An Inquiry into 'Ulysses: The Corrected Text' **

John Kidd.  Papers of the Bibliographical Society of America, 1988.  Digital 1989.  Online 2007. 

* * 2nd file of a book by John Kidd. Link to Introduction , Table of Contents , Author , Joyce , or Textology .

Transcribing the Rosenbach Manuscript

The first announcement in print of a project to edit Ulysses from manuscript sources came in the James Joyce Quarterly in 1976, at the head of A. Walton Litz's review of Ulysses: A Facsimile of the Manuscript, edited by Clive Driver:

    Textual study of Ulysses has recently entered a new and exciting phase, marked by the appearance of several important collections of previously unpublished material. Our understanding of how the novel was written can now be greatly refined, and that most difficult end-product, a reliable text, suddenly seems a realistic goal.  (Hans Gabler and Michael Groden are already engaged in preliminary work leading to a definitive text.) (JJQ 14:1, p. 101)

Mr Litz is correct in emphasizing that the publication in facsimile of source documents can be a helpful step toward a critical text. A text may be established from materials otherwise little-known, but long and close scrutiny by many pairs of eyes increases the chances for proper treatment of complex issues. Some of those eyes will have to be directed at the originals, of course, not only at the facsimiles. The 1975 reproduction of the Ulysses manuscript at the Rosenbach Foundation in Philadelphia provides a perfect example of how one document can transform our understanding of a literary work. At the same time, many of the central conclusions reached by its editor, Clive Driver, were disputed immediately upon the facsimile's publication. The debate about the status of entire episodes of the Rosenbach fair copy intensified with Hans Walter Gabler's 1977 claim that to handle the "dynamic process of artistic creation" in Ulysses, "current editorial theory will require extension by a rationale of 'continuous copy-text' " (The Library 32:2, p. 182). As a way into Ulysses: A Critical and Synoptic Edition, the Rosenbach Manuscript is the ideal prolegomenon.

Until the Rosenbach Manuscript (hereafter called R) was reproduced in facsimile, it was generally thought to be a unified holograph in the direct line of transmission. Clive Driver took it to be such after collating the holograph with both the 1922 first edition (22U) and the Little Review (LR) serialization (1918-1920). The third volume of the set is a reduced photofacsimile of the first edition with handwritten notation of differences between R and the 1922 text as well as variant readings from the fourteen Little Review episodes.

By an oversight that would prove serious, the fragmentary extant typescripts were not collated. Driver assumed variation between R and LR to have occurred largely due to authorial revision on the typescripts, a reasonable assumption given Joyce's well-known habit of extensive revision at every level. Sensing that the reduced facsimile might not easily accommodate all levels of revision and corruption, and weighing the huge task of gathering the far-flung documents in microfilm, Driver collated only the Rosenbach Manuscript, the first published edition of 1922, the Little Review and two errata lists for the first edition. By Litz's account, he and Michael Groden, Philip Gaskell, and Hans Gabler jointly puzzled over the discrepancies between R and the typescripts at SUNY-Buffalo. Their conclusions, which were at once startling and convincing, were published as Litz's piece in JJQ, Gaskell's TLS review, Gabler's review in The Library, and an appendix to Groden's "Ulysses" in Progress.

One need not here restate the evidence that for ten episodes of Ulysses the Rosenbach Manuscript was typist's copy and for eight episodes it was collateral to a working draft given to the typist. (One episode, Calypso, may have been typed twice, meaning that R was typist's copy, but that an intervening and lost stage occurred. Understandably, this unsmoothed wrinkle in accounting for one of eighteen episodes leads to the common statement, slightly inaccurate, that nine episodes of R are in -- and nine out of -- the main line of descent.) While Joyce did send changes to his typists, the absence of marks on R at points of difference between R and the typescript in eight episodes makes it unlikely that R served as typist's copy. The affected episodes are, to use the Homeric names Joyce gave them in his notes and letters, but never in his book: Lotus Eaters (5), Hades (6), Aeolus (7), Lestrygonians (8), Scylla and Charybdis (9), Sirens (11), Nausicaa (13), and Oxen of the Sun (14). (Litz's JJQ stemma, p.109, accidentally omits episode 8, though he includes it elsewhere.) Episodes 1-3, 10, 12, and 15-18 are now accepted as the portions of R in the direct line of transmission. These nine episodes go by the names of Telemachus, Nestor, Proteus, Wandering Rocks, Cyclops, Circe, Eumaeus, Ithaca, and Penelope. For Calypso, episode 4, R is probably the source of the first of two typings. (It should be understood that in addition to forbidding the use of the Homeric names in the published text, Joyce never used episode numbers in the book proper, an authorial prerogative violated for the first time by Gabler. In The Corrected Text, each episode begins with a large numeral paradoxically placed in brackets as if to signal its own inappropriateness to the reading text of the novel. Now all trade editions except one of the two Penguins are further marred by episode numbers at the foot of each page and running marginal line counts on all pages.)

At first it was thought that there might have been two typings of episodes 4-9, 11-13, and 15-18. Gaskell suggested this in the earliest of the four major responses, but he soon came around to Gabler's theory about the onetime existence of working manuscripts from which R was fair copied. These lost drafts were revised again before typing, although the act of fair copying doubtless brought revisions to mind which were entered back into the draft. If Joyce revised as he copied but failed to register a change in the draft, then R might represent a point of development more advanced than the one actually transmitted. This hypothesis, which first appears in Gabler's 1977 Library review of the facsimile, leads Gabler from his evaluation of Clive Driver's Bibliographical Preface into the formulation of a rationale for the 1984 Ulysses. In effect, Gabler's review is an announcement of his own edition-in-progress. The review is replete with original hypotheses about the composition of Ulysses, and supplies details of high importance; oddly enough, it is not listed in the bibliography of the 1984 edition. This may be because much of the evidence brought into play in the review is contradicted by the edition which followed it by seven years. Because the edition itself neglects to reconcile earlier claims, or even acknowledge them, the mislaid arguments of 1977 deserve reconsideration here.

Single Rosenbach Leaves as Typist's Copy: Nausicaa leaf 55

One recent hypothesis about the Rosenbach Manuscript is particularly intriguing: not only are some episodes in and some out of the line of transmission, but certain individual --though unspecified-- leaves of the "out" episodes may in fact be "in." Michael Groden put it this way in his 1985 review of the Synoptic Edition :

    It used to be assumed that, even though the relationship between the Rosenbach Manuscript and typescript varied among episodes, it was constant within each episode.... But Gabler discovered the extant manuscript is not entirely outside the line of transmission for the other nine episodes. Rather, the relationship at times seems to vary from page to page in those nine.... (143)

This speculation has been repeated by the Foundation staff in The Rosenbach Newsletter (No. 2, Sept.1985, p. 2):

    In fact, the new edition notes that not only are nine (half) of the episodes entirely within the line of transmission of the text but that all the others have at least some bearing on the line of transmission, their relationship to the typescript and printed text varying from page to page.

Were this so, for episodes whose fair copy is on the whole out of the line of transmission, certain leaves would intermittently carry the text Joyce gave his typist, saving us some of the guesswork about the lost working draft. This possibility is appealing: if portions of the typist's copy are in fact not really lost, we can collate the fortuitously preserved leaves against the typescript, and reject all differences as transmissional errors. Although Gabler postulates several such migrating faircopy leaves, he identifies only one, Nausicaa leaf 55.

The chief obstacle to confirming the "discovery" of mixed lineal status for some episodes is a shortage of evidence presented in 1984. Hundreds of leaves are available for classification as fair copy or working draft, but the editor never bothered to tell us which leaves he thinks went to the typist and which did not. For the eight episodes judged out of the line, the 1984 editor remains silent or vague--with one notable exception discussed below. Lotus Eaters may have "individual leaves" (p. 1732) and Oxen of the Sun "some individual leaves" which had gone to the typist (p. 1745). No mention is made of leaves being in the line for five of the eight relevant episodes: Hades, Aeolus, Lestrygonians, Scylla and Charybdis, and Sirens. For Lotus, Nausicaa, and Oxen the mixed status is casually granted without specifics. Because the edition is an eclectic assemblage of documents and the texts they carry, the relationships for each leaf should have been clearly stated.

Indeed, in only one instance does the editor identify a specific fair copy leaf from these eight episodes as being in the line of descent: Nausicaa 55. Two substantive emendations of a famous passage of Ulysses hang on the assumption that Nausicaa 55 was inserted into the working draft used by the typist, then safely returned to its place in R. The 1984 synopsis itself provides all the data necessary to conclude that even this one leaf cannot have played the role in transmission attributed to it-- much less support a wider theory of mixed manuscripts.

Every student of Joyce knows the erotic climax that resulted in fines for the Little Review publishers in 1920. After Bloom's beachfront onanism, a counterpoint to his last flush rises over the bay:

    A lost long candle wandered up the sky from Mirus bazaar in search of funds for Mercer's hospital and broke, drooping, and shed a cluster of violet but one white stars. (22U 361.26)

In rejecting the 1922 "A lost long candle" for "A last lonely candle", Gabler provides in a textual note for Nausicaa 13.1166 a paleographic argument which develops into a miniature thesis about the succession of drafts for Nausicaa:

    814.16 last lonely] The early draft alters its original "lonely last" to "last lonely" by deleting "lonely" and re-inserting it between the lines in a way that Joyce misread it as "long" in copying. "Last", which is clear in the draft, is ambiguous in R and is misread as "lost" by the typist. Restoring the uncorrupted draft reading and analysing the transmissional corruption in this manner implies that fol. 55 of R was copied directly from the early draft and belonged to the minority of faircopy leaves incorporated in the typist's copy. (84U 1744)

Because the edition does not transcribe the extant early draft, we should set aside for a moment the question of whether Nausicaa 55 was copied from it. But the texts of both the Rosenbach Manuscript and the typescript are available in the synopsis to test the judgement that leaf 55 was "incorporated in the typist's copy." Although not demarcated in the synopsis, R 55 begins in mid-sentence with "beautiful calm" (13.1163/ 361.23), and ends with "Filthy" (13.1185/362.07). In the third line of the synopsis after R 55 begins is a half-bracketed addition at "level R." At the head of the Nausicaa textual notes, R is defined as "final working draft revision in the act of fair-copying, not entered back into the final working draft document" (p. 1743). In other words, the Rosenbach Manuscript occasionally has revisions which occurred to Joyce as he copied, changes he then failed to enter on the (apparently now lost) working draft ultimately given to his typist. Anything labelled level R is a judgement that the typed descendant of the lost working draft differs from R due to a revision made in R. The opposite could be the case, of course: the typescript may differ from R because of a post-Rosenbach change in the lost working draft.

In the case of Nausicaa 13.1164 (361.24), it was decided that the difference between the fair copy R "looking down so peaceful" and the typed "looking down" reflected an addition Joyce made to R but never entered on the lost draft given to the typist. Consider the resulting curiosity: the textual note at page 1744 states that this leaf of R was the typist's copy; but the synopsis states at page 814 that this leaf has a phrase never given to the typist. The only fair-copy leaf specifically identified in 1984 as going to the typist of Nausicaa is synoptically annotated as being out of the line of transmission. Contradicting the textual note, the synopsis indicates that the typist worked from a lost draft for this segment.

Gabler is right that the typist never saw the fair copy, but that does not insure that the extra words "so peaceful" are a post-working-draft addition that Joyce forgot to enter back into the typist's copy. On the contrary, they are as likely either a phrase overlooked by the typist or a post-Rosenbach deletion by Joyce. The insertion into the synopsis as a level R addition, and the inclusion in the reading text, is simply a guess. (That the synopsis is constructed of guesses has eluded many reviewers who think a computer sorted out the levels of accretive text while excluding all errors of transmission.)

Dropping down fifteen lines in the synopsis, still within the transcription of R 55 and later accretions, we find the evidence that categorically eliminates the fair copy from consideration as the typist's source. In an anthropomorphism echoed later in Finnegans Wake, the rock of Howth "felt gladly the night breeze lift, ruffle his many ferns" (on the fifth proof, Joyce revised "many" to "fell of"). The early draft has only the verb "lift", and R has only the verb "ruffle", but the typescript has "lift, ruffle", the 1922 reading (362.01). If accurate, the footnote to "lift" at 13.1179 (362.01) documents the impossibility of R being the typist's source:

    lift,] STET aP, tC; ABSENT aR.

Lacking "lift", R 55 was not the typist's source for "lift, ruffle his many ferns". Rather, this segment of Nausicaa was copied by Joyce from the extant early draft "aP" ("lift") into a lost final working draft ("lift, ruffle"), then typed. In addition to misanalyzing the genetic relation of the documents in the note, two errors in the synopsis prevent users from reconstructing the revision process. The absence of "ruffle" from the early draft is not recorded, and the footnote just cited above gives "lift" a comma in the lemma which it lacks in the draft. The draft reads "lift his many ferns" (JA 13:237.26).

Like all of Nausicaa, leaf 55 of the Rosenbach Manuscript was copied from the lost final draft, preserved in an exceptionally clean state, and never underwent the odyssey postulated in order to make the strained emendation of "A lost long candle" (1922) to "A last lonely candle" (1984).

Nausicaa 55: Paleography and Paper

The preceding analysis has been textual, comparing words and punctuation in three documents to determine relationships. We found nothing textual to support the 1984 thesis of mixed leaves in the Rosenbach Manuscript. Yet paleography is part of the textual note's argument as well: "deleting `lonely' and re-inserting it between the lines in a way that Joyce misread it as `long' in copying" (p. 1744). The early draft (JA 13:237.13) does not corroborate this statement. The paragraph began "A lonely last candle", then Joyce scored "lonely" and with a caret inserted a word before "candle".

To my eyes, the insertion is "long" for three reasons. The inscription has four letters, and "lonely" (six letters) cannot have been intended because there is not a second "l". The closing "g" matches that of a revision to "long" in line 24, and the "long" ... "long" revisions might work in tandem. The third argument against Joyce's misreading "lonely" as "long" is that, when copying the draft, he would have read "A lonely last" a millisecond before encountering his insertion. Were there any doubt about the insertion, "lonely" would be before him as a reminder. Suggested by the writing itself (four letters ending in "g"), a revision to "long" eleven lines below, and the deleted but unobscured "lonely" nearby to prompt Joyce if needed, I conclude that "lonely" was revised to "long", the reading of the draft, the Rosenbach MS, and all editions before 1984. Having acknowledged in its note that R reads "long", as does the typescript, the 1984 emendation of its copytext depends on a questionable interpretation of an early draft and rejection of autograph and all editions.

Even the physical makeup of the Nausicaa portion of the Rosenbach Manuscript tells us that the probabilities are against Gabler's guess that leaf 55 was copied from the extant early draft (instead of the later draft, as was the rest of Nausicaa) or that it migrated out of its sheaf to join the typist's source. The tests which might be applied to these manuscripts include differences of paper, number or placement of fastener stab marks, disruption of foliation numbers, stains matching other documents (food, ink, water, cosmetics), changes in ink, text markings to alert the typist to the beginning or end of copy, extra space at the bottom of a leaf, paper folds, changes in the slope or margins of text, or stray markings attributable to the typist. By all of these tests, leaf 55 fails to stand out. A survey of the Rosenbach Manuscript original in Philadelphia suggests that the term "fair copy" is nowhere more applicable than in the pedestrian uniformity of the 60 numbered leaves of Nausicaa. The physical manuscript of Nausicaa, unlike the less uniform episodes with mixed papers, wrong foliation, etc., does not corroborate the conjecture that leaf 55 was copied from the early draft or ever left the Rosenbach Manuscript to join the lost final working draft as typist's copy.

Thus does one leaf in the Ulysses holograph cast doubt on both the broadest claims and the minutest responsibilities of the 1984 edition. It is the only piece of manuscript singled out to uphold a theory that the Rosenbach Manuscript is a mixture of fair copy and working drafts whose transmissional status varies literally, we are told, from leaf to leaf. The edition's synopsis and the Rosenbach leaf itself witness the contrary.

In the attempt to reconstruct the genetic progress of leaf 55, we were hampered because major differences between the early draft and R (such as the absence of "ruffle") are not recorded and a footnote grants a comma to a draft reading which lacks it. Beyond the breakdown of the synopsis, however, is the questionable choice of "lonely" against "long". Is "lonely" of the early draft "uncorrupted" (as the note asserts) or did Joyce revise to "A long last candle" as in the final manuscript and in one of the most famous passages of the published book?

The theory of "continuous manuscript text" gives nothing to fall back on here because both "lonely" and "long" belong to the so-called "copy-text" of Joyce's autograph. The combination of self-contradiction, transcription error, and omission of draft readings grants little reason to set aside what Joyce wrote out at least twice and passed in the revised typescript, revised proofs, and all editions. "A lost long candle", Joyce's final autograph inscription, should stand; the confusions which led to its recent disappearance from Ulysses are serious not only in the details, but for the shadow they cast on the widely repeated claims of a "discovery" about the textual authority of the Rosenbach Manuscript.

The Peril of Working from Facsimiles

In his Library review of the Rosenbach facsimile, Gabler argued that textual criticism of Ulysses cannot proceed directly from the facsimile of the Rosenbach Manuscript, despite the high quality of its photography, because "the clear differences in colour of ink, and slope and pressure of hand and pen" in the original are less clear in the facsimile. For example, there are "distinct sections in black or in blue or blueish ink in the original manuscript which no longer stand out as such in the reproduction.... Nor can pencil entries in the original in every instance be safely told from inked ones" (p. 178).

Although the reviewer felt these details important in 1977, his own edition contains no record of ink colors, or even a notice of where entries are made in pencil. The transcription of the manuscript was not made in Philadelphia, but from the facsimile, and that transcription was proofread against the facsimile, not the original. This can be demonstrated with recourse to the two volumes of the manuscript reproduction and the annotated and reduced 1922 Ulysses in the third volume.

A participant at the 1985 Joyce conference in Philadelphia where the manuscript was displayed, Coilin Owens, pointed out to me an example of the failure of the 1984 transcript to register what was visible in the original. In Philadelphia I was able to confirm his impression. In Telemachus leaf 14, lines 23 and 28, is the repeated word, in much larger hand than the surrounding text, "honey". Despite the exaggerated letters of "honey", in both cases a gap before the next word shows an erasure. Although not evident in the facsimile, the original in each instance strongly suggests "marmalade" as the erased longish word. But "marmalade", which initially took up so much space in the line, yielded first to "jam" (the "j" descenders are still evident in the reproduction), also erased before "honey" was written in.

Working from the facsimile in Europe, the transcribers simply recorded an erasure, not knowing what was beneath. The transcript was not checked against the original, so at 1.334 and 1.337 in the synopsis of revisions stand symbols for illegible erasures. Clive Driver's annotation of the legible erasures is recorded in volume three (a four-pages-to-one reduction of the 1922 Ulysses), at the left margin next to 1922 pages 12.11 and 12.14. Driver singles out the remnants of "marmalade" and "jam" on page 3 of his introduction to volume 3, which makes Gabler's silence about them doubly puzzling.

Falling between the "marmalade" erasures is a another change, of "Our Lord" to "O Lord" by erasing the "ur" which is detectable in the original leaf 14, though faint in facsimile. Again, the synopsis gives only the symbol for an illegible erasure. The first line of leaf 14 also has an erasure which, however, is not recoverable in the original. Preceding Mulligan's exclamation "Janey Mack, I'm choked!" at 1.323 (11.35) should be an erasure symbol in the synopsis, but there is none. The revisions to leaf 14 carry over to leaf 20 where "spread jam on" (or "over") is erased to become "let honey trickle over". None of the erased words were decipherable in the facsimile, so at 1.476 (16.01) two erasure symbols bracketing "ead" are all that the transcribers could come up with. Although absolute certainty about these erasures is not possible, a few weeks in Philadelphia would have yielded some worthwhile conjectures as well as a few certainties where the Synoptic Edition now provides not even a guess.

A cardinal rule of critical editing -- Check all transcriptions against originals -- was not observed. That a reviewer could complain that the facsimile did not adequately differentiate blueish from blackish ink, and subsequently publish an edition that never mentions inks and does not distinguish ink from pencilled entries, is a revealing contrast between principles and practice. Failure to collate the transcription of the Rosenbach Manuscript against the original is, frankly, incredible. As the practice of the edition makes evident, all documents available in facsimile -- drafts, typescripts, and proofs -- were similarly transcribed with insufficient recourse to originals. Where the printer went to work on manuscripts with pen or blue pencil, the 1984 transcriptions from facsimile confuse authorial and scribal inscriptions which in the originals are as distinct as black and white. Before moving to those later documents transcribed out of the Joyce Archive, we should spend more time here with the synopsis errors that result from use of the Rosenbach facsimile, an excellent publication in itself, but no substitute for the original.

(Two afternoons at the Rosenbach Museum in 1986 turned up many errors in the Synoptic Edition which occupy considerable space below. These examples are only meant to point out the nature and range of error. A comprehensive errata list would of course require some weeks in Philadelphia.)

Wandering Rocks: Joyce's or Budgen's Hand?

On the final leaf of Wandering Rocks in R, Joyce wrote that "PP. 32-48 were written by my friend Francis Budgen at my dictation from notes during my illness January-February 1919". The 1984 synopsis designates Budgen's hand "S" and additions or corrections to Budgen in Joyce's hand as "A" (not to be confused with level A in other episodes where post-Rosenbach changes are made by Joyce in letters or notes to his typist). Within passages bracketed with S or A levels, the synopsis attempts a distinction between what Budgen first wrote from dictation, and his own "overlay" (Gabler's undefined catch-all term).

The Rosenbach facsimile is not suitable, however, for detecting with confidence differences of pen and pencil, and the synopsis fails to capture the "intradocument revisions" (again, Gabler's phrase) it undertakes to report. For brevity, I will comment only on mistranscriptions in three leaves of the last segment of Wandering Rocks in Budgen's hand, although all the leaves 32-48 carry layers of pen and pencil from dictation. In these seventeen leaves are Joyce's ink additions to Budgen's pen and pencil, and the possibility of some pencil markings being Joyce's can only be tested by further study of the original in Philadelphia.

The base text in Budgen's scribal section is in ink, but throughout are pencil additions and corrections, some of which simply repair or reinforce the wording and punctuation already present. For example, in line 16 of Wandering Rocks leaf 44, the large pencil comma after "Ormond hotel" is not an addition, as it appears in facsimile, but an enlargement of an underlying ink comma visible in the original. The same is true of the apostrophe in "subsheriff's" in line 21. But the commas in line 21 "office," and 19 "Dedalus," do appear first and only in pencil, being genuine revisions within the document. The apostrophe in "lord's" of line 26 also originates as a pencil revision of Budgen's draft from dictation; only in the original is it clear that this apostrophe is not a speck or paper flaw.

Now for the synoptic treatment of these pencil changes in leaf 44 (all falling on 22U 242): the commas are not reported as revisions (though two of the four are), but as base text of the same level as the ink; the apostrophe in "lord's" is not recorded at all. The phrase "lords deputies", as it seems to read in facsimile, should have set off an alarm. There is of course the British plural form best known in "Lords Justices." Yet other possibilities suggest themselves: "lord's deputies" or "lords' deputies" or "lords, deputies" or "lords deputy". The ambiguity of "lords" versus "lord's", which may never be penetrated, abides in the archaic phrasing plucked from the mind of Hugh C. Love, M. A.: "mindful of lords deputies whose hands benignant had held of yore rich advowsons." (10.1203/242.30)

The unrecorded apostrophe in "lord's" has a parallel in the unrecorded pencil comma punctuating the ink of "Deep in Leinster Street, by Trinity's" in the margin of Wandering Rocks leaf 47. For 10.1263 (244.16), the footnote assigns the comma to the typist and rejects the punctuation of all editions. On the contrary, the typist was being true to copy by including it. Leaf 47 also has a pencil underline for the last word of "his blub lips agrin" (10.1273/244.26). As mysterious as the underline is, it ought to have been reported so that others might try to make out its significance.

The synopsis for leaf 45 of Wandering Rocks registers the "overlay" (Gabler's term) of a comma after "french primer". The phrase and its comma are in continuous ink (10.1228/243.19). Having set out to record (or guess at) late changes in punctuation, the facsimile transcribers did not detect the pencilled addition of commas after "gaily" and "gravely" in line 14, after "guests" in line 16, after "Menton" in line 26, and after "Buildings" in line 27. The five pencil additions are reported as if part of the base text in ink. That "Kavanagh's" of line 2 gained its "h" as a pencilled afterthought is unmentioned (these examples begin 10.1211/243.04).

Most serious in the synopsis for Wandering Rocks leaf 45 is the omission of the marginal "Fowne's" in pencil, an alternate spelling for "Fownes's" street. The synopsis, footnotes, textual notes, and Afterword make no mention of "Fowne's", a spelling found on many Dublin maps from 1900 to the present, including the map for Wandering Rocks provided in Gifford's Notes for Joyce, page 210. A compression of the possessive form of the English name Fownes has led to the short lane near Trinity College being spelled "Fownes's", "Fowne's" and "Fownes" in Irish publications. Either Joyce or Budgen decided to use the shorter form of the street name, but the typist did not follow the marginal correction.

If "Fowne's" is rejected, the other pencil changes on the leaf may be suspect as well, raising the question of the relative authority of ink and pencil text in Budgen's hand. The 1984 edition fails to distinguish media within the episode -- while silently, or accidentally, omitting some apostrophes, commas and marginalia. The Afterword has not even a paragraph on the composition history of Wandering Rocks.

Spurious Erasures

"Details of the autograph inscription -- deletions, erasures, insertions and illegible words or letters -- are recorded." This promise in the Afterword of the Synoptic Edition (p. 1901) incurs an extra transcription burden because deletions, erasures and blurred words are generally not decipherable from facsimiles, and erasures tend to disappear altogether. A morning spent holding original leaves up to a lamp in the Rosenbach Museum made it clear to me that the facsimile was Gabler's principal source for erasures, and an unreliable one.

In the facsimile of Scylla and Charybdis leaf 25, the word "evil" almost leaps out from line 8, so much darker is it than any other nearby word. The synopsis for Scylla 9.828 (198.30) reports this as the erasure of an illegible word and insertion of "evil" as a revision later than the original writing of the leaf. One might speculate on the earlier formulation in Stephen Dedalus' declaration that "A father ... is a necessary evil." One might wonder whether Joyce first wrote "burden" or "agent" or "foil". Shakespeare being under discussion in the National Library, Shakespearean pronouncements on fatherhood could be culled and sifted. Any guess at Joyce's earlier wording would be wasted because the revision to "evil" is a photographic illusion. The original paper is unerased and it is uncertain why the word is so much darker than its neighbors. And the erasure reported beneath "Yes" of Scylla and Charybdis 9.303 (184.34) is not to be found at R leaf 9, line 21, meaning that of only three erasures reported by Gabler for this episode two are spurious.

In Proteus Stephen reflects on fibs he told his schoolmates, that his uncle was a judge, another a general. Untruth is ugly: "Beauty is not there. Nor in the stagnant bay of Marsh's library where you read the fading prophecies of Joachim Abbas." (3.107/40.06). The synopsis for Proteus records an erasure beneath "Abbas". Joachim de Flora, or Joachim Flora, -- Floris, Flore, Fiore -- Joyce may have written. To search for a name beneath "Abbas" is futile because nothing has been erased there.

(Only four lines above in the synopsis, the wrong symbols are used to record a pencil deletion of "air" and insertion of "wind" above it. The angle brackets elsewhere denote only deletion and immediate replacement in the initial inscription, ink in this passage. Later "overlay" revision within a document requires raised carets missing here. This type of error is common in the synopsis but could be avoided if changes from ink to pencil were noted. See Table of Symbols in each volume's preliminaries.)

Calypso, the fourth and second shortest episode, has only seventeen leaves, but use of the Rosenbach facsimile created two imaginary revisions to muddle the genetic record of Ulysses. Under "see" at Calypso 4.133 (56.28/R leaf 4.39), the synopsis mistakes a reinforcement of the "s" for revision over an erased word. The illegible letter said to be deleted beneath "six" in Calypso 4.505 (66.27/R leaf 16.12) never existed, except as an illusion created by the facsimile.

Undetected Erasures

Nestor, the second and shortest episode, has an erasure in its first line, the irrecoverable early name of the schoolboy Cochrane whom Stephen Dedalus quizzes about Pyrrhus and his unhappy victory. This erasure is reported in the 1984 synopsis, but within Nestor's brief seventeen leaves, only two and a half percent of R, or one fortieth of Ulysses in manuscript, ten erasures are unrecorded. Obviously seen to have erasures when held up to a lamp, the worn paper seems flawless in facsimile. Table A records the overlooked Nestor erasures by leaf and line, with double underline indicating the point of paper wear, and unusually wide or narrow word spacing reproduced.

Some of these breakdowns in transcribing Nestor deserve comment. In addition to the erasure not visible in facsimile for R leaf 8 line 10, 84U fails to decipher the erasure it does record underneath "head". In the original, the "h" is not over an erasure and appears to have been part of the original word. Looking closer, "r" is discernible beneath the present "d", suggesting that the "h--r" beneath "head" was "hair". From "illdyed hair" to "illdyed head" may not be a surprising transformation for Mr Deasy, but it does draw attention to his head, perhaps to his headstrong way of thinking about things, his false attitudes, transparent hypocrisy. The small revisions of Ulysses, what

Table A. Nestor erasures overlooked in the synopsisTable B. Spurious erasures reported in the 1984 synopsis.

1984 1922 R leaf.line

2.48 25.21 2.40 Had Pyrrhus not fallen

by a beldam's hand

2.116 27.19 5.01 their cries echoed


2.197 29.26 8.10 his illdyed head.

2.237 30.26 9.33 what money is.

2.314 32.29 12.15 Again: A


2.333 33.10 13.08 Emperor's horses at


2.345 33.22 13.31 voice spoke .

2.384 34.27 15.07 Hooray! Ay! Whrrwhee!

2.415 35.21 16.08 Mr Field, M.P.

2.438 36.09 17.05 Do you know that?


1984 1922 R leaf.line


1.373 13.11 15.42 No erasure before Your


2.369 34.11 14.28 No erasure before to


3.109 40.07 4.42 No erasure before


3.298 45.07 11.33 No erasure before I


4.133 56.28 4.39 No erasure before see

Scylla and Charybdis

9.303 184.34 9.21 No erasure before Yes

9.828 198.30 25.08 No erasure before evil


12.1173 310.37 37.51 No erasure after he


15.829 435.12 20.19 No erasure before

with hangdog


16.381 579.02 9.53 No erasure before too

16.1214 601.06 33.15 No erasure before


they may reveal about the whole, are a principal justification of any genetic record. There is little point in publishing a partial record of erasures, a partial recovery of overwritten text, if the only hindrance to completeness is the time it would take to examine the original.

At least one of the overlooked Nestor erasures is obvious even in the facsimile. At leaf 12, line 15, it is clear in facsimile that the capital of "Again" is written over a minuscule "a". This was overlooked, as well as the erasure of a full inch to the left margin which has left a light smear in the blank, and has blurred the tail of the "y" in "boys" in the line above. Instead of a new paragraph beginning "Again: A goal" (the "A" is far too large to be "a"), there was one paragraph which may have been:

Shouts rang shrill from the

boys' playfield and a whirring whistle.

[??Shouts] again. A goal.

As it stands now in R, the final inscription is:

Shouts rang shrill from the

boys' playfield and a whirring whistle.

Again: A goal. (2.313/32.28)

All of the above examples of erroneous transcription of the Rosenbach Manuscript were gathered in two days at the Rosenbach Foundation in 1986. When the unreliability of the 1984 synopsis was established, it was necessary to check all the reported erasures against the original. This meant searching the entire synopsis. Erasures are reported in 139 places; these were checked against the original by Mr. Charles Phillips.

The eleven erasures reported in 1984 for which there is no trace in the Rosenbach Manuscript (table B) are revealing. Gabler reported erasures in all but the final chapter, and his spurious entries are spread through eight of those seventeen episodes. The initial survey of erasures was made from the facsimile, an understandable procedure. But it appears that the necessary confirmation against the original was neglected not only in parts of the manuscript, but throughout it. For Eumaeus both the reported erasures are spurious. For Scylla and Charybdis all but one are spurious. Instead of reporting 3, 2, and 0 erasures for Scylla, Eumaeus, and Penelope, Gabler should have reported 1, 0, and 0. Mr. Phillips also spotted forty erasures not reported in 1984 and confirmed the ten listed for Nestor above. Further work with the original will precede publication of any final account of the forty newly discovered points of Joyce's revisions.

More Transcription Problems: The Telemachiad

Now to return to Gabler's 1977 disappointment in The Library that the Rosenbach facsimile did not adequately show "clear differences in colour of ink." The concern for changes in ink color is valid, and it is regrettable that Gabler's interest when reviewing the facsimile was not sustained. The synopsis has a symbol for "space reserved in autograph" and when this space is later filled in, one might be alerted by a change in ink for brief stretches. The symbol is abused, however, and inserted in the synopsis where neither ink nor handwriting justify the conclusion that at one time there were textual lacunae to be later filled.

For Telemachus leaf 24 (1.576/18.25) "Muglins" is bracketed by carets, meaning revision at "first overlay level," and preceded by the symbol for a space left by Joyce with the intention of adding something. Yet the slight downward slant in the last word of the line "save for a sail tacking by the Muglins" is simply Joyce's compression of "Muglins" as he neared the paper's edge. Joyce was also dodging the descending "p" in "empty" overhanging the "l" of "Muglins". At the right hand margin of Telemachus are, among others, these downward slanting words (followed by leaf number): hyperborean (4), edge (5), always, pointing (6), shears, window (7), nothing (8), called, where (10), memories (11), virginium, heard (12), Mulligan (14), messmates (15), droned (16), wandering (17), again, seven (18), That's, Mulligan (19), under (22), Muglins (24), etc. The implication that Joyce left space for "Muglins" while he checked a geographic reference is unsupported by the manuscript itself. Yet the user of the 1984 synopsis will get the impression that Joyce was unsure of the name of the shoals at the southeast boundary of Dublin Bay. The misinformed user who assumes that the "Muglins" were tacked onto the scene may then overlook that Joyce included a phonetic anagram of "Mulligan", while describing Haines in quasi-homeric tones:

    Eyes, pale as the sea the wind had freshened, paler, firm and prudent. The seas' ruler he gazed over the bay, empty save for a sail tacking by the Muglins. (cf. 1.573/18.22)

The scene seems to be built around the Muglins with the intent of merging Haines, Mulligan and Muglins. To imply that "Muglins" is a revision is to mask the method. Incidentally, the scene is echoed a few pages later in Nestor as Stephen observes Mr Deasy: "The seas' ruler. His seacold eyes looked on the empty bay:" (2.246/30.35). The Telemachus preposition in "over the bay" has changed in Nestor to "on." Joyce may have been trying to remedy the inconsistency in an unpublished postcard to Claud Sykes not recorded in the 1984 edition, though it is in the Joyce archives of SUNY-Buffalo.

For Nestor 2.140 (28.07) the synopsis has "arm" in angled brackets for deletion, followed by "arms". By this awkward notation we are asked to believe that Joyce revised "in her arm" to get "Yet someone had loved him, borne him in her arms and in her heart." Nestor leaf 5 has no such verbal revision. Joyce's initial inscription of "arms" had such a small squiggle for an "s" that he touched up the word exactly as he repaired "loops" 19 lines above on the leaf. There both the "p" and "s" were written over more clearly, but the synopsis is silent. The "<arm> arms" notation is spurious.

Even to have attempted such distinctions without devoting the necessary time to collating the transcriptions against the original is imprudent. One can only conclude that if a printed record of the preeminent holograph document of literary modernism is worth having for study, the work must begin anew, from scratch. The 1984 offering is incomplete, erroneous, and distracting.

Paper Folds

Gabler's 1977 Library review, despite its brevity, has a series of cavils with the Rosenbach facsimile worth recapitulating in order to judge his edition. In 1977, Driver's record of paper folds is called "incomplete, confined as it is to Cyclops and Circe, and omitting reference to Calypso, and especially to Eumaeus" (p. 179). As if anticipating his own edition's claim to distinguish the levels of revision in Ulysses, Gabler concludes that "anyone hoping to unravel part of the manuscript's history by matching paper folds must work with the original leaves." Paper folds are never again mentioned in Gabler's writings on Ulysses.

Driver pointed out a vertical fold in all of Circe, fols. 1-50, and the leaves immediately following fol. 50 happen to occupy a page of the 1984 Afterword, the longest such attention to paper in the edition. Leaf 53 is there distinguished from its neighbors as "falling into neither bibliographic sequence," on the basis of "Page size and paper quality" (p. 1884). Its most remarkable feature passes unmentioned: Circe 53 is the only leaf of the episode creased into quarters. Having been overlooked by Driver, the folds of leaf 53 should be central to Gabler's argument that it "may have come either from a different sheaf of paper or from an earlier draft." A little further on, we read that "56a conforms bibliographically to fol. 54." How the one "conforms bibliographically" to the other is not stated there, but can be here: the vertical folds in 1-50 also crease 54 and 56a. Phrases like "bibliographic sequence" may sound grand to the lay person, but having criticized his predecessor's procedures, the new editor might have mentioned the folds in the very leaves singled out for elaborate "bibliographic" analysis in reconstructing Joyce's revision of Circe.

A Late Addition to Sirens?

Suggesting that the Rosenbach leaves "contribute, in strictly circumscribed areas accessible to bibliographic and palaeographic analysis, to a critically relevant recovery of stages of the composition," Gabler's Library review announces one such "recovery" that later shaped the synoptic display in the new edition:

    In this respect, the three initial leaves of Sirens must surely be a source of delight to any attentive reader, so clearly do they reveal the genesis of the chapter's double opening. Evidently the orchestral tuning-up by way of a segmented concatenation of the episode's linguistico-musical motifs was prefixed to the narrative opening proper as a 1 3/4-page manuscript addition. (181)

The clang and clash of this last sentence may have distracted readers from double-checking the "bibliographic and palaeographic analysis" being set forth. There is scant evidence that the prologue to Sirens ("Bronze by gold heard the hoofirons ...") is an addition to the secondary opening, "Bronze by gold, Miss Douce's head ..." The 1984 synopsis marks the first 63 lines of Sirens with raised carets for "revision (addition) within one document at overlay level," yet neither the Rosenbach Manuscript nor the synopsis itself uphold Gabler's thesis that the prologue was a late addition.

The prologue leaves 1 and 2 are foliated consecutively within the Sirens sequence of 1-44, and the same ink, hand, and margins are shared by all leaves. Joyce began both the prologue and the body of the episode at the top of a leaf, but that alone does not imply that the prologue was written out later than the rest of Sirens. On the contrary, Joyce intended there to be a visible expanse between the "Begin!" which concludes the prologue and before the second opening. These intentions are not at all clear from the Synoptic Edition because one of Joyce's most important instructions to his printer has been omitted. In the Joyce Archive page proofs for Sirens:

    Laissez ici un espace sans astérisque[.] Mieux serait commencer plus [deleted: pas then inserted] bas page 245 afin l'espace se trouve après le 1 ère our 2 ième ligne de page 247 (JA 24:172)

The printer chose not to follow Joyce's request to make the first lines of Sirens start lower down the page. As it fell out, the second "Bronze by gold" rings out not near the head of page 247 but at the foot of page 246. But these instructions suggest that when the second "Bronze by gold" heads leaf 3, it is not because these lines were copied out before leaves 1 and 2. Rather, Joyce wanted "Bronze by gold, Miss Douce's head by Miss Kennedy's head, over the crossblind" to be at the head of page 247. It is a shame not to have Joyce's instructions to his printer in the Synoptic Edition.

Gabler seems to have believed in the following sequence for the composition of Sirens: Joyce copied leaves 3-44 from his working draft, did not foliate the concluded piece, then drafted the prologue, added the new draft segment to the typist's source, wrote out a duplicate for his fair copy, and then numbered the leaves 1-44. The documents are more economically accounted for by assuming that Joyce copied the entire episode from a completed working draft, leaving a blank quarter page at the overture's end on leaf 2 to mark a shift to the re-opening at the top of Sirens 3, a space he also ordered in the message to Darantiere omitted from the Gabler apparatus.

One anomaly has escaped mention in the literature on the Rosenbach Manuscript: despite the similarity of ink, handwriting, margins, and continuous numbering within Sirens 1-44, leaf 2 is a thinner, darker, and waxier paper than any other in the episode. To judge it as earlier, later, or concurrent with the surrounding leaves requires a paper analysis of the entire Rosenbach Manuscript, a task begun in Driver's "Bibliographical Description of the Manuscript" but unadvanced since 1975.

Joyce to Typist Claud Sykes

Driver accounts for some touching-up in the first three episodes of R by citing Joyce's published correspondence with his American typist Claud Sykes. Gabler cites Letters I and II for nine readings, but Driver identifies three Joyce revisions sent to Sykes which are not mentioned in 1984, and a fourth is only partially recorded. Each of these four letters reveals something about the progress of the early episodes from the lost drafts.

For Nestor 2.349 (33.26) the synopsis records a deletion of "And now it has come." without citing Joyce's December 1917 card (Letters I, 109) instructing the deletion. Yet Driver found this an important clue to the stages of revision for the episode as a whole: "It is probable that the other heavily deleted lines were also marked by Sykes, for this manner of deletion occurs only in these episodes" (p. 19). Rather than simply scoring through or, as Joyce increasingly did, putting matter for deletion in parenthesis, this heavy blacking out occurs at leaves 13 and 29 of Telemachus, 2, 8, 12, 13, 16 and 17 of Nestor, and 9 of Proteus. No comparable blottings are to be found in the subsequent six hundred leaves, though at Wandering Rocks leaf 29 "Old Antichrist" has several cancel scores.

The 1984 synopsis treats all these expunged phrases and sentences as authorial deletion, silently disagreeing with Driver's conclusion that they are scribal. If the phrases were expunged by Sykes, they witness a far larger set of revisions sent to him by Joyce than were surmised in 1984.

The fact that entire sentences were cancelled and not inserted elsewhere also presses against the broad 1984 thesis of an "accretive" working method. These sentence-long blots in the Rosenbach Manuscript suggest that passages presumed by the 1984 editors to have been lost in typing may have been deleted by Joyce himself. The assumption that the disappearance of any sentence after the fair copy was given to the typist must be an error of transmission depends on an assessment of the sentences blotted on the fair copy. If the heavy deletions were made before sending the manuscript to Sykes, the shape of the episode may have been finalized and any omission accidental. But if Joyce made heavy deletions after the return of the manuscript and typescript, a sentence present in MS and absent in TS might require a different analysis. The issue of cancelled sentences vis-Ã -vis phrases absent in transmission but not cancelled in MS is problematic, and Gabler's silence on Driver's valuable guess is worrisome. These questions remain: Are the deletions by Joyce's hand or Sykes'? Were they made before or after typing?

The deletion of "And now it has come." (it would fall in Nestor 2.349) is reported as a revision in Joyce's hand. This assignment of the deletion to Joyce is crucial. In the 1984 synoptic scheme for this episode, raised carets mark Joyce's revisions in the Rosenbach Manuscript and level A is a revision sent by card or letter after the fair copy R was made. Level A changes carried out in Sykes' hand are labelled A-S. (All writing not in Joyce's hand in the extant documents is identified as S or s, meaning "scribal." That Sykes was the first such scribe is alphabetically coincidental.) The same undated postcard instructed Sykes to revise "wheywhite" to "wheysour", which he did by underlining the original word and pencilling the replacement in the margin of Nestor leaf 7. Another revision from the card of Letters I, 109, "Good man. Good man." is pencilled in the margin of Nestor 10, as "Good man, good man." (the only form of the phrase recorded in 1984) with a guideline to the supplanted "Quite right. Quite right." Both pencilled changes are labelled level A-S. But at Nestor 13, the third directive of the postcard is fulfilled when "And now it has come" is thoroughly blotted in ink. The synopsis records the deletion as Joyce's without the expected footnote to Letters I, 109. Of three changes requested in the postcard, two are carried out in pencil and one in ink. If Sykes was instructed to delete the sentence yet left no mark at the superseded passage, leaving the change to be caught by Joyce, we can posit a category of deletions which, though eventually carried out, were not marked by Sykes on the manuscript. Silent unmarked deletion suggests that other phrases, too, could be deleted on the typescript without corresponding marks in R. Possibly, then, Joyce did not later follow up by inking out every intended deletion. The 1984 synopsis calls the Nestor 13 deletion Joyce's. If Joyce himself made the heavy ink deletion he had asked of Sykes, there must be a category of revisions fully executed in the three copies of the typing but unmarked by Sykes in R. And had Joyce forgotten to blacken the deleted phrase, an editor might have erroneously "restored" the unwanted "And now it has come." of Nestor leaf 13.

The implications for a reading text of the three episodes typed by Sykes are tremendous. Revisions, such as R's "land" to transmitted "country" on the first page of the work, as well as deletions, suddenly seem authorial. Variation between the fair copy text and that transmitted by the lost typescript can be seen in an entirely different light, affecting dozens of substantives in the first three episodes. Instead of full annotation of the passages Joyce altered through the mails, in 1984 is found an incomplete record of letters and no discussion of Driver's important guesses about revision in R. A text of the first three episodes cannot be established without a hypothesis on the relations of manuscript, correspondence, and revision on the lost typescripts.

For an addition at Proteus 3.249 (43.32), "Shattered glass and toppling masonry.", Sykes' incorrect pencilling of "topling" is emended with reference to the lost typescript (tB) rather than the authorial instruction, not mentioned, in Letters II, 416. The extant working draft also reads "toppling". We have the phrase twice in Joyce's hand (card and draft) but neither corrective to Sykes' slip is brought into the synopsis.

For the revision of "wheywhite milk" to "wheysour milk", in Nestor 2.166 (28.33), a footnote cites Letters II, 415, without mentioning that there "whey sour" is in two words. The one word form is, however, in an earlier card to Sykes: "in description of the boy Sargent for `wheywhite' read `wheysour'" (Letters I, 109). In order to show authority for the compound form of Sykes' pencilled addition, the card with the one word "wheysour" ought to be noted.

The fourth correspondence which only Driver quotes has bibliographical implications, even if textually it remains neutral. In a card of 27 November 1917, Joyce urges Sykes to borrow Rudolph Goldschmidt's typewriter to commence Telemachus "at once". In closing, Joyce writes "For `plunged' in the description of Buck Mulligan searching his trunk for a handkerchief read `plunged and rummaged'" (Letters I, 109). Yet Rosenbach leaf 21 has only the correct "plunged and rummaged" at line 31. The page is cleaner than almost any other of the episode, lacking the erasures, underlinings, marginalia, strike-outs and pencil additions found on the majority of the episode leaves. Leaf 21 may have been recopied.

A Sampling of Transcription Errors

One of the advantages to Gabler of not making up a list of differences between his helpers' transcriptions of the Rosenbach facsimile and Clive Driver's collation of the original is that other scholars would find it hard to take sides without spending months collating. The disadvantage to Gabler is that he had no set of points to check repeatedly against the original to test whether his differences with Driver were real or manufactured by the facsimile. Within three lines of the Eumaeus episode Gabler mistranscribes the Rosenbach Manuscript three times. Two of these suggest unfamiliarity with Driver's collation. All three transcription errors would also have come to light if Gabler had prepared for publication a transcript of the pre-Rosenbach draft of Eumaeus, Buffalo notebook V.A.21.

The 1984 transcription of Bloom skimming a newspaper:

    Ascot meeting, the Gold cup. Victory of outsider Throwaway recalls Derby of '92 when Capt. Marshall's dark horse Sir Hugo captured the blue ribband at long odds. New York disaster. Thousand lives lost. (16.1242-44/601.35-37/R leaf 34)

Driver's collation from the original alerts us that two periods transcribed by Gabler are not to be found. Neither the period after "Gold cup" (which Gabler emends with a capital) nor the one after "disaster" are in the Rosenbach Manuscript -- which is no surprise, given that Bloom's eye is darting around the newspaper page. A look at the earlier draft shows that the periods are omitted there as well:

    Ascot Meeting. The Gold Cup [insert l. marg w/ symbol F above ambig punc:] victory of outsider Throwaway [deleted: at long odds] recalls the Derby of '92 when Capt. Marshall's dark horse sir Hugo captured the blue ribband at long odds [punc after Cup as:]. New York disaster [no punc, extra space] 1000 [poss 1100] lives Lost. (JA 15:357)

With this transcription at hand, anyone working from the Rosenbach facsimile would be prepared for its lack of periods between "cup" and "Victory" and after "disaster". Certainly they are unneeded in context. Driver got the punctuation right, but without the evidence of the early draft, he made the same error as Gabler in transcribing the Rosenbach's "sir Hugo" as "Sir Hugo". Transcription of the moderate ambiguity of the fair copy could be aided by checking its predecessor, which emphatically reads "sir Hugo". So all three of Gabler's errors would have been avoided by recourse to the early draft, and two of them by comparison with Driver.

Gabler emended the lower case "cup" of the Rosenbach and all editions, a needless change given that the phrase "Gold cup" occurs throughout the book. Instead of using the early draft as a grab bag for emendations, Gabler could have more profitably collated it against the Rosenbach and published a transcript of the draft. My own transcription of the draft from the Joyce Archive is only tentative, pending study of the original. The period after "Capt" may be an illusion created by reproduction.

The Imaginary Level R

It should surprise no one that when an author makes a fair copy to be set aside for sale or presentation he may enter last minute changes in the clean copy which never enter the main stream of transmission into print. An editor noticing variants unique to the fair copy will have to decide whether they are the final readings intended, or incidental lapses. Every act of copying a long work will introduce changes, not all of which can be called revisions, especially when they are not transferred into the final drafts leading to publication. It is said "the critical edition reclaims some one hundred substantive readings from the fair copy" (p. 1881). That is an astonishing number of substantives, given that Joyce himself was responsible for altering his final working draft W as he was making a clean copy to set aside. In the synopsis and clear text of 1700 pages, it is no easy task to isolate and judge the 100 words which are drawn from the fair copy to replace the version passed on by Joyce to his typist.

Having found all the R level changes, I consider the majority unjustified emendations of the published text, which even the 1984 editors admit is Joyce's own at each of these points. Almost none of these collateral Rosenbach readings are demonstrable final revisions. The argument that the collateral R readings are both later and superior to what Joyce gave his typists never overcomes two obstacles: Joyce himself wrote and conveyed the words as published; the finality of one wording or another is largely undecidable for almost every phrase purportedly reclaimed by the edition. To reclaim a given word, another must be removed from the published text; caution should reign.

Users of this edition will have to spend many days searching the synopsis to gain a sense of these 100 changes of wording. The new phrases could easily have been identified in a long paragraph or appendix. Failing this, the historical collation could have painlessly accounted for every instance of a 1922 reading displaced by a level R revision selected in 1984. By design, at each point in the historical collation where 1984 disagrees with 1922, the documentary source of each rejected 1922 reading is given after the symbol \. For example, if a scribal hand on the third proof alters what Joyce wrote, the rejected 1922 reading is given, followed by the editions where it occurs, and concluding with \s3 to mean "rejected reading originated as scribal change on proof level 3." Where level R is labelled by the 1984 editors, they are arguing that Joyce revised himself in the Rosenbach fair copy but failed to do so in W. The draft was in turn typed and went into print after many phases of further revision. In none of the historical collation entries where Joyce's (lost) final working draft is overruled is the symbol \(aW) supplied. In some cases an extant early draft aP confirms the later but lost (aW). Never does the collation give aP as the point of origin for the rejected 1922 reading. To use \(aW) and \aP would betray the authorial nature of the rejected readings in a list presumably demonstrating how "5000 departures" have been "corrected." Rather than lay the cards on the table, a misleading \tC (which implies typist's change) stands at every point where \(aW), \aP -- or both -- should be. This not only misstates the authorial origin: it conceals one hundred level R alterations of the historical Ulysses within thousands of other typed readings. By Gabler's own view, none of these one hundred 1922 wordings are typing errors -- all are Joyce's own words. These one hundred "level R" phrasings are among several hundred emendations of what Joyce wrote that are disguised as "corrections" marked by \tC in volume three when the typed readings are also found in the Rosenbach or early drafts.

There are a few defensible changes of the known Ulysses to be drawn from the collateral variants of the Rosenbach MS, and they are worth examining, if only for contrast with the majority of the hundred alterations which were better left unmade. As Joyce was copying out the Rosenbach MS from the lost working draft, he occasionally made a deletion and replacement which is not paralleled in the typescript. Take for example Bloom's comment on how politicians need to press the flesh, have an iron gut and an innate Wanderlust: "Want to gas about our lovely land" (8.463/156.16). The typescript agrees with the base inscription in R, which was later altered by a thin line through "Want" and an ambiguous "Or" over "to" (R leaf 11). There are not one hundred as Gabler claims, but only about 20 detectable late revisions in R. Such cases suggest three possibilities: Joyce failed to enter the revision back into the working draft from which he was copying; he made the change in the draft in such a way that it was overlooked by the typist; or he thought better of the change and reversed himself. Joyce could be expected to retain the original reading in his draft while not restoring it in the fair copy precisely in order to keep it "fair." There are, after all, hundreds of final draft revisions witnessed by the typescript for which there is no hint in the fair copy. Where Joyce reversed himself, Rosenbach may show no trace of his change of heart.

One finds most authors reverting at points to the earlier of two readings in their working papers. In folder 1, leaf 1 of manuscripts for Leaves of Grass (1860) in the Barrett Collection at Virginia, Whitman deleted "Victory" from line 13, interlined "Vista" with a caret, and later in lighter ink deleted "Vista" and rewrote "Victory". He also might put a row of dots under a word to restore it after his own cancellation. (See Fredson Bowers, editor, Whitman's Manuscripts, 1954).

Confronted by an apparent revision in the Rosenbach Manuscript of Ulysses that Joyce may have neglected to make in his transmitted draft, the editor must still decide whether the typescript may represent the finally intended form. When there is physical evidence of revision in R, and the editor elects to follow the last reading of R, rarely would anyone dispute the choice. Of the difficult choices to be made, such are the least vulnerable to second guessing.

Still, the same leaf of Lestrygonians which suggests that Joyce revised "Want to gas about the country" to "Or gas about the country" contains in the preceding paragraph a far more troublesome phrase. Bloom compares James Stephens (the revolutionary, not Joyce's poet friend) with Garibaldi the liberator and unifier of Italy. Or at least he comes to mind: "Irish Garibaldi." The typescript from the lost working draft omits "Irish", leaving a one-word sentence, "Garibaldi" (8.461/156.14). Here there is nothing in the manuscript to guide us. Which is the earlier reading, which the later? Or is the absence of "Irish" a typing error? Did Joyce add "Irish" in his draft in some way the typist overlooked? Did he intentionally delete it? Did Joyce add "Irish" in the fair copy and forget to enter the addition just as he forgot to put "Or" in the next paragraph? One must fall back on the copytext reading. The "continuous manuscript text" is "Irish Garibaldi", R being the only manuscript witness. But Gabler does not follow R, declaring the absence of "Irish" a level B revision in the lost draft. There is no textual note to justify the choice, only a blithe synoptic deletion. Nor is this deletion listed anywhere as an emendation of copytext. Because the editors decided that Joyce deleted the word in the lost draft (although we have no evidence it ever was in the draft) the phrase without "Irish" instantly becomes the copytext or continuous manuscript text. Had the editors decided "Irish" was an addition, level R, Joyce forgot to enter into his final draft or was a plain typing error, tC, then any objection to adding "Irish" to the reading text of Ulysses would be met with the retort that, being in "the author's own hand," it is part of the "continuous copy-text" and therefore all editions which lack "Irish" are in error. Much to our surprise in this instance the text as published in 1922 stands unaltered, a potential restoration foregone.

The second leaf of Sirens has in line five "Preacher is he." after "Namine Damini.", although "Preacher is he" is absent from the typescript which also has "Naminedamine" (11.43/246.16). The ink is uniform throughout the leaf, and the spacing is no more irregular in line five than others. The synopsis labels "Preacher is he" an R-level addition, but "Naminedamine" a B-level revision away from R. There is no physical evidence that "Preacher is he" is an addition in R; being absent from the typescript it is either Joyce's deletion or his or the typist's oversight. Level R should not be invoked because there is no way to recover whether or not Joyce forgot to add the phrase to his lost draft, or added it to W which was mistyped, or had written it earlier in aP as well as W which was however mistyped, or deleted it from W after he had finished with R. However one analyzes the difference between R and the typescript, there is no evidence of the Rosenbach Manuscript having a late addition, the position Gabler takes. The question must be narrowed to whether Joyce wanted the phrase in the book. Level R is assigned because Gabler felt Joyce wanted it in his book; he could have called the omission a tC error and plausibly retained the phrase. If Joyce was still making additions as he wrote out R, why is the R spelling of Namine Damini in the same line of the same document rejected in favor of Naminedamine? What evidence is there that within line five of leaf two, Joyce spontaneously added three words as he made his fair copy, later combining the first two words into a compound ending in "e" and not "i", yet neglecting to notice that three words were missing? One could as easily argue that R is superior for the entire line, not just the "addition." The fusion of "Namine Damini" into "Naminedamine" may be part of the broader revision of the overture to the Sirens -- a revision so dependent on the musicality of each line that there was no place for "Preacher is he" when all was sung and done. The 1984 choice to assign part of a line to level R and part to level B is unconvincing and complicated beyond need. The deletion could be Joyce's, it could be the typist's. After the undemonstrated synoptic assertion that Joyce deleted "Irish" from "Irish Garibaldi", justifying "Preacher is he" with the popular claim that Joyce wrote by accretion, not deletion, rings untrue.

The danger of confusing the fair copy with the final draft is evident in a daring 1984 replacement. Bloom's recollection of an errand is changed from "Must see him about Keyes's par" (22U 250.07) to "Must see him for that par" (11.187). The synopsis for that line records a revision at level R which deletes "about Keyes's" and substitutes "for that". Given no evidence of revision in Sirens leaf 7, we cannot assume that R's "for that par." is a substitution for "about Keyes's par." On the contrary, it is equally probable that the typed phrase is a revision postdating R. Surely no one can imagine that the reference to Alexander Keyes has any origin but Joyce. The full paragraph of Bloom's musings as published:

    By Cantwell's offices roved Greaseabloom, by Ceppi's virgins, bright of their oils. Nannetti's father hawked those things about, wheedling at doors as I. Religion pays. Must see him about Keyes's par. Eat first. I want. Not yet. At four, she said. Time ever passing. Clockhands turning. On. Where eat? The Clarence, Dolphin. On. For Raoul. Eat. If I net five guineas with those ads. The violet silk petticoats. Not yet. The sweets of sin. (22U 250)

In the seven lines of this paragraph about salesmanship and the puff that Keyes wants before renewing his ad, there are six differences of wording between R and the typescript. Gabler judges only one of the typed readings not the final draft version. Here are the Rosenbach and typescript differences:

    walked roving Bloom,] roved Greaseabloom,
    ABSENT ] their
    wheedling.] wheedling at doors.
    for that] about Keyes's
    make] net
    ABSENT ] The violet silk petticoats.

In addition to the six typed differences from R, the typescript carries five revisions in Joyce's hand for this page as well as his attentive correction of misstrikes and dittography (JA 13:61). Joyce made one revision on first proof ("wheedling at doors." to "wheedling at doors as I."), and there were no errors of transmission to the 1922 book or beyond. The paragraph as we have known it, then, is fully authorized. The typescript apparently is an accurate transcript of the final working draft accepted by an author studiously attentive to the passage. The 1984 change of "about Keyes's" to accommodate "for that" (found only in the superseded, collateral R) is unwarranted. Among the six phrasings separating R and the typescript, "for that" is no more final or authoritative than the other five. There is no hesitation, cancellation, false start, or any other sign that Joyce was "revising" himself when he wrote out the Rosenbach version of the paragraph. I can see no reason to call "for that par" a late revision any more than calling it a cast off early version. When he might have made his wishes known, none of the six R variants were transferred into the typescript as Joyce made late revisions in the printer's copy.

In the Sirens paragraph under discussion (11.185 on), variation between the Rosenbach MS and the typescript is represented as both revision at level (B) credited to lost draft, and revision at level R as Joyce copied it out. In a footnote four lines below, a third position on simple Rosenbach versus typescript variation is offered. "O welcome back Miss Douce" is the typescript and 1922 reading, but the "O" is set off with a comma in the fair copy and 1984 (11.194/250.14). The variant is not considered a level B, R or V revision: the footnote gives aR for the lemma and (aW):tC to explain the rejected form. This adds a fifth level to the four already considered. Above it was observed that any difference between the Rosenbach and the typescript (in episodes typed from W) can be assigned level R, V, B, or tC. The first three are authorial revisions, the fourth a departure by the typist. In the synoptic style, if R, V, B, and tC do not supply the nuance desired, a composite (aW):tC is invoked. This symbol states that in the lost final working draft (aW) Joyce wrote something accurately typed in tC -- the editors may reject it nonetheless. It is remarkable that comparison of a page of fair copy and a page of typescript of the same passage can support five different "levels."

Scrolling back to the first page of Sirens, we find another deceptively simple revision in the "continuous manuscript text." In the sixth line of Sirens, R reads "Blew. Blue Bloom is on the", ending without punctuation before the short line turns, as in verse, to "Goldpinnacled hair". Certainly there is nothing un-Joycean about "Blue Bloom is on the / Goldpinnacled hair" (11.06/245.06). Yet a second interpretation of the line ending in "the" is available: Joyce may be loosing an intentionally unfinished phrase upon the winds which blow the darling buds of May. In any case, Gabler rejects the Rosenbach MS, the typescript, and all editions by inserting an unwonted period. Also in this line is a second of those "accidentals" that can mean so much in Joyce. In the typed lost final draft, "Bloom" is lowercase "bloom". The Gabler edition accepts the typescript over the fair copy to agree with all editions, and wisely so. This may be Joyce impersonalizing his hero for a moment. Just as Bloom took the pen name Henry Flower, here he becomes "the bloom upon the rye". Under the rules of the 1984 synopsis, this would be a straightforward level B revision made in the lost draft some time after the fair copy yet before the draft was typed. The synopsis, however, reports otherwise. "Bloom" is assigned level V as if Joyce inserted Leopold into the line as an afterthought in writing out R. Because the editors reject this "later" reading, it is not an R-level revision, but V-level. What is the copytext reading, the uppercase or lowercase form? Is it Mr Bloom or the blooming rust upon the rye? The manuscript has the proper noun, the typescript the common. Which is the continuous manuscript reading? Whichever was judged to be last in order of composition would be the continuous manuscript text. Since V comes after B in the sequence of levels, V is the later of the continuous revisions; of course, we really do not know whether "Bloom" followed "bloom" or viceversa -- levels B and V are conjectures, not documentary readings impartially recorded. In the terms of copytext editing, which is the copytext and which the emendation? Is there an emendation here? The edition occasionally refers to a "List of Emendations" (e.g. pp. xii, xiii); there is no list in these three volumes which includes "Bloom" or "bloom" for the sixth line of Sirens.

To depart from Sirens, which alone could yield another fifty examples of incautious conflation, we can turn to the second page of the Scylla and Charybdis synopsis. In substance, all editions, including 1984, read "Seven is dear to the mystic mind. The shining seven WB calls them" (9.27/177.01). (Gabler alone removes the abbreviation points to agree with R against the typed final draft. This would be unexceptional except elsewhere he supplies abbreviation points not in R ; if there is a policy on pointing, it is unstated; still, he does here follow the extant MS, if not necessarily the lost one.) The fair copy continues after "WB" with "says" over which is written "in" in the same ink, inconsequentially darker, creating "sings" followed by "of." The change to "sings" seems immediate because "of" would not normally follow "says" at a sentence's end. According to the sequence of levels in the synopsis, "calls them." was the lost draft reading which stood unchanged in the typing, "says" came later to Joyce as he copied out the passage, though he then settled upon "sings of." This order is improbable. It is proposed that "The shining seven WB calls them." became "The shining seven WB says" -- either concluding here or not, we cannot know -- which gave way to "The shining seven WB sings of." The last reading sounds wrong from the mouth of John Eglinton, although it would not be out of place in Oxen of the Sun or some other parodic passage. Where it stands, the sibillant alliteration is inferior to Eglinton's plain-spoken "The shining seven WB calls them." This last was probably the final and rightly published reading.

These few examples drawn from three episodes suggest that levels R, V, B and tC do not themselves chart Joyce's revisions, serving rather to uphold editorial decisions where the documents themselves are indifferent or even testify against the editors. Those who differ with the present analysis of these passages might nonetheless agree that the Gabler apparatus can easily shroud improbable guesses in an obscuring cloak of "levels."

The book An Inquiry into Ulysses: The Corrected Text continues with the topic || Shifting Levels ||