Studies in Bibliography 56 (2003-2004) 1-44

We cannot verify your location (Log In)
Tanselle's "Editing without a Copy-Text":
Genesis, Issues, Prospects
Richard Bucci

At the opening panel of the 2001 conference of the Society for Textual Scholarship, some interesting remarks about copy-text were delivered by John Unsworth, a member of the Modern Language Association's Committee on Scholarly Editions (CSE). Unsworth said that he had originally planned to tell his audience that "the Greg-Bowers theory of editing" or "copy-text theory" had once enjoyed "hegemony within the CSE," but no longer did, owing to challenges from outside the Greg-Bowers school, where the focus was on other "periods, languages, and editorial circumstances." Unsworth submitted this thesis to Robert H. Hirst, the chair of the CSE at the time, for his thoughts, and reported receiving the following reply:

You seem to imply that all this change is coming from outside the hunkered down group of copy-text editors! . . . it has been chiefly copy-text editors over the decades who have insisted on refining and changing the application of copy-text theory. After all, Tom Tanselle is the only editor I know who's actually published an essay advocating "Editing without a Copy-Text." And long before that, Bowers published his essay on "Radiating Texts," that is, texts for which the very idea of a copy-text was inapplicable. So from my point of view, the hegemony of copy-text theory (both inside and outside the CSE) is mainly in the eye of the beholder, as opposed to the everyday practitioner. Practitioners have always sought to broaden or change everything from the "final intention" goal to (in Tanselle's case) the very idea that any one text should be automatically preferred in cases of doubt. 1

Unsworth was kind to pass on this private communication, since it contains many points worthy of deeper consideration. Hirst's phrase "hunkered [End Page 1] down group of copy-text editors" appropriately summons the false image some critics have projected of adherents of the editorial approach inspired by W. W. Greg's 1949 essay, "The Rationale of Copy-Text," and developed by Fredson Bowers—the better to dismiss them as relics of a by-gone age. This criticism ignores the rich practical experience of "copy-text editing" and the satisfying—one could even say exciting—theoretical developments it has engendered over the years. Hirst also names two outstanding moments in this experience when he mentions Bowers's encounter with what he called "radiating texts," and, most importantly, G. Thomas Tanselle's essay "Editing without a Copy-Text," which appeared in the pages of this journal in 1994. 2 This essay, which is indeed one of the most important writings on editing to appear in recent times, is concise and not intended to be exhaustive of its subject in and of itself. It stands, however, upon a great body of knowledge, having arrived on the morrow of a long period during which many literary scholars were deeply engaged both practically and theoretically with the Greg-Bowers-inspired idea of copy-text. Tanselle has been the most insightful and far-seeing participant in this collective engagement, and so his recommendation to "move beyond" Greg's "often useful but nevertheless inherently restrictive concept" (p. 2), so that editorial problems may be understood more immediately and with less technical prejudice, should arouse intense interest.

I

Tanselle's essay focuses our attention on the point where Greg explicitly limits the role of editorial judgment, and then demonstrates that this seemingly modest restriction has had unexpected adverse consequences. We are reminded that Greg's "strong endorsement of editorial freedom" extends only to the text's substantives (Greg's term for the wording); the copy-text "accidentals" (his term for the spelling, punctuation, word division, and emphasis) are accepted almost automatically (p. 8). While Greg also insisted that the editor be free to emend either the substantives or the accidentals whenever there was cause to do so, his assumption that a copy-text was needed at all was, in Tanselle's words, "founded on a belief that there was usually insufficient evidence for reasoning about accidentals" (p. 9). The copy-text, according to Greg, is to supply the [End Page 2] accidentals when variant accidentals from other authoritative texts are not clearly superior—that is, obviously authorial or having more recent authority. Tanselle argues that if the copy-text is used as the "fall-back" text to decide among variant accidentals, and if copy-text accidentals and substantives are to be altered by the editor whenever there is cause to do so, then it stands to reason that the copy-text will tend to be treated as the fall-back text for the substantives as well. This amounts to the "tyranny of the copy-text" which Greg sought to avoid (p. 9)—that is, the copy-text as monolith, unyielding of any word or mark of punctuation that has not been decisively disestablished by the editor.

Greg's rationale presumes an ancestrally linear series of texts, from author's manuscript to printed editions. Other kinds of textual traditions exist, and in the late 1960s Fredson Bowers encountered the most common of these in some of the stories of Stephen Crane. The stories were printed more than once, but each time from different, now-lost documents of equal authority. Some appeared in one American and one British periodical, with one printing based on a ribbon and the other on a carbon copy of a typescript made from Crane's manuscript. Other stories were syndicated in American newspapers: the syndicate received a manuscript or a typescript of the MS from Crane, made a proofsheet of it, and sent copies to subscribing newspapers, which used them as printer's copy. The prepublication documents are all lost, so the extant tradition for each story consists of multiple newspaper or periodical printings. Each printing was independently derived of the author's manuscript and therefore all have equal authority. An interesting variation occurred when, surmised Bowers, second typescripts (also now lost) were made of the manuscripts of some of the stories that had appeared in periodicals, in order to furnish printer's copy for book collections. For these stories, all printings have equal authority, but the periodical printings descend from one typescript, and the book versions from the other. 3 [End Page 3]

Bowers gave the term "radiating texts" to the tradition he encountered because the multiple printings of each story "radiate" independently from their lost manuscript. Though Bowers recognized that each printing was therefore of equal authority, he still attempted, apparently, to base each critical text on a copy-text as defined by Greg—"apparently," because, as Tanselle pointed out, he chose his copy-texts "not for their authority but for the extent of their agreement with what he had already decided the text should contain." 4 That is, after comparing the texts of each printing, Bowers chose as copy-text the printing that departed least from what he believed were the readings of the lost source. He usually settled on the printing that was most often with the majority wherever there was a variant. Bowers's apparatus reported all substantive variants but only those accidental variants which had required him to emend his "copy-text." Generally, according to Greg's rationale, accidental variants in later editions in a linear series are assumed to be more corrupt than those of an early copy-text, so excluding them from an apparatus could possibly be justified. Radiating texts, however, are not ancestrally linear, and the excluded accidental variants came from documents of no less authority than those Bowers had chosen as copy-texts.

Bowers discussed radiating texts in a group of essays, the first of which, called "Multiple Authority: New Problems and Concepts of Copy-Text," was published in 1972. As the title indicates, Bowers maintained that editing a group of radiating texts involves choosing a copy-text, even if the choice is a "theoretically indifferent" one, made only for the sake of "convenience," after the editor "has reconstructed the lost, common printer's-copy. . . ." 5 Since Bowers chose his copy-text after he had established his critical text, the copy-text was completely outside his purposes, and imposed out of mere habit. His insights about radiating texts recalled the way editors of ancient and medieval works reconstruct a lost source when multiple manuscripts descending from it survive—as Tanselle suggested in "Classical, Biblical, and Medieval Textual Criticism and Modern Editing," an essay from 1983 containing many forward-looking discussions on the relationship between judgment [End Page 4] and method in editing in all literary periods. 6 Bowers, it may be added, even hit upon a simple guideline loosely applied in the editing of ancient texts, known in that field by its Latin name, "difficilior lectio potior" (the difficult reading is preferable). Bowers did not explicitly cite this guideline, but he caught its gist when he observed that a less common variant might be the authorial reading, since "a majority of compositors faced with an unconventional accidental may sometimes opt for normality, leaving the true authorial reading preserved only by the dogged or indifferent few." 7

Had Bowers pursued the relationship between modern radiating texts and situations faced by editors of older texts, instead of attempting to impose Greg's rationale on his problem, he might have felt comfortable enough to allow his practical insights to shape his theoretical overview. It was Tanselle who recognized the true implications of Bowers's insights, which he revealed in his 1974 article, "Editorial Apparatus for Radiating Texts." Here he recommended editing without a copy-text, explaining that the critical text might be constructed of all the independently derived printings, and supported by an apparatus recording all variants, substantive and accidental. 8 In 1979, Robert H. Hirst became the first editor to follow Tanselle's recommendations, in his treatment of radiating texts in Early Tales & Sketches, an edition of some of Mark Twain's early writings. Both volumes in this edition contain critical texts reconstructed from contemporary, independently derived reprints of passages that first appeared in letters in now-lost [End Page 5] issues of the Virginia City (Nevada) Territorial Enterprise. In his commentary in the first volume, Hirst cited Tanselle's 1974 essay, and stressed that for the radiating texts, "no copy-text is designated because none of the authoritative texts is genetically closer to the original than the other." 9 Following his own inclinations as much as Tanselle's recommendations, Hirst also reported in the editorial apparatuses of the radiating texts all substantive and accidental variants from all his sources.

When in 1990 Tanselle reprinted "Editorial Apparatus for Radiating Texts," he remarked provocatively that "the idea of editing without a copy-text, set forth briefly here in relation to one particular kind of situation, has further applications that ought to be explored." 10 That Tanselle himself undertook the exploration was to be expected. In "Editing without a Copy-Text" he reminds readers of Greg's warning concerning the "tyranny of the copy-text," in which Greg maintained that the failure to understand that accidentals are more often subject to casual alteration, and substantives to purposeful—and therefore, more potentially authorial—change

has naturally led to too close and too general a reliance upon the text chosen as basis for an edition, and there has arisen what may be called the tyranny of the copy-text, a tyranny that has, in my opinion, vitiated much of the best editorial work of the past generation. 11

Tanselle points out in his essay that like Bowers in his encounter with radiating texts, Greg too was "somewhat tyrannized by the idea of copy-text," since he also recommended choosing a copy-text in a situation where two or more texts are of equal authority—that is, when there would be no justification for presuming the accidentals in one document to be more authoritative than those in the other (p. 10). What clearly concerns Tanselle most, however, is that in situations where a copy-text is warranted according to Greg's rationale, it tends to interfere with good judgment, by extending its influence, despite Greg's wishes, over the wording of a text as well as the accidentals. The "role of the copy-text," [End Page 6] Tanselle remarked, "turns out to be that of supplying readings (of both substantives and accidentals) whenever there seems no other basis for deciding" (p. 9).

II

The history of editing has moved according to its own logic, marked, on the positive side, both with methodological advances and with compelling demonstrations of the importance of informed judgment. The two elements, method and judgment, tend to appear on the intellectual stage as opponents, and are sometimes identified, respectively, with the ideas that the past is best recovered by either objective or subjective means. The tension in this opposition has generated its share of pointless negativity, but like all dynamic relationships the struggle has its creative potential. In the cycle referred to by Tanselle on the first page of "Editing without a Copy-Text," in which editorial discussions are alternately dominated by objective or subjective orientations, come points where deeper, synthetic breakthroughs seem possible. Not that such breakthroughs necessarily emerge from the discussions to guide editorial activity, but that the potential for holistic understanding exists for editors to exploit, in the best work of those identified with both sides. Ultimately, an editor can draw the two sides into a unitary approach, as A. E. Housman did. Housman the scholar may be most widely remembered for the sharp arguments he advanced in the prefaces to his editions, and in his lectures, favoring informed judgment over the mechanical application of methods. He is also recognized for his great talent for conjectural emendation—the ability to detect and correct corrupt passages without the direct support of documentary witnesses. 12 Less known is that Housman did not edit by his emendatory power alone. He recognized the importance of all pertinent knowledge and understood [End Page 7] the significance of Lachmann's method of the genealogical classification of manuscripts better than those who believed in its oracular powers. On his many-sided considerations, including his careful recensions and careful handling of the recensions of others, did Housman's insights flourish.

So dazzling were some of these insights that even canny admirers failed to notice the full range of scholarship which supported them. In La genesi del metodo del Lachmann (The Origin of the Lachmann Method), Sebastiano Timpanaro cited the testimony of another that his own teacher, Giorgio Pasquali, once declared excitedly, "C'è uno solo che può far emendazioni, è il Housman" (There is only one who can make emendations: he is Housman). 13 Timpanaro adds that in a lecture he attended, Pasquali admired the genius of Housman's famous interpretation of Catullus 64. For all his admiration of Housman, however, Pasquali did not, Timpanaro notes, hold him in as high esteem as he might have. Pasquali appreciated Housman's great gift for emendation, his ear for language, and deep understanding of poetry, but believed these were humanistic talents, which came at the expense of a more developed scientific orientation to his subject matter. Pasquali's misconception was partly due to Housman's reputation as a critic of the routinized application of editorial methods, and to Pasquali's inability to obtain Housman's edition of Manilius, 14 "where," said Timpanaro, "the genealogy of the codices are delineated with a sure hand" (p. 103). To his teacher's view that Housman's genius was "unmethodical," and [End Page 8] even "antimethodical," Timpanaro countered by referring to his "profound `methodicity,' " according to which

rigorous methodological criteria always guide his emendatio: the material gathered by him, in the prefaces and in the notes to the editions, and in many articles, on the various types of corruptions and their origin, confirms what he always emphasized—that the `intuitive' element necessary to make conjectures must be confirmed by experience and reason; and in this same vein go the syntactical, stylistic, and prosodic-metrical observations that he always considered were necessary to support his conjectures (or his defense of variant traditions: these also exist, and they are, for the most part, excellent). (p. 104, 104 n. 42; translations original to this article)

What Timpanaro would have us understand is not so much that Housman struck some adequate balance between method and judgment, but more that Housman allowed his own intelligence to guide his approach to textual problems, selecting and applying the relevant criteria in accordance with the material and his purposes. The example of Housman illustrates the fairness of the view that the opposition of method against judgment is only apparent: method is rather a creation of judgment, a development of it, a particular judgment, concentrated and abstracted. A method, an analysis, or a rule is often independently developed numerous times, and may therefore have more than one author. La genesi del metodo del Lachmann contains the incontrovertible demonstration that the method to which Lachmann gave his name was not really his. Lachmann formally divided the process of editing a text with multiple surviving sources into sequential halves. In recensio (recension), the relationships between the sources are established through collation and the analysis of errors held in common, to determine, ultimately, whether one of the sources is the common ancestor of the others, or whether the common ancestor is lost and must be reconstructed of its independent surviving witnesses. In emendatio (emendation) the common ancestor that has emerged from recensio— whether a surviving exemplar or a conjecturally reconstructed one—is corrected. Emendatio, Timpanaro pointed out, was an art as old as late antiquity, revived by the humanists, and practiced with moments of brilliancy in subsequent centuries by figures such as Giuseppe Giusto Scaligero (1540-1609) 15 and Richard Bentley (1662-1742). Timpanaro then demonstrated that recensio was the outcome of the collective work [End Page 9] of several nineteenth-century predecessors and contemporaries of Lachmann, and its principles were already in place by the time Lachmann summoned them into service for his Greek New Testament (first published in 1831) and his famous edition of Lucretius (1850). Greg's rationale is more rightly named, since it is mainly his alone—though, like all good insights, it rests on the earlier, partial advances of others. Whether a method or a type of analysis was created by a single individual or was the outcome of an entire intellectual tradition, however, it can never be anything more than the distilled thoughts of human beings. Methodological and analytical approaches to editing emerge from the thinking of their creators possessed of some objectivity, for they likely were developed over time, as responses to a variety of experience. Yet they are not natural laws; they can only suggest ways in which judgment might be profitably focused in order to re-create the past. They may be continually tested against new evidence, and adjusted or abandoned, according to the limitations that are revealed by this. They are, in other words, within our control. To regard them as independent of human thought, as existing above and beyond judgment, is a conceptual error that is bound to impede the understanding and resolution of editorial problems.

"Editing without a Copy-Text" shifts the editorial point of view from Greg's methodological design to the immediate evaluation of the work being edited. This is as it should be. As every experienced editor knows, textual situations vary, so it may seem axiomatic to say that it is best to approach each new project without the preconceived intent to apply a particular method. Editorial history, however, mainly runs in the other direction, and the shift that Tanselle makes in "Editing without a Copy-Text" is also away from older and more restrictive (and more enduring) points of view than Greg's. That Tanselle carefully prepared the way for this shift is known to the readers of his many essays, in which, over the years, he has investigated a great many of the editorial and bibliographic questions that literary scholars face. The essays are rich in interesting, relevant details, and the discussions are clearly presented, if sometimes driven by an intense logic, which can seem inescapable when a particular point is being made. After all, however, most readers will agree that the true object of Tanselle's discussions has never been to make points but to stimulate serious thinking about texts, and countless passages are given over to fair assessments— and sometimes optimally judicious restatements—of alternative viewpoints.

The idea that editors might pursue different goals has been understood by Tanselle's readers as far back as 1976, when his essay "The [End Page 10] Editorial Problem of Final Authorial Intention" appeared in Studies in Bibliography. 16 Here it was suggested that critical texts may be prepared of earlier, unrevised versions of works, or of later, revised versions. The recognition that literary works are defined by a series of historical moments has emerged as an essential principle of critical editing, as the texts of works from more recent periods have come in for scholarly treatment. The chronology includes the moment of first publication, but also earlier and possibly later moments, whether or not these are also adequately represented in existing documents. The failure to grasp this principle has led both to editorial blunders and to the mistaken charge that critical editing is a platonic striving toward a single ideal text, by means of unprincipled or aesthetics-based eclecticism. Over the years, however, Tanselle has clarified and emphasized the centrally historical nature of literary works to such an extent that by now this may be overlooked only through prejudice.

Of course in 1976 it was assumed that readers of critical editions were most interested in the text the author wanted, no matter what point in the history of a work the editor was seeking to re-create. This Tanselle noted in "Textual Criticism and Literary Sociology" (Studies in Bibliography, 1991), at the beginning of a consideration of the ideas of Donald F. McKenzie, Jerome McGann, and others, which move editorial attention away from the author's text, toward the text as published and received. "Editing without a Copy-Text" appeared at a time when the ideas of McKenzie and McGann had already become popular, if not yet with many practicing editors, then with many who write about editing. In these writings, the emphasis on the social nature of texts is often accompanied by a rejection of the author-centered editorial past and Tanselle's role in shaping that past. The time at which "Editing without a Copy-Text" appeared, then, was not the most opportune for a widespread positive reception. Inevitable is the comparison with the essay it is meant to replace, "The Rationale of Copy-Text," which Bowers used to signal the dawn of a new age in American literary scholarship. Every age has its moods, and presently most editors in this field have the sense of a setting rather than a rising sun. But if an "age" of editions of American authors has entered into a decline, valuable knowledge and experience of the editorial problems of modern literature has accrued. Greg's essay presented the wisdom of a half-century of thinking about English Renaissance texts, which Bowers energetically applied to his astonishing array of editions, including works from each of five centuries. "Editing without a Copy-Text" comprehends both the wisdom [End Page 11] of Greg's considerations and another forty-five years of editorial practice and thought. While Greg was tentative about the reach of his recommendations, Tanselle—relying on further history and experience— can confidently advance an "overarching framework" (p. 21) for approaching all editorial goals, in all literary periods. He helpfully calls what takes place within this framework "constructive critical editing," emphasizing that editing is a form of "historical reconstruction," wherein each word, each mark of punctuation, is critically determined by the editor, according to his or her knowledge of the author and the author's associates, the physical evidence, and the purpose of the editorial project (p. 22). Constructive critical editing is therefore not an editorial method, but rather a highly informed state of mind, which, according to the design of each project, draws to its attention all the relevant evidence and applicable supporting methodologies. Greg's rationale, Tanselle points out, may be one of those methodologies, but used to its original, restricted purpose—as an aid, that is, to judge the authority of the accidentals, and not as a base text for relieving the editor of the responsibility for making editorial decisions.

III

Genuinely important discoveries of ways to evaluate physical evidence systematically generate enthusiasms which can sometimes temporarily obscure the abiding importance of judgment in editing. So it was with the discoveries of the "New Bibliography," and so it was with stemmatics. The discoveries of Lachmann and his predecessors certainly put an end to some bad editorial practices, such as the one which favored majority readings blindly, without considering whether the majority was constituted of derivative repetitions of the same error. So much basic confusion and so many worthless manuscripts did stemmatics clear away that some less restrained practitioners applied it in pseudo-scientific fashion, thus achieving insupportable results. On the one hand, it was claimed that the method produced correct readings whenever it did not produce impossible ones; on the other, eliminatio—that part of recensio in which codices wholly derivative of others surviving are eliminated from editorial consideration—was practiced falsely and with a vengeance, so that all that remained afterwards was a sole source. By these errors, described by E. J. Kenney as the "brutal simplification of the textual evidence," errant Lachmannians came close to anticipating by several decades that agnostic rejection of Lachmannism known as [End Page 12] the "best text" approach. 17 Housman, who ranked Lachmann as high as Scaligero and Bentley in his editorial pantheon, did not think so much of Lachmann's mistaken followers, whether they mindlessly believed that stemmatics could extract correct readings from any number of manuscripts automatically, or pretended that it was a just means of eliminating troublesome evidence. Of the motives of those who labored under either misconception, Housman reported, "They must have a rule, a machine to do their thinking for them. If the rule is true, so much the better; if false, that cannot be helped: but one thing is necessary, a rule." 18

The term "best text" is usually associated with the anti-Lachmannian approach to medieval literature introduced by Joseph Bédier some years after Housman made these remarks. It nonetheless accurately describes a commonly recurring approach to editing literature of any period. Housman was certainly familiar with earlier generations of it, for in the same preface quoted above he criticized the "precious precept of following one MS. wherever possible." 19 Housman's bold advocacy of critical judgment would have an important if not immediate effect on the editing of literature from the printed age. The founders of the New Bibliography were at first not averse to the best-text approach. Ronald B. McKerrow did not believe that the documentary evidence could sufficiently support much critical emendation of sixteenth- and seventeenth-century English literary texts. A response to the presence of unreasoned eclecticism in the Shakespearean editorial heritage, as well as to some contemporary scholarship which he regarded as overly speculative, McKerrow's skepticism, while understandable, led him generally to discount the role of judgment in editing. Hence, as Greg made known, McKerrow, in his edition of Thomas Nashe (1567-1601), held that an editor of a work existing in more than one edition, each deriving from the one preceding, had no choice but to base himself on the latest edition known to contain the author's modifications. This text McKerrow called the "copy-text," and he recommended retaining it more or less whole, even though he knew well that but for those late modifications, it was probably less reliable than the earlier text.

McKerrow's edition of Nashe appeared in the first years of the 1900s. [End Page 13] In later years McKerrow reversed the direction of his practical recommendations, though even then, when he seems to have turned his own theory of copy-text upside-down, he persisted in an agnostic outlook:

It might, indeed, be better if in the domain of literary research the words `proof' and `prove' were banished altogether from statements of results obtained, for they can seldom be appropriate. . . . Nothing can be gained, and much may be lost, by a pretence of deriving results of scientific accuracy from data which are admittedly uncertain and incomplete. 20

McKerrow may have held agnostic views generally, but here and elsewhere in his Prolegomena for the Oxford Shakespeare (1939) he seemed illogically to suggest that knowledge of the literary past is especially unobtainable. Paul Werstine lately argued that McKerrow's skepticism was provoked by the high-flown conjectures of J. Dover Wilson, and by Greg's (early and selective) endorsement of them. 21 Yet the Prolegomena also expressed a more positive outlook, in its formulation of a new view of copy-text:

Even if, however, we were to assure ourselves on what seemed quite satisfactory evidence that certain corrections found in a later edition of a play were of Shakespearian authority, it would not by any means follow that that edition should be used as the copy-text of a reprint. It would undoubtedly be necessary to incorporate these corrections in our text, but unless we could show that the edition in question (or the copy from which it had been printed) had been gone over and corrected throughout by Shakespeare, a thing in the highest degree unlikely, it seems evident that, allowing for the usual continuous degeneration customary in reprinted texts, this later edition will (except for the corrections) deviate more widely than the earliest print from the author's original manuscript. This deviation is likely to be mainly apparent in spelling and punctuation. . . . We may indeed, I think, take it as certain that in all ordinary circumstances the nearest approach to our ideal of an author's fair copy of his work in its final state will be produced by using the earliest `good' print as copy-text and inserting into it, from the first edition which contains them, such corrections as appear to us to be derived from the author. (pp. 17-18)[End Page 14]

By these remarks McKerrow seems to have opened the way to editorial judgment that he had previously barred. His death in the year following the appearance of the Prolegomena left his edition of Shakespeare unrealized, but here he seemingly signaled an intention to edit with an awareness of the problem Greg addressed more directly in "The Rationale of Copy-Text." Greg defined the dimensions of the problem with greater clarity and precision, especially by drawing the operative distinction between accidentals and substantives. By identifying and segregating these two categories of the problem of authority, Greg was able decisively to release editorial judgment from the constraints of McKerrow's early view. Among the examples Greg used to demonstrate what he was getting at was McKerrow's critical text of Nashe's The Unfortunate Traveller. 22 McKerrow based his text on the second edition, since evidence indicated that it had been revised by the author. But the editor also believed that Nashe was not responsible for all the changes, and that the accidentals of the second edition were less reliable. Unable to see his way clear to a rational eclecticism—which might have allowed his text to reflect what he knew about the author—he surrendered the better part of his judgment to the confines of what amounted to a "best text." 23

Worth nothing is that The Unfortunate Traveller is a romance in prose, and by using it as one of his central illustrations, Greg demonstrated that he was not focusing his analysis on a particular literary genre—that is, dramatic works—as is often assumed. Greg was also aware that the "underlying principles of textual criticism" were held in common across literary periods and languages. 24 He especially recognized the relation of the problems he was facing in the literature of the English Renaissance to those faced by editors of classical literature, and he introduced his discussion with an illuminating sketch of editorial trends [End Page 15] in the classics, concentrating on the tension between method and judgment. Of course Greg recognized the differences also, mainly nothing that editors of early modern literature concerned themselves with their authors' spelling, whereas editors of classical literature usually normalize spelling, since their source texts were at too great a remove from the original manuscripts to do anything else. But when Greg warned that "the classical theory of the `best' or `most authoritative' manuscript . . . has really nothing to do with the English theory of copy-text," 25 he did not mean to discourage readers from seeing connections between the two editorial fields. Rather, this warning had the special purpose of preparing scholars of early modern literature to accept what for some would be difficult propositions: that textual authority relevant to the reconstruction of a particular moment in the history of a literary work might be preserved in more than one document; that for the reconstruction to be credible, the editor must be free to draw upon all the authoritative documents, as well as upon his or her own thinking; and that governing power over the editorial process is the mind of the critic focused on this historical problem. Greg's copy-text is not a "best text" or a base text, since it is not meant to decide the wording.

It is fair also to say that Greg intended for his copy-text to decide even less than the wording, since he expected an editor to think about spelling and punctuation too, and alter copy-text forms whenever there was reason to do so. While Greg's rationale has a methodological appearance, it is reasonably conceived, since chances are that the earliest surviving document in a series is the one which will preserve the most authorial details. Greg did not believe, however, that the manuscript details of the Renaissance works he was concerned with could be generally restored by his rationale. Few manuscripts, of course, survive. Collateral holographic evidence is usually scarce, and so an author's customary spelling and punctuation patterns cannot often be identified. Overall norms for such details, furthermore, had not yet emerged, and one need not doubt the technical competence of Renaissance scribes, compositors, and proofreaders, nor their disposition to follow copy faithfully, to suspect that they would not hesitate to alter manuscript spelling and punctuation which they believed were deficient or erratic. In view of these factors Greg noted:

Since the adoption of a copy-text is a matter of convenience rather than of principle—being imposed on us either by linguistic circumstances or our own philological ignorance—it follows that there is no reason for treating it as sacrosanct, even apart from the question of substantive variation. . . . I see [End Page 16] no reason why [an editor] should not alter misleading or eccentric spellings which he is satisfied emanate from the scribe or compositor and not from the author. If the punctuation is persistently erroneous or defective an editor may prefer to discard it altogether to make way for one of his own. He is, I think, at liberty to do so, provided that he gives due weight to the original in deciding on his own. . . . 26

By his eloquent restatement in the Prolegomena of the idea of copy-text, McKerrow revealed that he had been uncomfortable with his earlier conception of it. The new conception that Greg took up, with its reasonable eclecticism and emphasis on informed judgment, was bound to make another kind of editor uncomfortable, one who, unlike McKerrow, was at ease only when hunting in a single text for obvious errors. Greg's rationale challenged editors to face the difficult editorial choices. Aware of the extremes to which classical stemmatics had been taken, the author hedged his recommendations against misuse, by emphasizing that his intention was to clear the way for the intellectual resolution of textual problems, and by his warning about the "tyranny of the copy-text." When the subtlety of Greg's thinking is taken into account, especially his pronounced distrust of even the copy-text accidentals, then the warning seems to have more to do with the original conception of copy-text than with his redefinition of it. The experience of Lachmannism, however, showed that methodological approaches can lose some of their theoretical subtleties in the course of a widely-based practical application. The very presence of a method can entice some editors to focus their energies on questions of its application (in the case of Greg's theory, the choice of copy-text, whether to emend it, etc.), rather than directly on the work being edited.

Owing to Fredson Bowers's strong personality and his unique interdisciplinary expertise, Greg's rationale was widely applied, as everyone knows, to the editing of nineteenth- and twentieth-century American literature. In this field, published works had not received much critical editorial attention, and so it was inevitable that the literary scholars who assembled to prepare the texts of the many editions coordinated by the CEAA beginning in the early 1960s would have had little or no experience with textual problems. Some early efforts were not surprisingly marked by a conservatism characteristic of inexperience, and provide ample evidence of the tyrannizing influence of a designated copy-text. Such a beginning could have been predicted, and over time—just as predictably—better results were achieved more often, as editors gained experience. It must be said, however, that the granting of undue influence [End Page 17] to the copy-text continued and continues, in editorial endeavors that stick to copy-text readings rigidly, or select a copy-text with the disguised or half-disguised purpose of excluding other evidence from editorial consideration altogether. Of the two transgressions, the former is the less troubling, provided that the rejected variants are recorded in the apparatus; the second is the more harmful, since the copy-text is chosen in order to withhold evidence from the reader. Either way, however, adherence to Greg's rationale is often proclaimed in the textual essay, while the ostensibly critical text more truly reflects a best-text approach. The failure to recognize radiating multiple authority remains a persistent problem. More than once in the field of American literature, for example, critical editions have adopted as copy-text a first book edition of a work that was also printed serially in a magazine, with both printings deriving from the same typescript copy of the author's manuscript. While the first book edition's accidentals may be followed carefully, with as small a detail as a broken comma reported in the apparatus, the accidentals of the magazine printing are simply ignored, owing to the mistaken notion that scholarly editing means never having to report accidental variants. Bowers, as we have seen, was partly responsible for fostering this notion, which, like the kind of copy-text choice described above, compounds the ill effects of misconceived editorial choices with a deceptively spare apparatus. The main thrust of Bowers's recommendations on apparatus, however, went in the other direction. Here he followed McKerrow's good example, laid out in the Prolegomena, and improved upon it over the years. In 1962 he revealed his plan for the informative apparatus criticus now familiar to scholars of American literature. The plan advised editors to report many textual details, including what came to be called "pre-copy-text variants"—varior evidence of revision found in documents preceding the selected copy-text. 27 It should further be remembered that Bowers was also responsible for those deeply penetrating writings on radiating multiple authority, which should convince modern editors of what is more commonly understood by editors of earlier texts—that the existence of multiple independent witnesses of a lost original is an editorial blessing and not a curse. From multiple independent authorities the substantives may be established more securely than from a single line of descent, while at the same time informed choice about the accidentals becomes possible.

With his bibliographical experience and familiarity with English [End Page 18] Renaissance studies, Bowers was well prepared to direct the attentions of American literary scholars to textual matters, and introduce critical editing into their field. Taking stock of the better preserved historical record of modern literature, Bowers did not simply transfer the lessons of editing English Renaissance texts to the new period. He rather extended the logic of Greg's and McKerrow's recommendations, guided quite naturally by the same overriding interest in what the author wrote. This interest, Bowers understood, could be pursued further in modern works than in works from more remote periods. He also developed the concept of the "author's final intentions," first named by McKerrow, 28 in recognition of the ample documentary record of many modern works, which often preserves more than one moment in the development of an author's intentions. The "ancestral series" on which Greg based his rationale was a series of printed texts containing at least two "substantive" texts—that is, texts carrying authority, such as the earliest, or a later one bearing an author's revisions. 29 For editors of modern literature, the series may include early draft manuscripts, a fair copy manuscript, typed or handwritten amanuensis copies, galley and page proof-sheets, prospectuses, periodical printings, and first and subsequent book editions (and possibly separate series of these in different countries). Any of these documents might contain the handwritten revisions of the author or an assoociate; its genetic development may be linear to a point and then radiate from there. Several extant "substantive" texts may predate the first book edition of the work. Bowers created a scholarly edition that made use of this evidence, in a critical text ordinarily (not always) reflecting the author's final intentions for the work at the time of his or her last revision of it, and an apparatus recording much of the history of the text to that point. The author's earlier intentions— whether expressed in a draft manuscript, or a first edition (when there was a later revised edition also)—would therefore be recoverable in the apparatus.

Like any good historian—for scholarly editors are historians of the written word—Bowers suited his approach to the evidence. The history of editing is defined not only by advances made within a particular subject [End Page 19] period, but also by the belated movement of scholarship through literary time. With each succeeding chronological period coming in for editorial attention, the level of documentary evidence rises, improving in both quantity and quality. Speaking very generally, the evidence available to editors of ancient texts is the most compromised; the situation improves, slowly at first, and then more dramatically, for editors of Medieval, Renaissance and early modern, and modern texts. As Greg noted, editors of ancient texts tend to normalize spelling and punctuation, since the evidence rarely permits them to know anything at all about their authors' preferences. Some authorial details may survive in Renaissance books, and preserving these in a critical text is the purpose of Greg's rationale. Obviously editors of modern literature can recover much more textual history, since they may be able to call upon multiple surviving "substantive" documents, including authors' manuscripts, and possibly external evidence as well (such as letters and other collateral documentation). The intentions of the author might now be understood in their development—how they changed over time, on the author's own initiative, or through a collaborative interaction, with a reader whose opinion the author valued, for example. The degree to which these intentions were respected in the publication process might also be discoverable, and where they were not respected, a cause might be revealed, such as careless typesetting, the application of a publishing-house style, or factors more deeply related to the substance of the work in question. The author's text, for example, might have displeased the publisher, an agent, or even a government censor, for reasons ranging from the commercial to the political. Many editors who have focused on authorial intention are familiar with these relationships and factors, having recorded the evidence of them in their editions and analyzed their significance. Their interest in the author helps them understand that in such situations, authors have sometimes had no recourse but to alter their works, or allow them to be altered by others, according to the demands of those with the power.

The essence of editing is in the treatment of the historical evidence. Unfortunately, the vitally important historical aspect of author-centered editing has not always been recognized by its contemporary critics. Those lately emphasizing the social nature of texts have thus negatively evaluated eclectic texts as things unto themselves, without referring to the scholarship—which ought to be published in the apparatus—on which they depend. Of course some editions are more competently edited and therefore more informative than others, but the uneveness in quality has not been an essential target of these criticisms. A well-edited [End Page 20] work, however, should contain the evidence of how the author's text was "socialized."

IV

An unintended consequence of the recent disputes over editorial aims and methods has been the uncovering of a great deal of confusion about the meaning of copy-text and what constitutes the "Greg-Bowers approach." This development is potentially promising, since latent misunderstandings are the more damaging. The shift in editorial focus, from preconceived method to immediate thought, suggested by Tanselle in "Editing without a Copy-Text," provides the logical grounds for clearing the confusion away. Thoughtful Greg-Bowers practitioners, social textual theorists, or adherents of any other editorial approach ought to find in this essay the intellectual inspiration to achieve their particular aims to a high standard of scholarship.

While "Editing without a Copy-Text" attempts to move editing beyond the limits of the particular methodological preconception of the Greg-Bowers approach, it also preserves and proceeds from the approach's universal advantages. Among these are, first of all, a great body of practical editorial experience, from which emerge the other advantages—the intense focus on textual history, the high scholarly and technical standards, and the tradition among its best practitioners of the free exercise of informed judgment. Sustaining this common editorial heritage, it may even be said, is the better part of Tanselle's aim. Today's editorial climate, however, is clouded in places with barely qualified rejections of the entire Greg-Bowers experience. While the critics have made many valid points, they have also fallen too often into fundamental misunderstandings. If these are allowed to stand, then little advantage will be taken of Tanselle's essay, and so any assessment of the essay's prospects must engage the criticisms. The most convenient way to do this involves considering the social text as advanced in Jerome McGann's writings, since here the criticisms are reasonably conceived.

Jerome McGann has been the most conscientious in attempting to give the sociological approach to editing a theoretical foundation, and in his writings one finds many fair assessments and criticisms of the Greg-Bowers editorial approach. These are marred, however, by some less informed remarks, indicating that McGann has absorbed a few tired misconceptions about author-centered editing. Despite Tanselle's many painstaking demonstrations of their illogic, these misconceptions have appeared year after year as straw-men, set up and knocked down by the [End Page 21] opponents of author-centered editing, and sometimes mistakenly defended by those in favor of it. From these misconceptions a reductive shadow of the Greg-Bowers approach is cast, devoid of the approach's defining nuances and flexibility of application. The emphasis on the exercise of sound editorial judgment is especially absent, as is an awareness of the theoretical and practical growth the approach has gone through in recent decades. In place of variegated thought and rich editorial experience come the impoverished notions of the copy-text as a "best text," and of the critical text created according to the Greg-Bowers approach as a timeless, ahistorical, and therefore ideal representation of the author's intentions for his or her work—or, in McGann's words, "a pure abstraction." 30 This second notion is openly proclaimed; the first tends to steal into discussions or take hold of editions semi-surreptitiously.

Indicative of the problem is a passage in McGann's 1991 collection The Textual Condition in which two editorial outcomes are falsely set in opposition to each other—"the production of an eclectic text" and "the production of an edition which displays and analyzes the historical descent of the work." 31 The former idea is identified with the Greg-Bowers approach, and the latter is represented as being foreign to it. Yet the two outcomes are not logically opposed, and a scholarly edition of a work taking the Greg-Bowers approach should contain both an eclectic critical text reflecting a particular moment in that work's history, and a critical apparatus with the evidence necessary to reconstruct other historical moments. Bowers has been much criticized by McGann and other editors on subjects ranging from his overall interest in reconstructing authorially intended texts, to the particular choices he made in the works he edited. What is interesting about a great deal of this criticism is that it is based, at least in good part, on the evidence that Bowers published in his editions. One may go as far as to say that the second of McGann's two editorial outcomes—"the production of an edition which displays and analyzes the historical descent of the work"—owes more to Bowers than to any other editor of anglophonic literature. Early on in his project of transferring to the field of American literature the more exacting standards of English Renaissance bibliography and editing, Bowers declared it the duty of editors to place all their "textual cards on the table—face up." 32 He demonstrated what he [End Page 22] meant in his series of editions, beginning with the Centenary Hawthorne, which presented literary works as they had rarely been presented before. A critical reconstruction of an author-focused text which aimed at what Bowers called an "inferential authorial fair copy" 33 was accompanied—indeed, inextricably linked with—an unusually full apparatus criticus, containing not only editorial reasoning (textual notes), but the history of the work in its variants from document to document. Bowers even reported authorial alterations in the manuscripts he preferred as his copy-texts. Along with these illuminating innovations, however, Bowers's editions, as was subsequently shown, also contained not a few errors (great and small), needless inconsistencies, and even some apparent sloppiness of form. That Bowers did not always practically fulfill the promise of his pioneering editorial outlook, however, should not be allowed to obscure his achievements. His editions left a good deal of room for improvement, but much of this could be (and was) made on Bowers's own terms—with more careful historical collations, clearer presentations of the relevant textual evidence, a better understanding of the relationship between the critical text and the apparatus, and a greater appreciation of pertinent biographical and historical evidence relating to the author and the work being edited.

Since Bowers presented his critical text within the larger context of a scholarly edition, his editorial choices were by and large open to general scrutiny—that is, the alternatives which might be argued over were published in the apparatus. The regular exception, of course, was the accidental variants, which Bowers for the most part did not report. The guidelines of the Center for Editions of American Authors (CEAA), which attempted to codify Bowers's approach to editing, did not, as a practical matter, require scholarly editions to report them. In his own editions, Bowers cast this practical compromise in more theoretical terms—citing in the Centenary Hawthorne, for example, not only the "copiousness" of the accidental variants as cause for omitting them, but "their basic lack of significance" as well. 34 The unfortunate appearance [End Page 23] of this claim—that accidentals are basically insignificant—in the edition to which many apprentice editors looked for guidance had the predictable effect of routinizing the omission from scholarly editions of a class of evidence which at least sometimes was important. Bowers preferred to take an author's manuscript as his copy-text when this was possible, and so generally it was the accidental variants of the first edition that went unreported. The many critics of Bowers's preference for manuscript copy-texts—most prominently, James Thorpe, Donald Pizer, Philip Gaskell, and Donald McKenzie—would not, however, have been satisfied with mere lists of the missing variants. While they disagreed among themselves on a variety of issues, they all believed strongly that the critical text itself ought to reflect the process by which an author's manuscript accidentals were subject to printing-house modifications. Yet given that the alternative accidentals would not be reported in the apparatus, Bowers's choice was, from a scholarly standpoint, the less inconvenient one. First-edition accidentals are, after all, in first editions, copies of which are usually easier to consult than authors' manuscripts. Harder to investigate would be variant manuscript accidentals left unreported in an edition using a first-edition copy-text. McGann endorsed the argument in favor of first-edition copy-texts in A Critique of Modern Textual Criticism, holding that it is "clearly more sound than Tanselle's and Bowers's, for it takes better account of the social dimension which surrounds the process of literary production." 35 McGann's statement shifts the definition of copy-text, if Greg's rationale is at all being taken into consideration, since that was designed not to emphasize the social dimension of texts, but to keep as close as possible to the author's manuscript accidentals. We hardly need wonder what Greg's recommendations would have been if authors' manuscripts of the works that he studied had survived.

Despite his overall emphasis on the socialized text, McGann does favor fairly heavy editorial intervention in certain circumstances— though his reasons for doing so are not obvious. He defends, for example, a regularized and modernized "reading text" for early modern works, "to preserve the continuity of a . . . cultural resource." 36 While arguments for modernization have long been made, McGann's is peculiar in its aggressively negative assertions against those who eschew modernization. Tanselle, as McGann notes, has made the obvious point that casting a work of literature in the wording and spelling of a different time period is an "ahistorical" practice; for this McGann accuses him [End Page 24] of failing to "recognize the historical dimension of all literary productions," and failing to understand that "[e]very literary production is `ahistorical.' " As if it were not hard enough to understand how the same person could be guilty of both these transgressions at once, McGann goes on to chastise scholarly editors who seek to conserve Shakespearean spelling and vocabulary, for burying "the factor of the intended audience" under their "social and institutional ideology." 37 Whatever may be McGann's precise meaning here, what comes through generally is, on the one hand, the belief that an editor's prejudice against the competence of modern readers to understand old texts is a sufficient basis for radically altering them, and, on the other, the denial of scholarly legitimacy to attempts to move closer, by following the evidence, to what might have been the text of the author's final manuscript. According to McGann such efforts are deceptive and reveal a "hypnotic fascination with the isolated author." 38

Or so it would seem. In "Textual Criticism and Literary Sociology," Tanselle appraised McGann's initially keen appreciation of Hans Walter Gabler's edition of Ulysses. 39 Tanselle pointed out that despite its use of some counterproductive novelties—including a confused redefinition of the term "copytext" (without the hyphen), to mean both the constructed critical text and a stage in its construction—Gabler's method of editing was essentially author-focused. Tanselle further showed that McGann's endorsement of Gabler's edition as textually innovative was misplaced. While the mode Gabler chose for presenting the textual history of Joyce's work may have been unusual, the same kind of history is also present in any good scholarly edition with a substantial apparatus. McGann once believed that Gabler's arrangement of this history was a fundamental improvement on Bowers's, while Tanselle explained that such arrangements may reasonably differ as long as the history of the text is expressed adequately. McGann eventually underscored Tanselle's point, by backing away from his earlier appraisal. In his essay on Gabler's Ulysses, McGann praised the editor's "synoptic continuous manuscript text" as a "brilliant editorial reconstruction" that allowed "seriatim reading" of Joyce's work as it had developed. 40 A few years later, however, he questioned whether Gabler's edition could "illuminate" issues of [End Page 25] textual variation, and ironically asked, "Would anyone think that Hans Gabler's edition of Ulysses is a work to be read?" 41

Tanselle noted that Gabler's understanding of "copytext," the term he attempted to redefine, was problematic to begin with, reporting Gabler's belief that "By common consent, an editor chooses as the copy-text for a critical edition a document text of highest overall authority."42 42 McGann expressed a similar understanding of the concept of copy-text in his essay:

In the post-Greg context, the term signifies what an editor chooses to take as the text of the highest presumptive authority in the preparation of an eclectic, or critical, edition. . . . The copytext serves as the basis of the critical edition that is to be produced. The theory is that the readings of the copytext will be taken over in the critical edition unless other readings . . . are positively shown to carry a higher authority. In this theory, copytext is practically equivalent to some document or set of documents. 43

This definition of copy-text is not McGann's own, of course, but it illustrates well enough the difficulty that even the most astute scholars have had with Greg's conception of divided authority. McGann has also written subsequently of "establishing" a copy-text, and of the copy-text becoming "eclectic." 44 This difficulty caused mischief in the interesting appendix of A Critique of Modern Textual Criticism. Here, under the title "A Possible Objection" (pp. 125-128), McGann discusses exceptions to his overall recommendation of an editorial focus on the socialized text. He concludes that in situations of textual "expurgation, suppression, and mutilation," editors had better focus on recovering and presenting the text of the author. The inconsistency of this position remains quite striking. One could expect an interest in texts as social products to sharpen when dramatic differences separate authors' texts from the versions permitted by publishers or governments or some other controlling social factor. The examples McGann used to illustrate the need for author-centered rather than social editing include the published works of John Cowper Powys, which, according to McGann, were ill-affected by the author's fear of lawsuits and a post-war paper shortage. These are social factors, of course, as surely as others whose effects McGann wants editors to respect, such as the imposition of printing-house styles or publishers' wording changes on authors' texts. In his 1986 study "Historicism and Critical Editing," Tanselle well expressed the dilemma [End Page 26] McGann faced by selectively advocating a socialized text: either, he argued, McGann is recommending a socialized text only when the editor finds it preferable—which amounts to a call for aesthetics-based editing— or he is recommending it only when the author preferred it—which shifts the overall editorial focus back to the author, where McGann supposedly does not want it. 45

The dilemma is complicated by a tangled discussion involving copy-text, which turns on the apparently needless question of whether the expurgated, published version of Powys's novel Porius—the version that, we are told, was "drastically cut back" because of the paper shortage—or the unpublished, unexpurgated typescript text, should be used as the copy-text of a critical edition of the unpublished text. In 1983, when McGann's discussion appeared and most scholarly editors were concerned with representing an author's final intentions, it was generally thought that some works exist in versions too different to be adequately presented in an edition containing a critical text of but one of the versions.46 46 Editions had already appeared, however, that ran counter to this supposition, or at least demonstrated that scholarly editions could present a great deal of textual history, including the details necessary to reconstruct more than one version of a work. One of these editions, Early Tales & Sketches, the previously mentioned collection of some of Mark Twain's early writings edited by Robert Hirst, presented critical texts of the early versions of magazine and newspaper stories that Mark Twain later revised—sometimes extensively and sometimes more than once—for a series of book collections. The variants from the later versions were recorded in the apparatus. An editorial precedent for treating a situation even more closely related to the one McGann describes, in which an author's work was grossly altered for clearly practical reasons, had also been set by that time. The Northwestern-Newberry Typee, edited by Tanselle, Harrison Hayford, and Hershel Parker, appeared in 1968, and presented a critical text reflecting Melville's original intentions for his work, as expressed in the first British edition, before he acceded to the demands of his American publisher to soften his criticisms of missionaries. In line with these examples of author-centered editing (though without reference to them), McGann favors presenting readers with an unexpurgated Porius; once this view is taken, it would [End Page 27] seem no choice remains concerning a copy-text, since we have been told that there are but two significant documents—the author's original typescript and the expurgated book version. By "copy-text," however, McGann has McKerrow's meaning in mind, since Greg's rationale is only a tool for preserving in a later, revised text, an earlier and presumably less corrupt level of accidentals. McGann does mention accidentals, but the meaning of his discussion about copy-text obviously turns on the widely differing wording of the two versions of the novel.

Mix-ups like these have plagued Greg's conception of copy-text for most of its history, partly because its adherents have employed the term as loosely as its critics. Greg recognized two meanings for the term, corresponding respectively to textual situations defined by a single authoritative document, and those involving multiple authorities whose texts relate to each other in linear fashion. His unique contribution—his "rationale"—addressed only the second type of situation. Bowers, with Tanselle's help, came to understand that it could not address situations defined by multiple authorities which descend independently from a common ancestor—that is, a nonlinear tradition. For clarity, he might have added that neither can Greg's rationale be applied to works surviving in a single authoritative source—a situation which, logically speaking, is also nonlinear. Yet when creating critical texts of such works (as Hawthorne's Fanshawe, for instance), Bowers and most CEAA-CSE editors have called the single authorities their copy-texts. Greg also endorsed this wholly different use of the term copy-text, meaning the document whose text the editor "copies" out and then corrects:

If the several extant texts of a work form an ancestral series, the earliest will naturally be selected, and since this will not only come nearest to the author's original in accidentals, but also (revision apart) most faithfully preserve the correct readings where substantive variants are in question, everything is straight-forward, and the conservative treatment of the copy-text is justified.

Here the "copy-text" is the only "substantive" text. Greg then reemphasized his special rationale for situations defined by "more than one substantive text"—where the copy-text is followed "in accidentals," but "allowed no over-riding or even preponderant authority so far as substantive readings are concerned." 47

Terminology is sometimes regarded as a secondary question by innovative thinkers such as Greg, but one need not subscribe to the views of Derrida to see that the labels by which ideas are known can sometimes decide whether they are understood rightly or not. As Greg noted, McKerrow [End Page 28] "invented the term `copy-text' " in his edition of Nashe, "giving a name to a conception already familiar," being "that early text of a work which an editor selected as the basis of his own" (p. 19). The precedence of this definition of copy-text, its simplicity, and its compatibility with the sound of the term itself probably explains why it continues to be used in the way McKerrow defined it. Had Greg labeled his innovation "the rationale of multiple authority editing" or even "the rationale for an accidentals text," some confusion might have been avoided. As it was, Greg's modified definition of copy-text never took full hold in the minds of many professed adherents, and so it is not really surprising that its critics also should have found the concept difficult to grasp. McGann, for example, cites "An Introductory Statement," issued in 1977 by the CSE (the successor of the CEAA), in which it is asserted that a "primary requirement for any responsible edition is that it include a statement identifying the document which supplies the copy-text—that is, the text which the editor is following as the basic text." 48 The ancestor of this document, the CEAA's 1972 "Statement of Editorial Principles and Procedures," similarly defined copy-text as "that individual manuscript or proof or state of an impression which forms the basis for the edited text; in other words, it is the text which the editor follows at all points except those where he believes emendation to be justified." 49 Both definitions were augmented with extended discussions of how the copy-text should be chosen and treated, especially in the 1977 document, and generally presented thoughtful and clear guidelines for constructing accurate, informative editions.

As official statements, both documents were aimed at imparting vital bibliographic and editorial wisdom to scholars engaged in editorial projects who may not have had much experience in scholarly editing. In line with their educative purposes, the statements may also have deliberately emphasized that aspect of "The Rationale of Copy-Text" in which Greg seems to reduce the importance of the distinction he himself drew between accidentals and substantives, in order to encourage editors to alter either whenever there was reason to do so. Tanselle, a creator of the CEAA and CSE statements, would likely have been responsible [End Page 29] for this emphasis. In his influential essay of 1975, "Greg's Theory of Copy-Text and the Editing of American Literature," he stressed the "pragmatic"—and therefore provisional—nature of Greg's distinction, reminding editors to follow the accidentals of the copy-text only when they had no reason to do otherwise. He noted that resorting to the procedural part of Greg's rationale in such circumstances "is more satisfying than tossing a coin"—an endorsement whose obvious meagerness was meant to illustrate the relative positions of procedure and judgment in Greg's essay. Later in the 1975 essay, Tanselle reasserted that "nothing in Greg's theory . . . prohibits the emendation of accidentals in the copy-text when one has grounds for doing so." 50 This remark was part of a brief response to Paul Baender, who, in his article "The Meaning of Copy-Text," had suggested that the retention of the concept of a copy-text was out of keeping with Greg's main observation, that textual authority might reside in more than one document. Baender was a CEAA editor and inspector, and in this published form of a paper he first read in 1967, he endorsed the use of Greg's rationale for certain situations, while describing other situations for which he believed it was not suited. One of these involved the presence of multiple independently descended witnesses of a lost original—exactly the problem that Bowers would encounter a few years later and acknowledge as insusceptible to Greg's rationale. In hindsight, Baender's early identification of what Bowers later called "radiating texts" is eye-catching, as is his further questioning of whether the concept of a copy-text was ever appropriate, now that editorial "principles have become eclectic." Baender suggested, for example, that McKerrow in his Prolegomena, Greg, and Bowers "may not have realized the full implication of their eclecticism, which in the long run rules out the designation of a single text, basic text, or copy-text when there is more than one text of substantive authority." In an attempt to understand why the concept was retained even though McKerrow's early single-text rationale for which the term was invented had been discarded, Baender guessed that perhaps it was because "with respect to accidentals there still remained a single-text criterion." 51 This was the remark that elicited Tanselle's objection, quoted above; in saying this, however, Baender was not quite expressing his own belief, but rather giving his estimate of the belief of others, and the remark by itself does not indicate an aversion to the reasonable alteration of copy-text accidentals. In "The Meaning of Copy-Text: A Further Note," an earlier and more expansive answer to Baender, Tanselle [End Page 30] upheld "Greg's theory of copy-text" as applicable "to all situations," while attempting to dispossess Baender of his eccentric insistence on the interchangeability of the terms "copy-text" and "printer's copy." 52 This obvious misconception certainly harmed the reception of Baender's other observations, which were also expressed too briefly, perhaps, to encourage exploration of their potential implications. Baender, furthermore, seems never to have pursued these matters, even in his own editorial practices. Instead, in those early days before much editorial experience had accrued in the field of American literature, Bowers and Tanselle sought to foster a position according to which the copy-text of Greg supplied "fall-back" authority for substantives as well as accidentals, while stressing the importance of subjecting the whole copy-text to a thorough critical examination. The plain intention of taking this approach was to maximize editorial judgment, but the approach also risked obscuring Greg's special contribution to copy-text theory; potentially elusive, in other words, especially for newer editors, was Greg's notion that for certain situations, the copy-text is only and at most an accidentals text: "The true theory is, I contend, that the copy-text should govern (generally) in the matter of accidentals, but that the choice between substantive readings belongs to the general theory of textual criticism and lies altogether beyond the narrow principle of the copy-text." 53

In the CEAA and CSE statements quoted earlier, the copy-text is represented first as the overall authority for the substantives and accidentals (unless emendation is warranted), whereas in the situations Greg wanted to emphasize, the copy-text would be at most ("generally") the authority for the accidentals. Logically, according to Greg's rationale, no direct causal relationship should exist between the wording of the copy-text and the wording of the critical text. If the two texts agree, it should not be because the wording of the copy-text has simply been followed in the critical edition; rather, the wording of all "substantive" (authoritative) texts in a series collectively establish the wording of the critical edition—where they agree and there is no other contradictory authoritative evidence. Assuming, for example, the goal of a critical text reflecting an author's later final intentions, it might even be said that the wording of the later, revised text is more relevant, even in those places where it agrees with the earlier copy-text, since a revising author who allows some wording to stand might be conferring upon the unchanged passages the fresh authority of his or her new intentions. Where the earlier and later authoritative texts do not agree, the editor chooses [End Page 31] from among the variants, according to the chronologically limited set of authorial purposes from which the critical text is being derived. In this example, an editor would favor the later variants (minus errors and changes not ascribed to the author) for the critical text, and report the earlier ones in the apparatus.

Those editors of American literature who have understood the meaning of Greg's rationale may not realize that the term "copy-text," in its wider contemporary use—in fields such as Chaucer and Shakespeare criticism, for example—usually carries its original meaning. Editing with a copy-text, furthermore, ordinarily means following the text of a particular document wherever possible, as this typical editorial statement, taken from a (modernized) edition of Shakespeare intended for the college classroom, indicates:

Every effort consistent with critical sense has been made to adhere to the declared copy-text . . . , and unnecessary emendation, that pricking devil, has been carefully eschewed. When the copy-text, however, resisted all reasonable attempts to make sense of it, readings from another early printed text or from other editions have, of course, been admitted, but in all such cases the emendation has been placed in square brackets to warn the reader that the text at this point is open to question. 54

The copy-text is followed conservatively in all matters, that is, for each and every play, whether it survives in a single substantive text or several. The explanation of the use of the brackets invites readers to trust only copy-text readings, and regard what has not come from there with suspicion. Similar statements can be found in other editions of Shakespeare, and also in modern editions of Chaucer's works, where the existence of multiple independently descended manuscript witnesses of all the tales would seem to discourage the assigning of preponderant authority to any one.

In "Editing without a Copy-Text" Tanselle acknowledges that the "basic meaning of the term `copy-text' has remained stable from McKerrow's time onward—that is, the documentary text used as the basis for a scholarly edition" (p. 11). On the other hand, he also notes that the term "copy-text editing" is often used by Greg's critics to refer to the editorial practices of his adherents—to the use, in other words, of Greg's rationale. The term therefore signals, after all these years, either of an old pair of opposites—a best-text approach, or an eclectic one based on historical textual analysis. Greg was originator of this duality, and was comfortable using the same term to describe the counterposed approaches, [End Page 32] since he felt each had its place. Many of those who took notice of Greg's altered meaning of copy-text, however—whether to adopt it or criticize it—had trouble keeping it apart from the earlier meaning, while editors who were unaware of Greg's considerations (or McKerrow's second thoughts) gave the term in its original sense a wide currency. Whether because the term was too well suited to the meaning McKerrow first gave it, or because the CEAA/CSE upheld an interpretation of Greg's rationale that allowed the copy-text to be used as the fall-back authority for the substantives as well as the accidentals, or simply because easy-to-use "best-text" approaches will tend to drive out more nuanced eclectic ones—whatever the cause or causes, that is—the further existence of Greg's special sense of the term copy-text, as an accidentals text, is in question. In "Editing without a Copy-Text" Tanselle recommends that in constructing critical texts scholarly editors abandon Greg's copy-text (while, of course, preserving his rationale concerning accidentals), so that it will not be held up falsely, as a truth-giving mechanism, nor used to disguise virtually noncritical reprints of previously published texts as eclectic, critical texts. Should Tanselle's recommendation gain wide acceptance, especially among Greg's followers, the meaning of the term copy-text would, paradoxically, cease its sixty- or fifty-year internal struggle, and resume its original and untroubled one-dimensional appearance.

V

If misconceptions about copy-text and Greg's rationale have "mesmerized" (to borrow a term of Greg's) author-focused and society-focused editors alike, it is the latter group alone that has, mistakenly, criticized the authorially intended text as ideal in the philosophical or aesthetic sense. Accordingly, eclectic texts are seen as manifestations of the idealistic philosophical views of those taking the Greg-Bowers approach, views which cause them to pursue a phantom, perfect form of the works they edit. Tanselle has shown that this criticism is not really new: in his entertaining biography of Fredson Bowers, he recorded what must be its earliest appearance, in remarks made by Leo Kirschbaum at an MLA conference in 1959. Kirschbaum believed that an editorial focus on the author's text was out of place in works designed for theatrical performance, and he saw a "Platonic bias" in Bowers's application of Greg-based eclecticism. 55 Kirschbaum was reacting to the very first practical demonstration [End Page 33] of Greg's copy-text recommendations, made by Bowers in the first volumes (1953, 1955, 1958) of his four-volume edition, The Dramatic Works of Thomas Dekker. A quarter-century later, McGann expressed concerns similar to Kirschbaum's:

The idea of a finally intended text corresponds to the "lost original" which the textual critics of classical works sought to reconstruct by recension. Both are "ideal texts"—that is to say, they do not exist in fact—but in each case the critics use this ideal text heuristically, as a focussing device for studying the extant documents. Both classical and modern editors work toward their ideal text by a process of recension that aims to approximate the Ideal as closely as possible. Both are termini ad quem which, though not strictly reachable, enable the critic to isolate and remove accumulated error.

For the critic of modern texts, the classical model upon which his own procedures are based frequently does not suit the materials he is studying, and has often served, in the end, to confuse his procedures. Because this textual critic actually possesses the "lost originals" which the classical critic is forced to hypothesize, his concept of an ideal text reveals itself to be— paradoxically—a pure abstraction, whereas the classical critic's ideal text remains, if "lost," historically actual. 56

Critical texts of course do exist. By saying that "they do not exist in fact," McGann presumably meant that they do not exist until scholars perform the necessary historical research and create them. In other words, critical texts do not present the text of any single preexisting document. When speaking of older works, however, it must be recognized that for the most part, the preexisting documents may themselves be described as critical editions. That is, the scribes who made them ordinarily attempted to correct errors that they perceived in their sources, and not infrequently ventured ambitious emendations, whether by conjecture or by consulting manuscripts other than the one they were immediately copying. Many manuscripts are, furthermore, the products of a more thoroughgoing conflation of multiple sources. Most editors, whether they are editing older or more recent works, rather than working "toward their ideal text," seek to reconstruct, according to the limits of the evidence, texts that can be said to have existed "in fact," as long as factual existence is granted to the intentions of the author, whether or not these were ever accurately recorded. In the course of reconstructing the last surviving common ancestor of the extant witnesses, for example, the editor of a classical work may discover that this text—the "archetype"—contained errors. If the editor corrects them, then he or [End Page 34] she is departing from the archetype, and moving toward what the author wanted—whether or not he or she ever got it. The same holds true for the editor of a modern work who finds that the author's wishes were contravened by the publisher, typesetter, or even his or her own hand, in the process of writing down the words formed in the mind. Factual existence, that is, can also be assumed of Mark Twain's intentions to write "Becky" and "Rebecca" where he referred, mistakenly and in different books, to Judge Thatcher's daughter as "Bessie" and to the heroine of Scott's Ivanhoe as "Rachel."

Tanselle raised the important conceptual points relevant to this issue in "Textual Instability and Editorial Idealism," the 1996 essay (see note 55) that ought permanently to have dispelled the mistaken notion that editors, acting as editors in exercising their judgment, are idealists. He noted that the preoccupation of editors concerned with authorial intention is hardly as McGann has it—with a "pure" artistic form, poorly reflected in the material world—but rather with the more pedestrian problem of "the difficulty of getting words transferred accurately to a physical surface" (p. 12). McGann's misreading partly stems from "his failure to consider intentions as historical events" (p. 13). Furthermore, Tanselle points out, editors create critical texts reflecting authors' intentions not to serve as perfect, stable, or definitive versions of their works, but more basically "because intended texts are not available in documents and therefore cannot be studied without the exercise of critical judgment, leading to attempted reconstructions" (p. 13).

For McGann, the idealism manifest in the eclecticism of the author-centered editor goes hand-in-hand with a quixotic desire to fix a "single `text' of a particular work . . . as the `correct' one." If this were true, then his purported counter-emphasis on the "indeterminacy of the textual situation" would be justified. 57 In his 1994 essay "The Rationale of Hyper-Text," in the context of an argument against printed forms of scholarship and in favor of electronic forms, McGann expressed his distaste for scholarly editions of the Bowers type as "infamously difficult to read and use." 58 [End Page 35] This remark may explain his willingness to believe that author-centered editors present the critical text as the only "correct" text. Yet what is "difficult" to understand is not the scholarly edition (assuming it is competently constructed), but the "indeterminacy of the textual situation" itself, at least for some works. A scholarly edition allows readers to view a work in its development, by presenting the essentials of its history in the apparatus. The presentation is made mainly by recording the variants, rather than printing multiple versions of the work, but this form is convenient for study, since it focuses attention efficiently, on the changes. If the edition is well-organized, the difficulties inherent in grasping a remote and complex history are minimized with no oversimplification of the evidence. 59 [End Page 36]

The introduction of philosophical language into any discussion ought to signal the deepening of understanding, but the casual laying on of the term "idealist" to author-centered texts and editing has not been helpful. Those who have used it have, most basically, failed to recognize that all scholarly editions present "ideal," or hypothetical, re-creations of past moments in the life of literary works. This is true whether the main purpose of the edition is to present the unaltered text of a document, or an authorially intended critical text. 60 Either way, a moment from the past is resurrected by a scholar or group of scholars, who collect and evaluate the surviving materials and decide how best to present them, in a process involving human judgment from beginning to end (even when the choices are not explicitly acknowledged as choices).

Editors preparing an edition of a work of modern literature must often first decide which moment or moments in the life of that work to present—since sufficient documentary evidence for re-creating several may have survived. As has been discussed, a Greg-Bowers-style scholarly edition usually offers a critical text reflecting one more or less discrete level of an author's intentions for his or her work, while supplying in [End Page 37] the apparatus what evidence exists of other levels, as well as a record of errors, and of alterations made by persons other than the author, such as those that yielded the first published text.

A social textual editor constructing a scholarly edition around a socialized text must also make a series of decisions. The first and most important of these involves choosing the fundamental conception of socialization that shall govern the text. Two distinct conceptions seem possible: the social text can be viewed either as the outcome of the collaborative intentions of all who took part in the process of publishing a work, or simply as the published text received by its first readers. These two different forms of socialization correspond to different moments in the history of a work, and though these moments may have followed closely on each other, it is unlikely that the same text, nor even the same kind of text, could accurately represent both moments. The earlier moment, the "collaborative" social text, probably cannot be adequately represented by reproducing the text of any one preexisting document, and so an eclectic critical text is here indicated. The later moment, the "received" social text, is by definition the text of a particular document, and to represent it, an essentially noncritical presentation is called for. 61 To re-create the collaborative social text, the editor must deal with the problem of errors, since, presumably, all involved in the publication process intended for the text to be without errors. Of course once social textual editors admit that human intentions are historical facts, then they are, despite their different goal, operating within the same concept of history as author-focused editors. The decision to correct errors, furthermore, means that social textual editors are sharing a concern with their author-centered counterparts. Here they must resist the temptation, apparently felt by some with social textual interests, to limit their corrections to obvious errors. The reader does not really need a scholarly editor to find obvious errors; however, obviousness is not an essential characteristic of errors. There are also errors that are not obvious, and to correct these, access to original documents and specialized knowledge is needed. For scholars to restrict their own editorial judgment artificially just at this point, where it might be of help to readers, is, it must be said, intellectually irresponsible. On the other hand, an editor presenting a "received" social text would not correct original errors, [End Page 38] since here the object would be to reproduce the text experienced by its early readers, and the errors were, of course, part of that experience.

A further issue concerning the nature of texts is worth remarking upon at this point, as it helps further reveal the inadequacy of the idea that author-centered critical texts are idealistic abstractions, while preexisting texts are faithful expressions of material reality. It is the observation, made briefly by Tanselle in "Editing without a Copy-Text" and discussed more expansively by him elsewhere, that the texts of documents are not literary works in themselves, but rather sets of instructions for the re-creation of these works. 62 Readers of A Rationale of Textual Criticism, the slightly revised published version of Tanselle's 1987 series of Rosenbach lectures, will be especially familiar with this insight, which memorably relates literary works to works of music or dance, while distinguishing them from works of painting or sculpture. In the former grouping are works conveyed in "intangible" media—language, sound, movement. Their "tangible" artistic compositions—documentary linguistic texts, musical scores, dance notations—are unlike the tangible creations of painters or sculptors in that they are not the works themselves, but rather the forms into which intangible works must be converted so that they may be reconstituted and experienced. These forms consist of complex series of marks—as many (for a novel, for example) as a million or more letterforms, symbols, and spaces. Being highly complex, texts of documents are therefore also potentially highly unstable, prone to heavy levels of accidental alteration every time they are copied. While scholars might agree on what constitutes the text of a particular document, they will likely come up with different answers to the question of how accurately it records the instructions of the author. By embracing the roles of those who worked with the author, social textual editors might expect the problem of the textual fidelity of documents to diminish. Where, for example, the author-centered editor would find infidelity, the social textual editor would presumably endorse a text in which a publisher's copy-editor has regularized an author's punctuation or toned down his or her colorful language. Yet for an editor interested in the collaborative social text, there remain the questions of how accurately the text as first published reflects the intentions of the copy-editor, where alterations were desired, and also of the author, where his or her text was allowed to stand. As for the editor desiring to present a received text to modern readers, he or she faces a number of questions about how [End Page 39] best to re-create that early reading experience. The editor may have to decide which of several early published forms of the work to re-create, and whether the chosen text—say it is the first book edition of a novel— should be reproduced in an exact transcription or in photographic facsimile. The text of the chosen edition will have to be investigated by comparing numerous copies. If press variants are discovered, these should, of course, be recorded in the scholarly edition, but their presence would require a basic textual decision, about whether the scholarly text should reflect a particular variant state or the edition as a whole. 63 Answering the questions inherent in a social textual undertaking might easily demand as much critical thinking as is expended in author-centered editing, and whether social textual editors seek to present critical or essentially noncritical texts, they should find the author-centered editorial past full of useful lessons (both positive and negative) on the construction of scholarly editions—especially in the matter of scholarly apparatus.

Opposition to the concept of authorial intention and the belief that the eclectic text is an ahistorical, idealistic abstraction turn out to be counterproductive of thoughtful editing, no matter what editorial goal is being pursued. By accepting instead of rejecting historically determined eclecticism, social textual editors not only would avoid falling into logical contradictions, they would find themselves able to correct unintended errors more efficiently, by making optimum use not only of their own powers of observation but also of the ascertainable knowledge of the habits and activities of those originally engaged in the publication process. In discarding logically untenable objections to the Greg-Bowers tradition, furthermore, social textual editors might better appreciate the revealing textual histories available in editions whose critical texts reflect [End Page 40] an authorial focus. They might discover that the word "final" in "final intention" has sometimes meant that the contributions of an author's welcome collaborators have been admitted to the critical text, since its aim may be to reflect the text the author wanted, whether or not these wants were influenced by others. Author-centered editing is, of course, also sensitive to what authors did not want. From this sensitivity too comes much that should be of absorbing interest to the social textual editor. Scholars interested in the social text would hardly want to limit their understanding of printing-house styles, for example, by remaining in the dark about the author's competing preferences, if those can be known; nor, when faced with expurgated texts, would they want to investigate the differences between authors' and publishers' texts (as well as the causes of the differences) any less thoroughly than do author-focused editors. The pressures brought by others upon the author to alter his or her text are evaluated with care by the author-centered editor, to determine whether the resulting changes were accepted in the spirit of willing collaboration, or were forced upon an author who had little choice in the matter. While social textual editors may not be interested in a critical text that remakes this result in the author's favor, they should nonetheless welcome the presentation in the apparatus of the evidence of how the author's text was forcibly socialized (if this was the case), and find ways to match or improve on such presentations in their own editions.

Such typical editorial problems help reveal that the social and authorial focuses in editing have many concerns in common. What they do not have in common, however, is a history. All of that—the experience, the insight, the technical skill—has been gained in pursuit of the author's text, which is why it is imprudent, to say the least, for those interested in the social text to continue denying the validity of the Greg-Bowers approach. Many benefits would consist in accepting its validity and absorbing the lessons of its historical experience, including, most obviously for the social textual editor, how best to identify and account for non-authorial contributions to the development and production of texts.

In the compelling final pages of "Editing without a Copy-Text," the author conveys the essence of his recommendations by calling the reader's attention to the proven editorial conception of radiating texts. The advantages to reading and scholarship of having more than one independent documentary witness of the same lost text are self-evident. That a critical reconstruction of the lost common ancestor, and its accompanying apparatus of rejected variants, do not practically require a base- or copy-text has been proven. The designation of a copy-text in such situations, [End Page 41] furthermore, needlessly obscures the textual genealogy, and potentially clouds the editor's judgment. Once it is established that a group of sources descend independently from the lost ancestor, then any text in the group may be right against the others at any point of variance. All editorial choices must therefore be active choices; none can be made by default, by resorting, thoughtlessly, to a copy-text. At points where variants "seem `indifferent,' " Tanselle writes,

an editor may of course choose a reading from the text that supplies the largest number of other reeadings; but the decision is still an active one, in which one of the factors taken into account is the apparent general reliability of a particular text. The process remains one of building up a new text rather than making changes in an old one. (p. 19)

This is the approach that Tanselle proposes be taken to all textual situations, not only those involving radiating multiple authorities. 64 If critical editing has permanent scholarly value—and the foregoing discussions have attempted to show that it has, no matter what editorial goal is being pursued—then editorial judgment must not be unreasonably restricted. As we have seen, even the most thoughtful editors have fallen victim to the "tyranny of the copy-text"—despite Greg's warning about it. Accordingly, Tanselle, rather than issuing a stronger warning, has proposed that the truest way to stay clear of unreasoned results is to remove the copy-text from the editing process, since it has proven after all to impede serious thinking about the wording of critical texts.

Tanselle's recommendation may be said to consist of two parts: 1) as explained in his essay, the concept of editing without a copy-text, and its practice, modeled on the experience with radiating texts; and 2) the many lessons of a long editorial past, on which his concept depends and which he assumes will have been understood by its practitioners. These lessons contain both negative messages (such as warnings about the potential deceptiveness of facsimiles, the tyranny of the copy-text, or the limitations of mathematical schemes for evaluating variants), as well as many positive ones. The lessons of the New Bibliography, the observations of Greg and others about textual genealogy, the insight about the potential superiority of difficult or unusual variants, paleographic analysis—all editorial and wider historical knowledge, that is—retain their value within Tanselle's framework. To edit without a copy-text, editors [End Page 42] focus on the editorial problem with the maximum amount of relevant knowledge obtainable. This knowledge aids, rather than substitutes for, editorial judgment. Judgment is "clearly in the dominant position" (p. 20). The distinctiveness of Tanselle's approach is emphasized in his assessment of the likelihood that an editor employing Greg's rationale but without recourse to a copy-text might "select" some of the same readings that would have been "retained" had a copy-text been designated. Here Tanselle rightly insists that the distinction between the two approaches is nonetheless essential:

the difference between these two justifications for the same decision is not superficial: it goes to the heart of what critical editing is. The key point is not whether an editor would make the same decision by following Greg's rationale or by designating no copy-text but still following Greg's argument for the presumptive authority of the text closest to an authorial manuscript. . . . The important point is that the former approach places a rule above reason (as any recourse to a fall-back position must do), whereas the latter restructures the problem so that the editor's decision (even if it is the same decision) results from the positive step of taking a reasoned action (p. 19)

Tanselle has taken care that his recommendation embraces all critical-historical editorial goals, including the author's "finally intended" text, earlier or later authorially intended texts, and socialized texts emphasizing the contributions of the publisher's staff. His plan for "constructive critical editing" (p. 22) is therefore "a framework that liberates editorial judgment from the concept of copy-text while being neutral in itself as to the goal toward which that judgment should be directed" (p. 21).

Perhaps, as it becomes widely adopted, every truly important editorial insight, such as Lachmann's or Greg's, is fated to go through a period of overwork. Stemmatics was worked very hard for about fifty years, before Housman and others pointed out the uselessness of much of its product and reminded those who were listening that no mechanism can rightly substitute for an editor's good judgment and expert knowledge of his or her subject. As the writings of Giorgio Pasquali, Sebastiano Timpanaro, L. D. Reynolds and N. G. Wilson, E. J. Kenney, Bruce M. Metzger, and many others reveal, the best scholars of literature from the remote past have long since recommended a thoughtful approach to editing, appreciating the limitations as well as the benefits of methodologies and formulaic analyses, drawing upon them as needed, to aid, rather than override, thought. Fraught as the editorial outlook for more recent literature now appears, here too it seems inevitable that the most thoughtful scholars will move in a similar direction, if they haven't already [End Page 43] begun the process. The shift in editorial orientation may take a while longer, since divisive, extraneous prejudices against author-centered editing, which have needlessly accompanied the appearance of social textual concerns, have yet to fade from the scene. But sooner or later, the characteristically generous and inviting spirit of G. Thomas Tanselle's profound essay will attract to it the best editorial minds of all orientations.

Appendix

Observations on the collaboration between Cooper and John Fagan, reprinted with the permission of Thomas and Marianne Philbrick from their Textual Commentary for their CE edition of Cooper's 1844 novel Afloat and Ashore (AMS Press, 2004).

The surviving proof sheets of Afloat and Ashore . . . provide a revealing glimpse into the relation of author and printer in the production of mid-nineteenth-century American texts in general and, more specifically, they supply an extensive demonstration of the process by which Cooper and Fagan collaborated to produce the novels of the writer's last twelve years. For the first time the entire sequence of steps that led from the author's manuscript to the printed pages of the first American edition can be followed in detail. It now becomes clear that Fagan's compositors not only corrected obvious errors in the manuscript and introduced the usual number of errors of their own but freely initiated stylistic "improvements," many of which, in the absence of the proof sheets, would be virtually impossible to distinguish from Cooper's revisions. Moreover, the proof sheets display Fagan's important role in the process, for it was he who initially proofread the typeset copy, comparing it to the manuscript and catching many of the compositors' errors and innovations (but missing many others), and it was he who acted as a copy-editor by querying phrasing that he regarded as infelicitous and proposing changes for the author to consider. Finally the proof sheets afford a full picture of Cooper's role, as he responded positively or negatively to Fagan's queries, corrected some of the errors that had eluded Fagan's proofing, and continued the revision of his text that he had begun with his manuscript alterations—all the time, it would seem, working without further reference to the manuscript.

The text that resulted from this process, the basis for the first American edition and for all subsequent printings of Afloat and Ashore, is a highly imperfect one. As the proof sheets reveal, in Part 1 alone more than 600 substantive departures from the author's manuscript, consisting of compositorial errors as well as innovations, escaped the scrutiny of Fagan and Cooper. Some of those printers' variants merely blunt the edges of Miles Wallingford's narration, as when (at 1.226.26-27) Emily Merton's "bright, blue, English eye" becomes her "light, blue English eye," or muffle the tones of a character's speech, as when Marble, asserting that he has enough needles and thread to "set up" a slop shop, is made to say that he has enough to "supply" one (at 1.435.17). But many other such variants do serious damage to Cooper's meanings, as when the compositor drops whole lines of the manuscript (at 1.459.14-17 and 1.471.79) or misreads "into the colony" as "in the valley" (at 1.16.22), "even more" as "comrade" (1.23.6), "gleaned" as "gained" (1.66.4), "sly" as "shy" (1.166.5), "hint" as "point" (1.181.16), "more" as "men" (1.278.25), "sheers" as "shores" (1.359.27), "Callao" as "called" (1.383.27), "minx" as "miss" (1.475.16), "artlessness" as "restlessness" (1.515.17), and even "south" as "north" (1.231.1).[End Page 335]

In setting the text of Afloat and Ashore from Cooper's manuscript, Fagan's compositors applied what would seem to be a house style in the treatment of capitalization, spelling, and punctuation. Thus at the same time that the compositors correct a good number of the mechanical errors in AMS, in many more instances, though with varying consistency, they obscure Cooper's acceptable, if at times unconventional, accidental forms. For example, the printers regularly capitalize his lower case "state" (in reference to the American political division) and use the lower case for his capitalized "Street" (as in "Wall Street"); they print his "favor" as "favour" and his "any thing" as "anything"; and they often normalize his characteristic treatment of the comma, by which it functions only sporadically in its familiar grammatical role of signaling such things as the distinction between restrictive and non-restrictive clauses but persistently as an indicator of the pauses that a speaker might make, almost in the manner of phrase-marks in a musical score.

The styling of Cooper's text by the compositors extends to his words and word-order. They frequently tinker with his prepositioned idioms, as when they change "edifice in stone" to "edifice of stone" at 1.16.26; they update his verb forms, as when they change "sprung" to "sprang" at 1.50.17; they fuss with his connectives, as when they change "Then" to "Besides" at 1.212.24 or "for" to "as" at 1.413.4; and they rearrange his adverbs, as when they change "got subsequently" to "subsequently got" at 1.52.16.

But beyond such piddling refinements and wholly apart from the hundreds of compositorial misreadings of AMS is a large class of substantive variants in the uncorrected proof (subsequently referred to as PR) that can only be seen as attempts by the printers to re-write Cooper. These variants have nothing to do with styling, nor are they forms that have any visual resemblance to those of AMS. Rather, they are free inventions, apparent efforts to improve upon the author. Thus AMS "named" becomes "called" in PR (1.50.26), "in that day" becomes "at that time" (1.55.12), "steeples" becomes "spires" (1.55.22), "senses" becomes "soul" (1.60.10), "square-built" becomes "square-rigged" (1.62.11), "direct" becomes "enclose" (1.68.17), "portion" becomes "part" (1.77.25), "threshing" becomes "good threshing" (1.78.18), "passage" becomes "voyage" (1.80.21), "oddities" becomes "peculiarities" (1.81.1), "the tale" becomes "it" (1.82.12), "in the spring" becomes "early in the spring" (1.82.17) and "as soon as" becomes "at the time" (1.92.4). These examples are drawn from just two of the thirty chapters of Part 1, but they are representative in kind of this whole class of variants.

All of the extant proof of Afloat and Ashore except that for the first seven paragraphs of the Preface to Part 1 is marked with notations—corrections, queries, and occasional comments—made distinctly in black ink and written in a hand that is clearly John Fagan's. Fagan's copy-editing (hereafter referred to as FPR) performs a number of different functions. For one thing, it corrects many printer's errors, most of them typographical, such as the use of the wrong font, the misalignment of type, or the setting of an inverted letter. Some of those corrections, as when FPR supplies AMS text that the compositor has dropped, could only have been made by reference to the manuscript. The many occasions on which Fagan overlooked compositorial departures from AMS, however, suggest that he did not read the proof against the manuscript but only consulted the manuscript when he encountered obvious anomalies in PR.

In addition to making corrections, Fagan continues in proof the process of styling that the compositors had initiated in PR, attending to spelling, capitalization, and punctuation and making a few minor substantive changes on his own authority, such as replacing PR "as much as" with "so much as" at 1.115.27-116.1, PR "quick" to "quickly" at 1.357.15, and PR "sung" to "sang" at 1.370.16. But nowhere does FPR embody the bold substantive changes to Cooper's language that PR so frequently (and silently) does. Rather Fagan almost always poses substantive changes as proposals or queries, at times only underlining the forms that he regards as questionable and at [End Page 336] others both underlining the form and suggesting an alternative in the margin along with the abbreviation "qr." The queries that he addresses to the author range from such small matters as changing PR "dove" to "dived" at 1.362.5 (a proposal that Cooper rejected) to issues of plot and character: at 1.453.14-16, for example, he marks Miles' joke about the width of Neb's grin during the battle with the Malay pirates and then comments marginally, "This remark seems scarcely to consist with the gravity of Wallingford" and at 1.455.13 he underlines Miles' mention of the Bay of Naples, querying, "Will Wallingford see Naples?" Although Cooper usually responds to Fagan's stylistic queries by revising the text, he never makes changes in response to Fagan's comments on his narrative, and some he strikes out with a thoroughness that suggests a degree of fury.

Footnotes

1. "Reconsidering and Revising the MLA Committee on Scholarly Editions' Guidelines for Scholarly Editions," address by John Unsworth from the panel on "New Directions for Digital Textuality," Eleventh International Interdisciplinary Conference of The Society for Textual Scholarship, 18-21 April, 2001, Graduate Center of the City University of New York; quotations taken from printed handout.

2. Studies in Bibliography 47 (1994): 1-22, repr. in G. Thomas Tanselle, Literature and Artifacts (Charlottesville: Bibliographical Society of the Univ. of Virginia, 1998), pp. 236257. Tanselle notes that he first presented this paper on 12 July 1993, "as a Book Arts Press lecture during Rare Book School at the University of Virginia" (p. 1); the citations in this paper refer to the SB printing.

3. A typical example is "An Indiana Campaign," a story that was syndicated by Bacheller, Johnson and Bacheller, and printed in the Kansas City Star, the Buffalo Commercial, the Nebraska State Journal, the Minneapolis Tribune, the San Francisco Chronicle, and the St. Louis Post-Dispatch. Bowers demonstrated that these six newspaper printings, as well as subsequent printings in the English Illustrated Magazine and in Crane's book, The Little Regiment and Other Episodes of the American Civil War (New York: D. Appleton, 1896), were all based on separate copies of a master proof, and therefore of equal authority. This proof was made, Bowers surmised, of a typescript of Crane's manuscript. The typescript (either the ribbon or carbon copy) was, furthermore, probably the source of another printing, in Bacheller's own Pocket Magazine. Thus, in all, nine printings radiated from the lost typescript, independently transmitting its authority. A variant example is "The Revenge of the Adolphus," which appeared in Collier's Weekly, Strand Magazine, and Crane's book, Wounds in the Rain: War Stories (New York: Frederick A. Stokes, 1900). Here, Bowers believed that the two periodical printings were based on a single typescript made from the manuscript, one on the ribbon and the other on the carbon copy, while the book printing was based on a second MS-based typescript; see Tales of War, The University of Virginia Edition of the Works of Stephen Crane, Volume VI, ed. Fredson Bowers, with an introduction by James B. Colvert (Charlottesville: Univ. Press of Virginia, 1970), pp. lxix-lxxv, cxxix-cxl.

4. "Editing without a Copy-Text," p. 15.

5. Library 5th ser., 27 (June 1972): 81-115, quotations from pp. 101-102. Bowers's later considerations of radiating texts appear in "Remarks on Eclectic Texts," Proof 4 (1975): 31-76, and "Mixed Texts and Multiple Authority," Text 3 (1987): 63-90.

6. Studies in Bibliography 36 (1983): 21-68, see pp. 65-66.

7. Bowers, "Multiple Authority: New Problems and Concepts of Copy-Text," p. 99; for the customary version of this insight, see L. D. Reynolds and N. G. Wilson, Scribes and Scholars: A Guide to the Transmission of Greek and Latin Literature, 3rd ed. (Oxford: Clarendon Press, 1991), pp. 221-222; and Paul Maas, Textual Criticism, translated from the German by Barbara Flower (Oxford: Clarendon Press, 1958), p. 13 (C § 16 (a)). On this subject, Sebastiano Timpanaro has explained (through the English translation of Kate Soper): "anyone who has anything to do with the written or oral transmission of texts (including quotations learnt by heart) knows that they are exposed to the constant danger of banalization. Forms which have a more archaic, more high-flown, more unusual stylistic expression, and which are therefore more removed from the cultural-linguistic heritage of the person who is transcribing or reciting, tend to be replaced by forms in more common use." (The Freudian Slip: Psychoanalysis and Textual Criticism, London: NLB; Atlantic Highlands, N.J.: Humanities Press, 1976, p. 30 [English translation of Il lapsus freudiano: psicanalisi e critica testuale, Firenze: La nuova Italia, 1974].)

8. Library 5th ser., 29 (September 1974): 330-337; Bowers took note of this article in both "Remarks on Eclectic Texts" and "Mixed Texts and Multiple Authority" (see note 5), but viewed Tanselle's discussion as an intriguing practical suggestion, without recognizing its theoretical significance. Some years earlier, on the other hand, Paul Baender had recognized that a copy-text would be out of place in textual situations defined by multiple independent witnesses of a lost original: see the discussion toward the end of section VI of this essay.

9. Early Tales & Sketches, Volume 1 (1851-1864), ed. Edgar Marquess Branch and Robert H. Hirst, with the assistance of Harriet Elinor Smith, The Works of Mark Twain (Berkeley, Los Angeles, London: Univ. of California Press), pp. 658, 659 n. 236.

10. G. Thomas Tanselle, Textual Criticism and Scholarly Editing (Charlottesville: Bibliographical Society of the Univ. of Virginia, 1990), pp. 167-176, quotation from p. xiii of the preface.

11. "The Rationale of Copy-Text," Studies in Bibliography 3 (1950-1951): 19-36, quotations from p. 26. Greg's now-famous essay was delivered for him in 1949 at that year's conference of the English Institute; it was also reprinted posthumously, with a few revisions and corrections he left in manuscript, in W. W. Greg, Collected Papers, ed. J. C. Maxwell (Oxford: Clarendon Press, 1966), pp. 374-391. References in this paper are to the first, SB printing.

12. The term is somewhat misleading, since it implies that a reading produced of an editor's thinking is generally more doubtful than a reading present in a documentary source. Good emendations, however—and many of Housman's were very good—can be self-evidently authorial, and may restore or repair a passage that is no longer truly represented in any surviving document. In recognition of one such emendation, G. P. Goold was moved to remark, "It verges on the miraculous that Housman, unaware of what actually happened, was able, by sheer intuition of the poet's words, to restore them" (Elegies of Sextus Propertius, ed. G. P. Goold [rev. ed., Loeb Classical Library, Cambridge: Harvard Univ. Press, 1999], p. 17). Goold's enthusiasm will be familiar to anyone who has encountered such an authentic restoration, though Housman himself might have contested the idea that his emendation (of Propertius 2.12.18) was based on "sheer intuition": see the following discussion. Tanselle discussed conjectural emendation in "Classical, Biblical, and Medieval Textual Criticism and Modern Editing," calling "delusory" the suggestion that readings based on documentary evidence are necessarily more objective than those based on an editor's thoughts; see Studies in Bibliography 36 (1983): 25-27.

13. Timpanaro, La genesi del metodo del Lachmann, [4th ed.], con una Presentazione e una Postilla di Elio Montanari (Torino: UTET Libreria, 2003), p. 103, 103 n. 39. The earliest version of this work appeared in Studi italiani di filologia classica, nuova serie 31 (1959) and 32 (1960). Timpanaro's source for Pasquali's remark on Housman is the published form of Otto Skutsch's centenary address, Alfred Edward Housman, 1859-1936 ([London]): Athlone Press, University of London, 1960), p. 7.

14. In his Storia della tradizione e critica del testo, Pasquali acknowledged that Housman's Manilius was "inaccessible" to him (see latest reprint of 2nd edition, Firenze: Le Lettere, 2003, p. 392 n. 3). By indicating the importance Pasquali attached to editorial method, the mistaken criticism of Housman paradoxically reveals just how close in outlook these two great editors were. Like Housman, Pasquali is mainly recognized as an uncompromising advocate of thoughtful editing. His famous book grew out of a long critical review (published in Gnomen 5 [1929]) of Textkritik (1927), Paul Maas's (intentionally) severe disquisition of stemmatics. Storia is a monumental demonstration of the unique and concrete character of each textual tradition, and of how the differences limit the usefulness of abstract editorial principles. For Pasquali (and it must be acknowledged that Maas did not disagree with him on this point), there was no acceptable substitute for erudition, careful investigation of each textual situation, and the free exercise of informed editorial judgment. Storia was first published in Florence by F. Le Monnier in 1934, and a second edition was brought out by the same publisher in 1952, with a new preface by the author and three appendixes (including one by Paul Maas). This edition was previously reprinted in 1974 by Mondadori (Milano), and by Le Lettere in 1988.

15. Scaligero, a Huguenot of Italian heritage but born in France, is known there as Joseph Justus de l'Escale, and to English speakers as Joseph Justice Scaliger, or the younger Scaliger, to distinguish him from his well known father, humanist physician and philologist Giulio Cesare Scaligero (1484-1558).

16. This essay has an even earlier history, since Tanselle prepared a version of it for another journal in 1968; see the asterisk on p. 166 of SB 29.

17. Kenney, The Classical Text: Aspects of Editing in the Age of the Printed Book (Berkeley: Univ. of California Press, 1974), p. 138.

18. Preface to D. Iunii Iuvenalis Saturae [Satires of Juvenal], ed. A. E. Housman (London: E. Grant Richards, 1905; repr. Cambridge: Cambridge Univ. Press, 1956); quotation taken from extract in A. E. Housman, Selected Prose, ed. John Carter (Cambridge: Cambridge Univ. Press, 1961), p. 58.

19. Ibid., p. 58.

20. McKerrow, Prolegomena for the Oxford Shakespeare: A Study in Editorial Method (Oxford: Clarendon Press, 1939), p. vii.

21. Werstine, "Editing Shakespeare and Editing Without Shakespeare: Wilson, McKerrow, Greg, Bowers, Tanselle, and Copy-Text Editing," Text 13 (2000): 27-53. In a 1930 appraisal, Greg drew attention to the allure and the dangers of Wilson's talents ("The Present Position of Bibliography," Bibliographical Society address printed in Library, 3rd ser., 11 [December 1930]: 241-262, repr. in Collected Papers, ed. J. C. Maxwell [Oxford: Clarendon Press, 1966], see pp. 217-218 of Collected Papers). Fredson Bowers discussed Wilson's scholarship as a point of tension between McKerrow and Greg in On Editing Shakespeare, a collection of lectures and articles (Charlottesville: Univ. Press of Virginia, 1966), see nn. 2-3, pp. 181-186.

22. McKerrow repeatedly referred to his critical editions as "reprints"—thogh he emended his copy-texts. Greg called this peculiarity "symptomatic," though he did not say of what; later, Tanselle offered clarification: " `symptomatic'—that is, of McKerrow's pervasive reluctance to give rein to individual judgment" ("The Rationale of Copy-Text," p. 24 n. 9; Tanselle, "Greg's Theory of Copy-Text and the Editing of American Literature," Studies in Bibliography 28 [1975]: 176 n. 9).

23. That McKerrow was a reluctant conservative has been lately and sensitively demonstrated by Marcel De Smedt in his engaging review, "R. B. McKerrow's Pre-1914 Editions" (Studies in Bibliography 55 [2002]: 171-183); obviously aware of the issues on which the copy-text debate turned, De Smedt focuses some attention on works within McKerrow's edition of Nashe that survive in more than one authoritative text. One need not accept De Smedt's view that Greg "misleadingly" applied the "best text" (p. 179) label to McKerrow's choice of late copy-texts to agree with him that McKerrow (not, in this respect, unlike Bédier) was a thoughtful editor and did not treat these copy-texts with undue reverence.

24. "The Rationale of Copy-Text," p. 23.

25. Ibid., p. 20.

26. Ibid., pp. 29-30.

27. Bowers, "Some Principles for Scholarly Editions of Nineteenth-Century American Authors," paper read on 22 November 1962 at a meeting of the South Atlantic Modern Language Association, published in Studies in Bibliography 17 (1964): 223-228, see p. 228.

28. Bowers, "Some Principles for Scholarly Editions of Nineteenth-Century American Authors," Studies in Bibliography 17 (1964): 227; cf. McKerrow, Prolegomena, p. 6: "For scholarly purposes, the ideal text of the works of an early dramatist would be one which . . . should approach as closely as the extant material allows to a fair copy, made by the author himself, of his plays in the form which he intended finally to give them."

29. The term "substantive text" was ordinarily used in this way; Greg's use of the word "substantives" to mean the wording of any document was thus doubly unhappy: it could suggest to readers that punctuation was insignificant, and also that wording, by virtue of its being wording, was authoritative.

30. McGann, A Critique of Modern Textual Criticism (Chicago: Univ. of Chicago Press, 1983; repr. with a foreword by D. C. Greetham and preface by the author, Charlottesville: Univ. Press of Virginia, 1992), p. 57 (all citations to reprint edition).

31. McGann, The Textual Condition (Princeton: Princeton Univ. Press, 1991), p. 50.

32. Bowers, "Textual Criticism," in The Aims and Methods of Scholarship in Modern Languages and Literatures, ed. James Thorpe (New York: Modern Language Association, 1963), pp. 23-42; 2nd ed., 1970, pp. 29-54, quotation from 2nd ed., p. 54; Bowers made similar statements in his 1962 paper, "Some Principles for Scholarly Editions of Nineteenth-Century American Authors" (p. 228), and in "A Preface to the Text" in The Scarlet Letter, The Centenary Edition of the Works of Nathaniel Hawthorne, Volume I, ed. William Charvat, Roy Harvey Pearce, Claude M. Simpson, Fredson Bowers, and Matthew J. Bruccoli (Columbus: Ohio State Univ. Press, 1962), p. xlvii.

33. Bowers, "Textual Criticism," in Thorpe 1970, p. 33.

34. The Scarlet Letter, The Centenary Edition of the Works of Nathaniel Hawthorne, Volume I, p. xli; the assertion about the insignificance of accidentals was repeated in volumes II-IV of the edition (in "The Centenary Texts: Editorial Principles," the revised version of the essay which appeared in Volume I as "A Preface to the Text").

35. McGann, A Critique of Modern Textual Criticism, 1992, p. 113.

36. Ibid., p. 104.

37. Ibid., pp. 112, 113.

38. Ibid., p. 122.

39. Studies in Bibliography 44 (1991): 83-143, see pp. 103-113.

40. "Ulysses as a Postmodern Work," in McGann's Social Values and Poetic Acts: The Historical Judgment of Literary Work (Cambridge: Harvard Univ. Press, 1988), pp. 173-194, quotations from pp. 175, 181; McGann's reason for not typographically emphasizing "Ulysses" in his title is not explained.

41. McGann, The Textual Condition, pp. 52, 96.

42. Tanselle, "Textual Criticism and Literary Sociology," p. 106 n. 30.

43. "Ulysses as a Postmodern Work," p. 177.

44. The Textual Condition, pp. 72, 73.

45. Studies in Bibliography 39 (1986): 1-46, see p. 23.

46. For these works it was assumed that an apparatus record of the variants necessary to re-create the other versions would be too complex to use easily. Such assumptions— another holds that accidental variants are always too numerous to print—are easily made, and rarely tested; see Tanselle's study, "The Editorial Problem of Final Authorial Intention," in Studies in Bibliography 29 (1976): 167-211, especially the discussion in section III, pp. 191-207.

47. "The Rationale of Copy-Text," p. 29.

48. McGann, A Critique of Modern Textual Criticism, 1992, pp. 6-7; McGann quotes a lengthier passage, and refers to his source as "CEAA/CSE Introductory Statement, 3" (p. 131 n. 9). His quotation and citation vary in some details from the version studied for this article, which is: "The Center for Scholarly Editions: An Introductory Statement" (New York: Modern Language Association of America, 1977); in this version the relevant passage appears on p. 2.

49. "Statement of Editorial Principles and Procedures: A Working Manual for Editing Nineteenth-Century American Texts," rev. ed., Center for Editions of American Authors (New York: Modern Language Association of America, 1972), p. 4.

50. Studies in Bibliography 28 (1975): 167-229, quotations from pp. 174, 180, 201.

51. "The Meaning of Copy-Text," Studies in Bibliography 22 (1969): 311-318, quotations from p. 314.

52. Studies in Bibliography 23 (1970): 191-196, quotation from p. 195.

53. Greg, "The Rationale of Copy-Text," p. 26.

54. The Riverside Shakespeare, 2nd ed., ed. G. Blakemore Evans et al. (Boston: Houghton Mifflin, 1997), p. 67.

55. Quoted in "The Life and Work of Fredson Bowers," Studies in Bibliography 46 (1993): 1-154, see p. 105; the persistence of the charge of aesthetic idealism made against "eclectic" editors is partly recorded in Tanselle's later essays, "Textual Instability and Editorial Idealism," SB 49 (1996): 1-60, and "Textual Criticism at the Millennium," SB 54 (2001): 1-80.

56. McGann, A Critique of Modern Textual Criticism, 1992, pp. 56-57.

57. McGann, The Textual Condition, p. 62.

58. Text 9 (1996): 11-32, quotation from p. 13; quotations in this note and the next are taken from the 1996 version in Text. Additional forms in which McGann's "Rationale" appeared are recorded in Tanselle, "Textual Criticism at the Millennium," Studies in Bibliography 54 (2001): 35 n. 54. In his essay, McGann slights the advantages of books, while making exaggerated claims for electronic media. Careful criticism of the "information revolution" is called for from scholars who understand more than its potential advantages. A belief in the inevitability and even the omnipotence of computers has characterized some of the computer advocacy coming from the administrative hierarchies of many libraries, for example, which is causing fewer scholarly books and journals to be purchased, and even the "de-accessioning" of collections, to make room for more computers. Secondary consequences include the depressed state of academic publishing, as scholarly editions of literary and historical works, which were once assured of finding their way into thousands of libraries, can now count only on hundreds. The continuing necessity of books and printed materials from a humanistic standpoint can be demonstrated, in part, by revealing the comparative—and often hidden—disadvantages of computers and computerized texts: their elitism, reflected in the great overall expense and complexity of computers; their impermanence, being subject to, among other inconstancies, the pursuit by their designers, including the U.S. military and the computer companies, of a radical policy of planned obsolescence; the lack of privacy that attends their use; and—not least—their lower compatibility, compared to "hard copies," with human perceptive and cognitive faculties.

59. McGann emphasizes the inconvenience of the codex form in his "Rationale of HyperText," the better to assert the "revolutionary" nature of the computer: "The change from paper-based text to electronic text is one of those elementary shifts—like the change from manuscript to print—that is so revolutionary we can only glimpse at this point what it entails" (p. 28). The introduction of printing was not a perfect boon to learning (inferior manuscripts were sometimes widely printed; valuable ones were sometimes discarded after serving as printer's copy), but it did make reading easier, or at least no more difficult. McGann's "hypertextual" alternative to printed scholarly editions is, in theory at least, an electronic archive of all the substantive texts of a work in their entirety. Where the apparatus of a print edition traditionally records variants alone, keyed to a single critical text, McGann's "noncentralized" electronic archive would contain all the texts, arranged so that none "is privileged over the others" (p. 31). Such an archive, of course, is to be welcomed, but it must not be regarded as a substitute for original documents or for scholarly books. "Navigating" through a hypertext archive to arrive at the textual history of a work so presented is, for example, a fairly complex process, according to McGann's description of it (p. 31)—and more complex than reading a traditional printed presentation in a book, at least in that it requires special computing skills and access to expensive machinery and Internet connections. A further problem may arise in reader reluctance to read through multiple complete texts of long works, such that this form of presentation could easily obscure, rather than highlight, textual variation. Finally, and not least important, is that like earlier claims about the "paperless office," the claim regarding the "change from paper-based text to electronic text" is in itself misleading. Few readers, that is, will want to read even one version of a novel on-screen, let alone several. The alternative being bulky, unbound computer printouts, the electronic text might be seen as a step backward from the codex, rather than a revolutionary improvement, since for the most elementary purpose—reading—it must rely on a technically primitive form of the technology it is supposedly replacing. McGann is obviously sensitive to many of these issues, but, in "The Rationale of HyperText" at least, he does not clearly separate himself from a kind of triumphalism characteristic of some computer advocacy, warning dissenters, for example, that "no real resistance to such developments is possible, even if it were desirable" (p. 11).

60. It is useful for the discussion that follows to note that Robert Hirst, with his edition of the letters of Mark Twain, has called into question the assumption that the most satisfactory way to represent the texts of manuscript documents is by means traditionally regarded as "noncritical," such as photographic or type facsimiles, or systems of "genetic" transcription. The disadvantages of these alternatives are well known. Photographs unaccompanied by transcriptions force the reader to decipher unfamiliar handwriting (including details of revision) at one remove (or more) from the original; type facsimiles can demand complex feats of typesetting, and at points the principle of visual identity must be abandoned altogether, and some hybrid system developed. Genetic transcription conveys manuscript details through editorial commentary (whether in verbal or symbolic form) inserted into the author's text. That the result is frequently illegible, more closely resembling a textual apparatus than the text of the original document, accounts for why most scholarly editions of letters have opted to make critical presentations in "clear text." Here the editor presents the reader with the text as revised, as if for publication, with the details of revision reported in the apparatus. Legibility is no longer a problem, but fidelity is. For Mark Twain's Letters, Hirst devised a middle way, according to his view that the distinction between noncritical and critical is basically important for scholarly texts, but more a matter of emphasis along a continuum rather than a stark division. His compromise, called "plain text," is a critical text, in that it relegates some manuscript details to an apparatus, though the text includes more of these details than many professedly noncritical systems. Hirst achieved his result by deepening his understanding of nineteenth-century letter-writing and typographic conventions, and by rationalizing the use of symbols and other typographic equivalents accordingly (including strike through horizontal rules for deletions, sublinear carets for insertions, and italics for underlining); see Robert H. Hirst, "Guide to Editorial Practice" (revised, February 2002), in Mark Twain's Letters Volume 6: 1874-1875, ed. Michael B. Frank and Harriet Elinor Smith (Berkeley: Univ. of California Press, 2002), pp. 697-725.

61. Some years ago Tanselle identified and discussed exactly these two possible varieties of the social text, as "a publisher's intended text" requiring "critical reconstruction" on the one hand, and an "unaltered reproduction" of "what the publisher in fact did publish" on the other. Tanselle sees the latter variety as the most useful complement to authorially intended critical texts; see "The Text of Melville in the Twenty-First Century," his contribution to Melville's Evermoving Dawn: Centennial Essays, ed. John Bryant and Robert Milder (Kent, Ohio: Kent State Univ. Press, 1997), pp. 332-345, quotations from pp. 337-338.

62. "Editing without a Copy-Text," see pp. 5-6; the longer discussion appears in A Rationale of Textual Criticism, A Publication of the A. S. W. Rosenbach Fellowship in Bibliography (Philadelphia: Univ. of Pennsylvania Press, 1989); see especially chapter 1, "The Nature of Texts," pp. 11-38.

63. A view of the "edition as a whole" can be attained with more or less clarity, according to the number of copies examined proportional to the original total. The concept of "ideal copy," a bibliographical description of all variations present in the copies of an edition as they left the control of the publisher, emerges in this context. In the course of defining this concept more than twenty-five years ago, Tanselle had cause to refer to the bibliographical responsibility of editors of noncritical editions. These remarks are especially pertinent: "Just because editors of noncritical editions do not have to make critical choices among individual variant readings does not absolve them of the responsibility for knowing what variants exist within the edition they are concerned with; choosing a copy for reproduction is itself a critical choice, and it should be as informed a choice as possible. When a descriptive bibliography has sorted out the various states comprising the ideal copy (or, it might be better to say, the ideal copies) of an issue, the editor of a noncritical edition can use this information with great profit. But when that bibliographical work has not been performed, the responsible editor of a noncritical edition has no alternative but to undertake the task. Editing, even of noncritical editions, cannot be divorced from descriptive bibliography and from the concept of ideal copy." ("The Concept of Ideal Copy," Studies in Bibliography 33 [1980]: 18-53, quotation from p. 37.)

64. That Tanselle's model for constructive critical editing is the approach to radiating texts already proven by him and Bowers, among others, has implications for the scholarly apparatus, though these are not spelled out. The author's name alone should, however, be indication enough that the scholarly standards inherent in editing without a copy-text are very high, and involve the scrupulous disclosure of bibliographic and textual evidence.