Among the extraordinary collection of Joyce manuscripts recently acquired by the National Library of Ireland, the two drafts of the « Sirens » episode stand out as the most fascinating. Not only do they reveal significant elements for the interpretation of that particular episode, but they afford us an unprecedented view of the main turning point in the history of Ulysses: the transition between Joyce’s early manner and his mature mode,1 what is probably the decisive step towards the formal complexity that made his work the paradigm of modernism.
My purpose, in the present article, is to give a summary account of these drafts, based on the notes that I was able to take before they left Paris, and to open the discussion about their significance, a discussion that will probably go on for a long time in the future. It is to be hoped that the owners of the documents and of the copyright will soon allow an edition to be published or, even better, a fac‑simile to be produced and widely disseminated. Then every one will be able to check the facts that are presented here, supplement them with other observations and reach conclusions that will be better grounded, if only because they will be based on more detailed and more leisurely observations.
Since the first ‘Sirens’ draft is included in the same copybook as a draft of « Proteus » (a puzzling and suggestive concatenation), I will also summarily describe this very remarkable draft (it is the earliest extant Ulysses draft), but my findings about it will be even more tentative. I will also say a few words about a page of Homeric notes at the end of the copybook.

The coverless copybook

In the National Library’s provisional inventory, it is described as a « Notebook, no covers, 22 x 17.8 cm, graph paper, 28 pages; pp 1‑9 Proteus, pp 10‑27 Sirens, p 28 notes from Homer. » We should add that the « Proteus » pages are unnumbered while the « Sirens » pages are numbered from 1 to 18.
This copybook is probably the most interesting single document among all the extent Ulysses manuscripts. It is quite unlike anything in the published text or elsewhere in the archive. When I first saw it, I thought for a few moments that it could be very early indeed, a series of epiphanies followed by an unpublished story for Dubliners, later to be recycled into, respectively, « Proteus » and « Sirens. » A second look is enough to see that this cannot be the case, for various reasons of style and plot, but this first impression is characteristic of the uncanny feeling produced by this manuscript. We recognize the words: the vast majority of them have made their way into the final text, undergoing relatively little change in the process. We recognize whole passages. But the general outlook is so completely different that we hesitate to recognize the episodes.

Ur‑Proteus: the broken stream

The visual bafflement is particularly great in the case of the « Proteus » draft. Instead of the dense and continuous text we are used to, which conveys the (fallacious) impression of a uninterrupted stream of consciousness spanning the episode, we have here a series of fragments, 15 or 16 sections2 of different lengths, carefully separated by rows of asterisks. Every section can be easily identified with a passage from the final text of « Proteus »:


Reading the equivalent passages of the final text in the above sequence gives a very good idea of what this draft is like, and one gets an even better approximation if one takes the trouble of reading the corresponding passages in the Buffalo draft (V.A.3, reproduced in JJA 12) which was considered the earliest extant version until the recent discovery. Many additions and modifications have been made, but the chief thematic material is the same, as well as the narrative and stylistic options: the blend of first person monologue and third person narration, with some transitional sentences that may or may not be free indirect speech, is already effective, and we see it gaining in fluidity with the revisions.3
The fragmentary nature of these short segments is all the more surprising and raises an important question. Is it simply the accidental appearance of a composition in progress, a succession of passages that Joyce wrote in this copybook as they came to his mind, or gathered there before integrating them in a future continuous text? Or is it a stylistic device reflecting a deliberate aesthetic choice, a mode of presentation that was later discarded in favour of a different option?
It seemed safe to assert that Joyce had renounced the fragmentary form (dear to other figures of Modernity, before and after him) relatively early: when he stopped considering that his epiphanies were in themselves essential works of art that should be disseminated in the main libraries of the world and started using them as building material for more continuous productions. After that, the fragment stopped being an aesthetic object for him and even what might look superficially fragmentary is quite the opposite. (See the impeccable chronology and the tight reflexivity of the diary segments of the Portrait or the careful arrangement of the sections of « Wandering Rocks » as the components of a complex spatiotemporal labyrinthine structure...). Giacomo Joyce4, however, stands as a counterexample (a temporary regression?) and this draft seems to be another, for material as well as aesthetic evidence is in favour of the second hypothesis.
Materially, this version does not look at all like a set of notes or even a very rough draft. The main body of the text is neatly written, although it is subjected to some currente calamo corrections and in some places to several layers of marginal redrafting. One would presume that it is at least partially based on previous writing, either a still earlier complete version, or a series of separate drafts, perhaps on loose sheets, of the various sections. It does not seem to be the same kind of document as the fragmentary first drafts of « Cyclops » and « Circe » in the Buffalo collection (V.A.8 and V.A.19). The difference is particularly striking if one compares it to the end of the ‘Sirens’ draft in the same notebook, where many snippets of writing, sketches of future passages, are crowded haphazardly on the pages in what was evidently a burst of invention.5
Aesthetically, the episode makes perfect sense in this fragmented form. The poetical prose of Stephen’s musings is, in a way, even better set off than in the final version. The underlying narrative is at least as easy to follow, starting as it does with Stephen’s decision not to go back to the tower, and alternating observations of what he sees around him on the strand, reminiscences, fantasies and speculations. Now, saying that the fragmentation is purposeful is not to say that the episode is not truncated: it is certainly odd that it would end with the waking of Paris (U 3.209‑215). It is possible that Joyce changed his mind as he was going, got dissatisfied with this version, abandoned it before it was finished and started another, presumably continuous, one. Why this dissatisfaction? Perhaps because the openly arbitrary segmentation gave the episode too much of an arty, fin‑de‑siècle flavour. Probably also because of the general trend, noted earlier, towards continuous forms (a superficial formal continuity undermined by an increasing underlying enunciative, stylistic and — eventually in Finnegans Wake — linguistic discontinuity). Whatever the reasons may have been, the fragmented form was discarded and the form that we know emerged on the Buffalo draft.
It would be important to determine whether the draft under consideration is the immediate antecedent of the Buffalo draft. A careful collation of the two manuscripts would be necessary to provide a definitive answer. At first sight, several parts are sufficiently similar to have been copied from one to the other and the reshuffling of the passages might have been effected directly, probably with the help of an outline on a loose sheet, but how can we account for all the missing parts? For instance the « Ineluctable modality of the visible » passage, the opening of the episode as we know it, is entirely absent from the early draft, and its layout in the Buffalo manuscript seems to indicate that it was faircopied from another source, not invented as it was being written... So we must either infer an intermediary complete version between the two drafts or suppose that Joyce used some partial drafts of the missing sections in combination with the NLI draft to produce the Buffalo draft.6 These partial drafts could very well be part of the same (pre‑existing) series from which the NLI draft was presumably copied from, or they could have been prepared on purpose in order to fill the gaps in the new outline. My favourite hypothesis would be a combination of the two (but close analysis of the extant documents and, who knows, new discoveries to come, may go in another direction).
As a matter of fact, it is quite possible that we already have access to one of these intermediary partial drafts — or at least to a trace of its existence. It would correspond to our section 4 (equivalent to U 3.106‑127), devoted to Stephen reminiscing about his reading of the prophecies of Joachim Abbas7 and to his speculations about the multiplicity of the Eucharist as envisaged by Occam. The original section 4 of the NLI draft only covers U 3.106‑119, the rest is an addition, squeezed in the margin in two different stages. The marginal addition is hurried, and ends in the form of a note rather than a draft.8 What is missing, if we compare it to the Buffalo draft (and the final text), is the Eucharist bell ringing simultaneously in the course of several concurrent services. Now, traces of a draft, or series of notes, for precisely that passage can be found in the Buffalo notebook VI.C.16.
The text, as reconstructed by Danis Rose and John O’Hanlon from Madame Raphael’s incomplete copy, began with the words SD in Marsh’s library’ (this would not be the beginning of the draft, but a kind of heading) and ended with the words « hears again bells, »9 but we unfortunately cannot tell what came in between. This is enough, however, to suggest that, in what is known as « The Lost Notebook », Joyce drafted a version of this passage10 that is anterior to the Buffalo version but posterior to the NLI version — and more specifically posterior to the marginal addition, that is itself obviously posterior to the main body of the NLI draft. Of course, we do not know how much time elapsed between all those stages, but if the Buffalo draft was written in the Fall of 1917, this would suggest Summer 1917 as the latest probable date for the main body of the NLI « Proteus ».

Sirens without Bloom

Unfortunately, this does not help to date the « Sirens » draft that immediately follows in the coverless notebook. The only thing that we can tell is that it was necessarily written after the « Proteus. »11 The « Proteus » ends on the bottom of the ninth page and the « Sirens » starts on the top of the tenth without anything marking an ending or a beginning, but the writing is clearly different. One can guess that Joyce decided to use the empty pages of this notebook after the « Proteus » was finished or abandoned — but we don’t know when he did it. The interruption could have been relatively long and we cannot infer that « Sirens » was written immediately after « Proteus » or that it was meant to follow it in the outline of Ulysses — but there is no proof of the contrary.
Now if we look at the draft itself, we do not experience anything like the visual shock produced by the fragmented « Proteus » draft. The beginning looks indeed like the beginning of the « Sirens » as we know it — except, of course, that there is no « overture », but we could not have expected to find this in an early incomplete draft, since it is a motival compendium of the episode and could only have been compiled a posteriori. We find all the familiar elements: bronze and gold are immediately there and so is the viceregal cavalcade, the tall silk and such expressions as « Mind till I see »(U 11.73). The memorable phrases « Bronze by gold » and « ringing steel » appear as interlinear corrections.12 The only striking (but nevertheless minor) departure is the colour of the dress of Lady Dudley: it is pink instead of green, the eau de nil, so important in establishing the aquatic atmosphere of the chapter as we know it, is replaced by (or rather, eventually replaces) crushed strawberries.13 The manipulation of language is much less obtrusive than it is in the final version, but this is exactly what we might have expected, since we already know that Joyce increased this aspect at each stage of the development of the episode, from draft to fair copy, from fair copy to typescript and at each of the proof stages.
In the next pages, the episode similarly proceeds along familiar lines with dialogues almost identical to those we know: the rude boots and Miss Douce’s haughty answer, Miss Douce’s dermatological problems, the giggling conversation about the awful old wretch in Boyd’s, Simon Dedalus and his flirtatious cordiality, Lenehan’s pathetic efforts, the tuning fork, Simon’s half‑hearted singing at the piano, Boylan’s entry, the Sonnez la cloche performance, Ben Dollard and Father Cowley’s conversation, Dollard’s attempt at Love and War... But by the time we get to page 10 of the draft, we cannot fail to notice that something capital is missing from the chapter as we know it: Bloom is entirely absent! No trace of his approach to the Ormond or his entering the dining room. He is not mentioned by the other characters or the narrator and, more importantly, he is absent as a point of view. Up to that point in the draft, the episode is entirely confined to the dialogue in the bar, with connecting narrative passages in the third person. There is no monologue at all (or any other form of representation of stream of consciousness, and very little subjectivity) and we find only faint forebodings of the counterpoint that we have come to associate with « Sirens » (snatches of song come from Simon in the music room during Lenehan’s interview with Boylan, the solitary twinkling of a diner’s bell in the restaurant accompany Miss Kennedy’s compassion for the blind piano tuner). We are much closer to the style of Dubliners than to the « initial style » of Ulysses, let alone the style of the middle stage. From this point of view, the fact that the « Sirens » should follow the « Proteus » in the coverless notebook is notable information; if it was not for that, we could have sworn that it belonged to an earlier period. This apparent regression is very puzzling. After the dense and claustrophobic subjectivity of the Telemachiad, did Joyce want to move to a more impersonal, dialogue‑oriented presentation? After the lyric, the dramatic mode?
In the present state of our documentation, it is impossible to recover precisely Joyce’s project at the time he wrote the beginning of this draft, but what is particularly fascinating is that we can follow the details of the process that led him to change his mind as he was writing and to move to the contrapuntal style of ‘Sirens’ and more generally to the style of the middle stage of Ulysses.
The turning point is on page 10 of the draft, but it can probably be situated even more precisely. On the occasion of Simon, Dollard and Cowley’s nostalgic evocation of the famous concert at which Dollard gave « Love and War » with Goodwin at the piano, there is no mention of Bloom’s saving the situation and Molly’s used‑clothes business. Cowley inquires about another singer present at this concert, called Marie Fallon, with the name « Fallon » immediately struck out and replaced currente calamo with « Powell ». Simon answers that she is alive and married, but an ellipsis indicates that his sentence is interrupted by Dollard’s singing and the reader is never told whom she married. This kind of thinly disguised aposiopesis becomes significant if we remember that Major Powell was, according to Joyce himself, the original of Major Tweedy.14 This would be enough to suggest that Marie Powell is an early name for Marion Tweedy‑Bloom, or a figure that shares some of her characteristics, and this is further corroborated by a later modification which strikes out the end of Simon’s sentence and replaces it with the phrase we know from the final text: ‘My Irish Molly O’ (U 11.512) and a still later marginal addition indicating that Marie/Molly actually comes from the rock of Gibraltar (see U 11.514).
After that, the dialogue proceeds for a few lines and then stops abruptly in the middle of the page. Obviously, an interruption, of indeterminate duration, occurred. After a blank of several lines (later occupied by additions), writing resumes, in a darker ink and a slightly more cramped hand, with the name of Mr Bloom. The paragraph is a mixture of third person narration, interior monologue and free indirect style conveying Bloom’s point of view on the barmaid.15What has happened? It seems possible that the irruption of Marie/Molly‑Fanlon/Powell/Tweedy/Bloom in the dialogue brought with her not only the idea of her husband but also the necessity of varying the point of view and the technique used so far. If Molly is born of a rumour (Joyce’s original plan to write a story about Hunter who was reputed to be a cuckold) she is too sensitive material to be abandoned to malicious gossip alone. Introducing her husband’s voice seems like a good way to counterbalance the vox populi. Hence perhaps the idea of a new departure in the development of Ulysses,16 the introduction of more than one point of view in the episode, in the form of counterpointed voices (in a loose sense of the phrase).
It is possible that at first Joyce had in mind something like the present structure of « Nausicaa »: a simple cinematographic shot‑reverse shot, with the second part of the chapter given entirely to Bloom. But such a structure, not being combined with the elaborate parody of the first part of « Nausicaa, » would have been too simple for Joyce’s purpose here, so it was deferred for later use. (The thematic resemblance is striking enough to suggest that this paragraph is indeed the origin of the « Nausicaa » episode: Bloom exchanges knowing glances with the young woman, wonders about her virginity, is fascinated by her quasi‑masturbatory gesture on the beerpull17 culminating in an orgasm of gushing beer).18
Almost immediately, we see Joyce preparing for the orchestration of something much more complex. A multitude of short segments follow one another feverishly, later expanded in the margins, sometimes leaving blanks that are later filled in with a different writing instrument. They are passages to be used like bricks to build the second, as yet unwritten part of the chapter, but also to be inserted in the first, already written part. Most of these segments are crossed out with red or blue crayon.
Alternating with Bloom’s snatches of monologue, there is more nasty gossip, in particular about the Blooms (some of it is not crossed out and never made it to the next draft). This time we have the impression that we have the seed of the « Cyclops ». We even find the interjection « begob » so characteristic of the nameless debt collector who reflects half of that episode. But without the contrasting mock‑epic passages, such an exercise in meanness would also have been too simple, a kind of regression towards Dubliners. Something else had to be invented.19
At this stage, we still find no trace of the early passages introducing Bloom in the chapter, his walk along the Quay to the Ormond or his visit to the stationery shop. Joyce is clearly embarking into a polyphonic organisation, but the details of the arrangement are far from settled. Fortunately, the other « Sirens » draft acquired by the NLI documents the next step. Before we discuss this draft, however, we should say a word of the last elements to be found in the coverless copybook: a page of Homeric notes and a loose leaf related to the draft « Sirens ».

Homeric notes

The last page of the copybook is occupied by a set of Homeric notes. It is difficult to relate these notes to the neighbouring drafts in a definite way. The fact that they are written on an external page (the equivalent of a back cover for this coverless copybook) makes it almost impossible to determine their time of inscription. We can make the following guesses: a) the direction of writing of this page relatively to the two drafts does not make it very likely (but not impossible) that it was inscribed first, before the notebook was turned over for a fresh start; b) the page cannot have been filled after the completion of the ‘Sirens’ draft, since the expansive pressure of the draft on the last pages is such that it would have welcomed an extra blank page if one had been available, and it did require, as we will see, the usage of at least two supplementary loose leaves; c) the notes could follow the « Proteus » draft and precede the « Sirens »; or d) they could follow the first spell of writing that took the « Sirens » draft to page 19 of the notebook (numbered ‘10)’), before the radical change that characterizes the second part of the draft.
The contents of these notes will have to be scrutinized and compared to Joyce’s other sets of Homeric notes. The page is headed « Lacedemon » and the first third is devoted to a kind of genealogical schematisation of the two unions of Menelaus, with a slave and with Helen. A series of correspondences is enumerated, for instance Menelaus is equated with Uncle Hubert, O’Rourke,20 Captain O’Shea, Joe Casey, J.H. Parnell, S.D.’s brother, Ulysses (some of these people are eclipsed by their brothers, others are cuckolds). Helen is equated with Kitty O’Shea, Aunt Josephine, Molly Bloom (this proves at least that her role was already defined when these notes were taken)... Below this, a list, numbered and then partially renumbered, of the events of Telemachus’s visit to Lacedemon, with a boxed subdivision relating mostly to the interrogation of Proteus. Several interrogation marks seem to indicate uncertainty of usage and interpretation. This is clearly the case for Helen’s pouring of Nepenthe, which is tagged with a parenthesis: « (Latin?) ».21 Almost half of the events in the list are crossed out in blue.
When Joyce took those notes, was he contemplating a « Lacedemon » episode, as a transition between « Nestor » and « Proteus »? We know that in 1915, he planned a Telemachiad with four episodes.22 At a later stage, Joyce seems to have mined these notes for the writing of ‘Nestor’ and the expansion of the ‘Proteus’ draft, but also for other episodes.

Supplementary page

Inserted in the coverless copybook is a loose leaf of similar dimensions but made of a different paper. Written on one side only, it is numbered « 20) » in the same way as the « Sirens » part of the copybook and it bears three « Sirens » related, disconnected passages, similar to those found in the last 8 pages of the draft and crossed out in the same way. The draft stops on page « 18) », so Joyce must have used (at least) two extra pages: this leaf and a missing leaf.23

The Fuga per Canonem copybook

Compared with the astonishing coverless copybook we have just described, the other « Sirens » related manuscript acquired by the NLI is extremely predictable.24 It is predictable in so far as it is the first part of an extant draft, Buffalo V.A.5 (JJA 13:32‑56). The pages of the new document are numbered in pencil from « 1) » to « 20) » while the Buffalo document goes from « 21) » to « 35) ». One stops in the middle of a sentence (U 11.785), between subject and verb and the other picks up exactly at the same place.
As was to be expected, the first « half »25 looks almost exactly like the second. There is one difference, however: in the NLI manuscript, the text is written on versos as well as rectos, while in the Buffalo manuscript it is written only on rectos, the versos being reserved for additions. It seems probable that Joyce, when he began a second notebook after having completely filled the first, decided on the new layout because he thought that half the 24 available pages would be amply sufficient.26 Another reason may have been that the second part of the episode was at a much less advanced stage than the first, so Joyce foresaw that he would need more free space for alterations.
What is really remarkable in this draft, however, is what is written on the inside of the front cover: a detailed list, in Italian, of the parts of a Fuga per Canonem. We cannot pretend that this is totally unexpected, since it is the technique assigned to « Sirens » in all the schemata provided by Joyce, but consideration of the finished text (or of the previously available manuscripts) did not appear to corroborate this identification. It seemed that the main characteristic of a fugue (the superposition of voices) could not be observed at the level of the general architecture of the chapter (that they are indeed impossible to achieve on a large scale in a work of literature). I must confess that I thought that the label was either a hoax or a very superficial identification.27 The presence of these notes on the threshold of the second draft and the juxtaposition with the earlier version proves that this was not the case. Joyce did strive to turn the episode into something equivalent to a Fuga per Canonem. We must understand that the superposition is diegetic, not literal, it occurs in represented time, not in the time of performance (or narration, or reading), for the mind’s eye, not for the eye/ear of the reader/listener. The new manuscripts also prove that this fugal idea is not consubstantial to the episode but a second thought, something induced by the need to integrate the different voices and perspectives that Joyce, half way through the first draft, had decided to include in it.
The consequences of this relatively late choice are enormous. It inaugurates the radical congruence between style and subject matter that is so much in evidence in the published schemata. This chapter devoted to the seductions of music assumes a musical form with a vengeance: not a vague Paterian allusive nostalgia, but the intricate structure of one of the most forbidding musical forms. Applied to barroom singing of facile operatic music, humming of music‑hall tunes, not to mention smacking garters, twanging elastics and resounding farts, this imposing model cannot but acquire a parodic dimension. We can see this as the starting point of the parodic strain that characterizes the style of the central chapters of Ulysses.
The combination of the two « Sirens » drafts gives us the impression of observing at close range a crucial turning point in the history of Ulysses — one could say in the history of literature. We see how the change of the name « Marie Fallon » into « Marie Powell » seems to trigger a chain of consequences that, after planting the seeds for ‘Nausicaa’ and ‘Cyclops’, completely change the face of the episode and of the novel.
We have the impression that we can follow the most important links in that chain. Are we actually missing anything? Is the draft in the « coverless copybook » the immediate antecedent of the draft now split between the NLI and Buffalo, or should we postulate an intermediary draft?
A simplified stemma shows how the new documents (in bold and underlined) are positioned in the development of the episode :


Could Joyce have gone from the early, fugueless draft to the fugal NLI+Buffalo draft without an intermediary stage? This seems plausible for the first 9 pages of the early draft and their transposition into the first 14 pages of the « Fuga per Canonem » copybook: there is not that much of a difference and it would not have been too difficult to insert the new passages, sketched at the end of the notebook and on the loose leaves, into the main body of the already drafted text. After that, however, there is no real first draft in the « coverless copybook, » only a multitude of hastily sketched short passages. A glance at the Buffalo V.A.5 copybook in the Archive is enough to be certain that it is not a first draft: it is clearly copied from a previous source before being enriched by numerous additions. That source is lacking, so we must imagine either a lost second draft, a partial draft covering the last two thirds of the chapter, or at least a series of partial drafts on loose leaves integrating the most difficult passages.
What is definitely lacking is any drafting of the overture. It is not yet present in the Fuga per Canonem copybook. This only confirms prior deductions that had determined that the overture must have appeared at the stage of the lost fair copy.28 There seems to be a starting point, however, in the form of a note in the margin of the first page: « repeat / phrases / episode » (or « episodes »). This can be interpreted as a self‑prescription: Joyce is reminding himself that he should write an opening made of repeated phrases from the episode. The location of the note seems to support this interpretation. But such explicit instructions are extremely rare in Joyce’s manuscripts, so it could also be a complement to the definition of the Fuga per Canonem, written on the adjacent cover, which does not mention « episodes » (episodes in a fugue are transitional, freer passages between stricter canonic sections). We can even imagine that Joyce wrote this in one way and later interpreted it in a different way, and that this was the origin of the overture, so incongruous in the context of a canonic fugue... This is pure speculation, but the recently discovered manuscripts have surpassed our wildest speculations. Who knows what will come up next?

1 This corresponds to the transition between what Joyce himself called the « initial style » of Ulysses and the middle stage posited by Michael Groden in his fundamental « Ulysses » in Progress (Princeton: Princeton University Press, 1977). One can consider on the one hand that the « initial style » is an extension of Joyce’s pre‑Ulyssean experimentations, and on the other hand that the late style of Ulysses (and, arguably, the whole of Finnegans Wake) is a radicalisation of the critical turn effected at the middle stage.

2  The hesitation about the exact number is due to the fact that sections 6 and 7 are not separated by asterisks, but material and textual evidence suggests that they should still be considered as two different sections. They deal with completely different subjects and will eventually end up in separate parts of the final version of the episode. The absence of asterisks can be explained by the fact that section 6 ends at the bottom of a page and the bottom margin is entirely filled: in this carefully laid out manuscript the asterisks would have looked odd at the top of the next page.

3  For instance a sentence that introduces too obviously Stephen’s perceptions is struck out.

4 The Ur‑Proteus is perhaps closer to Giacomo Joyce than to any other Joycean text. The Paris waking of the last section of the draft is a barely modified version of the Trieste waking in Giacomo. This is even more obvious in the draft than in the final version (or in other extant manuscripts) from which some elements have disappeared (the adjective « testudoform, » the reference to bugs).

5  On the other hand, it does not look either like the early notebook acquired at the same time by the NLI, which constitutes a kind of anthology of passages neatly copied out of previously written works, without any signs of redrafting. The « Proteus » text is still open, in a state of flux.

6  It is remarkable that marginal additions are not crossed out separately, as is usually the case in Ulysses or Finnegans Wake drafts (and as it is the case in the « Sirens » draft further in the copybook). Several separate additions, destined to be integrated in different places, are crossed out with one single crayon stroke. This may indicate an a posteriori check after a reshuffling of elements from different documents rather than a systematic process of transposition from one document to another.

7  No allusion to Swift is yet perceptible in the NLI draft.

8 It is interesting to remark that, in this marginal addition as in the final text (U 3.124), Occam is called « invincible doctor » while those words are not present in the Buffalo draft. This should be a warning against too strict an application of the principles of textual filiation. We should not infer from this that the NLI draft is posterior to the Buffalo draft! The epithet was certainly associated closely enough with the name of the medieval philosopher for it to resurface spontaneously after it was discarded or misled at an earlier stage.

9  James Joyce, The Lost Notebook, edited by Danis Rose and John O’Hanlon, Edinburgh, Split Pea Press, 1989, p. [28].

10  Or took notes for this passage that were used for a later draft, or were conflated with the NLI draft to produce the Buffalo version of the passage.

11 The fact that Autumn 1917 is the terminus ad quem for the « Proteus » draft, does not of course imply that it is the terminus a quo for the draft that follows it. On the other hand, we can be certain that the « Sirens » draft is later than the « Early Ulysses Notebook » also acquired by the NLI, since two elements from this notebook, crossed out with red crayon, are used in the « Sirens » draft: the description of Simon picking chips from his thumbnail (U 11.192‑3) and the recipe prescribing borax and cherry laurel water with glycerine for sunburns (U 11.116‑22). Michael Groden has brought to my attention Rodney Owen’s book James Joyce and the Beginnings of « Ulysses » (Ann Arbor, UMI Research Press, 1983) and his hypothesis according to which the echoes of the « Alphabetical Notebook » and of Giacomo Joyce in the first third of the « Sirens » suggest that this part was written before Joyce’s departure from Trieste in July 1915, while these documents were still available to him. The supposed echo from Giacomo Joyce (Miss Douce implicitly compared to a fowl on U 11.141‑2) is not yet present in this draft, but the two echoes from the « Alphabetical Notebook » (Simon’s weak buccinator muscle on U 11.512 and Mooney’s sur mer on 11.264) are indeed there. This cannot be decisive, for it seems very possible that Joyce could have remembered those or (as Owen himself remarks) that he could have transferred selected entries from the ‘Alphabetical Notebook’ into a more portable document. But Owen’s prescience about the existence and nature of the newly revealed documents has proved so accurate (see Michael Groden’s article in this number of the JJQ), that we should treat his uncanny deductive powers with the greatest respect.

12 Another interlinear correction ascribes Miss Douce’s laughter to her wet lips. The process of autonomization of the parts of the body (see André Topia « ‘Sirens’ »: The Emblematic Vibration » and Derek Attridge « Syntax and the Subject in ‘Sirens’ in M. Beja, Ph. Herring, M. Harmon et D. Norris eds., James Joyce: The Centennial Symposium, Illinois University Press, 1986) is starting very early.

13 In the Linati Schema, the colour assigned to the episode was still « coral ». Eau‑de‑Nil is inserted on the following draft, next to the crushed strawberries.

14  “Major Powell—in my book Major Tweedy, Mrs Bloom's father?’ Letters I 198 [Letter to Mrs William Murray, 21 December 1922].

15 He is looking more specifically at her eyes. The considerable importance of eyes, mirrors and gazes, so striking in a chapter dedicated to music (see D. Ferrer, « Echo or Narcissus », James Joyce: The Centennial Symposium, op. cit., p. 73‑4) is already noticeable in the first draft. It is also remarkable that the celebrated ending of the episodes comes about as the result of a vacant gaze: Bloom is looking away from the frowsy whore and pretends to be absorbed in the contemplation of a shop window (not yet named as Lionel Mark’s). It is only in an interlinear addition, expanded in the margin, that Emmet’s last words come to occupy that window.

16 We do not know if « Scylla and Charybdis » was already written or what it looked like at this stage.

17  All the later details are already there: the polished knob, the firm baton emerging out of the sliding ring of the gently touching fingers (U 11.1111‑1117)

18  More openly than in the final chapter where the gushing overflow is blended with the musical description (U 11.705‑709).

19 Michael Groden remarks: « we might conjecture that Joyce began with the opinionated, bigoted narrator and then gradually developed the series of parodies as a complement to, and a check on, the authority of the naturalistic description. But the opposite was the case: Joyce created the parodies first, the barroom scene came soon after, and the narrative voice developed last. » (« Ulysses » in Progress, op. cit., p. 124). At the level of the genesis of the whole book, it seems that a version of the barroom scene was invented first (the « Sirens » episode in its primitive form). Within that frame some elements of the Unknown narrator’s voice were tried out, long before the defining characteristic of « Cyclops » (the contrasting heroic parodies) was imagined.

20  Obviously O’Rourke, prince of Breffni, not Larry O’Rourke.

21  This resurfaces in « Lotus Eaters » (U 5.350).

22  Letter to Stanislaus Joyce of 16 June 1915 (SL 209).

23 The fact that leaf « 20) » is written on one side only might suggest that it is the last of the series: Joyce would have filled both sides of the leaf numbered « 19) » and one more side of this one to exhaust his material. But this would imply that the filled verso of « 19) » remained unnumbered, which would be inconsistent with Joyce’s practise in the notebook. It seems more likely that Joyce used the versos only of his loose sheets in order to make their manipulation easier. In that case, there is nothing to stop us from supposing that the extant leaf « 20) » is not the last of the series and that several others have been lost.

24 It is described in the provisional inventory as a « Notebook, blue cover, blank white label, 22.2 x 17.4 cm, lightly ruled graph paper, 10 leaves, all pages written, numbered 120. »

25  The correct fraction is slightly over sixty per cent: 20/33 in terms of manuscript pages, 785/1295 in terms of lines of the final text in the Gabler edition.

26   It turned out that Joyce had miscalculated: for the last three pages of the chapter, he was forced to retrace his steps and to write them on the empty versos.

27 And wrote it (see « Echo or Narcissus »,op. cit.).

28 See Daniel Ferrer, « ‘Practise Preaching »: variantes pragmatiques et prédication suspendue dans un manuscrit des ‘Sirènes’ », Writing its own wrunes for ever, Essais de génétique joycenne/ Essays in Joycean Genetics, D. Ferrer and C. Jacquet eds, Tusson, Éditions du Lérot, 1998, p. 34.