1994 Prospectus: Creating an Electronic Archive of Piers Plowman

The Nature of the Problem

The culturally important texts of the medieval past survive only in defective and fragmented handwritten documents. The production of critical texts from these imperfect documentary matrices is the most fundamental task of medieval scholars. Each generation of scholars makes new contributions to this ongoing historical project. At present, as a result of computer technology, medievalists (like other scholarly editors) stand on the verge of a major reconceptualization of the nature of text. We are as eager as printers in the age of incunabula to exploit the speed and reliability of the new technology, the unprecedented advantages of the new medium. But we are perhaps more wary than those Renaissance printers about the consequences of exploiting a new technology. We do not wish to sacrifice the specialized philological and editorial skills and judgments appropriate to the study of handwritten documents and complex textual traditions. We want to retain for the discipline those skills that have been laboriously learned in centuries of print editions. Many students and scholars, we recognize, are likely to remain people of the printed book. To that end, we expect eventually to accommodate such readers by producing a traditional printed critical edition derived from the more spacious electronic text. Our primary goal, however, is to advance the reconceptualization of text while retaining the scholarly rigor of the print tradition in order to make specialized philological and editorial skills and knowledge an integral part of creating the texts of the twenty-first century.

During 1993-1994, as a Fellow at the Institute for Advanced Technology in the Humanities, I have been the representative of the Piers Plowman Electronic Archive, a collaborative effort to create a multi-level, hypertextually linked, electronic archive of the textual tradition of all three versions of the fourteenth-century allegorical dream-vision Piers Plowman as well as documentary editions of the scribally created splices of the A and C versions. Robert Adams, Eric Eliason, Ralph Hanna III, Thorlac Turville-Petre, and I are convinced that Piers Plowman with its extraordinarily perplexing textual situation, only slightly less complex than that of the Greek New Testament, is the ideal English vernacular text for developing a model for computer-generated editions of medieval texts. We also believe that editorial collaboration assumes new meaning in relation to electronic technology. Editors of electronic texts, unlike earlier editors of printed editions, need not suppress or conceal editorial disagreement nor impose spurious notions of authority. They may, instead, exploit editorial disagreement and embrace the provisional nature of scholarly editing.

The Historical Setting of the Textual Problem

William Langland's Piers Plowman is a fourteenth-century alliterative, satiric, allegorical quest poem. Instantly popular among literate city tradesmen and county gentry as well as among sophisticated ecclesiastical consumers of Latin theological and devotional works, Piers Plowman survived the Catholic world whose spiritual values and strivings Langland had mirrored, maintaining its popularity after its Middle English had ceased to be spoken. Its popular appeal in its own time is reflected in John Ball's Blackheath sermon and in his letter to the men of Essex in the Peasants' Revolt (1381) exhorting them to "do welle and bettere." Its appeal to city and court is evident in Thomas Usk's reference to the poem in his Testament of Love (c. 1384-5). Through the remaining years of the fourteenth century and well into the fifteenth, other alliterative poets echoed Langland's reformist concerns. Three sixteenth-century printings by Robert Crowley suggest how readily Langland's depiction of the spiritual strivings of a fourteenth-century Catholic world was adapted by succeeding generations until finally it was rebaptized as a proto-protestant poem, its reformist arguments taken to have anticipated and hastened the advent of protestant England. As such, it was known and loved by Spenser, Milton, and John Bunyan, whose Pilgrim's Progress would have been a different work had Piers not survived the fourteenth century. Of its era, only Chaucer's work has had a greater impact or a more lasting.

As we have noted, the recovery of an authorial text of Piers Plowman poses editorial problems only slightly less complex than the problems of editing the Greek Bible. Langland himself composed three versions of his poem (conventionally referred to as A, B, and C) during a period from the late 1360s until his death, perhaps c. 1390. [1] Of the 54 surviving manuscripts, none is an autograph, none can be dated certainly as of the poet's lifetime. The poem's very popularity poses a grave difficulty for modern editors: all of the surviving medieval copies are full of errors, both subtle and blatant, serious and trivial. Like most other medieval poets, Langland lacked control over the reproduction of his work. Enthusiastic early readers often produced inexpert copies for their own use. These copies became in turn the bases for yet other copies, with each copying accumulating fresh errors, conjectures, "corrections," and contamination within and between versions. Authentic lines were garbled or omitted. Inauthentic lines were introduced when scribes acted as amateur, self-taught editors, sometimes mixing lines from the poem's three authorial versions and occasionally adding words or lines of their own.

However, it is not simply the problem of the naive, careless, and corrupting scribe that accounts for the state of the Piers Plowman text. The degree of intelligent reader participation is almost unparalleled in the period of handwritten books. Some early readers and copyists were obviously aware that the poem existed in different versions and, wanting to produce the best version available, they conflated manuscripts and versions. This "contamination" is exactly what makes the poem's textual situation impossible from a traditional recensionist point of view. As a result, obstacles to the recovery of an authentic text exceed those inherent in editing any other English vernacular document of comparable length.

A Brief History of the Modern Text

For the past century and a half, major scholarly efforts have been mounted to provide accurate and authoritative texts of all three versions of Piers Plowman. In 1885, after twenty-odd years of heroic labor, the Reverend Walter W. Skeat produced an Oxford University Press parallel-text edition, the basis for Piers Plowman studies until 1960.[2] Skeat reliably determined which lines belong to which of the three versions and established the sequence of composition. Some of Skeat's beliefs about the poem, however, are no longer tenable. For instance, he believed that the Laud manuscript of the B version might have been the poet's autograph, a mistake which led him to rely too heavily on its authority. Skeat himself acknowledged, in a private letter in 1909, that many erroneous readings remained in his texts and that much work remained to be done. He noted the new principles for dealing with highly contaminated manuscript traditions enunciated by B. F. Westcott and F. J. A. Hort in their edition of the Greek New Testament, published about the same time as his own parallel-text of Piers Plowman.[3] Wescott and Hort had made it seem feasible to restore an authorial text from garbled and contaminated manuscripts by employing a whole host of heuristic principles inferred from observations of tens of thousands of instances of actual scribal behavior in the copying of the gospels and epistles. Adapting these methods to Piers Plowman, R. W. Chambers, A. G. Mitchell, and J.H.G. Grattan began re-editing Piers Plowman during the first four decades of this century.[4] They began to transcribe and collate the manuscripts. Two major European wars, however, prevented their bringing the edition to fruition, and George Kane, one of Chambers's most brilliant pupils, inherited the project in 1946.

Over the next decades, Kane and two collaborators, George Russell and E. Talbot Donaldson, transcribed and collated all the known manuscripts in a renewed effort to produce texts closer to Langland's original versions than Skeat had attempted. In 1960, Kane's magisterial edition of the A-Text, the earliest and shortest version of the poem, appeared to almost unanimous scholarly acclaim.[5] Fifteen years later, the Kane-Donaldson edition of B was published; this time, however, scholarly reaction was mixed.[6] Literary and textual scholars voiced dismay at the editors' tendency to discard distinctive stylistic features and many of the unique readings of the archetype (the no longer extant original of all the surviving B manuscripts) in favor of different, homologous ones drawn from the A and/or C versions. Although some of that criticism was simply ill informed, significant questions were raised. Some reviewers worried that Kane and Donaldson had attempted to disguise the embarrassing possibility that Langland's second effort, like Wordsworth's revised Prelude, was not uniformly better than his first. Less charitably, certain reviewers speculated that the editors, by privileging the readings of their reconstructed A text over the B archetype, were unconsciously attempting to forestall criticism of Kane's choices among variants in the A textual tradition.[7]

In the meantime, Derek Pearsall has published what he called an "interim statement," a semi-best-text, semi-eclectic edition of the C text, a provision until such time as the Kane-Russell C edition appears. [8] A. V. C. Schmidt has produced an alternative edition of the B text, apparently basing his work upon Kane and Donaldson's collations of the B text, and has announced his intention to publish a parallel-text edition.[9]

II. The Piers Plowman Electronic Archive

Work on Piers Plowman has traditionally been in the vanguard of Middle English editorial practice. In this historical context, we propose an innovative editorial archive that will take advantage of the electronic revolution in textual studies. In addition to creating the kind of edited critical text enabled by print technology, we plan to offer in electronic form all of the significant pre-modern textual evidence relevant to the poem.

It would be hard to imagine an English literary text that would profit more from this revolution. Degrees of accuracy and consistency formerly unattainable by even the most scrupulous editors are now within reach. Our editorial decisions will not depend on hand-produced collations. Our analysis will not be hampered by hundreds of sortings, marred by inevitable slips of eye, attention, and memory. After we have made our initial transcriptions and begun to deliberate on the text of the archetype and the critical version, we will have nearly instantaneous access to eight different manuscript versions, to concordances of each manuscript version as well as to the archetype and to machine-readable texts of modern editions of all three versions. In addition, we will have access to electronic texts--prepared over the past dozen years for my metrical researches--of all the other major Middle English alliterative poems. We will, as well, have access to electronic concordances to each of those texts. Moreover, we have designed our project so that transcriptions of the remaining manuscripts can be added to the Archive as they become available.

We can promise readers of Piers Plowman a text substantially better, more true to what the poet wrote than any of the editions now available. To scholars and editors, an electronic text offers unprecedented advantages. Unlike earlier, printed critical texts, the electronic text will permit manipulation of the individual manuscripts, the reconstructed archetypes, and the critical texts on a CD ROM disk. Its design will provide for adding manuscript, historical, linguistic and other kinds of evidence as well as individual concordances to each manuscript, to the archetypes, and to the critical texts. The extremely plastic nature of an electronic text is conducive to representing the actual textual tradition of Piers Plowman, one filled with an unusual degree of ambiguity and uncertainty. An electronic edition can accommodate scholars who prefer a "best text" documentary edition as well as those who want the best possible modern editorial reconstruction. In the world enabled by electronic texts, theoretical differences formerly thought to be irreconcilable need no longer prevent dissemination of textual evidence, as in the unfortunate case of the Dead Sea Scrolls. Unlike printed editions which historically have presented themselves as "authoritative," often excluding editorial uncertainty or suppressing disagreement among editors, our electronic text openly embraces the provisional nature of scholarly editing.

We claim no permanently definitive status for our critical edition. Its construction will serve to exhibit many of the possibilities of the new technology. We can, nevertheless, make permanently available in electronic form all the manuscript texts upon which critical texts must be based, thus facilitating the creation of yet other editions. As recent consideration of the metrical structure of the poem has revealed defects in earlier editions, so future discoveries will prompt other editors to construct different theories of the text. The greatest part of our editorial work--transcriptions and collations in particular--will establish a data base permanently useful to both editors and literary critics. Such work--because it can be absorbed into future critical editions--need never be replaced. Students and critics of the future can interrogate such an Archive in ways unimagined by its creators.

Uses for the Archive

We expect that students of late medieval literary culture, both novices and experienced professionals, will benefit from the Piers Plowman Archive. It should benefit historians, historical linguists, dialectologists, literary historians, paleographers, and others who are interested in the material production of books and the study of literacy.

The reception of an important cultural document such as Piers Plowman is to an unusual degree reflected in its manuscript copies. In the Archive, these manuscripts will be represented both by electronic color facsimiles and by transcriptions marked with SGML. These representations will provide swift, non-impressionistic ways to observe and compare the manuscripts. Color electronic facsimiles will permit paleographers and other textual scholars to enlarge and enhance discrete items on the page or to subject portions of the text to color or gray-scale analysis. Teachers of paleography and their students will use the color facsimiles to study scribal patterns of abbreviation and suspensions or to compare different styles and modes of layout and page composition or ornament. Historical linguists and dialectologists will be able to search and manipulate a hyperlinked matrix of several million word forms derived from several dozen dialect areas for studies of syntax, morphology, phonology, orthography, and lexicography. Lexical collocations, every one of them, can be searched and subjected to analysis. Metrists and linguists who study stress patterning will be able to copy texts and mark them specially for their own searches and tabulations.

Recent interest in the history of literacy and the material production of medieval books has focused renewed attention upon scribal behavior. Everywhere in the manuscripts of Piers Plowman is plentiful evidence that its scribes took an unusual degree of interest in this text which served a variety of functions for a variety of factions. Though it would be easy to over-emphasize the role of the scribes as critics and interpreters, the early reception of the poem is nevertheless extant primarily in their interventions and annotations.[10] The structured database of the Archive offers students of scribal behavior unprecedented access to efficient comparisons and accurate statistical compilations. Once the archetypal text has been reconstructed, cultural historians and textual scholars can efficiently gather evidence on rates and forms of scribal error, contamination, substitution patterns, coincidental variation, and so forth.

The Archive also offers new possibilities for literary analysis. Professor Larry Benson at Harvard, among others, has begun to demonstrate some potential uses of the computerized text for literary criticism. He has prepared a fully lemmatized electronic concordance for his Riverside Chaucer with a sophisticated set of search and analysis software tools. Other scholars are preparing similar lemmatized concordances to Gower's Confessio Amantis and the works of Hoccleve. Professor Benson, in a sophisticated comparative analysis of word frequency counts in Chaucer's "Man of Law's Tale" and in Gower's version of the same story in the Confessio Amantis, provides a splendid model for literary study informed by computer research.[11] Structured data bases offer increased precision to the observation of Langland's poetic style. The possible uses of the Archive for literary critical analyses are limited only by the imagination of the critic.

Finally, new classroom applications and new bases for teaching editions and translation will be available as a result of the Archive. Electronic facsimiles will bring immediacy and new techniques to the study of poetic style, of paleography, and of the history of the language. Students, teachers, and editors who work at a distance from major manuscript collections will find the Archive a useful resource.

History of the Archive

Robert Adams and I first began to discuss re-editing the three texts of Piers Plowman when defects in the Kane-Donaldson edition became evident. My work in 1986 on the metrical structure of the alliterative long line and later of the metrical structure of Langland's distinctive form of the long line convinced me that Kane and Donaldson--like pre-Tyrwhitt editors of Chaucer--had worked in essential ignorance of Langland's meter, misapprehending both his alliterative practice and the rhythmic structure of the halfline. As a result, in both the A and B versions, they sometimes rejected authorial readings and replaced them with unmetrical and unauthentic lections or conjectures. Moreover, their list of variants did not steadily record metrically substantive variants. Other scholars during the same period of time pointed to yet other difficulties with the theory and execution of Kane's and Donaldson's B text. After their years of intense scholarly labor, we still lack an authoritative edition of the B version of Piers Plowman.

From 1987 to 1989, Adams used a relational database program on his Macintosh computer to collate segments of the B text, elaborately comparing the lections within the B tradition with those of the A, C, and Z versions. Though his program did not permit a sufficiently rapid collation to make the project feasible, his preliminary set of collations showed very clearly that the B archetype could be constructed from the eight manuscripts we propose to transcribe initially. Moreover, it convinced both Adams and me that the B archetype often revealed a rationale for Langland's revision that had apparently eluded Kane and Donaldson. In focusing too narrowly on differences and agreements among the A, B, and C versions at the level of the half-line, Kane and Donaldson often failed to account for possibly authorial changes at the level of the sentence or paragraph. At that point, Adams and I agreed to edit approximately a tenth of the poem to demonstrate to ourselves how such an edition might be done.

We purchased microfilm copies of the B manuscripts and the more important manuscripts of A and C to provide a basis for our investigations. I continued work upon the morphology of Langland's dialect; Adams, who continued to collate the B manuscripts and to compare them with the other versions, derived a theory for editing the poem recently elucidated in his 1992 article in Studies in Bibliography. [12] We also began to talk with Eric Eliason, Ralph Hanna, and Thorlac Turville-Petre about the possibility for combining our efforts and knowledge to try to develop a radically new mode of editing, one that answers simultaneously to the demands and needs of conservative "best text" editorial theorists and to interventionists. The five of us met at the Kalamazoo International Congress on Medieval Studies in May, 1991, and mapped our strategy.

Since then, we have worked as a team on various aspects of editing Piers Plowman in an electronic form. We have investigated various software options for manipulating documents, finally settling on Peter Robinson's program COLLATE. In June, 1991, I conferred with Marilyn Deegan and Peter Robinson at the Oxford Computing Centre and enlisted their enthusiastic support for the project. Both have served as consultants on all aspects of our editorial work. About the same time, Hanna did a trial edition of eighty- four lines from passus V to satisfy himself that he could reconstruct the archetype and recover a more probable text than that of the Kane-Donaldson critical edition. Adams and I began late in 1992 to transcribe the 353 lines of passus 3 in manuscripts CFGHmLMRW. We have collated them with COLLATE and have begun the process of preparing a sample reconstruction of the B archetype. My work on the Archive is discussed more fully below in Section III.

Materials and Methods for Reconstructing the B Archetype

The manuscripts of the B text of Piers Plowman have been catalogued and described over more than a century of scholarly editorial and interpretive labor. Barring discovery of new B manuscripts, the fundamental bases for an edition of the poem have been identified and are readily accessible, both the originals and photocopies. We have collected microfilm copies of the manuscripts and have begun to prepare and annotate color digital facsimiles of four of the following manuscripts:

C Cambridge, Cambridge University Library: MS Dd.1.17

F Oxford, Corpus Christi College: MS 201[13]

G Cambridge, Cambridge University Library: MS Gg.4.31

Hm, Hm2 San Marino, Huntington Library: MS 128

L Oxford, Bodleian Library: MS Laud Misc. 581, (S.C.987)

M London, British Library: MS Additional 35287

R London, British Library: MS Lansdowne 398; Oxford, Bodleian Library: MS Rawlinson Poetry 38, (S.C.15563)

W Cambridge, Trinity College: MS B.15.17

In the first stages of the Archive we plan to provide editions of manuscripts of the A and C versions to facilitate comparison among versions. We have selected Trinity College, Cambridge, MS R.3.14, for the A version, supplying where necessary those lines omitted in that witness and correcting obvious scribal lapses. The C version will be represented by Huntington Library, MS HM 143.

Of the seventeen manuscript copies and Crowley's three sixteenth-century printed editions, only the eight listed above are necessary to establish the B archetype. To note that fact is not to claim that any manuscript evidence can on a priori grounds be neglected in the final stage of the editing project when the critical text is being established, but for establishing the archetype the eight manuscripts we have selected represent the major sets of affiliation among the manuscripts.

Seventeen manuscripts preserve the B-version of Piers Plowman in relatively uncontaminated form. None of these manuscripts is a direct descendant of any of the others, so each preserves an independent witness to their common ancestor (the B archetype), which was a copy already at some remove from the authorial text and already characterized by the presence of numerous errors. In addition, Robert Crowley's three printings of 1550, based on manuscripts that no longer survive, have some authority. On the basis of shared unoriginal readings and omissions, the witnesses may be divided into two major families, alpha (MSS R and F) and beta (MSS W, Hm, L, M, S, G, Y, C, O, C2, Bm, Bo, Cot, Ht, and H--as well as the three Crowley imprints [Cr1, Cr2, and Cr3]). Within the beta group, a large subset links together eight copies: G, Y, O, C2, C, Bm, Bo, and Cot. Of these, Bm, Bo, and Cot are the closest kin, probably produced in the same shop (c. 1410?) from haphazardly shared exemplars. Though among the oldest of the B-version manuscripts, these three copies attest to a text of B that is persistently inferior. Of the other five members of this group, the one most closely related to these three appears to be C, which, nevertheless, lacks some of their distinctive readings and shows no evidence of their tangled production history. Another persistent subset is that formed by MSS O and C2, which together witness to an inferior shared ancestor similar in many respects to the ancestor of the C(BmBoCot) group. In the early parts of Piers the genetic affiliations of G are obscure, but it seems most closely related in its later sections to MS Y; and both of these have closer affinities with the O/C2 subset than with the other four members of the larger group.

Beyond this major group, another beta subset is defined by MSS W, Hm, and S, together with the manuscript(s) inferentially behind Crowley's imprints. This subset typically reflects the B archetype more faithfully than the aforementioned group of eight. Relative superiority of text is also apparent in MSS L and M. Neither shows close affiliation with any of the other surviving copies (although M has been sporadically "corrected" from a source related to the WHmCrS group and appears related, by change of exemplar, to that group, in the last few passus of the poem). One other copy, MS H, attests to a B-version text from the Prologue through the beginning of Passus 5 but thereafter has as its exemplar a manuscript of the A-version. Its fragmentary nature, and the likelihood of its B section's having suffered sporadic contamination from the A exemplar, make it undesirable for our purposes.

In the light of these relationships, the eight manuscripts that we propose to use in reconstructing the B archetype seem ample, if not excessive. MSS L and M cannot be ignored because of their unique genetic positions. Two manuscripts from each of the large beta groups also seem necessary in order to guarantee an accurate fix on their respective ancestors. From the smaller of these two groups, W and Hm are intrinsically more valuable as well as more practical to work with than Cr and S. Both of the latter are quite late, with S showing considerable modernization of phrasing and diction and the Crowley imprints manifesting a confusingly incestuous relationship with each other as well as the likelihood of contaminated manuscript parentage. MSS G and C are employed to cover the comparatively rare circumstances where their large subset may have uniquely preserved the archetypal reading, but G (a sixteenth-century copy) also appears to have some independent value as a witness for the early passus. Finally, both MSS R and F, comprising the alpha family, must be used to establish the readings of their common progenitor in order to weigh them against the readings of the beta family in reconstructing the parent of both families, the B archetype.

Transcriptional Policies

Our first task has been to transcribe each of the manuscripts into ASCII files. Adams is about to complete a transcription of Hm, and Eliason has completed a first draft of C. Since June, 1993, my team of research assistants and I have made and proofread fully TEI-conformant transcriptions of manuscripts F, L, M, and W. Sean Taylor, a doctoral student at the University of Washington, working under the direction of Professor Paul Remley, has virtually completed an electronic edition of R, and if his finished and revised edition is accepted for publication by the Society for Early English and Norse Electronic Texts, he has consented to incorporate its text in the Archive. The last four weeks of my tenure at IATH in June will be devoted to transcription of G. The IATH editorial team will do three separate proofreadings of each transcription before each is dispatched to one of the senior editors outside the University of Virginia. We plan to have two of the senior members of the editorial team check the transcriptions against the photocopies (and when possible, the color digital facsimiles). At least one senior editor will read each transcription against the original manuscript. The transcriptions have been made with full SGML markup in accord with the guidelines of the Text Encoding Initiative.[14] Transcribers have provided annotations on the physical features of each manuscript, recording erasures, subpunctions, and other forms of deletion; suspensions and abbreviations; marginal and interlinear additions or corrections to the text as well as contemporary and later glosses; changes in scribal hand; and any information the editors deem relevant to potential users. Since our goal is to provide the lections of each manuscript in machine-readable form, we are in effect acting as conservative editors, presenting eight diplomatic texts of Piers Plowman. We will make no emendations, regularizations, or modernizations in these base manuscript files.[15]

We plan to provide a full codicological and partial linguistic description for each manuscript.[16] In the present state of the Archive only the physical description of F is available, and that is only partial and tentative, with full description to be written in the summer of 1994 with the manuscript in hand. Though as presently laid out on the page these descriptions look very like traditional ones in printed editions, they are in fact hypertextually linked data bases. Ultimately, users may search and compare and collate the descriptions of paleographic and codicological features on all 54 witnesses in ways difficult, if not quite impossible, with printed editions.

After the electronic transcriptions have been made and proofread, Adams and I will each independently run the COLLATE program to produce the corpus of variants from which we will construct the B archetype.[17] Using both computerized and conventional editorial techniques, each editor will have responsibility for two sections of approximately 1,400 lines. Consistency in this large task will be facilitated by our access to computer-generated concordances of all the manuscripts (and edited texts of all three versions) as well as by our ability to keep detailed, readily accessible records of editorial decisions. When these sections are complete, a third editor will collate them, registering agreements and reservations and suggesting additions or changes to the annotations. If, after discussion, editors cannot agree about the lection(s) of the archetype, they will write notes to record the exact nature of the textual difficulty. Eventually, all five editors will meet to discuss remaining difficulties and disagreements. We may reasonably expect that resolution of problems in one section of the text will cast light on disputed problems elsewhere. Though the primary goal of the editors during this stage is reconstruction of the B archetype, it is perhaps more important that whenever the editors disagree about the readings of the archetype, the nature of the problem will be fully presented with all the relevant evidence. Annotations to the B archetype will focus solely upon problems of determining its text.[18]

Effective representation of manuscript variants is a particular strength of an electronic text, since it will, when complete, permit a reader to see every variant from every manuscript by the simple expedient of calling up the line in a base text (which might be the critical edition, the archetype, or any other manuscript) and evoking the other manuscript readings in different windows or by consulting the output of COLLATE. Such a reader can also call up digitized facsimiles of the manuscript(s). Other filters can be set, depending upon the reader's interest, to include or exclude dialect divergences, metrically significant variants, purely orthographic differences, or simply those variants deemed by the reader to be substantive. Readers can focus as narrowly or broadly on variations among the manuscripts as their individual research interests dictate. They may choose "significant" variants or "all" variants (see sample collations). In order to achieve these results, of course, all the texts (and their textual and critical annotations) will have to be synchronized so that the reader at any one point in any text can readily access the same point in all the texts on the disk. Those hypertextual linkages are not available at this point in the project, though transcriptions have been prepared with a system using SGML markup for passus and line numbers to provide a base for them. The technical difficulties of synchronizing parallel texts can be overcome with available software by the use of canonic text divisions (i.e. passus divisions and line numbers).[19]

Producing a Critical Text

Though not entirely convinced by the claim that the B archetype is "markedly corrupt" (Kane-Donaldson 70), we agree that it is still not Langland's original text and does not reflect his final intentions for his text. The B archetype represents only an early scribal copy of the poem, not the poet's holograph. The critical editing remains to be done. From this point, the lections of all the manuscripts and the Crowley early printed editions, the readings (and variant readings) in the traditions of A and C, and of course the collective editorial and historical scholarship of the last two centuries must all be considered.

Insofar as schools of textual criticism are concerned, our approach, like George Kane's, would be labelled "eclectic." This does not mean that we are without principles or that our principles are borrowed in ragtag fashion from systems with more internal coherence. Instead, it signifies our commitment to inductivism, to allowing the specific facts of each textual situation to reveal the principle that explains its particular pattern of evidence. Likewise, it signifies our aversion to the rigid deductivism of traditional recension (allowing a family tree of manuscripts based on a highly selective body of textual data to determine original readings, regardless of the internal logic of any particular array of textual variants).

To call oneself "eclectic" in this sense also implies a belief in the possibility, as well as the value, of retrieving authorial readings from complex traditions of scribal transmission. Such a commitment ought to be taken for granted in an editor. But the erosion of epistemological and cultural confidence has reached such a stage in literary studies that fashionable theorists (and some practicing editors) are retreating to "best-text" (or "diplomatic") editing, the commonest of pre-modern approaches. The best-text editor merely chooses a "good" early copy to reproduce, occasionally emending obvious mistakes but avoiding subtle decisions. With some works, of course, surviving evidence makes any other approach unfeasible. The recent popularity of the method, however, is categorical rather than pragmatic. Chaucer, for example, is assumed to be unknowably remote. Moreover, any attempts to recover his social and political perceptions are taken to constitute an elitist disdain for those of his scribes. We reject both the skepticism and the naive populism of such opinions.

Real methodological issues divide eclecticists. The difference between George Kane's approach to the editing of Piers Plowman and ours reflects two of the most basic of those issues. The first is whether the evidence used to determine textual originality should be almost exclusively micro-linguistic (lexical and morphological issues affecting words and short phrases, with little regard to their origin in manuscripts of varying parentage and reliability) or whether editors instead should combine such considerations with evidence based on larger linguistic units (the sentence and paragraph) as well as that derived from what we know about the psychology of the composition process itself, the sociology of authorial dictional adjustment, and the transmission history of texts. In the particular case of Piers Plowman, it can often be shown that small, archetypal variations among the three versions of the poem, variations that Kane's narrowly philological eclecticism tends to screen out as scribal, correspond to subtle pressures emanating from macro-linguistic or paralinguistic sources and which are, therefore, quite likely to be authorial.

A second issue dividing eclecticists concerns the degree to which editors may reasonably hope to distinguish an original, authorial text from its scribal variants. Theoretically, all eclecticists acknowledge several paradigmatic textual situations where recovery of an original is impossible. In practice, however, radical eclecticists tend to overreach, magnifying the weight of tiny pieces of evidence so as to avoid having to fall back on an uncritical "best-text" at any point. Conservative eclecticists, by contrast, acknowledge the operational primacy of copy text or archetype in cases where evidence is negligible or ambiguous. What distinguishes their approach from that of "best-text" editors is that they will freely and systematically emend where measurable differences of likelihood among textual variants can be discerned, even if such differences seem merely stylistic or metrical. Such an approach governs our re-editing of Piers Plowman B.

A more detailed discussion of our eclectic editorial practice can be found in Turville-Petre's 1987 essay, "Editing The Wars of Alexander," Adams' 1992 article "Editing Piers Plowman B: The Imperative of an Intermittently Critical Edition," and my 1993 article "A New Critical-Diplomatic Edition of Piers Plowman B in Hypertext." Practical examples of the results of this method can be found in Hanna's edition of The Awntyrs off Arthure at the Terne Wathelyne and Turville-Petre's and my edition of The Wars of Alexander.

For the electronic critical edition, each member of the team will assume primary responsibility for about one-fifth of the poem, working in close collaboration with another member of the group. We will supply textual, historical, and interpretive notes. In the introduction we will describe the manuscripts, describe and defend our editorial methods, and provide instructions for using the various software packets for accessing and manipulating the texts and annotations.

I and my team of research assistants will continue at the same time to transcribe the remaining B text manuscripts and attach such transcriptions to the Piers Plowman Electronic Archive. Eventually, we and other editors will supply the manuscript transcriptions, archetypal and critical editions of both the A and C versions, and diplomatic editions of the conflated versions. These are early days in electronic text editing, but we anticipate that electronic publication will lend itself to collaborative efforts like ours and to the publication of separately produced portions in large archival projects like this one.

The Archive will be periodically updated, both with corrections and added texts as transcriptions of other manuscripts (of A, B, or C texts) become available. As we have noted, the Archive will eventually include full texts of every authoritative witness to all three versions, and archetypal and critical editions of versions A and C as well as the conflated versions. We do not imagine that our team will do all of this work. Rather, we expect to incorporate other scholarly work of a high standard into the Archive. Such work is already in progress.[20] For instance, two advanced doctoral students at the University of Virginia plan to edit manuscripts of Piers Plowman for their dissertations. One plans to edit the second Crowley printed text of 1550. The second will edit Huntington Library MS 114. Though rightly rejected by earlier editors as worthless for establishing the original text, this conflation of all three versions represents valuable evidence of the reception of Langland's text. Such manuscript evidence, once discarded, can now become part of the critical and historical discussion through the medium of electronic publication.

Dissemination of the Archive

We plan to present our final product in a variety of formats. The fullest of these will be a CD-ROM archival edition which will include the following features:

1. the 54 transcribed manuscript texts with annotations;
2. the text of the A, B, and C archetypes and annotations;
3. critical texts of A, B, and C versions and annotations;
4. software such as TACT, COLLATE, and DynaText for text retrieval and manipulation, for making concordances and indices;
5. color digital images of the 54 manuscripts;
6. software to manipulate the texts and annotations;
7. if it is feasible, a version of the critical text normalized to the spellings and forms of MS X or U of the C version of the poem as reflecting the forms closest to those of William Langland's dialect.

Rapidly changing computer technology makes it unwise to commit the project to any current proprietary software. If research libraries across the world follow the example of the University of Virginia's Alderman Library, or if the World Wide Web develops a more sophisticated DTD and the capacity for searches and text manipulation it presently lacks, we will not need to supply text manipulation software for the Archive. Nor is it at all certain that CD-ROM will be the most rational form of publication. These decisions must be made later. The important point is that no archive as expensive to produce as this one should be tied to any existing hardware platform or operating system. Relatively detailed SGML markup provides a basis for achieving platform independence. SGML, as interpreted by the Text Encoding Initiative, promises to become the international standard for publishing in both print and electronic forms. Even if it does not become the standard or if it is superseded, it nevertheless provides a highly structured text which can be adapted readily to any new international standard.

We intend also to distribute individually the machine-readable transcriptions of the manuscripts, as well as our reconstruction of the B archetype and the critical text on floppy disks in MAC, UNIX, and DOS formats. With Thorlac Turville-Petre, I have helped to form a new scholarly organization, the Society for Early English and Norse Electronic Texts (SEENET), designed to publish reliable and scholarly electronic texts of Old Norse as well as Old and Middle English manuscripts. SEENET and the University of Michigan Press plan to publish each of the eight diplomatic editions of manuscripts CFGHmLMRW as it is completed. The initial version of The Piers Plowman Electronic Archive will also be published by SEENET and the University of Michigan Press, with periodic updatings of the Archive as more manuscripts are added.

Printed books remain a very efficient means for conveying texts, especially for class use. Eventually, we will produce a traditional printed version of the critical text with full introduction and explanatory, glossarial, and historical notes. The printed edition will be designed to address a variety of potential users: graduate and undergraduate readers of the text primarily interested in the poem, their teachers, and specialist scholars with an interest in textual editing.

III. Activity at IATH from July 1993 - May 1994

My goals for the IATH fellowship year were as follows:

1. to create an efficient structure for a complex archive of transcriptions; edited texts of hypearchetypes, archetypes, critical texts of versions A, B, and C; facsimiles of all fifty-four witnesses; and an apparatus criticus for each text to include codicological, paleographic, linguistic, metrical, glossarial, and historical annotations, all hypertextually linked;
2. to develop transcription protocols for preparing documentary editions both the Archive and SEENET;
3. to develop a DTD for both the Archive and SEENET editions and to master the TEI markup guidelines appropriate to representing the structures of late medieval handwritten texts;
4. to experiment with various methods for making digitized facsimiles of manuscripts and early printed editions and for providing means of linking the facsimile images with their transcriptions.
5. to acquire both photographic images of the manuscripts and permission to publish this new kind of facsimile.
6. to make full and accurate transcriptions of five of the eight manuscripts needed to establish the B archetype.

Creating a Structure

The precise structure of each segment of the completed Archive cannot be predicted at this point without begging important questions about the relations among manuscripts and versions. At this point in our investigations, we are unwilling to declare that there are three, and only three, versions, each deriving from an authorial fair copy without further authorial revision, or, alternatively, that none of the manuscript variants in one or all three versions are to be attributed to authorial revision by some process of rolling revision. We think that Skeat was correct in identifying the three versions, but we intend to re-examine the evidence without bias. Powerful arguments by a variety of scholars have been advanced in support of both positions, and until we have completed our transcriptions and collations, we can only suggest a tentative structure for the final Archive. Hanna's most recent (as yet unpublished) work with manuscripts R and F promises to renew discussion of the theory, enunciated most powerfully by E. Talbot Donaldson, that at least some RF readings, not shared by beta family witnesses, represent authorial revision.[21] Therefore, whether the B portion of the Archive will consist of a single critical text with hypearchetypes of alpha and beta families or two critical editions of two states of B remains to be determined.

We have, nevertheless, established a nexus of structural relationships in which each text will ultimately find a place. At the lowest level of the Archive, each manuscript will be represented by linked digital copies, one a color facsimile, the other a transcription, each with appropriate introduction and annotation. At the next level(s), if the evidence requires and permits (re)construction of hypearchetypes or archetypal texts, the resultant texts will be linked hypertextually to both the facsimiles and the transcriptions of the primary documents, along with their appropriate apparatus criticus. If, as we anticipate, the view that there are just three authorial versions is supported by the evidence, we will establish a level for the critical editions of versions A, B, and C, each again supplied with the kind of annotations appropriate to its level of the Archive and again hypertextually linked to all the lower levels. Linkages between the extant witnesses and the highest level of the Archive can be multi-layered. With respect to the hybrid manuscripts in which different versions have been spliced together, the linkages will, of course, be more shallow.

No available hypertextual software is entirely adequate to our purposes, though DynaText is capable of doing most of what we require. We anticipate further development of SGML interpretive hypertextual engines well before the Archive is completed.

Developing Transcriptional Protocols

An archive consisting of 54 manuscripts and early printed books produced over a period of 150 years contains a great deal of linguistic and textual variety. It is, therefore, essential to establish a common policy for transcribing them, and I have drafted a protocol to guide transcriptions by the team of graduate research assistants and the senior editors. That protocol, which has evolved over the course of the year, will doubtless continue to change and expand as we encounter fresh problems in new manuscripts. Some part of the problem lies in the very flexibility and freedom permitted by the Text Encoding Initiative guidelines. In many case, there are several ways to skin the cat, and for a project of this size, it is well to agree to certain conventions of transcription early in the project. The current set of transcriptional protocols may be inspected by clicking here.

Developing a DTD for SEENET and the Archive

With the kind assistance of John Price-Wilkin and Thornton Staples, I have tweaked in minor ways the Oxford Text Archive DTD developed by Lou Burnard of the Oxford University Computing Service and the standard used by the Electronic Text Center at Alderman Library.

Experimenting with Digital Images

Initially we intended to make digital facsmiles from black and white microfilms of each Piers Plowman manuscript. There were a number of reasons to favor that course of action. Microfilm copies of most, if not all, manuscripts already exist, so the basic documents need not be subjected to further handling. Moreover, digitizing from microfilm is relatively inexpensive, a few cents for each opening. When, however, I began digitizing Corpus Christi College, Oxford, MS 201 (F), it became obvious that too much information would be lost by this method. In the worst cases, bleedthrough of the ink from the opposite side of a leaf made all or part of a page illegible. Even when a relatively clean page can be produced, the black and white image fails to convey a good bit of textual information encoded in color, as in Trinity College, Cambridge, MS. B.15.17, (W) fol. 13r, where minor staining also makes the microfilm difficult to read. One has only to compare color images of the same leaves to see how very much is lost with digitized microfilm. In the case of F, the results are astonishing. A leaf once impossible to read became merely difficult to read. Those of you with text manipulation software with the capacity to magnify the image or manipulate brightness or do gradience analysis may experiment with the image to see how easily it may be made yet more readable. In the case of the leaf from W, it not only becomes more readable but offers a reader access to a kind of textual information once available only to readers with access to major manuscript collections.

British Library, MS Additional 35287, is an early fifteenth-century copy with extensive alterations from several hands. Color digital images make it possible in most instances to recuperate erased letters and in some cases to determine what lies beneath passages of overwriting. The image presently available is of little intrinsic textual interest--a handful of erased final -e's can be recovered--but the ease with which they are recovered will suggest the advantages of digital color facsimiles over black and white microfilm.

A color image of the first folio of W shows how digital facsimiles can be used to enhance readings. As is fairly ordinary with vernacular manuscripts, the first leaf is marred both by dark stains and by rubbing. The color image made from a 35 mm slide is difficult to read as it first appears on the screen. The stained top may be cropped and magnified, and the text is immediately made more readable. However, when a color editor is evoked to change the relative brightness of the image, the portions of the text formerly obscured by the stain become very easy to read indeed. Rubbing has made the beginnings of lines toward the bottom of the leaf difficult to read on the digital image (e.g., the first few words at the beginnings of B.Prol.24-25), but cropped, doubled twice, and darkened slightly, the text becomes tolerably clear. It must be clearly stated that the manuscript itself is, at this point, not difficult to read with the naked eye. However, in some instances, digital enhancement will make it possible to read on the screen what cannot be seen at all on the manuscript without harsh light and magnification, but this is not such a case. That is, the image manipulation demonstrated here represents an improvement only for readers who lack immediate access to the manuscript itself.

We have thus far experimented with two different ways of making color digital images. Dr. Andrew Prescott of the British Library has provided for experimental purposes color images made on their newly acquired Roche/Kontron ProgRes 3012 digital camera. Full leaves or details can be scanned at 2000 x 3000 pixels in 24-bit color with truly astonishing fidelity to the originals. Readers familiar with Kevin Kiernan's Electronic Beowulf project will have seen the magnificent color images produced by this camera. Those who have not may view the results on the World Wide Web at the following address:

• http://www.uky.edu/~kiernan/eBeowulf/guide.htm.

No images are presently available while the British Library determines its policy on publication permissions for digital facsimiles.

In our experiments of 1993-94, we have also found that digitizing from 35mm color slides produces images of high quality. Though we might obtain a higher quality of image from 4" x 5" color transparencies, they are prohibitively expensive, and the results of digitizing the 35mm slide images suggest that in most cases the less expensive slides are more than adequate. A jpeg image of 750 kilobytes may be doubled four times without significant loss of quality or pixillation.[22] Using a Nikon LS-3510 Slide Scanner attached to a MacIntosh Quadra 800 running Adobe PhotoShop, Professor Christie D. Stephenson, Assistant Fine Arts Librarian at the University of Virginia, did an experimental scan of slides from two different manuscripts at the following settings:

Image Name   Scan Pitch     Scan Res  Output Res      File Size

vhigh.tif        1          	3175 dpi   1000 dpi         	30  MB

vhigh.jpg        1          	3175 dpi   1000 dpi          	2.4 MB



high.tif         2          	1587 dpi    500 dpi          	7.5 MB

high.jpg         2          	1587 dpi    500 dpi        	750 KB



med.tif          2          	1587 dpi    500 dpi          	2.7 MB

med.jpg          2          	1587 dpi    500 dpi        	328 KB



low.tif          5           	635 dpi     72 dpi          	1.2 MB

low.jpg          5           	635 dpi     72 dpi        	168 KB

The 30 MB tif files may be necessary for some manuscript pages, those with erasures or complicated stains. Such images may be doubled six times with a resulting image 32 times larger than the original. It should go without saying that 30-40 MB are expensive in terms of storage and slow to call to the screen and manipulate. The RS 6000 on which most of my work is done has 32MB of RAM, and it is noticeably slowed in dealing with such large images. The low and medium quality images are better than a digitized microfilm facsimile, but the set of high quality jpeg images are even more satisfactory; they have answered neatly to every demand I have made upon them in the course of the year. Such images tend to run between 700KB and 1.4MB per manuscript leaf.[23] They are clear and crisp, and they lend themselves to various forms of image manipulation. At around a megabyte in size they will run quickly and efficiently on most 486 DOS machines and on the RISC stations. I have not experimented with MACS.[24]

Acquiring Images and Permissions to Publish

Librarians are often excited at the prospects of the new computer technology, but they are also often justifiably cautious about its implications for their holdings. In most instances, permissions have been generously offered. The following librarians and their institutions have been extraordinarily generous in their assistance both in granting me access to the original documents and facilitating digital imaging:

Dr. David Cooper, Librarian, Corpus Christi College, Oxford

Mr. D. J. Hall, Senior Under-Librarian, Cambridge University Library, Cambridge

Mr. D. J. McKitterick, Librarian, Trinity College, Cambridge

Dr. Andrew Prescott, British Library, London

Dr. Patrick Zutshi, Keeper of Manuscripts and University Archives, Cambridge University Library

Endnotes

1. 
   
   
2. 1 Several hybrid versions--various combinations of A and C or B and C--exist. One such manuscript, Bodleian Library MS Bodley 851 (SC 3041), has been edited as a fourth version by A. G. Rigg and Charlotte Brewer, Piers Plowman: The Z Version, Pontifical Institute of Mediaeval Studies, Studies and Texts 59 (Toronto: Pontifical Institute of Mediaeval Studies, 1983). See also George Kane, "The 'Z' Version of Piers Plowman," Speculum 60 (1985): 910-930. Two manuscripts are rejected by George Kane and E. Talbot Donaldson as too corrupt to be of use in establishing the B text: Huntington Library MS 114 (Ht) and former Sion College MS Arc.L.40 2/E (S), now Tokyo, Takamiya 23. Kane and Donaldson characterize HM 114 as "a text produced by conflation of the three versions of Piers Plowman, sophisticated and with added spurious matter" [Piers Plowman: The B Version (London: Athlone Press, 1975): 14 n. 95]. The Sion College manuscript is a sixteenth-century modernization of the poem. Though without value for establishing the text, both are interesting in their own right.Back
   
   
3. W. W. Skeat, ed. The Vision of William concerning Piers the Plowman in Three Parallel Texts ...., 2 vols. (Oxford: Oxford University Press, 1886).< For recent surveys of the modern history of the text, see George Kane, "The Text," in A Companion to Piers Plowman, ed. John A. Alford (Berkeley and Los Angeles: U California P, 1988): 175-200; Eric Dahl, "Diverse Copies Have It Diverslye: An Unorthodox Survery of Piers Plowman Textual Scholarship from Crowley to Skeat," in Suche Werkis to Werche: Essays on Piers Plowman In Honor of David C. Fowler, ed. Miceal F. Vaughan (East Lansing: Colleagues Press, 1993): 53-80. David C. Greetham has generously let me see his forthcoming article on the same subject, and Charlotte Brewer has written a book-length study devoted to the modern history of the textual traditions of Piers Plowman.Back
   
   
4. B. F. Westcott and F. J. A. Hort, The New Testament in the original Greek, 2 vols. (Cambridge: Macmillan, 1881).Back
   
   
5. R. W. Chambers and J. H. G. Grattan, "The Text of 'Piers Plowman'," MLR 4 (1909): 359-89; "The Text of 'Piers Plowman': Critical Methods," MLR 11 (1916): 257-75; MLR 26 (1931): 1-51.Back
   
   
6. George Kane, ed. Piers Plowman: the A Version (London: Athlone, 1960).Back
   
   
7. George Kane and E. T. Donaldson, eds. Piers Plowman: The B Version (London: Athlone, 1975).Back
   
   
8. The Kane-Donaldson B text was met initially by a number of hostile reviews in the scholarly journals as well as a somewhat smaller number of admiring ones. For the most favorable discussion of the Athlone B text, see Lee Patterson, "The Logic of Textual Criticism and the Way of Genius," in Textual Criticism and Literary Interpretation, ed. Jerome J. McGann (Chicago: University of Chicago Press, 1985): 55-91. Nevertheless, serious voices have been raised in criticism of the Athlone B text, and recent discussions of editorial theory and practice have been almost unanimous in their dispraise of the methodology and textual results. See Charlotte Brewer, "The Textual Principles of Kane's A-Text," Yearbook of Langland Studies 3 (1989): 67-90; "Scribal vs. Authorial Writing in Piers Plowman," in Medieval Literature, Texts and Interpretation, ed. T. W. Machan (Binghamton, NY: MRTS, 1991): 59-89; and "George Kane's Processes of Revision," in Crux and Controversy in Middle English Textual Criticism, ed. A. J. Minnis and Charlotte Brewer (Cambridge: D. S. Brewer, 1992): 71- 96. The essays in Crux and Controversy, with the exceptions of those of Nicholas Jacobs and Ralph Hanna, reject the empirical methodology that inspired the Athlone editors. As we have noted, much postmodernist criticism of their text seems to us confused and illogical.Back
   
   
9. Derek Pearsall, ed. Piers Plowman, by William Langland: An Edition of the C-text (London: Edward Arnold, 1978); Back
   
   
10. A. V. C. Schmidt, ed. The Vision of Piers Plowman: A Critical Edition of the B Text (London: Dutton, 1978).Back
    
    
11. Students of Chaucer, Gower, and the Middle English romance have begun to use such evidence with critically useful results. See, for instance, Paul Strohm, "Jean of Angoulême: A Fifteenth-Century Reader of Chaucer," Neuphilologische Mitteilungen 72 (1971): 69-76; Barry Windeatt, "The Scribes as Chaucer's Early Critics," Studies in the Age of Chaucer 1 (1979): 119-142; Kate Harris, "John Gower's Confessio Amantis: the Virtues of Bad Texts," in Manuscripts and Readers in Fifteenth-Century England, ed. Derek Pearsall (Cambridge: D. S. Brewer, 1983): 27-40; and Nicholas Jacobs, "The Processes of Scribal Substitution and Redaction: A Study of the Cambridge Fragment of Sir Degarré," Medium AEvum 53 (1984): 26- 48; "Some Creative Misreadings in Le Bone Florence of Rome: An Experiment in Textual Criticism," in Medieval English Studies Presented to George Kane, ed. Edward Donald Kennedy, Ronald Waldron, and Joseph S. Wittig (Woodbridge, Suffolk: D. S. Brewer, 1988): 279-284.Back
    
    
12. Benson's unpublished paper "A New Lemmatized Chaucer Concordance," was delivered at the 6-7 November 1992 Toronto conference "Of Remembrance the Keye: Computer-Based Chaucer Studies." I am grateful to Professor Benson for sending me an electronic copy.Back
    
    
13. Duggan, "Notes Toward A Theory of Langland's Meter," The Yearbook of Langland Studies 1 (1987): 41-70; "Langland's Dialect and Final - e," Studies in the Age of Chaucer 12 (1990): 157-192; Robert Adams, "Editing Piers Plowman: The Imperative of an Intermittently Critical Edition," Studies in Bibliography 45 (1992): 31-68.Back
    
    
14. This manuscript was copied by an early amateur interventionist editor who changed the text with considerable freedom. It is nevertheless necessary to include it because in its agreements with R it testifies to a form of the B tradition not otherwise extant.Back
    
    
15. We have conferred frequently with Peter Robinson of the Oxford University Computing Services about markup for the manuscripts. Robinson is chairman of the TEI committee in charge of preparing the standards for manuscript markup. He is also editing an electronic edition of Chaucer's Canterbury Tales, and we have agreed to coordinate the practices of these editions of the two most important Middle English authors. We are, moreover, fortunate in having John Price-Wilkin as a steady consultant for this project.

    We agree with Charles Faulhaber that much remains to be done in connection with providing markup in electronic texts to avoid the manual creation of markup and hypertext links: "The requirement for such massive editorial encoding would mean that, as in the case of manually produced concordances, only the most important texts--as defined by the current critical climate-- would be edited, while, on the other hand, much of that encoding would be of little or no use to the vast majority of users. Thus, rather than limit users to what the editor has already done, or require the editors to do work whose [cost] is all out of proportion to its [usefulness], the proper strategy is to provide software tools which will enable users to modify the text for their own purposes. Only time and experience will establish the proper demarcation of editorial and user encoding, but on both sides many tasks must be carried out automatically in order to make hypereditions feasible" ["Textual Criticism in the 21st Century," Romance Philology 45 (1991): 140.]

    It is important to add at this point that we have been evolving markup criteria during the past couple of years, and some texts transcribed earlier in the process must be corrected in the light of present practice. Furthermore, the HTML DTD does not permit the full range of SGML markup, and the texts prepared for this report have in most instances had the more full TEI-conformant markup stripped out for display on Mosaic.Back
    
    

16. Actually, SGML markup permits editors both to record the manuscript reading and at the same time mark it with appropriate tags that will make it possible for users to search both for the manuscript form and the more conventional form. For instance, scribes will write the phrase "at ease" as "a tese." The transcription markup permits us both to record the literal manuscript form and the expected form for word searches. Such doubly recorded passages of text are not emendations, do not misrepresent the scribal form, and at the same time serve many of the functions of the traditional print edition's emendations.Back
    
    
17. We will describe only those linguistic features pertinent to determining the provenance of the scribal copy or to establishing the text, since the fullest possible phonemic and graphemic inventories will be enabled for users by the transcriptions and software. Our linguistic descriptions will be modelled on those of Duggan and Turville-Petre, eds., The Wars of Alexander, EETS SS 10 (Oxford: Oxford U P, 1989): xxv-xlii.Back
    
    
18. We have not found the DOS program UNITE flexible enough for our purposes. It nevertheless can serve in a gross fashion as a check against errors in the output of COLLATE.Back
    
    
19. It is, perhaps, obvious that the dual processes of establishing the archetypal and critical texts, though conceptually distinct, must in practice often be simultaneous activities. That is, the editorial note appropriate to establishing the archetype will often, when the two texts differ, prompt the note that will accompany the critical text.Back
    
    
20. For discussion of the problems involved, see C. Michael Sperberg-McQueen and Lou Burnard, eds. Guidelines for the Encoding and Interchange of Machine-Readable Texts, Document Number: TEI P1. Draft: Version 1.0 (The Association for Computers and the Humanities; the Association for Computational Linguistics; The Association for Literary and Linguistic Computing, 1990): 122-123; and Charles B. Faulhaber, "Textual Criticism in the 21st Century," Romance Philology 45 (1991): 141. My colleague Peter Baker delivered an informative paper on precisely the problem of representing manuscript variants in SGML markup at the 29th International Congress on Medieval Studies, May 5-8, 1994, at Kalamazoo, Michigan. A longer and more detailed version of the paper is to appear as "Computers and the Future of the Critical Edition" in a forthcoming collection of essays edited by Douglas Moffat and Vincent P. McCarren on editing Middle English texts.Back
    
    
21. Carter Hailey, an advanced graduate student at the University of Virginia, is writing a dissertation with Professor David Vander Meulen and me on a bibliographic survey of the sixteenth-century Crowley printed editions in preparation for an electronic edition. Professors Jennifer Miller (University of California, Berkeley), Samuel Overstreet (Maryville College), and D. Vance Smith (West Virginia University), have begun transcriptions of an AC splice and two manuscripts from C to be incorporated into the Archive, and other scholars have been or will be asked to contribute other important electronic documentary editions.Back
    
    
22. "MSS R and F in the B-tradition of Piers Plowman," Transactions of the Connecticut Academy of Arts and Sciences 39 (1955): 177-212. The theory was first enunciated by W. W. Skeat in relation to R alone in his 1869 edition of B (xii). Donaldson had changed his mind by 1975 and the publication of his and George Kane's text of B (63-69).Back
    
    
23. The tif images used in this research report were made from jpeg images and will not produce the same quality image as the original jpeg.Back
    
    
24. The JPEG images are done at the Photoshop setting one step above "good," yielding about 10:1 compression.Back