“For polemypolity’s sake”: Editing Finnegans Wake — A Consideration and Review
— “And it is surely a lesser ignorance to write a word with every consonant too few than to add all too many” (FW2010 091.16-18)
Wim Van Mierlo, Institute of English Studies, University of London
Ohio Dominican University
The Houyhnhnm Press edition
of Finnegans Wake, edited by Danis
Rose and John O’Hanlon, was published with some — but not very much — fanfare
in March 2010. A couple of notices
appeared in the national press.
It’s a shame, I think, for the
edition certainly deserves attention. With a design commissioned and printed by
the Stamperia Valdonega in
The edition, however, has already been greeted with skepticism—and to a large degree unfairly so.
Danis Rose’s reputation of course precedes him. Rose certainly has not endeared himself with particular claims he made about his ability to make Joyce more readable when his Reader’s Edition came out in 1997. As a result, Rose’s scholarly achievements, which include his arrangement of all manuscripts relating to Finnegans Wake in the James Joyce Archive, are unfortunately not always appreciated.
It is no surprise, then, that the few
responses we have had to the Rose/O’Hanlon edition broadly follow the same
tenet. No academic reviews have yet
appeared. The handful of reviewers that
wrote reviews for some of the better daily newspapers were obviously baffled by
what they had in front of them, and mainly repeated what they had read in the
press release. Two exceptions to this
were Michael Wood in the LRB, who
produced a lengthy, interesting and on the whole positive piece but who barely
gives any detail about the edition itself, and Terence Killeen, who gave the
edition a brief, but somewhat damning review in The Irish Times.
So more than the reaction to Rose’s reputation, it is the apparent skepticism about the possibility of editing Finnegans Wake tout court that stands in the way of an open reception of the Houyhnhnm edition. I find the idea that the Wake cannot—even should not—be edited quite problematic.
Two recent articles come to mind that made this point, one by Sam Slote, the other by Robert-Jan Henkes and Erik Bindervoet, both of which were published in Genetic Joyce Studies. They are not exactly the spearhead of a campaign against editing Joyce’s final work. Nor do I wish to single out their authors for blame. Each article is reasoned and measured in setting out the potential difficulties a critical editor of Finnegans Wake work might encounter, and the respective authors use their knowledge of Joyce’s working methods well to back up their arguments. I focus on them, however, because their views illustrate a deeper issue.
After pondering the difficulties of establishing rigid criteria by which an editor might remove errors from the text in “Sound-Bite against the Restoration”, Slote opined that only a “rigid conservatism” was the right option: not to edit at all. His reasons were twofold: “not all departures can be unequivocally gauged as being purely accidental” and not all errors can be “unquestionably determine[d]” to be “ruinous to the sense of a passage”. in “Finnegans Wake, the Corrected Text”, Robert-Jan Henkes and Erik Bindervoet put forward a similar case. To them, restoring words and phrases that have dropped out of the text is in particular a questionable undertaking, because “from their moment of demise” these words no longer “form part of the texture and the process of fermentation”. And thus reinstating them would in some way be like introducing an alien element that disturbs the creative universe. Concurring with Slote, therefore, they maintained that editing Finnegans Wake “is not necessary, not possible and not desirable”.
Admirable as their arguments are, however, the reasoning behind them reveals a critical blind spot about the methodologies of textual editing as a scholarly and critical practice, and so each of the articles in their final assessment misses an important point. Slote’s closing imperative comes out against editing on the grounds that emendations change the “logopoeic effect” of the passage. Henkes and Bindervoet agree. Emending the text would disturb its intricate pattern: “The Wake is too closely knit”. While these concerns may be broadly speaking true, they are also contradictory. What is that pattern if it is already disturbed by textual deterioration? What is that logopoeic effect if it is based on a faulty text? If emendation dislodges the linguistic and aesthetic integrity of the text, surely textual corruption has already done the same. The issue is not with the logic of the arguments, but with the appropriateness of the methodology: they raise objections to editing Finnegans Wake on literary-critical and readerly grounds, not as it should be on bibliographical and textual grounds.
In fairness, the position of these scholars is not simply dismissive of textual scholarship. After all, Henkes and Bindervoet edited Finnegans Wake, after a fashion, when they incorporated textual emendations and restorations in their Dutch translation from 2002. Aware of the contradiction, they felt compelled to do it because they wanted “to read each and every word Joyce [had[ added to his book”. Helpfully, also for the non-Dutch reader, they provided in an appendix a complete list of the transmissional departures they had identified. Likewise, Slote defends the recording of errors in a variorum: even if they must be “barred from the text”, they may be allowed in the paratext. I accept this solution; it’s elegant and informative, but it’s also the lowest common denominator. For a brief period in the 1990s, there were voices that called for everyone to be their own editor. This postmodern view was, however, short-lived, and rightly so. Editors not only have a responsibility to edit and to edit responsibly (a view propounded by Peter Shillingsburg and many other textual scholars), the a need for critical editions is real, lest readers get lost in the plethora of variant readings.
From the perspective of textual scholarship, there is nothing illogical or impossible about editing Finnegans Wake — which is not to say that the work doesn’t pose any problems, but more about that later. For the moment I want to dwell on the issue of error. It is of course true that error is an important part of the aesthetic of Joyce’s final book. “Missed Understandings” [FW2010 139.02] may be what drives the work, and actuates the pleasure we derive from it, but that does not foreclose the potentiality, or indeed the right, to produce a critical edition. Likewise, much has been made of the words that Joyce puts in Stephen’s mouth in “Scylla and Charibdis”, where the young man holds forth that “a man of genius makes no mistakes. His errors are volitional and are the portals of discovery” (U 9.228-29). However, to consider textual corruption equal to a programme of serendipitous enrichment by error is simplifying the matter, and overlooks the nature and impact of textual corruption. By this logic, the reading of U 2.199 should retain the erroneous “Mother dying come home father” from the Shakespeare & Co. text, for that too might be a portal of discovery. The fact of the matter is that the type of error Stephen has in mind is something completely different from textual error. For errors to be volitional, they must involve an act of authorial agency. (Incidentally, few critics seem to notice that what Stephen alludes to are the errors of Shakespeare the man — his forced marriage to the older, dominating Ann Hathaway, who allegedly subsequently cuckolded him — not Shakespeare the writer, a view of the relationship between life and art reminiscent of W.B. Yeats.) Textual errors result from agencies other than the author. They are not the author’s faults at all, and they are therefore certainly not volitional.
It is this second type of error—and this alone—that textual editors consider when preparing an edition. No matter what editorial rationale one adopts, emendations are always based on incontrovertible evidence derived from archival documents; where the archive is not clear or incomplete, conjectural emendations are necessary. Conjecture, however, is not an arbitrary decision; it is at worst an educated guess, at best a careful decision based on a critical examination of the available facts. Any conscientious editor would present a full explanation and argumentation in the Textual Notes to justify his choices. In the end, the value of a critical edition depends on the justification for the editorial rationale, on whether the rationale has been applied correctly, and on whether the evidence has been interpreted correctly. It cannot be judged on whether one “reading” is better or more meaningful than another “reading”.
Using these principles I will assess the Rose/O’Hanlon edition when I come to it. But first there are few other general matters to clear up about editing Finnegans Wake.
An apparent difficulty with the Wake is that it seems impossible to know where error occurs. In plain, “wideawake” English, no one has difficulty seeing that in the sentence “Father Conmee stropped three little schoolboys” (1922, p. 211) “stropped” is an obvious error for “stopped”. But who is to say that “everglaning” (FW 221.19-20) is wrong and “everglading” (FW2010 174.40) is not? Not even the full phrase “everglaning mangrovemazes” would settle the matter, for undoubtedly with some ingenuity one can construct a plausible interpretation. One possible association, apart from the obvious evergreen, is to glean. This might suggest something like mangroves/mazes that gather up and scrape together whatever is left behind, a form of ingestion that history performs, an act suggesting the inverse of Kate the Slop’s scavenging. Plausibility, however, is not a criterion in textual scholarship, for mostly one can make some sense of something that might actually be nonsensical, and therefore meaning — or logopoeics — in and of itself can not be a determining factor in scholarly editing.
Furthermore not all errors are obvious errors, not even in the nineteenth-century novel. In 1941 the eminent critic F.O. Matthiessen remarked on the aptness with which Herman Melville had chosen the phrase “soiled fish of the sea” in White Jacket, first published in 1850, to conjure up an almost existential sense of filth when the sailor falls into the ocean. “The discordia concors”, Matthiessen wrote, “the unexpected linking of the medium of cleanliness with filth, could only have sprung from an imagination that had apprehended the terrors of the deep, of the immaterial deep as well as the physical". Unhappily for the critic, an error had crept into Melville’s text in the Standard Edition of 1922: all editions before that, including the first edition, had correctly read “coiled fish of the sea” — a kind of sea eel.
The fact is that uncovering textual errors involves a rigorous process of collation by which all variant readings are recorded and compared against each other. As this is an evidence-based, information-gathering process, it doesn’t matter how normal or how difficult a text is. The principle is the same for “Auld Lang Syne” as it is for “Jabberwocky”, for Vanity Fair as it is for Finnegans Wake. The only text that is in fact impossible to edit critically is the text that exists in only one edition or exemplar.
Collating variants, it goes without saying, is not the same as emending errors. But it is an important and necessary first step towards making emendation possible. The process of correcting the text involves the use of the editor’s critical judgment, which — it needs saying again — is based on an interpretation of the textual evidence, not on an interpretation of the work in question. Different judgments, like different editorial rationales, lead to different editorial decisions and different editions. Such proliferation of editions is not a failure of the editorial process. The specter of the so-called definitive edition has long been laid to rest; among responsible editors the belief is now widespread that editions have a limited shelf life before new times and new research methods demand new editions.
This textual pluralism, moreover, is not something that is restricted to editorial methodologies; it is part (to adopt Jerome McGann’s term) of the “textual condition” to which any work of literature is subject. It surprises me in fact how little attention literary criticism pays to the materiality of the text—by which I mean the physical condition of the printed words on the page, not the semiotics of the text—and the instability that goes with it. The mantra of close reading demands— no doubt rightly — that we pay attention to what the text has to say. But we need to know first what that text is and where it comes from.
Not all texts are equal. Between author and printed book come a number of processes, carried out by individuals who do not always have a vested interest in the work they are producing, which have a considerable effect on the transmission. These processes of corruption, and the editor’s task of removing them, are generally not considered controversial when they pertain to earlier periods in literary history and to authors who did not see their own works through the press. For more recent works, however, and for strong-minded authors like James Joyce who exerted a high degree of authorial control, the problem of textual authority is refracted through a complex process oscillating between corruption and revision. While the author countermands errors introduced by typists and printers in his attempt to (re)gain control over his text, the elimination of error is reversed by the inevitable introduction of new errors. Even a meticulous person like T.S. Eliot was unable to create a stable version of his œuvre. As his own editor and publisher, he was in principle able to supervise every step of the production of what was in the end only a relatively small body of work. Yet with all conditions working in his favor he did not manage to bring his Collected Poems and Selected Poems completely in line with each other while they went through subsequent reprints. What happened was that the two collections were never reissued at the same time, so at one point he would send one instruction to revise or correct to the printer for the Selected Poems; at another point he would send another instruction for the Collected Poems. But at no point did he make a concerted effort to make these various revisions match up one edition against the other.
Textual stability may well be an illusion. Many other examples from English and World Literature, from all periods and in all genres, can illustrate this point. As a general rule, authors are unable to control the transmission of their texts, even when they try to. Others simply do not seem fazed by the prospect of textual pluralism, happily introducing variant readings and creating different versions whenever they see fit. Someone like W.B. Yeats, for example, revised not only to improve, but to give his poems and plays a new purpose and a new audience.
Despite Joyce’s obduracy that every word in his texts should be printed as he wrote it, he too was unable to control his handiwork. His temperament, on the one hand, prompted him to devise an elaborate system whereby the successive cycle of notetaking, drafting, revising, and expanding was intended to manage the rich materials of the work. That same temperament, however, also worked against him. His mind focused on creation, he easily overlooked the increasing number of errors that came with repeated copying necessary to satisfy his constant need for clean fair copies, typescripts and galleys.
While here and there Joyce does catch and correct errors, their number is disproportionate to the ones he misses. With a work like Finnegans Wake, whose chapters were often revised only at long intervals, it is not so outlandish to think that at times Joyce himself was tripped up by what he had written.
Editing Finnegans Wake — anyone will recognize — is no sinecure. The richness of the work and the longevity of its composition on the one hand, and on the other the relative complexity and certainly the size of its surviving archival record make the task something of a challenge. But these do not make it impossible. For one, editing Shakespeare or the New Testament is just as difficult, if not more so. For the New Testament about 4,000 early documents exist, some complete, some incomplete, which has resulted in colossal number of variants that scholars have not yet succeeded in collating all. With Shakespeare, as is well known, none of the texts we have for his works has come down to us in authorial witnesses. Critical editions of these texts are by necessity conjectural. Yet I know of no critic who propagates that they should not be critically edited at all. Editions by their very nature do not present “ideal” texts, but texts mediated through an editorial rationale. What makes editions valuable, however, even when they cannot be “definitive”, is the application of rigorous methodologies which safeguard editors from making arbitrary, subjective interventions. One of the essential functions of textual scholarship as a discipline is to invent, discuss, question, refine and update these rationales.
* * * *
So how would one edit Finnegans Wake? It is clear that traditional copy-text editing, as developed by W.W. Greg and modified by Fredson Bowers and G. Thomas Tanselle, cannot work. The principle that Greg recommended in “The Rationale of Copy-Text” (1950/51) is based on the selection of what is on bibliographical grounds the most reliable text. Depending on the circumstances, this text can be the first edition, if it was seen through the press under the author’s close supervision; the final manuscript as it was submitted to the printer, or any other text or edition that best represents the author’s final intention. This copy-text, then, guides the editor in his editorial choices wherever needed. In the case of “substantives”, the editor will use his knowledge and critical analysis of the textual evidence to select the correct readings from the variants. Where the evidence is indeterminate, which is most often case with “accidentals” such as punctuation, capitalization and so on, he will follow the copy-text, on the grounds that it has already been established that this text is generally more consistent with the author’s wishes.
For Finnegans Wake, no complete final authorial manuscript or
typescript exist, and so the choice would normally fall on the first edition as
the copy-text. The 1939 text, however,
as published by Faber and Faber in
How then did Rose and O’Hanlon edit Finnegans Wake? As
The editorial process is remarkably simple and elegant, mainly owing to three factors. First, Joyce was a very methodical writer whose method of composition (which David Hayman has dubbed the revise-and-expand technique) is quite linear. Second, the archive for Finnegans Wake is unusually complete; while there are several, obvious gaps in the draft levels, these are in most cases relatively minor. Third, while errors are introduced at each stage in the writing, the level of error only increases during the later stages of the book’s genesis, as a result of the increased bulk of the writing and the growing gnarliness of the language. (If I am allowed an impressionistic judgment, I would say that the main culprits who introduced the most errors are the printers of transition and, even more considerably, R. MacLehose, the printers for Faber. MacLehose not only introduced a lot of printing errors, they were also responsible, possibly owing to an entrenched house style, for normalizing many specific features of the intended mise-en-page, such as the extra spacing before “riverrun”, which they moved from its intended position half-way down the line to a normal indent.) Furthermore, errors tend to come only once—which may strike one as a peculiar contravention of the old printer’s devil. A word that is mistyped, or a passage that becomes garbled, does not tend to be garbled further. The ones that do (I have not found any, but I am sure there are some) are the exception.
To illustrate the process, let me take again the example of everglaning/everglading. The phrase “everglading mangrovemazes” (BL Add.47477-273; JJA 51:397) first appeared as part of a longer insertion in the left-hand margin of the first set of galley proofs for Finnegans Wake, pulled by the printers on 19 and 29 January 1938. As Joyce was wont to do, he had all his autograph additions retyped and keyed to the galleys by a typist, presumably to avoid the chances that the printers would overlook some of them. In this case, however, the typist typed “everglaning mangrovemazes” (BL Add.47477-272v; JJA 51: 396) and that is how the phrase appears in the published text. Rose and O’Hanlon restore “everglading”.
This error, as it happened, was not the only one the typist committed. On the same page, she (or he, for we do not yet have a detailed record of Joyce’s typists) in fact conflated and muddled two separate insertions. It was an easy mistake to make. One insertion is tightly crammed into a corner and woven around the other, practically without leaving a gap, so that the flow of the new text from left to right was quite confusing. What Joyce had added was “but throughandthoroughly proconverted”, and “with animal variations amid everglading mangrovemazes and propounded for cyclological beorbtracktors”. What the typist read was “but throughandthoroughly proconverted, and propounded for cyclological beorbtracktors”. Coming to “with animal variations amid everglading mangrovemazes”, she (or he) realized she (or he) had made a mistake. The word “beorbtracktors” appears just a fraction below the words next to it, pushed down ever so slightly by the writing above it. The typist then cancelled “beorbtracktors” in the typed overlay and added the word to “with animal variations amid everglaning [sic] mangrovemazes and beorbtracktors”, thereby creating not one but two errors, which resulted in at least one elliptical clause (FW 220.30 and 221.19-20). It takes a keen eye to spot this mistake in the manuscript. Rose and O’Hanlon justifiably reconstruct what Joyce had originally written.
But what if at some point Joyce changed his mind and actually preferred the error? The mistake would still not be volitional, though it might be fortuitous. Vladimir Nabokov considered this a real possibility. He believed no text should ever be corrected; “after all”, misprints he said, “may be supposed to have been left uncorrected by the author”. Joyce too, we know, was open-minded about serendipity. His creative instincts fed on accidental finds. He not only, as he once boasted, had the ability to do anything with words that he wanted; anything that came his way in terms of facts, stories, names, places, inventions, beliefs, rumors, etc. could always in some shape or form be fitted into Finnegans Wake. The book has thus become the most immense repository of history and human culture. So how can one rule out—as the textual doubters wonder—that Joyce did not passively authorize these serendipitous changes?
Strictly speaking, one cannot. Not with the Wake; nor with any other work of literature. But to be honest, it is not very productive to dwell on this. Where there are known unknowns—fine. Uncertainty can lead to conjecture. But where there are unknown unknowns, one engages in unfounded speculation. From a historicist and editorial point of view, passive authorization is a totally unworkable concept.
Passive authorization, it needs pointing out, is a notion that crops up almost only in debates on the editing of Joyce’s texts. I have rarely encountered it anywhere else in editorial theory. The reason for this should be apparent: it has no function in textual scholarship. Authorization is either active and present or it is absent. If it is absent, it leaves no trace. That which has left no trace cannot be introduced as evidence of anything. Admittedly, one can imagine that an author silently accepts alterations in his text—but how do we know? If we do know, authorization is not really passive. If we do not, we have not even got a way of ascertaining where it might occur and where not.
Take for example the lamentation “Úalu Úalu Úalu! Quáouáuh!” (FW 2010 003.24-25), which was printed in 1939 without the accents (FW 004.02-03). The phrase first entered the text thus on the first complete fair copy (level I§1.*2/2.*2), dated 29 November 1926 (BL Add.47472-5; JJA 44:106), with the accents; they were left out on the typescript and its carbon copy (prepared by Miss Weaver, who presumably could not do them on her English machine) (I§1.3/2.3 and 1 I§I.3+/2.3+), but were inserted again by hand on “Úalu”, though not on “Quaonauh”, which, incidentally had an additional error, when Joyce’s “u” was mistaken for “n” (Private Collection; JJA 44:146). In transition 1 (I§1.4/2.4) “Úalu” still appears correct, and one accent has been restored in “Quáonauh” (BL Add.47472-75; JJA 44:205), which suggests an intervention by the author in between the typescript and transition page proofs. On the missing second or duplicate proofs also the “u” was corrected, for “Quáouauh” appears correctly in the published issue. In the Galleys (first set, I§1.6/2.6) for Finnegans Wake “Quáouauh” appears unaltered, but now the three accents on “Úalu” have gone missing again; BL Add.47476a-1; JJA 49:005). The text was published in 1939, with finally only the accent on “Quáouauh”. Now from this one can assume that between 1926 and 1939 Joyce had changed his mind about the accents and that he intended the phrase to read “Ualu Ualu Ualu! Quaouauh!” Having intervened in “Quáouauh” at least four times, he had had sufficient opportunity to restore all other accents as well. But opportunity cannot count as evidence. To deduce Joyce’s intention from a hypothetical situation is not good editorial practice, particularly when Joyce’s failure to correct is easily explained as oversight—a simple case of human limitation. Authors (and we ought to know this from our own practice) rarely read proofs against the earlier copy from which the text was set. And so it is easy to see how even the most meticulous proofreaders overlook errors.
That said, since the work of D.F. McKenzie, editors do accept that agencies other than the author (printers, copy-editors, friends of the author and so on) play their part in the constitution of the text, though their authority is usually limited to issues of spelling, punctuation, page design and other small interventions which affect the text. But this is not passive authorization per se; this is delegation, for which, as Scheibe also says, the author leaves clear instructions (Scheibe, 177). Either the archive provides clues to this effect or one must reliably invoke house style or other printing house customs to which the author submits. (Those who reject the imposition of a house style—such as Joyce, for example, who insisted on dashes over “perverted commas”—leave clear instructions too.)
In Joyce’s archive, it is true, instances may be found where he deliberately chooses to adopt a transmissional error. When Joyce hired Madame France Raphael in 1933 to transcribe all the unused items from his notebooks — not an easy undertaking, given Joyce’s sloppy handwriting and the fact that she had little or no English. Not surprisingly, her work resulted in a substantial number of alterations. Joyce was well aware of this, and in most cases corrected her mistranscriptions when he began using the new notebooks, sometimes remembering what he had originally written, sometimes changing Mme Raphael’s garbled entry into something different. Only in very few instances did he accept his secretary’s mistakes, realizing the comic possibilities of her fabrications. So, for example, “/\c on vibrating bed” (VI.B.22.160) was transcribed by Raphael as “/\ convibrating bed” (VI.C.13.239), and ended up in Finnegans Wake as “convibrational bed” (FW 394.03). This is an example from the notebooks, however; I have not yet found a similar instance where Joyce unequivocally accepts a transmissional departure caused by a typist or printer. This is not to say they might not exist. But surmising (or even knowing) that there are errors that Joyce accepted does not amount to an edict that all errors must be left standing.
On the whole, for an editor to err on the side of error is wiser and more pragmatic than to slide into unfounded speculation. At least that is what keeps editing scholarly and critical.
There is a further point. When authors give the bon à tirer for their texts, they authorize the whole of the text,
but they do not separately authorize each individual word. This is not a paradox. Nor a merely theoretical point. Oversight and error are part of what it is to
be human. There are some very well-known
examples of authors who have “authorized” publications that contained the most
egregious errors. In Volume 2 of the
1800 edition of the Lyrical Ballads
15 lines of verse dropped out of the text of “Michael” on p. 210, leaving two
thirds of the page blank in a poem that otherwise is not subdivided into
sections. One can only wonder how it was
possible that neither Wordsworth, nor the printers noticed the gap. A more peculiar example, however, is the case
of Goethe’s Die
Related to passive authorization is another argument that is repeatedly made against editing Finnegans Wake, and that is that Joyce was constantly revising his text. So it is at least possible that Joyce might have changed a word or phrase that actually contained an error. Or it could be that an editor restores an original reading for a corrupted passage, but then the possibility exists that Joyce might have had occasion to revise that part of the text. Neither argument in fact fully holds. In the first scenario Joyce’s revision removes the error: that Joyce revised a section that was actually corrupt is simply not relevant, for the revised represents a new intention that replaces what was in the text, regardless of whether what was there was correct or not. The second scenario is again purely hypothetical, and thus it cannot sensibly inform editorial decisions. What has no evidential base cannot be used to arbitrate the validity of the text. Even so, the possibility is real, but if the logic of the argument in effect says that a passage that is corrupted by textual error but not revised by Joyce might have been revised if the error had not occurred, the probability of this happening seems on the whole rather small. Given that omissions, though substantial in number, are only one type of error, and given that only a fraction of these omission consists of passages of any significant length (most are individual words and phrases), the impact on the text remains rather limited.
Interestingly, the problem of revision poses itself in reverse. One important difference that distinguishes the archive for Ulysses from that of Wake—and thus also what distinguish Gabler’s editorial practice from that of Rose and O’Hanlon—is that for Ulysses we have the Rosenbach manuscript, which, though a composite document, comes close to what Sally Bushell calls an act of contingent completion. The existence of this document prompted Gabler to differentiate between documents of creation—all drafts prior to Rosenbach — and documents of transmission — all document that came after Rosenbach. The distinction furnished him with a fixed point before which variant readings ought not to be considered final. For Finnegans Wake, no such document or point exists, which poses at least a theoretical problem: how far back can or should one go to retrieve authorial readings?
For their edition, Rose and O’Hanlon appear to go back as far as is necessary to retrieve authorial readings—to the earliest extant draft if need be, even beyond to the notebooks where these appear to corroborate doubtful readings. Their reasoning may well have been motivated by the desire to find an “original” text. Yet the nature of the composition history is such that choosing any other point in the timeline of Work in Progress one cannot go beyond is tantamount to arbitrariness.
* * * *
It goes without saying that the surviving evidence does not always offer straightforward solutions for straightforward problems. Again, even though the archival record is at times incomplete and hard to interpret, the difficulties that arise are not unique to Finnegans Wake. Chapter II.1 for which there are substantial lacunae in the extant documents no doubt offers many examples. I will give one further down. Apart from an early draft, two very incomplete typescripts (all dating c. 1932), and some extra draft materials (presumably used for revisions on the transition proofs) for only one section of the chapter, no documents survive until the revised pages from The Mime of Mick, Nick and the Maggies (the layer of revisions dating from late 1937). With so much missing, these proof pages are already likely to contain a substantial number of transmissional errors. Notwithstanding the enormity of this gap, for the book as a whole I have found few instances where there are any really problematic cruxes. My sampling has been necessarily incomplete, but given the nature of the work I had expected more problems. For each instance that I discuss below, I will also give my sense of whether Rose and O’Hanlon’s handling of the issue is justifiable or not.
The first example picks up on my earlier discussion of the accents in “Ualu Ualu Ualu! Quáouauh!” (FW 004.02-03). Rose and O’Hanlon are confident in their restitution of the accents, since their editorial principle points them back to the earliest, uncorrupted reading. There is a further piece of evidence, however, which complicates the matter. Joyce in the “Corrections of Misprints in Finnegans Wake”, which he prepared with the help of Paul Léon in the summer of 1940, made a further alteration: he now also removed the accent in “Quaouauh!” (VI.H.4.b-1; JJA 63:352). On the face of it, the accent-less reading represents Joyce’s final intention. Yet one needs to ask what motivated Joyce to make this correction. Was it really a new intention? In that case one must take Joyce’s correction as another intervention that actually undoes an error, as I have argued above. Or was it simply the most economical manner of restoring balance? In that case one could be inclined to see Joyce’s correction not as a new intention at all, but just a quick way to patch up a fault. Joyce’s correction in other words would be rejected and the text would then revert to its original authorial reading. As a document produced after publication, the “Corrections of Misprints” may warrant different treatment and the evidence provided in it given different weight. Editors must choose in such matters between alternatives that may be equally viable. Either way, all words in the phrase must have accents or none at all; not just the one as in the 1939 text.
My second, more complex example is taken from chapter III.3 and pertains to a duplicate set of marked pages from transition 15, revised by Joyce some time in the summer of 1929 (III§3B.10 and III§3B.10’) in preparation for the Babou and Kahane edition of Haveth Childers Everywhere (1930). On one set Joyce noted the following two additions in the left margin, one slightly below the other: “blueeyed man in the lowbelt suit” and “, my nomesuch,” (BL Add. 47484b-347v; JJA 59:056). Furthermore, some confusion arises from the fact that the central text has a few insertion marks which do not have any overlay associated with them. The first insertion, moreover, appears to read “A blueeyed man” etc., but the capital “A” is not a indefinite article, but a siglum which connects “, my nomesuch,” to “blueeyed man”. The likely intended reading, therefore, is this: “The caca cad ^, my nomesuch, ^blueeyed man in the lowbelt suit^^! I protest it that he is by my wipehalf!” On the second set, however, and in the same location a hand other than Joyce’s adds different insertions that are incompatible with the instructions on the first set: “The caca cad! ^The snakeeye! ^Strangler of green parrots!^^ (BL Add. 47484b-351v; JJA 59:064). Although in the James Joyce Archive these duplicates are consecutively ordered, the actual sequence of revision is now impossible to determine. The typist instructed to make a new copy apparently faced the same problem. She resolved the issue creatively, it seems, and seeking to make sense of the confused overlay kept the readings of the second set, but placed the apposition “a blueeyed man” (with the indefinite article, which she changed to lower case) next to the nearest possible subject. The resulting passage runs thus: “the pupup publication of libel by any Ticks Typsyloon a blueeyed man in the lowbelt suit to my nomesuch that highest personage at moments holding down the throne. So to speak of beauty scouts, and the celluloid art! The caca cad! Snakeeye! Strangler of green parrots. I protest it that he is by my wipehalf!” (VI.E.13; JJA 59:073). This is clearly a non-authorial intervention which according to Rose and O’Hanlon’s editorial principles will have to be undone. However, the difficulty is that both revisions on the marked pages from transition are authorial, so both carry equal weight. There could be a clue in the various stray insertion marks which might indicate that Joyce was not entirely certain where to place his additions, and which therefore might provide an argument for following the second reading. Even so, this does not bypass the amalgamation created by the typist. There appears little alternative for the editor but to accept the readings produced by the typist, for in any case Joyce subsequently made several alterations to the passage, which have enshrined it in its current location.
In their edition, Rose and O’Hanlon, using the evidence that was available to them, follow a different course. In a way their decision to emend the passage follows both the principle of returning to the earliest, non-corrupted version while also ensuring that both Joyce’s authorial revisions were retained in the text, albeit that they introduce some transpositions of their own. In the passage below in which I quote the text as emended by Rose and O’Hanlon, I have tried to separate the relevant layers of the text using different colours; green text represents those parts of the text as it appeared in transition 15 and which Joyce subsequently marked up; red text corresponds to the two contradictory revisions which Joyce added on the duplicate set; black text indicates all subsequent additions to the text until publication:
into my preprotestant Caveat against the pupup publication of libel by any tixtim tipsyloom or tobtom towley of Keisserse Lean to that highest personage at moments holding down the throne. So to speak of beauty scouts in elegant pursuits of flowers, searchers for tabernacles and the celluloid art! Happen seen sore eynes belived? The caca cad! A bloweyed lanejoynt, waring lowbelt suit, with knockbrecky kenees and bullfist rings round him and a false roude axehand (he is cunvesser to Saunter’s Nocelettres and the Poe’s Toffee’s Directory in his pisness), the best begrudged man in Belgradia, who doth not belease to our paviour, he walked by North Strand with his Thom’s towel in hand. Snakeeye! Strangler of soffiacated green parrots! I protest it that he is, by my wipehalf, my nomesuch! He was leaving out of my double inns while he was all teppling over my single ixits. So was keshaned on for his recent behaviour. Sherlook is lorking for him. Allare beltspanners! Hourspringlike his joussture, immitiate my chry! As urs now, so yous then! Get your air curt! Shame upon pipip Private M—! Shames on his foulsomeness! (FW2010 415-26-416.02)
On the one hand, the editors’ intervention is a near-successful attempt to mitigate the textual difficulties. The whole section from “into my preprotestant” to “The caca cad!” removes the typist’s chance placement of “a blueeyed man in the lowbelt suit to my nomesuch”. The emendation is in part predicated on their interpretation of the location of “, to my nomesuch”,, which the typist also adapted to fit her interpretation of the context, and which Rose and O’Hanlon place at the unallocated caret after “wipehalf”. On the other hand, they choose the “first” insertion, now altered to “A bloweyed lanejoynt, waring lowbelt suit, with knockbrecky” etc., as the appropriate description for the Cad, and simply leave the “second” insertion, which now reads “Snakeeye! Strangler of soffiacated green parrots!”, in the same position as it appears in the 1939 text: after “Thom’s towel in hand” (see FW 534.27). (Strangely, they do not restore the definite article in “The snakeeye”, as Joyce had originally put it in the margin.) The solution is ingenious, but I am not sure if it follows the edition’s implied principles. Access to a full collation or the isotext would no doubt have clarified their reasoning, although I cannot see how the surmised linearity of their rationle, which as they strongly imply in their “Editorial Methodology” is what drives the conversion from isotext to edited text, actually works in this instance. Moreover, the whole emendation rests on the apparent primacy of “blueeyed man” over “snakeeye”, of level III§3B.10 over III§3B.10’. The arrangement in the James Joyce Archive was, of course, Rose and O’Hanlon’s as well. I am not able to see any conclusive evidence on which they might have based the order.
Unknown to them, the reality behind the revision of this passage was different. A new document which forms part of the Jahnke bequest at the Zurich James Joyce Foundation, now classified as MS Finnegans Wake 12, vindicates the typist. A third set of marked transition 15 pages, keyed to notes on extradraft sheets (some of which are in Lucia’s hand), provides a missing link between the duplicate set and the typescript. On this third set, Joyce clearly and carefully marked the position of the overlay, which corresponds to the placement found on the typescript (VI.E.13). The typist, therefore, did not intervene at all, but (barring a few typos) faithfully followed Joyce’s instructions after he realized he had made a mistake. This goes to show how possible new discoveries in the future will also make new emendations, and new editions, necessary. Rose and O’Hanlon, who obviously did not see this document, will no doubt reconsider their assessment of the textual detail. However, the larger point remains that the evidence that was available to them warranted a more conservative approach.
Another example of an amalgamation can be found in Chapter II.1. On the first typescript the Costumers are described like this: “a bundle of representatives who are sloppily served by” etc. To this simple phrase Joyce makes the following additions and alterations: he turns “a bundle” into “a draughthoarse bundle”; he revises “representatives” to “representative civics, each of whom is a jactitator” (BL Add.47477-25; JJA 51:009). The typescript that follows soon after omits some of these changes: “a bundle of representative civics who are slopilly [sic] served by (BL Add.47477-48; JJA 51:011). The typist committed many similar omissions so that one cannot infer Joyce’s hand behind these changes. When Joyce returned to the passage after publication of The Mime, he evidently was trying to fix the phrase that was wrongly typed up five years prior. Now the phrase, with Joyce’s insertions, reads: “a bundle of a dozen of representative locomotive civics ^, each^ inn quest of outings, who are ^still more^ sloppily served by” (BL Add.47477-150v; JJA 51:204). It survives as such in the published text of Finnegans Wake (FW 221.04). Too much of the record is missing (as I explained above) to know exactly what happened to “jactitator”—i.e., whether it ever returned; whether “outings” was a creative intervention on Joyce’s part to get closer to the original; or whether “, each” was used here for to get closer to the original typesript or whether it was reinserted because it had dropped out a second time before “inn quest of outings”. In their edition, Rose and O’Hanlon amalgamate this later reading with the lost reading from the earliest typescript, which results in “representative civics inn quest of outings, each of whom is a jactitator, who are still more sloppily served by” (FW2010 174.26-27). But for this to work they actually undo what is undeniably an authorial revision that superseded what was there at the earlier stage.
The intricacy of the
preceding examples makes that no solution is entirely satisfactory. But at
least in one instance Rose and O’Hanlon introduce a correction which simply
seems wrong-headed. On one of the sets
of marked-up pages from transition 15,
Joyce adds the following: “under
the advicies from Messires Norris, Sotheby[,] Yates and Weston, inc”. (BL Add.47484b-347v; JJA 59:056, slightly simplified).
In the typescript that follows this level, the typist appears to corrupt
the name of the famous
* * * *
The words “as I understand it” are important. As Rose and O’Hanlon have only provided the broadest descriptions of their editorial principles, I cannot even be sure that the examples I have given or the objections I have raised are at all valid. Not having the rationale set out in detail, and complemented by collations in a full apparatus criticus, is obviously a flaw. But there is a more serious problem.
Despite Rose and O’Hanlon’s claim that the emendations they admitted to the text follow the logic of the work’s composition, they have introduced changes which vitiate that logic. Rather than by following the textual evidence, they feel justified to do this in order to clarify the sense. This is where they stray outside of the domain of textual scholarship and into the domain of editing to taste. In their Afterword they make the following questionable claim. “Correcting the many manifest errors” which occurred in the text’s transmission was, they write, only “the less important of the editorial tasks”.
The greater task lay in the restoration through emendation of the syntactical coherence of individual sentences as they underwent periodic amplification under the writer’s revising hand. What is important is that the root sentence, considered as a logical linguistic structure expressed through syntax, retains its essential structure irrespective of its often complex expansion. In practice, yet not invariably, damage to this coherence was corrected by Joyce or one of his helpers. Otherwise it is visible in collation as a simple error. In other instances the loss or part-absence of the syntactical structure was not noticed and, as the sentence was further amplified, the damage intensified, often to the extent that its original and essential coherence is irrecoverable short of a full genetic analysis. (p. 36)
In more than one respect this is rather an unfortunate statement. Rose and O’Hanlon’s direct admission that they are willing to go beyond tangible evidence to correct the text of Finnegans Wake is simply indefensible in scholarly terms. What is also unfortunate is that it will remind readers of similar statements Rose made in his “Reader’s Edition” of Ulysses. The decision to correct what is “not noticed” is essentially a decision to assume an authorial, rather than an editorial, role. It sounds similar to the distinction Rose makes in the Reader’s Edition between “error” and “fault”. The semantics are maybe somewhat clumsy, yet at the root of it lies an actual difference: errors are caused by typists and typesetters; faults are caused by Joyce himself in the process of composition or during self-copying. (I am very much in favor of the German position on this—which is also the position adopted by Gabler—that editors correct texts; they do not correct their authors, even when the author makes a blatant mistake.) As a result, many will now feel even more reluctant to engage with the edition. Nonetheless, I believe Rose and O’Hanlon to be over-stating their case, caused no doubt by their own prose which suffers a little too from damage to its coherence. Two points are important to note. The first is that syntactical consistency is a feature of Finnegans Wake, but it is one that is hampered by persistent revision and corruption. The second point is that damage to the coherence will show up in the collations of variant readings. So only “in other instances” and what follows in the editor’s statement is suspect from a bibliographical point of view. As I have shown, instances of Rose and O’Hanlon’s creative interventions in the text can easily be found, but the majority of their emendations can be defended on purely textual grounds.
* * * *
Scholarly editions must be judged on their rationales and on the editor’s consistency in applying those rationales. For these are matters that make the scholarly debate possible; arguments about whether a particular reading is right or wrong only make sense when they happen not in a vacuum, but in the context of editorial theory and practice. The problem with the debate between Kidd and Gabler over Ulysses was that Kidd attacked Gabler for having made the wrong corrections, whereas his disagreements—as he finally acknowledged in his long piece in Papers of the Bibliographical Society of America—were about Gabler’s methods. Different methods lead to different editions.
Do I believe that Danis Rose and John O’Hanlon have created a good edition? Despite the problems and issues I have noted, the answer is yes. If given the chance, would I edit it differently? The answer again is most likely yes.
The ethos of modern scholarly editing is to mediate between the work and its various textual emanations. This is no easy a task, and it carries great responsibility. At the basis of it lies an understanding of and an appreciation for the history of the text. This understanding goes hand in hand with respect for the textual pluralism that is part of the textual condition. Particularly with a work like Finnegans Wake whose text was composed over such a long period of time the idea of textual mutability is ever-present. The creation of Finnegans Wake is something that happened quite organically, though not without direction, and resulted from different, perhaps even contradictory, intentional moments. Accepting such a position has a considerable impact on editorial theory, and thus on the way we edit the text of Finnegans Wake and other works like it. However, it does not make the task of editing futile, impossible or unnecessarily intrusive. One can draw a parallel with the notion of restoration in art history and architecture. On the face of it, restoration is the reverting of a damaged object to its original, pristine state, but in fact it is a process that mediates between the object in its original state and the object in its current state, in the course of which scholars gather further knowledge about the place and the time in which object existed and exists. (There certainly has been a movement away from restoration to conservation: the prevention of further deterioration.) Scholarly editing has in other words a lot in common with restoration. While it creates to all intents and purposes a new text, it also mediates between the various textual emanations—its different versions and variants—of the work and creates an awareness of its textual history. This is how textual editing is inextricably linked with reception history.
As I have said before, “critical” edition no longer means “definitive” edition. The printed medium does not easily allow for the inclusion of all possible choices and all possible versions; it cannot do justice to the fluidity of the textual condition, and so it is tempting to attribute finality to a critically established text, but this is a finality which it does not have. (Even Rose did not consider his Reader’s Edition the ultimate and final edition [Introduction, xi].) Yet it is important to remember that the clear-reading text is always only a part of the scholarly edition. Critical editions do not replace the original edition; they complement it. In the words of Peter Shilingsburg, “knowing who was responsible for each specific thing in a text affects how we understand it”, and so our research demands that editions teach us about “the history of the composition, revision, production and distribution of texts” (p. 32, 31). For this reason, the absence of a critical apparatus or unavailability of the isotext is a serious impediment. But even despite the edition’s other shortcomings, critics—and readers—of Finnegans Wake are the poorer if they choose to ignore the 2010 edition. For if nothing else, it teaches them that there is no such thing as the text of Finnegans Wake.
Wim Van Mierlo, Institute of
 A shorter version of this article has appeared in the Autumn 2012 issue of the James Joyce Literary Supplement.
 To put this figure in perspective, Jack Dalton 45 years ago announced up to 7,000 corrections, J. P. Dalton, "Advertisement for the Restoration”, Twelve and a Tilly, eds. Jack Dalton and Clive Hart (London: Faber and Faber, 1966), 129. Robert-Jan Henkes and Erik Bindervoet identified 2,228 mistakes, Robbert-Jan Henkes and Erik Bindervoet, “Finnegans Wake, the Corrected Text”, Genetic Joyce Studies 4 (2004): http://www.geneticjoycestudies.org/GJS4/GJS4%20RJE%20Corrected%20Text.htm [accessed 10 May 2011].
 Michael Wood, “Quashed
 Numerous commentators who remarked on the impossibility and undesirability of editing Joyce’s final book have preceded them. See, e.g., Roland McHugh, The Finnegans Wake Experience (Berkeley: University of California Press, 1981), 79.
 Critics like to point out that Finnegans Wake indulges in error. Examples to the contrary, however, also exist, e.g., “the compositor of the farce of
dustiny however makes a thumpledrum mistake by letting off this pienofarte effect as his first act” (FW2010 129.01-02).
 Tim Conley draws an
intriguing parallel between textual errors and “discernible errors of continuity” in films, citing
the “infamous incorrect ordering of two chapters in the
 F.O. Matthiessen, American Renaissance: Art and Expression in the Age of Emerson and Whitman (New York: Oxford University Press, 1941), 392. Matthiessen does not give a citation for the edition he used.
Rose, I do not put editorial judgment on a par with “common sense”; such
judgment is more complex than that (Danis Rose, “The New Edition of Finnegans Wake: Editorial Methodology —
A Very Brief Overview”, HOUYHNHNM —
Finnegans Wake [
 See Peter
l. Shillingsburg, From Gutenberg to
Google: Electronic Representations of Literary Texts (
 Clive Hart, A Concordance to Finnegans Wake (Minneapolis: University of
Minneapolis Press, 1963), iv. See also
Finn Fordham, “The Corrections to Finnegans
Wake: For ‘reading’ Read ‘readings’ (VI.H.4.b-2; JJA 63:352)”, James Joyce:
The Study of Languages, ed. Dirk Van Hulle (
 Danis Rose and John
O’Hanlon, Afterword, [Companion Booklet
to] Finnegans Wake, by James Joyce, ed. Danis Rose and John O’Hanlon (
 Vladimir Nabokov, Eugene Onegin: A Novel in Verse, by Aleksander Pushkin, trans. Vladimir Nabokov, Vol. 1: Introduction and Translation (Princeton: Princeton University Press, 1990), 16.
 Siegfried Scheibe implicitly speaks about passive authorization: “What we can and must assume is that the authorized witness documents provide the authorized textual versions, in that everything they contain was willed or accepted by the author” (“Theoretical Problems of the Authorization and Constitution of Texts”, Contemporary German Editorial Theory, eds. Hans Walter Gabler, George Bornstein and Gillian Borland Pierce [Ann Arbor: University of Michigan Press, 1995], 184). Note, however, that Scheibe talks about “authorized witness documents”, which is not the same as the published text. Hans Zeller fundamentally disagrees with Scheibe saying that passive authorization is essentially non-verifiable: the “surprising absence” of an author’s (dis)approval of outside intervention is not to be used as a scholarly criterion, “Record and Interpretation: Analysis and Documentation as Goal and Method of Editing”, Contemporary German Editorial Theory, 27-28. Peter Shillingsburg concurs: “to be passive is to be inactive” and therefore “passive authorization is a contradiction”, “A Resistance to Contemporary German Editorial Theory and Practice” in editio: Internationales Jahrbuch für Editionswissenschaft 12 (1998), 144.
 One of the most adamant defenders of a social theory of editing in English is Warwick Gould, who recognizes the authority of Thomas Mark, W.B. Yeats’s editor at Macmillan, and George Yeats, the poet’s widow, in substantial readings in A. Norman Jeffares’ edition of Yeats’s Poems (London and Basingstoke: Papermac, 1989). See also Warwick Gould. “W.B. Yeats and the Resurrection of the Author”, The Library, 6th ser., 16 (1994): 101-134. For McKenzie’s best exposition of his theory, see Bibliography and the Sociology of Texts (Cambridge: Cambridge University Press, 1999).
 In fact, the opposite viewpoint, as proposed by Siegfried Scheibe, is merely a theoretical point — that authors always authorize their work in whole and in all its parts (185). I am indebted to Burghard Dedner for his insights on this matter.
 See Hans Zeller, “Structure and Genesis in Editing: On German and Anglo-American Textual Criticism”, Contemporary German Editorial Theory, eds. Hans Walter Gabler, George Bornstein and Gillian Borland Pierce (Ann Arbor: University of Michigan Press, 1995), 116 and Shillingsburg, Gutenberg to Google, 177.
 The issue also occurs in other areas of life, in particular law. The rapidity with which lawmakers enshrine new laws and the complexity of the text of these laws are often at odds with each other, causing not infrequent ambiguities or vagaries that lawyers and tribunal need to sort out at a later date.
 Finn Fordham in his conscientious deliberation of this question comes down in favor of invention (“Corrections to Finnegans Wake”, 48-50). Rose and O’Hanlon appear consistently to have opted to overrule Joyce’s late corrections. See, for example, “Irenews” (FW 254.10) versus “Ireneus” (FW2010 200.30), which Fordham discusses in his essay.
 To represent Joyce’s revisions linearly, I use a simplified system of diacritics adopted from the Critical and Synoptic Edition of Ulysses. Insertions appear between carets (text ^inserted text^ text); cancelations appear between pointed brackets (text <cancelled text>); substitutions appear as a combination of both “(text ^<cancelled text> revised text^ text).
 I am enormously grateful to Nicholas Morris who provided me with a detailed commentary on this intervening stage in the revision of the transition 15 page.
 Wake criticism has only recently begun to recognize that the unit of meaning in the book is not limited to the level of the word, but also the sentence and the paragraph. The point was recently made by Michael Wood in his review in the London Review of Books.
 See John Kidd, “An Inquiry into 'Ulysses: The Corrected Text'“, Papers of the Bibliographical Society of
 I hesitate to use the word textual instability, for it conjures of a poststructuralist discourse which is of quite a different order than the kind of pluralities the textual editor encounters.
 See, for example, Paul
Eggert, Securing the Past: Conservation
in Art, Architecture and Literature (
 Many scholars mistakenly believe that a book which has a critical introduction at the front and some annotations at the back is a critical edition. What makes an edition not only a critical but truly a scholarly-critical edition is the inclusion of all variant readings in a critical apparatus. Market-forces, unfortunately, work against the production and publication of multi-volume scholarly editions, as a result of which detailed information of the textual lives of our greatest of works of literature is being withheld from students and researchers, whose lack of textual awareness is exacerbated by the disappearance of historical bibliography from university curricula.