:: :: University of Virginia Library

If one considers systematic textual criticism to have begun with the Alexandrian librarians of the third century B.C., then one can say that for twenty-two and a half centuries the ultimate goal of textual criticism was—almost without exception—the establishment of texts as intended by their authors. The controversy that has always characterized the field was concerned with how best to approach this goal, not with whether this goal was the proper one. Sometimes editors recognized that authors may have changed their intentions over time; but if there was some doubt about whether early or late intentions were to be preferred, the focus was still on authorial intentions. Even when editors of ancient writings stated that they were reconstructing the text of the lost common ancestor of the extant manuscripts, knowing that it could not be equated with the text intended by the author (if indeed there was a single author), they were nevertheless attempting to move backward in time with the aim of coming closer to the author's intended text (or the text intended by the various creators of the work) than anyone had previously come. Most editors of the past have simply assumed, without giving much thought to the matter, that the purpose of critical editing was to correct the texts that have survived in documents, so as to bring them more into line with what the authors of the works intended.

In recent years, however, several writers on textual criticism have questioned this assumption, and their arguments have been so widely noted and discussed that the issue can probably be regarded as the dominant one in current theoretical debates. The line of argument runs as follows: authors cannot normally bring their works to the public without the assistance of other persons, such as scribes, printers, publishers' editors, and publishers, who in various ways alter the texts that pass through their hands; literature is thus a collaborative art, the joint product of a number of people; a concern with what authors alone intended is therefore artificial, since works can only be produced in the forms (both linguistic and physical) that the social process of publication

gives to them, and it is in those forms that readers encounter and respond to them (forms that keep changing over time as new publishers and new readers deal with them in different terms). This position shifts the emphasis from the individual creators of literary works to the social mechanisms for the dissemination and reception of particular texts of those works.[1] For the first time in twenty-three centuries, a significant segment of that body of individuals who consider themselves textual critics and editors are critically examining the place of authorial intention among the goals of editing. This is not to suggest that there has been any diminution in the proportion of editions focusing on authorial intention but only that more editors than ever before are cognizant of the reasons for considering an alternative approach.

This trend in textual theory is part of a larger movement in literary studies. During the same years, deconstruction, the "new historicism," and reader-response theory—among other approaches—have supported a turn away from the authorial and the canonical. Language is seen to betrary those who attempt to express themselves through it, and meaning is found to emerge from historical contexts and from the encounter of readers with texts. Although there has always been little (far too little) interchange between textual and literary critics, they do inhabit the same intellectual world, and it is to be expected that the concerns of both groups should be touched by the same currents of thought. One of the effects on textual criticism has been a new emphasis on the instability of texts, on their indeterminate nature. In this context, some theorists have regarded the editorial aim of reconstructing an authorially intended text as a misguided attempt to fix a text in a single form, hiding rather than revealing the fluidity and openness that are characteristic of texts. The idea that texts are the ever-shifting products of converging social forces is compatible with those approaches to literature that elevate linguistic analysis, historical associations, and readers' responses over the effort to receive a communication from an individual in the past.

Editors of nineteenth- and twentieth-century literature have faced this set of issues for a longer time than some of them realize, and certainly before the time when a concern for social approaches to textual criticism was much in the air. The frequent survival of authors' manuscripts from these centuries has meant that editors dealing with this period have often had to decide between an author's manuscript and a first printed edition as the best choice for copy-text. When they chose

the first edition, their decision was generally made in the name of authorial intention, for they argued that the first edition contained the kinds of adjustments (especially in spelling and punctuation) that the author expected and desired the publisher to make. But whenever one speaks of an author's expectation that certain matters will be taken care of, one is involving other persons in the completion of the work and embracing a view of literary works as collaborative products. Although there can be instances in which an editor may decide that an author's uninfluenced intentions are best represented in a first edition, one can say in general that the choice between manuscript and first edition often reveals whether the editor's primary interest is in the product of an individual creative mind or of collaborative action. Greg's rationale of copy-text, favoring the former, has come in for considerable criticism in recent years by those who object to its orientation and thus to its wide influence. It is a healthy situation in any field for positions perceived as orthodoxies to be challenged, and the recent debates have unquestionably been useful in bringing increased recognition to the instability of texts and to the ways in which they are continually being shaped by society. An interest in the texts that emerged from the process of publication and were available to readers at given times is obviously one (but not the only) valid approach to the past, and it had not previously been altogether overlooked; nevertheless, its new visibility is welcome.

This attention would be more welcome, however, if the advocates of a socially oriented textual criticism wrote on its behalf with greater clarity and coherence. Unfortunately, many of the recent discussions are so carelessly presented that they could convince no one; but since thoughtful readers will understand in any case the value of looking at texts as social products, they may derive some new insights from these essays despite the illogical presentation. In what follows, I shall survey some of the theoretical writings on textual matters that have appeared during the second half of the 1980s, taking the social approach to textual criticism as the obvious theme.[2] I shall concentrate first on some

recent work of D. F. McKenzie and Jerome J. McGann, whose writings in the late 1970s and early 1980s (discussed in my survey five years ago) had first brought significant attention to the social view and who are still generally regarded as its most prominent exponents (though their emphases are not identical). It will then be necessary to examine several writers who are concerned, in one way or another, with the integrity of individual versions of works. This concern is a manifestation of the same set of attitudes, for an emphasis on versions reinforces the idea of textual fluidity. And even when the emphasis is on prepublication versions, some writers (notably a group of French and German critics) view

such versions as evidence for assessing authors' attitudes (sometimes welcoming, sometimes apprehensive) toward the transition from private creativity to public professionalism. As for post-publication versions, one of the most interesting cases is drama, where the contrast between a text close to the playwright's manuscript and a text reflecting what occurred in performance epitomizes the dichotomy between uninfluenced authorial intention and the joint intention that emerges from the collaborative process whereby the playwright works with others to bring a performance into being. Finally, a brief overview of other recent essays, showing the prevalence of these issues, leads to a reconsideration of editorial apparatus, for not surprisingly this subject has surfaced repeatedly in the writings of the last few years: in many ways it lies at the heart of these debates and offers a means of reconciling the extreme positions that have been taken.

I

In 1985 the inaugural series of Panizzi Lectures at the British Library was delivered by D. F. McKenzie; published as Bibliography and the Sociology of Texts (1986), these lectures have reverberated through many of the subsequent discussions of textual matters.[3] Those familiar with McKenzie's writings of the previous ten years will recognize the ideas developed here; but unfortunately these lectures are often as confused in argument as his Wolfenbüttel piece on Congreve and lack the eloquence and power of his Bibliographical Society address.[4] The reason for the influence of these lectures is that their angle of approach is in tune with the intellectual temper of the times, not that particular ideas

are offered with such rigor that the sheer force of the argument carries the day. Many of the principal points, when abstracted from their presentation, are in fact sound, and not particularly new. One can readily agree, for example, that bibliographical studies are historical studies (p. 3); that textual criticism encompasses the examination of texts transmitted in any form, not just in ink on paper or parchment (p. 4); that it is important to study "the social, economic and political motivations of publishing, the reasons why texts were written and read as they were, why they were rewritten and redesigned, or allowed to die" (p. 5); and that "the material forms of books, the non-verbal elements of the typographic notations within them, the very disposition of the space itself, have an expressive function in conveying meaning" (p. 8). None of these points is startling, and none, it seems to me, can be denied. Any cogent exposition of these matters would always be welcome and could provide a constructive point of reference for further discussion; McKenzie's lectures do not serve this function because the central points are not argued coherently and indeed are sometimes trivialized.

The opening pages of the first lecture ("The Book as an Expressive Form") reveal these weaknesses. McKenzie takes his first task to be a new definition of the scope of bibliography, and he begins by examining W. W. Greg's 1932 statement that bibliographers are concerned with printed texts as inked type-impressions on paper, as "arbitrary marks," not as groupings of words with meanings. McKenzie's style of thinking is represented by his first observation on this statement: "it remains in essence the basis of any claim that the procedures of bibliography are scientific" (p. 1). But surely what makes a procedure "scientific" (in the usual definitions of the term) is the method followed, not the exclusion (or inclusion) of a particular body of evidence (in this case the inked impressions as symbols for letters, which form words of a language). In any event, McKenzie finds Greg's statement "no longer adequate as a definition of what bibliography is and does" (p. 2). Because McKenzie is particularly interested in books as "expressive form," in book design as a purveyor of meaning to readers, he asserts that "bibliography cannot exclude from its own proper concerns the relation between form, function and symbolic meaning" (p. 2). But then he recognizes, a few paragraphs later, that bibliography has "consistently studied" book design, production, distribution, and collecting, as well as textual transmission: "no part of that series of human and institutional interactions is alien to bibliography as we have, traditionally, practised it" (p. 4). He further understands that "Physical bibliography—the study of the signs which constitute texts and the materials on which they are recorded—is of course the starting point." And then he adds, "But it cannot define

the discipline" (p. 8). By this time the reader will wonder what the point of his criticism of Greg is. Greg obviously understood the interconnections among all these various "bibliographical" activities (as other parts of his essay show), and he practiced many of them himself; as an editor he took the meanings of words into account, but he felt, just as McKenzie does, that the physical evidence must be thoroughly investigated first. When he said that bibliographers regard letter forms as arbitrary marks, he was of course describing what McKenzie calls "physical bibliography" and not attempting to characterize the whole interrelated congeries of bibliographical studies. Greg and McKenzie would seem not to be at odds after all.[5]

What, then, is the basis for McKenzie's repeated worry about Greg's statement? Why, at the end of the lecture, does he say that "Greg's definition of what bibliography is would have it entirely hermetic" (p. 19)? One cannot believe that McKenzie is so literal-minded that he objects to Greg's use of "bibliography," in a context that makes clear what he meant, instead of "physical bibliography." Suppose, for the sake of argument, that Greg did wish only physical bibliography to be called "bibliography": what difference would that have made, since he plainly grasped the relationship between it and all the other aspects of textual study? If, as McKenzie acknowledges, the entire textual cycle, from composition by an author to response by a reader, has generally been understood to be linked together and deserving of integrated study, what does it matter whether it is called "bibliography" or something else? What difference does it make whether we think of that integrated study as a single field or as a group of related fields? Why is the label important if the substantive connections have been made? As it turns out, there is a reason—an embarrassing reason: "As long as we continue to think if it [bibliography] as confined to the study of the non-symbolic function of signs, the risk it runs is relegation. Rare book rooms will simply become rarer. The politics of survival, if nothing else, require a more comprehensive justification of the discipline's function in promoting new knowledge" (pp. 3-4). The phrase "if nothing else" does not accomplish its mission of making the point seem casual and subordinate.

To speak at all of the "politics of survival" trivializes the discussion beyond redemption. A "discipline" does not exist for the purpose of self-perpetuation; if it requires "justification" as a political strategy for survival, it had better be allowed to die.[6] McKenzie's politicizing of scholarship crops up again when he says that Greg's "confinement of bibliography to non-symbolic meaning, in an attempt to give it some kind of objective or 'scientific' status, has seriously impeded its development as a discipline" (p. 8). It is a misreading of history, as a strategy of "survival," to claim that Greg's statement, or the view it represents, impeded progress in any way. Bibliography, in the broad sense, has benefited immensely from the recognition of the role physical evidence plays in the study of textual transmission, and Greg was one of the scholars who established that field of study; it is understandable that, in addressing the Bibliographical Society in 1932 on its fortieth anniversary, he would emphasize physical bibliography, for the relation of physical evidence to literary study was the most far-reaching insight that had emerged under the aegis of the Society. Ten years later, in a fiftieth-anniversary paper, he stated that "the object of bibliographical study is . . . to reconstruct for each particular book the history of its life, to make it reveal in its most intimate detail the story of its birth and adventures as the material vehicle of the living word." Greg was thinking along the same lines as McKenzie—but thinking much more clearly: after pointing out that bibliographical and textual criticism were so interlocked that it was difficult to regard them as separate fields, Greg added, "This is not a matter on which I desire to lay stress, for it is largely a question of terms and therefore of relatively minor importance."[7]

What is of much more importance than deciding which activities we wish the term "bibliography" to cover is examining the attitudes we bring to our work. There can be no objection to thinking of bibliography as "a sociology of texts" (p. 8) if by that is meant an openness to all

the kinds of research that bear on the history and influence of textual transmission. But McKenzie, despite his inclusive view of bibliography, has a curiously limited approach to historical investigation. His well-known skepticism about analytical bibliography[8] emerges here, as when he says that compositor studies have displayed "virtuosity in discerning patterns in evidence which is entirely internal, if not wholly fictional" (p. 7). If a pattern—or, more precisely, a conclusion based on a perceived pattern—is "entirely internal" (that is, unsupported by information in a document external to the book under examination), is it necessarily false or unworthy of serious consideration by historians? How does he think that facts (i.e., what we take to be facts until they are shown to be unsatisfactory) get established in the first place? Do we not search for patterns in external evidence, too, as part of the process of attempting to make sense out of a welter of data? He had earlier noted that bibliographical analysis "depends absolutely upon antecedent historical knowledge" (p. 2). But how is a body of "knowledge" built up if not by examining all kinds of evidence, and does not "antecedent" knowledge continually have to be revised as new pieces of evidence, or new interpretations of evidence, emerge? Analysis does not depend "absolutely" on antecedent knowledge, for the possibility must remain open that the results of analysis will overturn previously accepted views. When McKenzie later admits that physical bibliography is "the starting point" (p. 8), one has to wonder how sound a start it is if it excludes the kinds of analysis of physical evidence that are generally grouped under the term "analytical bibliography." More fundamentally, one has to worry about an approach to history that bans a basic category of evidence. Bibliography, he insists, is "the historical study of the making and the use of books and other documents" (p. 3). But his history, as it emerges, has a program, which is not hospitable to analytical bibliography; and by rejecting a large body of bibliographical evidence, he makes it impossible for readers to respect his approach as an openminded search.

A major part of his program is to understand the physical book as "an expressive form"—and therefore to look at physical details not as evidence of the book-production process but as indicators of the cultural values surrounding that production and as determinants of readers' responses.

His principal illustration of this point, occupying the entire latter half of the first lecture, is the epigraph to W. K. Wimsatt, Jr., and M. C. Beardsley's famous 1946 essay "The Intentional Fallacy." Noting that the four lines quoted from Congreve's prologue to The Way of the World differ from the "authorized version" of 1710 in wording at one place, in punctuation at four places, and in capitalization at six, McKenzie argues that the 1946 readings alter the meaning of the passage to one that better suits Wimsatt and Beardsley's purpose. Whereas Congreve was asserting the rights of authors to establish their own meanings, he says, the 1946 quotation reverses the sense, encouraging "audience and readers to discount the author's meaning" (p. 14). This example of textual variation is not an apt illustration of all the points that McKenzie wishes it to demonstrate, and one wonders why he was content to offer it at length (unless for the amusing coincidence between the apparent content of the passage and the phenomenon it is cited to illustrate). It is in fact a much more conventional illustration than McKenzie seems to think, for what it demonstrates are the dangers of modernizing and of accepting any text that comes to hand. As McKenzie admits late in his discussion, Wimsatt and Beardsley in all likelihood took their quotation from a popular anthology, not from the 1710 edition; and although they presumably selected the passage because it seemed to support their argument, they were apparently not responsible for misquoting it to give it that meaning.[9] Furthermore, the changes in punctuation and in capitalization were no doubt made by the anthologists in an effort to modernize the text for students. The problems that unauthorized variants and modernized texts create for unwary critics have often been commented on; and it is no revelation to be shown that variant wording, punctuation, and capitalization are textual matters, for they are regularly regarded as such. What McKenzie's argument calls for is an illustration of the way in which format, type design, and layout can

convey meanings and are thus "textual." In his earlier Congreve essay he did cite such elements of design as integral parts of Congreve's intended texts, and it is puzzling that he does not do so here. He claims to have dealt with the signification of "the fine detail of typography and layout" (p. 16); but his discussion of Wimsatt and Beardsley scarcely touches on those matters, and the lecture therefore does not really engage its avowed topic, "the book as an expressive form."[10]

Despite the failure of the Congreve example to serve its requisite function, readers of the lecture are presumably willing in any case to agree in theory to many of McKenzie's conclusions, for there can be no question that the physical forms in which texts are presented to the public are the products of printers and publishers (and sometimes authors as well), that these forms reflect cultural influences, and that they in turn affect the interpretations of readers. One accepts these points as matters of common sense, not because McKenzie has provided any new insights into them.[11] Indeed, he confuses the issues because he does not coherently distinguish the procedure for locating authorial intention from that for assessing works as communal products. After pointing out that in reading contemporary editions of Congreve we must consider the contributions of the printer and the publisher to the "interpretation of Congreve's meaning" (p. 17), he asks, "Who, in short, 'authored' Congreve?" But this question (surprisingly) implies a single "Congreve," a single text, and does not recognize the distinction between the text of the printed artifact and the text of the literary work as intended by Congreve (or others). Yet earlier he had enumerated some of the issues in the history of the book as "What writers thought they were doing in writing texts,[12] or printers and booksellers in designing and publishing them, or readers in making sense of them" (p. 10). No single text can accommodate both the first (the authorially intended text) and the second (the text as published). Instead of asking who

"authored" Congreve's works, we have to ask whether we are interested in what Congreve himself "authored" or what he "authored" in collaboration with printers and publishers. There is of course a legitimate interest in both, as well as in the meanings that readers have derived at different times from the texts available to them. But McKenzie sums up this variety in a troublesome way: "My argument," he says, "therefore runs full circle from a defence of authorial meaning, on the grounds that it is in some measure recoverable, to a recognition that, for better or worse, readers inevitably make their own meanings" (p. 10). The sequence is hardly a "full circle," for readers' meanings can arise from any public text and do not necessarily derive from one intended by the author, if indeed such a text has ever been "recovered." And to justify the search for authorial meaning "on the grounds that it is in some measure recoverable" misstates the basis for historical research. We never know whether anything is recoverable, nor do we know when we have in fact recovered something; all we can do is attempt to move in the direction of recovering whatever we have decided is worth recovering.

Deciding that the past can be of interest or relevance is the crucial matter, not how recoverable it is; and McKenzie's naïve view of this question undercuts the peroration of his first lecture, as it has weakened all that went before. He says that critical movements from "New Critical formalism" to deconstruction "share the same scepticism about recovering the past" (p. 19). Of course they do—and they share it with all thinking individuals. What distinctively characterizes those movements is—in varying degrees—a lack of interest in the past, a rejection of the past as a useful concept. Whether a rejection of the past is a rejection of "human agency" is another question, and not one obviously to be answered in the affirmative, as McKenzie does. After all, the "critical self-absorption" that he sees in these approaches is also an example of the human effort to come to terms with the verbal artifacts that surround us. Our studies cannot so easily be robbed of humanity as McKenzie imagines, nor can bibliography so easily reinstate the human element by dealing with "discoveries as distinct from invented meanings" (p. 19). It would be far more understanding of humanity to recognize that discoveries are invented meanings, for they are known only through the exercise of human judgment. McKenzie has become so convinced of the idea that recent criticism is anti-humanistic that he reserves his last line for proclaiming that bibliography can correct this direction by showing "the human presence in any recorded text" (p. 20).[13] Leaving aside his assessment of

contemporary criticism, one may question his vision in its own terms, for—despite his earlier inclusion of authorial intention among the concerns of bibliography—he now seems to exclude it. The emphasis on successive texts as recorded forms, he says, "testifies to the fact that new readers of course make new texts, and that their new meanings are a function of their new forms";[14] as a consequence, one approaches these texts "no longer for their truth as one might seek to define that by authorial intention, but for their testimony, as defined by their historical use" (p. 20). Why authorial intention is not one instance of human agency in the past is left unclear. McKenzie is waging an unnecessary battle if he thinks it will be difficult to convince anyone that texts as they change through time provide evidence for readers' changing responses; but the incoherence of his remarks about authorial intention and its place in historical study will nevertheless set up a barrier between him and his audience.

These problems persist through the second and third lectures: there is scarcely a paragraph not weakened by them. The second lecture, "The Broken Phial: Non-Book Texts," begins with several pages that continue the emphasis on expressive textual forms as "less an embodiment of past meaning than a pretext for present meaning" (p. 25). In other words, "Meaning is not what is meant, but what we now agree to infer" (p. 26). But one of the present meanings or inferences—though anyone is free not to be interested in it—is what we believe we can conclude about the meanings of a work at a given time in the past. Among those past meanings are the authorially intended ones; and a concern with authorial intention does not, as McKenzie seems to believe, contradict the idea of textual instability (p. 28), for authors' intentions shift with time, and our reconstructions of their intended texts can never be definitive. To say that the concept of the authorially intended text "has largely collapsed" (p. 28) is merely (and ineffectively) provocative (and indeed is at odds with the calmer—and true—statement, a few lines later, that it "no longer compels universal assent"). The ensuing statement that "The only remaining rule seems to be that we must not conflate any one version

with any other" (p. 29) is simply incorrect. This "rule" is not in fact the only one—or the only respectable one—now being followed. And the implied connection between a declining interest in authorial intention and a rising distaste for eclecticism is not made clear; a connection can be made, but theoretically, of course, the arguments for and against eclecticism are relevant to any editorial goal.

The core of the second lecture is an examination of "non-book texts," and McKenzie takes up, in turn, landscape, maps, photographs and film, and theater. That texts in all media pose textual problems and that "textual criticism" encompasses all such problems are not new ideas.[15] Perhaps McKenzie is not claiming that they are, but he does seem to think he is exhorting "bibliographers" to broaden their outlook. What he is calling them to is unclear, however. His first illustration—the significance of landscape for the Arunta tribe—is meant to show how "the land itself" can be "a text" (p. 31), how topographical features can have "a textual function" (p. 32). But he confuses his point by explaining that these natural features "form the ingredients of what is in fact a verbal text, for each one is embedded in story . . . and supports . . . the symbolic import of a narration." In that case, he is talking about landscape as a visual supplement to, or embodiment of, a verbal text; but it is not the verbal narrative that makes the landscape a text, for the landscape is a text in its own right, a nonverbal text made up of physical objects, a text that can be read in the way sculpture can. McKenzie, here and later, does not clearly distinguish between, on the one hand, the nonverbal media in which communication can take place (as in painting, music, and dance) and, on the other, the various nonbook vehicles that are used to transmit works in the medium of language. He does, in some examples, deal with visual and kinetic texts;[16] but he never makes clear that

the relation of photographs, for example, to printed and manuscript books is of a different order from the relations among sound recordings, computer disks, and books as conveyors of verbal texts. His discussion is meant to argue for "the centrality of a textual principle in bibliography" (p. 43); but that principle is obscured by his failure to make these discriminations.[17] And would it not be more constructive to emphasize the centrality of textual criticism to all fields? Whether we define "bibliography" as essentially textual is less important than whether we understand the basic role that bibliographical and textual investigation plays in every field. Whether "bibliography" subsumes all such investigation is of less moment (except in academic politics) than whether the work gets done, and gets done by those who in each instance bring to it a knowledge of the field concerned.

The third lecture, "The Dialectics of Bibliography Now," showing little sense of progression, offers two more examples of the book as expressive form (drawing on Locke and Joyce) and one extended example of the textual study of a film (Citizen Kane). It is something of a letdown to find out that a principal point of the latter discussion is to demonstrate that films, like books, have "texts" and that "the word now has a meaning which comprehends them all" (p. 56). Obviously films and books (and all other physical objects) display patterns of details that we can agree to call "texts"—though whether we actually do call them that is a trivial matter, further trivialized by the insistent restatement that "the discipline [bibliography] comprehends them both [films and books]" (p. 59). As before, the more meaningful point is passed over: that films provide examples of textual problems in a different medium from literature, not just in a different form of transmission. Film is a different art from literature, whereas literature stored in a computer and literature stored in a book are not two different arts.[18] When McKenzie

says that bibliography is "committed to the description of all recorded texts" (p. 51), the term "recorded texts" glosses over an important question: whether it refers only to the concrete forms through which works in intangible media are primarily transmitted (that is, books, sound recordings, motion picture films, and so on) or whether it also includes works that exist in physical form (that is, paintings, sculptures, buildings, and so on). His imprecision in thinking about this question is illustrated by the observation that "Whereas libraries have held books and documents as physical objects, computer systems have been mainly concerned to retrieve content" (p. 60). But computer systems also inevitably hold their "content" in physical objects (tapes and disks); and librarians have generally been more concerned with "content" than with the preservation of objects, as their practice of replacing one edition of a work with a "reprint" (or microfilm) of it, or with another edition, shows. The two parts of the sentence are not parallel in focus, and the statement thus obscures, rather than illuminates, the relationship between books and computers.

The weaknesses of these lectures are epitomized in the opening sentences of this third lecture, when McKenzie summarizes his two contrasted "concepts of 'text'" in this way: "One is the text as authorially sanctioned, contained, and historically definable. The other is the text as always incomplete, and therefore open, unstable, subject to a perpetual re-making by its readers, performers, or audience" (p. 45). These two sets of attributes do not in fact distinguish the concept of the authorially intended text from that of the collaborative text (produced by publishers, readers, actors, and so forth, both contemporary with the author and later). What they actually describe is a very different dichotomy, that between texts of documents and texts of works. The text of a surviving document is "contained" and "historically definable"— though how it relates to the author's, or anyone else's, intention is a debatable question. The texts of works in intangible media must always be reconstructed from whatever physical and oral evidence comes to hand and inevitably reflect the predispositions of those doing the reconstructing; thus the texts of works—both authors' intended texts and the texts preferred by others—are "unstable" and "subject to a perpetual re-making." One of the summarizing statements in this lecture is the

assertion that "bibliography is of its nature . . . concerned specifically with texts as social products" (pp. 51-52); but McKenzie apparently fails to see that the attempt to reconstruct authorially intended texts is one of the many activities that readers can engage in as they evaluate the socially produced evidence that survives for their examination. These lectures would scarcely have warranted the space I have devoted to them here if they had not been written by McKenzie and had not, as a result, been given considerable attention by others. They do have some significance as an indication of a current direction in editorial thought, and it is disappointing that they cannot be greeted as an effective manifesto; but their laxity of argument makes them an unstable foundation on which to build.

II

Jerome J. McGann's prominence as a spokesman for the social approach to texts is largely due to A Critique of Modern Textual Criticism (1983),[19] though he has written a number of essays on editing, and his view of the nature of texts is of course evident in his many other writings. A basic statement of his position in the period under consideration here can be found in his review of McKenzie's lectures.[20] His general approach is substantially the same as McKenzie's, and perhaps for that reason he overlooks some of the flaws of those lectures. In any case, he serves McKenzie well, sometimes summarizing McKenzie's points more

100

effectively than McKenzie himself had done (despite his excessive reliance on fashionable jargon). There are two particularly valuable emphases in McGann's review. One is his insistence on the inseparability of the activities traditionally called "textual criticism" and "literary criticism." The other emerges from his understanding of the process of reading a good critical edition with apparatus: "a process by which the entire socio-history of the work—from its originary moments of production through all its subsequent reproductive adventures—is postulated as the ultimate goal of critical self-consciousness" (p. 21, col. 3).

There are some loose ends, however, that prevent his argument from being satisfying. He defines language to be "more properly conceived as an event than a medium," as "an extended field of communicative action"; one therefore "has to take the entirety of the language event as the object of interpretation" (p. 20, cols. 3-4). It follows that the same "text" will wear "different faces" in different situations (p. 21, col. 1). The problem arises in his accounting for these faces:

one might usefully distinguish "the text" (or the poem as a purely linguistic event) from the "version" (or the immediate and integral physical object "through which" the "text" is being executed), and make yet a further distinction of "text" and "version" from the "work" (the term to stand for some more global constitution of the poem). There is a "work" called Paradise Lost which supervenes its many texts and its many versions; to William Blake that work was one thing, whereas to William Empson it was something else; and of course to any one of us the work we call Paradise Lost can be, will be, reconstituted once again. (p. 21, col. 2)

This tripartite classification (of what?) into "text," "version," and "work" is too imprecise to be helpful. If we agree to consider "text" the name for the "purely linguistic event," we have to shift our definition from the one McGann proposed a few paragraphs earlier, where the "text" (almost always in quotation marks) is declared not to be a linguistic event ("'The text' will not be located as the words on the page immediately before one's eyes"). Furthermore, why should "text," the linguistic event, be distinguished from "version," in McGann's sense of the physical object, if the point of the whole argument is that they are inseparable? "Text" and "version" are not two of the "faces" that a "text" (in a different sense) can wear; either the combination of "text" and "version" together (the words and their physical presentation) turns a different face to us under different circumstances, or else the "version" is the "face" that a "text" wears. In the latter case, the "text" (linguistic event) has an independent existence and would seem to be incongruent with the earlier definition of language as "an extended field of communicative

101

action" (the extended field surely including the physical presentation). Even more surprising is the inclusion in this classification scheme of "work" as a "more global constitution of the poem." Presumably this vague phrase means that there is, after all, some meaningful sense in which a poem has an existence apart from its individual occurrences. If the "global" work "supervenes its many texts and its many versions" (note the "its"), how is it to be defined without conceiving of language as an intangible medium and depriving material texts of their postulated primary status? That to each of us "the work we call Paradise Lost can be, will be, reconstituted once again" suggests, as textual critics have traditionally thought, that the physical presentation of verbal texts offers us not the works themselves but the evidence out of which we "reconstitute" the works. This view does not contradict the idea that a document is "meaning-constitutive . . . in every dimension of its material existence" (p. 20, col. 2); but it subordinates that process of deriving meaning to the reconstitution of a work that has no tangible existence. McGann's concept of "the work" thus undercuts his larger argument and aligns him with an approach to texts that he seems at other times to find defective.[21]

This indecision is apparent in another passage as well. McGann identifies three kinds of reading, "linear" correlated with "text," "spatial" with "version," and "radial" with "work" (p. 21, cols. 2-3). But asserting that critical editions are "typically structured so as to enforce spatial and radial reading processes along with the linear process" once again paradoxically raises questions about the role of physical form in the creation of meaning. A critical edition has its own physical form; and the readings of other editions, reported in its apparatus, are a part of that form. If such an arrangement permits "radial" reading—that is, reading texts with the knowledge of other texts in one's mind—it can only do so on the "linear," not the "spatial," level, even if the physical features of the earlier editions are described, because no verbal description of visual effects (nor even a facsimile) can substitute for the actual visual presentation. It follows either that physical form is dispensable as an element in reading or that "radial" reading is not being posited as a full engagement with multiple texts.

The role of physical form is a central issue not only for literary sociology but also for traditional analytical bibliography, which McKenzie and McGann find too restrictive. McGann, seemingly with approval, summarizes McKenzie's "initial, critical remarks on theory of bibliography,"

102

particularly his criticism of Fredson Bowers for saying that "historical bibliography is not, properly speaking, bibliography at all" (p. 20, col. 1). McGann, like McKenzie, does not take into account the shifting usages of the word "bibliography" illustrated by such a quotation or seek to find a stable concept linking apparently divergent views of bibliography. The context of Bowers's statement (in his Encyclopaedia Britannica article on "Bibliography" in 1960) makes clear that he was distinguishing "analytical bibliography," the examination of physical evidence in books for clues to their production history, from historical studies of book-related topics, like type, paper, binding, printing, and publishing. He was certainly not implying that bibliography (i.e., analytical bibliography) was not a tool of historical research or that it did not draw on and contribute to historical studies of book production. Just how much of the whole realm of book history is to be called "bibliography" is unimportant, a mere matter of labeling. What is fascinating in this concern with the definition of bibliography is the downplaying of one kind of physical evidence and the elevating of another. Analytical bibliographers look at the physical evidence that can reveal information about production history; literary sociologists are concerned with the physical details that may have affected readers' responses (or that publishers—and sometimes authors—believed would affect those responses). The work of the analytical bibliographers is seen by the literary sociologists as feeding into an intentionalist view of literature, and as authorial intention loses favor, so must analytical bibliography do so. Although analytical bibliography was largely developed by editors interested in establishing authorially intended texts, it uncovers facts of printing history that are obviously not tied to any one editorial theory. What happens in the printing shop is part of the social process by which texts of works are disseminated and is of direct relevance to literary sociology. But McGann, like McKenzie, chooses to regard analytical bibliography as somehow inimical to history, as something now superseded by the "return to history" that has resulted from the sociological approach.[22]

103

In the course of his comments on critical editions, McGann cites "the excellent recent edition of Ulysses edited by Hans Gabler";[23] he had earlier written at length about this edition in an essay entitled "Ulysses as a Postmodern Text: The Gabler Edition,"[24] which offers further opportunity for examining his position on textual matters. The title under which he later collected the essay, "Ulysses as a Postmodern Work," is misleading, for it is not Ulysses but Gabler's edition of Ulysses (1984) that he sees as a "postmodern" work. The 1922 first edition is the "appropriate modernist Ulysses," whereas Gabler's Ulysses explains the work to us in "a peculiarly appropriate postmodern form" (p. 192). What makes it "postmodern" is its "synoptic text" (Gabler's term), full of variant readings and symbols aimed at showing the compositional development of the work. According to McGann this text is "distinctly postmodern" because "the style is impersonal and maintained in a surface mode ('languaged'); the procedure is intertextual and self-referencing; the form of order is stochastic" (p. 189). To the extent that this

104

characterization is accurate, it can be applied to any editorial apparatus (a point to which I shall return).[25] However "postmodern" the edition may be, one may wonder why it interests McGann, for Gabler's aim is to trace the history of the composition of the work, and thus the focus is on Joyce as author, not on the collaborative forces of the production process. But McGann does look at the edition itself as the product of a particular set of social forces: Joyce's text, he says, is "enmeshed within an editorial network" that reflects two sets of "determinants"—the viewpoint of "an internationally mobile scholar," with training at Virginia and access to computer facilities at Tübingen, and "the institutional history of Joyce scholarship up to the present time." The form of the edition "replicates the conditions of its production," and its details are "no less than a coded set of interpretive clues for understanding, and using, the work" (p. 188). This observation does not, however, provide a reason for singling out Gabler's edition, since any edition can be approached in this way. McGann himself immediately says, "The same is of course true for other literary works, whose meanings are a function of their material, institutional, and social histories."[26]

One must therefore look elsewhere for the distinctiveness of the Gabler edition, for the reason to discuss it in the context of those scholarly works that "immediately establish themselves as epochal events" or in relation to the "few seminal works" of modern textual study (p. 173). To say that the Gabler edition ought to be "a required object of study for every scholar working in English literature" (p. 174)[27] necessitates noting those features that set it appart from other editions. McGann's aim, accordingly, is "an exploration of the general methodological significance of the edition in its immediate historical context." The significances he points out can be summarized under four heads: (1) Gabler's focus on the composition process; (2) his handling of copy-text; (3) his

105

deployment of apparatus; and (4) the implications of the edition for critical analysis. In each of these areas McGann's reasoning is faulty, and in none of them does he succeed in showing how Gabler's edition is distinctive. His analysis does not finally serve to clarify the details of his "sociohistorical empiric" (p. 186).

As for the first matter—the focus on Joyce's composition—McGann appears to imply that Gabler's emphasis on distinguishing evidence of composition from evidence of transmission is a distinctive feature of the edition and thus is somehow different from what intentionalist editors have traditionally done. But there is nothing new, for example, in saying that "Joyce is sometimes merely his own scribe, a textual transmitter and not a textual maker" (p. 177); this statement is simply another way of saying that editors must be critical even of texts in the author's hand, since the author, like anyone else, can make mistakes in writing. McGann quotes, evidently with approval, Gabler's operational distinction between "documents of composition," in which "'the text is held to possess full authority, unless it can be shown to be faulty,'" and "documents of transmission," in which "'the text is held to be potentially faulty, unless it can be proved to possess authority'" (pp. 176-177). Gabler recognizes here that the texts of individual documents are often unlikely to be either exclusively compositional or exclusively transmissional.[28] But the lack of parallelism in the two parts of the statement ("to possess full authority, unless . . ." versus "to be potentially faulty, unless . . .") leads one to see that in fact the texts of both categories of document are potentially faulty and must be questioned at every point. The two-part classification of documents is more a hindrance than a help, since the texts of extant documents are unlikely to correspond exactly to stages of textual development: a single extant document may contain the only surviving clues to several compositional and transmissional stages of the text. In order to associate such stages with discrete documents, Gabler is sometimes led to citing hypothetical lost documents, containing in Joyce's hand the authoritative revisions that appear in extant transmissional documents (not in his hand). But having admitted that Joyce himself can be at times only a transmitter rather than a creator, Gabler cannot by this maneuver realistically postulate lost documents that are entirely compositional—and therefore it is hard to see how this approach is more than a confused way of stating

106

what editors have always understood. They have always assumed, of necessity, that some of the readings in extant documents first appeared in documents now lost. What else could editors focusing on authorial intention have done but to extract from extant documents the evidence for understanding the compositional history of a work? McGann's discussion actually adds some confusion not present in Gabler;[29] but what is more significant is that it fails to show how Gabler has made any new contribution to textual theory or practice in his concept of compositional and transmissional documents.

Regarding copy-text, McGann says that "the problem of copytext in this edition focuses attention upon everything in the edition which is most interesting and important" (p. 178). Despite his own misunderstanding of the concept as conventionally used,[30] McGann does see that Gabler's usage is unconventional, since in the Ulysses edition "copytext" means the eclectic text that Gabler has constructed, not the documentary text (or texts) serving as the basis for that construction. This pointless shift in definition cannot be what is important, however. Presumably what is claimed to be significant is the way in which Gabler has gone about constructing his "continuous manuscript text." McGann seems to think that the existence of a plethora of prepublication documents—a common situation for twentieth-century authors—necessitates a new editorial goal: "Whereas critical editors of earlier works tried to reconstitute some (now lost) state of the text—ideally, the earliest

107

possible or 'original' text—critical editors of modern works are already in possession of the kind of 'original' text which those other editors were trying to recover" (p. 179). Critical editors of earlier works, however, do not necessarily focus on "the earliest possible or 'original' text"; critical editors of later works, even when they have the author's fair-copy manuscript, still have to determine how it relates in every detail to the author's intention, if that is what they are interested in; and critical editors of all works seek to reconstruct the texts of works as they stood at particular past moments, texts that can never be assumed to coincide with those of any surviving documents. McGann is nevertheless correct in pointing out that the text Gabler arrives at by building up a text from a succession of prepublication documents is different from "a fully and systematically corrected edition of 1922" (p. 180): whenever any variants are regarded as indifferent, the choice of copy-text (in the conventional sense) determines some features of the critical text. But McGann's distinction between these two approaches to the choice of copy-text is puzzling:

Gabler's is an imagination of Joyce's work and not its reconstitution. Gabler invents, by a process of brilliant editorial reconstruction, Joyce's Ulysses (as it were), a work that existed, if it ever existed at all, for Joyce the writer rather than Joyce the author. Gabler's edition does not give us the work which Joyce wanted to present to the public; rather, it gives us a text in which we may observe Joyce at work, alone, before he turns to meet his public. (p. 181)[31]

In the first place, critical editors—by definition—always produce an "imagination" or a "reconstruction" of a past text. Furthermore, is not the distinction between "Joyce the writer" and "Joyce the author" a way of talking about two kinds of intention? Editors have regularly distinguished between an author's prepublication or private or artistic intention

108

and an author's more inclusive intention that incorporates various ways of accommodating the expectations or demands of others in the publication process. The two kinds of intention may indeed imply different choices of copy-text, but not necessarily; only the editor's judgment in assessing the nature of the surviving materials and the author's working habits can lead to a decision about copy-text (which may of course be a decision to choose different copy-texts for different parts of a work). McGann has not demonstrated that Gabler's treatment of copy-text is "interesting and important," if by that he means that it pushes forward our thinking about the concept.

McGann makes even higher claims for Gabler's presentation of textual evidence in the form of a "synoptic text" that incorporates the compositional variants, coded with diacritical marks and symbols. Gabler himself declared, "As a form of apparatus to be read and used as a text, the synoptic presentation of Ulysses in progress from manuscript to print is the innovative feature of this edition" (p. 1901). McGann goes further and says that Gabler's synoptic text "completely overhauls the way we might think about the text as a whole" (p. 181, italics his): by giving "priority of importance" to the synoptic text as a text for "seriatim reading" (p. 175), Gabler shows what is "entailed in the idea of textual instability" and allows a "number of different Ulysses . . . to occupy the space of critical possibility" (p. 181). There are no grounds for regarding Gabler's edition as pioneering in this respect, since an "inclusive" apparatus (incorporated in a running text) has been used by many editors to show manuscript revisions; the 1962 edition of Billy Budd, Sailor, edited by Harrison Hayford and Merton M. Sealts, Jr., is a famous earlier example of a genetic text accompanied by a reading text.[32] Whether Gabler's approach to apparatus is innovative is less significant in any case than the general question of what this kind of apparatus accomplishes. McGann's view at the time of this essay was apparently that the placement of superseded variants in a running text gives them greater prominence and offers the reader a better sense of the instability of texts. But everything that he says about Gabler's synoptic text could be applied as well to the kind of apparatus in which variants are listed at the foot of a page or the end of a text. Gabler's diacritics he calls "a grammar of an artificial language" (p. 181), and he believes that, when we have mastered it, "we shall have gone a long way toward understanding the nature of texts in general" (p. 182). The "surrounding diacritics"

109

emphasize the "fragility" of the text, in which "one reading is marginalized by another"; we are "made conscious that this text is to be fundamentally characterized as a thing of many real and concrete details which are, at the same time, extremely fragile, and put together in strange, stochastic orderings" (p. 191). The alternative system of reporting variants in footnotes or lists is also a "language," and when its conventions are mastered it tells us the same things. Variants placed in a running text are not necessarily more prominent, or easier to read in context, than those placed elsewhere: it is all a matter of becoming familiar with a set of conventions. That McGann may have come to this opinion is suggested in his review of McKenzie when he includes critical editions among the productions that encourage "radial" reading, the reading of several versions of a work simultaneously. It is clear that he is not thinking only of editions with genetic texts because he describes how "one moves around the edition, jumping from the reading text to the apparatus, perhaps from one of these to the notes or to an appendix, perhaps then back to some part of the front matter which may be relevant, and so forth" (p. 21).[33] This positive description of how a critical edition "allows one to imagine many possible states of the text" makes admirably clear the way in which an appended apparatus can be just as stimulating and productive an aid to reading as an inclusive apparatus is. Gabler's Ulysses, then, is not particularly significant for its handling of apparatus.

Finally, McGann stresses one particular critical implication that he finds in the Gabler edition. He argues that those critics who see textual meaning as the product of readers are not inclined to be concerned with the kind of textual indeterminacy that arises from the existence of variant documents. They see texts as "angelic rather than human" (p. 185), as having an ideal existence apart from physical embodiments. If we view them as human products, however,

the interpretive act "constitutes meaning" (as we now say) only in terms that are licensed by the received sociohistory of the text. And that sociohistory, for texts, is constituted at its most elementary level as a set of empirical documents whose meaning is intimately bound up with the sociohistory of the documents. (p. 185)

110

McGann is making a valuable point if he is saying that reader-response critics overlook an important factor in readers' production of meaning when they do not take into account the physical features of documents. But how he wishes to link this observation, or some such observation, to the Gabler edition is not clear. "Gabler's edition," he says, "helps one to see the formal limits which always constrain the generation of texts" (p. 182). These "formal limits," this "sociohistory of the documents," cannot have much to do with the formal arrangement of text on pages or with the physical appearance of the documents, for the amount of information on such matters to be derived from the Gabler edition is not extensive (nor is it from any edition without facsimiles). We are left, then, with the text itself, encompassing alterations within individual documents and variants between one document and another; and if McGann is merely saying that the Gabler edition reports such alterations and variants, he is not distinguishing it from hundreds of other editions. "What Gabler's edition shows," he tries to explain, "is that unstable 'texts'—texts that are 'in process' or 'indeterminate'—always appear in material forms that are as determinate as the most 'stable' text one might want to imagine" (p. 186). But it is not true that texts "always appear in material forms": some texts exist only in the mind, and others have a public existence only in oral recitation. Furthermore, documents that no longer exist did have determinate texts when they existed, but their texts can now only be conjectured—and are therefore indeterminate, since no reconstruction can be certain. It is thus hard to see how McGann is stating anything more than the obvious fact that the texts of extant documents are determinate. His aim of incorporating the study of those documents within the process of literary criticism is laudable; but curiously his way of pursuing the point seems to reinforce the old separation of textual from literary criticism by suggesting that textual criticism and analytical bibliography set limits to be followed subsequently by literary criticism. Bibliographical analysis of a given document can indeed set limits on conjectures about the production of that document and hence sometimes about how the text of the document came to be what it is; but such analysis cannot operate on nonexistent documents, and any facts it seems to have established regarding extant documents cannot limit the emendations that may be proposed for reconstructing a text that does not at present exist. This limit is set by the informed judgment of the individual doing the reconstructing, and interpretation is thus tied to textual criticism in a more basic way than McGann's essay manages to make clear.[34] Gabler's edition is in any case

111

no more appropriate for illustrating this line of argument than any other edition that records variant readings; indeed, it may be particularly inappropriate, for its citation of inferred documents as sources of readings de-emphasizes the distinction between existent and conjectured texts as underpinnings for critical editing.

McGann's discussions of Gabler and McKenzie offered two natural opportunities for him to clarify what was left confused in the Critique. But neither essay makes matters clearer, except possibly at one point: in the Gabler piece, he says that "'genetic' texts . . . may be conceived either as mirrors of composition or mirrors of production" (p. 182). If this statement indicates a recognition that authorial intention and the results of the collaborative process of production are two independent (and in most instances mutually exclusive) goals of critical editing, it would mark a significant advance over the thinking reflected in the Critique. The statement in fact still displays some confusion by making "'genetic' texts" rather than so-called reading texts its subject: the choices entailed in producing a clear (i.e., "reading") text cannot normally accommodate both goals simultaneously, but a "genetic" apparatus can be constructed to do so. Possibly McGann would have made this point if he had written the Gabler essay after the McKenzie review, where his account of apparatus seems more in line with it. But no further clarifications, so far as I am aware, have appeared in his other recent pieces on textual criticism,[35] although in a 1986 interview he explicitly amplified the Critique by accepting authorial intention as one of the editorial goals that can emerge from a study of the full compositional and production history of a work.[36] This development in his thinking

112

is welcome because it will make more coherent his argument for an inclusive conception of textual criticism. Up to now, his effectiveness has been seriously undercut by a lack of rigorous thought, at least as reflected in his often careless prose. He has nevertheless played an important role in recent years by causing more textual critics than ever before to pay attention to other goals for critical editing than the construction of authorially intended texts and thus to define their activities in relation to a broad range of critical endeavor.

III

McGann's interest in Gabler's approach to editing illustrates how the genetic aspects of textual study can be seen to reinforce, and be congruent with, a sociological view of literary history. How authors inscribe and revise their private manuscripts can reveal their attitudes to the public world of literary publishing. The pre-publication growth of a work through various textual stages reflects the same openness and instability that are seen in the post-publication vicissitudes of texts; and every text, whether in an author's manuscript or a published edition, is obviously the product of a set of sociohistorical forces. Gabler sees his own work as growing out of a German and French movement that emphasizes the study of authors' manuscripts and the process of composition. Several European textual critics of this general orientation are becoming better known in the English-speaking world through the recent publication of some of their papers in English translation (although an essay by Hans Zeller, editor [1958- ] of C. F. Meyer, has been available in English since 1975).[37] Through Gabler's efforts, this movement was well represented at a Charlottesville conference in April 1985 honoring Fredson Bowers on his eightieth birthday, and a few days later some of the same speakers participated in the biennial conference of the Society for Textual Scholarship in New York. As a result, eight papers by six French and German scholars appeared in the 1987 volume of Text and the 1988 volume of Studies in Bibliography (those from the

113

STS conference in the former, the Charlottesville conference in the latter).[38]

Gabler's own contribution[39] makes no advance over his 1981 paper that set forth the rationale for his synoptic text of Ulysses,[40] but it does contain some sentences that concisely capture the underlying tenets of this approach. For example: "A work revised in successive stages signals the author's free intentional choices at any given textual stage, and the aggregate of stages may justifiably be considered to embody his final intentions with regard to the work as a whole" (p. 110). Thus "authorial rejection cannot be equated with editorial rejection" (p. 110): "The text in the determinate record of its instability falls to the editor therefore not for the fulfillment of its real or assumed teleology, but for the description and analysis of its documentary existence" (p. 111). Any apparatus of variants, of course, provides an account of documentary evidence, and Gabler's essential complaint about what he calls "the conventional model of the critical edition" is that its apparatus subordinates the genetic record to a "stable reading text of unquestioned privilege" (p. 107). Although some editors have no doubt been so foolish as to think that their critical texts commanded "unquestioned privilege," most editors—having laboriously worked through all the documentary evidence—recognize that any clear reading text is only one of the texts that can be derived from the evidence. And surely, after going to great pains to record variant readings, their intention is not to "annihilate" them—though Gabler believes that "What is near-to-annihilated . . . in the established critical edition is the superseded authorial variant, relegated as it is to apparatus lists in footnotes or at the back of the book, together with the bulk of rejected transmissional errors" (pp. 109-110). As in his earlier piece, Gabler reduces an interesting conceptual question—the nature of texts as reflections of works in process—to the practical level of methodology for reporting evidence. Some forms of apparatus may be better than others in certain situations, but the decision to record variants in appended lists does not imply that genetic study is unimportant. A more basic flaw, however, is Gabler's failure to distinguish clearly between texts of extant documents and stages in the development of a work. He does recognize that authorial revisions "leave a record when, though only in so far as, committed to paper"—and, he should have added, not all such pieces of paper have survived. But he

114

still can say that textual instability is, for the editor, "determinate," being "confined within the complex, yet closed system of the words and signs on paper" (p. 111). This statement can mean nothing more than the obvious point that editors with a historical orientation are not free (as authors are) to put whatever words they like into a text. It does not illuminate the imaginative role of the historical scholar in reconstructing from fragmentary physical evidence the growth of a work in an intangible medium.

Somewhat more useful as an introduction to critique génétique (as this approach has come to be called in France) are two essays by Louis Hay, founder of l'Institut des Textes et Manuscrits Modernes in Paris.[41] The one in Text sketches the history of modern editing in Germany and France, noting that "editing has always embodied the main ideological and cultural concerns of its day" (p. 117), and concentrates on the last half-century, when the attention has been on "a new kind of scholarly object: the text as it grows" (p. 118). Hay cites some illustrative editions, beginning with Friedrich Beissner's (1943-62) of Hölderlin, and describes a growing interest in the technology of examining manuscripts; but he leaves for his essay in Studies in Bibliography any real speculation about the nature of texts as implied by this work. That SB essay conveys some sense of the excitement experienced by this group of scholars as they began to explore the "avant-texte," the "pre-text"—that is, the versions preceding the published one.[42] But it is finally unsatisfactory in its attempt to explain the theoretical foundation of genetic criticism. Answering the question of his title, "Does 'Text' Exist?", obviously involves distinguishing various concepts of "text," and the question is of course meant to suggest that "text" as a fixed entity does not exist; but no clear concept of "text" emerges. Despite Hay's recognition (p. 69) that "The ink on the page is not the writing itself" (meaning, presumably, that texts of documents are not texts of works or versions of works), his discussion does not seem firmly grounded in this important distinction. "Pre-text," on which genetic critics focus, leads up to "text," which "is considered achieved when published," for "the author's intention . . . becomes manifest in the act of publication" (p. 71); even rough

115

drafts can be promoted from "pre-text" to "text" by being published (p. 72). This distinction seems superficial, for the act of publication does not change the ontological status of physical texts; the texts of all documents, unpublished or published, may equally fail to reflect accurately the intangible works of language that they purport to represent. In analyzing two versions of Paul Éluard's poem that became "Libertḗ," he says that the "first, distinct work was one of the possibilities of the text, though it was neither integrated nor subsumed in the second work. . . . Perhaps we should consider the text as a necessary possibility, as one manifestation of a process which is always virtually present in the background" (p. 75, italics his). The shifting meanings of "text" here (and its unclear relation to "work") are symptomatic of the lack of sharp focus throughout.

A similar slipperiness in the handling of concepts mars the two essays by Klaus Hurlebusch, coeditor of the Hamburg Klopstock edition (1974- ).[43] The far more substantial piece, in Studies in Bibliography, attempts to contrast two "editorial concepts" that reflect two opposing views of authorship, growing out of "the fundamental conflict between an individual and a social identity" (p. 102). One concept, "reception-oriented," regards the author as a "community being" (p. 125), primarily concerned with a finished, public product to be received by readers; it results in editions that emphasize an established text over an apparatus of variants. The other, "production-oriented," sees the author as principally engaged in personal expression, developing ideas "as independently as possible of considerations for others, of readers' expectations" (p. 124); it results in editions that emphasize genetic process. In the former, "what the author ultimately intends is an imagined and affirmable idea of himself as a person, in relation to which his creative ability of expression is secondary"; in the latter, "the identity of the author whose perceptions continually change finds expression not so much in the work as in the process of perception, i.e., the author's working procedure" (pp. 122-123).[44] Hurlebusch is struggling here with a dichotomy

116

that all editors must give thought to, but his formulation of it is not helpful. For one thing, it assumes a sharp and recognizable distinction between "the productive function of writing" and "the revisional one" (p. 123), as if the operation of a "writer activating his powers" (p. 122) is separate from that of a writer engaging in the "redaction and revision of his own texts" (p. 125)—and as if revision is to be equated with a move towards conventional expression. Furthermore, it assumes that, after authors are classified according to this distinction (Hölderlin, for example, being "a primarily text-producing author" [p. 123]), the editorial emphasis to be accorded them follow naturally.[45] There is no recognition that every author may legitimately and usefully be provided both with critical editions (containing critically constructed texts, with records, however presented, of variant readings) and with reproductions of documents. Hurlebusch makes a strong case for the importance of some authors' manuscripts as visual objects, every detail of which is reflective of the creative process; but he does not indicate that all manuscripts can be approached in this way, even those by "reception-oriented" authors.[46] In addition, his comments on intention, like Hay's, fail to make careful discriminations. Authorial intention sometimes seems to mean intention to publish: "If the author has actually . . . released the contents and textual composition from his control and submitted it to . . . a publisher's, . . . it is true to say that this version is intended by the author" (p. 109). But then we are told that "every

117

state of a text . . . only represents the intended state as a whole, and not in every detail" (p. 110). It is hard to see how such a distinction can be productive, since textual intention must always be concerned with textual details and since any "state of a text" is given its character, "as a whole," by those same details. His other essay, an insignificant piece in Text on editing diaries, furthers the reader's puzzlement by observations that seem out of line with this essay.[47] One has to conclude that Hurlebusch, though clearly of a conceptualizing turn of mind, has been presented to English-speaking readers in essays that do not demonstrate clarity of thought and do not advance editorial theory regarding the nature of authorship.

Of the remaining three papers in this group of translations from European textual critics, the one by Gerhard Neumann, one of the editors of the Kafka edition (1982- ), should be touched on here, for its concluding paragraphs serve to epitomize some of the conceptual problems common to the whole group.[48] After explaining how Kafka was a writer torn between the desire for intimate self-expression and the desire for public recognition, he sets forth "three fundamental concepts which all bear on what we have to understand by 'text' in the modern situation" (p. 98): "script," "work," and "a 'fluid composite' which is never wholly the one nor the other." "Script" (or "writing") reflects "the flow of creativity"; "work" reflects "the area of 'authorship' as an institution and the communicative function of literature as a social phenomenon"; and the "fluid composite" reflects the "intertextuality" that results from placing particular "textual units" in different arrangements and collections. It seems unwise to use the word "work" to refer to published texts (imputing to what an author publishes an unwarranted degree of textual

118

authority); but Neumann's usage stems from the same tendency observed in the other writers to exaggerate the finality conferred by authorized publication and to compartmentalize creativity in the prepublication stages—a tendency to stress the author's search for identity at the expense of sufficient consideration of the practical realities of the publishing world. Neumann's need to create a "fluid composite" category, however, shows that he sees, at least in some respects, how the private and public categories overlap. For Kafka, he says, "one has little choice but to bring out all three aspects in an edition." Once again there is the concern with how an edition can, through its formal arrangement, emphasize one or another characteristic of authorship and a seeming lack of understanding that—far from having "little choice"—an editor of any author might positively desire to "bring out" these conflicting aspects of authorship, recognizing their relevance in all cases. He is right, of course, to conclude that editing involves interpretation (being, he unfortunately adds, "no longer a mechanical task of unquestioning reproduction"); but when he says that interpretation is necessary "to do justice to modern concepts of author, work and text" (p. 99), one must wonder how this triad relates to the earlier one, where "work" was one of three categories of "text." This whole group of papers regrettably lacks rigor and thus makes no significant theoretical contribution to editorial thinking. But the attention now being given to prepublication material by European literary and textual critics is welcome in any case, for the importance of such material is unquestionable—as editors in English-speaking countries have also long recognized (witness the original manual of the Center for Editions of American Authors in 1967).[49]

An interest in textual genesis is inseparable from an interest in "versions" of works, and a number of English-speaking textual scholars have in recent years focused on "versions"—most conspicuously a group of Shakespeareans including Steven Urkowitz, Gary Taylor, Michael Warren,

119

and E. A. J. Honigmann.[50] One result of this movement has been the inclusion of two versions of King Lear in the new one-volume Oxford Shakespeare (1986) on the grounds that conflating the first quarto and the Folio texts (as customary in the past) mixes two discrete stages of Shakespeare's work on the play.[51] Such arguments for maintaining the identity of versions—whether by scholars writing in French, German, or

120

English—often founder on the question of what role critical editing can play in the production of editions of versions. No defense is needed for reproducing or transcribing the texts of documents (handwritten and printed) exactly as they stand: such texts are historical facts, and it is useful to have them made available. But some editors wish to purge those texts of readings—such as typographical errors—that were not intended. In departing from the texts of documents, they must face the question of defining what kinds of emendations are to be allowed; and no matter what categories are named (even if only typographical errors, which are of course not always obvious), there is the possibility that some instances will be detected only by noting the variants in other documents and deciding that some of them were the intended readings all along. One is not mixing versions simply by drawing readings from different documents, since the texts of documents cannot be equated with the texts of versions—a fact recognized in the original decision to present a critical text. After all, traditional critical editors interested in authors' final intentions are not trying to mix versions but to recreate one—one that is not present in satisfactory form in any surviving document. But recent discussions of the importance of versions have too often been guilty of a double fallacy: believing that eclecticism (drawing readings from different documents) necessarily involves the mixing of versions, and believing that critical editing can ever be other than eclectic.[52]

Donald H. Reiman published in 1987 a spirited defense of the production of scholarly editions containing versions,[53] but it does not entirely avoid these confusions. Reiman advocates what he calls "versioning" rather than "editing" (p. 169): "there are good reasons to redirect our energies away from the attempt to produce 'definitive' or 'ideal' critical editions and, instead, to encourage the production of editions of discrete versions of works" (p. 179).[54] This summary statement suggests

121

that what he favors are unaltered documentary texts (facsimiles and transcriptions), as indeed he does for some purposes. But he also includes within the concept of "versioning" the presentation of texts in which the editor has corrected "any typographical errors" and perhaps some other errors: pointing and orthography may be emended, but not "in any wholesale way, lest what results is some version other than the one being advertised" (p. 178, italics his). He is thus not opposed to the critical editing of versions; but he does not address the question of how the process of editorial judgment entailed in recreating pre-final versions is different from the one required for recreating a finally intended version. He is in fact sanctioning for pre-final versions what he calls "'definitive' or 'ideal' critical editions" when devoted to finally intended versions. Apparently it is not the judgment involved in critical editing that bothers him, but rather what he sees as its excessive concentration on final intention. Such a conclusion is surprising, however, in view of his sarcastic remarks about editorial judgment—as when he says that an editor who engages in "versioning" may not require "the preternatural power to divine the unstated moods or preferences of dead authors" (p. 178). He fills most of a page detailing how editing would be different in a world where "versioning" were standard. Among the points enumerated: "Editing might not always then require the detective skills to root out all the surviving evidence about the author's involvement . . . in a work throughout its entire textual history"; "there would be less need for the editor to hypothesize events and attitudes where the crucial evidence concerning the author's involvement in the text is lacking"; there would be no need to decide which variants "resulted from the author's grateful acceptance of the publisher's or compositor's suggestions and corrections and which ones resulted from reluctant acquiescence in, or unawareness of, each particular change" (p. 177). What is particularly troubling about this litany is the implication that non-"versioning" editors have engaged in a great deal of unnecessary work and idle speculation. On the contrary, the topics listed are relevant to any study of a work's textual history; and informed speculation is an essential ingredient in all historical reconstruction. Reiman seems to feel, paradoxically, that eclectic texts with full apparatus inhibit debate, whereas separate texts of individual versions encourage it.[55] It was not

122

necessary in any case for him to attempt to denigrate eclectic texts as he promoted single-version texts: the value of the latter has never been in question, as his interesting survey of examples indicates.[56] But single-version editing has no doubt received less discussion, and Reiman's vigorous championing of it may serve to bring it further attention.[57]

A consideration of authorial versions in the context of the social or collaborative status of authorship eventually leads one to think of the theater, for the preproduction version of a play is likely to be different from the one that emerges from the production process, and the collaborative character of theatrical production raises in extreme form the question of how authorial intention in a work of language is to be conceived. Recently T. H. Howard-Hill has turned his attention to the implications for play-editing of the nature of drama as a genre, and the resulting essay, as one would expect, is impressive in its scholarship, its sharpness of insight, and its style of argument (although one may question some of its assumptions). "Modern Textual Theories and the Editing of Plays"[58] provides, first, a searching criticism of Greg's rationale of copy-text, and, second, a forceful statement of the view that playwrights' intentions can be fulfilled only in performance. The two matters are separable (and the second is the one directly relevant here), but the connection is clear: Greg's rationale leads toward the choice for copy-text of the text closest to the playwright's manuscript; but the text of a (later) document associated with the theater, which is likely to contain the playwright's revisions made during the course of production, may seem to reflect the playwright's intentions better than a preproduction text. Thus, for Howard-Hill, Greg's rationale "applies least successfully

123

to the genre that brought about its conception" (p. 89).[59] Since under Greg's procedure the alterations judged authorial from a later text could in any case be emended into an early text, the heart of Howard-Hill's criticism of Greg is the treatment of so-called indifferent variants (although the value of preserving an early copy-text's accidentals is also questioned). Howard-Hill's position is that, if some variants in a later text seem clearly authorial, then some (or all) of the indifferent variants in that text might be authorial, too, and they should not automatically have to yield to the readings of the (early) copy-text. Discussing this question does not seem very productive, however, since everything depends on how an editor in a particular situation decides which variants are to be called "indifferent."[60] What is of greater interest is the concept of dramatic authorship implicit in Greg's rationale. The choice of a copy-text close to the manuscript and the goal of reconstructing an inferential authorial fair copy show that Greg was interested in the form

124

of plays as conceived by their authors working alone and that he wished to exclude from critical texts the alterations introduced in the process of staging. The appropriateness of this approach is indeed an important subject for debate.

Howard-Hill believes that editors who follow Greg "reject the playwright for the author, the dramatic script for the literary manuscript" (p. 103). Playwrights understand, he says, that their manuscripts are "the raw material in a process of shaping a play for performance that requires the collaboration of many contributors: theatre personnel, actors, directors, and the author himself" (p. 105). Thus the author writes a manuscript "expecting (to various degrees) that the script will be modified in the theatre" (p. 105). It follows that "a playwright's intentions are represented best (if perhaps not completely) in a manuscript associated with the theatre" (p. 112), and any editor whose goal is to reconstruct a preproduction manuscript "averts his face from the theatre for which the dramatist wrote, and presses boldly backwards into the primitive jungle of the author's drafts" (p. 108). That "jungle" has its own interest, of course, as we have been reminded by the critics just discussed, among others; and Howard-Hill would presumably agree that there is always a value in editions that focus on preliminary versions. But his particular concern is with works as finally intended by their authors, and from his vantage point a rationale for play-editing that purports to stress final intention and emphasizes preproduction documents is "textually-regressive" (p. 90), or "editorially regressive" (p. 101). Part of this argument would be accepted by anyone: that a play does not exist in its intended medium except in performance, and performance is necessarily collaborative. No one can object to editions that attempt to record the text followed in a play's initial (or some other) production, for the versions that reached the public in performance are obviously of historical interest. What is objectionable in Howard-Hill's presentation is his insistence that the only legitimate critical texts for representing playwrights' final intentions are those based on performance texts (or such textual evidence as there is of what actually occurred in performance). His account is notably unbalanced in not sufficiently recognizing that alterations made for performance (even if agreed to by the playwright) do not always please the playwright.[61]

125

It is indicative of a fundamental misconception underlying Howard-Hill's argument that he sees a meaningful opposition in "the dramatic script" versus "the literary manuscript," "the playwright" versus "the author." Presumably he means that the "author" of a "literary" piece like a novel or poem works alone, whereas the "playwright" of a "dramatic" piece must engage in the collaborative activity of production, and therefore an editor who uses a preproduction manuscript of a play as copy-text is behaving as if the work being edited were of a "literary" rather a "dramatic" genre. "Literary" authors, however, must participate in the collaborative (and altering) process of publication if their work is to reach the public, and thus the real opposition is not between "drama" and "literature" but between versions of works (of whatever genre) as intended by their authors alone and versions of works (of whatever genre) as they emerged from the collaborative efforts required to bring them to the public. The analogy between production and publication (of which Howard-Hill approves [p. 104]) is worth exploring further. There is an obvious way in which the analogy is inexact, for what a performance of a play offers is an actual work of drama, whereas what a book provides is not a work of literature (the medium of which is not paper and ink) but a set of instructions, a script, whereby one can recreate a work of literature through the act of reading. The publication of a play in print is directly comparable to the publication of a novel or poem: in both instances the printed text is the means by which works in intangible media can be transmitted. In the one instance, the work involves actors and props on a stage (and is thus a mixed-media work, using sound, movement, and visual effects as well as language); in the other, the work consists of language alone, and the recreation of the work can take place within the mind or in private recitation.[62] A more exact comparison

126

is with music: the first performance of a symphony constitutes its publication as a work of music (that is, publication in the medium in which it was meant to be experienced); the publication of its score on paper simply makes more widely available the instructions for recreating the work.[63]

Thus critical editors of drama are in a position no different from the editors of other works in intangible media (including literature): they all must rely on surviving documents (which are not the works themselves) and strive to reconstruct from them the texts that were intended by particular persons (whether authors alone, or authors in collaboration with others) at particular points in the past.[64] One of the basic

127

choices any editor has is whether to concentrate solely on an author's intention or to focus on the combined intention of the group (including the author) overseeing public dissemination—for the two interests can almost never be satisfied by a single clear text. An editor who chooses as copy-text the fair-copy manuscript of a novel rather than the first edition set from it is doing the same thing as the editor of a play who selects the final preproduction manuscript as copy-text. In both cases emendations can be made from later texts if there are variants that seem to reflect the author's uninfluenced intention—which is, after all, the aim of these editors. The choice between authorial and collaborative intention and the crucial distinction between intention and expectation have been much discussed,[65] and it is in the context of those issues that the editing of plays should be viewed, for plays do not present a unique situation. Howard-Hill apparently wishes to be understood as believing that there is no historical validity to the concern for authors' (at least early play-wrights') uninfluenced intentions: "To postulate amongst professional playwrights of the early period the romantic author pouring his inspiration into his early drafts which alone conveyed his intentions, having been afterwards corrupted by grubby theatre professionals, is the worst kind of historical error" (p. 106).[66] Even if it had been couched in less slanted language, this statement would have been unacceptable. Writers of all periods, as human beings with feelings and opinions, have had their own personal preferences regarding the form and content of their work, even if their status as professionals sometimes required them to accept or initiate alterations that seemed to them less desirable. And as long as anyone is interested in understanding the workings and accomplishments of minds of the past, the task of attempting to determine authors' own intentions and to reconstruct texts reflecting them is a valid activity of historical recovery. Howard-Hill's contrary position is weakened

128

by his presenting it as the only defensible approach rather than as one alternative, as one of several paths to the past, each of which gives us a different perspective.

IV

The pervasive presence of textual and literary sociology (with its related challenges to authorial intention) in the current climate is indicated by the attention it receives in general assessments of the present state of textual studies. For example, Fredson Bowers, in his presidential address to the Society for Textual Scholarship at its biennial conference in 1985,[67] devoted half his time to a critical examination of McGann's approach, voicing his suspicions of "any overall theory that denigrates an author's intentions by sharing them with social milieu as a central fact" (p. 10) and concluding that "Critics who mistrust our developed conventional editorial theory do not seem to be fully aware of its flexibility when properly applied, its consciousness of the shift in the nature of problems and of the methods for dealing with them as the centuries pass" (p. 11). Two years earlier, Paul Oskar Kristeller's presidential address[68]—not published until 1987—had considered "traditional" textual scholarship to be on the defensive and argued that, although "economic, social, and political developments are necessary conditions for many or even all cultural or intellectual developments of the past," they are not "sufficient causes that would adequately explain the concrete texts and documents with which we are concerned, or their specific form and content" (p. 6). Both scholars affirm traditional approaches, not because they are unwilling to consider alternatives but because the recent challenges seem (as Bowers says) to be based on a misunderstanding of the traditional approaches and attempt (as Kristeller says) not only "to supplement traditional scholarship (which would be quite acceptable) but to discredit it and to replace it" (p. 2). Nevertheless, they both see the challenges as a central fact of the present moment.

The founder of the Society for Textual Scholarship, D. C. Greetham, has constructed a characteristically wide-ranging description of the situation at present by setting out a series of "ideological pairings" between

129

textual and literary theories, showing that the various literary approaches emphasizing authors or texts or readers have their counterparts in textual theory.[69] Social textual criticism of course receives its due in this scheme as one of the reader-based theories, and Greetham says that it has become "a major focus for debate" (p. 23) as a result of McGann's Critique, which provides its "clearest statement in recent textual theory" (p. 11). In Greetham's earlier account of the state of editing medieval materials (which is in fact a lively treatment of issues involved in any editing),[70] McGann's approach and the related emphasis on versions are linked to "current critical positions—the death of the author, the primacy of the fragment, the deconstruction and aporia of the work as a consistent framing of one person's intention" (pp. 61-62). Greetham, in essays such as these, has been actively promoting greater understanding of the connections between textual and literary theory, and he was given a useful opportunity for furthering the cause when he was asked to serve as commentator on the papers delivered at a "Symposium on

130

Textual Scholarship and Literary Theory," organized by the Society for Critical Exchange and held at Miami University (Ohio) in March 1987.[71] His even-handed remarks are firmly critical of literary theorists who see textual criticism as a static field intellectually unrelated to literary theory (a view represented at the conference); at the same time he takes for granted that the period when authorial intention had its "greatest practical influence" in textual criticism has passed.

Another textual conference, "New Directions in Textual Studies," was held in March—April 1989 at the Harry Ransom Humanities Research Center in Austin. On this occasion Ian Willison attempted a general survey of the connections between editorial theory and the history of the book, stating at the outset that as a historian of the book and authorship he is "obliged" to favor the social approach to textual matters represented by McGann and McKenzie.[72] It is surprising that a historian of books and authorship would see a necessity to choose between the social and the authorial approaches to editing, when each concentrates on a different aspect of the total picture and the two taken together might be thought to produce a more rounded view.[73] An earlier conference, one on "textual hermeneutics" held at Canberra in May 1982, produced some controversy that appeared in print over the next several years.[74] At the conference Stephen Knight attacked "old-fashioned positivist, text-and-author centered editing" (p. 44) and indicated that his own aim in editing Chaucer is to produce a text "with the fullest socioliterary

131

potency" and that he chooses between "equally possible variants" the one with "the maximum possible historical tension, the reading which loads the text most strongly with ideology" (p. 49). Harold Love later responded, with some sarcasm, on behalf of intentionalist editing;[75] and Stephanie Trigg then noted some of the deficiencies of both Knight's and Love's papers, accepting Knight's point that (in her words) "any edited text is an ideologically loaded construct" (p 20) but preferring to stress the reading audience as the locus for the production of meaning.[76]

Numerous other essays deal with these matters in one way or another. Peter Shillingsburg, for example, provides an exposition of the obvious considerations involved in the "social contract" approach to editing, sensibly affirming editorial pluralism;[77] James L. W. West III focuses on the "act of submission" as a key moment in the textual history of a work, a moment that serves to illuminate the contrasting attitudes of editors following authorial and social theories;[78] Leonard N. Neufeldt describes

132

editorial work as the product of a specific time and place, influenced by particular critical theories and all the other "forces of institutionalization in the field of literature";[79] Hugh Amory criticizes an introductory bibliographical and textual manual for its emphasis on the "obsolescent" Greg-Bowers tradition and its neglect of McKenzie and McGann;[80] and many studies of particular authors or fields explicitly confront (however effectively) the competing claims of the authorial and the social.[81]

133

I have cited all these essays to suggest how insistently the issue of social textual criticism has arisen in recent discussions and how frequently it has been linked with developments in literary theory. The connections between textual and literary theory have not as often been seen, however, from the other direction: writers who think of themselves primarily as literary theorists or critics have not been very cogizant of the related debates taking place among textual critics. McGann is of course an exception to this generalization, and so is David Gorman, whose essay "The Worldly Text" is a remarkable survey of current directions in literary and cultural theory, directions that are "worldly" because they go beyond the purely linguistic.[82] Gorman includes "textual studies" as one of three areas for detailed discussion, the others being "the theory of social action" and "the philosophical questions raised by historicism," each chosen because he believes it "suggests important new avenues of research in cultural history" (p. 183). The essay, which is well-informed and judicious, may possibly be the first survey from a literary theorist, addressed to an audience interested in literary theory, that takes adequate account of developments in textual criticism. He ends his section on textual critics by saying that "The level on which their theoretical debate is taking place is very high indeed, and one that should put many theorists of interpretive criticism to shame" (p. 198).

Nearly all of the essays I have mentioned, here and earlier in this survey, have touched—in more or less detail—on the matter of textual apparatus, on how textual evidence is to be reported in an edition; and some other essays, dealing exclusively with apparatus, have recently appeared. Although apparatus may seem a less intellectually interesting subject than theories about the nature of verbal texts, one can readily see why the form of apparatus has become a central concern at a time of challenge to traditional intentionalist editing, for the standard presentation

134

of a clear text with appended apparatus (on the same page or elsewhere) has been interpreted as reinforcing the concept of a single, closed, definitive text. There are two levels on which discussions of apparatus can take place, and they should be carefully distinguished, for they are often intertwined in the same arguments. One level, the lower one, has to do with readers' convenience, or the "usability" of an apparatus. Objecting to an apparatus solely because it seems cumbersome, or because it can be expected to discourage readers from using it, expresses a concern at the practical level of clerical procedure, not at the conceptual level of theory. All of us would like the books that we read and study to be convenient to use, but we put up with inconvenient ones all the time, because we have no choice but to use them as they are. It is unquestionably a flaw in an edition to have an unnecessarily awkward apparatus, but not nearly so severe a flaw as to have left out an essential category of information; at least the information is on record, even if it is less easy to retrieve than one might wish. But objections to an apparatus can also be raised on a more serious level, when arguments are made that a particular presentation of material is substantively misleading, incompatible with the historical situation being depicted or with the scholarly goals of the editor. Although such arguments do necessarily involve a questioning of formal conventions, their real concern is with the communication of meaning through the modes of expression classified as apparatus.

Two recent articles illustrate these levels. Don L. Cook[83] surprisingly challenges the assumption that "the more information we can give the user of a critical edition about the genesis and evolution of a written work, the more useful our volume will be" (p. 82). Certainly he is moving against the current trend represented by the European proponents of genetic criticism. Few would disagree with his point that the publication of revisions in prepublication documents "should result only from the thoughtful consideration of the service they can provide to the serious scholars who will be using the volumes, not from considerations of respectability or continued funding" (p. 89)—or from the mechanical following of some rule. But it is hard to see how such information would ever be useless to scholars, and Cook recognizes that editors themselves must always be aware of pre-copy-text alterations. The reason not to publish the information, then, is on the practical level: such lists are likely to be long, and therefore expensive (pp. 83, 87); they are also

135

likely to be complex, and users may thus "become exhausted and finally disillusioned" in attempting to use them (p. 84).[84] To condone the withholding of avowedly important information on these grounds seems a counsel of despair.[85] Another article, by Ted-Larry Pebworth and Ernest W. Sullivan, II,[86] recommends that versions of works representing "markedly different semiotic entities" or independent textual traditions (p. 44) be presented as separate texts, each with its own apparatus. Although the authors do complain that some information may not be "easily discoverable or recoverable from the traditional lengthy and complex apparatus format" (p. 44) or that "some bibliographical training and considerable industry" is required to retrieve it (p. 47), this kind of problem is not their primary point—which is rather that a consolidated apparatus may in some cases misrepresent the textual history of a work by merging the histories of independent traditions and suggesting that the version printed as the text is more authoritative than the other independent versions (p. 46). They may have exaggerated the degree to which "traditional" editions are guilty of this practice; in any case one can say that there is another established tradition as well, in which separate editions are prepared for versions that are so different as to be judged distinct works.[87] Pebworth and Sullivan's proposal is a

136

refinement on this tradition, most valuable in its suggestions for handling versions of which only certain lines or passages require printing in full. One should not, however, draw from their article the conclusion that independent versions are printed separately for the purpose of clarifying the apparatus; they are printed separately because the editor judges them to be independent versions (or works) demanding such treatment, and separate apparatuses follow as a matter of course. The essential problem with a consolidated apparatus for independent versions is not that it is inconvenient but that it is a reflection of an inappropriate treatment of the texts.

These essays do not question the use of apparatus as an accompaniment to clear texts; but theorists who stress the indeterminacy of texts often object to the "privileging" of any one selection of variant readings, regarding a clear text with an appended apparatus as an inappropriate elevation of certain readings and subordination of others. Many such editors in recent years have argued strenuously for inclusive texts—texts, that is, in which variant readings, or some categories of them, are inserted directly into the linear text, accompanied by any necessary sigla or diacritics. The essence of Gabler's rationale for his presentation of the text of Ulysses is that he wishes to show the "diachrony" of the revisional stages of the work by placing each authorial variant in its "contextual relations" within the text; apparatus is the central concern, for the traditional apparatus, he feels, results in "fragmentation" of the text as a whole, the text conceived as the totality of all the author's revisions.[88] A common theme among scholars of the genetic school, however different their actual handling of variants, is that the traditional apparatus falsifies the historical situation by minimizing the significance of certain variants, subordinating them to an editorial construct, instead of giving them their due as equal partners in an inexorable chronological procession. A clash of editorial theories finds its battleground in the lists of apparatus.

An awareness of the two levels of discussion about apparatus can

137

help one analyze these arguments. One might at first assume that the objections of the genetic critics to traditional apparatus are on the more significant of the levels: they may seem to be saying that the traditional apparatus distorts the story they have to tell, that it reflects a misleading presentation, or fragmentation, of the text. But further reflection casts serious doubt on this assumption. The kinds of variants that genetic critics usually deal with are not the kinds Pebworth and Sullivan discuss: that is, students of textual genetics normally confront revisions made in a linear series, not variants that produce, or result from, independent traditions of transmission. Revisions that form a single sequence (and do not produce versions to be regarded as discrete works) lend themselves perfectly well to consolidated presentation; unlike variants that cluster into distinct groups, each telling a separate story, they form a continuous narrative.[89] If, then, a consolidated treatment of such revisions does not falsify (but rather clarifies) the picture of how the text developed, one must conclude that the choice between the two forms of consolidated treatment—texts with inclusive apparatus and texts with appended apparatus—is a decision on the level of efficiency and convenience for the user. Both forms bring together the evidence from separate documents to produce the editor's reconstruction of a historical process; neither one is inherently incompatible with that goal. The form the apparatus takes in these instances is not an unimportant matter, but it raises practical, not theoretical, issues.

Some of the editors who have opposed the appended style of apparatus have argued that it subordinates the readings thus recorded, taking them out of the context of the text as a whole, and complicates the reader's effort to see a continuous narrative. This position is often regarded as advanced or "radical," challenging the conservatism of an established tradition. In fact it is a traditional position, assuming the primacy of linear reading; the argument, on the other hand, that an appended apparatus does not subordinate material recognizes the more complex ways in which serious reading is performed. When we think of readers extractng all they can from books, reading intensely and productively, we do not picture them moving dutifully from one line to the next, but rather we see them jumping forward and backward, comparing one statement with another, bringing one point into the context of another—precisely the process McGann has effectively described as

138

"radial reading."[90] Editors who construct "traditional" appended apparatuses expect serious readers to behave in this way. The form of the text that they choose to present in linear style no doubt reflects their priorities (as, indeed, does the form other editors choose for ordering the readings of inclusive apparatuses), but they are not trying to suggest, by placing the apparatus at the foot of the page or the end of the text, that it is not to be read in the process of trying to understand the text. The Introductory Statement (1977) of the Center for Scholarly Editions, rightly regarded as an endorsement of intentionalist editing, described the importance of having variant readings present within the same volume that contains the rest of the text:

Textual scholars are not the only ones who use textual apparatus: any literary critic—indeed, any careful reader of the text—may, in considering a particular passage, wish to know whether any other versions of that passage have ever appeared. It may be vital to a particular interpretation to know what readings—if any—an editor has rejected as nonauthorial or superseded. If this information has to be searched out in the special-collections department of a particular research library, the matter may never be pursued. But if it is easily available in published form, such as a list in the edition that the critic is using or in a standard edition to be found in many academic and sizable public libraries, one can reasonably expect that the question is much more likely to be investigated. The ready availability of textual data, in other words, is likely to result in better-informed and more fruitful discussion of the writings involved. (pp. 3-4)

The case could be put still more strongly, but even here there is the recognition that apparatuses are to be used actively in reading. Whether one decides to insert variants into a running text or to record them in appended lists properly turns on the details of the individual situation, not on the preconceived notion that one or the other is necessarily easier for the reader to use. Each involves a set of conventions, and in general one can as readily become accustomed to the one system as the other. Some people have found Gabler's inclusive record of Joyce's revisions usable and effective, and others have not; some have been appreciative of Bowers's appended record of William James's revisions, and others have not. Readers, like editors, may disagree about which system should be used in particular cases; but it is not logical to regard one or the

139

other system as inherently more practical or convenient or informative for the consolidated presentation of readings from different documents.[91]

If both appended and inclusive apparatus can alternatively serve to present a conflated record of prepublication authorial revisions in a meaningful fashion, both can also be used effectively in presenting nonauthorial alterations that emerge from the initial and later publication process. The idea that an apparatus gives readers access to differing texts of a work (while at the same time showing the evidence underlying editorial decisions) is of course the traditional reason for providing apparatus; and it may yet offer a means for reconciling the social and the authorial approaches to editing. In the presidential address mentioned earlier, Fredson Bowers, speaking as one who favors authorial intention, remarked, "If he makes proper use of the apparatus, a cultural historian may find what he needs to know from a thoroughly edited work" (p. 8).[92] An editor approaching texts from the other direction

140

could say that students of authorial intention can find what they need in the apparatus. As long as it is not realistic to imagine that many works can routinely be the subject of alternative critical editions, some means must exist for accommodating different critical approaches within the same volumes. Apparatus (both inclusive and appended) has obviously been the standard device for accomplishing this goal, and the recent dissatisfaction with traditional intentionalist editions has often been directed at an assumed subordination of the readings in apparatus (particularly appended apparatus). All critical editors, like all other critics, bring points of view to their work, and in that sense an emphasis is placed on one approach in each instance. There is no reason why editors cannot, when they choose, take the social instead of the authorial point of view and represent the other in the apparatus. The key is the recognition that apparatus (whether inclusive or appended) presents alternative texts that can be read without difficulty, once the conventions are understood. In a thorough edition, the editor's own point of view does not deprive readers of reading alternative texts. Surely serious readers are not thwarted by the practical necessity (which everyone faces all the time) of accommodating themselves to specific (and sometimes unfamiliar) conventions and routines. Jo Ann Boydston, who shrewdly chose "In Praise of Apparatus" as the title of her 1989 presidential address to the Society for Textual Scholarship, emphasizes how apparatus, far from being a form of subordination, is "the history of a text," recreating "everything that happened to a text, from the author's conception of it throughout its life"—"a story of suspense and discovery, a true textual drama." What one can find in an apparatus, she concludes, is "a stimulating and highly productive intellectual adventure."[93] When apparatus is viewed in this spirit, most of the quarrels over formal arrangements shrink into insignificance, and a fruitful understanding of the validity of alternative textual theories can prevail.

These considerations lead to the question of what a critical text constructed according to a social textual theory would amount to. One would select as copy-text the text that best reflects the intentions of those persons responsible for the public presentation of a work at a given time, and one would emend that text to correct readings not intended by them. Presumably the resulting text would not be very different

141

from the text of a published edition (or, for earlier periods, a scribal manuscript) that actually circulated, for the emendations in most cases would consist entirely of the correction of so-called typographical errors (or slips of a scribe's pen).[94] Theoretically, of course, one might choose to be interested in the intentions of one, rather than all, of the persons involved in the production process—focusing on the intentions of a publisher's editor, for example, and attempting to eliminate the alterations made by compositors and copy-editors. But it seems unlikely (except perhaps in the case of certain well-known editors, like Maxwell Perkins) that the interest in any one of these persons would outweigh the interest in the joint product of all of them. The question then arises as to whether an editorially constructed critical text has any advantage over facsimiles and transcriptions for the student of literary sociology. The critical text would weed out certain slips (such as typographical errors) that could not have been an intended contribution of the production process; but those slips, after all, were in fact one of the results of that process and were a part of the texts that were presented to the public. One wonders, therefore, whether it is worth while to prepare critical texts reflecting the collaborative process of publication, when facsimiles and transcriptions can come closer to showing what the readers of a given time actually had at their disposal.

This point of view is by no means novel: it is, indeed, a traditional one, and it brings us back to the most basic decision that all editors must make. Every editor must decide whether to present the texts of documents (and thus use a noncritical presentation, such as photographic facsimiles or literal—"diplomatic"—transcriptions) or whether to go beyond the documents and attempt to construct the texts of works as they were intended by one or more persons in the past (and thus use a critical presentation, in which documentary texts are emended to bring them closer to the intended forms as conjecturally established). Both approaches have long histories. When editors of the past chose facsimiles and transcriptions, they were no doubt thinking primarily of the value of making documentary evidence widely available; they naturally understood that the texts present in medieval manuscripts and later printed editions were the texts that had emerged from the publication process

142

and had been given to readers, but they did not always announce this point because (aside from its being obvious) it had not been made a prominent issue for discussion. Nevertheless, facsimiles and transcriptions do serve the interests of literary sociology and cultural history, and students of those fields do already have a great mass of editions appropriate to their needs. It is true that editors who have decided to produce critical texts have concentrated almost exclusively on authorial intention; but even those who gave little thought to other approaches were obviously aware of the existence of noncritical or documentary editions, serving other purposes, and realized that they had chosen one editorial path rather than another. There has always been an implicit understanding that the reconstruction of authorially intended texts is not the only possible approach to textual study, even if the alternatives were not so precisely delineated or so insistently advocated as they have been in recent years. And can one, in retrospect, blame critical editors for focusing on authorial intention? They may not have considered alternatives; but we, looking back, can see that any other goal for a critical edition generally makes less practical sense. If one is making the critical effort of constructing a text that recreates some moment in the textual history of a work, there is usually little to be gained by choosing any goal other than an authorially intended text—for authorially intended texts are rarely, if ever, to be found perfectly embodied in surviving documents, and their attempted recovery requires acts of informed critical judgment, whereas socially produced texts are available intact in documents that survive and are normally (at least in the era of printing) in no need of reconstruction.[95]

Textual and literary theorists can, and will, continue to debate the nature of texts, but editors have to face the practical question of how their procedures are affected by the theoretical positions they hold. When they examine the choices before them, they will see that their alternatives have not been changed by the debates. There is no escape from the eternal dilemma posed by works in the medium of language (or in any other intangible medium): do we accept the texts of artifacts,

143

which are primary evidence of the forms of works that were disseminated at particular times, or do we create new texts from that evidence, hoping through the trained historical imagination to come closer to what the authors (or other producers) of the works intended? Editors who have contemplated the conflicting demands of the social and the authorial theories of the production of texts are still confronted with the old choice between documentary and critical editions; and as a practical solution they may well decide, as editors before them have generally decided, to prepare critical texts if their primary interest is in authorial intention and to produce facsimiles or transcriptions if their primary interest is in surviving documents, either as records of the genetic history of texts or as the collaborative products of the publication process. If editing thus goes on as before, it nevertheless will not go unchanged, for the framework of thought within which editorial choices are made will have been more fully articulated as a result of the discussions of social textual criticism. The study of the past inevitably involves thinking about the role of individuals in history as against the role of social process. Both must be investigated: the intentions of individual creative minds will always be a valid subject for textual critics to pursue, as will the forms of texts that reached the public, shaped by the social forces of a given moment. The two are complementary, and any claims that one supersedes the other are obviously naïve. Partisanship is a natural element in attempts to revitalize what is perceived as a neglected concern, and exaggerated claims are a part of the process. But the lasting legacy of the recent debates, after the partisan controversy has taken its place as an episode in the history of scholarship, will be a greater awareness of the theoretical alternatives for textual study and a wider understanding of the position of textual criticism in intellectual life.

Notes

[1]

The fundamental distinction between the texts of works and the texts of documents, which will emerge at various points in the following pages, I have elaborated most fully in A Rationale of Textual Criticism (1989). (Cf. also note 63 below and my "Textual Criticism and Deconstruction," Studies in Bibliography, 43 [1990], 1-33.)

[2]

This is the fourth in a series of surveys covering the period since mid-century. The three earlier essays, originally published in Studies in Bibliography in 1975, 1981, and 1986, have been brought together as Textual Criticism since Greg: A Chronicle, 1950-1985 (1987). (Any reference to these essays in the present piece provides the SB citation first, followed in brackets by the page reference to the 1987 book.) As in the earlier essays, I again focus on discussions of general theoretical significance in the English language and therefore do not mention many of the articles that concentrate on textual problems in particular authors' works and many of the reviews of specific editions. During the period surveyed here, a useful volume of essays containing reflections on editorial history appeared in the series of Toronto Conferences on Editorial Problems: Editing and Editors: A Retrospect, ed. Richard Landon (1988), including essays on medieval Latin texts (Leonard E. Boyle), the Greek New Testament (Bruce M. Metzger), eighteenth-and nineteenth-century writings (Donald H. Reiman), and American literature (David J. Nordloh). (The Toronto series has also recently included Editing Early English Drama, ed. A. F. Johnson [1987]; Editing Modern Economists, ed. D. E. Moggridge [1988]; and Editing Greek and Latin Texts, ed. John N. Grant [1989].) Three significant surveys of earlier editorial history are Richard W. F. Kroll, "Mise-en-Page, Biblical Criticism, and Inference during the Restoration," Studies in Eighteenth-Century Culture, 16 (1986), 3-40; Joseph A. Dane, "The Reception of Chaucer's Eighteenth-Century Editors," Text, 4 (1988), 217-234; and C. O. Brink, English Classical Scholarship: Historical Reflections on Bentley, Porson and Housman (1986). Some historical comments on the nineteenth- and twentieth-century editing of medieval manuscripts appear in J. A. Asher's "The Textual Criticism Connection," Te Reo: Journal of the Linguistic Society of New Zealand, 29 (1986), 305-311. And a surprisingly detailed narrative of the editing of American literature under the CEAA (with a glance at recent issues) occupies much of Guy Cardwell's "Author, Intention, Text: The California Mark Twain," Review, 11 (1989), 255-288. Another historical account of the development of modern editing appears as the first chapter (pp. 1-29) of Mary-Jo Kline's A Guide to Documentary Editing (1987), prepared for the Association for Documentary Editing (ADE). Her account understandably emphasizes the editing of the texts of documents and the different traditions followed by editors of literary and historical figures; although she has conscientiously informed herself about the field of literary editing, she obviously speaks as one who has given more thought to historical annotation than to textual matters. (The bulk of the book deals with the routines and procedures for managing editorial projects; I have criticized its treatment of photocopies on pp. 39-41 of "Reproductions and Scholarship," SB, 42 [1989], 25-54. For a firm, yet generous, review of this book, see T. H. Howard-Hill, "Documentary Editing: Critical, Noncritical, Uncritical," Review, 10 [1988], 149-154.) More recently, the ADE has sponsored another volume, Editing Documents and Texts: An Annotated Bibliography (1990), by Beth Luey with the assistance of Kathleen Gorman. Although its goal is to cover both literary and historical editing, this checklist of about 900 entries, which are annotated with brief unevaluative comments, is slanted toward the editing of historical documents, for it cites many pieces from Documentary Editing on individual editions and omits many similar discussions, especially of literary editions, published elsewhere. But it is of course useful for the large number of references that it does bring together. (Joseph R. McElrath, Jr., in a brief historical description of recent editorial debates published in the journal of the ADE, emphasizes the divergence between "literary" and "historical" editors—unfortunately called "LITs" and "HITs" throughout—as if that were the meaningful dividing line; see "Tradition and Innovation: Recent Developments in Literary Editing," Documentary Editing, 10, no. 4 [December 1988], 5-10. This paper forcefully illustrates that the distinctions urged in Jo Ann Boydston's admirable ADE presidential address in 1985 have not been heeded, and she properly criticized several aspects of this paper in a letter to the editor; see her "The Language of Scholarly Editing," Documentary Editing, 7, no. 4 [December 1985], 1-6, and her letter in 11, no. 1 [March 1989], 28.)

[3]

Three thoughtful reviews, which should be read by anyone interested in pursuing the implications of McKenzie's general position, are those by Hugh Amory (Book Collector, 36 [1987], 411-418), T. H. Howard-Hill (Library, 6th ser., 10 [1988], 151-158), and Jerome J. McGann (London Review of Books, 18 February 1988, pp. 20-21). I share with Amory and Howard-Hill a feeling that these lectures are seriously flawed, but each of us emphasizes somewhat different matters (and, indeed, we are not always in agreement). Some comments on Amory and Howard-Hill appear in notes 5, 6, 7, 10, and 11 below; I discuss McGann's review in more detail at the beginning of part II below. An instance of the publicity that McKenzie's lectures have received is Roger Chartier's article ("Meaningful Forms") in the first number of Liber (October 1989; distributed to English-speaking readers with the TLS for 6-12 October 1989), pp. 8-9. Chartier considers the lectures "a brilliant manifesto in favour of an ambitious new definition of bibliography" and uses them uncritically to introduce the subject of "the effects on meaning produced by material forms." Chartier also cites McKenzie (p. 162) in "Texts, Printing, Readings," in The New Cultural History, ed. Lynn Hunt (1989), pp. 154-175; in this essay Chartier locates his field (as his tripartite title suggests) at "the crossroads of textual criticism, the history of the book, and cultural sociology" (p. 175)—yet no text exists, he claims, "outside of the support that enables it to be read" (p. 161).

[4]

I have discussed these two papers in SB, 39 (1986), 14-18 [122-126].

[5]

Howard-Hill, in his review (see note 3 above) provides a fuller analysis of Greg's essay and McKenzie's distortion of it (pp. 152-153). John Barnard, in "Bibliographical Context and the Critic," Text, 3 (1987), 27-46, quotes (p. 28) a passage from the same 1932 essay of Greg's, stating that a text is a "living organism," which at each stage of "its descent through the ages" is "in some sense a new creation, something different from what it was for an earlier generation"—a point of view strikingly similar to McKenzie's. Barnard specifically aligns himself with McKenzie and McGann, asserting that "bibliography needs to widen its aims" (p. 40). (Greg's essay, "Bibliography—An Apologia," is available in his Collected Papers, ed. J. C. Maxwell [1966], pp. 239-266.)

[6]

The line "Rare book rooms will simply become rarer" is not only a poor joke; it is an incoherent intrusion. The implied connection between rare-book rooms as institutions and bibliography as a professional discipline cannot be taken for granted and would be difficult to support by a rational argument. Amory, in his review (see note 3 above), comments on this line of McKenzie's with characteristic flippancy, suggesting in his own way that intellectual pursuits have a higher aim than institutional self-preservation: "the prospect has its charms. Many a rare book room rests on the ruins of another, and a thoroughly [sic] reshuffling of the deck might advance bibliography more effectively than years of supine institutional possession" (p. 415).

[7]

"Bibliography—A Retrospect," in The Bibliographical Society 1892-1942: Studies in Retrospect (1945), pp. 23-31 (quotations from pp. 27, 30). Both Amory and Howard-Hill, in their reviews of McKenzie (see note 3 above), seem to me to be overly concerned with the question of how "bibliography" is to be defined; and by returning to this matter at the ends of their reviews, they both conclude with passages that are far from their strongest.

[8]

Set forth most fully in "Printers of the Mind: Some Notes on Bibliographical Theories and Printing-House Practices," SB, 22 (1969), 1-75; see the responses by Peter Davison, in "Science, Method, and the Textual Critic," SB, 25 (1972), 1-28, and by me, in "Bibliography and Science," SB, 27 (1974), 55-89 (esp. pp. 73-78), reprinted in Selected Studies in Bibliography (1979), pp. 1-35 (esp. pp. 19-24). A recent brief assessment of the effect of McKenzie's essay appears in Robert Kean Turner, "Accidental Evils," in Play-Texts in Old Spelling, ed. G. B. Shand with Raymond C. Shady (1984), pp. 27-33 (see p. 33).

[9]

Whether or not they misquoted it unintentionally is not clear. It is true, as McKenzie says (p. 18), that the substitution of "wrote" for "wrought" occurs in the anthology edited by George H. Nettleton and Arthur E. Case, British Dramatists from Dryden to Sheridan (1939); but that fact alone does not seem sufficient evidence from which to conclude that Wimsatt and Beardsley used this anthology. Their quoted passage differs in punctuation at six points from the Nettleton-Case text, whereas it agrees perfectly with the punctuation in two other popular American anthologies of the time: Plays of the Restoration and Eighteenth Century, ed. Dugald MacMillan and Howard Mumford Jones (1931); and the Modern Library Twelve Famous Plays of the Restoration and Eighteenth Century, introd. Cecil A. Moore (1933). These two anthologies do have "wrought" (and one of them—the 1933—differs from Wimsatt-Beardsley in the spelling "dulness" rather than "dullness"). But the question remains open whether greater weight is to be given to the agreement in punctuation on the one hand or the agreement in the reading "wrote" on the other.

[10]

Further shortcomings of this illustration are perceptively noted by Howard-Hill on p. 155 of his review (see note 3 above).

[11]

It is therefore surprising to find Amory, in his review (see note 3 above), saying that the current critical climate, reflected by McKenzie, places on libraries the "burden" of preserving "decaying reprints that in theory are primary sources—of something or other, anyway, if the literary theorists could only spit it out" (p. 417). Bibliographers, long before the current critical theorists, have repeatedly recognized that all printed items are primary sources for the history of printing and publishing and that "reprints" are primary sources for the study of authors' reputations. Amory is perfectly well aware of these points; yet for some reason he is willing to make the foolish statement, "There is such a thing as too much evidence."

[12]

Writers are really concerned to write works, of course, not texts, though in order to transmit those works they must produce specific texts, which may or may not faithfully represent the works.

[13]

Bibliography, that is, in the broad sense. One of his earlier criticisms of analytical bibliography was its supposed exclusion of the human. Presumably one of the connections he sees between analytical bibliography and the New Criticism (cf. p. 7) is that the former "has obscured the role of human agents" by ignoring the "inevitable dependence upon interpretive structures" (p. 8), just as the latter has eliminated "a concern for the complexities of human agency in the production of texts" (p. 19). Analytical bibliographers and textual critics in the Greg-Bowers tradition, however, have by no means been unaware of the fact that they were dealing with the actions of human beings in the past: one of the best illustrations of this point is Greg's classification of variants into substantives and accidentals as a way of reflecting the attitudes of Renaissance compositors toward their copy. (On the New Criticism and analytical bibliography, see also note 22 below.)

[14]

New meanings, of course, can also occur without new forms, as earlier editions are read in new contexts.

[15]

See, for example, Vinton Dearing's opening statement in the preface to A Manual of Textual Analysis (1959): "The method of analysis described in the following pages may be applied to the transmission and embodiment in any form of any idea or complex of ideas" (p. vii). In his expanded work Principles and Practice of Textual Analysis (1974), he opens his first chapter with the more explicit statement, "Messages are not confined to words or even to sound, but may consist wholly of visual images"; hence, "There is no limitation whatsoever in textual analysis on the type or types of transmitters. Textual analysis is . . . a completely general discipline of very wide specific applicability in the arts and social sciences." Dearing has recently summarized his distinction between the "genealogy of texts" (the subject of "textual analysis") and the "genealogy of media," in "Textual Analysis: A Kind of Textual Criticism," Text, 2 (1985), 13-23.

[16]

But sometimes puzzlingly. When he mentions television weather maps, he raises this question: "Should we not at least be asking . . . at which point one stops a kinetic image to keep a record for posterity?" (p. 38) The answer is obviously no: a single still image (or a number of still images) would falsify the record, since the text consists of moving images. Two pages later, after discussing a comic-strip rendering of Shakespeare, he says, "I hasten to add that I am not endorsing the form as a suitable one for Shakespeare" (p. 40). What can this mean? The comic strip is of course not a theatrical performance; it is a work, inspired by Shakespeare, in another medium. Both these comments (like the treatment of landscape) suggest a reluctance on McKenzie's part to accept nonverbal texts on their own terms. I do not comment here on his brief account of drama (pp. 40-41) —which he begins by claiming, "The relation of textual criticism to the realities of theatrical production has always been one of embarrassed impotence"—because I take up the same issues (of the relation of play scripts to performance) in discussing Howard-Hill below (see note 63).

[17]

In his discussion of maps, he says that, "not as books but as texts, bibliographical principle embraces them [maps] too" (p. 37). What is "bibliographical principle"? If it is the "textual principle" that is central to bibliography, then the statement is merely a tautology.

[18]

The idea that a work of literature employs an intangible medium is not Platonic, as McKenzie seems to suggest in the second lecture (p. 24). One might more readily see as Platonic McKenzie's reference to "a kind of ideal-copy text, transcending all the versions and true to the essential intention of the 'work'" (p. 29), or his statement that "All the versions [performances of a play] imply an ideal form which is never fully realized but only partly perceived and expressed by any one" (p. 41). McKenzie does not seem to be grappling here with the intangible nature of the medium of verbal works, which makes the work itself always indeterminate. And he is certainly not suggesting that the ideal form is the one envisaged by the author. His argument does not require a concept of ideal form, and in the context of that argument these statements are incomprehensible.

[19]

I have discussed this book in SB, 39 (1986), 19-27 [127-135], along with several of his essays that have since been collected in The Beauty of Inflections (1985): "The Text, the Poem, and the Problem of Historical Method," "Shall These Bones Live?", and "The Monks and the Giants: Textual and Bibliographical Studies and the Interpretation of Literary Works." More recently, a linked pair of important criticisms of McGann's Critique has appeared: David J. Nordloh's "Socialization, Authority, and Evidence: Reflections on McGann's A Critique of Modern Textual Criticism," Analytical & Enumerative Bibliography, n.s., 1 (1987), 3-12; and Craig S. Abbott's "A Response to Nordloh's 'Socialization, Authority, and Evidence,'" ibid., 13-16. Nordloh provides an admirable summary of Greg's position to show how McGann misunderstands Greg and Bowers, and he then argues that McGann's "notion of socialization introduces a dangerous vagueness" into editing (p. 7), making it no longer a "serious intellectual endeavor, circumscribed by evidence and limited by coherent, identifiable premises" (p. 12). He is correct to point out that McGann "seems less interested in fully defining the concept [of socialized textual authority] than in employing it as a weapon against other editorial principles" (p. 5); but Nordloh, in turn, does not go very far toward imagining the uses of a socialized approach by insisting that, "for intelligibility and for coherence, authority must be defined as precisely as possible in terms of active authorial creation" (p. 11). Abbott reinforces Nordloh's criticisms by noting that McGann "exaggerates the intransigence of Bowersian editing" (p. 13); and he predicts that the result of the current debates will be "a wider variety of editorial approaches, not the replacement of one approach with another" (p. 16). (Although neither Nordloh nor Abbott refers to McKenzie, their articles are joined under the heading "The Sociology of the Text.")

[20]

"Theory of Texts," London Review of Books, 18 February 1988, pp. 20-21.

[21]

In the preface to his latest book, Towards a Literature of Knowledge (1989), McGann repeats this three-part classification of the "levels" of "operation" at which one can study the "social text" (p. ix).

[22]

McGann, following McKenzie, pairs the New Criticism with analytical bibliography (the "New Bibliography"), calling them, respectively, "versions of hermeneutical idealism and textual positivism" (p. 20, col. 2). Analytical bibliography may be considered positivistic, but not textually so. Nevertheless, McGann's succinct characterizations are more reasonable than McKenzie's strained comparison of the two. Analytical bibliography and the New Criticism, McKenzie believes, share a "view of the self-sufficient nature of the work of art or text" and "showed great ingenuity in discerning patterns" without concern for "precedent or subsequent processes" (p. 7). It is true of neither the New Criticism nor analytical bibliography that they have no regard for "precedent" (that is, history). In the case of analytical bibliography, Bowers, for one, emphasized what he called the "postulate of normality" in interpreting bibliographical evidence: that is, one must examine the evidence from a given piece of printing in the light of what is presently accepted—on the basis of previous investigations—as the normal practice for the period or shop involved (see Bibliography and Textual Criticism [1964], pp. 64-77). As for the New Criticism, Cleanth Brooks says in the preface to The Well Wrought Urn (1947), a book often considered the epitome of the New Criticism, "If literary history has not been emphasized in the pages that follow, it is not because I discount its importance, or because I have failed to take it into account" (p. x); and in an appendix (after the body of the book has taken up a series of poems from Donne to Yeats in chronological order) he says, "I certainly have not meant to imply that the poet does not inherit his ideas, his literary concepts, his rhythms, his literary forms" (p. 197).

[23]

He quotes McKenzie's statement that the new edition could not "represent the physical form of Ulysses as it was first published." Then, after summarizing McKenzie's presentation of information provided by John Kidd, he observes, "This limitation turns out to be an important one, because Joyce appears to have used the page sequences and lay-out of the 1922 edition as part of the work's semiotic system" (p. 21, col. 3). According to McGann's theory of reading, however, would not such a limitation be important in any historical study, whether or not the author is known to have made use of the physical details of bookmaking?

[24]

Criticism, 27 (1985), 283-305; reprinted, with slight revisions and a new title ("Ulysses as a Postmodern Work"), in his Social Values and Poetic Acts: The Historical Judgment of Literary Work (1988), pp. 173-194. The latter text is cited here. My comments below on the Gabler edition are limited to points raised by McGann. For a detailed criticism of the Gabler edition, see John Kidd, "An Inquiry into Ulysses: The Corrected Text," PBSA, 82 (1988), 411-584. And for accounts of the much-publicized controversy over the Gabler edition, initiated by Kidd, see Robin Bates's "Reflections on the Kidd Era," Studies in the Novel, 22 (1990), 119-141, and two essays by Charles Rossman, also in Studies in the Novel, "The Critical Reception of the 'Gabler Ulysses': Or, Gabler's Ulysses Kiddnapped" (21 [1989], 154-181), and a sequel to be published in the fall 1990 number. Examples of the way in which the Gabler edition (as well as McGann's view of it) has prompted discussion of textual theory in general are Ira B. Nadel's "Textual Criticism, Literary Theory and the New Ulysses," in Assessing the 1984 "Ulysses", ed. C. George Sandulescu and Clive Hart (1986), pp. 122-139; Patrick McGee's "Is There a Class for This Text? The New Ulysses, Jerome McGann and the Issue of Textual Authority," Works and Days, 5, no. 2 (1987), 27-44; and McGee's "The Error of Theory," Studies in the Novel, 22 (1990), 148-162.

[25]

McGann also says that "Gabler's edition calls attention to a peculiar generic quality of modernist [postmodernist?] writing: that its subject is often the act and process of writing itself" (p. 182). But all editions with apparatus recording authorial revisions call attention to "the act and process of writing."

[26]

His switch to "literary works" here—when he had been talking about an edition of a work—is perhaps meant to suggest that the "work" is inseparable from the apparatus in which it is "enmeshed" in a given edition, the two becoming a single new entity.

[27]

And, he says, the availability of many documents in published facsimile (such as The James Joyce Archive and the Rosenbach facsimile) "has made it possible for any student anywhere in the world to follow Gabler's work in the most minute and exacting detail" (p. 174). That no facsimile can ever be trusted to this extent is set forth in my "Reproductions and Scholarship," SB, 42 (1989), 25-54, which includes some comment on the deficiencies of the Joyce facsimiles and on Gabler's excessive reliance on those facsimiles (p. 32). John Kidd (see note 24 above) has discussed this matter in detail (pp. 433-448); and Gabler's admission that he did not fully collate his transcription against the Rosenbach original appears in Robin Bates's article (also cited in note 24 above).

[28]

As he also does elsewhere. For example: "Since by their autograph overlay the typescripts and proofs partially acquire the status of documents of composition, the question arises of how far the authorial presence affects, and penetrates, their basic level of transmissional transcription" (p. 1893). This statement describes an issue that editors have always dealt with; Gabler's classification of typewritten and printed documents as essentially transmissional, however, proves to be an impediment to a clear exposition of the issue.

[29]

A primary instance is McGann's contradictory description, within the space of one page (p. 176), of the authority of the 1922 first edition. He first notes that Gabler places the 1922 readings in the historical collation, not in the synoptic text, because "the first edition is prima facie a part of the 'full record of corruption'"; half a page later he calls the first edition "the 'ultimate stage of compositional development.'" The latter phrase is quoted out of context from Gabler and produces an inconsistency that was not present in Gabler's discussion: Gabler said, "The first edition admittedly represents the closest approximation to be found in one document of the work at its ultimate stage of compositional development" (p. 1894). Another misleading statement of McGann's is the following: "Because Gabler wants to assemble a text of the work's compositional development . . . and because he regards the act of composition as an entirely isolated and personal affair, he always sets a privilege on autograph manuscript texts. The typescripts, the proofs, and the first edition involve the intervention of other, purely transmissive authorities, and hence they fall outside the process of compositional development" (p. 177). This way of summarizing Gabler's procedure does not allow, as Gabler does, for the use of such documents as proofs and published editions as evidence of authorial revision in lost documents.

[30]

McGann's explanation of the "usual understanding" never mentions the distinction between substantives and accidentals and defines copy-text simply as "what an editor chooses to take as the text of highest presumptive authority" (p. 177); he thus ignores, in what he calls "the post-Greg context," Greg's primary contribution, the suggestion that one text may serve as the primary source for accidentals, another for substantives. Gabler's understanding of copy-text is similarly skewed: "By common consent, an editor chooses as the copytext for a critical edition a document text of highest overall authority" (p. 184). Greg's point was that no single document might have "overall authority."

[31]

A few pages earlier, McGann had described Gabler's text as "the work of Ulysses as Joyce actually produced it in a continuous act of writing and rewriting" (p. 175). The word "actually" does not allow for the role of judgment and conjecture involved in producing "an imagination of Joyce's work" (or in imagining "a continuous act of writing and rewriting")—or, indeed, in preparing any critical edition. Furthermore, McGann should have questioned whether it is proper to imagine all intended readings as having appeared in Joyce's hand—that is, whether it is proper to imagine lost documents that present Joyce's intended readings without error. It is always possible that some intended readings may never have appeared in the author's hand (such as a typist's or printer's correction of an author's slip of the pen). Gabler is therefore holding an untenable position when he says that "the final state of the text's development is considered reached when it is last fully and correctly written out in the author's hand" (p. 1901). Given Gabler's focus on intended texts abstracted from documents, he is free to emend errors in extant documents and to imagine the intended texts of inferred documents; but he cannot claim that such intended texts were always "fully and correctly written out in the author's hand." His tying of intended readings to handwriting, and thus to documents (whether extant or inferred), is a central weakness of his whole approach.

[32]

In the case of Billy Budd, Sailor there is only one manuscript to report, but it is a complicated manuscript showing many revisions that reflect several stages of composition. On the differing significance of inclusive apparatus in a transcription of one document and a consolidated record of multiple documents, see note 91 below.

[33]

More recently still, McGann has advocated a form of edition consisting of a printed reading text accompanied by a computer disk containing "the electronic hypertext"—that is, a record of all variants, accessible "through hypertext programs which would enable the reader to reconstruct any state of any particular text, and to organize those particular texts into any form or order within themselves or in relation to each other" ("Which Yeats Edition?", TLS, 11-17 May 1990, pp. 493-494). Although he does not use the term "redial reading" in this review, a computerized apparatus would clearly facilitate such reading.

[34]

I am not claiming that McGann does not understand this point but that his prose fails to convey it.

[35]

Such as two slight pieces for the Society for Textual Scholarship. The first, "Interpretation, Meaning, and Textual Criticism: A Homily," sums up his "socio-historical" approach to textual scholarship in terms of a "schematic outline" moving from the "originary textual moment" to the "immediate moment of textual criticism," via a series of "secondary moments of textual production and reproduction" (Text,, 3 [1987], 55-62). The second, "The Textual Condition," restates his plea for "bringing about an end to the schism in literary studies" between textual and literary criticism (Text, 4 [1988], 29-37). One of his main points in the latter piece cannot be too often emphasized: "Scholarship is interpretation, whether it is carried out as a bibliocritical discourse or a literary exegesis" (p. 37).

[36]

The interview, effectively conducted by David Gorman in the fall of 1986, was published the next year as "An Interview with Jerome McGann on Textual Scholarship as Literary History and Ideology Critique," Social Epistemology, 1 (1987), 163-173 (see p. 165). His point here has a different emphasis (as he recognizes) from his statement in the Critique that author's intentions are "only one of many factors to be taken into account" and that they may sometimes "determine the final decision" (p. 128); cf. my discussion of this statement in SB, 39 (1986), 22-25 [130-133]. (McGann's awareness, in the interview, of "the possibility of different kinds of editions for different works and authors" is welcome, but how it "moves against certain currently dominant views and procedures in editorial method" is unclear.) Later, in the preface to Towards a Literature of Knowledge (1989), McGann speaks of "the network of intentionalities which constitute the field of the social text," a "field of intentions" made up of various "intentional structures or agents," none of which "will ever be equal to the entire set" (p. x); but he does not in this context comment on the editorial possibilities of focusing on particular intentional agents.

[37]

"A New Approach to the Critical Constitution of Literary Texts," SB, 28 (1975), 231-264. Zeller's emphasis on the integrity of every "version" of a work is linked with a social view of literary production: he believes that "uninfluenced artistic intentions" (p. 248) cannot be separated from the "play of forces from all sides" on the author (p. 244)—"the magnetic needle of the author's wishes is quivering in the field of non-aesthetic forces" (p. 245). I have commented on Zeller's article in "Problems and Accomplishments in the Editing of the Novel," Studies in the Novel, 7 (1975), 323-360 (see pp. 329-331) and more briefly in SB, 34 (1981), 30-31 [72-73].

[38]

For the literature of this general movement, see (in addition to the editions referred to below) the documentation to Hurlebusch's SB essay (note 43 below).

[39]

"The Text as Process and the Problem of Intentionality," Text, 3 (1987), 107-116.

[40]

"The Synchrony and Diachrony of Texts: Practice and Theory in the Critical Edition of James Joyce's Ulysses," Text, 1 (1984 for 1981), 305-326; I have commented on this essay in SB, 39 (1986), 37-39 [145-147].

[41]

"Does 'Text' Exist?", SB, 41 (1988), 64-76; "Genetic Editing, Past and Future: A Few Reflections by a User," Text, 3 (1987), 117-133.

[42]

After mentioning the varieties of apparatus employed by these scholars to show both "the genesis in its process and the final state of the text," Hay makes an odd observation: "they all mark their distance from the traditional apparatus of variants, abandoning the viewpoint of pure erudition for the problematic of the pre-text" (p. 70). The original French does say "l'érudition pure"; apparently "érudition" here carries the sense not of learning but of established and incontrovertible fact. See p. 152 of "'Le texte n'éxiste pas': Réflexions sur la critique génétique," Poétique, 62 (April 1985), 147-158.

[43]

"Conceptualisations for Procedures of Authorship," SB, 41 (1988), 100-135; "'Relic' and 'Tradition': Some Aspects of Editing Diaries," Text, 3 (1987), 143-153.

[44]

He offers two very different examples of the former category. Unquestionably the selective recording of variants in the Weimar Goethe edition (1887-1919) reflects a lack of concern for genetic study. But Friedrich Beissner's edition (1943-62) of Hölderlin contains detailed records of manuscript variants, and Hurlebusch places it in this category because "the author is still granted the decision how his works are to be read" (p. 111), with a final alteration "not seen as the last of several textual options, but as the 'only possible form' or the 'consummate form' of expression" (p. 113). Hurlebusch regards Beissner's presentation of a text representing the author's final intention as a sign of his subordination of genetic process, despite the thorough genetic record. Karl Goedeke's edition of Schiller (1867-76), on the other hand, is considered a "production-oriented" edition, even though its apparatus (like Beissner's) presents manuscript variants "according to the pattern of the apparatus criticus in classical philology" (p. 115), because the main text in each instance is the earliest one and the aim of the edition is to provide "a history of Schiller's mind" (p. 114). Hurlebusch sensibly does not make the form of the apparatus the determining characteristic here; but he might have noted that Beissner was not slighting the composition process simply because he wished to offer readers a text reflecting the author's final intention, for his edition illustrates how a study of the genetic record is essential to the construction (and critical reading) of such a text.

[45]

For example, he concludes his section on "reception-oriented" editions in this way: "As long as there are authors who sufficiently clearly lay down their decisions on the versions in which they wish their texts to be read, and as long as there are readers willing to submit their souls to authorial guidance, this editorial concept cannot be considered outdated" (p. 114). But an editor's decision to aim at an authorially intended text is independent of the degree to which the author has made the task easier—and, in any case, critical editors must evaluate for themselves every purported authorial decision (as must critical readers, even if they ultimately wish to "submit their souls to authorial guidance"). An "editorial concept" is valid or invalid (not "outdated") according to its own logic and coherence; it may fall out of favor (and seem "outdated"), however, as fashions in critical approaches to literature shift.

[46]

He invents an unnecessary dilemma when he asks, "Are they ['the writers' working procedures'] of interest merely by their workshop handling of texts and variants, and not also as the author's medium of expression?" (p. 124). That he sees these two as separable is shown by his answering the question affirmatively in regard to alterations that can be classified as "revision" (rather than "production" or creation).

[47]

He asserts, for example, that diaries "must be edited in a way which preserves their individuality and uniqueness as documents of non-intended transmission while maintaining as far as possible the conventions of text rendering and readability which belong to the area of 'tradition' documents" (p. 146). (A "tradition" document, in his terminology, contains a text intended for publication; a "relic" document contains a text not intended for publication.) Or again: "the task of the editor is twofold: on the one hand, he has to preserve the nature of the document and, on the other, he must present a readable text" (p. 148). The quite proper emphasis in the SB essay on the importance of the physical features and idiosyncrasies of private documents would seem to make the "conventions of text rendering and readability" irrelevant.

[48]

"Script, Work and Published Form: Franz Kafka's Incomplete Text," SB, 41 (1988), 77-99. The other two essays are Jean-Louis Lebrave's "Rough Drafts: A Challenge to Uniformity in Editing," Text, 3 (1987), 135-142; and Siegfried Scheibe's "Some Notes on Letter Editions: With Special Reference to German Writers," SB, 41 (1988), 136-148. Lebrave, who is working on computer systems for analyzing manuscript variants in Heine, believes that standard approaches to recording variants do not facilitate the analysis of the "genetic memory" (p. 135) embedded in rough drafts and other avant-textes. Scheibe, an editor of Goethe, Georg Forster, and Christian Gellert, discusses general guidelines for determining the contents of editions of letters.

[49]

See the paragraph containing the statement, "Cancellations in pre-copy-text forms will normally be fully reported" (Statement of Editorial Principles, p. 8). Robert Murray Davis has recently suggested that students of textual genesis and the creative process should draw on what composition teachers have learned about "recursion," a nonlinear, looping pattern characterized by pausing, moving backward, rewriting, and so on: see his "Writing as Process: Beyond Hershel Parker," Literary Research, 12 (1987), 179-186. His title of course implies that this understanding of composition is an advance over the concept of the determinate period of creativity presented in Hershel Parker's Flawed Texts and Verbal Icons (1984)—on which I commented in my previous survey (SB, 39 [1986], 27-34 [135-142]). (For two additional views of Parker, see James McLaverty in Review, 8 [1986], 119-138, and Don L. Cook in Documentary Editing, 9, no. 1 [March 1987], 5-8.) Parker has presented a summary of his position in "'The Text Itself'—Whatever That Is," Text, 3 (1987), 47-54.

[50]

For references to their work, see my comments in the preceding survey, SB, 39 (1986), 36 [144], note 68. More recently, Steven Urkowitz has said that "today we confuse bibliographical expertise with textual omniscience," in "'Well-sayd olde Mole': Burying Three Hamlets in Modern Editions," in Shakespeare Study Today, ed. Georgianna Ziegler (1986), pp. 37-70 (see p. 68). And Stanley Wells, in "Revision in Shakespeare's Plays," in Editing and Editors: A Retrospect, ed. Richard Landon (1988), pp. 67-97, advises editors to "get out of the strait-jacket of attempting to provide texts that aspire to definitiveness and aim rather at a reasoned plurality" (p. 97). See also Grace Ioppolo, "'Old' and 'New' Revisionists: Shakespeare's Eighteenth-Century Editors," Huntington Library Quarterly, 52 (1989), 347-361. Other scholars, reflecting experiences in other fields, have also taken up this matter. S. M. Parrish, for instance, thinks of each version as "a vertical slice cut through the continuum of text" (p. 346) and defends "the autonomy and the validity of each steady state of the text" (p. 349); see his "The Whig Interpretation of Literature," Text, 4 (1988), 343-350 (in which the "Whig" view disregards early versions as superseded by "an inner logic of inexorable growth" toward a final form [p. 349]). Another recent proposal for separate editions of versions comes from John T. Shawcross, in "Scholarly Editions: Composite Editorial Principles of Single Copy-Texts, Multiple Copy-Texts, Edited Copy-Texts," Text, 4 (1988), 297-317: "for certain works, particularly where multiple authoritative texts exist, a single version of the text is not sufficient for a scholarly edition; rather, significant texts should be offered as the disparate texts that they are" (p. 301). Edward Mendelson, in a piece entitled "The Fading Coal vs. The Gothic Cathedral or What to Do about an Author Both Forgetful and Deceased" (Text, 3 [1987], 409-416), distinguishes an interest in early versions from an interest in late versions by the two images of his title (but the "cathedral," despite its suggestiveness, is apparently not meant to imply collaborative effort). On the fundamental question of how to distinguish between versions to be presented separately and those (if any) to be incorporated into an eclectic text, see the references recorded in note 87 below.

[51]

See the editors' commentary volume, William Shakespeare: A Textual Companion, by Stanley Wells and Gary Taylor with John Jowett and William Montgomery (1987), pp. 16-19 (a general argument for Shakespearean revision) and p. 509 (an explanation of the "radical departure from traditional editorial practice" for Lear). (For assessments of this notable volume, see Peter Davison's review in the Library, 6th ser., 10 [1988], 255-267, and MacDonald P. Jackson's review in Shakespeare Survey, 41 [1989], 228-241.) Some of Taylor's statements in the general introduction—such as "Editorial controversy, like all other forms of discourse, is an instrument of power. . . . Editors are the pimps of discourse" (p. 7)— resemble those in his paper on "The Rhetoric of Textual Criticism" in Text, 4 (1988), 39-57, in which he links the "revisionist revolution in the editing of Shakespeare" to "a change in the rhetoric of textual criticism" (p. 53) and describes the "rhetorical strategy" that elevated the idea of two Lears from the status of an "iconoclastic heresy" to the position of being supported by an apparent "mass movement" (p. 46). He sets forth the revisionist position in "Revising Shakespeare," Text, 3 (1987), 285-304—where he sees it as "the new revisionist onslaught" against "the entrenched conflationist orthodoxy" (p. 302). A recent example of the revision theory in operation is John Jowett and Gary Taylor's "The Three Texts of 2 Henry IV," SB, 40 (1987), 31-50.

[52]

I have made these points about versions in SB, 34 (1981), 62-63 [104-105], note 75; SB, 39 (1986), 44 [152], note 85; and A Rationale of Textual Criticism (1989), pp. 79-82.

[53]

"'Versioning': The Presentation of Multiple Texts," in his Romantic Texts and Contexts (1987), pp. 167-180. The first part of this book, entitled "Romantic Texts" (pp. 15-180), brings together (along with the first appearance of "'Versioning'") nine pieces that had previously appeared elsewhere, seven of them reviews and essays relating to specific editions of nineteenth-century poets and the other two being the valuable two-part historical survey of the editing of the Romantics that he published in 1982 (Studies in Romanticism) and 1984 (Text). See also "Textual Criticism in Nineteenth-Century Studies," Nineteenth-Century Contexts, 11 (1987), 9-21; and "Gentlemen Authors and Professional Writers: Notes on the History of Editing Texts of the 18th and 19th Centuries," in Editing and Editors: A Retrospect, ed. Richard Landon (1988), pp. 99-136. Another recent piece of his is "Gender and Documentary Editing: A Diachronic Perspective," Text, 4 (1988), 351-359—where he confusingly links "slips of the pen" with "indifferent educational opportunities" as two causes of unintended readings in the texts of documents (p. 352).

[54]

Whether his preference for single-version editions is a matter of principle or convenience is not entirely clear. At several points he complains about the difficulty of using lists, and once he even says that a critic could "learn more" from comparing discrete versions than "by trying to unravel a conflated eclectic edition" (p. 173). (One can disregard, however, his statement in his prefatory note that he has found some eclectic editions to be "untrustworthy" [p. 167]: there are no doubt scholars capable of producing untrustworthy editions of single-version texts also.)

[55]

"The surviving pertinent materials would be before the reader and critic, the teacher and student, who could participate in the debate over issues that really matter until a new consensus emerged, rather than wait passively for a new canonical text to be imposed from 'above'" (p. 177).

[56]

Yet he seems to believe that the great volume of writing about critical editing has led to "some prejudice against the 'uncritical' presentation of primary manuscripts or early published versions" (p. 176).

[57]

One of his suggestions should be heeded: the publication of photofacsimiles of printed texts emended with paste-over alterations in an appropriate type face (pp. 178-179). In fact, this procedure is not limited to single-version editions; many eclectic texts produced by full critical editing would not incorporate emendations of such quantity or kind that a resetting of the copy-text would be necessary.

[58]

Library, 6th ser., 11 (1989), 89-115. A related brief essay of his, "Playwrights' Intentions and the Editing of Plays," Text, 4 (1988), 269-278, covers much of the same ground as the latter half of the Library essay. In a somewhat earlier article—"The Author as Scribe or Reviser? Middleton's Intentions in A Game at Chess," Text, 3 (1987), 305-318-he made some of the same points, but it was evidently written before he had developed the position on indifferent variants taken in the Library essay (cf. p. 318, note 18, of the 1987 piece with the comments on indifferent variants discussed below).

[59]

A few sentences later he says that it may also be unacceptable for other genres. One fault he finds with it is that it "interprets authorial intention primarily on the level of the document rather than on the level of the work" (p. 90). How can one locate the work except through the evidence of the texts of documents?

[60]

Howard-Hill obviously wishes Greg had decreed in such situations that the later text carries presumptive authority for indifferent variants (which would have left him closer to McKerrow). (I am using "indifferent variants" to mean those variants that seem equally balanced to an editor, taking everything into account. Howard-Hill distinguishes two uses of the term, one—"the usual interpretation"—referring to the "significance or aesthetic value" of the readings themselves (p. 98), the other referring to the authority of the sources of the readings. The two must ultimately coalesce, however, for any inference regarding the source of a reading, or inability to infer one, is influenced by a critical evaluation of the reading itself, in the light of what is known of the author's habits of expression.) Howard-Hill accepts at face value Greg's statement that "there can be no logical reason for giving preference to the copy-text" in the choice between indifferent variants, and he proceeds several times to call the practice illogical (as on pp. 97, 99, 100). Actually, the whole drift of Greg's argument provides the reasoning: the copy-text is chosen for its genealogical proximity to the text of the author's manuscript, on the grounds that texts deteriorate as they are recopied or reset in type; thus those readings (substantives and accidentals) in later texts that differ from copy-text readings may be presumed to be non-authorial, except when a case can be made that the variants result from authorial alteration or from repair (by someone) of the kind that the editor would have had to undertake anyway. One may disagree with this entire line of reasoning, but one cannot label as illogical the idea of retaining the copy-text readings in cases of indifferent variants, for it is a consistent part of the whole. In practice, one can decide in a given instance that the circumstances demand the conferring of presumptive authority on all the substantive variants of a later text (or certain categories of them); I am only trying to suggest why it is understandable that Greg did not state this position as a general rule. As for Greg's position leading—in the hands of some editors—to a "new 'tyranny of copy-text'" (p. 90), one can never hope to eliminate this danger (of which Greg was well aware), whatever one's approach. The art of critical editing has always centered on the delicate process of guarding against, on the one hand, excessive reliance on a favored text and, on the other, overconfidence in one's ability to improve it.

[61]

In his concluding paragraph, he states his "most important" recommendation in this way: "authorial intentions relinquished to the theatre by design and custom should be completed by an editor in accordance with his understanding of the author's intentions as reflected in surviving documents, and of the theatrical milieu in which the playwright wrote" (p. 115). This statement does not clearly distinguish two possibilities. Aspects of a play "relinquished to the theatre" were sometimes completed in ways that were not in accord with the playwright's intention; but in that form they became part of the performance text actually used. Howard-Hill earlier stressed the importance of the editorial adoption of such texts (as from promptbooks), because plays are collaborative products and playwrights expect changes to be made in their work. The emphasis in this statement, however, is on the editor's own emendations designed to produce a performance text in line with what can be known of the playwright's intention. A text reflecting the playwright's intention for performance is a very different editorial goal from a text reflecting the handling by theater personnel of details relinquished by the playwright.

[62]

A playwright who wishes to have a different text for the reading audience from one followed on the stage is of course creating two works, the one in print being like a novel (or a "closet drama") in employing only language as its medium. Even when the texts are the same, there may be features of the text, as in stage directions, that seem written primarily for the benefit of a reading audience; but as long as the text intended for the reading audience is the same as the text intended for performers, it is difficult to conclude that certain non-dialogue features of the text were meant to create a distinct work for the reading audience, since they also influence the performance of actors. The relations of reading text to performance are productively explored by Randall McLeod (writing as "Random Cloud") in "The Psychopathology of Everyday Art," The Elizabethan Theatre, 9 (1986), 100-168, which makes a similar point: McLeod argues against normalizing the irregular naming of characters in speech prefixes, for he believes that it (like punctuation) affects not only readers' responses but also actors' interpretations. (Cf. Philip Brett's discussion of the way in which the modernizing of musical notation affects performance, in "Text, Context, and the Early Music Editor," in Authenticity and Early Music: A Symposium, ed. Nicholas Kenyon [1988], pp. 83-114.)

[63]

Howard-Hill's argument is not helped by an eccentric use of the word "work," which does not in his definition encompass a play. He speaks of "the essential distinction between authors of works and playwrights" and explains, "The distinction I make is between an intention to publish by performance ('play'), and one to publish in print ('work')" (p. 108). My comments in this paragraph, and the definitions of "work" and "text" implicit in them, are derived from my A Rationale of Textual Criticism (1989), passim (on drama, see p. 85). (This paragraph and the next can also serve as my response to McKenzie's discussion of drama in his Panizzi Lectures, where he says, rather obscurely, that "the sociological dimension of production and reception . . . confirm[s] the textual nature of each element in a play" [p. 41].) My distinction between works and documents is entirely misunderstood by Margreta de Grazia in "What Is a Work? What Is a Document?", in The New Historicism and the Editing of English Renaissance Texts (photocopied papers, separately paginated, from the Renaissance English Text Society MLA panel organized by Thomas L. Berger, 1989): she believes that the traditional view (and mine) can be represented by such statements as "A work has an author, a document does not. A work is subject to critical interpretation, a document is not" (p. 1). It is no wonder that "this distinction is approaching collapse"—the "claim" of her paper, though a more challenging claim to prove would be that it ever existed. She somehow finds that attention to Shakespeare's plays as performed, along with the printing of multiple versions, shows an "indifference to a distinction ['the work/document distinction'] that was once basic" (p. 8); she does not see that every physical text is a documentary text and that every one of them is also an attempt to transmit a work. Another article, besides Howard-Hill's, that in my view fails to see the similarities between the texts of plays on paper and the tangible texts of works in other intangible media is John Glavin's "Bulgakov's Lizard and the Problem of the Playwright's Authority," Text, 4 (1988), 385-406. Glavin asserts that "the fusion of text and theatre . . . subverts the inscription that evokes it," and this "subversion radically fractures the relation between drama and other major literary genres" (p. 387)—as if fiction and poetry reach the public without intermediaries. Indeed, his statement that "The novel or poem is printed, read, and studied as written" seems oblivious to the fact that private and public texts of such works are so frequently different as to have given rise to the debate between intentionalist and social textual critics.

[64]

Even though a play exists as a work only in performance, historical research into the texts of plays—like playwrights' own composition—must rely on documents (i.e., on instructions for performance). Thus an editor seeking to reconstruct the text of a play as intended by its author for performance need not be bothered by the fact (or possibility) that the play was never actually performed in that version. The play (as a work) may not have existed, but plans for it existed in the playwright's mind and, in greater or lesser degree, in the form of text on paper. Those plans or instructions (in the form of text) are the reality that such an editor is attempting to recover.

[65]

For my own comments, see SB, 29 (1976), 183-191 (reprinted in Selected Studies in Bibliography [1979], pp. 325-333), and A Rationale of Textual Criticism (1989), pp. 77-78.

[66]

The idea that an interest in the autonomy of the author is "romantic" (which has become fairly common in anti-intentionalist writing on textual criticism) is challenged by Tilottama Rajan in "Is There a Romantic Ideology? Some Thoughts on Schleiermacher's Hermeneutic and Textual Criticism," Text, 4 (1988), 59-77: "My own contention is that not only is the ideology of the author as sovereign subject that subtends much textual criticism a modern reconstruction that finds only partial authority in the Romantic period but that the origins of the contemporary questioning of this authority initiated by McGann are also to be found in the Romantic period" (pp. 59-60).

[67]

"Unfinished Business," Text, 4 (1988), 1-11. During the period under review here, Bowers has given particular attention in print to the complex issues of regularization: "Readability and Regularization in Old-Spelling Texts of Shakespeare," Huntington Library Quarterly, 50 (1987), 199-227; "Regularization and Normalization in Modern Critical Texts," SB, 42 (1989), 79-102; "The Problem of Semi-Substantive Variants: An Example from the Shakespeare-Fletcher Henry VIII," SB, 43 (1990), 80-95.

[68]

"Textual Scholarship and General Theories of History and Literature," Text, 3 (1987), 1-9.

[69]

"Textual and Literary Theory: Redrawing the Matrix," SB, 42 (1989), 1-24. This essay can in some respects be supplemented by Greetham's "A Suspicion of Texts," a provocative introductory essay for the general reader published in Thesis: The Magazine of the Graduate School and University Center (City University of New York), 2, no. 1 (Fall 1987), 18-25. Near the end of the essay he suggests some of the ways that theoretical orientations reveal themselves in published editions, and his final line is that "a suspicion of texts is in fact one of the fundamental requirements of the critical mind."

[70]

"The Place of Fredson Bowers in Mediaeval Editing," PBSA, 82 (1988), 53-69. Greetham's interest in placing medieval textual problems in the context of current debates in textual criticism is also illustrated in "Challenges of Theory and Practice in the Editing of Hoccleve's Regement of Princes," in Manuscripts and Texts: Editorial Problems in Later Middle English Literature, ed. Derek Pearsall (1987), pp. 60-86. He offers here a descriptive list of seven "theoretical options . . . available to the editor," ordered in "declining degrees of 'fidelity' to the documentary state and history of the work" (p. 62). The list begins with "the photographic facsimile" and ends with (in sixth place) "the Slavic textological model, with its emphasis on the potentially equal status of all remaniements as textual witnesses" and (in seventh place) "a 'social' textual theory such as that envisaged by Jerome McGann" (p. 68). (One may wonder whether these last two are in fact at the farthest extreme from photographic facsimiles in terms of "'fidelity' to the documentary state and history of the work," for they both could be well—perhaps best—accommodated by the use of facsimiles. They are both at the opposite extreme from final authorial intention, but facsimiles do not necessarily represent authorial intention.) In the course of the essay he makes a statement that explains the form of a number of recent textual essays: "The conservative turned radical always seems to feel that his earlier conservative credentials need a greater demonstration of their frailty than do his new radical precepts, which can be regarded as articles of the revealed truth" (p. 70). Greetham here concludes that "a 'social' school of textual criticism is invalid where intention is recoverable" (p. 80); a detailed account of his method of reconstructing Hoccleve's intended text of Regement of Princes by incorporating into it Hoccleve's practice in accidentals, as established by analysis of holographs of other works, is provided in "Normalisation of Accidentals in Middle English Texts: The Paradox of Thomas Hoccleve," SB, 38 (1985), 121-150 (see p. 127, note 10, for his criticism of McGann's treatment of the relation between Lachmannian genealogical analysis and Gregian copy-text theory).

[71]

The papers from this conference and Greetham's response are to form a special issue of Critical Exchange (which, as of this writing, is not yet available). I am grateful to Professor Greetham for providing me with a copy of his commentary.

[72]

He adds that he finds support for his view from authors' statements of their own intentions—ignoring not only the difference between intention and expectation but also the historian's task of analyzing critically the motivations underlying statements made by individuals in the past. Willison's paper does in fact show the ubiquity, from medieval times to the present, of the split between authors of works and disseminators of texts. The papers from the Texas conference will be published in a special number of the Library Chronicle of the University of Texas (the collection is not yet available, as of this writing). I am grateful to Mr. Willison for giving me a copy of his paper.

[73]

Another historian of books and publishing, John Sutherland, has ridiculed the intentionalist approach and depicted those advocating the social approach as adversaries of a repressive establishment. See "Publishing History: A Hole at the Centre of Literary Sociology," Critical Inquiry, 14 (1988), 574-589; reprinted in Literature and Social Practice, ed. Philippe Desan, Priscilla Parkhurst Ferguson, and Wendy Griswold (1989), pp. 267-282. My response to this slight piece, which trivializes its subject by approaching the careers of McGann, McKenzie, and Robert Darnton in terms of academic politics, appears in Literature and Social Practice, pp. 283-287 ("Response to John Sutherland").

[74]

See Stephen Knight, "Textual Variants: Textual Variance," Southern Review [Adelaide], 16 (1983), 44-54; Harold Love, "Sir Walter Greg and the Chaucerian Force Field," Bibliographical Society of Australia and New Zealand Bulletin, 8 (1984), 73-81; Stephanie Trigg, "The Politics of Editing Medieval Texts: Knight's Quest and Love's Complaint," ibid., 9 (1985), 15-22.

[75]

He describes Knight's "distortions of Greg" as "a classical example of a Bloomian tessera—a 'strong misreading' of an anxiety-causing precursor designed to create imaginative space for the textual ephebe" (p. 76). More usefully, he points out that a social goal for textual criticism is "a matter of editorial philosophy which can be applied to any work of literature whatsoever" and is "not necessarily dependent on the work being the product of collective composition" (p. 79). But he muddles this point by asserting that, in choosing between indifferent variants, "Knight's principle is certainly no less logical than the idea that one should follow the reading of the 'best manuscript' or Greg's recommendation that one should accept the reading of the copy-text" (p. 80). If one can apply social textual criticism to any work, as of course one can, then one's method of handling indifferent variants must be consistent with it, and Knight's is more appropriate for this purpose than Greg's. No doubt Love had this point in mind, but what he says does not make entirely clear that social and authorial emphases cannot sensibly be mixed in a single edited text—and thus he does not elucidate as much as he might the very real problem, in Knight's piece, of allowing authorial intention to guide some, but not all, editorial decisions. Love weakens his argument even further by suggesting at the end that an edition following Knight's rationale would be "a twentieth-century artefact, not a medieval one" (p. 80). Although he understands that all editors, including "traditional" ones, bring their own "ideologies" to their work, he does not proceed to recognize that texts constructed to reflect authors' intentions and those constructed to recreate the "dialectical tensions of the period of origin" are equally the products of the historical imagination of scholars living and working at particular times and inescapably imbued with the sensibility of those times.

[76]

She believes that "medieval texts, far from being the innocent victims of superimposed critical theories, or coloured by different editorial practices, are only produced as we read them" (p. 21).

[77]

"An Inquiry into the Social Status of Texts and Modes of Textual Criticism, SB, 42 (1989), 55-79. Shillingsburg says that "this survey of competing views of editing has convinced me even more that editing is a critical enterprise that not only involves criticism but is in fact a form of literary criticism." Like others before him who have made this discovery, he adds, "I believe no theorist should say that his method is the only responsible one, though I think it is possible to discover that some methods are irresponsible" (p. 74).

[78]

"Editorial Theory and the Act of Submission," PBSA, 83 (1989), 169-185. West concludes that an intentionalist editor can accept some of the alterations made in a text by persons other than the author if those alterations seem to reflect fulfillment of the author's active intentions (and the resulting edition is both "an act of literary criticism and of biography" [p. 185]). He regards this position as intermediate between the strict intentionalist view, which would not allow any alterations not made by the author, and the social view, which would favor the reproduction of the received text accompanied by notes of "who did what during the compositional process" (p. 185). In this scheme, West believes that the two extremes are "rigid," whereas his recommended intermediate approach "increases the element of critical thinking in the creation of a scholarly text" (pp. 184-185). Some "rigid" intentionalist editors probably do exist, but being strictly intentionalist requires judgment, not rigidity. Any attempt to construct an authorially intended text demands "critical thinking," and the decision to accord a delegated authority to some of the changes made by authors' personal editors or publishers' editors does not necessitate a greater infusion of it, but rather an application of it toward somewhat differently defined ends. West has not proposed a new approach to critical editing but only another perspective on where the line between intention and expectation falls.

[79]

"Neopragmatism and Convention in Textual Editing, with Examples from the Editing of Thoreau's Autograph Journal," Analytical & Enumerative Bibliography, n.s., 1 (1987), 227-236 (quotation from p. 230). Neufeldt speaks for many other editors when he says, "Recent and current speculation about literary theory, theory of language, theory of discourse . . . has threatened a number of models that formerly offered a satisfactory starting point, modus operandi, and conceptual focus for editing texts. . . . Our eyes—including our editorial eyes—are being retrained" (p. 227).

[80]

"New, Old, Anglo-American, Textual Criticism," PBSA, 80 (1986), 243-253 (quotation from p. 245)—a review of William Proctor Williams and Craig S. Abbott's An Introduction to Bibliographical and Textual Studies (1985). This often acute review misleadingly depicts the "Anglo-American tradition of textual criticism" as an "immoveable orthodoxy" resulting in "institutional and theoretical certainties" (p. 244)—ignoring, among other things, the emphasis by many (perhaps most) of its practitioners on the lack of definitiveness of its products. In contrast, he argues, "McKenzie's call for a deeper attention to the materiality of the text and McGann's appeal over the heads of Greg-Bowers to the precedents of classical scholarship both entail a welcome return from the tyranny of method to ratio et res ipsa" (p. 250).

[81]

See, for example, Gerald M. Maclean, "What Is a Restoration Poem? Editing a Discourse, Not an Author," Text, 3 (1987), 319-346; David S. Hewitt, "Scott and Textual Multiplepoinding," Text, 4 (1988), 361-373; Herman J. Saatkamp, Jr., "Final Intentions, Social Context, and Santayana's Autobiography," Text, 4 (1988), 93-108; and two essays by Arthur F. Marotti, "Malleable and Fixed Texts: Manuscript and Printed Miscellanies and the Tranmission of Lyric Poetry in the English Renaissance," in Is the Typography Textual? (photocopied papers, separately paginated, from the Renaissance English Text Society MLA panel organized by Carolyn Kent, 1988), and "Manuscript, Print, and the English Renaissance Lyric," in The New Historicism and the Editing of English Renaissance Texts (see note 63 above). Each of these essays refers to McGann on its first page and gives considerable attention to a socialized theory of textual criticism; none of the writers, however, makes the point that both the authorial and the social approaches are independently applicable to every work. (Marotti comes close to doing so in the second paper but wishes—as he says in the first—to "dissociate" his work from "the usual textual and bibliographical program . . . that is informed by a textual 'idealism' that effectively eradicates those interesting socioliterary processes in which texts are historically embedded" [p. 1]. Saatkamp appends to his essay the naïve recommendation that "critical authorial editions" of living authors be undertaken, allowing the authors to determine their intended texts—as if the existence of such editions would alter in any way the scholar's task of assessing authorial motivation and reconstructing the texts intended at particular past moments.) These issues have also become prominent in the study of other arts. In music, for example, attention has been given to "authenticity," meaning fidelity not to composers' intentions but to the details of contemporary performance; see Authenticity and Early Music, which includes Philip Brett's excellent essay (see note 62 above) surveying the history and issues of music editing in the context of the theoretical debates among literary editors.

[82]

"The Worldly Text: Writing as Social Action, Reading as Historical Reconstruction," in Literary Theory's Future(s), ed. Joseph Natoli (1989), pp. 181-220. The section on textual criticism, "Beginning with a Text" (pp. 190-198), deals largely with Hershel Parker, Jerome McGann, and me.

[83]

In "Some Considerations in the Concept of Pre-Copy-Text," Text, 4 (1988), 79-91. Cook has also written, during the period under review here, a skillful account of "Preparing Scholarly Editions" for a general audience: see Humanities [National Endowment for the Humanities], 9, no. 3 (May/June 1988), 14-17. (His comment on apparatus in this piece is simply that an editor should record "fully the evidence on which the editorial decisions are made, including all the variant readings.")

[84]

Another reason, which Cook cites from the Indiana edition of Howells, is that serious scholars will wish to examine the manuscripts themselves; but this argument ignores both the usefulness of genetic information to a wide variety of readers and the scholarly contribution made by editors (experts in their authors' handwriting and working habits) in deciphering documentary texts, which are often complex or unclear.

[85]

Another scholar critical of standard apparatus is Barry Gaines, though the specific target of his attack is very different: in "Textual Apparatus—Rationale and Audience," in Play-Texts in Old Spelling, ed. G. B. Shand with Raymond C. Shady (1984), pp. 65-71, he contends that "no one is really interested in reconstructing the copy-text from the apparatus which accompanies a critical edition" and that, indeed, anyone would be foolish to try to do so when microfilm and other facsimile copies are available (p. 68); and the historical collation, he feels, is "simply a record of what the editor has endured to earn the right to proceed with the edition" (p. 69). The ineffectiveness of his argument is suggested by this false analogy: "When scientists publish their conclusions, they are not asked to include as an appendix all their worksheets" (p. 69).

[86]

"Rational Presentation of Multiple Textual Traditions," PBSA, 83 (1989), 43-60.

[87]

See, for example, my "The Editorial Problem of Final Authorial Intention," SB, 29 (1976), 167-211 (reprinted in Selected Studies in Bibliography [1979], pp. 309-353), especially the third section and the references cited there. An essentially similar approach to the one I proposed in some detail has been sketched in broad terms by Giovanni Aquilecchia in "Trilemma of Textual Criticism (Author's Alterations, Different Versions, Autonomous Works): An Italian View," in Book Production and Letters in the Western European Renaissance: Essays in Honour of Conor Fahy, ed. Anna Laura Lepschy, John Took, and Dennis E. Rhodes (1986), pp. 1-6: "I would maintain," he says, "that it is part of the critical editor's task to distinguish . . . between mere alterations which do not affect the structure of the work or a part of it on the one hand (to be recorded in the apparatus of a critical edition) and different versions of the work or part of it, sometimes amounting to different works altogether on the other" (p. 4). Fredson Bowers pursues this question further in "Mixed Texts and Multiple Authority," Text, 3 (1987), 63-90. After distinguishing "mixed authority" in a single-text tradition (produced by authorial revision) from "discrete multiple authority" (produced by thorough rewriting), the former amenable to eclectic editing and the latter not, he examines situations with "radiating multiple authority" (p. 74), compatible with eclectic editing but posing special problems because of the multiple "sets of independent documentary evidence," often of equal authority (p. 75). This essay supplements his landmark essay on "Multiple Authority: New Problems and Concepts of Copy-Text," Library, 5th ser., 27 (1972), 81-118 (reprinted in Essays in Bibliography, Text, and Editing [1975], pp. 447-487). On authority, see also his "Authority, Copy, and Transmission in Shakespeare's Texts," in Shakespeare Study Today, ed. Georgianna Ziegler (1986), pp. 7-36.

[88]

See "The Synchrony and Diachrony of Texts" (note 40 above); quotations from p. 311.

[89]

Some genetic critics may wish to present facsimiles and transcriptions of any or all of the documents that preserve such revisions, simply because any text as it stands in a document is of interest, for it is what exists in physical form. But in these cases a consolidated presentation, showing the development of the text as reflected in a series of documents, is also always in order.

[90]

And that theorists of reading and writing call "recursion," as Robert Murray Davis has pointed out (see note 49 above). This view of reading does not in itself make the contents of books a visual art, despite the similarity of the process to the way our eyes roam over a painting or a sculpture. The way in which we study the physical object containing a verbal text does not alter the status of the text: it may be a set of instructions for the recreation of a work of language, or it may be a visual work as well; but its classification in this regard is not determined by readers' techniques for perusing it.

[91]

An argument can be made, for example, that appended apparatuses have a particular advantage for texts intended for publication in that they allow for clear texts, and published texts are characteristically clear. The texts of private documents, on the other hand, are typically rough, with canceled and alternative readings; and one can argue that editions of individual private documents are especially well served by inclusive apparatuses, which retain the roughness in the linear text. This theoretical consideration no longer applies, however, when (as in Gabler's Ulysses) the readings of more than one private document are brought together in a single record. In such cases an inclusive apparatus produces a composite text, not a literal rendering of an individual documentary text; its primary purpose is thus to serve as a record, not to present a particular physical arrangement of words and punctuation, and the decision to use an inclusive rather than an appended apparatus to accomplish this purpose is a practical, not a theoretical, one. (I have made the argument to use appended apparatus for the texts of works intended for publication and inclusive apparatus for the texts of individual documents not intended for publication in, among other places, "Some Principles for Editorial Apparatus," SB, 25 [1972], 41-88 [esp. pp. 46-47]; reprinted in Selected Studies in Bibliography [1979], pp. 403-450 [see pp. 408-409]. I have also suggested, more than once, that practical considerations in certain situations may alter this recommendation, as in "Literary Editing," in Literary & Historical Editing, ed. George L. Vogt and John Bush Jones [1981], pp. 35-56 [esp. pp. 44-45].) Albert J. von Frank, in "Genetic versus Clear Texts: Reading and Writing Emerson," Documentary Editing, 9, no. 4 (December 1987), 5-9, takes a practical point of view in examining these alternatives for individual documents and advocates the production, on the computer, of both forms of apparatus. Robert H. Hirst, in Mark Twain's Letters, ed. Edgar Marquess Branch et al. (1988- ), has developed a "plain text" system in which inclusive apparatus is used for some details and appended apparatus for others "in order to make the text as complete and informative as possible without destroying its legibility" (1, xlvi). Peter L. Shillingsburg (who believes, with Gabler, that authorial variants are part of the totality of a work but who does not object to appended apparatus) has addressed the separate question of indicating (or, perhaps one should say, nominating) the agent responsible for each variant: see his Scholarly Editing in the Computer Age: Theory and Practice (1986), pp. 111-113, which I have discussed (in its 1984 edition) in SB, 39 (1986), 40-41 [148-149].

[92]

See note 67 above. Bowers's basic statement of the requirements for an apparatus is "Notes on Editorial Apparatus," in Historical & Editorial Studies in Medieval & Early Modern English, for Johann Gerritsen, ed. Mary-Jo Arn and Hanneke Wirtjes, with Hans Jensen (1985), pp. 147-162—cited in my previous survey along with his earlier exchange with Paul Werstine (SB, 39 [1986], 37 [145], note 69). Bowers's most detailed examination of the apparatus for recording manuscript variants is "Transcription of Manuscripts: The Record of Variants," SB, 29 (1976), 212-264, which emphasizes clear text with appended apparatus but proposes a system of transcription equally applicable to inclusive apparatus.

[93]

This address is to be published in the fifth volume of Text. I am grateful to Professor Boydston for allowing me to have a copy of the paper as delivered.

[94]

Such texts are precisely what the Library of America series is producing in those instances in which the text of an already existing scholarly critical edition is not used; but the aim in so doing is not an emphasis on publishers over authors. The series is committed to texts that already exist (critically edited only to the extent of correcting typographical errors); but within that constraint it tries to choose the texts that best (if, inevitably, not fully) represent the auhors' intentions. Correcting typographical errors (which are always listed in the Library volumes) does bring published texts closer to their publishers' intentions—but of course to their authors' intentions as well.

[95]

Every surviving manuscript book from the pre-print period is a social product, but many manuscript books that contained significant texts no longer survive. A standard procedure among editors of early manuscripts, when faced with multiple texts of particular works, is to attempt the reconstruction of the common ancestor of the surviving texts. The result of this process is a critical text recreating a social product, not an authorially intended text (though the editors of such texts have assumed that they were moving in the direction of an authorial text, as some of their critical judgments in the process of recension make clear). For works primarily transmitted in printed form, there is less occasion for reconstructing socially produced texts; but when no copies of a particular edition are known, it might sometimes be feasible, from the evidence of earlier or later editions or manuscripts, to attempt reconstructing it.