It is a convenient and pleasing Romantic myth that the true work of art springs full-blown from the unconscious mind. Revision comes from the conscious intellect or will, and this, as Wordsworth wrote, “is the very littleness of life,…relapses from the one interior life that lives in all things.”1 Some years ago, a novelist—Muriel Spark, I believe—was asked how she was able to write so many books in such a short space of time. She replied, “I write very fast and I never correct.” This is the ideal. Few writers are so fortunate. Most revise and, as they do so, create more problems than they resolve.

One of Balzac’s most interesting tales, Le Chef d’oeuvre inconnu (The Unknown Masterpiece), deals imaginatively and succinctly with revision. It was a subject close to the author’s heart: his books generally went through several versions before and after publication. However, no work of his was more completely or profoundly rewritten than Le Chef d’oeuvre inconnu.

The scene is laid in Paris in 1612, and the central figure is an invention of Balzac’s, a demonic personality who might have stepped out of the fantastic tales of E.T.A. Hoffmann: the old Frenhofer, the greatest painter of the age (all of Balzac’s important characters possess their qualities in the superlative degree, and no moderately talented artist could play a significant role in his work—even Wenceslas Steinbock in La Cousine Bette, when he loses his talent, becomes obsessively and spectacularly incapable and the hopelessly mediocre Pierre Grassou sees his work sold under the names of Rembrandt, Rubens, and Titian, becomes the favorite painter of the bourgeoisie, and enters the Academy). The two other painters in the tale are historical: the young Nicholas Poussin, just starting out as an artist, visits the atelier of the already established Franz Porbus, and meets Frenhofer there. For ten years Frenhofer has been working on one painting, a life-size portrait of a nude woman lying on a velvet couch—“but what are ten years,” he says, “when it is a question of wrestling with Nature? We do not know how long it took Lord Pygmalion to make the only statue that walked.”

Frenhofer will show no one the picture: to finish it, he says, he needs a model of absolutely perfect beauty. Poussin has such a mistress, the young and modest Gillette, who adores him. He persuades her with difficulty to pose in the nude for Frenhofer, who will in return allow him to view the unknown masterpiece. Alone in his studio with Gillette, Frenhofer compares his painting to the living form of the girl, and decides that it is finished, more beautiful than reality.

He lets Poussin and Porbus into the studio and places them before the work: they see only a

confused mass of colors contained within a multitude of bizarre lines that make up a wall of paint…. Moving closer, they saw, in one corner of the canvas, a naked foot that came out of this chaos of colors, tones, indecisive nuances, a sort of fog without form; but a delightful foot, a living foot. They stood petrified with admiration before this fragment that had escaped from an incredible, long and progressive destruction…. “There is a woman underneath,” cried Porbus, pointing out to Poussin the finesse of the superposition of colors with which the old painter had successively charged all the parts of his figure in trying to make it perfect…. “There,” said Porbus, touching the canvas, “ends our art on earth.”

Lost in admiration of his own work, Frenhofer does not comprehend that his ten years of revision have destroyed his painting, and Poussin loses his mistress, for Gillette cannot forgive his having sacrificed her deeply felt modesty simply to see a picture.

This moral tale of the terrifying effects of revision underwent wholesale revision after its 1831 printing in a periodical, L’Artiste, and its reappearance with some corrections in book form the same year. Six years later, in 1837, Balzac republished it, considerably enlarged and with a different ending as part of the seventeenth volume of his Etudes philosophiques. In this version, definitive except for some retouching in Balzac’s own copy of La Comédie humaine, the old painter, observing the reaction of his fellow artists, realizes the disaster, and throws the two younger painters out in a blind rage. That night he burns all his pictures and dies mysteriously.

The additions of 1837 are largely discussions of the theory of painting, in which Balzac ascribes to his seventeenth-century artists the ideas current in the 1830s: the supremacy of the colorist over the draftsman, for example. The anachronism is compounded in the reader’s mind by the development of art since Balzac’s day, by the suspicion that Frenhofer’s superposition of colors, his multitude of bizarre lines, his chaos of tones and indecisive nuances, might be found more sympathetic today than the banal life-size nude on a velvet couch. It is more significant, however, that the isolated foot that comes out of this chaos would have had a charm already in Balzac’s time precisely because it is a fragment, and Balzac’s description brings out this charm magnificently:


This foot appeared there like the torso of some Venus in Parian marble risen from the debris of a city destroyed by fire.

Perhaps the most extraordinary textual change made in 1837 is an apparently small one. Frenhofer’s refusal to display his picture to anyone else is a parallel to Gillette’s reluctance to pose in the nude for anyone except her lover. Before yielding he expresses his resistance with passion:

The work I keep under lock and key is an exception in our art; it is not a canvas, it’s a woman! a woman with whom I weep, I laugh, I talk and think. Do you want me to abandon ten years’ happiness as one takes off a coat? To cease in a single moment, being father, lover and God? This woman is not a creation, but a creature.

This is the version of 1831. In 1837 the final sentence was altered:

This woman is not a creature, but a creation.

It is wonderful to be able to reverse the terms in this way, and the sentence still makes sense with no change of context. It is clear that for this to happen, the meaning of the words have shifted but then, as Lichtenberg once wrote, whoever decreed that a word must have a fixed meaning?

Placed so near to “creation” in both versions “creative” means not only a living being but one created, and Frenhofer’s admission that he enjoyed playing God brings us to the first woman, Eve. A “creature” implies a living being, and “creation” only something made. What is imposed by the contrast is woman against portrait, the experience of life against the object. The two are fused ambiguously in both versions of this passage, but their opposition is the theme of The Unknown Masterpiece. Poussin loses his mistress for the sake of the portrait; Frenhofer has made his portrait the substitute for a woman, and his ten years of happiness destroy his work.

The two versions can act only as a paradox—or different paradoxes. To make sense of both, the meaning of “this woman” must shift. In “not a creation, but a creature,” we have “this portrait of a woman is alive”; in “not a creature, but a creation,” it changes to “this woman is something I have made.” What is disconcerting about the revision is the alteration of values. In the 1831 version, the living being takes precedence over the made object. By 1837 Balzac’s hubris has increased, and the work of art is nobler than the woman. (This change has its source in Balzac’s own temperament: after consoling a friend for the death of his mother, Balzac is said to have continued, “And now let us talk about something important: should I make the heroine of my new novel get married?”)

The change from 1831 to 1837 is, when you come to think of it, a parallel to the story. Both Frenhofer and Poussin allow the work of art to take precedence over the living being: Eve becomes not a creature, but a thing; the work becomes a fetish. The variant of 1837 reveals a moral deterioration of the author that reflects the tale, as if Balzac were corrupted by his subject (it is significant that the earlier version is not only more humane but more directly effective, the later version more subtly insidious).

Another variant reveals the same process. Poussin endeavors to persuade Gillette to pose nude for Frenhofer, and assures that her modesty will not be violated. In the periodical version, he says:

Il ne verra pas la femme en toi, il verra la beauté: tu es parfaite! (He will not see the woman in you, he will see beauty: you are perfect.)

In the first edition in book form a month later, we find:

Il ne pourra voir que la femme en toi. Tu es si parfaite! (He will only be able to see the woman in you. You are so perfect.)

In the first version “woman” is physical, sexual, and vulnerable: the woman in Gillette will be protected from the gaze of Frenhofer. In the second, woman has become a concept, abstract and general. This suggests the way revision in Romantic art moves away from direct experience to a mediated reflection.

In the case of The Unknown Masterpiece, however, there is no point in judging one version superior to the other. It is clear that a perception of the richness of meaning in these passages depends on a comparison of the different states of the text—the meanings may be implicit in each individual version, but they are more easily revealed when one version is superimposed over the other. In this sense, Frenhofer’s “masterpiece” is less an allegory of the dangers of revision than an image of a critical edition with all the variant readings displayed to the reader—above all when we reflect that we would probably have preferred the magical appearance of Frenhofer’s disaster to the more banal work he thought he had painted and that Porbus and Poussin all too reasonably expected to see.


Balzac’s description of the picture does not correspond to the ordinary process of revision, in which the difficulties are smoothed away and the original awkwardness covered over. In the chaos of colors, tones, and decisive nuances we seem to see all of the different versions superimposed. The different variant states of The Unknown Masterpiece constitute a more profound and original work than any individually published text.


It is an odd and even somewhat perverse experience to read a novel or a poem in what is called a critical edition—that is, an edition which lays out all the stages that the work went through from manuscript through the successive editions, and exhibits all the variants. Out attention is constantly and abruptly halted in mid-progress to consider the change of a comma to a semicolon, the addition of a paragraph, the excision of a phrase. The reader has the illusion of sitting in the seat of the author: he can inspect, and choose between, alternatives; regret lost opportunities, evaluate each improvement. At every step, he is distracted from the text by another text.

This kind of edition is sometimes called a “variorum” edition, but this is—or used to be—incorrect: a “variorum” edition until recently was one that displayed all the different notes and commentaries made by the various editors of a classical text. It represented the history of the critical tradition inspired by the work. There is no point in attacking what is by now a well-entrenched solecism (sanctioned by The Variorum Edition of the Poems of W.B. Yeats and Byron’s Don Juan, a Variorum Edition), but the pedantry is useful: not only the terminology but the aspect and the function of a critical edition have changed radically over the centuries.

In the fifteenth and sixteenth centuries the notes to an old work were often printed not only at the foot of the page, but at the top and the sides as well: the text of Virgil’s Aeneid, Petrarch’s Canzoniere, or the Gospels, for example, was published in large type and surrounded by a commentary in small print drawn from many sources. The work took its life and significance from the circumscribing glosses: the commentary was literally the context of the literary document.

The variant readings were most often incorporated with the commentary, the purpose of which was generally to explain how the new edition was more correct than all previous ones: the editor’s emendations were argued for and justified. Correcting the text of a Greek or Latin author became a major academic industry by the eighteenth century, and classical studies seemed almost to be reduced and restricted to putting out improved editions, attempts to establish a definitive text. Methods of correcting and editing were made systematic in the nineteenth century, although individual voices like A.E. Housman’s were raised in protest against the thoughtlessly mechanical techniques of correction that had evolved. Variant readings and explanatory commentary were separated into two kinds of footnotes, often printed separately and with a different typeface.

The principle of the variorum edition was gradually extended from classical authors to writers in the more vulgar tongues, beginning with Shakespeare and Dante. The problems of getting an acceptable text of Shakespeare, of correcting the multitudinous faults, obvious and not so obvious, of contemporary folio and quarto printings, made it appear that the problems of editing him were similar to those met in editing Sophocles or Terence. This was soon found to be an illusion, and correcting Shakespeare became a highly individual and specialized enterprise.

Meanwhile the interplay between text and footnotes began to fade away. By the twentieth century explanatory footnotes had often ceased to be footnotes and were printed separately at the end of the chapter or the book. (In McKerrow’s 1904–1910 edition of Thomas Nashe, often held up as the very model of a modern critical edition, variant readings are printed as footnotes, but the commentary is reserved for the final volumes.) Many publishers have insisted on the format which relegates the commentary to the end in a separate section, and have cited economy of typesetting as a justification. However, publishers are not motivated entirely by greed, in spite of appearances; the reason for separating the notes from the text is based much more on philosophical and aesthetic considerations, since publishers today have insisted on printing this way even more tenaciously, although computers have made the difference in cost negligible.

There is an obvious preference for a clean page of text, unburdened by explanation or any other supplementary matter, although it is easy to see that no text from the past can stand on its own and be enjoyed without misapprehension by a modern reader. In the new edition of Balzac, both variants and commentary are separately printed (the variants in italics) at the end of each volume; in the critical edition of Byron now coming out (edited by Jerome J. McGann), variants are printed on the page but the commentary follows at the end, and the explanatory notes of Byron himself are mixed in with the commentary of the editor, a method that certainly does violence to the intentions of the author, and the way he expected his work to be presented.

A clean page of text invites us to concentrate on the work. On the other hand, it is considerably more distracting to be obliged continually to turn to the back of the book for the information necessary to comprehension. The decision to print a page with a pure text uncontaminated by notes is, in the end, the publisher’s answer to an ontological question: What is a work of literature? It appears to be a well-defined aesthetic object that can survive the passage of time because it can be detached from the culture and the age that produced it, and still remain intelligible; it can be removed as well from the successive interpretations of generations of readers. This may not be a satisfactory description of a work of literature, but we may assume that publishers believe that it will sell.

This concept of a pure, definitive text is, however, the stumbling block for many critical editions of modern works. Moreover, it betrays a lack of awareness of the change in the nature of a critical edition. To edit a classical author—Plato, Sophocles, Seneca, or whoever—is to seek to construct a single text free of the mistakes made by successive copyists. No autograph has come down to us from classical times, and our knowledge of the literature is dependent on the copies made many centuries later; we are often faced with different versions, all of them corrupt in one way or another. The role of the editor is to purge the text of the corruptions and to reduce the different versions to a single one.

The purpose of a critical edition of a modern work, on the other hand, is more often than not to multiply versions. Its reason for existing is to show us the way a novel or a poem developed from the sketches and drafts to publication, to reveal the improvement or deterioration through successive editions, and to allow us to perceive how its reception has formed our present understanding. Whereas the classical editor constructs a single, often fictive, text, the editor of Balzac or Wordsworth breaks down the single text to which we have become accustomed into a chronological series of texts, some fragmentary, and some complete. A history of the developing commentary on the work, demanded by the old sense of variorum edition, enables us to appreciate how any single version can change its significance according to the perspective of the readers.

Some works, like The Unknown Masterpiece of Balzac, are enriched by a critical edition, and there are still others that are not fully comprehensible without one. The first modern book of this kind is Montaigne’s Essays, a precursor of later work in this sense as in so many others. In his copy of the second edition of 1588, already a considerable enlargement of the first, Montaigne filled the margins with new matter, sometimes lengthy developments or even extensive observations on subjects distantly related to the text. It has now become standard to indicate the various layers of composition: at many places, the indication is necessary to follow the argument, which would otherwise be totally obscured by the interpolations. The discourse is often clarified, too, by one’s knowing that a given paragraph is a comment by Montaigne written many years after the original essay. “I am myself the matter of my book,” he wrote; and with a volume written over more than two decades, we need to know which of the ever-fluctuating and diverse Montaignes he is portraying: timid beginner or confident author.

“I also delight in the company of beautiful, well-bred women,” wrote Montaigne (belles et honnestes femmes)—but “beautiful” is added in the margin, and it is legitimate to wonder whether Montaigne became more susceptible to beauty as he got older, or simply more frank about his preferences. This kind of speculation is, in fact, the very stuff of his own thought. In 1588, he wrote, “The public interest requires betrayal and lying”; sometime later, he added “and massacre.” The fact of the addition reveals what has been happening to France with the progressive acceptance of the horrors of civil war. Our knowledge that “massacre” is a marginal afterthought gives greater ironic intensity to Montaigne’s comment: “Let us resign that commission to those who are more obedient, more supple.”

The problem of representing these different strata in Montaigne is not a great one since he almost never excised an earlier passage. He built up his essays by a series of deposits. A few words were altered, but, for the most part, phrases, sentences, paragraphs, and whole pages were inserted as they occurred to Montaigne with very little regard for the original structure of each essay. His book was, deliberately, consciously, an image of his apparent inconsequence. The unity that we find in the Essays is not the result of a simple plan, and does not come from the orderly exposition of ideas; it lies in the movement of Montaigne’s thought, the way he glances off one subject to another, the means he finds of indicating his meanings without making them explicit. For this reason, decomposing his work into the various layers renders it more, not less, readable, increases its power, directness, and effect.

The editor of the new edition of Byron, Jerome J. McGann, has written persuasively about the conflicting claims of different versions,2 but he seems to envisage the printing of more than one version almost as a measure of desperation. “For the editor of late modern works especially,” he writes, “the first and crucial problem is not how to discover corruptions, but how to distinguish and finally choose between textual versions.”3 In many cases—though by no means in all or even a majority—choosing one version over all the others would mean a waste of the editor’s lengthy, painstaking, and elaborate labors. It often turns the various readings into an unreadable labyrinth in which what is significant is swamped by a mass of niggling detail. The crucial problem, it seems to me, is to decide when different versions are sufficiently interesting to demand full treatment, and to find a way of presenting these in a readable fashion.

This is important, above all, for the poem in which revisions have been so considerable that it would be difficult to maintain that the first and last versions are the same work. These cases may be rare, but they do not deserve to be dressed in the straitjacket of some editorial system. McGann has recounted his problem with one such poem of Byron’s, The Giaour, a Fragment of a Turkish Tale. This remarkable work, a romantic tale based on Byron’s rescue of a girl who was to be tied in a sack and thrown into the sea at the order of the Turkish governor of Athens, is less a fragment than a series of “disjointed” fragments, as Byron himself called it in the “advertisement” (the introduction). It began as a manuscript of 344 lines, and grew to 684 lines by the time it was first published; in the seventh edition, made only six months later, it had reached 1,334 lines, after which Byron ceased to add to the fragments.

It is clear that a poem of 344 lines is not the same work as a poem of 1,334 lines, even if (as is not the case) all the words of the manuscript reappear intact in the seventh edition. Moreover, McGann informs us that the punctuation is radically different in the first and seventh editions, the first being punctuated rhetorically—that is, following the vocal pauses made by a speaker, and the seventh more strictly according to the syntax; the first edition is therefore closer to the way Byron himself punctuated, and reveals his sense of the rhythm, while the seventh conforms to the way he felt his writing should be respectably presented to the general public. A publication that represents The Giaour only by the seventh edition while allowing the reader to excavate the other versions from the tiny print at the bottom of the page and in the back of the book, does not live up to the claim of its title, The Complete Poetical Works.

A printing on facing pages of the first and seventh editions, along with simple notes on the difference between the original manuscript and the first printing and an indication of the various expansions that appear in the intermediate editions between one and seven would have made it possible not only to reconstitute the various versions but to read them with pleasure. McGann did envisage a printing on facing pages, but largely for the purpose of representing the different systems of punctuation; he would have given with the first edition the added passages as they appeared in the manuscripts that Byron sent to the press, without the editorial punctuation: he claims, somewhat ruefully it seems to me, that “the exigencies of the whole edition obviated this possibility.”4 It is not clear why.

Only a few poems of Byron ask for such special treatment, and since the publisher is demanding almost sixty-four dollars for the paperback and one hundred twenty-five for the hardback volume, the reader interested in Byron could be granted more consideration. As things stand, it is easier to become acquainted with many aspects of the earlier versions from the edition published by Murray in 1900 and edited by E.H. Coleridge. At least there, the variants are printed as verse and not as a prose jumble.

We find ourselves here facing two bibliographical fetishes. The first is the question of “copy text” (i.e., the text that is taken as the basic one to be edited and published), a subject upon which bibliographers have spilt more ink than on any other—for many of them, indeed, choosing a copy text is their reason for existing. The truth is that for most books, the vast majority, in fact, the choice of copy text does not make much difference (after the first printing, the novels of Flaubert and the poems of T.S. Eliot, for example, underwent no revision). For a small number the conflicting claims of different copy texts are so strong that all should be represented: whether by variant readings or by the complete printing of all versions. Shakespeare’s King Lear and Hamlet and the poetical works of Pierre de Ronsard are among the most famous examples that contain such problems, and we are gradually becoming aware that the first half of the nineteenth century presents us with several others.

The other fetish is the belief that the author’s final version has the greatest authority and is therefore privileged. Here aesthetics has become oddly confused with law, and critical evaluation becomes an affirmation of property rights. An editor’s judgment seems often to be supplanted by the traditional rules of copyright, and an author who has been dead for more than a century is still allowed to decide which version of his work we shall be permitted to read. In most cases, this fetish does no harm, as authors are often no worse than any one else at correcting their works. There are, however, cases in which successive versions do not produce greater polish or enrichment, but tend to weaken and even to betray the original conception. That so many examples of this progressive deterioration are found in the early nineteenth century suggests that we are not dealing with the psychological difficulties of individual authors but with a problem of style. In recent years, the extension of the study of revisions and variants to this period has deepened our apprehension of Romantic art.


Romantic artists die young, or else lose their genius and sink into insanity, drug addiction, or respectability. This is a principle of Romantic ideology, not a fact of life, and there are many exceptions; nevertheless it is astonishing how many writers, artists, and composers of the first half of the nineteenth century conformed faithfully to the model. Dying young, of course, is not the prerogative only of Romantic artists: the lives of Du Bellay, Pascal, Marlowe, and Mozart were cut sadly short like those of Schubert, Géricault, Novalis, Kleist, Keats, Shelley, and Byron. Moreover, renunciation of a brilliant career was not unknown before 1800; the most interesting examples are those of Congreve, the finest playwright of his generation, who retired to become a gentleman, and Mariotto Albertinelli, the contemporary of Raphael, who after painting a masterpiece like the Visitation, now in the Uffizi, retired to become a cook.

What is unprecedented about the Romantic movement is the number of important figures who follow a few spectacular years with several decades of pedestrian work lit only by intermittent flashes of earlier genius. These include the major writers of Germany, France, and England who began to work in the very last years of the eighteenth century: Friedrich Hölderlin, Friedrich Schlegel, François René de Chateaubriand, Etienne Pivert de Senancour, Samuel Taylor Coleridge, and William Wordsworth. A little later, the generation of composers usually called Romantic, born around 1810, has a similar history: Chopin died at thirty-nine, Liszt completed almost all his major works by the time he was forty, Schumann by the age of thirty-one, Mendelssohn before he was twenty.

The most important achievements of Wordsworth, Coleridge, Schlegel, and Senancour were completed between 1797 and 1807; in 1805 Hölderlin was overwhelmed by schizophrenia, and spent the next thirty-eight years passively writing unremarkable verse, after a decade of work so powerful and innovative that it was fully appreciated only a century later. These years between 1797 and 1805 were filled with the most daring experiments in literature. They were a time of despair after the failure of the French Revolution and of half-ironic, half-serious hope for a new spiritual rebirth, a transformation of the world through art. Senancour produced the Reveries on the Primitive Nature of Man and Obermann. The latter inspired Franz Liszt and Matthew Arnold, and was called “the Bible of Romanticism” by Sainte-Beuve. In 1798, Schlegel founded the Atheneum, a literary review, with his brother August Wilhelm, and published there the eccentric, richly paradoxical Fragments which constituted a manifesto of German Romanticism and laid a foundation for most of the basic ideas of modernism—its experimental, progressive nature, the attack on established classical forms, a literature of visionary imagination alongside the portrayal of even the most repellent aspects of modern life. (Schlegel suppressed these fragments, his most remarkable achievement, when he later collected his works.) These were also the years of the Lyrical Ballads of Coleridge and Wordsworth, of “Kubla Khan,” of The Prelude, and of almost everything else that we still read today of their poetry. Even the magnificent first book of Wordsworth’s The Excursion was drafted at this time.

It was understandably impossible to sustain the creative energy of these brief years. To a large extent this was a literature of opposition. Above all, it was drawn directly from memories of adolescence, and as these memories receded into the last their evocation became more and more artificial, or else the writer found himself with a fully developed manner and no content. After 1805 Wordsworth drew with increasing difficulty upon the feelings that had provided the material for his finest work, and turned to producing a long series of sonnets in favor of capital punishment. The impoverishment was as evident to contemporaries as to posterity.

Chateaubriand alone of these writers was able to continue to produce work in later years comparable to his first masterpieces. His case is exemplary: what he did was to continue to publish and develop his earliest work. He returned from America aged twenty-four carrying an enormous manuscript (he claimed improbably that it later saved his life during the wars against the young revolutionary government of France by stopping a bullet). He extracted one book after another from this, starting with the Essay on Revolutions, The Genius of Christianity, and The Voyage to America. The successive revisions weakened what we have been able to learn of the original: in The Natchez, a grand epic of American Indians, the influence of travel literature and the Marquis de Sade was sadly watered down by injections of Milton and classical reminiscences. The late autobiographical work, however, retains much of his youthful power: considerable portions had been written much earlier, and the title, Memoirs from Beyond the Tomb, is revealing—Chateaubriand wrote as if already dead and he had less need to compromise.

Revision always means compromise, but rarely has a style been created in which it was so difficult to preserve the original inspiration. Goethe thought that Romantic art was a disease; in a sense he was right—healthy attempts at correction most often ended up by destroying the essential character of the first drafts, replacing the initial fever with something tame and bland. This is evidenced not only in larger issues, in the attempts of Senancour, Chateaubriand, Coleridge, and Wordsworth to blur the evidence of their earlier political and religious sympathies, but in the smallest details of diction. The opening lines of Wordsworth’s “Home at Grasmere,” drafted most probably in 1800, originally read:

Once on the brow of yonder Hill I stopped
While I was yet a School-boy (of what age
I cannot well remember,…)

By 1814 this had become:

Once to the verge of yon steep barrier came
A roving School-boy. What the Adventurer’s age
Hath now escaped his memory:…

The manner has become archaic, the vocabulary trite (“roving School-boy”), and the change from first person to third has reduced the immediacy of expression and gives a stilted air to the passage.

Sometimes the change is apparently minimal but disastrous. Lines 24 to 30 originally read:

…Who could look
And not feel motions there? I thought of clouds
That sail on winds; of breezes that delight
To play on water, or in endless chase
Pursue each other through the liquid depths
Of grass or corn, over and through and through,
In billow after billow evermore.

Wordsworth changed “look” to “gaze,” and “liquid depths” to “yielding plain”; he added an enjambment to the last verse by writing

In billow after billow, evermore

Perhaps he was disturbed by the insistence on “liquid depths” in what followed, but “yielding plain” is second-rate Pope, the triteness reinforced by “disporting.” Wordsworth himself recognized that his revisions generally made things worse, but he could not prevent himself from trying. What we have here is an attempt to disguise and distance the original thought.

The project of The Cornell Wordsworth, already much advanced, is to publish all the early manuscripts; “diplomatic” transcriptions, which show typographically all the layers of revision, are provided, along with photographs of the most important manuscripts. The demands of the lover of poetry who wants to read an uninterrupted text are amply fulfilled, since “reading texts” of the most interesting poems or passages are given, often in different versions on facing pages. (These reading texts present the manuscript versions as the poet would have read them aloud to friends, with none of the obtrusive indications of correction and revision.) These volumes are expensive, even though the project must be heavily subsidized; but I presume that the reading texts, which often make up only a quarter of the total pages of each book, will eventually be collected in one volume, making Wordsworth easily available in his freshest and most immediate form.

A revival of interest in Wordsworth has been stimulated and, indeed, created by the publication of these early versions, starting with De Selincourt’s printing of the 1805 version of The Prelude in 1926. The complacent, serene, and even stodgy Wordsworth often presented by the work published during his lifetime has given way to a more tormented writer. (It is true that the more disquieting Wordsworth could have been construed from what was already before us, but the discovery was facilitated by the new material.)

The most striking aspect of the early versions is their greater concentration. The two-book Prelude of 1798–1799 contains all the “spots of time,” those intense experiences of childhood and early adolescence which return as memories to sustain and fortify us later: they are not dramatic events, but ordinary sights invested by the imagination with a visionary power. The thirteen-book Prelude of 1805 does not present all this material in the opening books, but diffuses it as well through books V, VIII, and XI. The two-book Prelude discovered after the final version, makes the odd impression not of a preliminary sketch, but of a ruthless abridgment, a stripped-down selection with a fierce intensity.

This newly found intensity, which has been given back to Wordsworth by recent studies, is confirmed and even consecrated by the exhibition “William Wordsworth and the Age of English Romanticism,” now at the New York Public Library at Forty-second Street until January 2, which presents an astonishingly complete picture of the period through pictures, prints, books, and manuscripts. The display of manuscripts is generous: alongside the most important of Wordsworth’s drafts, there are Coleridge’s “Kubla Khan,” several sheets of Shelley and Keats, and Blake’s copy of Wordsworth’s Poems opened at the flyleaf, where Blake expresses his stern disapproval of some aspects of the work. Documents and books trace the effect of the revolutionary events in politics and the changing movement of taste: the agitation over the slave trade, the American and French revolutions, the quarrels over the picturesque, the fashionable interest in mountains. Painting is represented with a richness that one could hardly have hoped for. The full range of Constable’s genius is on display. Recent studies of Constable have paralleled the reevaluation of Wordsworth, and have recognized a similar violence and passion in the artist.

Best of all, the pictures do not simply illustrate the manuscripts, or the poetry displayed act merely as captions to the art: they interact to reveal the ferment and the energy of the period. Besides Constable, we find J.M.W. Turner, John Sell Cotman, Samuel Palmer, and Thomas Girtin at their finest. There is a chance to see rarely displayed watercolors like Cotman’s picture of industrial desolation, Bedlam Furnace, and Girtin’s supreme masterpiece, The White House at Chelsea. The new vision of Wordsworth that inspired the show allows us to appreciate the full power and eccentricity of these and other works and restores their intensity,

The excellent catalog gives the best popular account of the political significance of Wordsworth’s style and his break with the eighteenth century. The only fault is a certain provinciality: the influence of German poetry and thought on Wordsworth and others goes unmentioned, and the importance of the standard trip to Italy in the liberation of many of the English painters (John Robert Cozens, Turner, Richard Wilson, etc.) from the constraints of local tradition is not recognized. The novelty of English Romantic art was not specifically English. Nevertheless, this exhibition and catalog give the best synoptic view of recent studies of the period. Wordsworth is presented throughout almost entirely by the early drafts of his work, and the fruitfulness of these studies is well established.

There has been an inevitable modish reaction to the new tendency in criticism of Wordsworth, and a few critics feel that the preference for the early versions has gone too far. One such is David Bromwich, who attacked the 1799 Prelude and claimed that the text of The Borderers of 1842 is superior to the manuscript of 1797–1799 and has a greater dramatic intensity.5 It is difficult to see how anyone of modern sensibility could prefer a version of a tragedy in which the hero’s name has been changed from Mortimer to Marmaduke. Some of the alterations, of course, may be defended. In the crucial scene, where the villain (Rivers, later called Oswald) recounts his emancipation from the ordinary rules of conduct, the simplicity of one passage has been altered for greater smoothness (IV, ii, 141–145):

When from these forms I turned to contemplate
The opinions and uses of the world,
I seemed a being who had passed alone
Beyond the visible barriers of the world
And travelled into things to come.

This became

When from these forms I turned to contemplate
The World’s opinions and her usages,
I seemed a Being who had passed alone
Into a region of futurity,
Whose natural element was freedom.

Perhaps the repetition of “world” in the original was felt to be awkward, and the new last line adds greater clarity to the moral exposition. There is a regrettable loss of power, however, and the revision of the extraordinary verses that follow is indefensible. They present an important aspect of contemporary political thought, one which Wordsworth may never have fully accepted, but by which he had been tempted:

Is not shame, I said,
A mean acknowledgement of a tri- bunal
Blind in its essence, a most base surrender
Of our knowledge to the world’s ignorance?
I had been nourished by the sickly food
Of popular applause. I now per- ceived
That we are praised by men because they see in us
The image of themselves; that a great mind
Outruns its age and is pursued with obliquy
Because its movements are not understood.
I felt that to be truly the world’s friend
One must become the object of its hate.

The first four lines were cut, and the succeeding ones watered down, made more diffuse:

I had been nourished by the sickly food
Of popular applause. I now per- ceived
That we are praised, only as men in us
Do recognise some image of them- selves,
An abject counterpart of what they are,
Or the empty thing that they would wish to be.
I felt that merit has no surer test
Than obloquy; that, if we wish to serve
The world in substance, not deceive by show,
We must become obnoxious to its hate,
Or fear disguised in simulated scorn.

In the fourth line, “do” is patently introduced only to fill out the rhythm. “Obnoxious to its hate” is an absurd tautology (even worse than “not deceive by show”), and the last verse adds an unwanted precision which weakens the force of the thought. Still feebler is the way “to be truly the world’s friend” is pedantically spelled out as “to serve / The world in substance, not deceive by show.” The contempt for general human dignity and for popular democracy inherent in much revolutionary thinking is drained of most of its expressive force.

The new critical position on Wordsworth may derive directly from a reexamination of the manuscripts, but it must be acknowledged that presenting an author largely by those aspects of his work that he never wished to be seen by the public is a peculiar undertaking, and seems an immoral one to many. In 1835 Balzac canceled the serial publication of Le Lys dans la vallée, refusing to send the next installment, and sued the publisher. What outraged him was the attempt of the publisher to make some extra profit by selling an early draft of one installment to the distant Review of Saint Petersburg, hoping that the author would not get wind of it.

Balzac’s agreement with his publisher required the initial setting of his work in a typeface which would not be used for the publication and from which only one proof was to be made; on this the author was to enter all the corrections, and it represented a second manuscript which had then to be reset in a different typeface. It was this first proof that was sent to St. Petersburg, and was printed there, including at one point, with comic effect, Balzac’s directions to himself, “CONTRAST” in capital letters, intended to remind him to insert a letter from the heroine. This proof was rather like an initial typescript, and Balzac’s rewriting increased the difficulty for the typographers who found his handwriting a chore to read. (“I’ve done my hour of Balzac,” he once heard a typesetter shout; “Who’s next?”)

It is easy to imagine Balzac’s justifiable anger when this half-completed version appeared in print—but we must not hastily deduce that he was ashamed of it. He had this first proof bound luxuriously in folio by an important binder; he had the other successive sets of proofs bound as well and presented to his friends. Balzac was proud of the way he built up his novels in a series of rewritings, enrichments, expansions, and retouches. He would have considered the present study of these states not so much flattering as deserved. The biography of a book would have seemed as natural to him as the biography of an author—equally vital, equally organic.

After the original publication in book form, Balzac altered the scene of the death of his heroine and struck out a long passage in which she bitterly reproaches the much younger hero for having too much respected her, for not having seduced her. “If you had been less submissive,” she cries, “I should live, I should be able to care for the happiness of my children, to get them married, to guide them in their lives. Why did you not take me by surprise at night?” These tactless phrases, and others like them, were removed by the author on the advice of his mistress, Mme. Hanska, and of a recently deceased mistress, Mme. de Berny, as well. He wrote ambiguously about the decision:

As my pen crossed out [these sentences], I did not regret a single one, never heart of man was more strongly moved, I seemed to see before me that great and sublime woman [Mme. de Berny], that angel of friendship, smiling as she smiled when I employed that rare power which consists of cutting off a limb and feeling neither pain nor regret in correcting and vanquishing oneself.

He then added about his heroine these words for himself: “Oui, il faut ne laisser que soupçonner les regrets” (“We must allow only a suspicion of regret”).

Many writers have experienced the necessity of eliminating the disastrous improvements of editors, of writing stet throughout their previously corrected typescripts, of restoring the vulgarity and the solecisms that rendered their thought with greater energy. Here it is the author himself who has damped the vulgar explicitness of his original. I do not think that the editors of the new Pléiade text were necessarily wrong in choosing the less powerful final version for the basic text and relegating the first published form to small print in the back of the book, particularly in light of all the changes in format that Balzac made in order to integrate this novel into the Comédie humaine, including the disappearance of the division into chapters. What is disquieting, however, is that the decision was made more or less automatically, not because the editors considered the final text as the most satisfactory.

Editors and biographers dislike the injection of what they call subjective elements into the choice of “copy text.” It is difficult to see whether anything coherent is signified by this prejudice. A taste in these matters is anything but arbitrary: a decision to choose the 1805 or 1850 version of The Prelude is more like a reasoned decision to invest in a certain kind of stock than like a preference for raspberry over apricot jam. Subjective choice of copy text seems to imply the use of critical intelligence: objective choice is the kind that can be exercised without disaster by an idiot. It goes without saying that, in bibliography as in any other field of human endeavor, no one has ever invented a system that will preserve us from foolishness.

“Dramatic irony,” was defined by Friedrich Schlegel in his essay On Incomprehensibility, as “when an author has written three acts, then turns unexpectedly into another man and now must write the last two acts.” Romantic art presents us with numerous examples of such “dramatic irony,” less often in works for the theater, where tradition tends to impose strict limits, than in novels and poems and in the more innovative musical forms. Paradoxically—or not so paradoxically, after all—the closer a work is tied to the artist’s life, the less he seems to be able to exercise his critical sense, above all when he returns to the work after a lapse of time. Schumann’s second editions of some of his finest conceptions like the Davidsbündlertänze, when he reprinted them more than a dozen years after they were completed, show a consistent attempt to remove some of their most original features. The most regrettable changes show how far he had distanced himself from the young composer he once had been—he added a large number of indications of repetition, so that many of the eight-bar phrases that appeared fleetingly and then quickly disappeared are to be given a new, heavy emphasis.

The Romantic belief in the importance of the unconscious will in artistic creation begins to appear more sensible and reasonable than we might be willing to acknowledge. Many an author has had the experience of creating a work that, as it grew, seemed to develop a logic and even a kind of will power of its own: the author carries out a plan of which he is only half aware. Bernard Shaw once remarked that he began his plays with no idea how they would go on, that the characters seemed to go their own way as he continued to write. The work at a certain point develops its own intention for which the author is only an agent. Even language had, for the Romantic artist, a momentum and an energy independent of the will of the artist. Novalis claimed:

It is the same with language as it is with mathematical formulae—they constitute a world in themselves—their play is self-sufficient, they express nothing but their own marvellous nature, and this is the very reason why they are so expressive, why they are the mirror to the strange play of relationships among things.6

Some revisions betray the work’s intention. Often these revisions are imposed on the artist from outside. Between the quarto and folio versions of Shakespeare’s Othello, a law prohibited the use of profanity on the English stage: it would not be unreasonable to choose the folio as copy text and to restore the oaths that had then been banned. The second version of Mozart’s Don Giovanni is dramatically weaker than the original one for Prague, in spite of the beauty of the new arias added: these were forced on Mozart by the demands and limitations of the singers.

With Romantic artists, we reach a generation often disconcerted by the implications or intentions of their own works. When this happens, the revisions become a betrayal of the work when they are not a form of tinkering. We might say that the writer has ceased being an author and has turned into an interfering editor of his own work. With a number of works—“The Rime of the Ancyent Marinere,” “The Ruined Cottage,” Hölderlin’s “Patmos,” Liszt’s Petrarch Sonnets—the editor is forced to consider which revisions are developments of the intention, and at what point the changes begin to betray rather than to enrich.

This no doubt opens the door to a certain kind of irresponsible editing, a descent back into the Dark Ages of publishing when an editor altered and repunctuated as he pleased and made up eclectic texts of all the little bits of different versions that he liked best. I do not believe this is probable. I should be happy, however, if more editors would acknowledge the conflicting claims of multiple versions, and realize that the establishment of a definitive text is, only too often, an imposture.

This Issue

December 17, 1987