University of Virginia Library

Recent Editorial Discussion and the Central Questions of Editing
by
G. Thomas Tanselle

The late 1970s were an extraordinarily active and interesting period for those concerned with editorial matters. In addition to the continuing stream of scholarly editions, these years saw the beginning (1976) of a Center for Scholarly Editions (CSE) within the Modern Language Association of America, the inauguration of an Association for Documentary Editing (1978) and of an interdisciplinary and international Society for Textual Scholarship (1979), and the founding (1979) of a nonprofit corporation, Literary Classics of the United States, Inc., for publishing reliable texts in a form easily accessible to the general reading public. What these organizations have in common, aside from their interest in editing, is a breadth of vision. The fact that the CSE is the successor to the Center for Editions of American Authors (CEAA) is symbolic: whereas the CEAA was limited to the consideration of editing projects in American literature, the CSE is prepared to consult with editors of any kind of material from any country; in its 1977 Introductory Statement [1] it emphasized that editors of diverse works "have a common ground for coming together" and pledged "to promote greater understanding among editors in all fields." The Association for Documentary Editing—though the original impetus for its organization came from historians editing the papers of statesmen—welcomes as members editors from all disciplines, and editors of literary and philosophical works have been active in it.[2] Similarly, Literary Classics of the United States, although it is committed to publishing works that can be regarded as American, is not limiting its purview to works that are "literary" in a narrow sense, and it recognizes that a responsible

textual policy can accommodate itself to a variety of kinds of editions.[3]

This climate of openness and cooperation is highly desirable; but its increasing presence should not be taken as a sign that the editorial issues argued about with considerable vehemence during the previous fifteen years have been settled. The familiar debates—such as whether Greg's rationale is appropriate for post-Renaissance material or whether the CEAA was too rigid in its standards—have continued and have often been as ill-informed as before. It would be wrong to suggest that the late 1970s did not have their share of unpleasant and fruitless editorial controversy. But I think it can also be said that the discussion began to take on a somewhat different aspect during these years: arguments based essentially on emotional reactions did not as often hold center stage, and somewhat more commentary appeared that raised thoughtful questions about central issues. Tom Davis, in an important review of the CEAA enterprise,[4] was being overly generous when he referred to "the sophistication of the scholarly debate" that the CEAA editions generated; but there is no doubt about the sophistication of his own piece, and it is a prime example of the newer, more serious criticism of the CEAA.

When in 1975 I surveyed the editorial literature that had grown up around the CEAA,[5] I noted that much of it seemed to have arisen from conflicts of personality or temperament. Quantitatively these writings may have served a purpose, directing more attention than is customary to the activity of scholarly editing (though also reinforcing the view that editors are a contentious lot); but qualitatively this literature left much to be desired. The time now seeems appropriate to extend the survey through the remainder of the 1970s, not only to continue the record (for the development of the debate is of interest in its own right as an episode in the history of modern scholarship) but also to see if an examination of this literature cannot serve to identify and clarify certain basic issues of editing. Some of the pieces require little attention, for they are simply restatements of points of view that I have commented on in the earlier survey; but others, even if they are sometimes inept or illogical, touch on fundamental questions that are worth exploring further. The process of working through these discussions can, I trust, prove to be a fruitful way of approaching those questions. Although the specific subjects are

frequently the CEAA or Greg's rationale, the issues obviously go beyond American literature and Greg; many misunderstandings have come about through a failure at the outset to be clear about what these underlying issues are, through a failure to make certain elementary distinctions. One of the unfortunate effects of a protracted controversy can be to envelop the issues with a greater aura of mysterious complexity than they actually possess. Editors have difficult enough decisions to make in the process of producing critical editions without needlessly complicating the conceptual framework within which their activity must take place: deciding between two variant readings, for instance, may indeed be a complex affair, requiring an involved discussion, but explaining what one hopes to accomplish by making such decisions should not be a difficult task. A look at some recent editorial discussions can lead, I think, to a realization of how simply the basic questions can be framed.[6]

I

The additions during 1974 and 1975 to the debate inspired by the CEAA editions contribute little that pushes forward one's understanding and do not require any extended consideration. They are of some interest, however, in illustrating how the misconceptions and misunderstandings of the earlier discussions continue to appear and how in fact certain ill-considered arguments become almost stock responses, destined to come up whenever the CEAA is mentioned. Morse Peckham, for instance, delivered a curious paper on "The Editing of 19th-Century Texts" at a symposium on "The Nature of Linguistic Evidence" at the University of Michigan in March 1974.[7] This paper is partly concerned with repeating certain ideas in his 1971 essay, "Reflections on the Foundations of Modern Textual Editing,"[8] such as the view that "an author correcting his own work has the same relation to the text that an editor has" (p. 124). That earlier essay deserves serious attention for its analysis of the nature of written communication, but the aspect of it repeated here—and repeated without any new supporting arguments—is the weakest part of the whole. What the more recent paper emphasizes, however,

is the "randomness" in verbal behavior and textual transmission and the problem such randomness poses for editors.[9] Because of the possibilities for variation at every stage of the transmissional process and because of the dispersal of the evidence through hundreds or thousands of copies of individual impressions of printed editions, editors cannot—as Peckham correctly observes—be in a position to know when they have seen all the evidence. But this situation need not fill one with the degree of pessimism that Peckham exhibits; his reaction is seemingly colored by his own dissatisfaction with and resignation from the Ohio University Browning edition, an event that occurred a few months before his paper and that is alluded to in it.[10] He is right to suggest that a more sophisticated statistical approach be taken to the search for variants in nineteenth-century books; but to regard CEAA texts as failures because of what he calls a "policy" of collating[11] only five or six copies of individual editions (p. 133), or to conclude that "reliable editions cannot be created" (p. 142), is to overreact to the unfortunate truth that inductive investigations are never certainly "final." To be sure, some CEAA editors have from time to time been guilty of exaggerated rhetoric, and presumably that is what Peckham is referring to when he speaks of "unjustified claims about completeness or definitiveness" and describes editors "boasting what a superb and thorough-going job they have done" (p. 136); but those editors at the same time have normally been careful to put on record just what their research consisted of and surely did not fail to understand that additional information might later turn up that would alter their conclusions. When Peckham says at the end that "the current way of making editions creates a closed or dead-end situation which requires that the effort be repeated, and must fail again," his use of the word "fail" suggests that a work is not successful if it is superseded. But of course the accumulation of knowledge proceeds precisely by building on what went before, and it is naïve to think of scholarship in terms of finality.[12] There is no question that some editors are more thorough than others in their procedures, and much of what Peckham says about the importance of recognizing the human element in printing and the need for more awareness of statistical procedures in analytical bibliography is well taken. But his insistence on attacking the supposed claims of definitiveness in CEAA editions only serves to confuse the issues and weakens what could have become a helpful positive statement about the responsible handling of evidence. This is not the first, nor the last, time that the issue of the "definitiveness" of CEAA editions has diverted attention from more basic and substantial concerns.[13]

Peckham's paper—valedictory since he announces in it his resignation from the Browning edition—can be paired with another valedictory essay of the same year, Paul Baender's "Reflections upon the CEAA by a Departing Editor."[14] Baender in fact comments on Peckham's paper and makes an effective rejoinder to parts of it (see pp. 136-139). Although Baender is a "departing editor" (from the University of California Press edition of Mark Twain) because he finds himself not temperamentally suited to editorial work, he believes that the CEAA editions are "eminently worth doing" (p. 141) and places in proper perspective some of the superficial criticisms they have received. In the light of his effective exposure of certain "clichés" that continually turn up in writings critical of the CEAA,[15] it is surprising that his own criticisms also include some arguments that are clichés. For example, he objects to the "trivialities" and "excesses of trivia" that he finds recorded in some CEAA editions and suggests that if people had "taken the outrage of Mumford and Wilson seriously" the proliferation of "meaningless data" would not have occurred (p. 135).[16] To protest about the quantity of data recorded in CEAA editions has certainly become one of the clichés of criticism. But it is difficult to fathom why people become so incensed over the presence of information they do not find of use. Perhaps some editors are overly generous in supplying evidence they have considered in the

process of preparing an edition; but readers who are not interested in the information are free to ignore it. Criticizing an edition for containing too much so-called "trivial" information is itself a trivial complaint. Baender views this criticism more sensibly later when he says that "pedantic excesses" are "relatively minor concomitants of the editorial process" (pp. 143-144)—but at the same time he unexpectedly admits, "I too find little importance in word-division lists." These lists have been a favorite target for unreflective critics, but it is hard to see how anyone who has been involved in establishing a text could fail to understand their significance. I do not suppose that any editor regards them as particularly appealing, but editors nevertheless recognize that these lists play an essential role in the task that an edition aims to accomplish: when an editor has taken pains to establish a text—punctuation as well as words—it is illogical not to inform readers about which line-end hyphens should be retained in making quotations from the text.

Baender's essay in other respects, as in this one, does not advance the frontiers of editorial debate; but some of the questions he raises should perhaps be enumerated as further examples of his own category of "clichés." For instance, he blames the CEAA inspectors for some of the "trivia" in the published editions, noting with apparent distaste that "Generally the inspectors are greatly concerned with accidentals" (p. 135) and referring to "the inspectors' nitpicking" that causes editors "to worry over the details of their textual descriptions" (p. 136). Though some inspectors have doubtless been guilty of excesses, to regard the questioning of details as "nitpicking" suggests a lack of sympathy with the kind of attention to detail that scholarly editing necessarily involves.[17] Of more substance is Baender's dissatisfaction with the term "final intention"; but his brief discussion is no more than a gesture, for he ignores the interesting aspects of the question by taking "final" to mean "last," without regard to motive (he claims that if certain of Mark Twain's revisions of The Innocents Abroad for the Routledge edition are rejected as adaptations for a British audience, "the criterion of final textual intention has been compromised" [p. 140]). On another important aspect of intention—the relation to authorial intention of revisions dictated by the publisher—he makes an admirable statement ("an individual has the power and privilege of self-expression and of changing his mind," and "other individuals do not have the privilege of altering that self-expression

or of forcing that change of mind"); but when he adds, "I believe the CEAA has not sufficiently recognized that a writer's acquiescence in his publisher's alterations may also be construed as self-expression" (p. 141), he raises—but does not explore—what has become one of the most ubiquitous issues in discussions of modern editing. Baender's piece is essentially thoughtful, and for that reason certain stock reactions to the CEAA stand out in it all the more flagrantly; their presence in such an essay suggests how pervasive they have become.

Another paper of 1974 written by a person generally sympathetic to the CEAA but at the same time exhibiting superficial criticisms of it is Joseph Katz's paper for the November 1974 Toronto Conference on Editorial Problems. Its title announces that Katz's theme is the endlessly asserted "rigidity" of CEAA policy: "'Novelists of the Future': Animadversions against the Rigidity of Current Theory in the Editing of Nineteenth-Century American Writers."[18] Katz offers a useful discussion of some of the textual problems in editing Frank Norris; but when he tries to show that these problems "strain a strict application of current editorial approach" (p. 69), he resorts to several of the old clichés. He is particularly concerned about the notion of a "definitive" edition, asserting that the paucity of prepublication material for Norris's novels makes a "definitive edition" impossible (pp. 70-71, 74) and stating that he cannot conceive of a "definitive edition" of any work: "Every edition with which I am familiar could be upset completely by the emergence of some document not available to the editor" (p. 67). But surely no scholar imagines that a piece of work can be produced that could not be overturned by the discovery of new documentary evidence. To claim that CEAA policy requires editors to undertake a task that everyone agrees is manifestly impossible is merely to set up a straw man. Katz does say that "the problem is in part a matter of rhetoric"; but when he adds, "so are most editorial problems," one wonders how to take other parts of his own essay. Like his comments on definitiveness, his discussion of "author's final intentions" is in fact a rhetorical exercise and does not get to the interesting aspects of the issue, but it appears to have been written under the impression that a matter of substance was being taken up. What bothers him about the phrase "author's final intentions," he says, is the assumption "that there is always and invariably a single work" (p. 67)—"that an author himself always and invariably pursues a work until it has reached its one ideal state—or that he wants to, or that he ought to want to" (pp. 74-75). It seems unlikely, however, that many people have ever held such an assumption. Certainly Greg himself and those who

have in general followed his rationale have recognized that various "final intentions" might exist with respect to different versions of a work that deserve to be taken as independent entities.[19] By using his space on a non-issue, Katz has neglected to look into any of the intriguing questions that the concept of intention raises, such as how one decides when to treat a version of a work as if it were a separate work. Similarly, he begins with the often-repeated, but insubstantial, criticism that current editorial thinking assumes "the existence of a proven editorial routine which will produce definitive texts of works written at any time, in any place";[20] and at the end he makes the point in an equally familiar way by saying that "the problems faced by scholars concerned mainly with editing dramatic texts of the English Renaissance are not the problems that confront the editor of a nineteenth-century American text." These assertions are again matters of rhetoric and skirt the significant issues. Obviously there are differences of detail among textual situations from different periods, but there are some underlying central questions of editing that remain constant and that must be thought about before one can responsibly proceed to those details. And to claim that "current theory" entails a rigid routine is not only to confuse theory and practice but also to miss the flexibility built into the basic CEAA statements and exhibited in the various CEAA editions. This is not necessarily to argue on behalf of the CEAA editions but only to point out that Katz's paper does not go past certain stock responses to consider the real issues that are involved.

A paper that deals more seriously—but in the end no more success-fully—with problems raised by the concept of authorial intention is Hans Zeller's "A New Approach to the Critical Constitution of Literary Texts," which appeared in printed form in early 1975.[21] This paper serves as a convenient summary for English-speaking readers of a position that has been gaining acceptance in Germany. Zeller's argument essentially

is that any authorial alteration of a work creates a new "version" of the work and that each version is a independent entity which should not be emended by an editor with authorial variants from another version. This position, which limits the critical role of the editor to the detection and correction of errors within each version, is said to rest on the view that any alteration of a work produces a different relationship among all the elements of the work and on the belief that an author's artistic intentions cannot be disentangled from various practical and worldly concerns. Both these points have much to be said for them, but they do not necessarily lead to the conclusion Zeller draws. One can argue that the editor is in a position to attempt to discriminate among kinds of revisions and among various motivations for revision and that not to attempt such discriminations is to ignore many of the nuances of the textual situation, the result being an oversimplified picture. Presumably no one would deny the value of editions of individual versions of a work, but a more eclectic approach also has its particular merits. Zeller's position deserves to be considered as part of the recurrent challenge to the soundness of the critical editing that has followed from Greg's rationale;[22] his arguments, however, do not finally call into serious question the usefulness of an approach that allows an editor, as an act of critical scholarship, to make judgments not simply about the correction of error but also about the emendation of readings revised by the author.[23]

Later in 1975 Studies in the Novel devoted a special number, edited by Warner Barnes and James T. Cox, to "Textual Studies in the Novel" (vol. 7, no. 3). Among other contributions,[24] this collection includes an article by John Freehafer, "Greg's Theory of Copy-Text and the Textual Criticism in the CEAA Editions" (pp. 375-388), which the editors in their preface describe as singling out "for specific (and often harsh) criticism many of the practical and theoretical problems inherent in the CEAA editions" (p. 319). Freehafer does make a few valid criticisms and helpful suggestions—such as lamenting the customary omission in CEAA volumes of a historical record of variants in punctuation and spelling, and pointing out the desirability of making available facsimiles of significant documents in the textual history of particular works—but these observations occur in the midst of many others that misrepresent what Greg's rationale entails or repeat certain well-rehearsed but ill-informed objections to it. His expression of surprise that two editors following Greg but choosing different copy-texts of Dryden's The Indian Emperour should come up with texts that vary at a great many points (p. 377) indicates the level of his response to Greg's rationale. He takes the divergence as an indication that the editors have adhered too mechanically to their copy-texts instead of recognizing that an expectation of identical results presupposes a precision alien to any procedure relying on critical judgment. It is no criticism of Greg's rationale, of course, to say that it has been misused by some editors as a means for avoiding judgment. "The popularity of Greg's theory of copy-text as it has thus come to be applied," Freehafer claims, "may be partly because it often relieves the textual editor of the difficult task of trying to recognize authorial corrections, afterthoughts, or revisions, especially in accidentals" (p. 382). Freehafer does recognize Greg's emphasis on critical judgment, and the phrase "as it has thus come to be applied" makes this sentence tautological. Whether or not CEAA editors have in fact avoided making textual decisions is another matter, and one not related to Greg's rationale. The old question of the applicability of Greg's approach to nineteenth-century texts is one that Freehafer is inclined to answer in the negative, but his reasoning again refers to an inflexibility that is not part of Greg's

own statement. The CEAA editor, he says, faced with series of authorially revised editions that are common in the nineteenth century, is likely to choose an early text "as a source of readings more often than he should" (p. 379); but Greg made allowance for selecting revised editions as copy-texts when circumstances warrant. Freehafer's piece requires little discussion and in any case has already been sufficiently analyzed in "The Center for Editions of American Authors: A Forum on Its Editions and Practices," which immediately follows his essay (pp. 389-406).[25]

Two more essays of 1975 represent extremes, one being less important and the other more important than any of the discussions thus far examined. What is probably the most uninformed and irresponsible criticism of the CEAA, Peter Shaw's "The American Heritage and Its Guardians," appeared in late 1975 in the American Scholar.[26] Because of its publication in a general journal, Shaw's article has perhaps reached a wider audience than any of these others. No doubt some readers without any background in textual matters have been persuaded by it; but anyone acquainted with the CEAA editions and the recent series of editions of the papers of American statesmen knows how far off the mark Shaw is when he claims that the editions of the statesmen generally show more fidelity to the historical documents than do the "eclectic" editions of literary figures. Whether or not one approves of the CEAA editions or of Greg's rationale, one can see that Shaw misunderstands the "eclecticism" of modern critical editing and does not recognize why the unrecorded alterations that occur in many of the historical editions constitute a weakness. The article is full of confusions and repeats many of the clichés about the CEAA without adding significantly to the discussion. Shaw was simply unprepared to write such a piece, and it can safely be dismissed.[27]

At the other extreme is an essay by Fredson Bowers, "Remarks on Eclectic Texts,"[28] published at about the same time. This long discussion is an essay in definition: it explores, with many illustrations, what is meant by eclecticism in editing and shows the scholarly significance of critical editions. As a reaction against the unprincipled eclecticism of some editors in previous centuries, there has been a twentieth-century current of distrust of eclecticism. This distrust has taken many forms, as the essays of Zeller and Shaw suggest. Thus Bowers's discussion goes to the heart of one of the principal issues raised by critics of the CEAA editions. After examining in detail the application to various kinds of situations of an approach that aims to establish a text that will best represent its author's intentions, he arrives at an eloquent statement summarizing why the results of this approach "should constitute the finest flower of textual scholarship": "the main scholarly demand is for an established critical text embodying the author's full intentions (not merely one segment of them in an inevitably imperfect form) insofar as these can be ascertained by an expert who has had available all documentary sources and has devoted time and study to their transmissional history and authority" (p. 527). Bowers recognizes that facsimile editions have their uses and understands that in some cases two versions of a work are so different as to make any attempt to construct a single eclectic text inappropriate. But he shows that normally the expertise of the editor is not put to its highest use if it does not result in a single critical text: "Literary critics, historians, general scholars, students of all kinds—these need as authoritative a reconstruction of a full text as the documents allow, not editions of the separate documents, except when the distance is so great as to make eclectic reconstruction impossible" (p. 528).[29] Bowers's careful and thorough discussion will now be the standard reference to cite when questions are raised about the meaning or value of eclecticism in modern scholarly editing.[30]

Another essay of Bowers's, the following year, deserves mention here both because it provides a convenient and thoughtful general summary of the approach to editing that derives from Greg and because it opens the way to the serious reconsideration of Greg that was to occur in the following years. Delivered in an abridged form before the Bibliographical Society of America in 1976, "Scholarship and Editing"[31] principally offers an authoritative statement regarding modernization, eclecticism, radiating texts, choice of copy-text (Greg's rationale),[32] and apparatus. But it also comments on "exceptions to Greg's rationale" in which "the author can be shown to have devoted as much attention [in a revision] to the formal features of his text as to its material or substantive features" (p. 179). Such texts "appear to call for a relaxation of Greg's principle," he says, "not at all for the reasons adduced for Ben Jonson [i.e., the thoroughness of the substantive revision in Every Man in His Humour] but instead because the author can be shown to have paid such particular care to the essentials of his formal presentation" as to make a later edition "more authoritative in its accidentals on the whole" than the first. This kind of situation can be said to offer an exception to Greg's rationale in the sense that it requires the editor to depart from the common pattern of adopting as copy-text the text closest to the author's manuscript. Calling it an "exception," however, should not lead one to infer that it is not taken into account in Greg's essay. Greg cites Every Man in His Humour at one point (p. 390) as an example of substantive revision, but he also uses it as an illustration of an author's supervising "the printing of the new edition," correcting the proffs and taking responsibility for the revised

edition "in respect of accidentals no less than substantive readings" (p. 389).[33] Bowers issues a useful caution, in other words, against interpreting Greg with an inflexibility alien to the spirit of his essay. Whereas Greg criticized "the tyranny of the copy-text" (p. 382) by arguing in favor of the use of editorial judgment to determine authorial revisions or corrections that should be incorporated into the copy-text, Bowers points out a different kind of tyranny: Greg's concept of divided authority, he says, "has been so welcome to recent textual critics that they have had a tendency to overreact against any other rationale" and thus "have been loath to accept any suggestion of a return to unified authority even when the special situation warrants it" (pp. 179-180). This statement succinctly summarizes a situation brought about by the success of Greg's essay; but one would be incorrect to attribute the situation to a weakness or oversight in Greg's position, since Greg covers the possibility that undivided authority can reside in a later edition and emphasizes flexibility in approaching the great variety of "cases of revision" (p. 390).[34] So long as one arrives at a sensible editorial rationale, it does not of course matter whether or not one finds support for it in Greg's essay. But so much has been written, pro and con, about the relation between Greg and the CEAA editions that it is worth noting just how much is actually encompassed within his rationale. Bowers's essay, by calling attention to a new kind of textual tyranny, provides an occasion for doing this at a critical point in the history of recent textual discussion, just before the appearance of two major analyses.

These two major essays, which must be carefully considered by all who are concerned with modern editing, were published just about a year apart in 1977 and 1978, the first by Tom Davis and the second by Bowers himself. They raise more fundamental questions about the broad applicability of Greg's rationale and analyze them more cogently than all but one or two of the earlier CEAA-related discussions. Davis's piece,[35] a long review of the whole CEAA undertaking, is balanced and thoughtful: although Davis has some serious reservations, he is also able to praise the contributions to "literary, bibliographical, and historical scholarship" made by the CEAA editions (p. 63), and it is clear that, unlike some

previous commentators, he is approaching the issues with an open mind.[36] If it is an overstatement to describe, as he does, the "intense textual debate" generated by the CEAA as "profoundly stimulating and valuable," those adjectives may be applied to his own contribution; and he is right to make the interesting point that the extent and prominence of the continuing discussion mark "the beginnings of a replication of the scholarly audience that the editor of classical texts can expect" (p. 61). Davis concludes with the familiar complaint that CEAA editors have been too rigid in applying what they take to be rules derived from Greg, and he advocates "greater use of editorial judgement" (p. 71). What distinguishes his essay from most of the others that arrive at similar conclusions is the quality of the analysis leading up to the conclusion. He concentrates on two CEAA "methods or rules" that are "adopted from Greg's paper" and "depend on logical inconsistencies": the distinction between substantives and accidentals and the idea of following the copy-text when variants are indifferent. Even though the careful reader is not likely to be in complete agreement with Davis, the intelligence of his analysis makes his argument a rewarding one to think through.

On the distinction between substantives and accidentals, Davis's position is that the concept, both in Greg's essay and in restatements by CEAA editors, is so fuzzy as to be useless and that, in this form at least, it should be scrapped. As he explains it, the trouble essentially is that the definition of substantives as words and accidentals as punctuation and spelling does not coincide with the further definition of substantives as elements of meaning and accidentals as elements of form. CEAA editors do, of course, recognize accidentals as having a kind of meaning distinct from purely formal features of typographic design, for they are concerned to reproduce the former but not the latter. The "absurd consequence," in Davis's words, is that "all CEAA editors are committed simultaneously to the dedicated pursuit of authorial spelling and punctuation, while being equally committed to the belief that these phenomena are meaningless" (p. 68). Obviously Davis has, for effect, made this statement sound as absurd as possible; a fairer way to put the matter would be to say that serious scholarly editors (whether associated with the CEAA or not) regard spelling and punctuation as meaningful parts of a text[37] but that

some of them have been careless in distinguishing nonverbal elements ("accidentals") of a text from the words themselves so as to suggest that the former do not convey meaning. Davis's treatment does dramatize the sloppiness of thinking that has often been associated, from Greg onward, with attempts to define "substantives" and "accidentals." A number of CEAA editors have from the beginning been dissatisfied with these terms and their implications, and several criticisms of Greg's terminology have appeared in print; indeed, much of what Davis says on this score is inherent in Morse Peckham's 1971 essay,[38] though Davis's expression of it is clearer and more forceful. In one sense all this attention is misplaced, for in practice editors have not thought in terms of "substantives" and "accidentals" when making emendations: any variant reading in a later text, whether in wording or spelling or punctuation, has been accepted into the copy-text if it can convincingly be attributed to the author. Of course, as Greg predicted, one can less often argue with conviction about revisions in spelling and punctuation than about revisions in wording, but the same process of thought has been applied to all variants. From another point of view, however, one cannot so easily dismiss the problem, for incoherence of argument deserves to be deplored whether or not it has practical consequences; the very fact that it may not have consequences is in itself troubling. Just because Greg stresses the casualness of his distinction between substantives and accidentals and calls it "practical" rather than "philosophic" is no reason to overlook the problems it causes. On the other hand, one can tell, with a little reflection, that what Greg proposes is not as illogical as the language he uses would seem to imply: it is clear that his distinction was not meant to describe the nature of written communication but only to indicate how people have in fact often reacted to spelling and punctuation. If people who can affect textual transmission, such as compositors or publishing-house editors, think of spelling and punctuation in general as elements of form, distinct from "content," or words, their treatment of those features and the kind of attention they devote to them may be affected. It does not matter whether they are right to make such a distinction; what matters to the scholarly editor, who is a historian, is whether people have indeed acted on this conception.

Davis is quite right, therefore, to stress "the actual or inferred practice of authors and printers" (p. 68). Although he is not the first to make this point, his use of it is particularly effective.[39] Following a lucid account

of textual and typographical meaning (and the way in which printers sometimes have control over "not only the signs in a text that are not considered to generate the meaning of the work, but also many that do" [p. 64]), he provides an excellent summary of Greg that emphasizes the behavior of authors and printers, beginning with this sentence: "Greg's central insight is that both writers and printers tend to make a distinction between the words, and the spelling, punctuation, italicization, paragraphing, hyphenation, and so forth, of the text" (p. 65). After this summary (the paragraph beginning at the bottom of p. 65 should stand as a classic statement of Greg's approach), he points out that his paraphrase unfortunately does not represent the way in which Greg or the CEAA editors present the rationale, and he proceeds to show the difficulties that ensue when the equation of words with meaning, and punctuation with form, is enunciated as if it had a "philosophic" truth beyond its "practical" value in describing the behavior of writers and printers. The fact that Davis can regard his fine summary as a "paraphrase" of Greg, despite the illogicalities that he subsequently specifies in Greg's discussion, shows that he is able to see through the surface confusions to the clear-sighted position that lies beneath. Similarly, I can assert that a number of CEAA editors, regardless of how they may have worded their summaries of Greg, understand his position in the same way that Davis does and would concur in Davis's paraphrase. That does not of course excuse lack of rigor in presentation, and I am not for a moment suggesting that Davis was wrong to criticize many of the restatements of Greg that appear in CEAA editions. But I would add that to concentrate on those statements is to stop short of investigating the way in which Greg's distinction has actually been employed in editing. That there may be a discrepancy between the rhetoric employed in some editions and the editorial approach to spelling and punctuation manifested in the editing itself is unfortunate, but to dwell on the former rather than to see through it to the latter is not finally to get to the heart of the matter. Davis's incisive treatment of the distinction between substantives and accidentals leads one to be somewhat disappointed that his analysis ends with criticism of the language used by CEAA editors; one would expect, from the level of his previous discussion, that he might have examined the extent to which the actual editorial practice of those editors is accurately reflected in their statements, just as he was able to "paraphrase" what Greg was getting at in the "Rationale" in spite of the language in which it was expressed. In any case, one can agree with Davis when he says, "It is to be hoped then that we shall not hear much more about accidentals and substantives, but about words, and punctuation, and the actual or inferred practice of authors and printers" (p. 68). A number of editors and writers on editing have for some time now been reluctant to use these terms or have in fact refrained from using them; and it is clear that the time has come when the terms should in most situations be abandoned. The real contribution that Greg makes can—as I shall suggest shortly—be taken advantage of without clouding the issues by the use of these questionably defined terms.[40]

The second of Davis's two major points is a criticism of Greg's suggestion that the copy-text reading be retained whenever the editor finds the variant readings indifferent. Unfortunately Davis's treatment of this question is less satisfactory than his treatment of the previous one, for here he only alludes to the central issue involved; he does not pursue it at all but instead turns to a less significant aspect of the general question and then falls short of his own standard of perceptiveness (amply demonstrated earlier) in handling it. He begins on a promising note by questioning whether Greg's generalization is applicable to later periods; nineteenth-century novels, he points out, are "much more subject to authoritative revision than those works with which Greg was concerned" (p. 70). "It is by no means inevitable," he continues, "that in the case of indifferent verbal variants non-evident compositorial error will preponderate; it is possible that the reverse is true." And if "such errors do not preponderate, then to treat all indifferent variants as compositorial errors is to increase one's chances of adopting the wrong reading." Clearly the question raised here is of crucial importance. Whether or not Greg's rationale is valid for post-Renaissance writings has of course been frequently asked, often in trivial ways; but the real heart of the issue is to determine whether the circumstances of publishing in certain periods were such that textual variations in later editions were more likely to be authorial than compositorial. If such were the case, then Greg's rationale would obviously not be applicable to those periods, for it would lead editors in the wrong direction, assigning presumptive authority to an

early text rather than a later one.[41] Instead of following this interesting line of thought, however, Davis lamely concludes: "in practice since there is no way of knowing which kind of reading preponderates [authorial revision or compositorial error],[42] no ascertainable harm is done by following Greg's principle, so long as its tentativeness is preserved." If it is true that one cannot generalize about the nature of the indifferent variants, then it is also true that no ascertainable harm would come from choosing an early, rather than a later, text as copy-text and assigning it presumptive authority in the case of indifferent variants; but neither would any ascertainable good come from it, and under those circumstances one would have no justifiable reason to follow Greg's approach. To believe that an otherwise unjustifiable approach would be permissible if its "tentativeness" were emphasized is not a very constructive position; one would do better to regard seriously Davis's earlier point that "taking the trouble to toss a coin would statistically speaking give one a better text." The fact is, however, that Greg's rationale, for all its tentativeness, is predicated on the belief that one can indeed generalize about the deterioration of successive texts in sixteenth- and seventeenth-century editions; otherwise there would be no justification for retaining the readings of an early edition at points where variants are indifferent. And if one is to employ Greg's approach in dealing with later writings, one cannot simply appeal to tentativeness but must confront the question whether the same generalization applies to other periods. This central question, brought up but not analyzed by Davis, is examined by Bowers in the essay to which we shall shortly turn.

First, however, we should note what Davis concentrates on instead. Taking as his theme the "tentativeness" of Greg, he points out the way in which CEAA editors and commentators have in his opinion hardened Greg's tentativeness into "something much nearer certainty." He calls it a "heresy" (against Greg, that is) to claim that by retaining copy-text substantives (as well as accidentals, of course) in cases of indifferent variants one maximizes the chances of incorporating the author's intended readings. It is true that Greg considers the choice of copy-text to depend "solely on its formal features (accidentals)" and says that "fidelity as regards

substantive readings is irrelevant" (p. 386); but these statements are not inconsistent with his view, regarding substantive variants in revised editions, that a later variant which is "either completely indifferent or manifestly inferior, or for the substitution of which no motive can be suggested, should be treated as fortuitous and refused admission to the text" (pp. 387-388). Whether or not Greg can be interpreted as supporting the position, however, is not the principal point, as Davis realizes. What Davis argues is that CEAA editors and commentators, who believe they are acting in the best interests of preserving the author's intention by emending the copy-text with later readings (both substantives and accidentals) only when there is compelling evidence to do so, are in fact guilty of yielding to "the tyranny of the copy-text." There is some irony in this situation, since everyone is agreed that Greg's aim was to give editors freedom to use their judgment in deciding individual readings and to liberate them from feeling bound to the substantives of the copy-text (or to all the substantive variants of a later text in which it is clear that some are authorial). The crux of the disagreement lies in determining what is meant by an "indifferent" variant. If an editor describes a particular variant as somewhat suggesting an authorial rather than a compositorial hand but finally considers the suggestion not to be strong enough to justify emendation of the copy-text, Davis believes that such an editor has not really regarded the variant as indifferent and has simply retreated to the comfort of sticking with the copy-text. Naturally, different editors might make different decisions on whether to emend at a given point: allowing editors to exercise judgment ensures that there will be no uniformity of result. But just because an argument can be made for emending need not lead one to think that the failure to emend signifies a backing away from a decision and a reliance on a mechanical rule. It is crucial in scholarly critical editing, after all, for editors to distinguish between their personal preferences and their informed judgment regarding what the author would have preferred. Calling a variant indifferent, therefore, is another way of saying that the evidence for its being authorial (or nonauthorial) is not, in the editor's judgment, sufficient to settle the matter. The trouble with Davis's discussion of this point is, first, that he does not make adequate allowance for the role of scholarly judgment in the determination of what is indifferent, and, second, that he does not investigate the question whether Greg's suggestion for the treatment of indifferent variants is valid for periods other than the Renaissance.[43] Davis ends with some examples from CEAA texts, intended to show the need for a fuller recognition of the "subjective and critical process" involved in critical editing, and he repeats the usual points (which are not in question) about the inevitable lack of definitiveness of the products of a critical process.[44] The last part of his essay is thus less impressive than the earlier part; but that early part takes its place as one of the key passages in the whole editorial discussion stimulated by the CEAA. If more of the discussion had been on such a level, editorial thinking would now be more advanced than it is.

The other major essay of this period, and one of greater significance, is Fredson Bowers's "Greg's 'Rationale of Copy-Text' Revisited."[45] This remarkable essay confronts directly the question of the applicability of Greg's rationale to periods other than those Greg himself was familiar with and suggests that certain cautions are in order in attempting to make such a transference. Because Bowers is the most influential champion of Greg's rationale, his reservations on this score are particularly striking. Champions of causes are often regarded as being inflexibly committed to the positions with which their names are normally identified, and some superficial critics of the CEAA have looked at various CEAA editors, including Bowers, in this light. Anyone who has followed Bowers's writings carefully, however, will recognize that this essay is not the first occasion on which his concern with developing an editorial rationale for works of the last four centuries has led him to adapt or move beyond what Greg specifically says. Perhaps the most important earlier instance is his essay on "Multiple Authority,"[46] which complements Greg's "Rationale" by taking up a class of textual relationships not covered by Greg (radiating texts, standing equidistant from a lost common ancestor, as opposed to a

series of texts in linear descent). In contrast to that essay, which supplements Greg, the recent essay provides an incisive analysis of what Greg does in fact say. Bowers's knowledge of Renaissance texts enables him to examine closely the illustrations cited by Greg and to explain, more precisely than had been done before, the way in which Greg's recommendations arise from the textual situations generally encountered in work on Renaissance literature. From that firm base, he can then explore how well what Greg was saying fits other situations more characteristic of later periods. It is hard to imagine anyone better equipped for this task: Bowers's extensive editing of works from many periods has provided him with an amazing range of detailed examples. This kind of knowledge is exactly what had been missing from previous discussions of the broad applicability of Greg, and thoughtful editors will need to work their way through Bowers's analyses.

As the essay makes clear, the significance of copy-text for Greg was rather different from what it has been for editors of nineteenth- and twentieth-century works. Greg was under no misapprehension that the accidentals of an Elizabethan printed book reflected much of the author's practice; for Greg the choice of an early edition to supply copy-text was simply an expedient measure designed to ensure that the edited text would have accidentals as close in time and place to the author's as possible. As one moves to works of later periods, however, with more fixed spelling and punctuation and for which more authorial manuscripts survive, the chances of choosing a copy-text that will preserve authorial accidentals increases. For the editors of later works, accidentals have "a literary interest, not merely a philological" one: these editors are likely to believe "that the accidentals are an inseparable whole with the substantives in transmitting the author's total meaning" (p. 125). Therefore for them the choice of copy-text "transcends the grounds of expediency and must be recognized as having a critical significance beyond that which Greg conceived"; and they will more often be able to exercise critical judgment in the choice of accidentals "on the same basis that Greg urges for the substantives" (pp. 128-129). Although this difference in point of view might affect the significance of particular choices of copy-text, I think one can say that it would not in itself require different general rationales of copy-text for different periods. One could operate, for instance, within Greg's general rationale, even though the different conditions associated with different periods might cause editors more often to find persuasive evidence for selecting a revised edition in certain periods than in others. What would remain the same would be the presumption in favor of an early text as copy-text over a later, except when there is convincing evidence to the contrary. But since the purpose of choosing a copy-text is solely to provide a rationale for handling indifferent

variants (substantives as well as accidentals), this presumption is fruitful only if there is reason to believe that texts normally deteriorate as they are reprinted or copied. We come back, then, to this issue as the central one. Bowers's analysis of the differences between Greg's concerns in his own editing and the concerns of editors of later writings is enlightening; but the question whether Greg's rationale is applicable to all periods (or periods other than the one for which it was originally designed) is a separate question, and one that turns entirely on the validity of the idea that texts usually deteriorate as they are transmitted. Bowers's attention to this point is therefore the aspect of his paper that is of the greatest moment.

There is no need here to go over the illustrations he discusses from post-Renaissance literature; they are all cogently set forth, and I have no quarrel with the textual observations made in them. What I wish to comment on is the generalization drawn from these examples. Bowers concludes that, in cases of linear transmission from manuscript to print or from one printed edition to another,

Although this statement specifically deals with substantive variants, a similar point is made about accidentals: The concept of "accuracy" or fidelity entailed here clearly involves more than compositors' skill in following copy; it also takes into account their conventional practices. The state of English spelling in the Renaissance was such that compositors could feel free to follow their own spelling habits or to alter spellings to facilitate justification. As a result, what they set was not "accurate" in the sense that it did not fully correspond to their copy; but the lack of correspondence was not—at least in many cases was not—owing to carelessness or lack of ability. In the same way, any attempt to assess the performance of compositors in later periods must take into consideration the practices that compositors were expected, or allowed, to engage in. As spelling became more fixed, compositorial freedom to introduce variant spellings was correspondingly reduced; but if another kind of compositorial interference took its place, the divergence between copy and print was not necessarily decreased. For example, the growing concern with "correctness" of spelling and with syntactical punctuation in the nineteenth century meant that compositors were often expected to impose the preferred forms on wayward copy; such compositors did not have the freedom of their Elizabethan counterparts to follow their own inclinations, but they might equally transform the copy they were working from.[48] Furthermore, the differences between manuscript and first edition or between one edition and the next are not entirely to be accounted for by the activity of compositors, for there may also be copyeditors or publishers' "readers" whose function it is to make certain kinds of alterations (including many that are substantive) before the compositors enter the picture. Such persons were less in evidence in earlier years, before publishing as a distinct operation had fully developed; but in later years (the last two centuries or so), during the very time when—one might argue—compositorial fidelity to copy generally increased, the structure of publishing grew in complexity, and new positions with responsibility for making alterations in printer's copy came into being. The question that must be asked, therefore, is not whether compositorial accuracy increased over the years but whether the printing-publishing process as a whole has in certain periods resulted in so little alteration of printer's copy that, in the absence of contrary evidence, one had better regard the later variants, both substantives and accidentals, as presumably authorial.

It is extremely difficult, of course, to generalize about such matters. Not all printers and publishers in a given period will have behaved the same way—nor will a single publisher necessarily have treated each author the same way. Editors must naturally search for any evidence bearing on these matters in individual cases; and the more they learn about the author and the printers and publishers involved, the less they are in need of a generalization to fall back on. But when they can find no direct evidence pertaining to a particular situation, they are forced to try to

generalize about the practices prevalent in the period in parallel circumstances. The papers of other authors and publishers are one source, and another is the published comments that appear in printers' manuals, publishers' style books, authors' guides, and the like. There have recently been some surveys of these books, showing—despite many individual variations—that it was not uncommon, through the nineteenth century and into the twentieth, for printers and publishers to regard spelling and punctuation as their preserve.[49] My own unsystematic examination of such books confirms this view. Not long ago, for instance, I came across Benjamin Drew's Pens and Types; or, Hints and Helps for Those Who Write, Print, or Read, published by Lee & Shepard of Boston in 1872—a book that I have not seen referred to in the published surveys. Its statements may of course reflect only the policy of Lee & Shepard, but on the subject of punctuation they do coincide with those of a number of other books. The distinction it makes between words and punctuation in fact resembles Greg's generalization about the practices of printers nearly three centuries earlier. Drew, after calling punctuation "the perfection of common sense," continues:

Drew's attitude toward punctuation is clearly a nineteenth-century one, but his assertion that authors pay more attention to words than to punctuation and his acceptance of compositorial alteration of punctuation as a routine occurrence are not very different from points made by Greg. Even the idea that the division between words and punctuation is expedient, not logical, emerges from this passage—since the admission is made that punctuation can affect meaning.[50] Above all, Drew makes clear

that the printer has authority over the punctuation: the author's punctuation is to be followed only if the manuscript is "fairly" or "presentably" punctuated. One cannot generalize from the passage, or the many others like it, and claim that the attitudes expressed there were universal. Nevertheless, evidence of this sort turns up frequently enough to make one feel hesitant about accepting a generalization for any period that would place presumptive authority in the case of indifferent accidentals with the later rather than the earlier readings. And the same can obviously be said with equal force for substantive variants, especially in the years after the publishing industry became well developed: the officiousness of nineteenth-century publishers' "readers" is notorious. What must be emphasized is that in thinking about copy-text one is concerned only with indifferent variants. Whenever one has good reasons for choosing particular readings, one has no need for a copy-text theory; and among the reasons that will prevent readings from seeming indifferent is knowledge one has about the practices of the printer and publisher involved. The question for copy-text theory is whether, in the absence of evidence that one finds persuasive bearing on a particular case, one can generalize about a period as a whole and say that the odds favor the authority of indifferent variants in a revised edition. The picture that continues to emerge regarding the processes of textual transmission in all periods suggests that it would be unwise to take such a position. Bowers's discussion confronts this central issue more directly than any other essay and ends up questioning the appropriateness of Greg's rationale for post-Renaissance literature. I do not disagree with his analyses of specific cases but would simply suggest a different emphasis: choosing readings from revised editions or selecting revised editions as copy-texts when the evidence points in that direction does not—even if such instances are frequent— negate the existence of publishers' editors who made revisions or of compositors who were trained in the tradition of certain printers' manuals and took punctuation as their domain. As long as such possibilities for nonauthorial alteration of texts exist—as they seem to in all periods—the conservative approach would appear to be to consider an early text the better choice when the editor does not judge the available evidence conclusive one way or the other.

In addition to Bowers's essay, 1978 saw the appearance of several essays and a book that ought to be mentioned here.[51] One of the essays, by

David Foxon,[52] is principally concerned with the same point as Bowers's. Arguing from his knowledge of the texts of Pope, Foxon suggests that Greg did not sufficiently distinguish between scribes and compositors: the latter were subject, Foxon says, to "the check of proof-reading, often by the author himself" (p. 121), as well as to the general rule set forth by Moxon that compositors were to follow copy strictly. An author's passing of proofs does not of course in itself tell one anything about that author's care in proofreading or imply that the text fully conforms to the author's intention. As for the printer's rule of compositorial fidelity, there were various occasions—as we have just noted and as Foxon himself observes— for departing from that rule, even though it is no doubt true that the simpler and easier course was to make no changes. Thus to say that Pope's compositors "followed copy rather than their own habits and inclinations, and this is what we expect from modern compositors" (p. 121) is not to say that as a general rule one should rely on compositorial fidelity. Foxon is right to notice that "correctors" at the press were also responsible for alterations[53] and to emphasize that the "crucial question" is what the author's practice was in correcting proof. He rightly sees where this leads (though I am not sure why it is a "frightening prospect," since it is to be taken for granted as the scholarly approach): editing a work "may well mean studying the whole publishing career of that author, the changes in his practices when writing manuscripts and when proofreading, the conventions of his printers" (p. 123). These are useful points for Foxon to have underscored; but it is not clear how his discussion is a modification of Greg. He has established that in reading proof Pope was as concerned with accidentals as with substantives; and in later editions Pope "continued to make revisions on a large scale," with the changes in accidentals outnumbering those in substantives—"later editions refine his punctuation and remove typographical inconsistencies" (p. 119). "In this context," Foxon asserts, "Greg's proposals are not applicable, and my experience suggests that the case of Pope is by no means unusual. The cause of our present difficulties is the belief that a single generalized rule can have universal application." All one can do in the face of such a statement, made at this late date by a perceptive and learned bibliographer, is to repeat once again—by now perhaps rather tiredly— what has been said time after time before: the only "generalized rule" that Greg proposes is for situations in which one has no basis otherwise for making a decision; to find that a number of authors revised accidentals as well as substantives so thoroughly for a late edition as to make that edition the best choice for copy-text in no way contradicts what Greg was saying.[54]

At roughly the same time the first of a series of articles under the rubric "Redefining the Definitive" was published in the Bulletin of Research in the Humanities. The inauguration of a series to examine "definitive" editions in the first number of this journal under its new title shows its intention to continue the interest in editorial methodology exhibited in 1971, when it (as Bulletin of the New York Public Library) contained a brief flurry of discussion about CEAA practices.[55] Although this first piece is devoted to particular textual decisions in one edition,[56] the second, later in the year, raises a general point about the treatment of punctuation in CEAA editions. In it Harry Knowles Girling argues that "CEAA-approved editors" restrict authorial freedom of punctuation by imposing printers' conventional practices on manuscripts or by preserving the compositorial practices of printed texts.[57] Authors who protest printers' regularizations of punctuation, he says, "would lament the embalming of printers' choices in a 'definitive' text of their work" (p. 303); and scholars who adhere to a printed copy-text are "expending their devoted labours in perpetuating the original flouting of authorial intentions" (p. 297). Girling's generalization here is off the mark: although some CEAA editors have in certain instances made unwise decisions about punctuation, CEAA editors for the most part would agree with Girling's insistence on the importance of authorial punctuation; the effort to incorporate authors' rather than printers' punctuation into scholarly texts has normally been one of their primary concerns. Despite this misplaced attempt to discredit CEAA editing, Girling's article is refreshing because his criticism is just the opposite of the more usual one: more often critics of the CEAA are inclined to say that authors expected their works to be styled by the printer or publisher and that selecting manuscripts

as copy-texts does not do justice to the authors' intentions. Girling, however, believes—in line with most CEAA editors—that authorial punctuation reflects "semantic and stylistic distinctions" (p. 297) and that "the punctuation to be observed in nineteenth-century authors' manuscripts seems to have its own alternative consistency" (p. 303; alternative, that is, to printers' practice). He carries this position so far as to argue that "the editing of a piece of holograph manuscript as copy-text raises the question of the necessity of emending at all" (p. 304). Obviously a facsimile edition of a manuscript has its place; but if one is preparing a critical edition, one cannot by definition rule out the possibility of emending. An author's manuscript may contain slips of the pen or readings later altered by the author, and a critical editor using the manuscript as copy-text would have to emend at those points. Emendations involve judgment and therefore are sometimes wrong; but the goal is to establish, as Girling would wish, the readings (including punctuation) of the author.[58] The value of Girling's essay is that it provides a forceful statement of the reasons for preserving an author's own punctuation and offers one of the most detailed analyses available within the context of editorial discussion showing the ways in which punctuation affects meaning. After examining the changes in punctuation in the New York Edition of The Princess Casamassima, he concludes that James "had succeeded in exemplifying a punctuation system unguided by rules or precedents, and controlled only by a sense of taste and style" (p. 318). James's approach, he goes on, "gives a clue to the way that all authors (and all pen-wielders, even the most humble) might regard punctuation, were they not bullied into accepting a system that was developed to suit the exigencies of hand-compositors and maintained to support the conformist pieties of copy-editors and editorial scholars." One can overlook Girling's inaccurate generalizations about scholarly editors in order to have his strong argument on behalf of the determination and retention of authorial punctuation, an argument that in fact supports the view of a large proportion of editorial scholars.

One more publication of 1978, appearing in September, is Philip Gaskell's From Writer to Reader: Studies in Editorial Method. This book consists of a ten-page introduction on the general principles of editing,

followed by twelve chapters, each of which examines the textual problems in a particular work, normally focusing on a single passage and reproducing the various relevant texts of that passage. In his introduction Gaskell improves in some respects on his unsatisfactory treatment of editing at the end of A New Introduction to Bibliography (1972).[59] For instance, he now recognizes the reasons why it may be defensible to allow copy-text inconsistences to stand, and he sees that there is a difference between authorial acquiescence and intention. As a whole, however, his introductory chapter unfortunately does not succeed in providing a solid foundation for editorial thinking. His eagerness to emphasize the idea that one cannot edit by rules causes him to neglect certain basic principles or distinctions that are necessary for clear thinking on editorial matters. In a single sentence he can bring together three distinct approaches to editing without explaining how radically different they are: the editor, he says, in selecting a copy-text, "may consider whether the choice of a particular version will result in a system of punctuation, spelling, etc., in the edited text which is more likely than another to represent the author's own intentions for the form of the work; whether the resulting system is better critically than one that would result from a different choice; and whether it is the system most likely to be acceptable to readers of the edition" (p. 6). He proceeds to suggest that when an editor has insufficient evidence regarding authorial intentions, "the editor might decide for instance to follow the version that is best on critical grounds." What he never explains is how such a switch would mean preparing an entirely different kind of edition; he never clearly distinguishes between scholarly critical editing—in which the editors' judgments are directed toward attempting to establish what the author would have wanted—and another kind of editing in which choices are based on the editors' own critical preferences. Such imprecision and lack of logic occur throughout the book.[60] The individual analyses are nevertheless useful up to a point, since they conveniently bring together some basic information about a wide range of texts; but any commentary Gaskell adds to the factual information must be taken with great caution.

Given Gaskell's declared "message" that "the editor should not base his work on any predetermined rule or theory" (p. vii), it is perhaps not surprising that he should have joined those who look on the CEAA as

rigidly doctrinaire, inflexibly insistent on a single rule. A number of his observations appear to stem from this image of the CEAA, but the one chapter that directly involves a CEAA edition is the chapter on Hawthorne's The Marble Faun (pp. 183-195). That chapter is particularly relevant in the present context and can represent the kinds of problems that occur elsewhere in the book.[61] After a brief account of the surviving textual evidence, pointing out that the compositors of the first edition "normalized the text freely" (p. 185), Gaskell discusses the choice of copy-text. The Ohio State Marble Faun, he says, "is based, in accordance with the principles of the series, on the manuscript, not on the first printed edition" (p. 189). The manuscript is indeed the copy-text, but otherwise this sentence is misleading, for it suggests that the CEAA required surviving manuscripts to be used as copy-texts. What the CEAA guidelines in fact say is that a finished or printer's-copy manuscript would "normally" become copy-text but that an author's habits of revision in proof "may modify this choice."[62] Gaskell, however, is convinced of the rigidity of the CEAA and describes the CEAA application of Greg's rationale as "both more comprehensive and less flexible than Greg's pragmatic approach," asserting incorrectly that "it denies the editor much critical discretion in the choice of copy-text" (p. 190).[63] In extending Greg's rationale from Renaissance works, for which manuscripts do not often survive, to later works, the CEAA has —according to Gaskell— adopted the "very dubious" assumption that, "in the absence of evidence to the contrary, a nineteenth- or twentieth-century author 'intended' the manuscript rather than the printed version of his work" (p. 191). Gaskell does not pursue the meaning of authorial intention or explore reasons why students of an author might legitimately prefer to read a text with that author's punctuation rather than a compositor's or copy-editor's. Instead he shifts the grounds of the discussion: because, he says, we often cannot know whether authors "specifically approve or disapprove the normalization of the text" that occurs in the process of printing and publishing, all we can state is "that this version is to be preferred to that on other—perhaps critical—grounds." Just what he means by "critical" becomes clear in the next paragraph, when he explains why he agrees in this instance that the manuscript should be copy-text. The manuscript, he says, has "excellent punctuation" that contributes to the rhythm of Hawthorne's prose "in a way that the mechanical details of the normalized printed texts do not"; thus "there is a good critical case for basing an edition on the text of the final manuscript, which is that it offers the most convincing punctuation of the text" (pp. 191-192). The implication is that if the printed text had contained more "convincing" punctuation, it would become the copy-text, regardless of whether it was the author's. What Gaskell has done here—as he does repeatedly—is to confuse the attempt to establish an author's own words and punctuation with the attempt to produce the artistically "best" text of a work. He claims, "There is no need to introduce the assumption, which cannot be supported by evidence, that Hawthorne intended or would have preferred his own punctuation to that of the first edition" (p. 192). But if one is engaged in producing a scholarly critical edition, one is by definition attempting to establish an author's own words and punctuation, not one's own preferences, and part of the "evidence" is one's developed sensitivity to the author's characteristics. Whether the punctuation is "excellent" and "convincing" is beside the point, unless it is excellent and convincing in a way that the editor has learned, through close study of the author, to associate with that author's punctuation.

The central issue Gaskell attempts to deal with here—the choice of copy-text as between manuscript and first edition, and the assessment of authorial intention in relation to this choice—deserves careful attention, but Gaskell's treatment only confuses the question.[64] He then turns to an evaluation of whether the emendations in the CEAA Marble Faun are "both necessary and right." In objecting to what he regards as excessive regularization, he makes some good points. But he nowhere alludes to the practice of other CEAA editions; and since at the beginning of the

chapter he calls this volume "a typical CEAA edition" (p. 184), he leaves the impression that the kind of regularization he describes is characteristic of CEAA editions—which is not the case. Indeed, the final two paragraphs of the chapter consist of entirely unwarranted generalizations, coming at the conclusion of a discussion of a single volume:

The idea that one volume—regarded for whatever reason as "typical"—can lead to such conclusions derives from Gaskell's view of the inflexibility of the CEAA and the resulting assumption that all volumes in the series are identical. Actually, the "main editorial principle"—if by that is meant Greg's rationale—liberates editors to use their judgment; so even if it were "inflexibly applied" it would result in a great deal of variation, which in fact is what exists. I would certainly not wish to defend all CEAA volumes or every decision made in them. But whatever opinion one holds of individual volumes, one must deplore the kind of glib generalization that Gaskell indulges in here. It is sad to have to recognize that his book, which one might reasonably have expected to be an important statement, marks no advance in editorial thinking and indeed does not rise above the level of many of the less satisfactory contributions to this continuing discussion.[65]

It remains to note briefly one or two comments from 1979, which unhappily show what little effect all this editorial debate has had on improving the caliber of some of the discussions that get into print. F. W. Bateson prefaces a textual discussion of The School for Scandal [66] with some remarks on Greg's rationale that exhibit the same mixing up of distinct approaches to editing already observed in Gaskell. Greg's "mechanical ruling," he says, growing out of work on Elizabethan drama, was more concerned with printers' corruptions than authors' revisions (p. 323), and he believes that "biblio-textual" research is "helpless before authorial revision" (p. 322)—an argument that obviously ignores the central role Greg gives to editorial judgment in detecting authorially revised readings. The novel aspect of his criticism of Greg, however, is his opinion that "Such textual philistinism as Greg's may be ultimately social in its origin" (p. 324); his astonishing view is that Greg, because he was wealthy and therefore interested in property, looked on literary works as the "private property" of their authors and wished to tie later editions and readers to authors' revised readings even when those readings were inferior. Everyone would agree that authors can be whimsical and unsystematic in making revisions: "A revision . . . may stay or go or revert to its original form as the author decides—on the inexplicable spur of

the moment, or in the light perhaps of the passage's immediate context, or reconsidered to conform to the general pattern of the work or his writings generally (or even his life)" (p. 323). Of course an editor could continue this process, but the result would not be of particular interest to students of the author; the scholarly editor must be concerned with what the author did. Bateson does not seem to realize that scholarly critical editors are not limited to believing that last is best; they must, and do, try to assess the motivation underlying revisions and decide when revisions produce essentially new works, and they will not always adopt the last revision. But they could not agree with Bateson that the "responsible editor" will select "whatever after due consideration he (the editor) believes the best authorial version to be, whether it happens to be early, middle-period, or last" (p. 324). Editors of most authors will sympathize with Bateson when he says that Sheridan "was continually questioning himself—and very often coming up with the wrong answer!" (p. 333). But for a scholarly editor to choose, in each case of authorial revision, the reading that seems preferable "considered as meaning, good English, good literature" (p. 335) is not the way to be "responsibly eclectic" (p. 324). The place to make those judgments is not within the text of a scholarly critical edition.[67]

Another comment from 1979 that reflects an unfortunate misconception—one

that is perhaps more widespread—occurs in J. F. Fuggles's essay reviewing the bibliographical research of 1978.[68] In the course of this review he says that the argument for the exact transcription of historical documents would be strengthened by examples showing how "the meaning or significance had been affected by modernization" (p. 10).[69] Presumably he means "normalization" as well as "modernization," for the issue is broader than the question of whether or not to modernize. But the principal, and troubling, point is his implication that tampering with the punctuation and spelling of the original is unacceptable only if it can be shown to alter the meaning of the text. The burden of proof, however, should be the other way round. If an alteration is thought not to change the meaning (or "clarify" it, as some editors prefer to say), then what is the purpose of making it? Readers are not normally prevented from understanding a text by oddities and inconsistencies of punctuation and spelling, and when these irregularities are characteristic of the author what is the point of altering them? It is hard to see why editors think they are accomplishing anything by straightening out the details of spelling and punctuation in a letter or journal simply for the sake of tidying up.[70] To say that the tidying up does not alter the meaning is no defense, since it provides no reason for making any change at all. In fact, of course, any shift from the author's spelling and punctuation is likely to affect nuances of meaning. But even if that were not the case, why should one not prefer to read a document as its author wrote it? Fuggles proceeds to repeat an argument that should have been discredited long before now, one that attempts to make a distinction between historical documents and works of literature: "scholars come to different kinds of texts for different reasons," he says, and "inevitably this will influence editorial procedures. The historian will come to the letters of Lord John Russell because he is interested, perhaps, in what Lord John said to a cabinet colleague on a particular subject. Milton's punctuation will excite interest in a way that Lord John's will not" (pp. 10-11). It is no doubt true that Milton's punctuation is of greater interest than Russell's, but not for the reason Fuggles implies. Fuggles seems to be saying that historians are interested in "content" and that students of literature are concerned with "form" as well. But both groups wish to understand as fully as possible what is being communicated by certain pieces of writing. In approaching that task one can no more rule out Russell's punctuation than Milton's; the punctuation in each case is part of the essential evidence that scholars must have before them. The notion that punctuation is of more significance in "literature" than in other writing cannot be supported, even if what constitutes "literature" could be identified. That it still has its adherents shows that there is yet some way to go before fruitful communication among editors in all fields can be achieved.

This glance over editorial discussions of the past six years has revealed, I trust, certain recurrent issues that have proved particularly troubling to those who have been moved to write about editorial matters. Some of the discussions have been irresponsible and others carefully thought out, but whatever their quality they keep returning to a small number of basic problems. Evaluating those discussions amounts to working out a coherent theoretical approach to editing, as I hope my remarks illustrate. But analyzing particular arguments as they come up in a chronological survey does not produce a systematic statement, though it can provide the underlying argument to support such a statement. As a result of working through these discussions, therefore, one should be in a position to formulate, and to read with understanding, a summary statement of editorial rationale.

II

The central questions of editing can be simply put. Answers to them require thought and do not always come easily. But the editorial debate that has appeared in print in recent years has sometimes made the search for answers more complicated than necessary by diverting attention to issues that exist only as a result of misconceptions. Although there has thus been a certain amount of wasted effort, one at least can now be fairly sure of the problem spots. I shall therefore take advantage of the existence of all this discussion, and analysis of the discussion, by offering a compressed statement regarding the basic areas of difficulty. I need not elaborate on the points here, because the arguments have been rehearsed in detail, and often more than once, in what has appeared during the last several years. A concise expression of some basic propositions can serve a purpose, indeed, because of the very voluminousness of the published discussion. And I think that a further purpose can be served, at this stage,

by making no reference to the CEAA or to Greg's rationale. Such a body of emotional reactions and stock responses has by now grown up around them that simply to mention them in a theoretical discussion deflects some readers' concentration in unprofitable directions.[71] Whether or not some ideas have been attributed to Greg that are not actually to be found in his writing is obviously a different question from whether or not those ideas are sound; but sometimes the two questions have become confused. There can be no doubt that Greg's rationale has been of crucial importance in the development of modern editorial thinking; but that thinking has carried Greg's line of reasoning beyond the point at which he left it. To keep returning to the question of whether Greg's rationale is applicable to periods later than the Renaissance, as some critics do, is a rather pointless way of proceeding, or at least places the emphasis on the wrong part of the issue. The important question is whether a particular way of thinking (which happens to have been influenced by Greg, but that is beside the point) offers a reasonable approach to editorial matters. The published debate suggests, I think, that the principal problems fall into three groups, and I shall describe what appears to me to be a reasonable approach to each of these three areas.

The first set of questions consists of the preliminary ones that any editor must decide at the outset, questions about what kind of edition is to be undertaken. The most basic distinction is between editions in which the aim is historical—the reproduction of a particular text from the past or the reconstruction of what the author intended—and those in which the editor's own personal preferences determine the alterations to be made in a copy-text. Scholarly editions conform to the first approach, and presumably scholarly editing is what this whole debate has been about. Yet elementary as this distinction is, it has been lost sight of in some of the discussions. Historical scholarship naturally involves judgment and subjective decisions. But there is a difference between using one's informed judgment to attempt to decide what readings reflect the author's intention and making subjective decisions reflecting only one's own tastes and inclinations; and this difference has sometimes been ignored by those who are eager to stress the subjective element in critical editing. After one has decided to undertake a scholarly edition, and understands what that means, one still has to decide whether the edition is to be critical (that is, using editorial judgment to determine when, and whether, emendations are to be made in the text) or noncritical (that is, reproducing exactly one particular text, without alteration). Choice

between these alternatives will be affected by the nature of the text to be edited: pieces of writing to be presented as documents are most appropriately provided in noncritical texts, whereas those to be presented as finished works are most usefully offered in critical texts.

This is really all that needs to be said in a general way about the choice among different kinds of editions—the absence of any reference to modernizing, to the nature of the intended audience, or to whether or not writings are "literary" being meant to suggest that these matters need not be taken into account. However, so much has been said about them— they have proved to be the most prominent red herrings of editorial debate—that it now seems impossible to pass over them with no comment at all. Regularizing and modernizing (their aims may be different, but they amount to the same thing) are ahistorical in orientation and therefore have no place in the historical approach to texts—which is to say, in scholarly editions. Punctuation and spelling have been the subject of a disproportionately large amount of editorial discussion, principally as a result of a doubly-mistaken assumption: that punctuation and spelling are somehow not full-fledged elements of a text, giving an editor license to be freer with them than with the words, and that most readers will have difficulty reading—or may refuse to read—texts that do not exhibit in punctuation and spelling the consistency or the particular practices they are accustomed to. The defensible position in each case is the opposite: that punctuation and spelling are integral parts of a text, affecting its meaning and impact,[72] and that readers do not generally find it a great difficulty or inconvenience to read a text containing spelling and punctuation from an earlier time. If readers are not so easily put off as some editors think, and if there is at least a chance (to understate the case) that alterations in spelling and punctuation may distort the meaning, then why should editors go to the trouble of regularizing and modernizing? And if one follows this line of thinking, the question of intended audience becomes irrelevant, because a text prepared for scholars will also be the appropriate one to present to students and to the general public. The intended audience may indeed be a factor—for economic reasons—in deciding whether a detailed apparatus is to be published with the text; but there is no reason why it should be a factor in determining the treatment of the text itself.[73] Neither should the question whether the work is belletristic have a bearing: it is delusory to think that punctuation

is more crucial to written "literature" than to the written communication of historians, philosophers, or scientists. That the line between "creative" writing and "factual" writing cannot be clearly defined is well illustrated by Hayden White's comments on what historical narratives "most manifestly are"—"verbal fictions, the contents of which are as much invented as found and the forms of which have more in common with their counterparts in literature than they have with those in the sciences."[74] Even scientific writing, though it may be different in some respects, uses words and punctuation to communicate ideas, and anyone wishing to understand a scientific work will need—just as much as readers of "literary" works—a text that is reliable in its punctuation as well as in its words.

The second large group of questions concerns the nature of authorial intention and how one is to handle the difficult distinction between intention and expectation. First it should be said that these are questions entailing critical judgment and are relevant only to critical editing. If one is producing a noncritical text, one is concerned to reproduce what actually appears in a particular document or impression, regardless of the extent to which it reflects the author's intentions. Of course, critical judgment about intention may enter into the choice of document or impression to reproduce in the first place and into the annotation supplied with the noncritical text, but it will not have a bearing on the readings of the text itself. In a scholarly critical edition, on the other hand, the aim is to emend the selected text so that it conforms to the author's intention; one can never fully attain such a goal (or know that one has attained it), but at least one can move toward it by applying informed judgment to the available evidence. Obviously many writers have different intentions at different times, and in such cases one must decide which version of a work one wishes to edit, for no one critical text can reflect these multiple intentions simultaneously. But when an authorial revision does not indicate a new conception of a work but simply a continuation of the process of perfecting the expression of the same conception, that revision can legitimately replace the earlier reading in a single critical (that is, eclectic) text. Judging when such eclecticism is justified and when it is not remains a difficult decision, but it is central to critical editing.[75] A related and equally basic problem is determining

what relation authorial expectation bears to intention. Sometimes one hears the argument that a first-edition text should be preferred to the manuscript printer's copy because the author expected certain alterations to be carried out in the publisher's office. Each case must of course be looked at individually, but two general considerations should be kept in mind: first, that readers of a scholarly critical text are primarily interested in what the author did rather than what a publisher's editor did; and, second, that authors give their "approval," in various ways and for various motives, to alterations that they do not really prefer (as anyone who has written for publication knows all to well). Whether one is speaking of changes in wording or changes in punctuation, it is too simple a view to claim that a particular author expected certain kinds of alterations to be made and therefore that the published text takes precedence over the manuscript text; in a specific instance the published text may indeed deserve to take precedence, but one must be extremely cautious about attributing authorial intention or preference to alterations simply because they were passed, in one fashion or another, by the author. Thus scholarly critical editors frequently find themselves trying to reverse the activities of earlier publishers' editors. In this connection, some comments recently made by Robert Gottlieb, president of Alfred A. Knopf, are particularly striking because they emphasize respect for the author's text and define the role of the publisher's editor as essentially one of encouragement and support. "Even now," Gottlieb says, "it irritates me to know that Dickens changed the end of 'Great Expectations' because someone told him to. It's bad enough that the change was a mistake, but it's even worse to know that someone convinced him to make it."[76] What the scholarly critical editor must do, as Gottlieb does here, is to consider the motivation underlying textual changes; such an editor must try to disentangle the author's own wishes from the other elements that shaped the published text. Authorial intention cannot be equated with expectation or with acquiescence, nor can one accept an author's statement of intention at face value. A critical edition, if it is to be worthy of its name, must examine such matters critically.

The third of these large central questions is the problem of the socalled "indifferent" variant: that is, how is the editor to choose among variants in those cases in which critical analysis of the evidence finds the variants equally balanced and provides no basis for a critical choice of one over another? This problem has been much discussed under the guise of examining how to select a copy-text. It is not necessary to have a copytext at all, of course, unless there are in fact some indifferent variants. If at every point of variation in wording, punctuation, and spelling one is able to make a critical decision as to which variant represents the author's wishes, one has no need to fall back on the concept of a text with presumptive authority. Obviously one might, in such a case, choose a particular text to mark up to reflect these editorial decisions, but that text would only be serving as a convenient basis for producing printer's copy for the new edition and as nothing more. The reason so much attention has focused on the choice of copy-text is not that it is a necessary first step in critical editing but that in most cases variants appearing to be indifferent do seem to occur, so that one needs a principle for favoring one text over another. Generally speaking, an editor has less to go on when judging variants in punctuation and spelling than when judging variants in wording, and for that reason the text chosen as copy-text often supplies most of the punctuation and spelling for the critical text. But the editor is free, of course, to make rational decisions regarding spelling and punctuation when the evidence permits; conversely, variants in wording can sometimes seem indifferent, and the impasse is resolved by adopting the copy-text reading. It is not logically necessary, therefore, to distinguish spelling and punctuation from wording in arriving at a rationale for selecting a copy-text, for a copy-text is simply the text most likely to provide an authorial reading (in spelling, punctuation, or wording) at points of variation where one cannot otherwise reach a decision. The way in which spelling and punctuation may sometimes usefully be segregated from wording results from the fact that persons involved in textual transmission have frequently regarded spelling and punctuation as elements that they could alter more freely, or be less careful about, than the words; in situations where this generalization can be thought to apply, therefore, it will be one of the factors involved in editorial decisions. As for a general rationale for choosing a copy-text, one can draw on testimony from all periods, as well as on common sense and everyday experience, to show that texts can be expected to deteriorate as they are transmitted. It follows, therefore, that a copy-text should be an early

text—one as near to the author's manuscript as possible, if not that manuscript itself—whenever the individual circumstances do not suggest a different text as the more reasonable choice. When they do, then by all means another text should be chosen: the purpose of the general guideline is not to restrict thought or to force particular situations into a common mold. One begins with the variants about which one can reason (from both external and internal evidence) and reach a conclusion; for any remaining variants, one must be guided by the trustworthiness—in general or in particular respects—that one can attach to each text. If this process leads to the choice of a later rather than an earlier text as more trustworthy, then one of course chooses the later text; accepting the general observation that texts deteriorate in transmission does not mean that individual decisions cannot constitute exceptions to the generalization.

These comments on three central questions have been, as I say, intentionally brief and unadorned, both to show that the issues are essentially clear-cut (even though carrying out a responsible editorial policy may require some difficult decisions and certainly requires knowledge and insight) and also to demonstrate the cumulative effect of recent editorial discussion, for these comments, after all, grow out of that discussion and try to build on it. However ineffective some of the discussion has been, the level of discussion now is undoubtedly higher than it would have been without these years of debate. I hope that my summary can serve to suggest the most fruitful directions in which the discussion can proceed from here.

Notes