from Victorian Studies Volume 41, Number 4

Centripetal Textuality

P. Aaron Potter

---

Permission to Copy

You may download, save, or print for your personal use without permission. If you wish to disseminate the electronic article, or to produce multiple copies for classroom or educational use, please request permission from:

Copyright Clearance Center
Professional Relations Department
222 Rosewood Drive
Danvers MA 01923

FAX: 978-750-4470/4744
Web address: www.copyright.com

For other permissions, use our online reprint request form.
 

---

The advent of electronic media, particularly in its most interactive incarnations, the Internet and World Wide Web, has been widely pronounced a crucial turning point in our relationship to text. Such claims may perhaps be most dramatically concretized in George P. Landow's assertions in the mid-eighties that the arrival of popularly available hypertexts would signal a new age of interaction between text and readers, offering both an opportunity for readers to access contextual information with greater ease than ever before, and the ability to modify and reorganize material on their own initiative--a departure from traditional straight-line narrative resulting in a revolutionary reassessment of the location of the author- and editor-functions for such texts. However, despite current wide-ranging use and access to hypertext editions of everything from personal Web pages to entire editions of canonical authors, the revolution in textual studies (and interpretive modes) has failed to take place. Still, a number of scholars continue to hoist Landow's banner of textual revolution and textual decentralization as though the end of editorial intrusion were just around the corner--notably, Jerome McGann in articles such as ''Radiant Textuality'' (1996), and the attendant work, ''The Rationale of HyperText'' (1995). These two articles are closely related to arguments in McGann's Critique of Modern Textual Criticism (1992) which take issue with the attempts of textual editors of the Greg/Bowers tradition to produce ''definitive'' editions of literary works by synthesizing multiple variant editions into single texts which reflect the authors' ''final intentions'' toward those works. It is a project which McGann considers both ludicrous, in light of the falsity of the very notion of ''final intentions,'' and a disservice to readers, who are robbed of the opportunity to interact with a text uncorrupted by interposing textual critics, a text manifesting its historical and social nature. In these two articles, McGann, like Landow, sees hypertext and electronic media as not merely the forums, but the very catalysts for his personal critical and ideological project. However, one can argue that electronic media and the emerging hypertextual format will achieve precisely the opposite of McGann's project: that in fact the very nature of these media, as well as the strong social response to them, will serve to intensify the traditional view of literary objects as transcendental ''works,'' and thus reinforce traditional relationships to text and textual criticism.¹ This essay proposes that Landow, McGann, and other critics who suggest that emerging electronic media will destabilize traditional textual paradigms, have severely misread the potential and import of such technologies. In point of fact, not only are most electronic media cultivated to perform precisely the type of centripetalization which McGann opposes, but McGann's own work in the electronic forum demonstrates the deficiency of his theories.

I. The Arguments for Decentralization

In order to evaluate McGann's arguments properly, it is necessary to define terminology which has become widely circulated, but perhaps also widely misused. The term ''hypertext'' was coined in the 1960s by Theodor H. Nelson, who defined it as ''nonsequentially read (or written) text'' (qtd. in Landow, ''Hypertext in Literary Education'' 174). Another pioneer of hypertext systems, Andries van Dam, goes on to define it thus:

Both an author's tool and a reader's medium, a hypertext document system allows authors or groups of authors to link information together, create paths through a corpus of related material, annotate existing texts, and create notes that point to either bibliographic data or the body of the referenced text. . . . Readers can browse through linked, cross-referenced, annotated texts in an orderly but non-sequential manner. (qtd. in Landow, ''Hypertext in Literary Education'' 174)

What this means in terms of both reading electronic editions of texts and browsing through systems such as the World Wide Web is already familiar to us. Rather than experiencing a text as a single, straight-line narrative running along the author's determined path from beginning to end, readers of hypertext are able to rearrange their experiences of text both in time and in depth. Information in hypertext is often broken down into small chunks, each item containing a series of links to other, related items, which can be followed or not at the reader's discretion. By clicking on words or markers in the original document, readers may follow the link to related material more or less on their own free-associational path. Even texts which were originally written in traditional media--straight-line narratives such as Victorian novels, the Bible, or last week's Time Magazine--are often broken down into smaller elements when translated into hypertext, thus inviting the reader to digress from a straight-line reading and to take whatever sidetracks offer themselves, whether leading only to a set of footnotes or commentary, or to a new stream of text or documents.

McGann's fundamental argument in ''Radiant Textuality'' is that this apparent empowerment of the reader in his or her relationship to text will result in a redefinition of text itself (or, if one accepts the premises of McGann's Romantic Ideology [1983], a return to a previously tenable definition). Texts will be apprehended as implicitly social constructions--what G. Thomas Tanselle has called ''libretto[s]'' for reading performances (23-24). In particular, readers will have the opportunity to pick and choose among multiple variant editions of a text, selecting the most valuable for their own social purposes at a given time, just as the original author (McGann argues) selected different versions of a work based upon then-present need. This is in direct contrast to the traditional work of scholarly editing, particularly as codified by Fredson Bowers and his followers, in which individual texts are regarded as shadowy reflections of an idealized, transcendent ''work,'' representing the author's ''final intentions'' toward his or her project. In the traditional process, the reader is presented with a synthetic edition, fabricated from several exemplars, which, in the editor's opinion, best reflects the ideal, transcendent ''work.'' McGann's concept of effective hypertext is one which removes such editorial processes, or at least renders such processes transparent, by presenting a complete schema of variant editions simultaneously, rather than conflating them, thus returning the variety and fluctuating nature of texts to the readers, and allowing readers to make their own choices regarding which texts are ''best'' for their purposes. This has the dual advantage of resituating the texts in their historical and social context (the only context, McGann contends, in which texts have meaning), and of removing the editorial practice of attempting to recreate the author's lost ''final intentions'' which, McGann maintains, is a fictional concept. Both ideas are extensions of McGann's familiar arguments from Critique of Modern Textual Criticism, and, to some extent, the ideas of social construction forwarded in Romantic Ideology.

However, McGann's position is also strongly reminiscent of Landow's arguments for hypertext as a means of reader empowerment. Landow has proposed that hypertext may result in a democratization of textuality (a notion very much in the Romantic tradition) as a result of both greater reader interaction with text (including the ability of readers to add their own material), and of the ability to incorporate context more immediately into text itself, through hyperlinks. Landow's work with hypertext at Brown University can be considered a significant precursor to some of McGann's methodologies, and perhaps even his theories, and in this light a brief contemplation of Landow's most influential project, The Dickens Web, and its reception may be illuminating.

Briefly stated, The Dickens Web project involved constructing a website which included the text of a literary work, in this case a Victorian novel, setting up a few initial explanatory and historical hypertext links, and then encouraging students throughout the course of the term to add new material and links to the original database. Classes were conducted in a computer lab and students were given instruction in how to create files and link them to the collective web, both to the source text and to other students' file areas. The Dickens Web has since been distributed, as a file template system, to several schools where it has met with some success (Landow, ''Hypertext in Literary Education'' 174-98). In particular, teachers have noted that students are much more apt to read historical and critical materials beyond what is required in class, and seemed to bring this enthusiasm to their final papers. Jonathan Smith, using The Dickens Web to teach Great Expectations (1861) at the University of Michigan, particularly claims that students taught with the aid of The Dickens Web produced final papers that were much more focused and much more interesting to their authors, stating that 42% of students said that they found their final research projects to be more interesting to them than usual (124).

In response to projects like Landow's Dickens Web, several universities have experimented with student-authoring of hypertext documents, particularly as the culmination of a hypertext-taught course. Peter Havholm and Larry Stewart of the College of Wooster have claimed great success in teaching the interrelationships between culture and text by having students produce sets of hyperlinked files, similar to World Wide Web pages, which relate multiple primary texts with appropriate historical and sociological commentary. However, they note that while this again exposes students to a broad array of contextual materials, it does not necessarily foster the critical thinking and commentary necessary for deep analysis of any given text.

Much of this activity may be seen as a precursor to McGann's projects and particularly to his commitment to the consideration of texts in historical and social context. However, McGann sees hypertext not merely as an improved glossary or history book, but as the key to eliminating the idealizing editorial function which he feels currently intrudes into most textual scholarship. As I have suggested, for McGann texts are written and rewritten within social and historical confines, the signs of which are falsely hidden by synthetic editions of texts which may span many years, rewrites, and social milieus. Fluctuation and social construction are the bywords of textual production, and, he believes, the bywords of electronic media as well:

[I]n many . . . cases one would like the possibility to make ad hoc or provisional choices among the full array of textual alternatives--to shift the point of focus at will and at need. One cannot perform such operations within the horizon of the book. A hypermedia project like the Rossetti Archive offers just these kinds of possibility. Unlike in traditional editions, ''hyper'' editions need not organize their texts in relation to a central document, or some ideal reconstruction generated from different documents. An edition is ''hyper'' exactly because its structure is such that it seeks to preserve the authority of all the units that comprise its documentary arrays. In this respect a hyperedition resembles that fabulous circle whose center is everywhere and whose circumference is nowhere. (''The Rationale of HyperText'')

To some extent, McGann acknowledges the limitations of the current state of the media--that many editions purporting to be hypertextual are, in fact, performing no service that an adequate paper edition could not perform as readily.² However, he also believes that truly hypertextual editions, such as his Rossetti Archive, will in large measure serve to undermine the search for textual stability which underpins the Greg/Bowers school of editorship and, more generally, our approach to literature and art. Multitudinously variant editions made available through electronic media will serve to negate the idea of a transcendent, extra-historical ''work,'' while the fluctuating, interactive nature of the World Wide Web and other electronic media will usher in a new age of engagement with literary products. However, McGann's claims seriously misread the current situation, particularly with regard to the nature of electronic media.

II. The Nature of Electronic Media

It is important at this point to recollect that original definition of hypertext as a ''nonsequential'' medium, one in which the reader, rather than the author, determines the pace, direction, and focus of the story. A pre-scripted television program, newspaper article, or even electronic document, is not hypertextual. This is a crucial distinction: merely putting Paradise Lost (1667) on an electronic disk does not make it a hypertext. For those who have read the poem in a five-dollar paperback edition, the experience of reading Paradise Lost would be changed, because it is being read on a computer--but it also would be changed by reading a thirty-dollar hardbound edition. Even appending a glossary, an index, and a series of critical articles would not change our hypothetical Paradise Lost into true hypertext.

Several reviewers have criticized recent electronic editions of canonical works on just these grounds. Stuart Moulthrop points a critical finger at what he claims are merely ''Hypertext incunabula'' (59)--editions of texts on CD-ROM and in other electronic media which claim to present a truly ''hypertextual'' perspective, and which instead merely break down their source material into chunks of data, facts, and word lists to be dissected. Such editions do no service that a reasonable paper dictionary or bibliography could not perform as adequately. In fact, Moulthrop suggests that the very disjunctiveness of this approach to literature--the reduction of texts to a series of individualized, separate pages, sound bites, animation clips, and facts and figures--may turn people away from electronic editions and back to the traditional paper texts which preceded them (an argument I will consider later in this essay).

Moulthrop and others have also suggested that, in general, the hypertext documents and services which are available today fall far short of Nelson's original vision of truly nonsequential text. While it is true that a user of the World Wide Web is free to follow or not any links which he or she encounters, the user is still limited by those links the editor, programmer, or author has chosen to set up. As Foucault reminds us in Discipline and Punish (1977), many systems which appear to offer greater freedom are actually just doing a more masterful job of hiding the ways in which they control the subjects' activities and experiences--a statement certainly true of hypertext, wherein the programming which creates hyperlinks is hidden from the reader, who sees only the words or objects to which the links have been attached, which may or may not indicate the structures and hierarchies underlying the hypertext document (Smith 128). Hypertext is clearly not an authorless (or editorless) text by any means, even if it does tend to share some of its author- or editor-functions with the end-user or reader.

Still, even given these limitations, hypertext does imply a certain breakdown of the traditional relationship between author and audience. J. Hillis Miller notes that in their direct links to their source material, hypertext commentary and criticism actually intrude much more into the primary text than does a traditional body of criticism or explication. In reading the University of Texas at Austin's hypertext edition of Pride and Prejudice (1813), for instance, one cannot help but notice the sprinkling of bright blue words, which are linked to other sites and material. Even if one does not click on one of these seductive sidetracks, one is at the very least aware of one's own resistance to that temptation.

These observations are crucially important to McGann's theory of hypertext. He claims, ''When a book is translated into electronic form, the book's (heretofore distributed) semantic and visual features can be made simultaneously present to each other. A book thus translated need not be read within the time-and-space frames established by the material characteristics of the book'' (''Rationale of HyperText''). McGann's treatment minimizes the possible effects of the ''material characteristics'' (if hyperlinks and similarly encoded structures may be considered as such) of hypertexts themselves. If a reader is unable, ultimately, to escape the controlling influence of a Web page editor any more than that of a text's editor, where is the revolutionary aspect of reader interaction? McGann's Rossetti Archive, for instance, has its own hierarchies, organizational structures, and implicit (and sometimes explicit) directions to readers. It contains several suggested entry points, and forward and back buttons which suggest a linear reading experience. In addition, search engines which direct browsers to the project all point to specific pages which delimit the structure of the project. Although in ''The Rationale of HyperText'' McGann claims that ''a HyperText is not organized to focus attention on one particular text or set of texts,'' such arrangements imply that there is just such a focus, and that (at least under current paradigms) hypertext is no more transparent a medium than any reasonable index.

Even more crucial to McGann's Romantic suggestions than the transparency of the medium, however, is his suggestion that hypertexts have the ability to reduce editorial intrusion by presenting a potentially infinite number of parallel texts, none privileged over another. The reader thus can choose, for instance, which edition of Wordsworth's The Prelude (1798-99) he or she wishes to read, rather than relying upon an editor's attempts to syncretize several variants into a false construction intended to represent the work of The Prelude as a whole. What is most interesting are the ways in which electronic media privilege texts in precisely the opposite way McGann predicts.

Electronic media are suited, first and foremost, for the reproduction of extant texts. By their very nature, electronic media make it easier to produce copies of an existing document--much easier than to produce a new object, whether that entails a creative labor resulting in a new electronic work or the translation of an analog (in this case real world, material) work into digital format. Once a document, text, picture, or other object has been digitized, it may be reproduced an infinite number of times, each copy a precise duplicate, down to every pixel and bit, indistinguishable from the original. To appreciate the significance of this, consider the behavior of most users producing Web pages that incorporate fragments or entire copies of literary texts. Rather than bothering to retranslate the text into the digital format by hand (whether by typing, scanning, or other processes), most users will first search for an extant copy of the text online, and then either create a link to the existing resource, or duplicate the material on their own Web pages--a much simpler solution, and one directly supported by the facility with which material in an electronic format can be reproduced.

Such guileless duplication actually minimizes the reader/user's social interaction with a text, rather than increasing it, as McGann predicts. In particular, it minimizes the ability (or impulse) to introduce variants into the document, the very sort of interplay which McGann claims is the most dramatic form of social interaction with a text, and which he believes may be encouraged by electronic distribution.

Further, in evaluating the implications of electronic media we must consider that the most popular current forum (and the one in which McGann is presenting his Rossetti Archive), the World Wide Web, contains within it certain paradigms which may further undercut his conclusions. In his work with Landow's Dickens Web, Smith notes that the interactive hypertext project failed to be the revolutionary teaching tool he had expected on the basis of Landow's advertisements. Students were exposed to more material concerning the historical period of Dickens's life and works, but found it more difficult to link this material thematically to the book itself. Smith blames this on what he calls the ''atomistic'' nature of current hypertext documents. That is to say, information is stored in such discrete cells (such as Web pages) that it seems curiously un-connected to any other material, even when the two are hyperlinked. Smith argues that by compartmentalizing knowledge in this way, hypertext may actually be a hindrance to the incorporation of text and context, book and history, into a single cohesive whole. Moreover, when Smith's students were able to make thematic links from one element of the hypertext to another, it was very rare for them to follow an entire series of links--which suggests to Smith that hypertext links can limit the breadth of vision which can and should take place during the study and appreciation of literature. Rather than thinking about the book and its context as a whole, from which lessons and conclusions could be drawn, students were reduced to single leaps of connection and logic which broke down after a single connection was made.

While McGann stresses the ease with which hypertext readers may navigate from text to text, and the parallel nature of the multiple editions available in his projected archives, each of those texts must be maintained in a separately enshrined, atomistic page, divided from its surrounding links by the designs of its editors. Each of those pages is, further, itself a form of idealized representation of a specific document. In A Critique of Modern Textual Criticism, McGann condemns the practice of replacing a manifold text's history of social and historical variety with an idealized notion of an original, absent ''work.'' But in digitizing his documents, McGann promulgates a similar sort of idealization, inasmuch as he is merely presenting an imperfect representation of an absent document. It could be argued that all McGann has done is widen the scope of idealization (at the level now of ''documents'' rather than ''works''), instead of doing away altogether with the transcendentalizing gestures of the Greg/Bowers school of textuality.³

We might even contend that the indeterminacy of the electronic form, the fluctuating instability which McGann celebrates as the opportunity for interaction between text, context, and commentary in his essays, approaches the very sort of transcendentalism which he most reviles in the Greg/Bowers school of criticism. The very immateriality of a computerized file approaches a model of ideality. If, as he asserts, The Rossetti Archive can never be finished, can only exist as an imperfect ''work in progress'' (''Radiant Textuality''), then what is the location of the historical and socially constructed document which McGann identifies as the only true state of textuality? Considering that the Archive is not merely a free-floating construct, but has a stated purpose and goal toward which it aspires, the user is left with only one option: to consider each moment in the project's life cycle to be a ''snapshot'' of its existence in a discrete historical and social moment, an imperfect reflection which constantly approaches the project's stated purpose (and thus its identity). In short, the ''text'' of the Archive is constantly a shadowy reflection of an ideal ''work.'' If the telos of the Archive can never be achieved, then its telos (and hence its fullest identity) exists only as a transcendentalized, ahistorical ideal--an unachievable goal situated in the same governing position as the ''work'' in the Greg/Bowers tradition.

Finally, it must be noted that electronic reproductions of non-electronic documents (that is, digitized images or texts which reflect analog objects, such as those populating McGann's archives) always call attention to the fact that they are merely imperfect reflections of those ideal, absent objects, the original documents. Walter Benjamin's essay ''The Work of Art in the Age of Mechanical Reproduction'' (1935) provides meaningful vocabulary and conceptual architecture for my observations here. By acknowledging that it is an imitation, a copy, however exact, directs the viewer's attention toward its original. In effect, the more efforts are made to disseminate copies of the object, the more the ''aura'' of that object (its air of authenticity and antiquity, its nature as an ideal against which all copies must be evaluated) is increased. This is a centripetal suggestion, one which may direct readers away from the electronic forum in their search for stability and authenticity.

III. Social Resistance to Decentralization

In light of Benjamin's assertions, J. Hillis Miller and Stuart Moulthrop point out that the growing explosion of information (in all media, but particularly in electronic venues) creates a reactionary tendency on the part of readers to stress the familiarity and relative immutability of static texts, such as paper books. This impulse is particularly strong with respect to literature which, as an important repository of a society's identity, is a domain to which overwhelmed members of the society will turn for security, and which is consequently seen as most threatened by the information explosion. In our fin-de-siècle world, typified by informational and societal variety and indeterminacy, people's needs for dependable institutions are reinforced. Few will thank critics such as McGann who seek to thrust those very institutions into instability--or, as McGann might have it, who seek to remove the false illusion of stability from institutions which are already, and always have been, volatile.

For an example of the type of reaction I am suggesting, we might consider a typical World Wide Web search for a single poem, such as Keats's ''Ode to Psyche'' (1818).⁴ Search results indicate that several links refer to a copy of the poem located at a single archive, demonstrating the reproductive incorporation mentioned earlier which minimizes user interaction with the text. Most tellingly, at least one page on which the poem appears presents it without notes, with no indication of textual authority, edition history, date, or that there are any variants whatsoever. Apparently, in this Web page author's mind, there is but one ''Ode to Psyche,'' permanent and immutable. While we know that textual variants exist, they are unimportant to this editor's conception of the work, and thus are ignored in the desire, the need, to enshrine a stabilized literary icon.

This is precisely the opposite of what McGann proposes will be the result of his Rossetti Archive and similar projects, in which he attempts to return editorial function to the reader and society.⁵ Clearly the user in question is not so concerned with a recontextualization and rehistoricization of a single, chosen variant among many--instead, the user presents what he or she (and all the other readers who refer to that page) see as a single literary object, a recentralized ''work,'' straight out of the Greg/Bowers tradition.

It may be argued, of course, that literary scholars (the primary audience for McGann's theories, if not his actual public Web practice) are not commonplace readers, and will be less reactionary in their response to his conception of volatile electronic media and its destabilizing influence upon texts and textual criticism. However, most of these scholars' audiences, their undergraduate students and society at large, are commonplace readers, and may therefore react as Moulthrop and Miller predict. And if the text is, as McGann proposes, a socially constructed performance, then the conditions and reactions of the society which receives the text are of paramount importance. An audience unwilling to accept an ''unfinishable'' project on the very basis of its indeterminate nature is not one which will easily submit to McGann's project. In the second place, there is no reason to believe that sloppy textual criticism, or at least the reductionist, syncretic criticism which typifies the Greg/Bowers school, will not be as prevalent in hypertextual editions as on paper--perhaps more so in light of the relative unfamiliarity of the medium and its consequent perceived threat to traditional scholarship.

Finally, we have the certain knowledge that the electronic medium itself has the potential to introduce enormous variety into the process of reading, variety which is itself unsettling enough to produce the turn to familiar, stable texts which Miller envisions. What McGann calls the ''bibliographical code'' of a text is open to a great variety of interpretations in electronic format: various fonts, colors, links to other material, and other effects may be easily incorporated into the media. As McGann himself enthusiastically asserts,

Essays can present all their documentary evidence as part of their argument (in notes and appendices, or in electronic links to the original documents--as in the present essay). They can also exploit fully the use of illustrations and images, including video film clips, as well as audio clips. (''Radiant Textuality'')

In addition, if one accepts the paradigm of the World Wide Web as an exemplar of the hypertextual experience, then each reading will also be informed by all the intertextual links which precede, follow, and interrupt the reading. In the face of such potential bibliographic variety (and its potential for confusion), it is easy to suppose that the ''radiance'' of electronic media may itself produce a centripetalization of its textual elements. Both Benjamin's essay on the increased status of the original in direct proportion to the profusion of its imitations, and the readily observable confusion (and even fear) engendered by the bibiliographic variety of the Internet, suggest that such a move is inevitable.

IV. Centripetal Textuality

There is ample evidence that McGann himself is aware of this resistance to embracing an unsettled, decentralized text, and that his own practices fail to demonstrate or endorse his proposed theories. McGann's work on The Rossetti Archive, which he touts in ''Radiant Textuality'' as a textbook example of what he hopes may be accomplished in the unstable medium of the Internet, is due to be published on a CD-ROM by the University of Michigan Press. The very work which McGann forwards as a social production, incapable of ever being finished, is destined to be enshrined forever in a fixed condition, a document precluding the possibility of further modification.⁶

This might, in turn, be seen as a result of the conviction in our society (and our academy) that information which is not in some way crystallized, and then run through the concretizing and finalizing filters of the publication process, is useless, suspicious, transient, or otherwise unreliable. Web pages may give useful information, but considering the ease with which they may be produced, readers must always question their authority or, failing that, their stability. In the indeterminate universe of the World Wide Web, it is already difficult to ascertain the authority with which statements are made--it is very easy, for instance, to set up a Web site proclaiming that you are a chief engineer at NASA, or a professor at Oxford, whether or not you possess any such credentials. Even when sufficient authority is determined, there is the issue of impermanence to contend with: Web pages and even entire Internet domains can arise and disappear almost instantaneously due to shifts in funding, mergers, or merely changes of interest. As McGann points out, most Web-based projects must be considered as ''works in progress,'' a designation which simultaneously protects an author or editor from accusations that his or her work is incomplete, while signaling the reader or viewer that the material itself must also be considered unreliable. Basing one's experience of a text, one's professional communications, or any other significant part of one's work upon such shifting sands, is a dubious tactic at best--another reason that, while audiences private and professional may turn to projects such as McGann's to initiate their research, they will likely rely upon less capricious sources for their own final determinations.⁷ McGann's plan to publish his CD-ROM lends both authority and stability to his work, but critically undermines the very notions of textual fluctuation that the work promotes.

University of California at Riverside

NOTES

I am indebted to Professor Robert Essick of the University of California-Riverside, for his instigation and support of my research into the nature of current hypertextual theory.

1 Throughout this essay, I will rely upon G. Thomas Tanselle's distinctions between ''text,'' ''document,'' and ''work,'' as delineated in his A Rationale of Textual Criticism (1992). The term ''text'' here refers to the arrangement of words in a poem or other literary product, while ''document'' denotes the piece of paper, the actual physical object upon which those words are inscribed (the manuscript, for example). ''Work'' is a more slippery concept, but represents a large part of the basis of the Greg/Bowers school of criticism. Briefly, the ''work'' is the transcendental, ''ideal'' form of the literary product, the ultimate source of its identity, and the form to which all more or less corrupt texts aspire. So, one may have several variant ''texts'' of, say, Moby-Dick (1851), inscribed upon any number of ''documents'' (from original holograph manuscripts to common paperback editions), but all of which are reflective and identified as versions of the same ''work'': for example, the cultural icon which we call Moby-Dick, beyond any individual variants or versions. It is this final abstraction with which McGann takes greatest issue in Critique of Modern Textual Criticism.

2 For example, this endnote functions in much the same way as a hypertext link, drawing the reader (at his or her discretion) out of the primary narrative flow, making contextual informative commentary, and then returning the reader to the interrupted primary text.

3 This certainly seemed to be the case during my work with Professor McGann and several other graduate students on a hypertext edition of Mary Robinson's Sappho and Phaeon (1796) (currently part of ''English Poetry 1780-1910: A Hypermedia Archive of Critical Editions,'' available through the University of Virginia Electronic Text Archive mentioned in ''Radiant Textuality''). While the scope of our project was perhaps severely limited by our own inexperience with the medium at the time, it progressed along what I perceive as fairly traditional lines of textual scholarship, and produced what may only be considered, I think, a derivative reflection of the original document--an object we ''idealized'' and attempted to approach as closely as possible in our translation of the work into the electronic format.

4 My exemplar search was performed on the AltaVista search engine at http://www.altavista.digital.com, using the search terms ''Keats + Ode + Psyche,'' on 29 May 1997.

5 For McGann's explicit statement of his purpose, see the online introduction to his Rossetti Archive, available through the address in the list of works cited.

6 At their website, the University of Michigan Press editors anticipate that they will begin publication of the Archive ''using a combination of CD-ROM and online technologies'' in 1998. It is revealing to note that the letters ''ROM'' in ''CD-ROM'' stand for ''Read-Only Memory''--that is, material which may be read on a computer, but which may not be written to or modified. See http://www.press.umich.edu/digpub/rossetti.html.

7 Another important audience to consider here is that of university tenure committees. One of the questions which impelled McGann to write ''Radiant Textuality'' was: ''Would you accept a dissertation written in hypertext?'' University professors, I submit, are presently as unwilling as most of society to forego the stable, familiar terms of discourse in deference to a pluralistic, unstable format such as, say, a Web page. We have every reason to suspect that a dissertation or tenure committee would not respond favorably to a work which was submitted to them as ''unfinishable'' or a ''work in progress'' from which the submitter had attempted to remove his or her editorial presence. There is a good reason that most ''hypertext'' scholarly editions are, as has been noted, extraordinarily similar in structure and aim to their paper counterparts.

WORKS CITED

Benjamin, Walter. ''The Work of Art in the Age of Mechanical Reproduction.'' Illuminations. Ed. Hannah Arendt. New York: Harcourt, Brace, Jovanovich, 1968. 219-54.

Bowers, Fredson. ''Textual Criticism.'' The Aims and Methods of Scholarship in Modern Languages and Literatures. Ed. James Thorpe. New York: Modern Language Association, 1963. 23-42.

Havholm, Peter, and Larry Stewart. ''Computer Modeling and Critical Theory.'' Computers and the Humanities 30.2 (1996): 107-15.

Katz, Seth R. ''Current Uses of Hypertext in Teaching Literature.'' Computers and the Humanities 30.2 (1996): 139-48.

Landow, George P. ''Hypertext in Literary Education, Criticism and Scholarship.'' Computers and the Humanities 23.3 (1989): 173-98.

------. Hypertext: The Convergence of Contemporary Critical Theory and Technology. Baltimore: Johns Hopkins University Press, 1992.

McGann, Jerome J. A Critique of Modern Textual Criticism. Charlottesville and London: University Press of Virginia, 1992.

------.''Radiant Textuality'': http://jefferson.village.virginia.edu/public/jjm2f/radiant.html (also available in Victorian Studies 39.3 [Spring 1996]: 379-89).

------. ''The Rationale of HyperText'': http://jefferson.village.virginia.edu/public/jjm2f/rationale.html.

------. The Romantic Ideology. Chicago and London: University of Chicago Press, 1983.

------. The Rossetti Archive: http://jefferson.village.virginia.edu/rossetti/rossetti.html.

Miller, J. Hillis. ''The Ethics of Hypertext.'' Diacritics 25.2 (Fall 1995): 27-40.

Moulthrop, Stuart. ''Traveling in the Breakdown Lane.'' Mosaic 28.4 (Dec. 1995): 55-77.

Smith, Jonathan. ''What's All This Hype About Hypertext?: Teaching Literature with George P. Landow's The Dickens Web.'' Computers and the Humanities 30.2 (1996): 121-29.

Tanselle, G. Thomas. A Rationale of Textual Criticism. Philadelphia: University of Pennsylvania Press, 1992.