formula: the first eight bars repeated, then a middle eight, then the first eight repeated. When I lived in New York there were said to be people who made a living writing middle eights, although I never met one. Nobody seems to want to talk about this, and an interest in metrics is widely assumed to be the business of pedants and beyond what ordinary people care about. They care about it more than anybody.

If we are deaf to metre, the deafness is recent. In a high-tech culture you expect solutions to be instantaneous, like switching on a light; it is not like lighting a taper or trimming a lamp. Talking about stanza-forms or middle eights is like trimming a lamp, and by the 1960s a lot of people were noisily despising it, in poetry and elsewhere. Now the tempo of debate has changed, in politics and in the arts. Douglas Adams, author of The Hitchhiker's Guide to the Galaxy, whom I remember as an unstudious student of English more than 30 years ago, wrote a satire shortly before his untimely death about a sage who was asked for the secret of the universe and solemnly answered "Forty-two". There is still a tendency to suppose that knowledge is a light switch. If you do not have a quick answer, you do not have an answer. Sir David Hare of the National Theatre has told how at Cambridge he once asked a lecturer how he knew some books were better than others, and lost faith in literary studies when the answer failed to satisfy him. To know, it was assumed, was to give an instant account. Paris was the epicentre of that brand of nonsense, and I used to annoy audiences there by asking them what a banana tastes like. In the shadow of Jean-Paul Sartre you were supposed to distrust the incommunicable and deny to the intuitive any right of place or cause to be.

Poets seek an audience for poetry, and the problem now extends far beyond the preservation of manuscripts. It is time to get real. If people are to be tempted to read, they need to be told something worth knowing, or reminded of what they may have forgotten, and here the tattered tradition of Modernism in the Pound-and-Eliot style does not help. Poets of that school were at best unhelpful in the search for knowledge and at worst destructive. They wrote as if truth did not matter or, if it did, could only be found (as Eliot once wrote in his Harvard thesis) at the end of a rainbow, like a pot of gold. He was not interested in intuitive knowledge, presumably, like the two-times table. It all sounded too easy. Some like Pound, Yeats and early Auden openly flirted with tyranny. Eliot made knowledge sound as arduous as learning Sanskrit or wearing a hair shirt, and in Thoughts after Lambeth in 1931 he announced that any secular society must fail; the best you could do was to "redeem the time" and seek to do something useful while you wait. A quarter of a century later one of the last giants of Modernism, Samuel Beckett, wrote a play called Waiting for Godot where Godot never came. All that may have been music to the ears of the great dictators and to terror groups. It is a world away from Wordsworth's "We must be free or die who speak the tongue that Shakespeare spake", but by 1803, when he wrote that sonnet, he believed in what soon came to be casually derided as the Whig interpretation of history. What poet would proclaim it now?

A world that abandons all sense of moral knowledge is one where poets humble themselves to the minor role of self-definition, and it is hard to see what indispensable role a poet could then lay claim to or possess. Reading becomes an attempt at self-discovery for those uncertain whether there is any self worth discovering; to look knowing feels like a better option than to know. So you are asked to entertain beliefs, at best, as in Larkin's little poem Water, which invites you not to accept a religion but rather to smile at the happy conceit and pass on. But belief constructed on nothing but a whim goes nowhere. It is to trifle with knowledge. In his last book, Truth and Truthfulness: An Essay in Genealogy, Bernard Williams bravely confronted the modern dilemma of respecting truthfulness without acknowledging truth, and left it looking problematical, which it is. It is a dilemma still to be resolved.

The solution for poets and philosophers alike is to give up on moral scepticism and accept that virtue is more than opinion. That means a lot of giving up, and it will not be easy. It means courting danger. There is no comfort in moral certainty, after all, only hard deeds and a demanding sense of duty, of work waiting to be done. A century ago Rainer Maria Rilke told a story about a French poet expiring in a hospital who heard a nurse mispronounce a word and corrected her with his dying breath. "He was a poet, and hated the approximate." That is to hold the poet in bounden duty to get things right, which he is. But to accept that duty, and that responsibility, the poet would have to recover a sense of what right is.

ÖйúAƬ