University of Virginia Library

I

Greg's contribution to the 1949 session of the English Institute, "The Rationale of Copy-Text"—read for Greg by J. M. Osborn on 8 September 1949—was first published in the third (1950-51) volume of Studies in Bibliography (pp. 19-36). (There is a certain appropriateness, therefore, in re-examining the essay, on the twenty-fifth anniversary of its original appearance, in the pages of the same journal.) Since that time it has been republished in the posthumous volume of Greg's Collected Papers (1966), edited by J. C. Maxwell, who incorporated into its text a few minor revisions and a new footnote, as indicated by Greg in his working papers.[4] The essay is not long or complicated and is expressed with Greg's usual clarity. That such an essay should have given rise to so much discussion, and even controversy, is not surprising, however, for it has the kind of simplicity frequently characteristic of great concepts—a sweeping simplicity that results from having penetrated beyond peripheral complexities and arrived at the heart of a problem. Just as it is not easy to achieve such simplicity, neither is it always easy for others to follow or accept it.

Greg begins by referring to the first use of the term "copy-text"—by R. B. McKerrow in 1904 in his edition of Nashe—and sketches the history of the idea of "the most authoritative text"; it is evident, from this kind of beginning and from later references to McKerrow's and his own changes of position, that he is presenting his ideas on copy-text as the outgrowth of an evolving train of thought extending back over many years. Indeed, his opening paragraph says nothing about putting forth a new theory but only that he wishes to consider the "conception" and "implications" of a change in McKerrow's position. Although he soon admits (p. 377) that he is drawing a distinction which "has not been generally recognized," his emphasis is not on the

novelty of his contribution but rather on the way in which it seems but a natural step in the line of thinking already pursued both by him and by McKerrow. In effect he is saying that he has finally come to recognize something which he had overlooked earlier and something toward which McKerrow had gradually been moving.

It is important to notice the historical framework of Greg's essay: for Greg, stepping into the discussion at a particular point in its development, accepts without further analysis certain ideas about scholarly editing—two in particular—which he feels have already been adequately established. First, he makes clear that he rejects "purely eclectic methods," in which an editor has no restraints placed on his freedom to choose among variant readings on the basis of his subjective judgments of their aesthetic appropriateness; the "genealogical method," developed by Lachmann and his successors in the nineteenth century, was, he says, "the greatest advance ever made in this field," because it provided a more objective basis for preferring one text over another. McKerrow's concept of "copy-text"—taking the term to mean, in Greg's words, "that early text of a work which an editor selected as the basis of his own"—is clearly placed in the context of the genealogical method, for it implies that an editor has determined, through genealogical analysis, the "most authoritative text" and therefore the one to which his own text should adhere. By introducing Housman's criticism of the mechanical application of this procedure (the fallacy of believing that the readings of the "authoritative text" which are not manifestly impossible are in fact correct), Greg suggests the direction in which his argument is to move. But he sees no necessity to argue the general superiority of genealogical methods over eclectic ones; at mid-twentieth century this superiority can simply be asserted. A second assumption is that one can reject without discussion the notion of choosing the last edition published during the author's lifetime as the most authoritative. Placing his comment in a footnote—and in the past tense—to suggest how little attention the idea deserves, Greg says, "I have above ignored the practice of some eccentric editors who took as copy-text for a work the latest edition printed in the author's lifetime, on the assumption, presumably, that he revised each edition as it appeared. The textual results were naturally deplorable" (p. 378). Obviously Greg is not saying that one should ignore late revisions which one has reason to think are authorial; but, he is implying, it is no longer necessary to bother refuting the assumption that the last edition in the author's lifetime is automatically the most authoritative.

Without going over ground which he regards as already established,

then, Greg begins to reflect on current editorial practice and observes that the situation facing editors of English texts is different from that facing editors of classical texts, since the preference for "old-spelling" editions is now "prevalent among English scholars," whereas editors of classical texts normalize the spelling. Greg explicitly says that he does not wish to argue the virtues of old-spelling editions but accepts this "prevalent" view—that is to say, he accepts the view that editions of English works for scholars' use should not involve normalized or modernized spelling and punctuation. It should be clear, therefore, that his essay deals with one particular, if basic, kind of edition and implies nothing about the relative merits of modernized editions for other purposes—a point sometimes overlooked. If the editor of English texts properly follows the general tradition of the genealogical method inaugurated by classical editors, and if he must be concerned with the spelling and punctuation of his text in a way different from classical editors, it follows that his conception of copy-text must contain an additional element. In fact, viewed in this way, as Greg says, "the classical theory of the 'best' or 'most authoritative' manuscript . . . has really nothing to do with the English theory of 'copy-text' at all" (p. 375)—because, under the classical theory, the spelling and punctuation are not involved in selecting the copy-text.

By the beginning of the fourth paragraph of his essay, Greg has led the reader, with astonishing ease, to see the current situation in English editing against the background of its development and to anticipate the distinction he is about to set forth between, on the one hand, spelling and punctuation, and, on the other, the words themselves. The rhetorical strategy of the essay demands proceeding explicitly to make this distinction before returning to an examination of McKerrow's changing position (which thereby takes on a new dimension), and this remarkable fourth paragraph (pp. 375-377) contains the essence of what is now referred to as "Greg's theory of copy-text." First of all, it makes the point that an old-spelling edition must rely on some contemporary document, for the "philological difficulties" of attempting to recreate or establish spellings for a particular author at a particular time and place are overwhelming. Second, in view of this practical necessity, it says, one must distinguish between the actual words of a text and their spelling and punctuation:

The explicit separation of these classes for separate editorial treatment is one of Greg's key contributions; the third major point of the paragraph is what that separate treatment amounts to. Separate treatment is justified, the argument goes, because copyists and compositors are known to treat the two categories differently; since they generally attempt to reproduce accurately the substantives of their copy but frequently are guided by their own preferences in matters of accidentals, it follows that later transcripts of a work may depart considerably from earlier ones in accidentals and at the same time be very close to them in substantives. What an editor should do, therefore, as a practical routine, is first to determine the early text which is to be his copy-text; then, Greg says, In other words, because a copyist or a compositor reproduces substantives more faithfully than accidentals, substantive variants in later transcripts or editions are more likely to be worth editorial consideration as possible authorial revisions than are variants in accidentals.

Now a few observations are worth making in regard to what Greg does and does not say in this statement of his "theory"—particularly as an anticipation of some of the points which, as we shall see, have been raised in recent years. To begin with, while the terms "substantive" and "accidental" are not very happy choices,[5] what is crucial to

the theory is the distinction itself, and one should not be distracted from it by other associations which these words have. The terms have by now become so well established in editorial commentary that it would be foolish to attempt to change them, even though their use tends unfortunately to give the impression to the general reader that editing involves an arcane vocabulary and mysterious concepts. The situation is ironic because Greg did not pretend to be dealing with any abstruse concepts: he merely hoped that these two words could serve as a shorthand means for making a distinction between what are popularly regarded as content and form in verbal expression, a distinction with which everyone, in one way or another, has come in contact. Indeed, he goes out of his way to emphasize the fact that he is not setting forth a philosophical theory about the nature of language but is only drawing a practical distinction for use in the business of editing.[6] Naturally he is aware that content and form are never completely separable and that the line separating meaning and formal presentation in written language is not distinct (and philosophically raises complex issues); but for his purposes it is enough to append a footnote (p. 376) acknowledging "an intermediate class of word-forms about the assignment of which opinions may differ and which may have to be treated differently in dealing with the work of different scribes." Since the purpose of the substantive-accidental division is to assist the editor in deciding what variants in a text can reasonably be attributed to the copyist or compositor rather than the author, the focus is pragmatic—on the habits of individuals—and Greg is therefore more concerned with providing a suggestive approach, which can be used with flexibility to meet various situations, than in defining as philosophic concepts two mutually exclusive terms. The procedural recommendation which concludes Greg's paragraph is similarly couched in practical, and flexible, terms: the reason for selecting a copy-text in the first place is the limited nature of historical knowledge about accidentals (the copy-text is selected "on grounds of expediency, and in consequence either of philological ignorance or of linguistic circumstances"), and therefore one should follow the copy-text in regard to accidentals—but "within reason." This last phrase underscores Greg's approach: one follows the "theory" when there is no persuasive reason for doing otherwise, but when one has reason to depart from it, a rigid application of it would be foolish. Because the editor generally has fewer means for rationally determining authorial readings in accidentals than in substantives, he generally follows the copy-text in accidentals; but Greg is not asking him to fly in the face of reason by adhering to this procedure in situations which are exceptions to the generalization. Nowhere does Greg claim that following his rationale will invariably produce "correct" readings; what he suggests is that it offers the safest approach when one has otherwise no particular reason for choosing one reading over another as authorial. The theory clearly is one of expediency.

The skillful organization of Greg's essay is nowhere better exemplified than in his return to the subject of McKerrow in the pages following this basic exposition of his theory. The rigidity of McKerrow's approach is the more evident in contrast, and the reader is now in a position to see its limitations; at the same time he recognizes how Greg's ideas developed from McKerrow's and how McKerrow was on the verge of the same insight as Greg. In the 1904 Nashe (which Greg quotes), McKerrow had held firmly to the view that an editor should take as his copy-text the latest edition which could convincingly be shown to contain authorial revisions; so long as some of the variants in that edition were authorial, all its readings should be accepted (since conceivably they could all be authorial), except when they were obviously impossible. McKerrow allowed for some editorial discretion in the determination of what was obviously impossible, but in general he was determined to preserve the "integrity" of individual texts. But by 1939, when he published his Prolegomena for the Oxford Shakespeare, he had come to believe that a later edition, even one with authorial revisions, should not serve as copy-text, for—with the exception of those revisions—it would be less likely to reflect the author's manuscript than an earlier edition, which stood that much closer to the manuscript. He thus understood, without explicitly stating, something very close to the distinction between substantives and accidentals, since he now believed that the edition closest to the manuscript preserved the general texture of the work better than later editions and that authorial revisions should be incorporated into the text of that edition. Although this position represented a considerable move away

from his earlier fear of eclecticism,[7] he was still not ready to allow an editor to combine readings from more than two editions. When the editor believed a particular edition to contain authorial revisions, he said, all the variants in that edition "which could not reasonably be attributed to an ordinary press-corrector" (that is, in general, all the substantive variants) must be accepted into the copy-text. By the time of his death, therefore, McKerrow was well on his way to the position finally advanced by Greg,[8] the essential difference between the two being in the amount of responsibility given to editorial judgment. For McKerrow, the editor uses his judgment in determining what edition should be copy-text, what edition, if any, contains authorial revisions, and what readings are impossible, but he cannot go further and reject some of the variants in that authorially revised edition as not authorial.[9] For Greg, the editor who has already made certain basic decisions should be allowed to go on and choose among the possibly authorial variants. The effort to eliminate as much editorial decision as possible, he believes, is misguided: Again Greg's emphasis is on the use of reason and discretion, as it is in the brief summary which follows immediately: "the copy-text should govern (generally) in the matter of accidentals, but . . . the choice between substantive readings belongs to the general theory of textual criticism and lies altogether beyond the narrow principle of the copy-text" (pp. 381-382). Greg is careful here to insert a qualifying adverb even in the first part of his statement, which deals with accidentals and thus the more mechanical part of his theory; but in the second part he makes clear that the handling of substantive variants is a matter of critical judgment and cannot be regarded as mechanical in any sense. Not to recognize that substantives and accidentals must be treated in different ways, he points out, has led in the past to a "tyranny of the copy-text"—a tyranny because its readings were thrust on the editor, without the benefit of his critical thinking about their merits.

The remainder of Greg's essay, amounting to about half of it, consists of illustrative examples and discussions of particular problems in the application of the theory but does not add any essential point to the basic idea set forth economically in the first half. After citing examples from F. S. Boas's edition (1932) of Marlowe's Doctor Faustus and Percy Simpson's edition (1941) of Jonson's The Gipsies Metamorphosed to show the operation of the "tyranny of the copy-text,"[10] Greg provides a second brief recapitulation of his rationale (pp. 384-385), reiterating the limitations of mechanical rules and concentrating on the nature of the editorial judgment required for dealing with substantive variants. That judgment depends partly on an evaluation of the circumstances of the production of the editions in which those variants appear and partly on the relative reliability of those editions

as suggested by the number of "manifest errors" in them; but the heart of the matter is the editor's evaluation of particular variants in terms of "the likelihood of their being what the author wrote rather than their appeal to the individual taste of the editor" (p. 385). Then, to provide more practical help, Greg expands on three points already introduced. First, he suggests that an editor may legitimately decide to alter some of the accidentals of the copy-text and thus provides a gloss on the expressions "within reason" and "generally" which he had inserted parenthetically in his earlier statements about following the accidentals of the copy-text. Spelling or punctuation known to be at variance with the author's can be altered, for instance, and, when substantive emendations are made on the basis of later texts, the spelling of such words can be made to conform with the habitual spelling (if there is one) of the copy-text. Second, he restates in somewhat more detail his belief that an editor should not accept from an authorially revised edition any substantive variant that seems obviously incorrect, that seems not to be a reading which the author would have inserted, or that seems completely indifferent. The latter point illustrates once again the expedient nature of what Greg is proposing: if a variant appears so indifferent to the editor that he has no basis for arguing either for or against its adoption, then he simply follows the copy-text reading as a practical means for deciding what to do. "In such a case," Greg points out, "while there can be no logical reason for giving preference to the copy-text, in practice, if there is no reason for altering its reading, the obvious thing seems to be to let it stand" (p. 386). Third, he makes explicit (pp. 389-390) what was only implied before, that the choice of copy-text itself varies with circumstances and that situations arise in which one must choose a revised edition as copy-text, as when an author is thought to have overseen a revised edition so carefully that its accidentals as well as its substantives must be taken to carry his approval, or when revision is so complex or pervasive that it is not meaningful to think in terms of emending the unrevised text with later readings (Every Man in His Humour, Richard III, and King Lear are cited).

In connection with all three of these points Greg again defends the use of editorial judgment. When discussing the first he says, "These [decisions to alter accidentals], however, are all matters within the discretion of an editor: I am only concerned to uphold his liberty of judgement" (p. 386). In his discussion of substantives he repeats the view emphatically:

And in the third instance, dealing with the choice of copy-text, he states that no "hard and fast rule" can be laid down but that, whatever text is chosen, the editor "cannot escape the responsibility of distinguishing to the best of his ability" between authorial revision and "unauthorized variation": "No juggling with copy-text will relieve him of the duty and necessity of exercising his own judgement" (p. 390). This sentiment is clearly the dominant motif of the essay; if McKerrow had been reacting against nineteenth-century eclecticism in restricting the role of editorial judgment, Greg is here turning toward more reliance on judgment, but within a framework that does not encourage undisciplined eclecticism. It is in keeping with his approach throughout that Greg ends by saying, "My desire is rather to provoke discussion than to lay down the law."

I hope that my account of Greg's essay, by its very repetitiousness, has shown that the essay itself consists of repeated statements of a simple idea. Three times he presents a concise summary of his theory followed by a discussion of particular points implied by it, as if he were turning over an object in his hand, focusing his attention alternately on the piece as a whole and on certain of its details. The simplicity of his proposal is certainly one of its most remarkable features and is a natural result of the emphasis on individual judgment, for a methodology inevitably becomes more complicated the more one tries to substitute rules for judgment in the handling of the various situations that may arise. In somewhat blunt language, Greg's theory amounts to this: it tells the editor what to do when he otherwise does not know what to do. If he does know otherwise—that is, if his analysis of all available external and internal evidence (including, of course, his own intimate knowledge of the author and the period) convinces him that a particular text comes closest in all respects to the author's wishes or that a particular variant is the author's revision—then he does not need further guidance. But when there remains a doubt in his mind, after thorough analysis, about whether, for example, the author gave close attention to the punctuation of a revised edition or

whether a particular altered wording, in a text which contains many clearly authorial revisions, was the author's, the editor does need further help, since he has gone as far as reasoning can take him and the results are inconclusive. All that Greg suggests, in effect, is that the editor can most sensibly extricate himself from this situation by keeping two points in mind: (1) successive editions based on earlier editions become increasingly divergent from the earliest edition in the sequence, particularly in such matters as punctuation and spelling, not merely through carelessness but through the natural tendency of compositors to utilize their own habitual forms; (2) when an author makes revisions in a later edition, he may be likely to give considerably less attention to spelling and punctuation than to the words themselves, and even some of the differences in wording in a revised edition may in fact result from the process of resetting rather than from the author's revision. It follows that the editor who chooses the edition closest to the author's manuscript as his copy-text when he does not have strong reason for choosing a later one, and who follows the readings of that copy-text when he does not have strong reason to believe them erroneous or to believe that a later variant in wording (or, more rarely, in punctuation or spelling) is the author's—that such an editor is maximizing his chances of incorporating the author's intended readings in his text. No one would claim—and Greg specifically does not—that this procedure always results in the correct choices, but it tells an editor how to proceed when he most needs such advice (when he has exhausted the available evidence without reaching a decision) and it is more satisfying than tossing a coin (since there is at least a rationale involved, based on a generalization about the incidence of human error and the behavior of human beings in dealing with written language). The fundamental common sense of this approach can be seen foreshadowed in Samuel Johnson's comments on the editing of Shakespeare, when he says that "though much credit is not due to the fidelity, nor any to the judgement of the first publishers, yet they who had the copy before their eyes were more likely to read it right, than we who read it only by imagination."[11] The probabilities favor the correctness of the first edition, and it makes sense to rely on that edition except when there is compelling evidence for not doing so.

Expressed in this way—which emphasizes the flexibility and lack of dogmatism basic to Greg's position—this "rationale of copy-text" would seem to apply to all situations. But it is important to raise the

question of its universality, for Greg's primary interest, after all, was in the printed drama of the English Renaissance, and all his illustrations are taken from that literature. Did he believe that his rationale was more widely applicable? He was dealing with a period from which relatively few manuscripts have survived, but can the same procedure be applied to texts for which manuscripts do survive? He was working with a period in which greater variations in spelling were tolerable than in later times and in which any editorial supervision of a printed text normally occurred in the printing shop rather than, as later, in the publisher's office with its more highly developed editorial routine; but can Greg's rationale be applied to the products which emerged from the very different publishing circumstances of later periods? Greg's own answer to these questions, I think it can be plainly inferred from his essay, would be Yes. It is true that he limits himself in his illustrations to the field he knows best and limits his more abstract discussion for the most part to printed books, but there are indications that he is thinking in broader terms. For example, in that crucial fourth paragraph, distinguishing substantives and accidentals, he twice refers to "scribes" and "compositors" simultaneously, suggesting that the way human beings react to the two categories is the same regardless of whether they are copying by hand or setting type. He goes on, in the paragraph which follows, to restrict himself to printed books for the historical reason that "the idea of copy-text originated and has generally been applied in connexion with the editing of printed books" (p. 378). The focus of the essay, it must be remembered, is historical: a new approach to editing is set forth as a corrective to what had been developing over the previous century. Since the principal developments in editorial theory had taken place in connection with the editing of Elizabethan and Jacobean drama, it was natural that he should set forth his criticisms of current procedure with reference to the same field—and convenient, also, since that was his own area of competence. But he clearly implies that he is dealing with a larger principle that could be illustrated in other ways than the one he has chosen. Indeed, he suggests that the editors he is criticizing might have taken a different approach if they had been more familiar with the problems of variation in works transmitted in manuscript. And then he adds: The implication certainly is that he is concerned with a basic concept[12] which might not be clear to one who has dealt only with a particular class of problems. And while his illustrations come from Renaissance drama, some of them do involve authorial proof-correction (Every Man in His Humour) and revised editions incorporating corrections derived from authorial manuscript (Richard III and King Lear). In any event, his whole approach, stressing expediency and judgment, suggests that he thinks of his procedure as one capable of fitting widely varied situations. When an editor judges that he has sufficient evidence for proceeding in a particular way, he has no need for a plan of expediency; but a lack of sufficient evidence is a common occurrence in dealing with works of every period, and Greg's rationale commands respect in such situations because it is based on what observation shows to be characteristic human behavior. If I have set forth accurately here what Greg says, then it would appear to be a self-evident proposition that his recommended procedure would serve in handling editorial problems involving manuscripts as well as printed books, arising in twentieth-century literature as well as sixteenth.

There is one kind of editorial problem, however, which clearly lies outside the scope of Greg's essay. To place presumptive authority for accidentals, as a general rule, in the edition closest to the author's manuscript presupposes an ancestral series, in which the line of editions—with each edition based (for the most part, at least) on preceding ones—leads back to the manuscript. Although some of Greg's examples involve complicated variations (such as the revisions incorporated in the folio text of Every Man in His Humour), in which a later edition is chosen as copy-text because of the extent and nature of fresh authority (authorial revision or recourse to authorial manuscripts), those examples do not include situations in which two or more texts stand in exactly the same genealogical relationship to a lost ancestor, with no earlier texts surviving. In such a case, Greg's approach offers no help in selecting a copy-text, for no one of these texts is nearer the manuscript (or the antecedent text) than any other. The inapplicability of Greg's rationale to this kind of situation is obvious,

once it is pointed out, but it has only recently been examined in detail. Fredson Bowers was confronted with the problem in editing Stephen Crane's syndicated newspaper pieces: the variant texts of a given piece, as they appeared in various newspapers, are all equidistant from the syndicate's proofs which had been sent to those newspapers; in the absence of the proofs, the editor is faced with several texts, any one of which could be chosen as copy-text under Greg's rationale. The solution, as Bowers sets it forth in "Multiple Authority: New Problems and Concepts of Copy-Text,"[13] is to combine the features of these "radiating texts," as he calls them, through a statistical and critical analysis of the variants. In effect, one has to construct a copy-text, and the more surviving texts there are the more accurately can the common ancestor (the lost syndicate proof) be reconstructed. From that point on, naturally, Greg's rationale takes over, and the text thus constructed may be emended with variants from later printings, as may happen with an ordinary copy-text. The essential difference is that, in the case of radiating texts, no one document can serve as copy-text, for no one of the radiating texts can be presumed to have reproduced the accidentals of the syndicate proof more accurately than another. Bowers's detailed exposition of his solution therefore becomes a major supplement to Greg; his essay—which incidentally offers an extremely useful statement of Greg's position—deserves to be taken as a companion piece to Greg's "Rationale," and the two essays together provide a comprehensive editorial theory.

Bowers's discussion of radiating texts, in other words, does not invalidate Greg's theory in any sense, but it does show one respect in which that theory is not all-encompassing. No comparable supplement to Greg's theory has been made in the twenty-five years since its first

appearance, though many questions have been raised. But these questions (such as the extent to which the idiosyncrasies of nineteenth- or twentieth-century authors' manuscripts should be preserved in print), often interesting in themselves, involve matters of editorial judgment, not the basic theory. It is unfortunately true that such questions have frequently been posed as an attack on the theory; and the failure to distinguish between the theory itself and the individual decisions of editors who are following the theory has rendered much of the discussion less useful than it might have been, if not wholly beside the point. My own summary of Greg in these pages has tried to emphasize those elements of his essay which anticipate the later criticisms. Seemingly it takes many words to explain something which is simple and many assertions to proclaim lack of dogmatism; but the simplicity and lack of dogmatism of Greg's rationale have apparently not been perceived by a number of people, for many of their criticisms are undercut by a recognition of those qualities. A renewed close examination of Greg's essay does not suggest to me any reason to question Bowers's description of its thesis as "the great contribution of this century to textual criticism."[14]