Corps de l’article
As many readers of Romanticism on the Net are aware, Ronald Tetreault of Dalhousie University and I are preparing an electronic edition of Lyrical Ballads for Cambridge University Press. This edition will include full texts of all the authorized editions of Lyrical Ballads published in the poets' lifetimes, the full text of the unauthorized Philadelphia Lyrical Ballads of 1802, full transcriptions of the surviving printer's manuscripts housed in the Beinecke Library at Yale, and over a thousand images of manuscripts and printed pages, including complete sets of the pages of the authorized editions of the collection. Our texts will be fully searchable, according to a variety of criteria, and we will provide images of rare printed variants, such as cancels and paste-ins. No library possesses the range of copies that we will reproduce, and no exhibition, even in the bicentenary year of 1998, will bring them together in one place. But in the virtual space of our edition, they all will be present.
Producing an electronic edition of Lyrical Ballads has presented us with a troubling dilemma. Partly because of its importance, and partly because Wordsworth and Coleridge scholars have cared very much about the accuracy of their texts, Lyrical Ballads is an exceptionally well-edited collection of poems. Facsimile editions began appearing a century ago, its printing history and bibliographical characteristics have been minutely examined, and in print at the moment are excellent paperback editions for classroom use, hardback facsimiles of the London editions of 1798 and 1800, and, of course, the greatest Lyrical Ballads edition of all, James Butler and Karen Green's magnificent Cornell Wordsworth edition, published in 1992.  On the one hand, this rich editorial tradition provided a very firm foundation on which we could build our electronic version. On the other, we were concerned that our efforts would merely seem to repeat work that was already well done. We needed to find in the electronic medium ways to present our material that would be clearly distinct from print editions. Principally, we needed to distinguish ourselves from the Cornell edition of Butler and Green.
In a paper entitled "Versioning Wordsworth," delivered at the 1997 SHARP conference, Tetreault has discussed one of the ways in which we addressed this dilemma. Rather than supplying a single "best" reading text, as editors of letterpress editions have traditionally done, we decided to supply a multiplicity of reading texts, which readers can study and compare with each other. Tetreault's argument builds on and responds to the work of Jack Stillinger, Jerome McGann, and Zachary Leader.  Like them, he is suspicious of privileging one version of a poem or collection of poems over another; like them he also believes that developments in textual editing and literary theory have destabilized our understanding of the mode of existence of a literary work of art. The computer environment, Tetreault concludes, allows us to represent textual instability much more clearly than does the printed book: as McGann has argued, it is the necessary next step for the scholarly edition to take.  Moreover, because of its authors' habits of revision, and the existence of several distinctly different authorized editions, Lyrical Ballads seems ideally suited for the electronic medium.
We have also attempted to exploit other capabilities of the computer. The most obvious of these is its ability to analyze large bodies of data quickly. Our edition will provide the first reliable concordance to the various editions of Lyrical Ballads, and rather than having to thumb through a printed book, users will be able to generate this kind of information instantaneously, at the touch of a button. In addition, it will be possible to search our texts to generate information about Wordsworth's and Coleridge's stanzaic forms, their use of blank verse, and, we hope, their use of rhyme. Once again, this information, which in the past could only be gleaned after many hours of labor, will appear instantaneously. We have also attempted carefully to encode manuscript transcriptions, making possible analytical enquiries about handwriting, erasures, revisions, and deletions, that could not be performed in printed editions. And, finally, we are certain that as literary scholars become more adept at searching electronic texts, they will use the search capabilities of our edition to assemble bodies of data that we, at the moment, can scarcely imagine.
We are also interested in the ability of the electronic medium to make it possible to study the history and bibliographical characteristics of the collection in ways that would not be possible without extensive international travel to library collections. In this respect, our intentions are very different from those of the Cornell Wordsworth editors. There, Butler and Green were attempting to represent, as best as they could determine, authorial intent.  As a result, for all of its many strengths, their edition actually reproduces no edition of Lyrical Ballads that was ever published anywhere. Readers wishing to discover what exactly was included in the various printings of 1798, 1800, 1802 or 1805, can discover in the editorial notes, introductory material, appendices, and critical apparatus of the Cornell volume most of the information they require. But the search would be time-consuming, it would involve complicated acts of imaginative reconstruction, and, since bibliographical training has been on the wane for decades, it would be, for many scholars and students, not even possible. Our electronic edition all but eliminates the need for this kind of reconstruction. We have keyed our texts to specific copies of Lyrical Ballads, with the intention of preserving all of the known printed textual variants, as well as recording the ways in which the collection itself changed from edition to edition. We have also included digital images of the printed pages of the collection, several in both low- and high- resolution formats. These images will make it possible for scholars to see and manipulate what has almost always been reported in the difficult technical vocabulary of bibliographical description. Such images will also make possible close examination of typeface and page layout, and, in the case of pages with manuscript revisions, users will be able to enlarge handwriting samples severalfold, and even to "unerase" erased or faded pencil. Historians of printing, bibliographers, textual editors, and those interested in the publication history of Lyrical Ballads should find such an edition useful, both as a tool in their own research, and as a convenient means of teaching bibliographical and editorial skills to the next generation of scholars.
Because of its importance in the history of the collection, its scarcity, and the close attention it has received from bibliographers, we will give special emphasis to the 1798 Bristol issue of Lyrical Ballads. We have included full texts of each of the five copies of the Bristol issue that contain Coleridge's poem "Lewti." These are believed to be the earliest printed copies of the collection, each is bibliographically distinct from the others (sometimes the differences are minor, and sometimes they are quite significant), and very few scholars have actually seen and examined all five. Indeed, as late as 1992, Butler and Green reported that only four such copies exist, neglecting one of the two at the Beinecke Library (44). We will also include full texts of two of the remaining copies of the Bristol Lyrical Ballads, in which "Lewti" has been replaced with Coleridge's "The Nightingale": a copy from the Rothschild collection, now housed at Trinity College, Cambridge, which contains Coleridge's manuscript corrections to the "Rime of the Ancyent Marinere" (high-resolution digital images of the corrected pages will also be included); and the unique British Library copy which includes Thomas Beddoes' poem, "Domiciliary Verses."
The London imprints of Lyrical Ballads will of course be well represented. We have included a full text of the London, 1798, issue, based on a copy at Simon Fraser University and accompanied by a full set of digital images of the printed pages. As it happens, this copy contains the rare "horsebehind" variant in "Simon Lee," first noted (to my knowledge) in the Cornell edition (43). We are also indebted to Simon Fraser University for digital images of the editions of 1800, 1802, and 1805, and our texts are again based on their copies. Like the Bristol imprint of 1798, the edition of 1800 will receive special emphasis, because it is the most bibliographically complex of all Wordsworth's publications. For volume I, we will include electronic texts and digital images of the cancelled "Christabel" leaf from the "Preface," and texts and images that reflect both the initial misprints in "The Idiot Boy" and their corrections. For volume II, we will reproduce the variant versions of both "Lucy Gray" and "The Two Thieves," and have devoted particular attention to what Coleridge called the "infamous Blunder of the Printer" in "Michael," where fifteen lines of the poem were omitted in the initial printing.  Here we are grateful to the Wells Wordsworth Collection at Swarthmore College and the library of the University of Colorado at Boulder. The Wells Collection has supplied us with a text and digital images from a copy of volume II which contains a paste-in on which the missing lines from "Michael" are printed (only two such copies are known to exist). The University Colorado has kindly supplied us with digital images of the "Michael" cancel, on which the missing lines were also supplied, and the new 27-item errata leaf, both of which were printed in later issues of Lyrical Ballads (1800). Few scholars who know about these variant printings have actually seen what they look like, and even fewer have had the opportunity to compare them closely with each other. With the publication of our edition, that will change.
These are some of the highlights of what the edition will include. Those interested in previewing our work more closely should visit the Lyrical Ballads Bicentenary website, which, when fully developed, will contain sample texts of three distinct states of the 1798 first edition, based on Bristol imprints from Princeton University and the British Library, and on the London imprint from Simon Fraser University. A selection of images will also be available, a tentative table of contents of the edition, and links to on-line studies of Lyrical Ballads, such as Mark L. Reed's study of the Bristol imprint, forthcoming in Studies in Bibliography, and Alan Boehm's recent ELH article (for those whose with subscriptions to Project Muse). There will also be links to information about relevant conferences and exhibitions, and to various on-line projects having to do with Lyrical Ballads. We intend this project as an experimental model for what literary scholarship in the electronic medium can accomplish, and we look forward to receiving comments from the scholarly community about our efforts.
"Representing Textual Instability"
William Wordsworth told the story of Margaret again and again. The affecting tale of the loss of her family, her decline and death, has a tragic inevitability, though in the end the sympathetic listener is offered a degree of consolation. The closing lines once read:
The Excursion, Book I, 1814 first edition and all subsequent editions through 1843
My Friend! enough to sorrow you have given,
The purposes of wisdom ask no more:
Be wise and cheerful, and no longer read
The forms of things with an unworthy eye.
She sleeps in the calm earth, and peace is here.
But in 1845, he altered the lines to provide a more conventionally Christian solace:
The Excursion, Book I, 1845 and all subsequent editions 
My Friend! enough to sorrow you have given,
The purposes of wisdom ask no more:
Nor more would she have craved as due to One
Who, in her worst distress, had ofttimes felt
The unbounded might of prayer; and learned, with soul
Fixed on the Cross, that consolation springs,
From sources deeper far than deepest pain,
For the meek Sufferer. Why then should we read
The forms of things with an unworthy eye?
She sleeps in the calm earth, and peace is here.
This is perhaps the most dramatic example that could be cited of Wordsworth's habit of revision. Over his long life through the many editions of his works, he repeatedly changed not only accidentals like punctuation, spelling, and capitalization, but words, lines, and even whole passages that can have the effect of sending a poem off in a whole new direction.
The alterations to Lyrical Ballads are not quite so striking, but they are no less interesting to the textual critic and the bibliographer. The publication of the first edition in 1798 is complicated by the succession of the London issue to Cottle's initial offering from Bristol, and each of these exists in several states of printing and assembly; indeed, it hardly seems that any two copies of the rare Bristol issue are alike. Even more important is the vast expansion of the collection into two volumes in subsequent editions, largely to accommodate additional contributions from Wordsworth. Coleridge often changed his mind too, pulling "Lewti" from the first issue and replacing it with "The Nightingale", allowing "Love" to be added to later editions, and making substantive revisions to "The Rime of the Ancient Mariner". Wordsworth's revisions to his poems in subsequent editions of Lyrical Ballads are even more numerous.
When we read these poems, then, we always need to remember that we are reading only one version of Wordsworth at any given time, and the same may be said for Coleridge. Unfortunately, the printed texts that scholars use, and even more the ones we teach to our students, are constrained to give us only one Wordsworth or another so that we never can grasp him whole. But there is no way round choosing a particular copy-text in establishing a printed edition. Traditionally, editors have stuck by the principle that the text the author leaves upon his death represents his final intentions, a sort of legacy based on the model of a last will and testament that has more to do with legal practice than with aesthetic considerations. Thus, the edition of 1849-50 published when Wordsworth was eighty years old has usually served as the copy-text upon which editions of his works are based. The inadequacy of this mode of representing a great artist's achievement has been noted by Jonathan Wordsworth, who laments that "on the whole poets are known by the best versions of their works: Wordsworth is almost exclusively known by the worst." 
This kind of unease with the portrait of the artist in old age has been registered among a whole new generation of Wordsworth editors in their mission to recover a younger, more vibrant figure. These have developed a snapshot of Wordsworth closer to the moment of inspiration by basing their texts on the earliest completed version of his poems. This trend has culminated in the Cornell Wordsworth, a series of scholarly editions begun in l975 and ongoing. These volumes aim "to bring the early Wordsworth into view," according to General Editor Stephen Parrish, by printing "clean, continuous 'reading texts' from which all layers of later revision have been stripped away." 
Recently, though, this drive to return to an originary Wordsworth has come into question. Decrying a "devaluing of secondary processes" seen in the so-called primitivism of this approach to Wordsworth, Zachary Leader in Revision and Romantic Authorship urges us toward a more complicated view of the author.  Leader wants us to understand the differing concepts of identity that underlie both the traditional and the revisionary editorial methodologies. Taking the final authorized version as your copy-text places a high valuation on clarification, refinement of meaning, and recuperation, and implies a concept of identity as something to be worked toward and achieved. Such a view of Wordsworth remains compelling, since what readers often value in the poet is the sense of stability and unity of self his work strives to convey. However, a difficulty arises when we recognize that Wordsworth's intentions are irrecoverable. Furthermore, even where they can be roughly determined, they remain in the realm of wishes that may not be fulfilled, of goals set but unachieved. On the other hand, adopting as copy-text the earliest completed version of a work implies a solidarity with Romantic notions of the spontaneity and vitality of the youthful self. One difficulty Leader finds with this view is its "uncritical absorption in Romanticism's own self-representations", which Jerome McGann warned against in his critique of the Romantic bias in Romantic criticism.  Another difficulty Leader raises is that this approach in actual fact masks an aesthetic preference, which after all could be the subject of considerable debate. Besides, preferring either the late or the early Wordsworth is to slight all those versions in between, where a poem finely polished may be marred by subsequent tinkering. The editor is not really faced with an either/or choice between early and late, but something far more complicated, as Leader points out:
Early versions are more vivid ... but late versions are clearer. Early versions, it is claimed, are more original ... but late versions are more correct. Moreover, as I have tried to show, these oppositions are rarely as clear-cut as partisans would have us believe.72
Certainly my experience as a close student of Wordsworth's texts tells me that there are occasions when I prefer lines in the first edition, others where I deplore them and am glad he made emendations. Sometimes I find the expression in the final text satisfying and sometimes not. Certainly there are times when I could wish that he had let well enough alone.
Such considerations lead to the conclusion that the protean Romantic self can never be caught in any single text. As Leader shows, an alternative to the concept of the unified self is to regard identity as variable and multiple, changing with respect to circumstances and over time. Understood this way, there is not one but many Wordsworths, the young poet in his twenties evolving toward the old man of eighty with many stages in between, each with some claim to mastery and each therefore with some claim on our attention. Such a view of the poet is advocated by Susan Wolfson, who finds a resistance to closure in Wordsworth's practice of revision:
Wordsworthian revision sustains the illusion of mastery but steadily postpones that achievement .... Revision is endlessly open, not simply because any field of vision is open to numerous, potentially infinite interpretations and organizations , but because each view discovers new motions, changes, and interchanges. 
The consequence of this concept of identity is a destabilization of the text, which can now only be represented adequately in its multiple versions. But the medium of print is not well-equipped to depict such an open-ended process. Print can only show change over time through a cumbersome apparatus of footnotes or end-notes. These place great demands on the reader, who must sort through, interpret, and plug in variant readings. Nor for reasons of space can the conventional book afford to print all versions as they evolved. The new electronic medium of hypertext might offer a solution, though Leader is dubious that anyone but the specialist scholar would want to use it. (74)
True, an electronic Wordsworth may not be eagerly sought out by the general reader; I've met very few lovers of reading who prefer a screen-image over an affordable book you don't need elaborate equipment to use. Indeed, such a Wordsworth edition may not in fact be meant to be "read" at all, at least not in the sense in which we still use the term today.  Instead, any edition of Wordsworth in the new medium would be meant to be explored and argued over. The virtually infinite extensibility of digital resources allows for the inclusion of multiple versions, and these linked hypertextually could then be placed side by side and compared. This new sort of edition could then become the basis for new interpretations and new debates among readers of the poet, who now would be empowered to make and defend their own choice of preferred text. In these exchanges, the options could be studied closely in a fashion that the linear structure of print makes difficult, and used at a distance by scholars who don't enjoy the privilege of proximity to a major research library.
Even more important, Wordsworth in hypertext may be the most effective way yet to represent Wordsworth in development. New media functionalities of electronic text, multiple windows, digital images, and interlinked hypertext may offer an avenue into the maze of this poet's diversity. Hypertext may thus offer a means of grounding in actual practice Jack Stillinger's theory of "textual pluralism", in which "a work" of literature becomes an abstract organizing principle around which every physically-embodied version (published editions, manuscript drafts, etc.) is grouped, as a text among texts to be viewed both separately and simultaneously with all its cognates.  Conveying this plurality of each poem's incarnation in words is the focus of my work on the electronic edition of Lyrical Ballads that I am preparing for Cambridge University Press with my colleague Bruce Graver, of Providence College in Providence, Rhode Island.
In the print medium the choice of a copy-text has always posed a dilemma: establishing a text in print has always meant that we must choose to privilege one version over others, and settle for a static representation of what might be better understood as a dynamic process. In the electronic medium where there is no one text but multiple versions, that problem disappears only to give rise to new difficulties. Where the text of a work is not one but many, readers must cope with an excess of information, must keep track of which version they are examining, and must be able to relate one version to another in an intelligible fashion. None of this can be accomplished without organization, yet the principles for organizing electronic editions remain to be established. The dizzying possibility of getting lost in cyberspace is the abiding problem of hypertext.
It is heartening to know that the problem of developing what may be called "aids to navigation" is by no means unique to the electronic medium. The history of the printed book gives ample instances of the gradual and sometimes painfully slow development of conventions we now take for granted, such as page numbers, tables of contents, title pages, and indexes. The struggles of Renaissance scholar-publishers such as Aldus Manutius to evolve the structure of the modern book are likely to be repeated by those who would work in the new electronic medium.
As the member of the editorial team responsible for hypertext design on the electronic Lyrical Ballads project, I have tried to address these problems of navigation by using tools associated with the SGML and HTML markup of texts. Standard Generalized Markup Language or SGML is a scheme for representing the logical structure of texts, so that treating a text in this way automatically builds in the benefits of order. So- called SGML-capable software such as Panorama Pro allows the viewing of suitably marked-up text on a computer screen, and provides a means of navigating through it. It does so by means of two parallel windows, the one on the right containing the document to be viewed with the one on the left functioning as a sort of table of contents. A simple mouse-click on any title in the table of contents window takes you directly to the poem in question. A similar operation could be performed on any one of the four lifetime editions of Lyrical Ballads, whether 1798, or the revised versions of 1800, 1802, and 1805. This software also allows the user to call up a digitized image of the page displayed on the screen, so that original typographical features like font and layout may be inspected. These page images must be carefully and clearly linked to their corresponding passages in the text presented on the screen. Of course, like all electronic texts, the document is searchable and therefore self-concordancing.
However, even under this system it is not easy to compare one text with another. Some high-power SGML browsers are capable of generating multiple windows, each with a text from a different edition. These windows can then be scaled to different sizes and moved about the screen, but besides these manipulations being troublesome there is no easy way to keep track of which version is contained in which window. The consequence is that the skills of the software user come to predominate over those of the student of literature. Just as in the early days of print, where electronic texts are concerned more thought needs to be given to how we are to make our way through complex non-sequential texts.
To make the electronic Lyrical Ballads more user-friendly, I have attempted to address the problem of navigation by experimenting with functionalities associated with the Internet delivery of documents over the World Wide Web. An array of windows seems essential to the display of the Wordsworth project, for as Sherry Turkle observes "windows have become a powerful metaphor for thinking about the self as a multiple, distributed system."  What seems most attractive about HTML standards at the present stage of development is that they give control over the use of frames, a system of layout in which windows are generated on the screen according to a pre-determined pattern. Though this pattern can be easily altered by the user, its grid provides an intelligible starting point for the display of multiple texts of the same poem on the screen at the same time.
The prototype pages (click to see them in full interactivity) I have developed so far take "Simon Lee" and "We are Seven" as test-cases, and consist of a galaxy of pages each composed of five frames: one narrow vertical column on the left plays the customary table of contents role, while a grid of four squares on the right allows the reader to compare four different versions of the poem on the screen at once. The text in each frame may be scrolled through manually, but by the use of HTML's internal anchors a simple click on a live hyperlink in the left-hand frame causes all four texts in the squares to scroll simultaneously to the same line. There are two advantages to this scheme of display: first, a degree of animation is introduced into the text that cannot be duplicated in print, and, second, a balance is struck between exploration and direction in the reader's examination of textual complexity.
There is a second unusual feature of this scheme that is meant to help readers find their bearings. The left-hand column in this HTML treatment contains no mere list of contents but what I call a "variant map" of the poem being studied. The variant map is a guide to revisions that were made at various stages in the poem's development. It is based on the poem to the extent that it reproduces the text of the poem wherever changes were not made, but whenever a change in any of the versions under consideration is encountered it substitutes a descriptive hyperlink for the variants themselves. The reader is thus alerted by a sort of palimpsest that an alteration has been made, and by clicking on the "hotspot" can summon up the parallel passages. Replacing the copy-text with a variant map turns the annotation process inside out in a way that seems appropriate to the dynamism of this medium, for the hiatus in reading caused by running across a link cues the reader to click on the spot and look to the right (in the direction of the normal flow of reading) to learn the word or character elided and to compare versions. Rather than footnotes which distract attention from a definitive text, the variant map is an abstraction of the poem which does not privilege one version of the text over another and that piques the reader's curiosity by means of gaps in the text to pursue the significance of revisions made in successive versions. Together with its links to the four display windows, the variant map makes possible a dynamic collation of variant texts that cannot be achieved in print.
Because this new medium reinscribes textual stability as a series of moments in a lengthy creative process, the dimension of time must somehow be added to those of space. In our project, it is hoped that by adding motion to comparative views, hypertext will enable us to represent this fourth dimension in a way readers will find comprehensible. If electronic texts are to justify themselves, they must move beyond the three-dimensional space of the book. It is not enough to produce texts that are indexed and searchable; like the Aldine editions of old, electronic texts must establish their own unique conventions and develop easily-understood forms of presentation that distinguish them from what went before. The new medium should neither replace nor reproduce the book, but must strive to do things books could never accomplish.
Lyrical Ballads and Other Poems, 1797-1800, ed. James Butler and Karen Green (Ithaca, 1992).
Jack Stillinger, Coleridge and Textual Instability (Oxford, 1994); Zachary Leader, Revision and Romantic Authority (Oxford, 1996); Jerome McGann, A Critique of Modern Textual Criticism (Chicago, 1983).
Jerome McGann, "The Rationale of Hypertext". n. pag. Online. Internet. 10/12/1997. <http://jefferson.village.virginia.edu/public/jjm2f/rationale.html>.
"Our aim . . . is to stress Wordsworth as author more than Lyrical Ballads as book," they write, in explaining the rationale for their choice of the copytext for "Michael" (127).
Collected Letters of Samuel Taylor Coleridge, ed. E. L Griggs, 6 vols. (Oxford, 1956) vol. II, p. 707.
See The Poetical Works of William Wordsworth, ed. Ernest de Selincourt and Helen Darbishire, 5 vols. (Oxford: Clarendon Press, 1940-49) vol. V, p. 39.
The Music of Humanity: A Critical Study of Wordsworth's 'Ruined Cottage' Incorporating Texts from a Manuscript of 1799- 1800 (London: Thomas Nelson, 1969) p. xiii.
Preface to The Cornell Wordsworth, in (for example) Lyrical Ballads and Other Poems, 1797-1800, ed. James Butler and Karen Green (Ithaca: Cornell University Press, 1992) p. v.
Zachary Leader, Revision and Romantic Authorship (Oxford: Clarendon Press, 1996) pp. 1-77.
The Romantic Ideology: A Critical Investigation (Chicago: University of Chicago Press, 1983) p. 1.
"The Illusion of Mastery: Wordsworth's Revisions of 'The Drowned Man of Esthwaite', 1799, 1805, 1850", PMLA 99 (1984) 932.
Sven Birkerts points to this distinction quite clearly in The Gutenberg Elegies: The Fate of Reading in an Electronic Age (Boston: Faber and Faber, 1994) p. 15: "It is precisely where reading leaves off, where it is supplanted by other modes of processing and transmitting experience, that the new dispensation can be said to begin."
Coleridge and Textual Instability (New York: Oxford University Press, 1994) pp. 118-140. Jerome McGann, in A Critique of Modern Textual Criticism (Chicago: University of Chicago Press, 1983), contributed significantly to the development of "version theory" as a mode of textual criticism; following this methodology means that "the first consideration which the critical editor must face is to distinguish textual versions and not ... to choose copy-text" (p. 114).
Life on the Screen: Identity in the Age of the Internet (New York: Simon & Schuster, 1995) p. 14. Turkle follows George Landow's Hypertext (Baltimore: Johns Hopkins University Press, 1992) and Richard Lanham, The Electronic Word (Chicago : University of Chicago Press, 1993) in writing of the way identity is understood as fragmented and discontinuous in the digital age.