Editing Texts

This paragraph deals with the critical debate initiated with a single essay by W. W. Greg, ‘The Rationale of Copy-Text’, Studies in Bibliography , vol. 3 (1950-51), pp. 19-36, reprinted in his Collected Essays (1966), pp. 374-391, and in other anthologies of textual criticism; see Bibliography and Textual Criticism. English and American Literature, 1700 to the Present, edited by O.M. Brack, jr., and Warner Barnes (Chicago, University of Chicago Press, 1969), pp. 41-58; Art and Error: Modern Textual Editing, essays compiled and edited by Ronald Gottesman and Scott Bennett (London, Methuen, 1970), pp. 17-36. An Italian translation appears in Filologia dei testi a stampa , a cura di Pasquale Stoppelli (Bologna, Il Mulino, 1987), pp. 33-51. Again a word of warning is required, since for people not expert in the criticism of English Renaissance texts some of what Greg is saying can seem obscure, and the central message of the essay has to be extrapolated with care. Rivers of ink have been spent in debate relating to Greg’s rationale, many litres of which in complete misunderstanding of what it really suggests, so that anyone wishing to react against the brief summary furnished here would do well to read further before jumping to conclusions. As the title of the article inequivocably states, what Greg suggests is not a theory, nor a rule, nor a law, but a rationale which prevails only in a certain type of interaction between an author and a printing shop.

As has been remarked several times in the course of this piece, the defining characteristic of the McKerrow-Greg-Bowers school is a concern for finding ways in which bibliography betters our understanding of textual transmission and thus, as a final outcome, improves textual scholarship. It is perhaps inevitable therefore to discover that it has produced its own, perhaps idiosyncratic, but ingenious and effective theory of critical editing, containing elements of novelty for scholars expeced in other traditions, especially those raised on a diet of Lachmann and Bédier. Again the pragmatic approach exhibited initially by McKerrow, coined into a ‘rationale’ by Greg, and further elaborated by Bowers, derives from an intense familiarity with the problems of editing Elizabethan and Jacobean texts written in a very fluid state of the English language. With respect to other editing traditions, built on the acquaintance with classical and medieval texts, it needs to be pointed out that critical scholarship of Renaissance English texts habitually distinguishes between ‘old-spelling’ and ‘modern-spelling’ editions: in the first the words are printed and spelt as in the copy-text, in the second, though the copy-text is still followed, spelling, punctuation and other features are modernised. The peculiarities of late Sixteenth and early Seventeenth-Century works in English often mean that unusual and idiosyncratic spellings derive from the author in person, and therefore deserve attention from the editor. With respect to other languages, problems of form include not only spelling and punctuation, but also word-division, insertion of hyphens, upper or lower case, and the use of italic to highlight key terms in the text (for instance in the first edition of Milton’s Paradise Lost in 1667).

The copy-text – the term was first coined by McKerrow in 1904 - is the manuscript or printed version chosen by the editor to furnish the basis of the critical edition and scholars previous to Greg, such as Paul Maas, had already commented on how the “tyranny of copy-text” interfered with an freedom of choice, since any departure from the base could lead to an editor being accused of taking unjustified liberties. Building on a hint furnished by McKerrow in his Prolegomena for the Oxford Shakespeare (1939), Greg distinguishes between the “substantive readings” and what he terms the form or the ‘accidentals’ of the text. In his definition the substantives are the words themselves and their meanings: if we change horse to house , only one letter is modified but the whole sense has changed; the accidentals are the shape of the text, “such in general as spelling, punctuation, word-division, and the like” (p. 376), that influence our perception of the text but do not alter the substance of the “author’s meaning or the essence of his expression” (ibid.): if, instead of horse , we write ‘horse’ or Horse or hoss or ’orse , the surface of the text is changed but not the reference to an equine. When Greg produced this brief essay, textual editing, despite the grumbles of a minority, generally required both the accidentals and the substantives to follow the copy-text, except where manifestly incorrect. Greg states instead that this principle is valid only for the accidentals, whereas, as far as the substantives are concerned, an editor must be free to trust his own judgement.

Students familiar with other critical traditions will not find anything disturbing today in the principle of selection between different substantive readings, independent from the choice of copy-text, since the procedure reflects the practice followed in numerous editions of classical and medieval texts. But, having defined the principle of a divided authority and therefore of the editor’s right to produce an eclectic text, Greg applies the method to cases of printed texts which in successive imprints have been revised, either by the author or by reference to a manuscript. His essential point is that, unlike manuscripts in which the relationship between the witnesses is difficult to define, in printed texts the transmission is linear and the various stages are known. It is important to understand that compositors always preferred to set from a previous printed text, if available, and therefore, even when the version was improved by reference to a manuscript, it was common practice to write the corrections into a copy of an earlier edition and send it to the printing shop. In such circumstances the new edition is bettered in terms of the substantives, but worsened in the accidentals, since the compositor introduces the new substance but imposes his own preferences on the styling of the text. Greg’s solution has the logic of sweet reason. Where it can be reasonably supposed that the accidentals of the first edition are closer to the author’s original manuscript and that those of the reprint have no additional authority, a critical text can be constructed which unites the form of the earliest edition with the substantives of the later ones. The rationale has been frequently applied in editions of Shakespeare’s dramas: for instance, Othello was first published in quarto form in 1622 (Q1), while the text produced in the First Folio benefited from extensive revision taken from a manuscript, whose different and improved reading were copied into one of the quarto reprints (probably Q4). Since the form of this last is without authority, editors have generally conjoined the form of Q1 and the substance of the Folio.

Greg’s rationale was taken up, at times over-enthusiastically, by Bowers and his disciples who applied it on an extensive scale to critical editions of the British and American novelists (the latter under the auspices of the Centre for American Authors (CEAA)). The phenomenon has occasioned much debate, most of it partisan, and various attacks have been launched against these reconstructions of texts according to authorial intention. For the benefit of Continental readers, who are not necessarily familiar with the nuances of the history of English as a spoken and written language, I here wish to clarify some unspoken assumptions behind these operations. Again it is necessary to understand how writers worked in the period concerned, as well as the economic and social enterprise marked by the novel. For writers such as Scott and Dickens, a successful career as an author gave the possibility of immense gains, but also placed them under constant pressure to produce their material in a short space of time. Both were prolific in their output, so that their collected editions occupy over a metre of shelf-space, and both worked in close collaboration with their respective printing shops. As natives of other tongues have learnt to their cost and displeasure, in the English language there is little correlation between spelling and pronunciation, while matters such as capitalisation, punctuation and so on seem equally haphazard. The British have never had the equivalent of an Academie Française with the authority to decide right and wrong in matters of language, while usage has always been dictated by the streets of London, with an infinite number of variables. These same problems were felt by the writers, who duly created the text, but left the ‘styling’ (i.e. the accidentals) to the experts in the printing shop. When both Scott and Dickens, towards the end of their careers, came to revise the texts of their novels for their respective collected editions, they both made alterations to the substance of the text, but showed no interest in the form, since, to put it bluntly, they did not consider it their business.

What Bowers insists on and is often ignored in the debate around Greg’s rationale is the fact that a copy annotated in this manner represents the last intention of the author of which we have knowledge and therefore ought to be the focus of our critical attention: for instance, in his introduction to the Dramatic Works of Thomas Dekker in 1953, he states: “In all cases the first editions – the only ones set from manuscript – provide my copy-text. Later editions have no authority except for two plays... which show revisions and corrections deriving from the author. For these two plays I use the methods of recent textual theorists [i.e. Greg]. I retain the ‘accidentals’ – the general texture of spelling, punctuation, and capitalization – of the first edition, the only one which has a direct relationship to the ‘accidentals’ in the manuscript that served as printer’s copy. Into this texture I introduce those revisions (chiefly ‘substantive’) for which, in my opinion, neither the compositor nor the printing-house editor but the author was responsible. For these two plays the critical text thus comes as nearly as possible to reproducing the copy of the first edition marked by the author for the printer of the second edition” (Cambridge, Cambridge University Press, I, p. ix); cfr. Idem, ‘Current theories of copy-text, with an illustration from Dryden’, Modern philology, vol. 48 (1950), pp. 12-20, rist. in Essays cit, pp. 277-288: 280). Of course it does not always happens that an author revises a copy of the first edition in order to prepare a revised version. Sometimes a later reprint is chosen, but in such a case, having accepted that what we reconstruct is the annotated exemplar, theoretically it is acceptable to move this layer of manuscript corrections backwards in time into a virtual copy of the princeps, as Bowers explains in his introduction to Henry Fielding, A history of Tom Jones : “By this procedure an editor of Tom Jones does not substantially reproduce the exact marked copy that was given to the press for the fourth edition, for this was the third edition which is at one step removed from the authority of the first-edition accidentals. Instead, by choosing the first edition as copy-text and then substituting for its substantives those revisions in the fourth edition that are considered to be authorial, not compositorial, an editor attempts to reproduce what would have been the characteristics of the marked copy if Fielding had annotated the first instead of the the third edition. That this procedure reconstructs a purely hypothetical printer’s copy for the fourth edition is of no consequence in comparison to the virtues that are achieved in wedding the authority of the first-edition accidentals with the general authority of the revised fourth-edition substantives” (Oxford, at the Clarendon Press, 1974, I, pp. lxx-lxxi).

The outcome of these various operations a text which in ‘historical’ terms has not circulated previously, and from this particular point have derived most of the challenges to the acceptability of the rationale. Again it is possible that part of the reaction is due to the prescriptive tone in some articles by Bowers, though not a few of the arguments involve a theoretical whirlygig, vitiated by concepts such as reception theory and the historical instability of texts, which fails to address the practical problem of an editor who has to produce a text other people will read. To take the most common objection, is the editor duty-bound to respect the form of the text that originally appeared and were read at the time? A positive reply does not take account of the extent to which such texts are compromised (at times by the author), or corrupt, or sometimes just inferior. Do we really want to read the 1850 text of Wordsworth’s Prelude in preference to the 1805 version recovered from the manuscripts in 1971? Textual editors have always recuperated and made available versions of texts that do not have the support of a historical tradition, but which are better than previous versions put into circulation. Obviously a scholar interested in the influence of Wordsworth’s Prelude on writers of the late Victorian period will work with the 1850 text, but modern readers will choose the 1805 (or the two-part 1798-99) version as the truest expression of Wordsworth the poet. The construction of an eclectic text, which nevertheless remains truer to what the author actually wrote than any alternative, is therefore a perfectly logical, successive step.

Plenty of cases inevitably spring to mind, for example English or American authors such as Richardson (who was also a professional printer) or Henry James, and Italian writers such as Ariosto and Manzoni, who thoroughly revised both the substance and the form of their works and for whom Greg’s rationale is inappropriate, but the long list of exceptions serves only to define what is not even a rule but a solution in given circumstances. In the definition of Gianfranco Contini, any critical text is never more than a ‘working hypothesis’ that speaks to its own time and to no other. Future generations will produce their own critical texts according to their own criteria, resources and technologies, and this is their concern, not ours. Above all historical truth is not obtained by placing a higher value on the safety of the editors in their own critical community than on what the author actually wrote.

For further bibliography and above all for the development by Bowers of Greg’s initial proposition, see Fredson Bowers, ‘McKerrow, Greg and “Substantive Edition”’, The Library , s. 5, vol. 33 (1978), pp. 83-107, and Idem, ‘Greg’s “Rationale of Copy-Text” Revisited’, Studies in Bibliography , vol. 31 (1978), pp. 90-161. The rationale has also been the object of three articles by G. Thomas Tanselle, brought together and published in monograph form as Textual Criticism since Greg: A Chronicle, 1950-1985 (Charlottesville, University Press of Virgina, 1987). The same scholar offers his own view of what textual editing should be, along Greg-Bowers lines, in the 1987 Rosenbach lectures, issued as A Rationale of Textual Criticism (Philadelphia, University of Pennsylvania Press, 1989). The debate has remained a lively one and Tanselle offers a wide-ranging round up more recent developments in ‘Editing without a Copy-Text’, Studies in Bibliography , vol. 47 (1994), pp. 1-22; ‘Textual Instability and Editorial Idealism’, Studies in Bibliography , vol. 49 (1996), pp. 1-60; and most recently in ‘ Textual criticism at the millenium’, Studies in bibliography.