University of Virginia Library

Search this document 


expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
collapse section 
The Meaning of Copy-Text by Paul Baender
expand section 

expand section 


Page 311

The Meaning of Copy-Text
Paul Baender [*]

In the past two decades the word copy-text has become standard in editorial nomenclature. Whether we take up editions of medieval works or of works from the sixteenth through the nineteenth centuries, we may expect to find the word copy-text in the editors' introductions, usually in the crucial place where they state what texts are the basis for their own and by what principles they have emended them. So common is the usage, and as a result so official the word appears, one is uneasy about the few recent editions that do not use it — that use instead expressions like "textual basis" and "basic text." Where have these people been, one wonders, and are they ignorant of the word? On the other hand one is comfortable with those who use it. They are informed, they are sound, surely their work can be trusted. More often than not it can be trusted, at least in principle, for in recent years there has been a wide diffusion of sensible editorial procedures, attendant upon a remarkable boom in the projection of new editions. But it is time to re-examine the word copy-text. The boom in editions has led to its general use, and general use has made it ambiguous and misleading.

There is often no ambiguity, or any that matters, in the two most common meanings of copy-text. Sometimes it clearly means printer's copy, as when Donald F. Bond says that "for most of the [Spectator] the copy-text used by 12mo is the printed 8vo sheets";[1] or when Robert E. Scholes says that James Joyce "obtained a set of proofs [of Dubliners] which became the copy-text for Grant Richards' printer";[2] or when Michael Shugrue says that the "copy-text for the present edition of The Recruiting Officer is a first quarto edition [sic], now in the Yale University Library."[3] At other times copy-text means entire editions that serve as basic texts, presumably after due assessment of press variants and other discrepancies among all extant or representative copies. Such is Lloyd E. Berry's usage: "The copy text for each work [of Giles Fletcher, the elder] is the first edition";[4] and the usage of Brian Morris and Eleanor Withington: "The copy-text for


Page 312
the present edition [of John Cleveland] is the first edition of 1647 . . . for all the poems it contains."[5] Copy-text as printer's copy and as basic text are not necessarily incompatible meanings, and they may in effect mean the same thing. For instance when an editor chooses a certain book from which to take xerox sheets for printer's copy, but then inscribes on them variants appearing in other specimens of the same edition and even in different editions, he can legitimately say that the particular book is copy-text (i.e., printer's copy) or that the edition is copy-text (i.e., basic text). These usages would conflict only if two men disagreed over the same editorial choice — one claiming that a particular book should be the authoritative text as well as printer's copy, to the exclusion of variants in specimens from whatever edition; the other following the more familiar principle of eclectic texts. But even here there would be no ambiguity as to the word copy-text; there would instead be a quarrel as to which judgment, and consequently which meaning (only one, not both), was appropriate to the situation.

But ambiguity has arisen nonetheless, and for two main reasons. One is the prestige that copy-text has acquired. Often it seems to imply authority beyond its denotation, as though the term itself ratified an editor's choice of text. This is one's feeling especially in cases where editors state merely that X edition is copy-text and do not give a rationale for selecting and emending it. They seem to be revealing as incontestable fact that X edition is the "best text," better by far than any other edition, perhaps the best possible. They may not intend this impression, meaning rather that while X edition is the preponderate authority, it is all the same a point of departure for an eclectic operation. But the stature of copy-text in current usage makes the hypothetical edition sound as prestigious as the word is in our respect. And like all banner-words — such as nature, natural, reason, reasonable — this one tends toward the superlative no matter how qualified our respect.

Another reason for ambiguity is that copy-text as word and notion is not suited to the full range and complexity of editorial problems. Let me give here two brief illustrations. Number one, suppose a case where four texts independently derive from a single original. Only the original is by the author in question, but it has been destroyed. The four derivatives are by unknown reporters of indeterminate accuracy. Each apparently has as many authoritative readings as the others, yet those of all four are often different. Whatever an editor facing this situation chooses to do, he can hardly use the word copy-text as basic text or best text without being misleading. If he arbitrarily reproduces one of the four and calls it his copy-text, he misrepresents its stature vis-a-vis the others, which seem no less authoritative. If he reproduces all four he has no copy-text, unless the term


Page 313
is no longer to distinguish among texts. If he produces a conflation he has no copy-text in an honorific sense, unless he is so vain as to identify his own typescript in this way. Number two, suppose a case where there are two editions of a book for which no manuscript survives. The second edition contains many extensive revisions indisputably authorial, but since the book was electroplated when first set, accidental variants occur only in pages reset and replated because of substantive revision. To avoid the awkward procedure of inserting the authoritative second-edition readings in a reproduction of the first, the modern editor of this book uses xerox sheets of the second edition as printer's copy, restoring first-edition accidentals in the reset pages where he thinks they have been corrupted. Is the second edition therefore copy-text? One might think so, but the editor says it is the first edition, on the principle that the earliest text has authority where accidentals are in dispute. As a result the first edition sounds better than the edition used as printer's copy, yet each has its own authority. Clearly the word copy-text amounts to overkill as an expression of the special authority of the first edition.

We must go back to the history of the word to understand how, despite its inadequacies, it has become a fixture in editorial usage. The history is familiar, and I shall not linger over the details. R. B. McKerrow invented the term in 1904, meaning by it at that time simply "the text used in each particular case as the basis of mine."[6] In his Prolegomena for the Oxford Shakespeare (1939) he kept this meaning but proposed an eclectic method to accommodate authority in editions besides that of copy-text.[7] Ten years later W. W. Greg refined upon the method in "The Rationale of Copy-Text," a paper read before the English Institute and published the next year, 1950, in Studies in Bibliography.[8] In this great paper he drew his famous distinction between substantives and accidentals — substantives being "those readings that affect the author's meaning or the essence of his expression" (p. 21), and accidentals those readings that affect mainly its "formal presentation" (id.), such as spelling, punctuation and word-division. Greg agreed with McKerrow's later thinking in the Prolegomena that in general the earliest reliable printing should serve as copy-text and should control the accidentals, and he agreed also that substantive variants in later editions might replace copy-text readings. His main refinement upon McKerrow, aside from the naming and precise formulation of substantives and accidentals, was to introduce criteria for selecting among substantive variants. Greg's principles have saturated modern editors, both directly and through Fredson Bowers, who as early as 1950 began an advocacy of them which he has maintained on several occasions over the later years. These are the men chiefly responsible for the currency of copy-text,


Page 314
and one does not have to read far in them to see that the basic reason for it is the good sense of their editorial principles.

But that is so far as the general currency is concerned; why they perpetuated it among themselves is another matter, and a matter of some curiosity. For there was a radical change of editorial premises from McKerrow's preface in The Works of Thomas Nashe to such a piece as Bowers's "Current Theories of Copy-Text."[9] Back in 1904 McKerrow was operating on a single-text or best-text assumption, and so copy-text was a perfectly legitimate coinage that meant in effect printer's copy as well as "the text used . . . as the basis of mine." But later, in Greg and Bowers, and even in McKerrow's Prolegomena, the word is kept though the principles have become eclectic. We may make a few common-sense guesses as to why they kept it. They may not have realized the full implication of their eclecticism, which in the long run rules out the designation of a single text, basic text, or copy-text when there is more than one text of substantive authority. Or, at least in the beginning of the new principles, they may have retained the word copy-text both to gauge their departure from the old and to preserve continuity. The reason with objective evidence to support it is that with respect to accidentals there still remained a single-text criterion. As Greg put it: "The true theory is, I contend, that the copy-text should govern (generally) in the matter of accidentals, but that the choice of substantive readings . . . lies altogether beyond the narrow principle of the copy-text" (p. 26). The belief behind this distinction was that the texture of accidentals in the earliest reliable printing would be closest to the author's and that variants in later editions would more often than not represent corruption. We may find the same distinction drawn in recent practice, for instance in the policy statements for the Centenary Hawthorne and the California Dryden. And so a word that starts out meaning one thing ends up meaning at times that thing still, but also something else.

This aspect of the history helps explain the predicament outlined in my last illustration, where xeroxes of the second edition served as printer's copy though the first edition, because of its authority in accidentals, was called copy-text. The illustration has actually occurred. As a textual inspector sent out by the MLA's Center for Editions of American Authors, I have encountered the practice in two of three projects visited. The two did not have personnel in common, neither knew what the other was doing, each discovered the same expedient. The second instance was only slightly different from the one I have described. The work being edited existed in three most relevant forms: manuscript draft, first printing in periodical form, and first book printing. The manuscript could not be copy-text, for there were vast substantive discrepancies between it and the periodical version. On the other hand there were about three hundred substantive


Page 315
readings that first appeared in the book version. The editors decided to call the periodical version their copy-text, but they took xeroxes of the first book setting for printer's copy. Now I do not mean to criticize either project for acting in this way. On the contrary, their practice shows that they followed sound editorial procedure despite this convention of copy-text. One would hope, therefore, to find other projects making the same adjustment, and since I have found it in two out of only three, and since it was not necessary in the third, I suspect it has widely occurred already. And I suspect it will continue so long as the notion persists that there is a single form of a work which must be called copy-text because of its authority in accidentals.[10]

If this notion were always justified we could not complain: the form with accidental authority would deserve distinction by some term or other, and copy-text would do as well as any. But the notion is by no means always justified, and Greg, the man who first stated it fully, never claimed it was. He confined his principles of substantives and accidentals to printed books of more than one setting, admitting that for works transmitted in manuscript circumstances might not warrant similar discriminations of authority. He also confined his principles to books whose pre-publication forms did not survive, and this restriction cannot be over-emphasized. For it helps us see that his preference for the accidentals of the earliest reliable printing was simply a prudential rule which fitted that circumstance and not necessarily others. Without a manuscript one could not know what the author's spelling and punctuation were, and so one could not know how often the copy-text represented them correctly. One had to choose among printed texts, which brought up the consideration that authors seldom changed accidentals even in revised editions but that compositors often changed them. The best solution under the circumstances was Greg's, for it would produce a texture of accidentals as close as one might hope to what the author intended.

What Greg might have proposed concerning works with extant prepublication forms is a moot question. Bowers, who accepts his principles for the circumstances just described, thinks they also apply to these. When an author's manuscript survives, he argues, this becomes copy-text (i.e., accidental authority), which is to carry Greg's principle intact into the different situation. Bowers takes his supporting examples from Hawthorne. There are thousands of accidental variants between the manuscripts and first printings of The House of the Seven Gables, The Blithedale Romance and The Marble Faun, almost all of them initiated by the printer. Though Hawthorne accepted the changes, Bowers says we must restore the manuscript points because they were his manifest preference and because "the


Page 316
real flavor of Hawthorne, cumulatively developing in several thousand small distinctions, can be found only in the manuscript."[11] There is no disputing Bowers here. Obviously, if accidental changes are to house style or compositor's style, if they are indifferent choices, and if we do not know the author's wishes or his instructions to the printer, our judgment should tip in favor of the manuscript readings.

But the survival of pre-publication forms does not in all cases merely add another stage for a retrogressive pursuit of copy-text. It is a question of what shape these forms are in and how faithfully the author wished them to be represented. Sometimes a surviving manuscript may be a rough draft which the author polished in later drafts that do not survive. The spelling and other points of the rough draft may be conveniently careless; for example, studded with abbreviations, lacking connectives, marked with dashes to be replaced at the author's leisure by more exact punctuation. At other times an author will accept — not acquiesce in but accept — accidental and even substantive changes introduced for one reason or another by typist, editor or compositor. Since the standardization of spelling and other formal matters some authors have meant publishers to style their manuscripts according to current usage, and so a writer may indulge in whatever mistakes and inconsistencies he pleases, expecting that someone whose job it is will clean them up. It is impossible to say in general how often these practices have occurred, and it may be difficult to decide whether a particular author has followed them. But the point is that they happen and that an editor must respect them. For the survival of pre-publication forms opens to him the most personal and idiosyncratic phase in the history of his text. Two cases will seldom be the same, three more rarely. An editor must determine so far as possible what in his case was the relation between pre-publication and printed forms, and he must be prepared to make a point by point assessment of variants, on the assumption that no rule may emerge for the sweeping authority of either form. He must also try to distinguish intention from oversight among the author's manuscript eccentricities, and he should choose the intention even if it means suppressing what Bowers calls the author's "real flavor." When we apply Greg's principles to printed texts we seek to recover what an author intended, and we adopt toward him an Arminian attitude: his wishes count for more than the corruptions that have happened to him. But the application of Greg to pre-publication forms can lead to an attitude of Calvinist rigor: an author's intentions may count for less than his mistakes and waywardness.

If, then, Greg's principles are not equally appropriate in every textual situation, and in some not at all appropriate, his sense of the word copy-text should be restricted to where it belongs. But if this is so, we should drop the word and keep only the sense. For it can do no good, and has


Page 317
done harm by way of ambiguity and other mis-communication, to tie a word to a special condition, particularly when the word has different meanings in the same field of scholarship. Keeping Greg's sense, we need only say that when our option is among printed texts, the earliest reliable setting should be authoritative in accidentals and that an eclectic procedure should be adopted for substantives. Where is the necessity of copy-text or any other single word to express these principles? Is the accidental authority of the earliest reliable setting not clear enough in the statement that it is authoritative? When our option includes pre-publication forms as well as printed texts, again we need only describe the emergent authorities, such as they may be. But in this situation, where much more is known about the development of the text, determinations of authority should be through particulars rather than general rules. Thus, for Hawthorne, we might say that the kinds of accidental variants between manuscript and first printing suggest an imposition of house style and that the manuscript readings are consequently preferable. For another writer we might have to say something different; that the accidental variants suggest both house style and author's galley revision and that the present text makes its choices through a hopeful eclecticism. In situations like these, where so much is known, formulas and key words that signify formulas could lull us into a undiscriminating certitude. It is when little is known that our need is greatest for general rules, to keep our guesses logical.[12]

So long as the word copy-text has a definite and consistent meaning there is no need to suppress it. At the outset I mentioned its two most common meanings in current usage, printer's copy and basic text, and I suggest that only one of them, printer's copy, is generally viable. Every edition of whatever work involves transmission of copy from editor to printer, whether typescript, book pages, photocopies of book pages, or some other form not yet devised. Thus copy-text in the sense of printer's copy could never be invidious. It would mean that version and physical form of text an editor chose to send his printer, and it would perform this designation as a simple matter of fact, without prejudice. If an editor prepared a typescript, for instance because of conflation or because the work survived only in manuscript, he would call it his copy-text. If he took xeroxes of a printing he would ideally choose the version easiest for a compositor to read and requiring the least emendation, and when no version fitted both conditions he would choose expediently between a clear printing that required many emendations and a vague one that required few. Nothing here precludes Greg's principles for selecting substantives and accidentals. Indeed a major advantage of copy-text as printer's copy is that it precludes


Page 318
no principles for the establishment of text because it is a function of none. Thus an editor would have to establish his text through principles relevant to his situation, and he would have to describe their bearing in detail: he could not rely upon a naming of copy-text to support doctrinaire procedures or to suggest a rationale he did not provide.

But do we need copy-text as well as printer's copy, two expressions for the same thing, when because of past associations one of them may still imply a higher quality? Perhaps not, though it might be convenient to reserve copy-text for what an editor sends his printer and to use the other term only historically, to mean what an author sent his printer. If copy-text in this sense turns out to be superfluous, it had better be discarded altogether. A loaded word is particularly difficult to un-load when its neutral sense is not compelling, and so we might quickly end where we are now, with copy-text also meaning basic text, best text, and text authoritative in accidentals. A return to this usage, much less a continuation of it, would be a hindrance to precise editing. Principles and methods for the establishment of text are already complex, and with the increased application of statistics and computers they will become more sophisticated. There is also a growing sensitivity among editors to special circumstances that qualify general rules. For these reasons we must be alert to necessary changes in editorial vocabulary. We must be ready to invent new terms for new kinds of facts and formulas, at the same time keeping terms like substantive and accidental so long as they refer to discrete phenomena. We must be prepared also to redefine or abandon words that have lost general relevance and common meaning.



Read on 3 November 1967 before the Bibliography section of the Midwest Modern Language Association, meeting at Lafayette, Indiana.


Donald F. Bond, "The Text of the Spectator," Studies in Bibliography, V (1952-53), 113.


Robert E. Scholes, "The Text of Dubliners: 'The Dead",' Studies in Bibliography, XV (1962), 199.


George Farquhar, The Recruiting Officer, ed. Michael Shugrue (1965), p. ix.


The English Works of Giles Fletcher, the Elder, ed. Lloyd E. Berry (1964), p. 50.


The Poems of John Cleveland, ed. Brian Morris and Eleanor Withington (1967), p. lxxvi.


The Works of Thomas Nashe, ed. Ronald B. McKerrow (1904), I, [vii].


See p. 12n.


W. W. Greg, "The Rationale of Copy-Text," Studies in Bibliography, III (1950), 19-36.


Fredson Bowers, "Current Theories of Copy-Text," Modern Philology, XLVIII (1950), 12-20.


In the case of the first instance, the editors might have identified the "Second Edition" (originally so called) as copy-text had they considered it an issue, not an edition.


Fredson Bowers, "Some Principles for Scholarly Editions of Nineteenth-Century American Authors," Studies in Bibliography, XVII (1964), 226.


In his latest article Bowers himself is perfectly aware of these qualifications, though he does not reconsider the word copy-text in the light of them. See "Old Wine in New Bottles: Problems of Machine Printing," in Editing Nineteenth-Century Texts, ed. John M. Robson (1967), pp. 20, 28.