 | |  |
Notes
Greg's paper was solicited for a bibliographical section in the 1949 English Institute meetings in New York City, was read in absentia on September 8, and printed in Studies in Bibliography, 3 (1950-51), 19-36. It was reprinted, with a few minor changes, in The Collected Papers of Sir Walter Greg, ed. J. C. Maxwell (Oxford at the Clarendon Press, 1966), pp. 374-391. Quotations are taken from this latter source.
A dissenter is John M. Robson, who in 1966 gave a reasoned defence of a choice of the final edition as copy-text for the Toronto edition of John Stuart Mill: see "Principles and Methods in the Collected Edition of John Stuart Mill," Editing Nineteenth Century Texts, ed. J. M. Robson (University of Toronto Press, 1967), pp. 96-122. If I follow correctly the implications of the argument, the grounds for Dr. Robson's disagreement are as much ideological as practical, and—although the comparison is not elaborated—rest on "the differences between printing and publishing practices in the Renaissance on the one hand, and the nineteenth century on the other, [which] suggest a proper divergence in editing procedures, a divergence as justified as that between procedures in editing classical and Renaissance texts" (p. 114). This is a question I propose to discuss. Greg's rationale has not found favor on the European continent. The Soviet Russian textologists seem to be most advanced in the acceptance of the function of editorial judgment acting within the analysis of textual transmission; but it is unclear what their practising attitude is to the matter of accidentals in relation to the rationale's precepts, and thus their acceptance of critical editing procedures appears to be confined to the second half of Greg's propositions without taking in the first. Otherwise, it would seem that the German school, basically one of variorum editing, rules. The best statement in English of this European editorial theory and practice may be found in Hans Zeller, "A New Approach to the Critical Constitution of Literary Texts," Studies in Bibliography, 28 (1975), 231-264.
Warner Barnes, "Nineteenth-Century Editorial Problems: A Selective Bibliography," op. cit., ed. Robson, p. 129. Barnes does not specify, but the views would need to comprise (a) the choice of copy-text on the basis of the authority of its accidentals, not that of the substantives, and (b) in a critical edition the necessity for the editor to exercise his judgment in the selection of substantive readings from more than one authoritative document, not necessarily the copy-text.
In his excellent general survey, combining theory and practice, which should be obligatory reading for all students, G. Thomas Tanselle argues that at least implicitly Greg was aware that his rationale was applicable to literature later than that of the Renaissance: "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," "Greg's Theory of Copy-Text and the Editing of American Literature," Studies in Bibliography, 28 (1975), 181. The point is not an important one, but some clarification may be useful. My personal acquaintance with Sir Walter Greg would lead me to believe that in this essay, as in all his work, he was cautious about asserting that his concepts were applicable to a wider field of evidence than that in which he chose to set them. The reader might endeavor to extrapolate them to test their broader efficacy, but Greg would have hesitated to suggest in any manner that this was his own intention or even wish. I suppose that knowing he was on firm ground in the period of his experience, he disliked to introduce the possibility that doubt might arise about his position if a reader began to make exceptions suggested by his special knowledge of other fields. Obviously, a reader sympathetic with Greg's position has a duty to test it in circumstances not envisaged in the essay; but Greg's personal attitude, I venture to say, would have been that he does so at his own risk. In my view, Dr. Tanselle's evidence for his suggestion that Greg was thinking in broader terms than the set of conditions he was illustrating from the Renaissance is based on a misinterpretation. For instance, one piece of evidence brought forward of Greg's reference to scribes and compositors (p. 181) does indeed suggest, as he asserts, that Greg took it that human beings react similarly whether copying by hand or setting type. But Greg was thinking quite plainly of medieval and Renaissance scribes, and of Renaissance compositors setting by hand. He would have distrusted the application of this generalization, in as precise a way as he had utilized it, to a period, say, of typist copying or linotype setting. The different conditions produce in many respects different kinds and possibly different quantities of error. Other indications that Dr. Tanselle finds of Greg's broader implicit purpose are true enough if one recognizes that Greg is looking back to medieval manuscript texts, many of whose problems are applicable to the scribal copy behind Elizabethan books and present in a few respects relatively close analogies with the hand-setting of type in the sixteenth and seventeenth centuries. Thus, when Greg refers to another period than his own, he deliberately looks back, and not forward as has been suggested.
Misunderstandings leading to adverse criticism of Greg's position are perhaps even more common than doctrinaire rigidity in its application. Tanselle (p. 197 and note) remarks that "one cannot help regarding many of the recent discussions (both favorable and unfavorable) . . . as naïve and parochial, and frequently as uninformed or misinformed." For examples, see p. 201 ff., principally pp. 203-204, 207-219.
See my "Current Theories of Copy-Text, with an Illustration from Dryden," "Multiple Authority: New Concepts of Copy-Texts," and "Remarks on Eclectic Texts," collected in Essays in Bibliography, Text, and Editing (University Press of Virginia, 1975); also, "Textual Criticism," in The Aims and Methods of Scholarship in the Modern Languages and Literatures, ed. J. Thorpe (MLA, rev. ed., 1970), pp. 29-54. The textual introductions in a number of editions approved by the Center for Editions of American Authors (now the Center for Scholarly Editions) have made valuable contributions. Dr. Tanselle's important survey article has already been mentioned and the various references in his footnotes offer a useful conspectus of the numerous discussions not mentioned here.
Briefly, although certain syntactical punctuation like a period to end a sentence was recognized, in general the punctuation marks indicate not syntactical units and relationships, as in the present day, but instead a graduated series of pauses rising in weight from comma to semicolon to colon. The present distinctive use of the colon versus semicolon was unknown, and it is highly doubtful that the principle of restriction and nonrestriction was present in anyone's mind for phrases and clauses, apart from pauses: it certainly did not govern the presence or absence of punctuation. The loose Elizabethan syntax and its punctuation work together very smoothly once one becomes accustomed to its flexibility and occasional uncalculated ambiguities. Editors of modernized texts have difficulty fitting the sprawling syntax into neat modern units of syntactical punctuation and find themselves forced into an excessive use of dashes, or into J. Dover Wilson's idiosyncratic use of a series of dots.
Although critics loosely talk about house rules in the Elizabethan printing shop, little evidence exists to support their position. Personal example, and the master-apprentice system, might encourage some degree of uniformity in the treatment of certain accidentals: thus there is evidence that Jaggard's apprentice Compositor E rather quickly acquired many of the prominent characteristics of the shop's premier Compositor B. Nevertheless, the spelling of the First Folio is so individualistic that the exact pages set by each of six compositors can be established on the evidence of their spelling habits, combined with a few idiosyncratic typographical conventions. Hence some compositors could be called more 'modern' than others within the same shop. It follows that at least certain accidental characteristics of any printed book would vary according to the compositor who set the pages, and uniformity of accidental texture within any book is ordinarily non-existent. There was so little house style in Jaggard's shop, even in matters outside of spelling, that no fixed system was enforced in the First Folio about the italicizing or non-italicizing of the names of countries and cities, or of territorial titles. Even simple typographical mechanics were not standardized, as witness Compositor C's habit of setting a space before a comma in certain circumstances.
The printer's housestyling of books in the nineteenth and the publisher's styling as the twentieth century wears on might in some sense impose uniformity of system on an unsystematic author; but this is seldom referable to the spelling, for example, since in most respects general standards of acceptability had been established and the stylization of spelling is largely (although not entirely) a question of the correction of error, with some occasional standardization in the United States according to Webster's. Light or heavy punctuation systems and matters of word-division will bulk larger in the styling of accidental changes, whereas in Shakespeare's day the problems of recovering authorial spelling for its intrinsic interest and as a guide to emendation has proved to be of more concern than ordinary questions of authorial punctuation. For a recent study of the statements in nineteenth-century printers' manuals about the following of copy, see John Bush Jones, "Victorian 'Readers' and Modern Editors: Attitudes and Accidentals Revisited," PBSA, 71 (1977), 47-59. On p. 58 the conclusion is drawn: "Granted the dominant view was that accidentals were the printer's particular concern, nevertheless as early as 1808 Nightingale was advocating at least limited querying rather than arbitrary correction, and after mid-century there was an unsteady but visible progression of opinion toward preserving authorial accidentals." On p. 59 the important distinction is made: "the fact (according to De Vinne) that house styling of accidentals took precedence over authorial intentions in late nineteenth-century periodicals raises doubts about their reliability as copy texts for short works and serialized novels first published in magazines." I have observed this distinction to hold in the essays that William James published in popular magazines like McClure's in the early twentieth century but not in general in the learned journals where his work first appeared.
Since Greg's day the science (or art) of identifying compositors by their spelling and sometimes by certain mechanical characteristics has advanced rapidly. But the further step of identifying the underlying characteristics of the copy is still awaiting more information. When no sample of autograph manuscript is preserved, to identify an early author's variable accidentals as distinct from the compositor's is conjectural at best.
As, for instance, in the mainly collateral texts of such plays as Hamlet, Othello, and Troilus and Cressida, in which two different authorities are represented by the early Quarto and the later Folio texts. Regardless of the substantives, an editor will choose the Quartos as copy-text because the accidentals reflect a texture more nearly contemporary with Shakespeare whereas the Folio's texture is posthumous and modernized. (This is to ignore other considerations such as the physically derived nature of the Folio from the Quarto texts owing to the use as Folio printer's copy of a Quarto brought into general conformity with a second authority by scribal annotation.) The question of these collateral texts will be discussed later, as well as the somewhat confusing use that Greg makes of 'bad quartos' as illustration of revised texts. At the present, however, it may clarify matters to point to Greg's speculation (p. 381) about McKerrow's position if he had lived to apply the Prolegomena form of his editorial theory to plays with more than one authoritative text "as in Hamlet and perhaps in 2 Henry IV, Troilus and Cressida, and Othello." What Greg has in mind here is the uncertainties of his own day about the authority of the variant texts of 2 Henry IV, Troilus, and Othello as against the then certainty that the Folio version of Hamlet represented a substantially different authorial text from that of the Second Quarto. At the present time scholars are inclined to believe that, perhaps even more than in Hamlet, the texts of Quarto and Folio Othello possibly, and certainly of Troilus, are different in their origin more by reason of the author than of the conditions of transmission, with QF 2 Henry IV considerably more in doubt as of true independent authority. However, Greg does not have in mind in his distinction the bibliographical question whether, in using an annotated quarto as printer's copy, the Folio text can properly be called 'substantive' in a pure sense in comparison with a play that exhibits no physical derivation. Here only 2 Henry IV among Shakespeare's non-bad quarto texts is any longer seriously in question whether the Folio was set from manuscript or from a marked-up quarto. It must be admitted, however, that although 'bibliographical links' establish the derivation from the Quarto of the Folio texts of Troilus and of Othello, these links have not been so firmly demonstrated in Hamlet. Hence at least in theory the anomalies produced by the Folio reproduction of the Hamlet Second Quarto's substantive errors are susceptible of conjecture that they derive from some common lost manuscript. If the Second Quarto were indeed set from Shakespeare's working papers (for which again there is only a belief based on the unusual spelling and the apparent difficulties in legibility of the manuscript), then the notion of the derivation of common error is more difficult to maintain, for it would require the scribe making the copy that ultimately produced the Folio to have mistaken various words in the identical manner as did the compositor of the Second Quarto—more of a possibility than a probability.
Renaissance scholars always have in mind the possibility if not the probability that a scribal transcript may intervene between the holograph and the printed book, and sometimes more than one. Hence among the problems in assessing the accidental authority of an early printed text is that of the difficulty of finding evidence as to whether the printer's copy was a holograph or a derived scribal manuscript. A late nineteenth-century parallel would be the question of setting from holograph or from professionally made typescript; in the twentieth century the question may arise whether a book were set from an authorial typescript or from a professional one ordered by the author's agent and very likely never looked over by the author before type was set from it. It is true that some international publishing conditions may in later times reproduce a few of the problems of early collateral manuscript texts, as in the quadruple transmission of Stephen Crane's "Death and the Child" without authorial revision accounting for this multiplicity of documents. But the special medieval conditions of copies being made from other copies at different removes is not present.
The use of the word reprint in Greg's essay is subject to some possible confusion. In a strictly textual sense it means a resetting of an earlier printed edition that has not been subject to authorial correction or revision and hence is completely unauthoritative in every respect. In a looser sense it is often employed to mean any edition reset from another, whether or not the author or some other agent had altered the copy from which the later edition was typeset or had similarly intervened during the course of printing as by proof-correction. The modern reader must be on the lookout for the different shades of technical meaning in writings on textual matters. In hand printing a reprint normally implies a resetting of the type; but the word is also used (instead of reimpression) for the rare cases of another printing from standing type, corresponding in modern terms to the 'reprinting' of a book from the same plates. The confusion derives from the fact that in one sense reprint is a textual word with certain important critical implications; in the other it is a mechanical term for the printing process. A purist would confine reprint to the first and reimpression to the second; but many textual critics employ the term in its mixed or general sense, trusting to the context to prevent misunderstanding. Greg's objection on p. 379, note 1, to McKerrow's very loose use of reprint to describe a modern old-spelling edition based on an early copy-text is not nitpicking, for McKerrow's choice of the term is in fact misleading for anything other than a diplomatic reprint, or transcript, of some early authority. Greg usefully writes 'revised reprint' for a reset edition modified by the author or by reference to some other authority.
The parenthesis refers exclusively to reported texts, or 'bad quartos,' which are memorial reconstructions by some agent(s) other than the author and hence break the transcriptional link with a holograph since their accidentals can have no relation to those of the original text being memorially reproduced.
This is the conventional method in all ages for authorial revision. In the seventeenth century John Dryden marked up a copy of the first edition of The Indian Emperour to serve as printer's copy for the revised second, and he did the same with the second edition to produce setting copy for the revised third. Henry Fielding not quite a hundred years later annotated a copy of the third edition of Tom Jones to provide the printer's copy for the only revised edition, the fourth. William James in the late nineteenth and early twentieth centuries customarily annotated pages from his periodical articles when he reprinted them in revised form in a book collection. I do not know of examples, but I can imagine that at the present day if the authorial annotations were very copious and difficult to read, to spare expense in the printing some publishers (or agents) might make up a typescript of the authorially marked printed copy for the printer's convenience. Especially if the author were not informed of this fact (as is probable he would not be) and so could not recollate the entire typescript text, not just his annotations, typist variations could enter the transmission indistinguishable from printer's variants not caught and altered in proof unless the typescript (and original copy) were preserved as evidence for future generations of textual scholars. There is an unsubstantiated rumor that for at least one or two plays in an early volume of the Cambridge University Press's critical edition of Beaumont and Fletcher's dramatic canon (1966—), the Press made such a typescript from the difficult marked-up xeroxes of the original copy-texts although, if so, the editors were given back only the xeroxes for proofreading copy.
Actually, within this group the technical authority is by no means comparable in anything but a general sense. How much of the differences between the Quarto and Folio texts of 2 Henry IV stem from the scribe who copied the manuscript behind the Folio text is unknown; indeed, the whole lost history of the transmission of these texts before print is highly conjectural. Although the Second Quarto text of Hamlet may have been set from Shakespeare's own working papers, it seems clear that the Folio version derives from a prompt-book that from time to time had been adjusted for performance and may not even be the original book; hence in practical terms few modern editors have taken the Folio Hamlet as literally of comparable authority with the Quarto for the generality of its readings. There is some chance, however, that the two versions of Troilus have an immediate or very close relationship to Shakespeare and in the technical sense are comparable for their substantives. Even if one stemmed from a fair scribal copy and the other from Shakespeare's own papers, I fancy the overall authority is little disturbed, no matter how one might view the formal details of the scribal copy, principally because the odds seem very good that an original and a revised Shakespearean document lie behind the two forms. My intentionally neutral statement that the author 'was equally concerned in the composition of both versions' would also cover the case of Othello (albeit in a quite different sense) if Greg is correct (and not Walker) in his estimate that Quarto and Folio each derives from the versions produced by two scribes copying the same Shakespearean manuscript for different purposes.
For an extended analysis of McKerrow's definition of 'substantive' edition and of Greg's subsequent expansion of this term to include mixed texts like Richard III, Troilus, or Jonson's Catiline, whether resulting from scribal conflation with another authority or from authorial revision, see my "McKerrow, Greg, and 'Substantive Edition,'" forthcoming in The Library. The problem in the "Rationale" resulted from Greg's abandonment of his position about mixed texts as held in the Editorial Problem and his return to McKerrow's narrow definition. However, two particular remarks in the "Rationale" can be interpreted only in terms of his original acceptance of mixed texts as substantive and so suggest either incompletely digested revision of his essay or else an inadvertent anachronism in his thinking. It may be remarked that in 1942 Greg believed that the Folio Hamlet and Othello were printed from independent manuscripts, not from annotated quartos (Editorial Problem, pp. 64, 108). In his Preface to the second edition in 1951 he acknowledged the scholarship that had shown Othello and confirmed Troilus as set from quartos but he allowed his earlier statements about Hamlet to stand.
The shift might be confusing to a reader intent on following every point in the argument. In practical terms whether an edition is substantive or not, in McKerrow's sense, is of some editorial importance. If two editions are printed from different manuscripts, concurrence of readings is generally good evidence for their authority (barring common error transmitted from some antecedent document), whereas if two editions are bibliographically linked this concurrence might mean only a printing error in the first passed on to the second by scribal oversight, as is usually supposed to have happened with good kissing carrion and pious bonds (or bands) in Hamlet Q2 and F1.
At this time, especially, printers much preferred even heavily annotated printed copy to manuscripts as setting copy. In my own view, not all of the annotated copy was necessarily furnished Jaggard by the Folio editors: it is just as possible in cases when the company had provided some manuscript other than the prompt-book (not allowed from their possession) that Jaggard himself had an available quarto annotated by reference to the manuscript for the convenience of his compositors. If a prompt-book had been the only manuscript, the theatrical company would necessarily have been responsible for the marked-up printer's copy that had been conflated with it.
The only question (which Greg does not mention) might be the practicability of retaining a bad quarto's accidentals as more representative of an Elizabethan author's period than, say, a good text printed much later which modernized the formal features of some now lost early document. But the chances for such occurrences are so limited as to be ignored. In the Cambridge edition of Marlowe, it seemed pedantic to attempt to restore the bad quarto's first-edition accidentals in the text of Doctor Faustus for those portions of the later good quarto based on the annotated third edition of the bad. In order to avoid a top-heavy apparatus of emendation, convenience suggested the propriety (since in neither case was there evidence for accidental authority) of using the later good edition as copy-text. Interestingly, on page 111 of Prolegomena McKerrow discusses the possibility of using the accidentals of the bad Q1 Richard III and decides that it was mainly convenience that dictated his choice of the derived Folio as copy-text instead.
Actually convenience cannot be ignored when the choice seems relatively indifferent because of one's ignorance. In both Beggars Bush and The Woman's Prize the manuscript behind the 1647 Folio seems to be closer in text to the authors' working papers but the printing was about fifteen to twenty years later than the inscription of the preserved manuscript. Nevertheless, the punctuation of the manuscripts would have called for an extensive apparatus of emendation had they been selected as copy-text. The difference in formal features did not seem sufficient to justify the considerable reworking of the accidentals that would have been necessary in a critical edition, especially since it was possible that the system of the Folio, being set from an older manuscript than the scribal transcripts, might still be closer to the authors. When nothing is known of an author's accidentals, a practising editor (as distinguished from a textual theorist) is likely to have some sympathy with Greg's remarks about expediency.
In his general account (IX, 72) Simpson also remarks on the fact that the press-corrections are heavier in the older works and drop off markedly in the newer, which he takes to be an indication that Jonson was bringing these older plays up to date. One may comment that if Jonson had indeed read pre-printing proofs, this disparity need not have existed since then his in-press alterations should have shown light polishing and second thoughts plus the correction of compositorial failure to follow copy. Instead, the evidence suggests the serious correction and revision at the press that is to be expected if these were the only proofs he saw. The disparity in press-correction between the older and newer plays also has another implication in that it automatically reduces assumptions about the amount of compositorial transmissional error corrected in press and increases the amount of Jonson's independent revision to be found in the press-variants in the older plays. In turn this suggests that the original marking of setting copy may have been less extensive than usually thought in respect to the kind of details represented by his press-corrections since these would mainly be new changes.
The closest parallels I can think of concern the relation between a preserved early draft manuscript or typescript like Stephen Crane's "A Detail" or "The Octopush" and the completely rewritten revision that was subsequently published from a now lost quite different copy. Even the William James manuscripts that must yield as copy-texts to his greatly worked-over book publications are not this distant from the revised form: the choice of James's books as copy-texts is made for other reasons than those that affect Every Man in his Humour.
This is a confusing footnote at best since it seems to use Greg's discarded expansion and not McKerrow's strict bibliographical definition of 'substantive edition.' Technically, the Shakespearean bad quartos that represent two substantive editions in the bad and good texts according to McKerrow's standard would be like The Merry Wives of Windsor where the bad quarto was not utilized as copy for the good Folio text. (Q1 and Q2 Hamlet, in my opinion, would also be substantive in this sense.) Plays like Richard III offer a paradox. Technically according to the "Rationale" the bad quarto (Q1) is the substantive edition since it alone was printed without reference to any other edition, and the Folio is a derived edition since its printer's copy was contrived by annotating a copy of the Quarto (Q3) with readings from a good manuscript. But to be technical in this matter is to be ridiculous textually, and other terms must be sought. It would have been helpful if Greg had not blurred the distinction between the two kinds of bad quartos, for this distinction powerfully affects his definition of the corresponding good texts.
Among other important differences, the accidentals of the earlier edition have no authority and the bad quarto would not be an acceptable copy-text even if it were practicable to introduce into its accidental texture the good substantive readings (and their accompanying accidentals) from the Folio. Moreover, when as is inevitable the Folio of Richard III is taken as copy-text, no one but a variorum editor would dream of recording the Bad Quarto's variant accidentals as Simpson felt obliged to do for the Quarto (and later editions) when he selected the 1616 Folio as copy-text for Sejanus.
The only question (which Greg does not mention) might be the practicability of retaining the bad quarto's accidentals as more representative of an Elizabethan author's period than, say, a good text printed in the eighteenth century with the texture of that period. But the chances of such occurrences are so limited as to be ignored. The First Folio's accidentals, although 'modernized' from copy, are not so vitally different from those of earlier quartos as to be anachronistic to a serious degree.
When in the manuscript of The Marble Faun Hawthorne was endeavoring to spell in the English manner for the English publishers of the first edition, he nevertheless retained neighbor as an -or spelling. William James is an -or speller but he invariably writes colour and usually honour. Hawthorne's attempts to style his manuscript in the British manner led him to adopt some -our spellings that were not acceptable by English standards of the day. Having chosen the manuscript as copy-text in the Centenary Edition of this romance, I felt that to alter Hawthorne's uncharacteristic but self-imposed -our to the American -or he used in other works (or even to the current British standards) would have been essentially to modernize the text. This is a different matter from my editorial position that British forms in William James's articles printed in English journals should be Americanized to his standard manuscript practice, observable in those preserved manuscripts underlying the British prints. Naturally, when James revised such articles for American book collection, he or the compositors gave them the normal American spellings in the editions that have been chosen as copy-texts. Thus the two principles are quite different and do not apply to each other even when James's British periodical articles might become the copy-text in default of other more authoritative documents.
This is less open-ended than it sounds because in this period (with authors like Jonson something of an exception) one can usually never satisfy oneself that such spellings are not authorial: ordinarily Elizabethan works are edited with no authorial manuscripts preserved to indicate what the author's accidentals were like. Moreover, given the variation in spelling the same word customary in most authors (and many scribes), to decide whether or not some word is authoritatively spelled requires more confidence in normality than most Elizabethan editors would wish to assume. The practical upshot is that the more eccentric the spelling, the more likely an editor is to impute it to the author.
Always granting that in early authors they usually cannot be distinguished from compositorial or scribal spellings and that the preservation of authorial characteristics in early literature is likely to be less than in later, much depending, however, on the care with which a later writer marks copy and reads proof.
In a far cry from the attitude of the editor of modern literature toward the importance of copy-text to the presentation of an author, Greg writes (p. 384), "The thesis I am arguing is that the historical circumstances of the English language make it necessary to adopt in formal matters [i.e. accidentals] the guidance of some early text."
Although Greg would not necessarily have wished to confine this statement to substantives, the context makes it clear that he is thinking only of substantives when he writes 'original reading' and that his chief concern in this passage is to defend the rights of an editor to choose substantive variants from authoritative editions other than the copy-text. How much he would have defended the same rights as applied to accidentals is problematic. In the "Rationale," at least, he does not remark on the anomaly that Herford-Simpson in Sejanus (F being the copy-text) add certain cases of apostrophus from Q which F had dropped but do not emend, according to the same principles, when F normalizes the Q classicizing spellings with one hand while adding a few with the other. Nor does he comment on the tyranny of the copy-text that persuaded the editors to retain what appear to be F corruptions of Q's use of apostrophus, as in He's for Q He'is. Rightly or wrongly, the general impression one is likely to receive from the "Rationale" is a concentration on editorial freedom to deal with variant substantives but little recognition of the comparable opportunities that exist with accidentals.
Especially when a reprint is line for line with its copy, considerable mechanical advantage accrues in hand setting to follow the accidentals of the original in order to speed justification of the line. I am no historian of publishing practices and have no idea when the modern custom of a publisher styling the manuscript, to be followed exactly by the printer, revolutionized the older system. What I know from observation is that Hawthorne's manuscripts in the 185os were not at all marked by Tichnor and Fields but were heavily styled at the Riverside Press, several thousand accidental variants from the manuscript press copy appearing in each first edition. (There is also some question whether Tichnor and Fields ever assisted Hawthorne in reading proof for these romances.) The Marble Faun manuscript was not marked, either, by Smith, Elder, its British publisher. In the first decade of the twentieth century Longmans, Green never attempted to style William James's manuscripts or typescripts: James dealt directly with the Riverside Press for each step of the books' production. Whether it was the passage of fifty years or the difference between a fiction writer and a philosopher, in contrast to Hawthorne the Riverside Press set James's copy with considerable fidelity. A test case is his posthumous Some Problems in Philosophy in which his customary heavy proof-correction is not present to interfere with the evidence. The Riverside Press departed from the typescript copy furnished it by Horace Kallen in only about one accidental for every five or so printed pages. In PBSA, 71 (1977), 59 (see fn. 9 above), Bush detects a movement from the mid-eighteenth century toward the substitution of copyreading for the proofreader's correction of authorial accidentals; but he remarks "How widespread this reader's function was is impossible to say," and the references he cites are few and rather vague.
Here the evidence of Some Problems of Philosophy is very useful, indeed, in giving us an insight into the fidelity of setting his works customary at the Riverside Press. In James's Psychology (1890) the Holt printer's fidelity in close to verbatim setting from printed copy suggests that when more numerous variants from copy are elsewhere found the creator of these variants (both accidental and substantive) was James himself.
After the editor had made a futile attempt with Pragmatism, the initial volume, to apply the more common Greg rationale and insert substantive book revision in the journals used as copy-texts.
In "Principles and Methods in the Collected Edition of John Stuart Mill (Editing Nineteenth Century Texts [1967], p. 117) J. M. Robson remarks that "An electronic eye may find otherwise, but to the human eye there is (in general) a uniformity in accidentals between the substantively revised and unrevised portions of the later editions, a uniformity which does not exist between later revisions and earlier revised passages." The sense of this is not wholly clear to me, but if it attempts to distinguish the uniformity throughout the latest revised edition, for which Dr. Robson is arguing as copy-text, from the lack of uniformity in earlier revised texts, the evidence is not necessarily an indication of Mill's extensive accidental revision in the uniform text but could be the styling applied alike to revised and unrevised sections by the printer of the final edition. Dr. Robson may well be right in his choice of copy-texts, but his evidence is suspect.
A complicating factor can be the publisher's revisions also introduced into a revised text, as happened with F. Scott Fitzgerald: see M. J. Bruccoli, Apparatus for F. Scott Fitzgerald's The Great Gatsby (1974).
Naturally, no one would argue that in the transmission of meaning accidentals are as important as substantives, although there is always the example of the letter in Ralph Roister Doister as well as the two versions of Hamlet's what a piece of work is a man speech to show the contrary. I am talking about editorial principle, insofar as it can be implemented, and not expediency in any individual case. Since accidentals in modern times have lost the philological significance that Greg thought worth preserving, if they have not gained correspondingly in importance as a part of the total complex of authorial meaning, then all later texts might be modernized and discussions of copy-text are idle.
However, one must always consider the role of authorial oversight; even the scrupulous William James sometimes inadvertently overlooked in one passage anomalies he had set right in another.
Few writers are likely to quarrel with normal printer's (or publisher's) styling when the sense is acceptable even though it may differ from their own accidental characteristics, which alone should concern an editor and which, incidentally, are likely to remain constant in later documents. For instance, Stephen Crane seldom put commas between adjectives in series although the printer almost invariably 'corrected' him. He made no effort to remove these commas when he read proof, but neither did he accept the lesson and insert such commas in his subsequent manuscripts. William James usually (although not invariably) omitted the comma in the British fashion before the and or or of a coordinate series of three or more, which American printers usually inserted. He would sometimes in proof remove this inserted comma but perhaps more usually not, although his own practice was never affected in his manuscripts. The question that goes unanswered is whether a writer's indifference, his oversight, or an unwillingness to demand resetting of such departures constitute 'approval' of uncharacteristic accidentals in a revised edition, especially when the setting copy itself exhibited certain categories of these accidentals in the writer's characteristic form.
In the Berg Collection of the New York Public Library is preserved a memorandum written by Hitchcock addressed to but not demonstrably seen by Crane which was sent by Hitchcock to the printer in order to correct nine readings in subsequent impressions of the Red Badge. The following semi-query was addressed to Crane but no answer is preserved in case he ever saw it: "I don't understand just what local dialect you give your men, but suppose it to be western New York—I thought 'derned' and 'yeh' did not begin till farther west, but it is doubtless all right." The corrections (some of which are sophistications) are not in themselves evidence of Hitchcock's intervention in the preparation of the typescript for the printer, before or after Crane had himself finally revised it in New Orleans. In fact, it could be suggested that the readings (and the dialect) are matters he would have taken up in any careful publisher's reading of the typescript setting copy. It may be noted that the one major substantive alteration Hitchcock ordered (Moreover for Too) caught only one of the four occurrences in the first edition (and manuscript) of this typical Crane locution and is a manifest sophistication.
One need not be too casual about these losses, of course: preservation of the proof would help to identify any publisher's alterations, and preservation of the typescript would establish the extent of the publisher's editing of the printer's copy as against Crane's New Orleans revisions. However, if the manuscript is taken as copy-text, the loss is not so great as if the first edition had been selected (or the manuscript not preserved), for the authorial markings in these two documents would not have been extensive (especially in the proofs) and most substantive authorial changes may be identified by comparison of the manuscript with the first edition.
Since I have extensively analyzed problems of copy-text, with illustrations, in "Multiple Authority: New Problems and Concepts of Copy-Text," Essays, pp. 447-487, with further discussion in "Remarks on Eclectic Texts," pp. 488-528, the account here is correspondingly brief.
If Jonson did revise some of the accidentals as well as the substantives in the working papers before sending them to the printer, such accidentals would be impossible for an editor to distinguish from those originally present in the working papers but inadvertently not copied faithfully in the fair authorial transcript. Of this nature perhaps are the characteristic parentheses that Simpson notices as appearing in the Quarto Masque of Queens in some places where they are wanting in the preserved fair copy. Of course, from the point of view of a critically edited text, the origin is of no practical account so long as such variants are established (a) as authorial, and (b) as not representing a revision in writing out the fair copy although permitted to stand in the working papers and thence transferred to print.
Greg adds that the editor should also record "the alteration whenever the sense is appreciably affected." This selective permissiveness would prevent a reader from reconstructing the original from the apparatus and hence he would never know in any given case whether he was reading the original or the editor's silent substitution. Greg's remarks on the treatment of accidentals in an apparatus are not to be taken seriously as prescripts.
If the ideal state prevailed that an editor would produce the same critical text regardless of the choice of copy-text, then the whole matter would be academic and the choice would rest purely on the convenience of the apparatus. But except for multiple-authority texts like newspapers set from a common syndicate proof and preserved in sufficient numbers for statistics to operate, this state is quite theoretical, for few cases permit an editor to be certain about the exact authority of each individual variant accidental. Thus the choice of copy-text will exercise an important influence on the neutral accidentals printed in the critical edition: all those in the copy-text not subject to treatment like substantives must be taken on faith as in general representing more authority than those in the other document(s).
The textual commentary in the Centenary Edition (1974), IX, 536-547, may be consulted for the details. More evidence is present, with more variety, in Mosses from an Old Manse (1974), X, 537-556. The transmissional problems of The Snow-Image (1974), XI, 411-422, do not differ in kind except for one or two tales. In all these stories the collected book editions were taken to be of lesser accidental authority than the earlier printings and hence were rejected as copy-texts since Hawthorne had made only a minimum of alteration either in copy or in proof. In short, many more accidental variants appear in the books than those that might be imputed to Hawthorne. Under these circumstances, authority is best preserved, on the odds, by altering the originals in the few cases when the book accidentals give evidence of authorial change than by accepting the book's accidentals en bloc and restoring the forms of the originals only as corrections of the errors in the book.
For example, three accidental press-corrections were made in the punctuation of the four pages of the inner forme of sheet B in the revised 1668 edition, but in this same forme two variant spellings from the first edition, two variant word-divisions, three variant capitalizations, and three punctuational differences were not touched although presumably compositorial. If the revised 1668 edition had been chosen as copy-text over the 1667 first, ten unauthoritative accidentals would have been reprinted in order to secure three authorial (not identifiable if they had not been determined by press-correction). However, the third edition of 1670 was also revised. In the same four pages this text makes four substantive alterations. If Dryden had also touched up the accidentals at the same time, on the evidence of the 1668 copy whatever he had altered would be lost among the three changes from 1668 in word-division, twenty-three in capitalization, and eight in punctuation. The revised 1670 edition, therefore, would depart from 1667 in a total of forty-four accidentals, of which only three by a fortunate chance can be identified as authoritative. The major part of the remainder are presumably compositorial, including those categories not press-corrected by Dryden in 1668.
This would include radiating texts in which a choice of copy-text rests on the convenience of the reader according to the ease with which he can refer to the apparatus. That is, it is most convenient, usually, to select that radiating text that needs the least correction from the others and thus one that has the fewest unassigned accidentals. When enough witnesses have been preserved so that the statistical tables of the accidental variants can be utilized as the basis for choice, there should be little if any need for an editor to rely on any accidental in the copy-text simply because it occurs in the copy-text, although he may of course take that fact into account when the copy-text document seems on the whole to be relatively faithful and all other evidence is indifferent, as by only partial corroboration from other witnesses. On the other hand, if an editor chooses to adopt G. T. Tanselle's ingenious suggestions for a new kind of apparatus for radiating texts, the need for an arbitrary copy-text vanishes: see his "Editorial Apparatus for Radiating Texts," The Library, 5th ser., 29 (1974), 330-337.
The necessary changes according to the textual results of this new discovery are incorporated in the text of the 1977 paperback edition published by the Wesleyan University Press, with page-line references to enable holders of the two-volume Wesleyan-Clarendon edition (1975) to correct their copies pending a second printing of the larger form.
Some useful surveys of two of the Shakespeare Folio's compositors' errors in setting against copy may be found in Alice Walker, Textual Problems of the First Folio (1953), especially pp. 31-35, 86-93, 119-120, 130-134, 143-149. Another such list can be found in John S. O'Connor, "A Qualitative Analysis of Compositors C and D in the Shakespeare First Folio," Studies in Bibliography, 30 (1977), 69-74.
One may add, as a contributing factor, the possibility with careless authors that an indifferent variant in the revised edition may also be the publisher's editor, as happens in Crane's revised Maggie, and, as mentioned, in F. Scott Fitzgerald's plate revisions. This publisher's intervention can come in a different line of revision, as between an author's manuscript and the copy finally given to the printer, as assumed in Crane's novels published by Appleton although the extent of Ripley Hitchcock's ministrations is still far from certain as compared with those of the printer.
If, as seems likely, Fielding's lost manuscript actually contained laid hold on and anywise it might not be enough to show that these locutions could be found elsewhere in his works. Unfortunately, even their appearance at a date later than the 1743 third edition might not provide positive evidence one way or another, for it is clear that some of the readings of the 1749 first edition of Tom Jones (revised about a year later in the fourth edition) repeated the same sort of original readings that had been revised out of Joseph Andrews in 1743 and 1748. What would be needed would be evidence that Fielding used both, whether indiscriminately or chronologically. In that case the third-edition reading could be taken with more confidence as an authorial revision. If the third-edition form never appeared elsewhere in Fielding, it would suffer under the presumption of error, of course.
This is an argument to be used with discretion, of course, when the compositorial stints are unidentifiable. In Tom Jones, for instance, I am not troubled by the revised fourth edition's normalization of a country landlady's 'Here's a great young Squire, and a many other great Gentlefolks of Quality' (537.6-8) to and many other, for it is a natural sophistication or eyeskip; and when at 834.30 the same change is made in a reported speech by Squire Western an editor may merely note to himself that the earlier occurrence now seems more probable as sophistication. It seems difficult to believe that Fielding would remove such an idiom from two dialect speaking characters: a literal minded compositor, or even a pair, seems the better explanation.
On this evidence it is probable that in Joseph Andrews two examples of the reverse movement, both in the fourth edition of 1748-49, are not revisions but compositorial changes to be ignored in favor of the copy-text: broke for 1-3 broken (151.10) and tore for torn (331.40).
Estimate of error outside of positive typographical misprints or quite contrary sense, which do not count as evidence, is partly subjective, of course. Nevertheless, in Tom Jones if one sets aside the cases when the fourth edition followed a variant in its corrupt third-edition copy, there appear to be no certain compositorial substantive errors departing from copy in Book I, only one in Book II, two in Book III, six in Book IV, nine in Book V, seven in Book VI, and so on. Given the quantity of text, this represents greater faithfulness than would be expected in an Elizabethan printing-house. These presumptive errors contrast with seemingly authorial substantive revisions as follows: twenty-three in Book I, eighteen in Book II, eighteen in Book III, thirty-nine in Book IV, forty-three in Book V, and twenty in Book VI.
One must repeat that all rules are off when dealing with variant pre-publication scribal transmission of authority as with Shakespeare's two-text plays, or Fletcher's plays with preserved scribal manuscripts in a different line. Usually in such plays the chances for tracing the origin of error in any bibliographical sense are minimal, at least with results likely to win a consensus.
| | |