University of Virginia Library

Search this document 


  

expand section 
expand section 
collapse section 
 1. 
 2. 
 3. 
 4. 
 5. 
  
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
Document or Process as the Site of Authority: Establishing Chronology of Revision in Competing Typescripts of Lawrence's The Boy in the Bush by Paul Eggert
expand section 

expand section 

Document or Process as the Site of Authority: Establishing Chronology of Revision in Competing Typescripts of Lawrence's The Boy in the Bush
by
Paul Eggert [*]

An unusually thorny problem can arise in the preparation of a reading text of a critical edition if multiple printers' copies are extant which were originally prepared by the author for simultaneous publication (typically, in New York and London) and if the author revised the copies differently.


365

Page 365
Developing practices to deal with such a case—the typescript copies of The Boy in the Bush—has revealed a confusion which the usual formulation of the goal of scholarly editing glides over. 'Scholarly editors may disagree about many things', G. Thomas Tanselle observed in 1976, 'but they are in general agreement that their goal is to discover exactly what an author wrote and to determine what form of his work he wished his public to have'.[1] The editorial method described in the present essay is aimed at achieving that goal by satisfying Tanselle's attendant call for the critical judgment of textual cruxes on a case-by-case basis; and it accepts as a matter of course his clear implication that, often, no single document will fully represent the author's intention for the work. Consideration of The Boy in the Bush typescripts has, however, highlighted the need for a further editorial distinction beyond that of document and work: a distinction between the process of composition and revision on the one hand and the documentary sites of the inscription of the process on the other. This distinction, if accepted, can justify a provisional but potentially clarifying separation of the 'authorial function' from the 'production function' even where the evidence of the former and the requirements of the latter reside in the very same documents and where the same person—the author—discharged both.

It is first necessary to outline the particular textual problem.[2] The Boy


366

Page 366
in the Bush is a collaborative novel, but all of the extant manuscript material is in the hand of D. H. Lawrence and none in that of Mollie Skinner, an Australian bush-nurse and amateur novelist whom he had met when he visited Western Australia in 1922 and who, at his encouragement, wrote and sent him for placement a novel about colonial life in the 1880s. He believed the novel was unpublishable and so, with her permission, rewrote it in September-November 1923 in California and Mexico, sending the manuscript on in two batches for professional typing and subsequently adding a new last chapter in January 1924 after returning to London and just prior to correcting the (by now) typed copies. Though widely dispersed, the autograph manuscript and two typescript copies are extant: the latter became the setting copies (known as TSIa and TSIb, respectively[3]) used by his American publisher, Thomas Seltzer and his English publisher, Martin Secker, both of whom published the novel in 1924.

Both TSIa and TSIb contain 543 pages, and bear Lawrence's revisions and corrections in the page range 167-528. The reason that his revisions do not appear on pages 1-166 is that in London he revised a now lost, earlier typescript—one copy only—of this section (which represents the first batch of manuscript). It was then re-typed (ribbon and carbon) while Lawrence got on with revising duplicate typescript copies of the rest of the novel (corresponding to the second batch). When he received the copies of the re-typed section he simply added them in without further revising them, thus constituting TSIa and TSIb.[4] In pages 167-528 there are approximately 180 substantive revisions consisting of about 100 single word changes, about 65 revisions of short phrases, with the remainder being quite substantial changes. As he revised TSIa and TSIb Lawrence had had to bring the novel into line with the new last chapter he had just added which had given the novel a significant shift in emphasis.

In the interests of producing a reading text of The Boy in the Bush as close as possible to the one Lawrence 'would have wished to see published',[5] the autograph manuscript was chosen as base text and a procedure envisaged of emending it from one of the typescripts wherever Lawrence himself had revised them—provided of course they copied the manuscript accurately.[6] This solution, relying on textual evidence unambiguously in the hand of the author, might have swept away the textual problems in one move were it not for a complicating factor which soon became evident. Although TSIa


367

Page 367
and TSIb (destined to serve as printers' copies) had been typed professionally, no doubt originally as ribbon and carbon of one another, they had later been mixed up, becoming composites of ribbon and carbon—the form in which they are found today. Evidence gradually mounted indicating that Lawrence was responsible for the shuffling when revising the typescripts—differently.

Page 468 of both typescripts provides a striking example of the process (see illustrations; the handwriting is Lawrence's). On the sixth line of the first heavily revised paragraph in the TSIa page is the handwritten sentence starting: 'This had made her rebel . . .'. At first TSIa read 'so terribly against him' before 'him' was replaced by 'the thought of him'. Two sentences down a nearly illegible 'But' was replaced by 'Now'. At the same places in TSIb may be seen 'so dangerously against the thought of him' and 'Now': TSIb has no erasures of handwriting here. Lawrence may have revised TSIa, made the additional alterations, and then transcribed the whose passage to TSIb, changing 'terribly' to 'dangerously', and making a range of other changes further down the page as he went. Or, he could have transcribed first, developing the passage in TSIb as he transcribed, and then, realising that the typescripts now read differently, made a rough attempt to match them up. Belated or half-successful matching-up occurs elsewhere in the typescripts, suggesting that Lawrence felt the tug of responsibility but lacked the iron will of consistency necessary to achieve the result. As an author in revision Lawrence could not help but take the opportunity presented by transcription to further shape his description of Monica's awed resentment of the novel's hero; but as a scribe Lawrence's qualities were only mediocre. His mind was not on that job; and anyway, he may have reasoned, if the English and American editions were not exactly the same, what would it matter?[7]

The evidence of page 468 alone suggests a convenient hypothesis: that Lawrence always worked on TSIa first and on TSIb second. If so, TSIb would have been the transcription copy throughout and would contain the last reading in most cases. But an editor, tempted to embrace such a simple solution, is apt to be reminded of the situation of the donkey A. E. Housman describes, hesitating between two equally attractive bundles of hay;[8] would the editor have shown one typescript to be a superior source of emendation simply by walking away from the other one?

If that ignominious fate were to be avoided, closer inspection would be necessary. A textual collation showed that sometimes TSIa appeared to have the last reading and sometimes TSIb. An explanation of what must have happened readily suggested itself. In transcribing changes from TSIa to TSIb (or vice versa), Lawrence had frequently altered them. Evidently


368

Page 368
illustration

369

Page 369
illustration

370

Page 370
pleased in some cases with the new reading, he then transferred it back to the first typescript (but not always); with some longer revised passages he transferred back only some parts of the new reading. And often he would enter a revised reading in one copy and fail to transcribe it to the other (or, possibly, decide not to). The inevitable result was that his publishers would not receive identical setting copy and that TSIa and TSIb would contain thirty-seven instances where emendations appear on only one typescript, and twenty-nine instances where both typescripts are revised, but revised differently.[9] The retrieval of a reading text from this unexpected textual tangle had begun to look decidedly problematic: would it be possible in relation to the last two-thirds of the typescripts where the variant revisions occur to discover the last-entered reading in each case?

To attempt to answer that question it would be necessary to reconstruct how Lawrence had physically treated the typescript copies as he revised. The typists had not paginated the last two-thirds of the typescript[10] and Lawrence's hand-written pagination of this section involved several attempts to get right. Had these paginations been carried out as Lawrence revised or after he had finished revising? Towards the end of the novel he discarded some pages when replacing them with some pages he typed (amateurishly) himself. Because the replacements contain exactly the same pattern of successive cancelled paginations as previous and succeeding pages the paginations must have succeeded the revision.[11] The alternation of ribbon and carbon copy in the two typescripts precisely at chapter-ends[12] further suggests that Lawrence had been trying to keep the typed sheets together in chapters as he revised, and in fact one of the cancelled paginations resulted from his having left chapter xi temporarily out of the count. These considerations made it possible to conclude that Lawrence was probably presented with the typescript copies either in one bundle (ribbon copy of a chapter followed


371

Page 371
by its carbon copy, then the ribbon copy of the next chapter followed by its carbon, and so on) or in complete, separate bundles.

However, this conclusion did not reveal the order in which Lawrence revised, for however he pulled the chapters off the pile or piles he went on to re-collate them pretty much at random. (And he must have known he had done this when he subsequently paginated; evidently it did not worry him.) He was not devoting much effort into getting the collation nice and tidy.[13] It would have served little purpose for an editor to have notionally re-sorted TSIa and TSIb into ribbon and carbon copy because textual evidence showed that while Lawrence frequently revised ribbon copy first and transcribed his revisions and corrections to the carbon copy, he did not always do so. (Even if he had meant to do so he could accidentally have pulled pages off the wrong pile—if there was more than one pile—or he could have shuffled the two copies of the particular chapter in his hands as he laid them down.)

Although it is likely that Lawrence revised a chapter at a time, he appears to have stopped to transcribe from one typescript to the other after each page. The evidence lies in his correction of a great number of misstruck and faint characters—up to ten per page. Where Lawrence corrects such mistyping on any page the proportion he transcribes to the other copy is high. If he had waited till he got to the end of a chapter, it is unlikely that he would have picked up so many by leafing back though. Moreover, within a chapter there is evidence Lawrence would occasionally swap from revising say, ribbon copy (and transcribing to carbon) to revising carbon copy (and transcribing to ribbon); he would even, on rare occasions, move from one copy to the other while working on a particular page. (After making an alteration he could have immediately transcribed, then continued to read down the page on which he had transcribed the alteration.)[14]

An editor confronting this criss-crossing trail of textual revision might try to simplify matters: he might decide to take a modified 'best text' approach and incorporate into MS those revisions appearing in only one of the typescripts, consigning those in the other to the textual apparatus. However there is no evidence with The Boy in the Bush that Lawrence preferred one typescript over the other, so that the editor's decision, like that of Housman's donkey, would be arbitrary. Another simplifying technique would be to try to establish a rough chronology of revision in the competing typescripts by counting all cases where Lawrence revised a reading in one typescript,


372

Page 372
changed it immediately to another reading, and only entered that reading into the other typescript. So, for instance, on page 404 of TSIa, Len Ellis's age is typed 'not seventeen', which Lawrence changed first to 'not eighteen' and then to 'only seventeen'. TSIb has the second revised reading alone. At such points TSIa is clearly the correction copy, and TSIb is the transcription copy. Provided such examples outnumbered those where the contrary was true, the editor might be tempted to prefer TSIb as the source of emendation throughout.[15] But such evidence would not demonstrate the chronology of the documents as a whole, only the chronology of the revisions at those points. Is there an alternative to this sort of rough and ready calculation, or is it the best that can be done?

The answer must depend on the circumstances of the case, but with The Boy in the Bush it has, in fact, proved to be possible to establish the chronology of revision, though to do so is to thread back and forth between the typescript documents. The practice developed for the purpose made it possible to treat the revision stage as a single version which is witnessed in its entirety by neither typescript alone but by both. Taken together, the typescripts witness and can be made to yield a continuous stage of revision. To have chosen one of the typescripts as the source of emendation of the base text manuscript would have been to downgrade the importance of Lawrence's active engagement as an author in the revisional process. Arbitrarily to favour one document over a process that was actually witnessed by both documents would have been tantamount to ascribing textual authority to Lawrence's fallibilities as scribe and collator—functions which were clearly not uppermost in his mind.

To have chosen both typescripts and to have prepared a parallel text edition would not have produced an 'ideal text' either, for such an edition would also have been based on the extant documentary forms of the text rather than on the revisional process. This last observation points to the special significance of the textual situation of The Boy in the Bush: that it has been possible to distinguish the function of author-in-revision from his function as a participant in the production process, working to provide his publishers with copy which would allow the bookmaking to continue. Of course it was Lawrence's need to provide corrected copy which gave him the opportunity to engage in a process of revision, so to that extent the distinction does not hold; but, once he was involved in it, his dual functions—author-in-revision as against participant in the production process—may be usefully distinguished.

Two observations arise from the distinction. First, it may be found applicable in similar editing situations involving typescript or proof copies (i.e. where the author is sent two proof sets with one intended as copy for another publisher's separate type-setting).[16] Second, the distinction, if it is found to


373

Page 373
have a wider applicability, may help clarify those familiar disagreements about choice of manuscript or first edition as base text where the holograph manuscript is extant but the author is known to have read proof and approved them. Both sides in the debate believe the practice they advocate is aimed at establishing the authorial text; but my distinction suggests they may be invoking, unawares, different criteria under the same name. The distinction being proposed might help clarify matters by underlining the importance of going beyond the question of whether the author was involved at proof stage to conceptualise the kind and level of engagement he had. The editor would be asking himself whether it were possible in the particular case to distinguish between the author's compositional-revisional processes on the one hand and his participation in the commercial production processes on the other. This is not to claim that the two processes are separate and discrete for clearly they are not, only that a clarifying distinction may be able to be made. Not to ask the question may be to put oneself at risk of shifting inconsistently from one criterion to the other whenever a hard decision is faced.

A summary of the various sorts of evidence encountered and made use of in dealing with the typescripts of The Boy in the Bush may be of some interest, for in fact it proved possible to track Lawrence's zig-zag path of revision with sufficient confidence to allow the preference of a reading in one or other typescript in all but one case of variant revision. Taken singly, each sort of evidence is indicative rather than conclusive, and its relative importance will change depending on the condition of the documents an editor encounters. When, for comparison, I checked the typescript copies of Lawrence's travel book Etruscan Places, some chapters of which exist in three copies all revised by him, the kinds of evidence differed somewhat.[17] Because the preparation of three copies meant two lots of transcription from the


374

Page 374
correction copy, the likelihood of revisional development in the process of transcription increased dramatically. This process left the typescript copies with more than The Boy in the Bush yields of what is the best kind of evidence: an autograph revision in one typescript which was then cancelled and replaced by a second revision before the second revision only was transferred to the next copy. Such evidence on any one page of The Boy in the Bush suggests Lawrence was reading and revising the first copy and transcribing to the second, so that if a variant revision appears elsewhere on that page in the two typescripts there is prima facie (although not conclusive) evidence as to which one was chronologically later.

The next best evidence proved to be the tiny revisions—misstruck characters and marks of punctuation—which Lawrence was especially liable to overlook, particularly if he had a number of other and more obvious revisions and corrections to transcribe on the same page. If an editor were dealing with an author who transcribed chapter by chapter rather than page by page, then the job would be much easier, for many more transcriptional failures would be generated as some or many of the tiny corrections were overlooked. The present method would become relevant in such a case if, following revision, pages of the typescripts had been combined at random, thus destroying the integrity of each chapter's status as a first or second revised copy.

The state of the carbon copy also provided clues (although rarely in the case of Etruscan Places where the carbon was fairly evenly inked): poorly inked carbon copy will frequently fail to reproduce lightly struck characters, and, if the carbon paper itself was misaligned, ends of lines may have been cut off or the last line not transferred at all. Where Lawrence did not correct words rendered illegible in these ways on a carbon-copy page but did enter other revisions on it, it is likely that he was reading the legible ribboncopy page and transcribing to carbon—and of course vice versa when he corrected them all, particularly if conveniently combined with, say, a spelling error which he corrected only on the carbon copy.

The size of the handwriting can also indicate the direction of revision: two sizes on the one page of one copy may indicate two-stage revision. Lawrence could have entered his revisions in that copy and put the page to one side; bringing the corresponding page of the other copy in front of him, he would then have transcribed the revisions. But if, in transcribing, he immediately saw the need for another alteration on the same page, this would mean returning to the first copy to enter the alteration there, his physical movement of hand and arm being liable to cause a different size (or slope) of handwriting from the correction he had originally entered. The ease with which lengthy interlinear revision is squeezed into the available space may also suggest on which copy the revised reading was entered first. Only when transcribing would Lawrence be aware of how much space the reading required to be easily legible and so be able to adjust the size of his writing and spacing to suit.

Once the correction copy has been provisionally distinguished from the


375

Page 375
transcription copy for a given page the evidence being relied upon must be tallied with evidence from surrounding pages (or part-pages), particularly in the same chapter (if the alternation of ribbon and carbon is by chapter). One assumes that Lawrence would tend to move from copy to copy in a regular rhythm of revision and transcription, but one must remain alert to any and every piece of physical evidence which suggests that, for whatever reason and at whatever point, he changed the direction. For pages which bear no variant revisions no decision need be made of course, although such pages may well have indicative evidence which may help with decisions having to be made about the chronology of revision on other pages nearby.[18] With The Boy in the Bush the assumption that Lawrence would naturally revise ribbon copy and transcribe to carbon proved to be unfounded, but that direction did seem more usual with the Etruscan Places typescripts.

Further specification of the method is probably unnecessary, especially given that the kinds of evidence and their relative weights will probably differ from author to author and case to case. The crux of the procedure may be summarised however as the need to pay attention not just to the readings normally thought textually significant but also to the apparently insignificant, to the accidents of typing and transcription, so as to recover the historical process of revision. It took the present writer a long while to see these kinds of evidence, even though he had been poring over the typescripts, pondering the patterns of revision he could see so much more obviously inscribed. The literary critic in him had to learn to take its cues from the textual detective. Certainty, he found, was rarely obtainable; but strong probability was, allowing a more finely honed way than alternative methods afford of approximating the text of final revision.[19]

While readers continue to require, and publishers demand, the finalising of such texts, the critically established reading text based upon a certain standard, typically that of the postulated recovery of authorial intention, remains a worthwhile goal. The goal is an idea, the perfect achievement of which would produce an 'idea-l' text. However the logic of the present argument,


376

Page 376
if extended, is that the 'ideal' text will be more closely approximated when the processes of composition and revision are, wherever possible, distinguished from and allowed to take precedence over the extant documentary forms in which the processes are inscribed. To define the goal of critical editing as establishing the text which fulfills (typically) the author's intentions is to elide two kinds of authority: authority deriving from the author-in-composition and -revision with authority deriving from the documents he was prepared to countenance or 'authorise' as sufficient for the production processes to continue. Although evidence for the first kind of authority can only be derived from the documents, the situation with the Lawrence texts described here demonstrates that a distinction between the two kinds of authority at least sometimes can—and, where possible, suggests that it should —be made.

Critical editing involves a determined attempt at an act of historical recovery of the diachronic processes of authorial inscription. I am simply urging that, where feasible, the editor should refine on the diachronics of revision and accept the consequences of the distinction I have proposed even if this means abandoning the integrity of two or more documents which witness a continuous stage of revision. This is especially the case when, as with The Boy in the Bush, the so-called integrity of the documents depends on little more than where the pages and chapters happened to fall as their author finished with them: whether on the table, on the floor, or wherever he dropped them.

Notes

 
[*]

This essay is a revised form of a paper given to an editing session at the 1989 MLA Conference in Washington, D. C. Thanks for helpful commentary are due to Fredson Bowers, Hans Walter Gabler, Peter Shillingsburg and David Vander Meulen. The essay has also benefited from the 'Afterword' to Gabler's Garland edition of Joyce's Ulysses (3 vols., New York, 1984), and his 'The Text as Process and the Problem of Intentionality', Text, 3 (1987), 107-116.

[1]

'The Editorial Problem of Final Authorial Intention', Studies in Bibliography, 29 (1976), 167; cf. also his 'Recent Editorial Discussion and the Central Questions of Editing', Studies in Bibliography, 34 (1981), 23-65. As a result of the impact of post-structuralist ideas and Continental styles of textual criticism upon Anglo-American editing, this agreement is less general than it formerly was. For a survey, see D. C. Greetham, 'Textual and Literary Theory: Redrawing the Matrix', Studies in Bibliography, 42 (1989), 1-24. For an account of editorial orientations which marginalise authorial intention, see Peter Shillingsburg: Scholarly Editing in the Computer Age, English Department Occasional Paper 3, Faculty of Military Studies, Duntroon, A.C.T., 1984 (2nd ed. Athens, Ga.: U. of Georgia P., 1986); and 'An Inquiry into the Social Status of Texts and Modes of Textual Criticism', Studies in Bibliography, 42 (1989), 55-78. However, if Chris Tiffin's description of the ideal of 'recuperating final authorial intention' as a 'most venerable and tenacious' concept means that critical editions will continue to be prepared according to this goal for some time yet, then the present illustration of the dynamics of authorial intention in revision may be of some evolutionary value: Tiffin, 'Final Intention, Revision and the Genetic Text: Editing Rosa Praed's My Australian Girlhood' in Editing in Australia, ed. Paul Eggert, English Department Occasional Paper 17 (University College ADFA, Canberra, A.C.T. / New South Wales U. P., Kensington, N.S.W., 1990), p. 132. My own feeling is that the critically established reading text will continue to have a place but that its pre-eminence can no longer be assumed: see my 'Textual Product or Textual Process: Procedures and Assumptions of Critical Editing' in Editing in Australia, pp. 19-40.

[2]

Some complexities of the textual situation not relevant to the present discussion are ignored here; for these and for otherwise uncited factual observations made in this essay see the Cambridge U. P. edition of the novel (1990).

[3]

Items E55f and E55e in Warren Roberts, A Bibliography of D. H. Lawrence, 2nd ed. (1982). The manuscript is made up of items E55a and E55b; Roberts gives locations.

[4]

Pp. 528-543 are the typing of the new last chapter of manuscript and received only spelling corrections.

[5]

General Editors' Preface, The Cambridge Edition of the Works of D. H. Lawrence (1980-).

[6]

No proofs survive, but in view of Lawrence's known correcting of the novel's proofs for Martin Secker it was also decided to take some readings from the first English edition; and because Lawrence corrected a duplicate set of Secker proofs for Seltzer (so that the latter could check his own independently set proofs against them), a few readings have also been taken from the first American edition.

[7]

Whether Lawrence consciously reasoned thus cannot be known, but he was aware that he was preparing copy for two separate typesettings and at this stage he did not know that Seltzer's proofs would be checked in-house against a duplicate, authorially corrected set of Secker's (see previous note).

[8]

'The Editing of Manilius', in Housman's Collected Poems and Prose, ed. Christopher Ricks (1988), p. 377.

[9]

This exposes a fundamental flaw in the Romantic idea which Aldous Huxley circulated about Lawrence as an author who never revised but instead rewrote his works entirely —as of course he famously did with Lady Chatterley's Lover. (See Huxley's Introduction to his edition of The Letters of D. H. Lawrence [1932], p. xvii).

[10]

I.e. pp. 167-528: see note 4.

[11]

The cancelled paginations were occasioned by Lawrence's errors in paginating. The evidence at this point, together with the regularity of the pattern of cancelled paginations in TSIa, indicate that TSIa and TSIb had already been assembled into discrete bundles (see next paragraph) before being paginated and that therefore pagination-stage succeeded revision-stage. (The paginations are, moreover, in pencil whereas the revisions are in ink.) Lawrence paginated TSIa first: it has the bulk of the cancelled page-numberings. By the time he got to p. 468 of TSIb he had got the numbering right, and so it has only the one page number (see illustration). Thus the cancelled paginations do not assist in determining the sequence of revisions.

[12]

With the exception of 3 pages (pp. 315, 425, 524) plus 3 others, the result of carbon reversal. The typescripts of Women in Love (TSIa and TSIb: items E441d and E441e in Roberts, A Bibliography) follow a similar pattern (information from David Farmer). So does the surviving partial typescript, most of whose twin is lost, of Aaron's Rod: see L. D. Clark's Cambridge edition (1987), pp. xxxiv-xxxv.

[13]

The psychology of a modern-day equivalent readily explains this inattentiveness: a teacher collating photocopies of a multipage handout he has prepared for his students is apt to feel that because he has already put the 'real' effort into writing the document, and because the collation of discrete copies is in comparison a merely mechanical matter, he can afford to let his attention wander. The results are usually unfortunate.

[14]

P. 2 of TSIa and TSIb of Women in Love furnish an earlier example of this: see photo-reproduction in Charles L. Ross, The Composition of 'The Rainbow' and 'Women in Love': A History (1979), pp. 157-158.

[15]

Lindeth Vasey uses this method, but with additional evidence, in her Cambridge edition of Lawrence's Mr Noon (1984).

[16]

The problem may also have parallels with textual situations dating from before the invention of the typewriter. Washington Irving's Conquest of Granada (1829) is one (ed. Miriam Shillingsburg, 1988). Irving wrote the autograph manuscript in Spain and gave it to Spanish scribes to copy. Because their uncertainties with the foreign language varied from slight to entire, the copying tended to be slavish rather than regularising in habit. Where the scribe could not recognise a word, a blank was left for Irving to fill in later: with the exception of one chapter his markings appear on almost every page. However, rather than just fill in the blanks by reference to the autograph, he sometimes thought of wordings he preferred and proceeded to alter the autograph to match. Sometimes he thought of a further improvement in this transcription and so had to return to the scribal copy to enter it. Occasionally the revisional track did not run to completion, leaving variant revised readings. Irving sent the autograph to his American publisher via his nephew, Pierre Irving who, believing the manuscript was inadequately pointed, took upon himself the job of re-punctuating it in such a way that the original punctuation was frequently obliterated or at least made dubious so often as to prevent its use as base text. Thus the editor adopted the scribal copy as base text on the grounds that it more reliably transmitted the authorial punctuation than did the autograph manuscript. She emends the base text from the autograph wherever it has a later substantive reading than the scribal copy.

[17]

Roberts, A Bibliography, items E117b, E117c and E117d. The investigation of these typescripts was too brief to allow a conclusion to be reached as to the applicability of the present method to the editing of Etruscan Places.

[18]

A revision entered on only one copy presents other difficulties: the author may have forgotten to transcribe, decided not to, or only thought of the new wording while transcribing other revisions and then neglected to enter it back into the correction copy. These considerations—which necessarily lack textual evidence—are imponderable. But what such revisions certainly show is that the author did intend to, and did, make the change at some point in the revisional process. The fact that the autograph change appears in one document and not in the other does not give the document in which it appears any superior authority except at that point—where it witnesses a step in the revisional process absent in the other copy.

[19]

From typescript revision stage(s) onwards textual transmission characteristically takes divergent paths towards separate publication (although crossover at proof stage can occur: see note 6). These divergent routes might suggest that the techniques of editing medieval or ancient literature would be applicable. But the problem under discussion is not the winnowing out of scribal corruption, for the author himself is responsible for most of the variance. When transcribing revisions he is of course acting in a scribal mode, but being the author he is liable, in any act of transcription, to change suddenly from scribal to revisional mode.