University of Virginia Library

The Rationale of Copy-Text [*]
by
W. W. GREG

WHEN, IN HIS EDITION OF NASHE, McKERROW invented the term 'copy-text', he was merely giving a name to a conception already familiar, and he used it in a general sense to indicate that early text of a work which an editor selected as the basis of his own. Later, as we shall see, he gave it a somewhat different and more restricted meaning. It is this change in conception and its implications that I wish to consider.

The idea of treating some one text, usually of course a manuscript, as possessing over-riding authority originated among classical scholars, though something similar may no doubt be traced in the work of biblical critics. So long as purely eclectic methods prevailed, any preference for one manuscript over another, if it showed itself, was of course arbitrary; but when, towards the middle of last century, Lachmann and others introduced the genealogical classification of manuscripts as a principle of textual criticism, this appeared to provide at least some scientific basis for the conception of the most authoritative text. The genealogical method was the greatest advance ever made in this field, but its introduction was not unaccompanied by error. For lack of logical analysis, it led, at the hands of its less discriminating exponents, to an attempt to reduce textual criticism to a code of mechanical rules. There was just this much excuse, that the method did make it possible to sweep away mechanically a great deal of rubbish. What its more hasty devotees failed to understand, or at any rate sufficiently to bear in mind, was that authority is never absolute, but only relative. Thus a school arose, mainly in Germany, that

taught that if a manuscript could be shown to be generally more correct than any other and to have descended from the archetype independently of other lines of transmission, it was 'scientific' to follow its readings whenever they were not manifestly impossible. It was this fallacy that Housman exposed with devastating sarcasm. He had only to point out that 'Chance and the common course of nature will not bring it to pass that the readings of a ms are right wherever they are possible and impossible wherever they are wrong'.[1] That if a scribe makes a mistake he will inevitably produce nonsense is the tacit and wholly unwarranted assumption of the school in question,[2] and it is one that naturally commends itself to those who believe themselves capable of distinguishing between sense and nonsense, but who know themselves incapable of distinguishing between right and wrong. Unfortunately the attractions of a mechanical method misled many who were capable of better things.

There is one important respect in which the editing of classical texts differs from that of English. In the former it is the common practice, for fairly obvious reasons, to normalize the spelling, so that (apart from emendation) the function of an editor is limited to choosing between those manuscript readings that offer significant variants. In English it is now usual to preserve the spelling of the earliest or it may be some other selected text. Thus it will be seen that the conception of 'copy-text' does not present itself to the classical and to the English editor in quite the same way; indeed, if I am right in the view I am about to put forward, the classical theory of the 'best' or 'most authoritative' manuscript, whether it be held in a reasonable or in an obviously fallacious form, has really nothing to do with the English theory of 'copy-text' at all.

I do not wish to argue the case of 'old spelling' versus 'modern spelling'; I accept the view now prevalent among English scholars. But I cannot avoid some reference to the ground on which

present practice is based, since it is intimately connected with my own views on copy-text. The former practice of modernizing the spelling of English works is no longer popular with editors, since spelling is now recognized as an essential characteristic of an author, or at least of his time and locality. So far as my knowledge goes, the alternative of normalization has not been seriously explored, but its philological difficulties are clearly considerable.[3] Whether, with the advance of linguistic science, it will some day be possible to establish a standard spelling for a particular period or district or author, or whether the historical circumstances in which our language has developed must always forbid any attempt of the sort (at any rate before comparatively recent times) I am not competent to say; but I agree with what appears to be the general opinion that such an attempt would at present only result in confusion and misrepresentation. It is therefore the modern editorial practice to choose whatever extant text may be supposed to represent most nearly what the author wrote and to follow it with the least possible alteration. But here we need to draw a distinction between the significant, or as I shall call them 'substantive', readings of the text, those namely that affect the author's meaning or the essence of his expression, and others, such in general as spelling, punctuation, word-division, and the like, affecting mainly its formal presentation, which may be regarded as the accidents, or as I shall call them 'accidentals', of the text.[4] The distinction is not arbitrary or theoretical, but has an immediate bearing on textual criticism, for scribes (or compositors) may in general be expected to react, and experience shows that they generally do react, differently to the two categories. As regards substantive readings their aim may be assumed to be to reproduce exactly those of their copy, though they will doubtless sometimes depart from them accidentally and may even, for one reason or another, do so intentionally: as regards accidentals they will normally follow their own habits or inclination, though they may, for various reasons and to varying degrees, be influenced by their copy. Thus a contemporary manuscript will at least preserve the spelling of the period, and may even retain some of the author's own, while it may at the same time depart frequently from the wording of the original: on the other hand a later transcript of the same original may reproduce the wording with essential accuracy while completely modernizing the spelling. Since, then, it is only on grounds of expediency, and in consequence either of philological ignorance or of linguistic circumstances, that we select a particular original as our copy-text, I suggest that it is only in the matter of accidentals that we are bound (within reason) to follow it, and that in respect of substantive readings we have exactly the same liberty (and obligation) of choice as has a classical editor, or as we should have were it a modernized text that we were preparing.[5]

But the distinction has not been generally recognized, and has never, so far as I am aware, been explicitly drawn.[6] This is not surprising. The battle between 'old spelling' and 'modern spelling' was fought out over works written for the most part between 1550 and 1650, and for which the original authorities are therefore as a rule printed editions. Now printed editions usually form an ancestral series, in which each is derived from its immediate predecessor; whereas the extant manuscripts of any work have usually only a collateral relationship, each being derived from the original independently, or more or less independently, of the others. Thus in the case of printed books, and in the absence of revision in a later edition, it is normally the first edition alone that can claim authority, and this authority naturally extends to substantive readings and accidentals alike. There was, therefore,

little to force the distinction upon the notice of editors of works of the sixteenth and seventeenth centuries, and it apparently never occurred to them that some fundamental difference of editorial method might be called for in the rare cases in which a later edition had been revised by the author or in which there existed more than one 'substantive' edition of comparable authority.[7] Had they been more familiar with works transmitted in manuscript, they might possibly have reconsidered their methods and been led to draw the distinction I am suggesting. For although the underlying principles of textual criticism are, of course, the same in the case of works transmitted in manuscripts and in print, particular circumstances differ, and certain aspects of the common principles may emerge more clearly in the one case than in the other. However, since the idea of copy-text originated and has generally been applied in connexion with the editing of printed books, it is such that I shall mainly consider, and in what follows reference may be understood as confined to them unless manuscripts are specifically mentioned.

The distinction I am proposing between substantive readings and accidentals, or at any rate its relevance to the question of copy-text, was clearly not present to McKerrow's mind when in 1904 he published the second volume of his edition of the Works of Thomas Nashe, which included The Unfortunate Traveller. Collation of the early editions of this romance led him to the conclusion that the second, advertised on the title as 'Newly corrected and augmented', had in fact been revised by the author, but at the same time that not all the alterations could with certainty be ascribed to him.[8] He nevertheless proceeded to enunciate the rule that 'if an editor has reason to suppose that a certain text embodies later corrections than any other, and at the same time has no ground for disbelieving that these corrections, or some of them at

least, are the work of the author, he has no choice but to make that text the basis of his reprint'.[9] The italics are mine.[10] This is applying with a vengeance the principle that I once approvingly described as 'maintaining the integrity of the copy-text'. But it must be pointed out that there are in fact two quite distinct principles involved. One, put in more general form, is that if, for whatever reason, a particular authority be on the whole preferred, an editor is bound to accept all its substantive readings (if not manifestly impossible). This is the old fallacy of the 'best text', and may be taken to be now generally rejected. The other principle, also put in general form, is that whatever particular authority be preferred, whether as being revised or as generally preserving the substantive readings more faithfully than any other, it must be taken as copy-text, that is to say that it must also be followed in the matter of accidentals. This is the principle that interests us at the moment, and it is one that McKerrow himself came, at least partly, to question.

In 1939 McKerrow published his Prolegomena for the Oxford Shakespeare, and he would not have been the critic he was if his views had not undergone some changes in the course of thirty-five years. One was in respect of revision. He had come to the opinion that to take a reprint, even a revised reprint, as copy-text was indefensible. Whatever may be the relation of a particular substantive edition to the author's manuscript (provided that there is any transcriptional link at all) it stands to reason that the relation of a reprint of that edition must be more remote. If then, putting aside all question of revision, a particular substantive edition has an over-riding claim to be taken as copy-text, to displace it in favour of a reprint, whether revised or not, means receding at least one step further from the author's original in so far as the general form of the text is concerned.[11] Some such considerations must have been in McKerrow's mind when he wrote (Prolegomena, pp. 17-18): 'Even if, however, we were to assure

ourselves . . . that certain corrections found in a later edition of a play were of Shakespearian authority, it would not by any means follow that that edition should be used as the copy-text of a reprint.[12] It would undoubtedly be necessary to incorporate these corrections in our text, but . . . it seems evident that . . . this later edition will (except for the corrections) deviate more widely than the earliest print from the author's original manuscript. . . . [Thus] the nearest approach to our ideal . . . will be produced by using the earliest "good" print as copy-text and inserting into it, from the first edition which contains them, such corrections as appear to us to be derived from the author.' This is a clear statement of the position, and in it he draws exactly the distinction between substantive readings (in the form of corrections) and accidentals (or general texture) on which I am insisting. He then, however, relapsed into heresy in the matter of the substantive readings. Having spoken, as above, of the need to introduce 'such corrections as appear to us to be derived from the author', he seems to have feared conceding too much to eclecticism, and he proceeded: 'We are not to regard the "goodness" of a reading in and by itself, or to consider whether it appeals to our aesthetic sensibilities or not; we are to consider whether a particular edition taken as a whole contains variants from the edition from which it was otherwise printed which could not reasonably be attributed to an ordinary press-corrector, but by reason of their style, point, and what we may call inner harmony with the spirit of the play as a whole, seem likely to be the work of the author: and once having decided this to our satisfaction we must accept all the alterations of that edition, saving any which seem obvious blunders or misprints.' We can see clearly enough what he had in mind, namely that the evidence of correction (under which head he presumably intended to include revision) must be considered as a whole; but he failed to add the equally important proviso that the alterations must also be of a piece (and not, as in The Unfortunate Traveller, of apparently disparate origin) before we can be called upon to accept them all. As he states it his canon is open to exactly the same objections as the 'most authoritative manuscript' theory in classical editing.

McKerrow was, therefore, in his later work quite conscious

of the distinction between substantive readings and accidentals, in so far as the problem of revision is concerned. But he never applied the conception to cases in which we have more than one substantive text, as in Hamlet and perhaps in 2 Henry IV, Troilus and Cressida, and Othello. Presumably he would have argued that since faithfulness to the wording of the author was one of the criteria he laid down for determining the choice of the copy-text, it was an editor's duty to follow its substantive readings with a minimum of interference.

We may assume that neither McKerrow nor other editors of the conservative school imagined that such a procedure would always result in establishing the authentic text of the original; what they believed was that from it less harm would result than from opening the door to individual choice among variants, since it substituted an objective for a subjective method of determination. This is, I think, open to question. It is impossible to exclude individual judgement from editorial procedure: it operates of necessity in the all-important matter of the choice of copy-text and in the minor one of deciding what readings are possible and what not; why, therefore, should the choice between possible readings be withdrawn from its competence? Uniformity of result at the hands of different editors is worth little if it means only uniformity in error; and it may not be too optimistic a belief that the judgement of an editor, fallible as it must necessarily be, is likely to bring us closer to what the author wrote than the enforcement of an arbitrary rule.

The true theory is, I contend, that the copy-text should govern (generally) in the matter of accidentals, but that the choice between substantive readings belongs to the general theory of textual criticism and lies altogether beyond the narrow principle of the copy-text. Thus it may happen that in a critical edition the text rightly chosen as copy may not by any means be the one that supplies most substantive readings in cases of variation. The failure to make this distinction and to apply this principle has naturally led to too close and too general a reliance upon the text chosen as basis for an edition, and there has arisen what may be called the tyranny of the copy-text, a tyranny that has, in my opinion, vitiated much of the best editorial work of the past generation.

I will give a couple of examples of the sort of thing I mean that I have lately come across in the course of my own work. They are all the more suitable as illustrations since they occur in texts edited by scholars of recognized authority, neither of whom is particularly subject to the tyranny in question. One is from the edition of Marlowe's Doctor Faustus by Professor F. S. Boas (1932). The editor, rightly I think, took the so-called B-text (1616) as the basis of his own, correcting it where necessary by comparison with the A-text (1604).[13] Now a famous line in Faustus's opening soliloquy runs in 1604,

and in 1616, Here Oncaymœon is now recognized as standing for on cay mœ on or ?ν Κα? μ? ?ν: but this was not understood at the time, and Oeconomy was substituted in reprints of the A-text in 1609 and 1611, and thence taken over by the B-text. The change, however, produced a rather awkward line, and in 1616 the and was introduced as a metrical accommodation. In the first half of the line Boas rightly restored the reading implied in A; but in the second half he retained, out of deference to his copy-text, the and whose only object was to accommodate the reading he had rejected in the first. One could hardly find a better example of the contradictions to which a mechanical following of the copy-text may lead.[14]

My other instance is from The Gipsies Metamorphosed as edited by Dr. Percy Simpson among the masques of Ben Jonson in 1941. He took as his copy-text the Huntington manuscript, and I entirely agree with his choice. In this, and in Simpson's edition, a line of the ribald Cock Lorel ballad runs (sir-reverence!),

whereas for blewe other authorities have flirted. Now, the meaning of flirted is not immediately apparent, for no appropriate sense of the word is recorded. There is, however, a rare use of the substantive flirt for a sudden gust of wind, and it is impossible to doubt that this is what Jonson had in mind, for no scribe or compositor could have invented the reading flirted. It follows that in the manuscript blewe is nothing but the conjecture of a scribe who did not understand his original: only the mesmeric influence of the copy-text could obscure so obvious a fact.[15]

I give these examples merely to illustrate the kind of error that, in modern editions of English works, often results from undue deference to the copy-text. This reliance on one particular authority results from the desire for an objective theory of text-construction and a distrust, often no doubt justified, of the operation of individual judgement. The attitude may be explained historically as a natural and largely salutary reaction against the methods of earlier editors. Dissatisfied with the results of eclectic freedom and reliance on personal taste, critics sought to establish some sort of mechanical apparatus for dealing with textual problems that should lead to uniform results independent of the operator. Their efforts were not altogether unattended by success. One result was the recognition of the general worthlessness of reprints. And even in the more difficult field of manuscript transmission it is true that formal rules will carry us part of the way: they can at least effect a preliminary clearing of the ground. This I sought to show in my essay on The Calculus of Variants (1927); but in the course of investigation it became clear that there is a definite limit to the field over which formal rules are applicable. Between readings of equal extrinsic authority no rules of the sort can decide, since by their very nature it is only to extrinsic relations that they are relevant. The choice is necessarily a matter for editorial judgement, and an editor who declines or is unable to exercise his judgement and falls back on some arbitrary canon, such as the authority of the copy-text, is in fact abdicating his editorial function. Yet this is what has been frequently commended as 'scientific'—'streng wissenschaftlich' in the prevalent

idiom—and the result is that what many editors have done is to produce, not editions of their authors' works at all, but only editions of particular authorities for those works, a course that may be perfectly legitimate in itself, but was not the one they were professedly pursuing.

This by way, more or less, of digression. At the risk of repetition I should like to recapitulate my view of the position of copy-text in editorial procedure. The thesis I am arguing is that the historical circumstances of the English language make it necessary to adopt in formal matters the guidance of some particular early text. If the several extant texts of a work form an ancestral series, the earliest will naturally be selected, and since this will not only come nearest to the author's original in accidentals, but also (revision apart) most faithfully preserve the correct readings where substantive variants are in question, everything is straightforward, and the conservative treatment of the copy-text is justified. But whenever there is more than one substantive text of comparable authority,[16] then although it will still be necessary to choose one of them as copy-text, and to follow it in accidentals, this copy-text can be allowed no over-riding or even preponderant authority so far as substantive readings are concerned. The choice between these, in cases of variation, will be determined partly by the opinion the editor may form respecting the nature of the copy from which each substantive edition was printed, which is a matter of external authority; partly by the intrinsic authority of the several texts as judged by the relative frequency of manifest errors therein; and partly by the editor's judgement of the intrinsic claims of individual readings to originality—in other words their intrinsic merit, so long as by 'merit' we mean the likelihood of their being what the author wrote rather than their appeal to the individual taste of the editor.

Such, as I see it, is the general theory of copy-text. But there remain a number of subsidiary questions that it may be worth-while to discuss. One is the degree of faithfulness with which the copy-text should be reproduced. Since the adoption of a copy-text is a matter of convenience rather than of principle—being

imposed on us either by linguistic circumstances or our own philological ignorance—it follows that there is no reason for treating it as sacrosanct, even apart from the question of substantive variation. Every editor aiming at a critical edition will, of course, correct scribal or typographical errors. He will also correct readings in accordance with any errata included in the edition taken as copy-text. I see no reason why he should not alter misleading or eccentric spellings which he is satisfied emanate from the scribe or compositor and not from the author. If the punctuation is persistently erroneous or defective an editor may prefer to discard it altogether to make way for one of his own. He is, I think, at liberty to do so, provided that he gives due weight to the original in deciding on his own, and that he records the alteration whenever the sense is appreciably affected. Much the same applies to the use of capitals and italics. I should favour expanding contractions (except perhaps when dealing with an author's holograph) so long as ambiguities and abnormalities are recorded. A critical edition does not seem to me a suitable place in which to record the graphic peculiarities of particular texts,[17] and in this respect the copy-text is only one among others. These, however, are all matters within the discretion of an editor: I am only concerned to uphold his liberty of judgement.

Some minor points arise when it becomes necessary to replace a reading of the copy-text by one derived from another source. It need not, I think, be copied in the exact form in which it there appears. Suppose that the copy-text follows the earlier convention in the use of u and v, and the source from which the reading is taken follows the later. Naturally in transferring the reading from the latter to the former it would be made to conform to the earlier convention. I would go further. Suppose that the copy-text reads 'hazard', but that we have reason to believe that the correct reading is 'venture': suppose further that whenever this word occurs in the copy-text it is in the form 'venter': then 'venter', I maintain, is the form we should adopt. In like manner editorial emendations should be made to conform to the habitual spelling of the copy-text.

In the case of rival substantive editions the choice between

substantive variants is, I have explained, generally independent of the copy-text. Perhaps one concession should be made. Suppose that the claims of two readings, one in the copy-text and one in some other authority, appear to be exactly balanced: what then should an editor do? In such a case, 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.[18]

Much more important, and difficult, are the problems that arise in connexion with revision. McKerrow seems only to mention correction, but I think he must have intended to include revision, so long as this falls short of complete rewriting: in any case the principle is the same. I have already considered the practice he advocated (pp. 23-25)—namely that an editor should take the original edition as his copy-text and introduce into it all the substantive variants of the revised reprint, other than manifest errors—and have explained that I regard it as too sweeping and mechanical. The emendation that I proposed (p. 26) is, I think, theoretically sufficient, but from a practical point of view it lacks precision. In a case of revision or correction the normal procedure would be for the author to send the printer either a list of the alterations to be made or else a corrected copy of an earlier edition. In setting up the new edition we may suppose that the printer would incorporate the alterations thus indicated by the author; but it must be assumed that he would also introduce a normal amount of unauthorized variation of his own.[19] The problem that faces the editor is to distinguish between the two categories. I suggest the following frankly subjective procedure. Granting that the fact of revision (or correction) is established,

an editor should in every case of variation ask himself (1) whether the original reading is one that can reasonably be attributed to the author, and (2) whether the later reading is one that the author can reasonably be supposed to have substituted for the former. If the answer to the first question is negative, then the later reading should be accepted as at least possibly an authoritative correction (unless, of course, it is itself incredible). If the answer to (1) is affirmative and the answer to (2) is negative, the original reading should be retained. If the answers to both questions are affirmative, then the later reading should be presumed to be due to revision and admitted into the text, whether the editor himself considers it an improvement or not. It will be observed that one implication of this procedure is that a later variant that 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—to the scandal of faithful followers of McKerrow. I do not, of course, pretend that my procedure will lead to consistently correct results, but I think that the results, if less uniform, will be on the whole preferable to those achieved through following any mechanical rule. I am, no doubt, presupposing an editor of reasonable competence; but if an editor is really incompetent, I doubt whether it much matters what procedure he adopts: he may indeed do less harm with some than with others, he will do little good with any. And in any case, I consider that it would be disastrous to curb the liberty of competent editors in the hope of preventing fools from behaving after their kind.

I will give one illustration of the procedure in operation, taken again from Jonson's Masque of Gipsies, a work that is known to have been extensively revised for a later performance. At one point the text of the original version runs as follows,

whereas the texts of the revised version replace maunds by stalkes. Now, maund is a recognized canting term meaning to beg, and there is not the least doubt that it is what Jonson originally wrote. Further, it might well be argued that it is less likely that he should have displaced it in revision by a comparatively commonplace alternative, than that a scribe should have altered a rather unusual word that he failed to understand—just as we know that, in a line already quoted (p. 27), a scribe altered flirted to blewe. I should myself incline to this view were it not that at another point Jonson in revision added the lines, where stalk, in the sense of going stealthily, is used almost as a technical term. In view of this I do not think it unreasonable to suppose that Jonson himself substituted stalkes for maunds from a desire to avoid the implication that his aristocratic Gipsies were beggars, and I conclude that it must be allowed to pass as (at least possibly) a correction, though no reasonable critic would prefer it to the original.

With McKerrow's view that in all normal cases of correction or revision the original edition should still be taken as the copy-text, I am in complete agreement. But not all cases are normal, as McKerrow himself recognized. While advocating, in the passage already quoted (p. 25), that the earliest 'good' edition should be taken as copy-text and corrections incorporated in it, he added the proviso, 'unless we could show that the [revised] edition in question (or the copy from which it had been printed) had been gone over and corrected throughout by' the author (my italics). This proviso is not in fact very explicit, but it clearly assumes that there are (or at least may be) cases in which an editor would be justified in taking a revised reprint as his copy-text, and it may be worth inquiring what these supposed cases are. If a work has been entirely rewritten, and is printed from a new manuscript, the question does not arise, since the revised edition will be a substantive one, and as such will presumably be chosen by the editor as his copy-text. But short of this, an author, wishing to make corrections or alterations in his work, may not merely hand the printer a revised copy of an earlier edition, but himself supervise the printing of the new edition and correct the proofs as the sheets go through the press. In such a case it may be argued that even though the earlier edition, if printed from his own manuscript, will preserve the author's individual peculiarities more faithfully than the revised reprint, he must nevertheless be assumed

to have taken responsibility for the latter in respect of accidentals no less than substantive readings, and that it is therefore the revised reprint that should be taken as copy-text.

The classical example is afforded by the plays in the 1616 folio of Ben Jonson's Works. In this it appears that even the largely recast Every Man in his Humour was not set up from an independent manuscript but from a much corrected copy of the quarto of 1601. That Jonson revised the proofs of the folio has indeed been disputed, but Simpson is most likely correct in supposing that he did so, and he was almost certainly responsible for the numerous corrections made while the sheets were in process of printing. Simpson's consequent decision to take the folio for his copy-text for the plays it contains will doubtless be approved by most critics. I at least have no wish to dispute his choice.[20] Only I would point out—and here I think Dr. Simpson would agree with me—that even in this case the procedure involves some sacrifice of individuality. For example, I notice that in the text of Sejanus as printed by him there are twenty-eight instances of the Jonsonian 'Apostrophus' (an apostrophe indicating the elision of a vowel that is nevertheless retained in printing) but of these only half actually appear in the folio, the rest he has introduced from the quarto. This amounts to an admission that in some respects at least the quarto preserves the formal aspect of the author's original more faithfully than the folio.

The fact is that cases of revision differ so greatly in circumstances and character that it seems impossible to lay down any hard and fast rule as to when an editor should take the original edition as his copy-text and when the revised reprint. All that can be said is that if the original be selected, then the author's corrections must be incorporated; and that if the reprint be selected, then the original reading must be restored when that of the reprint is due to unauthorized variation. Thus the editor cannot escape the responsibility of distinguishing to the best of his ability between the two categories. No juggling with copy-text will relieve him of the duty and necessity of exercizing his own judgement.

In conclusion I should like to examine this problem of revision

and copy-text a little closer. In the case of a work like Sejanus, in which correction or revision has been slight, it would obviously be possible to take the quarto as the copy-text and introduce into it whatever authoritative alterations the folio may supply; and indeed, were one editing the play independently, this would be the natural course to pursue. But a text like that of Every Man in his Humour presents an entirely different problem. In the folio revision and reproduction are so blended that it would seem impossible to disentangle intentional from what may be fortuitous variation, and injudicious to make the attempt. An editor of the revised version has no choice but to take the folio as his copy-text. It would appear therefore that a reprint may in practice be forced upon an editor as copy-text by the nature of the revision itself, quite apart from the question whether or not the author exercized any supervision over its printing.

This has a bearing upon another class of texts, in which a reprint was revised, not by the author, but through comparison with some more authoritative manuscript. Instances are Shakespeare's Richard III and King Lear. Of both much the best text is supplied by the folio of 1623; but this is not a substantive text, but one set up from a copy of an earlier quarto that had been extensively corrected by collation with a manuscript preserved in the playhouse. So great and so detailed appears to have been the revision that it would be an almost impossible task to distinguish between variation due to the corrector and that due to the compositor,[21] and an editor has no choice but to take the folio as copy-text. Indeed, this would in any case be incumbent upon him for a different reason; for the folio texts are in some parts connected by transcriptional continuity with the author's manuscript, whereas the quartos contain only reported texts, whose accidental characteristics can be of no authority whatever. At the same time, analogy with Every Man in his Humour suggests that even had the quartos of Richard III and King Lear possessed higher authority than in fact they do, the choice of copy-text must yet have been the same.

I began this discussion in the hope of clearing my own mind as well as others' on a rather obscure though not unimportant matter

of editorial practice. I have done something to sort out my own ideas: others must judge for themselves. If they disagree, it is up to them to maintain some different point of view. My desire is rather to provoke discussion than to lay down the law.

Notes