University of Virginia Library

II.

New data about the textual transmission of A Portrait of the Artist as a Young Man have thus emerged from the discussion of three documents from its publishing history. Their influence on editorial decision and procedure has been incidentally considered. It now remains to outline a comprehensive editorial hypothesis on the basis of which a critical edition could be envisaged.

The Text from Manuscript to Print

In its authoritative textual witnesses, A Portrait of the Artist as a Young Man presents an almost classic case of linear and uncontaminated textual transmission. The faircopy holograph manuscript (D) is the only primary authoritative text of the novel. From it, five texts of secondary authority descend in linear succession: the typescript (T), the first printed version in the Egoist serialisation (E), and the first (H), the second (B) and the third (J) book editions. None of these secondary stages of transmission of the text relates back to any earlier stage than the one immediately preceding it, nor is the text of D ever conflated or 'contamined' with any of the secondary stages of authoritative transmission.[42] In establishing a critical text it should therefore be possible in principle to apply W. W. Greg's editorial rule which postulates that a critical text reproduce the earliest accessible authoritative text in spellings, punctuation and all other accidentals as well as in the body of its substantive readings, and that variants from the texts of secondary authority be admitted only when they are

the result of correction and revision by the author and thus positively supersede the authority of the original readings.[43] The basic text of A Portrait is the author's manuscript. Authorial correction and revision intervened at each stage of transmission between D and J, thus conferring secondary authority on each of the textual witnesses T, H, B and J. It is the extent to which their variants are authoritative which must in each case be determined. For the text in H and B, the documents which contain the intervening authorial corrections and revisions survive. These are the errata lists to Chapters III and IV in EC-W, and EC-A, the printer's copy with Joyce's corrections to Chapters I, II and V, for H; and the 'nearly 400' authorial corrections (Y), plus the printer's copy, HB, for B. Thus the authority, or lack of authority, of the variants in the first and second book editions is demonstrable. The proofsheets of the two, or probably three rounds of correction which Joyce read for J have however not been preserved.[44] The authority of variant readings in J, therefore, can upon close and discriminating analysis of the total B-J variance be established by inference only. Lastly, and most seriously, the typescript made from D and used as printer's copy for E, that is to say one of the authoritative textual witnesses themselves, is almost entirely lost. There is consequently next to no documentary evidence available of possible authorial alterations before the text was typed, nor of typists' omissions or commissions, nor of authorial correction and revision of the typescript; nor can, other than by inference, printinghouse changes in E be separated from the total body of D-E variance. Here lies the rub; for in view of the large and weighty discrepancy in the text between D and E, it is only by successful differentiation of all these separate stages of authoritative and non-authoritative interference which, hypothetically, the text passed through from D to E that a true critical text can emerge.

The external facts with which to fill this hiatus in the textual transmission are these.[45] The faircopy manuscript—bearing the date 'M.S. 1913' on its holograph title-page—was (it is assumed) written out by Joyce between December, 1913, and late October/early November, 1914.[46] Chapters I-III were merely copied over from papers (now

lost) which had contained them in a virtually final textual stage for several years. But Chapters IV and V were in their final form only conceived during these months and written (though doubtless preliminary material existed for them, too) before they were copied to complete the faircopy manuscript. The typescript followed the manuscript in close pursuit, chapter by chapter. As from February 2nd, 1914, onwards the Egoist serialisation, too, was progressing in fortnightly installments, the inference is that each chapter of the typescript was prepared with considerable haste and received only superficial authorial attention before being dispatched to London. No proof of the Egoist text was read by Joyce.

The internal evidence of the D-E variants should confirm or modify the assumed external facts. In the transmission of the text from D to E, the issues most critically at stake are the nature of the typescript, the evidence (if any) of authorial correction and revision before the typescript left Trieste, and the nature and degree of printinghouse interference with the text as it appears in print. Collation reveals most immediately the variation in accidentals. Close to 600 commas have been added in E and superimposed upon a system of commas, colons, semicolons and periods (with only the occasional exclamation or question mark) which, except in its commas, has been left largely intact. As the workmen of three printinghouses in succession set the text for A Portrait, it can be asserted that, on the whole, the additional commas were put in by them. The three printers did very nearly equal thirds of the novel: of the total of 123.5 printed Egoist columns, Johnson & Co. set 41.5, Partridge & Cooper 41.5, and Ballantyne, Hanson & Co. 40.5 columns (approximately). But the distribution of added commas is such that Johnson & Co. in ten installments added no more than 60, or about three commas in each two columns of print, while Partridge & Cooper in 8 installments added 277 (7 per column), and Ballantyne & Hanson in 7 installments 229 commas (less than 6 per column).[47] There is, moreover, a considerable fluctuation in numbers from one installment to the next—Partridge & Cooper added 66 commas on July 1, 1914, their first installment, and only 8 a fortnight later—and even from page to page and column to column. This quite clearly reflects the punctuation habits of different workmen. Moreover, the scarcity of added commas in the Johnson & Co. section of the text—itself undoubtedly the work of more than one compositor—reinforces the conclusion that the later inundation of the Egoist text with commas

was a printinghouse restyling of the text. It strongly suggests that the typescript did not essentially alter the manuscript punctuation.

The Egoist departure from the manuscript in other accidentals, such as capitalization, and hyphenation or two-word division of compounds, is far more restrained.[48] There is throughout the sections of the three printers a fairly even sprinkling of added hyphenations or compound divisions, and of added capitals. A distinction of typescript and printinghouse characteristics does not clearly manifest itself. On the contrary, it seems likely that a good number of compounds were hyphenated in E because they happened to be divided from one line to the next in T, as a good number of others are evidently hyphenated in print because they were demonstrably so divided in D and thus, by inference, entered the text of T with hyphens. Other hyphenations, such as 'good-bye', or the inevitable printed forms 'to-day', 'to-morrow', etc., were undoubtedly made according to stylesheet by the printinghouse compositors, and sometimes possibly by a typist before them. Typist's and compositors' habits likewise would seem to be the cause of added capitalization, such as the almost invariable spellings Protestant, Jesuit, Jews, Church, Mass, etc. for Joyce's protestant, jesuit, jews, church, mass. But it is very important to note that the added hyphenations and capitalizations, while of course unauthoritative in the Egoist text, are yet not inconsistent with the over-all manuscript styling. A large majority of the hyphenated and capitalized nouns and adjectives which occur in the Egoist text preserve faithfully the manuscript readings. Hyphenations and two-word divisions of compounds as well as capitalizations were largely eliminated by Joyce himself when he corrected the text for H and B. But his new directions then amount to no less than a systematic restyling of the text in print with respect to these accidentals.

If the sometimes excessively liberal addition of commas in the Egoist text is regarded as a special case—and good reason for doing so lies in the fact that Joyce's original punctuation is both unorthodox and extremely light—the general treatment of accidentals in the printed text suggests, even more so than before,[49] that the workmen engaged on E were careful and competent. This creates a certain "climate of opinion" for the consideration of the substantive variants. There is, for example, an astonishing number of omissions of single

words, phrases and even whole sentences from the text in E.[50] Anderson infers, and I believe quite rightly, that in the majority of cases these are typist's errors. In particular, he persuasively demonstrates (pp. 171 ff.) how the style of Joyce's prose by its repetitive rhetoric lends itself to the omission of phrases and sentences. His explanation of such errors by means of literary analysis can often be strengthened by taking note of the bibliographical evidence: where words and phrases are repeated in the text, their inscription on the manuscript page is frequently such that a typist's eyeskip in copying appears as the most likely mechanical reason for the omission of phrases and sentences. By contrast, the omission of single words which occurs with fair regularity throughout the text is not strictly the same phenomenon, and not as clearly explicable by literary or bibliographical criteria. It should, however, by way of hypothesis, and as a calculated methodical expedient, be acceptable to group all omissions together and provisionally to designate all omission in the extant text of E as an area of typescript error. If an omission is thus taken to be an error by principle of method, the question becomes negligible whether in actual fact it was a typist's or a compositor's blunder. The important consequence within the editorial hypothesis is that all variants in question are regarded as not authorized and that the original manuscript readings would demand to be restored in their place. If, on the other hand, the general rule in individual instances appears inapplicable, very good reasons must be found for a textual omission in E to be accepted as an authorial cut and thus to be editorially respected.

Implied by such reasoning is the truth of the assumption that the typescript was only superficially read by the author before being dispatched to London. To infer thereupon from the variants themselves, i.e., from the accumulation of omissions in the extant text of E, that Joyce indeed missed a hundred or more such errors in the typescript would be an argument self-defeating in its circularity unless support for it be found outside the circle. This problem in its turn is secondary to the basic question—which yet remains to be tested—as to whether the author gave any attention at all to the text of the novel after completing the faircopy manuscript (and before correcting E for H). The editorial difficulties presented by the missing typescript would of course be considerably diminished if it could be positively demonstrated that he did not. Answers to the open questions must be sought by scrutinizing those groups of D-E variance which have not previously been analysed, and by relating the omissions to them.

The substantive variants in E—317 in all by my count—are omissions, additions, and substituted readings. The additions are invariably confined to single words. They are few in number and make up a large part of what must be considered corrections of the manuscript text, of which there are 29 in all throughout the novel. These corrections, even if they involve an additional word, are mostly obvious enough, as when "shuffling along . . . in old pair of blue canvas shoes" becomes ". . . in an old pair . . ." (61.19),[51] and they can often easily be accounted for as the unaided work of the typist. That the typist was held to correct without specific directions by the author—or that a compositor far from Trieste did so by force of circumstance, should an incomplete or erroneous reading, real or fancied, have survived into his copy—is rendered likely when a miscorrection occurs, or a pedantic observance of grammatical congruence in tense or number sounds conspicuous. Except when miscorrection or style-sheet rectification of grammar are obvious, an edited text will of course accept the complete rather than the incomplete readings, regardless of whether or not the authority of each single addition can be ascertained. In Chapter V at least, if not before, such editorial policy can be justified by observing three individual one-word additions, two of them corrections of incomplete manuscript readings and one a genuine textual revision, which cannot reasonably be explained as anything but authorial in origin. No typist or compositor would have known how to complete the sentences: "What was their languid but the softness of chambering?" (:languid grace; 233.9), or ". . . a stasis called forth, prolonged and at last by what I call the rhythm of beauty." (:dissolved by; 206.23. 'ended' would perhaps have been an unguided guess), nor can anyone but the author be thought to have changed Cranly's toothpick at 229.33 into a 'rude toothpick', thus weaving once more into the fabric of the text the main characterizing adjective for Cranly. On the strength of these variants alone, authorial attention to the text between D and E must be admitted and taken into account as a real possibility. Automatically, it becomes a major concern of the editorial hypothesis to define its nature and extent.

Thus, the readings substituted in E for good manuscript readings become the focus of attention: they become suspect of being authorized changes. The total number of altered readings is large, but many of them are immediately recognizable as errors (as for example the numerous

substitutions of singular forms for the plural, and vice versa), or at the very least as being 'indifferent' and thus not even cumulatively strong enough to prove their origin as authorial: definite articles alternate with indefinite articles, or articles with possessive pronouns; 'those' stands for 'these', alternative prepositions are introduced, or the relative pronouns 'which' and 'who' replace relative 'that' as it is frequently used by Joyce. Yet other variants, again clearly errors, are accountable for as simple misreadings of Joyce's handwriting: 'jacket' becomes 'pocket', 'cracked' becomes 'crooked', 'harsh' becomes 'hoarse', 'like' becomes 'little', 'Kenny' becomes 'Kenory', 'slap' becomes 'step', 'burned' becomes 'turned', 'diseased' becomes 'disclosed', 'hear' becomes 'bear', 'head' becomes 'lead', 'true' becomes 'fine', the nonce-word 'nicens' becomes the none-word 'niceus', and the sentence "A rim of the young moon cleft the pale waste of sky like the rim of a silver hoop embedded in grey sand" is made to read, "A rim of the young moon cleft the pale waste of sky line, the rim of a silver hoop embedded in grey sand." Once all such variants are discounted, only very few readings remain which deserve closer attention. There should be no doubt, for example, that in the list of Stephen's classmates in Chapter II: the names Greets and Mangan were altered to Kickham and Moonan by the author before they thus entered the printed text in E. There is precedence for the authorial change Mangan>Moonan in the manuscript (Anderson, p. 170). This renders authorial attention here all the more probable, though it just could mean that an observant typist on his own accord had altered the present reading to bring it into conformity with the others. That this was not the case is suggested by the number of instances in which the name Mangan still stands unaltered in E. Moreover, it would not do to seek different explanations for two changes at the same point in the text: and the alteration of Greets to Kickham can be authorial only. Similarly, the retitling of 'Father' Barrett as 'Mr' Barrett at 30.1 in the Christmas dinner scene in Chapter I would seem to be a Joycean correction. Authorial correction and revision further manifests itself where identical changes are spaced out in the text. For example, the 'avenue of limes' which leads up to Clongowes—and it was apparently an avenue of limes—is an 'avenue of chestnuts' throughout the manuscript. In print, the change has been consistently made once each in Chapters I, II and III.

These few variants taken together confirm that Chapters I to III received authorial attention at the typescript stage of transmission. Apart from 'chestnuts'>'limes' (24.10) and 'Father' > 'Mr' (30.1), however, there is in Chapter I only one more variant—'in the square' > 'there' (43.24)—for which under the guidelines of the hypothesis here developed authority can be claimed with some confidence. The remainder of the substituted readings in this chapter are either obviously erroneous, or misreadings of Joyce's handwriting, or else too indifferent in character to be made out as authorial in origin. The situation in Chapter II is similar. To Greets > Kickham, Mangan > Moonan, and 'chestnuts' > 'limes' (93.11), it would again seem safe to add only one or perhaps two more variants: 'turning back in irresoluteness' > 'turning in irresolution' (83.31); and 'watching her as he undid her gown' > 'watching her as she undid her gown' (100.35).[52] There is admittedly a group of three further variants which, occurring within a few pages of each other, might suggest an intermittently closer authorial attention to the typescript: 'arching their arms above their heads' > 'circling their arms above their heads' (74.6), 'the old restless moodiness had again filled his heart' > '. . . had again filled his breast' (77.20), and 'the patchwork of the footpath' > 'the patchwork of the pathway' (79.1). However, careful scrutiny of the original readings as they look in Joyce's handwriting makes it virtually certain that "circling' and 'breast' are really misreadings of 'arching' and 'heart'. The apparent cluster is thus reduced to a single variant. By noting further that 'footpath' in several other instances throughout the novel is Joyce's unvaried term for "pavement", one is led to reject 'pathway' as a typist's or compositor's unauthorized substitution.

Thus, where variants in E are substituted for good manuscript readings, Chapter I appears to contain but three, and Chapter II a maximum of five authorial corrections. Of this total of eight, six (or five) show concern with factual accuracy ('Mr' Barrett, and 'limes' [twice]) or internal consistency of the text (Kickham, Moonan, 'limes' again, and 'he' > 'she', if this was an authorial correction). The two others seem concerned with a greater appropriateness ('there' as substituted for 'in the square') or fluency ('turning in irresolution') of

expression. Under the criteria by which these eight variants were separated from a host of erroneous or indifferent readings, no variants at all—except the third instance of 'chestnuts' > 'limes' (108.34) early in Chapter III—can be made out with assurance in Chapters III and IV. It is only in Chapter V that the correcting and revising hand of the author is again unmistakably present. The corrections here appear to have been made as reticently, or superficially, as in the first two chapters, but, where they occur, to have been made for similar reasons. Owing to the length of the chapter, their total number is slightly higher than before. Yet six of them, that is two-thirds of a total of nine, are clustered within twelve pages of the printed text. 'His toothpick' > 'his rude toothpick' (229.33), which represents both a stylistic improvement and a concern for greater precision, has already been referred to. Precision and factual accuracy is also the aim of 'Drumcondra' > 'Lower Drumcondra' (188.32), 'unesthetic emotions' > 'not aesthetic emotions' (206.10) and 'northward' > 'southward' (238.23), while improvement of style and expression predominantly motivate the changes 'benevolent mirth' > 'benevolent malice' (210.27), 'ringless' > 'toneless' (227.23),[53] 'old swans' > 'a game of swans' (228.20), and 'brief hiss' > 'soft hiss' (232.31). To this latter variant, the ninth and last in the list: 'brief hiss' > 'swift hiss' (226.27) is related, which however would seem to require emendation. At 232.27: ". . . and a soft hiss fell again from a window above", 'soft' is the original manuscript reading, while at lines 226.27 and 232.31 the manuscript still has 'brief hiss'. In revision, 232.31 follows 232.27 to read 'soft hiss'. But surely it is the sentence at 226.27, which in the manuscript reads "A sudden brief hiss fell from the windows above him . . ." that both occurrences on p. 232 are meant to recall. The revision was presumably retroactive, the author going back to alter 'brief hiss' on p. 226 after having unified the readings on p. 232 to 'soft hiss'. I take it that 'swift hiss' at 226.27 is an error of the E compositor, who misread 'soft' as it was written in by hand in the typescript, and would therefore emend to 'soft hiss' on the strength of the parallel revision at 232.31.

Thus, in the field of substitute readings, where initially all variants were suspect of being authorial in origin, the number of authoritative changes in the Egoist text has been narrowed down to a total of 18. All other variant readings substituted in E for good manuscript readings, that is something like half of the 317 D-E substantive variants, must consequently be classed as unauthoritative. This large group of variants,

then, seems in view of a projected editorial hypothesis to be practically identical in nature with that of the omissions, and by our comprehensive analysis it is thereby suggested that all substantive variants in E, with the exception of a small number of narrowly definable and identifiable readings, are unauthorized. This is a result attractive in its consistency and, though essentially hypothetical, it gains all possible probability from the three-pronged approach to the evidence as divided into three distinct groups of variants. It should be recalled at this stage, however, that the entire large group of the omissions was approached above with the initial expectation of a total lack of authorial interference and has so far been only provisionally designated as an area of exclusively unauthorized variation. Before final conclusions are asserted the omissions should therefore be briefly surveyed once more with regard to the fact that a certain degree of authorial attention to the text has meanwhile been ascertained. The authorial correction of the text was, it is true, evidently reticent and probably superficial, and to have established it as a fact cannot therefore in principle change our conception of the group of the omissions taken as a whole. The characteristics are far too strong which indicate that they were largely typist's errors which went by unnoticed in the author's reading of the typescript. In the three chapters in particular which contain more than one authorial variant each, there is not a single omission which by its nature suggests that it, too, might be authorial in origin. It is only in the latter half of Chapter III that doubts arise whether all omissions observed should be blamed on the typist (or compositor). At the rhetorical climax of the last of the hell sermons there is a passage which in the printed text has three separate omissions in brief succession: According to the rules established by Anderson for the treatment of omitted phrases,[54] which have in principle been accepted above, there is no alternative to regarding these omissions as three errors of the typist. But thus to regard them means to accept that he nodded three times separately in rapid succession, and yet in a curiously systematic way. On literary grounds, on the other hand—and therefore by reasoning which lies outside the area of textual analysis based on the transmitting documents—it is tempting to see the author at work here, pruning an excess of repetitive rhetoric for the sake of stylistic improvement, and a heightened rather than a lessened impact of the words. Under this aspect, the three separate errors of the typist would appear transformed into a single tripartite authorial cut. Were this to prove the only example in the text where omission became suspect of being authorial in origin, an editorial decision to respect it as such would be very hard indeed to defend, however much one's instinctive literary feeling were averse to restoring the full manuscript wording. But very tentatively something like a case can be made out for a repeated incidence of authorial cuts in the second half of Chapter III, whereby these would become identifiable and separable as a group from the other omissions, and thus editorially acceptable as readings in the variant form of the printed text. Three such omissions occur a few pages after the passage quoted which could also conceivably be due to a desire to reduce a repetitiveness of expression (as indicated):

As it happens, it is in close vicinity to these passages that a later intentional deletion is recorded. The first of the errata on EC-W, fol. 2, is "delete 'of herrings'" and refers to "Frowsy girls sat along the curbstones before their baskets [of herrings]" (140.26). This may be pure coincidence, and it proves no more than that Joyce was in fact capable of making a cut in A Portrait—an attitude of authorial self-criticism not readily evident otherwise in this text. This deletion has no intrinsic similarity to the four examples of omission in E here considered, and it can hardly be taken to reinforce an assumption that they be of authorial origin. If of course it were true that they all are genuine cuts, then this would indicate that the latter half of Chapter III, portraying

as it does Stephen's intensely painful self-torture, gave particular pains in the writing and was textually fluid for longer than any other section of the novel. However, in the absence of the Chapter III typescript any argument of textual or of literary criticism in relation to the variant passages must remain highly speculative.

On the whole, then, the variant readings in E caused by the omission of words, phrases and sentences from the manuscript text can now confidently be declared unauthoritative, as can the large majority of those variants in E which are substitutes for good manuscript readings. Conversely: on the basis of the preceding analysis, we consider, out of a total of 317 substantive variants between D and E, only 18 substitute readings, most of 29 corrections of incomplete or obviously erroneous manuscript readings, and possibly six omissions (occurring in four passages in the second half of Chapter III) as authorized. With respect to the body of D-E variance, provisional rules for establishing a critical text of A Portrait of the Artist as a Young Man may be set out as follows:

Of the variants in E,

• 1. Admit all corrections of incomplete and erroneous D readings that are not obviously either miscorrections or stylesheet rectifications of grammar and syntax;
• 2. admit 18 authorial corrections and revisions;
• 3. do not admit other substitutes for good D readings, whether or not they seem individually possible as variants;
• 4. do not admit readings in E which are the result of omission of single words, phrases, or sentences of the D text (with the possible exception of 6 such variants in the second half of Chapter III);
• 5. do not admit the E variation in accidentals.[55]

The Typescript

The hypothesis developed in the preceding pages for the total of the substantive D-E variance, an hypothesis which in turn must serve as the basis for a comprehensive theory of the textual transmission capable of governing editorial decision in the establishing of a critical text of A Portrait of the Artist as a Young Man, has, it is true, a narrow foundation. It depends on an evaluation of those variant readings in E which are substitutes for good manuscript readings; and only if, by this evaluation, it is ultimately correct to accept no more than 18 such variants as authoritative, is it valid editorially to reject the other Egoist readings which belong to this group and, furthermore, to conclude by analogy that a great majority, if not all, of the Egoist variants resulting from the omission of manuscript readings, too, are unauthoritative. It is most fortunate, therefore, that—contrary to previous assumptions—there is no need to trust exclusively in the soundness of a logical construct. Rather, it has in fact been possible to trace large fragments of Chapters I and II of the Trieste typescript against which our hypothetical assumptions can be tested.

The typewritten fragments of the text of A Portrait come from the possession of Dora Marsden, founder of The Egoist, and still its editor when the first ten installments of Joyce's novel were published. They contain the text for most of these ten installments. Of 68 numbered typewritten pages of Chapter I, only pages 1-15, that is the entire first installment plus three sentences of the second, and the single pages 53 and 59 are missing. Of Chapter II, for which the page numbering starts afresh, pages 1-9 and 17-27, plus six lines of p. 28, are extant, which correspond to the first, the third and part of the fourth Egoist installments of that chapter. The last fragment breaks off in the middle of the last Egoist installment of the novel published under Dora Marsden's editorship. There are 71 pages and six lines of typescript in all.

Upon collation against both D and E, a great number of readings, variant and invariant, in the text of these typewritten pages immediately suggest that they are indeed part of the Trieste typescript. Yet wherever there is no significant variation between the text in the typescript and in E, the process of transmission could of course be thought of as reversed. Order and direction of transmission are determined by those readings or typographical characteristics only which are invariant between D and the pages of typescript but variant between these and E. It is thus above all the reproduction in full of Joyce's manuscript system of dashes in dialogue in the typewritten pages, as against the absence of intermediate and final dashes in all dialogue in Chapters I and II in E,[56] which places the text of these pages firmly between that in D and in E and identifies the fragments as part of the Trieste typescript (T). Consequently, such instances of variation between D and T as the apparent, or indeed obvious, misreadings of Joyce's handwriting also attain value as evidence to secure this identification. They demonstrate, furthermore, that T was copied directly from D. That the extant fragments of typescript in their turn were used as printer's copy for E is not merely rendered likely by the circumstances of their preservation, but is also demonstrable from errors of a typographical nature in the typescript,[57] and by a large number of marks (crosses, queries, and the like) which would seem to have been added in the printing-house.[58]

When so much has been established, both the D-T and the T-E variance which occurs within the stretches of text preserved can be more closely analysed. In accidentals, D-T and T-E variation is clearly distinguishable. In both chapters, additional commas are very rarely indeed introduced in T. The Chapter I pages add four commas, the Chapter II pages add three. The typical typescript error in punctuation is rather one of omission of commas. 13 manuscript commas are missing in Chapter I, of which some have been restored in print and some not, and 9 commas are missing in Chapter II, only one of which has been restored in E. The adding of commas to the printed text, although comparatively restrained in Chapters I and II, is thus clearly a compositorial commission. The increase in capitalizations, on the other hand, is pretty evenly distributed between T and E, while the printers again are responsible for the greater part, though not all, of the additional hyphenation of compounds in E. By contrast, it is the rare substantive variant that originates in E. What earlier collation established as D-E variance in substantives is in fact seen to be variation between D and T. This is as it should be according to the basic assumptions of our hypothesis for the transmission from D to E of the entire novel, and it is gratifying to find it confirmed for the sections of T preserved. A grouping of the D-E substantive variants, undertaken hypothetically before for the sake of a coherent textual theory, can now be based on firmer evidence. Moreover, to determine the exact source and point of origin of a variant becomes a practical necessity for defining the nature and degree of authority of the recovered intermediary textual witness.

The omission from the typed text of words, phrases and sentences presents once again the crucial textual problem. D does not indicate any omissions. On the evidence of the typescript, the likelihood does not increase that they represent authorial, and thus authoritative, cuts at a lost stage of transmission between D and T. As T was demonstrably typed from D, such a lost stage could in any case not be thought of as yet another full transcript of the text intervening between D and T. At the most, the possibility of authorial direction in the form of oral or written instructions to the typist might be considered if the need were felt to make the omissions out as authorial cuts. But not a single variant outside the group of the omissions would help to strengthen any speculatively posited authorial attention to the text between D and T. Even the simple corrections evident in T of incomplete or

obviously erroneous manuscript readings could have been made by a typist on his own initiative; it need not be assumed that the author was consulted for them. The correction of one incomplete manuscript reading, indeed, which was not caught by the typist is entered in T in the author's hand ('wondering' [66.23]), thus allowing the inference that the others of its kind in fact were done by the typist and consequently, on the evidence of this group of variants, no authorial correction intervened between D and T.[59] All further circumstantial evidence considered above, moreover, which suggested that the variants due to omission of manuscript readings were errors in typing is in no way invalidated by any new evidence from the typescript. The onus for the considerable degree of deterioration from manuscript to print of the text of A Portrait is still largely on a typist. He had to take all blame in absentia before. Now the extent of his inattention to the text can be more firmly assessed. With respect to omissions, the typewritten pages reveal that errors of omission were corrected, the corrections always being typed in, in only a few instances when they were caught in the course of the typing. A systematic reading by the typist of the finished typescript against copy seems not to have taken place. Beyond that, there is only one instance in the entire 71 pages and six lines of typescript preserved where the author himself caught a typist's omission and corrected it. It is this instance alone which must take all burden of proof,[60] in so far as proof can be supplied by the typescript, that the omission of manuscript matter in print was due to typescript error and does not represent intentional authorial cuts. The manuscript sentence: "Perhaps that was why they were in the square because it was a place where some fellows wrote things for cod" (43.24) is rendered in the typescript as: "Perhaps that was why they were because it was a place. . . .". The author adds 'there' above the typewritten line, indicating the place of insertion by a caret between 'were' and 'because'. Clearly, 'in the square' are words erroneously omitted by the typist. The error was caught by the author (as so many of its kind were not) and corrected without collation against the manuscript.[61]

The typescript thus bears marks which unmistakably show that it was read by James Joyce himself, though they indicate at the same time that his reading, too, did not take the form of a thorough collation against the manuscript. This goes some way towards explaining why, as errors, the many omissions of manuscript readings should have been so consistently overlooked. Moreover, few and far between as are the marks which signalize Joyce's presence, they are also proof that his reading (as hypothetically assumed) was perfunctory only, and probably hasty. In the extant fragments of typescript there are only three instances in all of substantive authorial correction. In addition to the insertion of omitted 'wondering', and of 'there' for omitted 'in the square' the third authorial correction is 'Father' > 'Mr' (30.1). That part of our hypothesis which posited a scarcity of authorial corrections among the Egoist readings which replace good manuscript readings would thus also seem to be confirmed by the typescript. As it has already been shown that the text appears to have been given no authorial attention between D and T, it follows that all variants of this kind, too, were introduced by a typist and are non-authoritative, and that authority belongs to those variant readings only which are seen to have been entered in the author's hand in T. On the narrow basis provided by the extant fragments of typescript, this is an argument mainly by negatives. None of the passages of text in Chapter II where authorial correction was assumed are preserved in T, so that it cannot be positively shown that those variants alone are authorial in origin which were hypothetically singled out as such, although it is apparent from as much text as is extant that no other variants replacing

good manuscript readings are authorial of which this was not expected. The same is true for the much longer fragment of Chapter I. Of the three variants in this chapter hypothetically assumed to have been authorial, the assumption has been confirmed without reservations for one ('Father' > 'Mr'), and with some modification, not touching the fact of its authorial origin, for a second ('in the square' > 'there'). The third member of the group, however, is still invariant in T: the manuscript 'avenue of chestnuts' leading up to Clongowes is still an 'avenue of chestnuts'. The reading must have been altered between T and E.

If it is to be maintained that 'limes' for 'chestnuts' is an authorized variant, an influx of authority in some form must be assumed at a stage of transmission after the typescript was authorially corrected to the extent observable in its extant fragments as here discussed, and before the printed Egoist text as typeset and proofread was published. The assumption as such is strengthened by the observation that 'limes' for 'chestnuts' in Chapter I is not an isolated instance of significant T-E variation. (There is no corresponding variation in the Chapter II fragments.) For the extant text of Chapter I, there is a total of nine substantive T-E variants. In two instances, T copies the manuscript readings correctly and there are marks of the printer's pen or pencil set against them in T which suggest that the subsequent E variants originated in the printinghouse and are therefore unauthorized. In the remaining seven instances, on the other hand, T also varies from D and it is the original manuscript reading which has in each case been restored in E.[62] In three of these seven cases the T variant is an error to which the manuscript reading is the obvious alternative; a compositor or printinghouse proofreader could easily have made the correction. Yet in the remaining four cases the correction of the T error can have been made only from a knowledge of the manuscript reading as restored. There are thus five instances of substantive T-E variation—interestingly enough confined to the second and third Egoist installments—which make it an inevitable conclusion that the text of Chapter I, or part of it, was referred to authority in the course of transmission between extant T (as authorially corrected) and E. In other words, the Egoist text of Chapter I was to some extent authoritatively proofread before publication. One possible explanation of this fact would be that proofsheets of the chapter, or of the two installments

concerned, were sent to Joyce in Trieste. No further indication, it is true, derives from a collation of corrected T and E that this was the case, nor is any documentary evidence available from letters or the like to show that Joyce ever proofread any part of the Egoist text. Yet if the typescript used as Egoist printer's copy was the only copy of T which ever existed, the variants observed would allow no other conclusion.[63] If however, T was typed, say, with a carbon copy, and thus existed in duplicate, one might assume that the copy of T which did not serve as printer's copy contained the additional corrections, entered by hand by the author and/or typist, and was used for proofreading by some agent other than the author. No immediate proof is available which of these explanations, if any, is correct. Nevertheless, authoritative proofreading assuredly took place between corrected T and E. It did not broadly affect the text, and the few variants involved assert their authority mainly on their own strength.

The hypothetical concept of the transmission of the entire text of A Portrait from D to E, such as it was derived from a comprehensive collation and analysis of these two textual witnesses, is thus in no way invalidated by further investigation of the transmission processes as controllable in those sections of the intermediary textual witness, the typescript, which happen to have been preserved. For most of Chapter I and for substantial fragments of Chapter II, that is for those stretches of text which correspond to the extant pages of typescript, the postulates of our hypothesis have been fully confirmed, and a clear differentiation of the several stages of transmission of the text from manuscript to print is amply indicated. The question which remains to be answered is what further inferences may be drawn for the full text of the novel by extrapolation from the positive facts established about the transmission of the first two chapters. The most important fact not yet considered of the extant fragments of T is that each of the first two chapters was typed by a different typist. This is clear from a difference in spelling habits: in the Chapter II fragments of the typescript, the name of the novel's hero is consistently spelled 'Stephan'; it was probably the author himself who painstakingly changed the spelling

back to 'Stephen' in almost every instance. Moreover, Joyce's system of dashes in dialogue is regularly complemented by additional punctuation at the end, and sometimes rather illogically also in the middle, of direct speech in Chapter I, whereas in Chapter II the manuscript punctuation of dialogue is copied faithfully without additional marks of punctuation. Statistics derived from the incidence of substantive variance further corroborate a differentiation of typists. There is an average of 2.3 substantive variants per printed column in the 27.5 columns of Egoist text for Chapter I, as against an average of 3.9 per column for Chapter II. Leaving out of account the 3+5 variants which in Chapters I and II are with some certainty authorial corrections and revisions, the figures are 2.2/3.7. These figures comprise substantive variants of typescript and printinghouse origin. But since it can be positively established that the printinghouse compositors were responsible for only a minority of the substantive D-E variants—in the sections of the text preserved in typescript, the compositors in Chapter I introduced 9 out of 53, or roughly the sixth part of the substantive variants, and in Chapter II 3 out of 29, or approximately only one-tenth—the averages per column of printed text give a fair indication of the relative trustworthiness of the two typists. The Chapter II typist was significantly less faithful to copy in substantives (though more so in accidentals); his spelling 'Stephan' may even suggest that his native tongue was not English.

When statistical calculations are extended over the remaining chapters of the novel, it is remarkable that the higher incidence of variation and error per column of printed text is confined to Chapter II. The corresponding figures for Chapters III-V (with and without authorial variants, and giving the typist the benefit of the doubt by assuming that the omissions in the latter part of Chapter III are authorial) are 2.4:2.1/2.35:2.35/2.1:1.9. They are thus very close to the incidence of variation in Chapter I (2.3:2.2). Noting that when the printed text introduces Joyce's dashes in dialogue from halfway through Chapter III onwards, it does so not in the strictly faithful manner of the typescript of Chapter II, but rather according to the pattern of that of Chapter I, combining the dashes with additional punctuation, one is tentatively led to conclude that the identical typist was employed on Chapters I and III-V, while a second person typed Chapter II only. The general uniformity of the kind of variants introduced and errors committed would appear to support the conclusion. The uniformity, it is true, extends over all five chapters. The statistics alone would, in the absence of any part of the typescript, not be very reliable as an aid to differentiating typists, since their figures indicate a varying amount

only, and not a difference in nature of variants and errors. But once a second typist for Chapter II has been identified by several aspects of his extant handiwork, it is of no great consequence that he happened to be prone to the same kind of failings towards the manuscript as was the typist for Chapter I. What is important is inferentially to conclude that he was responsible for Chapter II only, and thereupon to recognise that both the statistical figures and the uniform nature of the variants, that is quantitative and qualitative arguments together, go to indicate that Chapters I and III-V in the typescript were the work of one other person, and one only. If, then, this conclusion is correct, it should greatly aid an editor in his evaluation of the total body of the D-E variation in substantives and, in particular, increase the assurance with which he posits that only a small minority of the substantive variants in E are the result of authorial correction and revision of the typescript.

Finally, the facts and inferences can be played with to speculate about the relative timing involved in the completion of the several sections of the manuscript and the typescript. By two typists, Chapters I and II could have been typed simultaneously. At least, it is not impossible to conceive that they were begun at about the same time, or else that their typing overlapped for some days, or a week or two, if it was Joyce's intention to get as large a section of the novel as possible to London as quickly as possible after Ezra Pound's enquiry of December 15, 1913, about printable material. But Chapter II was certainly not completed in typescript when Chapter I was, or they would have been dispatched together. Actually, the same typewriter may have been used for both: there are no typed underlinings in either chapter, and parentheses have regularly been substituted by dashes. Thus it is equally possible that they were typed in short succession after each other. It is not known when Chapter II arrived in London. But Chapter III, although belonging to that part of the novel which is assumed to have existed in final form since 1908, did not get there until July 21 (Ellmann, p. 365). Perhaps a second typist was employed on Chapter II because the first one was unavailable between January and July—unless the completion of the typescript for Chapter III was delayed by an interruption in the writing out of the faircopy manuscript between Chapters II and III, or in the course of Chapter III. Only a physical examination of the manuscript itself might throw light on the circumstances under which Chapter III was copied out and typed.

On August 1 the war broke out, cutting off almost completely all postal connections between England and Austria. On September 1 Chapter III ran out in The Egoist, and on November 11 Joyce managed to dispatch the typescript of Chapters IV and V via Switzerland.

Provided that the typescript was completed in the regular order of the chapters, I would tend to infer that the typing of Chapters III to V proceeded at a fairly even pace between July and October, and consequently that there was no last-minute rush in the writing of the novel itself. The hiatus in publication between September 1 and December 1 has been taken to indicate that The Egoist caught up with James Joyce too soon (Ellmann, p. 365; Anderson, pp. 182 ff.). But surely it was primarily due to the outbreak of the war. It was probably an editorial unwillingness to have to break off in the middle of a chapter in case The Egoist would not be able to continue publication at all which made the installments of Chapter III on August 1, August 15 and September 1 as long as they are. At the rate of publication of the previous chapters, Chapter III would probably have lasted until October. Chapter IV is short, and if it was begun to be typed soon after completion of Chapter III in late July, it ought to have been ready sometime in August and could, but for the war, have been dispatched to London to succeed Chapter III without gap in The Egoist. After Chapter IV, there was still Chapter V to be typed, which is by far the longest chapter, constituting almost one-third of the entire novel. It is the state of its text in E which indicates that it was probably completed in typescript under no special time pressure. It has the lowest incidence of substantive variation and error and, moreover, it was given greater authorial attention in the typescript than any of the previous chapters. By contrast, Chapter IV is perhaps the only chapter which appears not to have been authorially corrected at all. Had it, by November, been sitting about in typescript for so long that Joyce in the end forgot to read it? Or was Chapter IV perchance the last of the chapters to be written? It is perhaps significant that it is the only chapter of which several of the faircopy manuscript pages still show signs of thorough stylistic revision. It is also the chapter which Joyce in July 1915 claims to remember so well as to be able to restore the censored sentences without the aid of his manuscript. A thorough analysis of the manuscript might throw light on the question, and also quite generally serve as a test to whether the preceding speculative inferences about the successive completion of manuscript and typescript are tenable.

The Book Editions

Our final discussion must be brief of the stages of transmission of A Portrait of the Artist as a Young Man from the first printing in The Egoist to the 1924 Jonathan Cape edition. The publishing history has

essentially been dealt with by Chester G. Anderson. He has also considered many of the editorially relevant facts and inferences to be derived from it in his discussion of the Egoist tearsheets which served as copy for H, and the Y corrections of H for B (see pp. 186 ff., 196 ff.). In so far as a modification of his views proved necessary, this has been given in the course of our analysis of EC-W, YTW and HB. Beyond that, it may suffice to summarize somewhat categorically to what extent the text of the novel as published in H and B may be accepted as authoritative. With respect to substantive readings, all those corrections and revisions represent a final authorial intention editorially to be respected which James Joyce himself undertook in full view of the text as transmitted invariant from the manuscript. There are no problems here in so far as the several rounds of authorial attention to the text are documented in EC-A and Y. The one question not previously considered is whether the Y corrections are the only stage at which the author took influence on the text for B. There are at least two substantive variants in B which would seem to warrant a closer investigation. At 215.27, the text in B reads "When they passed through the passage beside Kildare house . . .", as against ". . . beside the royal Irish academy . . ." in H, and at 96.11 'ineffectualness' (H) is altered to 'ineffectiveness' in B. Neither Y nor HB contain directions for these changes. But as late as November 25th, 1917, more than two months after the Southport printers had begun setting up B, Harriet Weaver was writing to Joyce: "All your corrections have been made, including those you asked for in your last letter." A careful scrutiny of all H-B variants, still to be undertaken, must endeavour to ascertain which were the additional changes which Joyce had requested; there can be little doubt that at least the first of the two alterations quoted were among them.

With respect to accidentals, the situation created by the author's corrections prior to H and to B is somewhat more complex, and for an editor methodically quite intriguing. When faced with the close to 600 additional commas in the text of the Egoist serialisation, Joyce went about restoring the light punctuation of his manuscript with considerable determination. This process continued into Y. His paramount concern doubtless was for the rhythms of his text, the essential qualities of which—"to liberate from the personalised lumps of matter that which is their individuating rhythm"—had been early adumbrated in the very first paragraph of the narrative essay "A Portrait of the Artist" of 1904. Yet for the flexible rhythms of his novel, its semicolons and its colons, placed and distinguished from each other with great subtlety in the manuscript, are as important as its commas. It is surely

an indication of the author's increasing distance in time from a state of complete immersion in the work that, in the course of his corrections, he was much more permissive towards the printers' alterations of the original semicolons and colons. As for the commas, of course, many of those added by printinghouse compositors were left standing in the text despite all authorial efforts to eradicate them. There can be no question for an editor but that he reproduce in a critical text as closely as is at all possible the manuscript pattern of punctuation.

However, Joyce's attention to the accidentals of the text as printed in fact took two directions. As has been noted above, his reduction of next to all capitalization and his preference in print for compounds written together as one word is tantamount to a restyling of the text with respect to these accidentals. In so far as it can be positively ascertained that they were made by the author, the changes are part of the total authoritative correction and revision of E and H. As such, they invalidate, in part, the pattern of the manuscript inscription which cannot be editorially restored whenever positive authorial direction is given to depart from it. Neither, however, should a critical editor (as opposed perhaps to a publisher's editor) for the sake of a new typographical consistency, and in an attempt to finish what the author began, go beyond his alterations. Quite a number of original capitalizations and word divisions in fact survived Joyce's restyling. These cannot but be left untouched in a critical text, which will thus, in the treatment of accidentals, appear thoroughly inconsistent.[64]

Inconsistency of the kind just advocated, however, is really an expression of editorial consistency based on a systematic enquiry into the extent of authority in each substantive textual witness. With regard to Joyce's Portrait, such consistency should be firmly extended to the last of the substantive texts, that of the Jonathan Cape edition of 1924 (J). Because its proofsheets have not been preserved, it presents greater textual problems than H and B. What attention James Joyce

gave to it can only be inferred from external facts and the internal evidence of textual variants. A complete collation of B and J to ascertain all internal evidence has not yet been undertaken. But more is to be known of the external circumstances pertaining to this last phase of authoritative publication of the novel than has hitherto been realised. As mentioned by Anderson, Joyce reported to Harriet Weaver from Saint-Malo on August 16, 1924, that he had been reading revises for Jonathan Cape for ten days and had dispatched the proofs that day.[65] The company ledgers show that the book went into print on August 28.[66] Antedating the August letter is a letter to Harriet Weaver of July 11, 1924, also from Saint-Malo. It states: "Then Mr Cape and his printers gave me trouble. They set the book with perverted commas and I insisted on their removal by the sergeant-at-arms. Then they underlined passages which they thought undesirable. But as you will see from the enclosed: They were and, behold, they are not" (Letters, III, 99 f.). I take this to mean that Joyce had received two sets of proofs from Jonathan Cape before July 11. The context of the letter ("I left Paris in the usual whirl of confusion . . .") suggests that the second set arrived in Paris after Joyce's eye operation of June 11 and before he left for Saint-Malo on July 6 or 7. Probably this was the set which he showed to Sylvia Beach who records her "amazement at the printer's queries in the margins".[67] The author's refusal to cut at the request of the printers had apparently been accepted by July 11. "The enclosed" in the letter to Harriet Weaver would seem to have been some token of consent from Cape to publish the text entire. The question is whether Joyce voiced his refusal by letter only or in a note accompanying the return of the corrected second proofs. Should he indeed not have had the time or energy to read them in the "whirl of confusion" before his departure from Paris, one might be left to wonder whether the proofs with the printers' underlinings were the revises which were finally not read until August. But Joyce was not only habitually prompt in matters regarding the publishing of his books, but, as the letter of July 11 also records, he was able after the eye operation to see to the proofing of the installments of the French translation of Ulysses due to appear in the review Commerce (Ellmann, p. 573). It is therefore likely that the second proofs were returned before he left for Saint-Malo, however superficially they may have been read. If this was so, the revises which he corrected for ten days in August were the third round of proofs read on the Jonathan Cape text.

James Joyce, then, obviously took full advantage of the one and only opportunity he was ever given to see A Portrait through the press. This does of course not mean that the edition into which his efforts went ten years after the completion of the manuscript presents a text to override all previous texts. The inevitable deterioration of the text in ten years of transmission has not been remedied in the edition of 1924. Nor has the text been extensively revised. Even a cursory perusal of J will satisfy an editor that the authorial proofreading was not by any means on the scale of Joyce's reading and revising in proof of the text of Ulysses. The comparison with the later work indeed emphasizes the remarkable reticence against change and revision which Joyce showed towards the text of A Portrait at every stage of correction. By analogy to the earlier rounds of authorial correction of the text, it should be expected that the proofreading of J which did take place affected both accidentals and substantives. That Joyce paid attention to the accidentals is proved, for example, when J, restoring the manuscript punctuation, reads: "What is this your name is?" (50.1) against the typescript error perpetuated through B: "What is this? Your name is?" But it is an exceptional case when it can be proved on internal evidence that the author was responsible for a variant in accidentals. Generally, the lack of documentation prevents editorial acceptance of such variation. The substantive variants in J, on the other hand, are susceptible to evaluation by which it should be possible satisfactorily to identify the authorial corrections and revisions. For example, the manuscript sentence: "The doomsday was at hand." (113.11) was transmitted invariant through H. B, by the authorial direction of Y, changed to 'Doomsday', yet J reverts to the manuscript wording 'The doomsday' in what must be considered as the author's final decision on this reading, overruling the revision of Y. Similarly, the typescript error 'fellows' at 43.19 was transmitted unnoticed through B. J restores the singular 'fellow' of the manuscript. Finally, a subtle revision at 77.35, 'moment' for 'movement', indicates the author's care in proofreading. With 'movement' the manuscript had picked up 'a movement of impatience' of 77.28, which itself is a recollection of 74.28 and is only incidentally woven into the description of the situation of tension in the conversation between Heron, Wallis and Stephen on p. 77. Only in the 1924 re-reading of the text did Joyce himself realise that the proper reference of line 77.35 should be to 'a shaft of momentary anger' of 77.12.

When all authoritative stages of the transmission of the novel have been analysed in detail, a comprehensive textual hypothesis to govern editorial decision in the preparation of a critical text may be formulated in a revised set of directions:

• 1. Base a critical text of A Portrait of the Artist as a Young Man on the holograph manuscript (D) of 1913/14.
• 2. In accidentals, accept the manuscript system of punctuation. Follow authorial directions as contained in EC-A and Y to restyle the manuscript capitalizations and word divisions.
• 3. In substantives, accept all authorial correction and revision as documented in the fragments of T, in EC-A and Y, or as ascertained by inference from an evaluation of the total body of substantive variance between each of the authoritative editions. Except for the corrections of incomplete manuscript readings in E (in so far as they are not manifest miscorrections or stylesheet rectifications of grammar and syntax), reject all non-auhoritative variation in substantives.

With our present knowledge of the publishing history and the nature of the textual transmission of the novel, a critical edition, and maybe even a definitive text, of A Portrait of the Artist as a Young Man is not impossible to attain.