Suzi Gablik, in The Reenchantment of Art, reminds us that for some time now the focus of art has been on the individual acting alone, defying the gods, defying society (1991: 5). Even the reaction to this ethos in what has been called postmodernism retains this disconnectedness, or rather extends it, by proposing nihilism, the death that is final because it is individual death. Thus we deny to society the locus of what is to be called life (Gablik 1991: 40). In literature this disconnectedness has been possible only through cognitive dissonance, for every act of publication is an act of making public, of making to the world the gift of the textual object, and publication has generally been a team effort in any case, as it is a complex maneuver: author, editor, publisher, designer, printer and distributor have all been required. The Internet seems to offer a new field for the play of individualism in publication, yet it is the most communal medium (Leppert 1996: 7) so far devised, as the give-and-take of communication between authors and readers becomes what is known as a "thread" or single intertwining strand of textuality. Where there is community, there is tradition, that matrix of temporally conditioned expectations by which the past community maintains itself through the present into the future. Traditions in text design are migrating from other media to the Internet. This paper will examine the possibility that the book tradition will continue to influence design in the new medium, helping to orient readers to the text at hand.
Among the many texts appearing online are those which have appeared before in print and hold enough attraction for generations of readers to have become what are called classics. As the Internet grows, questions arise as to how best to represent classic texts online. Some, of whom Michael Hart of Project Gutenberg is the best known (Neuman 1991: 365), have advocated using "pure vanilla" ASCII code, the lowest common denominator of text encoding, so that no one need be left behind while those having more buying power move on to more expensive and complex technology. Others feel no effort should be spared in marking up texts for research, which may require a more sophisticated technology to use, but offers the best chance of producing new knowledge. In the forefront of this latter movement are the proponents of TEI, the Text Encoding Initiative (Neuman 1991: 367). TEI is an implementation of the capabilities of the Standard Generalized Markup Language (SGML), a coding scheme that permits in-depth analysis of the parts of a text. SGML is currently the best approach for providing scholarly electronic editions to those few scholars doing computer-based analyses of texts, but may be "overkill" for the production of popular editions. Fortunately, there is a middle way between these extremes, offering much of the simplicity of ASCII with a glimpse of the power of SGML: HTML (Hyper Text Markup Language). HTML is a subset of SGML designed for transportation of hyperlinked documents, graphics, sound files, and motion pictures via a network such as the Internet. Users who have never otherwise attempted computer programming have discovered the ease of working in HTML. There are now more than fifty million Web-accessible documents (HotBot 1996), of which over five thousand are classic texts traditionally presented in codex book form, ranging from Homer's Odyssey to James Joyce's Ulysses (Ockerbloom 1996).
With software of a new type called a web browser, one can now consult a rapidly expanding library of texts in ways not possible previously. A text-only browser such as LYNX, when combined with speech software, can read online text to a visually impaired user. Browsers have search capability, so that each instance of a given word or phrase in a given work may be located and studied in context; every online edition is thus also a concordance. Selected portions of longer works in the public domain such as an act of Hamlet or chapter of Lord Jim can easily be downloaded, reformatted, printed out and used in class packets. Thus, although text read from a monitor is not as legible as from paper, electronic text is useful enough to drive a movement to provide such access. As noted above, however, there is disagreement on how to go about this.
At Cornell University, the Library of Congress, and elsewhere (McClung 1997), experiments are going forward in presenting the original pages of classic print editions in electronic facsimile, much as has been done through microfilm technology. Such images may contain visual information, such as marginalia in the handwriting of previous owners of the scanned print copy, which cannot without significant effort be presented by TEI or HTML, but have their own drawbacks. A single page scan takes from ten to a hundred times as much memory as stored text, and is accordingly slow to transport over a network. Also, an image does not easily support text searches, though dual editions are in production that will do so. For the time being, then, for networked access, SGML offers the best choice for a scholarly edition, ASCII is still suitable for the widest possible dissemination, and HTML, with its increasingly diverse options for presentation design, offers a solution for attractively formatted teaching editions.
At the University of Oregon I am experimenting with design of electronic texts suitable for teaching use. They are of varying length and complexity, from Philip Sidney's pageant known as "The Lady of May" (Sidney 1996) to Edmund Spenser's epic Faerie Queene (Spenser 1995). As techniques become available, they are tested first on shorter works, and then, if proven useful, applied to longer ones. The "Lady of May" is a handy test bed. In effect a one-act play, it contains a variety of design elements. It also interests students of literature and history, as it contains decent poetry, much humour and some characterization, and was performed before Queen Elizabeth I, whose presence was material to the progress of its plot. There is also a liberal sprinkling of obscure Latin phrases throughout, which need elucidation; this helps to motivate experiments with hypertextual notation.
"The Lady of May", performed in 1578, would perhaps have been lost to posterity but for the fame of its young author, killed at Zutphen in 1586. Interest in his writings remained strong throughout the 1590s, and a version of the pageant was accordingly appended to the 1598 edition, and subsequent editions, of the popular and frequently reprinted prose work, Arcadia. A facsimile of the first page of the 1605 edition shows the style of printing of the time:
Figure 1: "The Lady of May", first page of the 1605 edition; Sidney 1605: 570.
The iconographic conventions of the page are, despite the intervening centuries, largely familiar to us and would also have been familiar to medieval scribes. The header or title is long by our standards but presents the sixteenth century compositor with an opportunity to set centred lines of descending type size and length, a usually pleasing effect harking back to manuscript book design. Also of venerable origin is the ornate initial letter at the head of the main matter of the text. Some of the conventions have been abandoned in the centuries since: here, j is not yet in general use, and is still represented with i, use of u for v within words, or use of v for u at the beginning of words, is still common. Spelling is richly variable: reape, sweete, formallie. Note also the convention of printing verse in italic. Throughout much of the history of text, as here, economics does battle with legibility for the upper hand in design: for legibility, white paper and black ink serve to produce sufficient contrast to distinguish type easily, and type design has improved over the black letter (gothic type) in common use a few decades earlier. On the other hand, economics has led to the use of smaller type sizes  than formerly, and little white space is left between the blocks of text. The wide margins (not shown) are a product more of technological than aesthetic considerations: the presses could not produce more than four folio pages per impression, so these had to be grouped together in a rectangular pattern near the centre of the sheet to facilitate a consistent impression from the available hand-operated wooden press. When folded and sewn, the folios show a text-page closer to the gutter than to the outside of the page, and closer to the top than to the bottom.
Over the years, as printing technology has changed from hand platen presses to mechanically driven rotary presses and on to photo-offset lithography, the technical requirement for traditional page placement has been rendered obsolete, yet the tradition itself has force, and non-traditional designs, while offering a momentary exhilaration of freedom, generally have become quickly dated, while the old standard page remains. This may be due to what George Landow calls the rhetoric of arrival and the rhetoric of departure (1991: 82). Communication depends for success on the relative absence of elements that have little or nothing to do with the idea to be communicated; standardization of textual elements, from grammar and style to the use of titles, headings, running heads, and folio numbers, is intended to reduce the energy expended by the reader in extracting information from the page (Carlson 1989: 6). Serious deviations from tradition pose problems for the reader. The text in Figure 1 is four centuries old, yet we know our way around in it; it is familiar territory and we know where to enter and where to exit.
Figure 2: "The Lady of May", opening page (ed. 1860)
Figure 2 shows the opening page of "The Lady of May" in a copy of The / Miscellaneous Works / of Sir Philip Sidney, knt. / With A life of the Author and Illustrative Notes / By William Gray, Esq. / Of Magdalen College, and the Inner Temple, dated 1860 (Sidney 1860: 265). The rhetoric of scholarship in the nineteenth century frequently called for florid titles and titled editors. Editing, at that time, implied active and frequent intervention; the effects of Mr. Gray's heavy hand can be seen. Note the updated spelling on the page and the changed punctuation, especially the exclamation points sprinkled throughout. The passage used in the sixteenth century for a title has been dropped into an introductory note in six-point type, and a new title, which has come into usage in the intervening years, is offered without explanation. The archaism of the piece is marked by its being given an ornate dropped initial, as before, and also a headpiece, a device which was used in Sidney's time though the folio editions of our text lack one. The editor seeks to isolate the narrator's voice in small type, giving to the text more of the conventions of a play-book than its original warrants. These changes might loom large in the mind of an editor, but to the average reader, not much has happened to the text. Page design basics have not really been tampered with. The compositor in the shop which produced this artifact lived by much the same rules as those who produced our 1605 edition. The technology also remains relatively little changed after 255 years: the book is laboriously composed in hand-set types as before, though the press in this case is probably a rotary press, powered by steam, using stereotype plates produced from the galleys of type, an economical advance in book production.
Figure 3: "The Lady of May", Cambridge UP edition from 1962
Figure 3 shows a Cambridge UP edition from 1962 (Sidney 1962: 205; actually the Feuillerat edition of 1912); 102 more years have passed, and editorial conventions have changed, albeit very slowly. The publishers caution that the editor had not the best copy texts or manuscripts available, yet reprint his assertion that the text is reproduced without any deviations from the originals in the matter of spelling or punctuation. The page before us shows that much of the spelling and punctuation has in fact been restored, though the sixteenth century usage of u for v within words, v for u at the beginnings of words, and of i for j throughout, have not. Feuillerat is uncomfortable with the late title ("The Lady of May") but evidently feels it must be included, and so it appears here within square brackets, and as a running head. This shows that by 1912 the instability of text has been noted (McGann 1991: 182), and editorial practices instituted to stem the flow of blood, so to speak. The first three lines of the sixteenth century title have been restored, centered, in descending type sizes, perhaps as a bit of archaism to set the mood, but the rest have been moved into page-width justified text, as if they were the beginning of the main matter. The main matter is still signified, however, by the use of a dropped initial -- no longer ornate, but simply a larger type size of the same font. Gone is the headpiece, and the entire design has been constructed from a single typeface in various sizes depending upon usage. Although it does not show here, this is a scholarly edition in that variants have been recorded; they are appended at the end, by page number and by line number on the page.
The printing technology in use here is offset lithography, albeit from plates photographed from the edition of 1912, which was composed on the Monotype machine, with titles set by hand, and printed by letterpress from stereotypes as in 1860. That the Cambridge publishers have reprinted, with only their own preface and a bit of Feuillerat's, the entire 1912 edition is a sign that the overall book plan, page design, and type style are deemed adequate for a new printing after fifty years! The agreements on arrivals and departures made centuries ago between those responsible for producing texts and those who read them remain extremely stable, even though lithography makes possible complete freedom in page design, and regardless of the instability of text itself.
Lithography gives a blacker, more uniform letter, and is the technology of choice for conversion by scanning; Figure 3 yields a more legible image than Figure 2, and would produce fewer errors for OCR (Optical Character Recognition) scanning. Lithographic printing arrived in time to facilitate the wholesale conversion of canonical works into electronic texts and enable the use of new computer technology in humanities research. But just as technology has made conversion possible without the labour of retyping, so that new editions could be prepared whenever new scholarship became available, the very text pages that would most reliably support such conversion are frequently unavailable for this use, because they are still under copyright (Benedict 1995). Nineteenth century editions are very unlikely to give copyright problems, and for this reason are frequently found among the conversions that have appeared on the Web. If you wish to work from editions earlier than the nineteenth century, OCR becomes less efficient due to the older typefaces, irregularities in printing and stains on pages, and so forth. It may be necessary to type. This is the method which was used in converting "The Lady of May".
Figure 4: Introduction of "The Lady of May" to the computer age
Figure 4 shows the first of my efforts to introduce "The Lady of May" to the computer age. This electronic text edition  derives from the 1605 edition of the Arcadia (Sidney 1605: 570-576). Long s has been modernized, largely because it is unavailable for ASCII anyway, and catchwords and marginalia have been removed. Sixteenth/seventeenth century usage of i for j and of u and v has been retained, along with the original spelling. A few errors have been emended within brackets. Many italics, such as those used for proper names, have been omitted. Endnotes are indicated within braces. These are editorial decisions that are relatively little influenced by the medium at this point, because this is not a hypertext edition, nor is it in one of the formats determined by the visual iconography of desktop publishing, such as TEX or Postscript. This edition is Michael Hart's "pure vanilla ASCII", the basic character set devised originally for the 8-bit computers of the 1960s and 1970s. Aesthetically speaking, it is a bit of a step backward. Depending on the monitor available (mine had yellow characters on a black background) the viewer would see generally eighty columns by twenty-four lines of a fixed width font resembling Courier. Visually, the text of Figure 4 is not at all as attractive as the print editions of the previous three hundred and ninety-five years. Yet it represents a significant advance. Anyone can now take up a floppy disk containing the file storing the edition, "may.txt", and do a wide variety of things with it. Even the simplest word processors can do word counts, character counts, and what are called "string searches", in which a set of characters can be located successively in each of the contexts in which it occurs. This was the primary use envisioned by the creator of the original etext, Roberto Busa, working in the 1950s on St. Thomas Aquinas (Raben 1991: 343). Concordancing and linguistic software can do even more, and the file can also be converted into practically any format that has been or is yet to be invented, including typesetting for paper book production.
"May.txt" was not typed from the edition of 1860, as that edition seemed to me to have taken too many liberties with the text. Nor was it typed from the edition of 1598;  the 1605 edition was available on microfilm, and I worked directly from photocopies taken from that microfilm. It was impossible for me to produce a truly scholarly edition, as I had no resources for comparing early editions, let alone copy-texts, on my own; others had done this, and will continue to do this, better than I. I felt, however, that I had two new contributions to make: that I could competently produce a contemporary introduction to the text, exploring rhetorical issues raised by the work of the new historicists, and that I could produce an electronic edition which, while it might never be the best text, would be one of the first Early Modern English works to appear in the new medium, useful to students all over the world who might not otherwise have access to it in paper form. I believe those aims were achieved, as I have received electronic mail from students (and scholars) in many countries who seem glad to have had access to this and some twenty-four other texts that I have similarly produced (Bear 1996).
Not everyone is happy to see texts like "may.txt" appear on the Internet. Periodically, a thread of heated discussion erupts in online seminars such as HUMANIST or SHAKSPER as to whether such texts are useful editions. The gist of the objections is that effort should be expended primarily on work produced in circumstances like those of peer review in academic journals, such as on authorized editions for which they themselves will do the authorizing, the primary use of which will be for scholarly analysis. Even granting this authority, however, will not prevent the appearance of publications aimed at being read.
"May.txt" actually is constructed a bit like an MLA-style term paper, with a general title, note on the text, introduction, text, notes, and bibliography, in sequential order. As it is an electronic text, it is searchable, downloadable, printable, and readable, but its readability (with the exception of its being readable by speech software for the vision impaired) is its weakness. Scrolls were replaced by codex books, beginning about AD 400, at least in part because scrolls were accessible only sequentially. Codex books permit random access, so that a reader may readily consult a particular passage (Manguel 1996: 127). Computers permit even better access to the information in a text than does the traditional indexing and pagination of the codex book, but many, many users have not yet developed the skill in using search software. The tendency is to "scroll" down through the text, a behaviour that is a throwback to the rolled book, and one of the primary causes of the perception that electronic texts are not convenient. Also, the 24 rows by 80 columns format lacks aesthetic beauty, which has always been a consideration in the engagement of the reader with the text. It must be made more accessible, and more attractive, if it is to gain acceptance in the reading community.
The advent of the World Wide Web presented new iconic possibilities for those who seek to produce reading and teaching editions. This was not immediately so (McLaurin 1991). HTML was originally devised with only one graphic design model in mind, that of the text outline with nested levels of headers, which is the model most familiar to the creators of software manuals. An outline is easily converted into a concept map and vice versa, as it is a hierarchical and sequential linking of concepts. Thus, a visual flowchart of elements could be reduced to a pre-organized verbal model for teaching new users to master a program. This is a powerful paradigm for information transmission among the hard and social sciences, as the iconic page that appears on-screen conveys an impression of a single culture united in a belief in causation. But not all "information providers" share the rhetorical project of the sciences, and their design goals will accordingly differ. Users were not satisfied with a Web consisting of an infinity of links between black-on-gray outlines. Images, which the author of HTML envisioned would be mostly photographs and charts exchanged among scientists across the variety of computing platforms in use, began immediately to be used as design and even typographic elements. Sensing an opportunity, Netscape Corporation in 1994 leaped ahead of the committees that had been entrusted with the development of HTML protocols, and introduced codes for, among other things, centring of text and font sizing (Netscape 1994). Web page designers seized upon the unauthorized codes immediately, and unauthorized texts of classic works began, within months, to clothe themselves in the graphic elements of traditional book arts, acquiring the rhetoric of those arts without, so to speak, having to pay the hitherto-required dues thereof.  Combined with the hypertextual powers of HTML, the new design elements created a workable tool for re-presenting texts in an entertaining and informative telecommunications environment that rapidly gained popularity (Ockerbloom 1996).
Figure 5: "The Lady of May" formatted in HTML
Figure 5 shows how my text of "The Lady of May" responded to the early design opportunities of HTML. A list of links, or a kind of interactive table of contents, whisks the reader to the text-matter of choice: introduction or main matter or notes or bibliography. Notation no longer merely refers to notes, but puts the note on-screen. A click of the back button returns the reader to the context in which the notation appeared. Typographic design is now possible, and the ritualistic richness of the Elizabethan title is restored in full. The earliest version of this file accepted the default background colour of the browser (usually gray); it now specifies white (hexadecimal #ffffff), as do many other books in cyberspace (for example, Project Bartleby 1996). Margins also have been restored, in imitation of the white space around the text that had been dictated by printing technology in the days of hand-operated presses.
As HTML has evolved, more and more formatting control has gone to the information provider, with mixed results. Many, new to the idea of access to publication design, have based their documents on print advertising and television commercials, the design of which reflect massive research into capturing attention (and money). Pages have gaudy graphics, with blinking text, animated GIFs (Graphic Interface Format), tiled background images, banners, Java ant-races, and clashing text colours, often masking the absence of significant content. At the other extreme are government documents containing millions of characters of content, with no more structure than nested headers for chapter, section, and paragraph, all on the same gray background. Those working with classic texts, however, tend to be aware of the rhetorical power of traditional book design, and their ideas on the use of HTML to translate this power to the Web appear to be convergent. 
Traditional page design is still authoritative because of its familiarity. The reader, reassured as to the points of arrival and departure, is free to concentrate on the matter being communicated. There are, however, other traditional models besides those we have considered up to this point, and some of these might be worth examining as we consider the future of electronic text. Elements of HTML not previously available will make it possible for notes and glosses to pop up on-screen when their key, or referent, is clicked (as for a hypertext link). It is already possible to foreshadow this technique by using frames -- more than one window opened by the browser at one time.
Figure 6: "The Lady of May" designed for a teaching edition
Figure 6 (Sidney 1996) shows one way in which traditional design and frames can be brought together to create a teaching edition, with an interactive sidebar. In Figure 6, we are at the point of arrival. Both the page title and the HTML header proclaim that centuries-old title of the piece not found in the original, orienting the modern reader upon arrival. The paragraphs following the header explain how to use the notes, introduction, and bibliography, and make available a non-frames version of the text, for older browsers and LYNX. I have retained white as the background colour of the main matter, along with the brown and green links, but the Notes window has (at least on the monitors I've checked) a tan/ivory background and a (very) dark blue text colour, to distinguish it easily from the main matter.
Much of the information once found at the head of the file, including text source and acknowledgments, I have moved into the note that appears when the text is first accessed. Other links previously found near the top have been moved to the bottom of the file in the main (left) window, as newly arriving readers are apt to click on these if they see them first, leaving the frames environment abruptly and becoming disoriented.  The aim here is to make a variety of reading and study strategies available to the reader without unduly distracting from the narrative continuity of the text, an aim that echoes that of the modern pedagogical codex textbook.
Figure 6 serves as our point of departure from this present narrative, presenting an edition of "The Lady of May" separated by four centuries of technology and editorship from the first edition, yet retaining an iconic likeness that is not accidental. The rhetorics of arrival, of presence within a text, and of departure still follow an ancient law: that of conservation of energy. In social behaviour, such conservation is known by the name of tradition, and we are may do well to carry our traditions with us when seeking out new lands.
 I am aware that there is no inherent virtue in samizdat publishing of classic literature; my own first outing, probably still available in a few places as "ballads.txt", was a disaster of poor proofreading, and I was rightly taken to task in a number of postings to scholarly lists for releasing it in such condition. With freedom comes responsibility, and the publisher must rise to the occasion or leave well enough alone, I believe.
 While I do not have space here to demonstrate this assertion exhaustively, I suggest that the reader examine a few texts accessible from the Online Books Page (Ockerbloom 1996); the Columbia U's Project Bartleby 1996 offers a particularly fine instance. Or see an egregiously obvious instance: Milton against a background of a right-hand book page, with the actual gutter at far left (Milton 1991).
 These changes were suggested by beta testers of the file, who responded to a request for participation posted on three Internet discussion groups (also known, somewhat inaccurately, as listservs): RENAIS-L, SPENSER-L, and HUMANIST.