a codicology

The following is adapted from my doctoral thesis, “Migrant Textuality: On the fields of Aimé Césaire’s ‘Et les chiens se taisaient’” (“Migrant”).

Except for the actual text, all the relevant material evidence is detailed and organized in a series of codicological tables below.1 All of the rows in the tables, except for the headers, represent a page. These are organized from top to bottom in the order in which they are found today in Saint-Dié des Vosges—what we could call for brevity’s sake, the Saint-Dié witness. This order also corresponds to the pagination scheme labeled Page.

The first table (table a), lists the support materials used for the typescript. I recognize a total of five different types of paper—four if you ignore size—in the 108 loose sheets of the Saint-Dié witness. I distinguish these under the heading Paper as atemoya, atemoya+, batata, coco and durian. The most common type is atemoya, thin, of a yellowing white hue (21.6 x 28 cm). Those pages marked atemoya+ have sheets of the same material as atemoya, except they are a few milimeters larger lengthwise. batata and coco belong to two self-contained segments in Act 1. batata is thin and pink (21.6 x 27.8 cm). The twenty leaves made of this paper are only found towards the beginning of the play as it stands now. coco follows batata in the current pagination, and is thin and white with a light blue tint (21 x 27 cm). Last, but not least, durian seems to have been introduced at a late stage of composition, and mostly appears in revisions made to Act 2. This paper is thick white and also yellowing, at times with rough-hewn edges, indicating the leaves came from a larger sheet. (22.4 x 28 and 21.6 x 28 cm).


figure a | atemoya


figure b | batata


figure c | coco


figure d | durian

The document seems to be, in large part, a copy of a missing original, and traces of carbon paper are visible to the naked eye in several pages. At least two kinds of carbon paper were used: blue and black. The traces of carbon paper residue in the typescript vary in intensity: at some points the paper is very dirty, at others very clean. At times, the carbon residue outlines unreadable glyphs, line traces from a previous page on the roller. Some pages are marked orig, for original, because we can’t perceive any carbon residue on them—or at least not without more scientific rigor. These pages mostly place towards the end, and indicate a late stage of the process. None of the use patterns align with other patterns we find in the pagination schemes, neither with the type of paper, nor with the type of revision instrument, which leads us to conclude that Césaire had a consistent supply of two types of interchangeable carbon paper. In theory, we could use the patterns to isolate smaller bursts of writing, but I have not had occassion to do it.

black carbon paper

figure e | black carbon paper

blue carbon paper

figure f | blue carbon paper

Besides direct overwrites with the typewriter, the text was revised using a variety of instruments likely handheld by Césaire himself. And analysis of the handwriting—a slight forward slant, a tendency for flat diacritics, sharp connecting strokes, a peculiarly open p, etc.—confirms these revisions belong to the same hand that revised the ~1938 typescript of the Cahier d’un retour natal archived at the Bibliothèque de l’Assemblée Nationale in Paris.

The different types of handwriting instruments used in the the Saint-Dié witness are catalogued in table b by the pages in which traces can be found. Blue ink markings, ranging in hue from light to dark, are by far the most common, and they are almost universally present in the typescript. The final pagination itself is almost exclusively written in blue ink. Unless we are allowed a “thermal desorption and gas chromatography-mass spectrometry” of the blue ink (Bügler et al 982–988), we won’t be able to determine whether we are indeed talking about two different types of ink, or whether the same ink became darker with age, or under certain atmospheric conditions. In any case, I distinguish between them aided solely by comparison between samples. These are indicated in table b as blue ink and dark ink. Revisions made with blue ink range from corrections to substantial additions.

Distinguishable red ink markings can be found in a few scant pages belonging in the first two acts. The red ink plays an important role in erasing some of the historical traces from the typescript.

Pencil markings come in three varieties, ranging from most common to most rare: light pencil, dark pencil and red pencil. The light pencil is somewhere in the graphite hardness range of 1H and HB, probably your everyday No. 2. The dark pencil, with more clay in the lead, ranges between 2B and 4B. Most pencil markings seem to occur during the middle stages of composition, and they play an important role in the temporal shift of key episodes and pages. The red pencil is a soft pencil, and can only be found on four pages at the latest stage of composition, in the episode of the parliamentarians.

blue ink

figure g | blue ink

dark ink

figure h | dark ink

red ink

figure i | red ink

dark pencil

figure j | dark pencil

light pencil

figure k | light pencil

red pencil

figure l | red pencil

The witness has interesting features which push the boundaries of traditional codicology, and yet make fine additions to our catalogue. These range from the regular, to the highly peculiar. Some of the regular features are listed in table c. For example, some pages end half-way through, usually marked with a large X drawn across the empty bottom space of the page. I flag these in the table as Half Pages. These pages can sometimes give us clues about the insertion of a page in a previous sequence, and other times help us isolate blocks of innominate and itinerant text.

Starting on page 41, and intensifying after page 60 of the current pagination, the typist2 tries to use as much of the real estate on the page as possible, sometimes schewing the line towards the end of the page, as the paper began to slip from the platen roller. Perhaps paper became scarcer during the later stages of composition due to the American blockade? Perhaps the household became more frugal? This marked difference in the use of paper is indicated in table c as Max Use.

slanted text

figure m | oh, well

The last of the regular features is the uneven practice of underlining the speakers, marked with an x under the heading Underline. This pattern is consistently used on the pages composed using batata, then dissappears for a while, only to intensify during the composition of the middle pages on atemoya paper. This quirk may indicate two typists, or simply two coherent periods of composition. Further evidence suggests the latter, but we cannot rule out the former.

The Saint-Dié witness also yields many unique material clues, some helpful, some odd. We still don’t know the typewriter that was used, but some clues point in the direction of an answer. The type alignment in page 52 and 62, for example, is slanted—as if the paper locating guide was missing, or simply ignored. On many of the pages where carbon paper was used, four evenly-spaced columns of residue are visible to the naked eye. What machine produced such effects? A Remington? A Japy? An investigation of the type and rollers of period machines might answer the question.

Some of these unique traces give us very useful clues about the order of composition and revision. On page 70, for example, traces of blue ink matching the corrections on page 69 suggest the pages were stacked in order when the revisions were made. This also implies that these corrections were made after the typescript had achieved its current ordering, since page 70 belongs to a set of three pages that followed page 41 according to a previous pagination.

Other traces suggest working methods. The verso of page 71, for example, is a reverse mirror image of the front of the page, suggesting the typist accidentally placed a second, black carbon paper facing the wrong way. This implies that at least 3 copies of that page were attempted using the sandwich method: the missing original, (with blue carbon paper behind), our First Copy (with the flipped black carbon paper behind), and a missing third.

Some traces, alas, remain nothing more than a tease to would-be genetic critics, like the words “notions de base” in all caps and underlined, in the back of page 75.

stack and shuffle

We are now ready to retreive the last key: pagination. In total, we can count ten different pagination schemes spread throughout the typescript. The pagination schemes are catalogued in table d. The schemes are differentiated from each other by the writing tool used and their position at the top of the pages, and though they imply an order of composition, this is not always the case.

Most pages have more than one set of numbers on them. Three sets of numbers come from a hand other than Césaire’s: the cataloguer’s numeration, the first pagination, labeled P1, and a set of typewritten pagination on several pages, P3. The last numeration added by Césaire determines the actual order of the pages as we find them today (Page in tables a-d)3 This numeration will provide our canonical reference for this work, often in parentheses.4 The page sequence established by Page is reinforced by the added pagination of a Saint-Dié des Vosges librarian, which I assume to be Albert Ronsin, who received the witness from Claire Goll, after Yvan’s death. I’d divide the rest of the numbering into two periods, early and late. The late paginations P6-P8 begin when P6 introduces a continuous numeration that encompasses and organizes most of the text as we find it now, going from P6.1 to P6.103. The early ones P1-P5 mark a period of reorientation and experiment, where whole sections will change their place in the order or be heavily revised. The different pagination numbers also provide us with the principal criteria for re-organizing the text into smaller blocks to help us orient our genetic analysis of the text.

paginations p1, page and p9

figure n | left to right: P1, Page, P9

In order to best understand the process by which we arrived at the current pagination, we can reverse-engineer the pagination schemes. The results of the reconstruction are visualized in figure p below, starting in chronological order from left to right. As in table d, each numbered row in figure p represents individual pages. Each of the matte, lime, lavender and clay-colored columns in figure p represents the state of the text in nine different stages of development, extending from left to right from the earliest stage S1, corresponding to the first pagination, P1, to the Final stage, corresponding to the main pagination, Page. A stage of development is not to be confused for a pagination, even though they coincide in many cases. In other cases, a set of pages does not receive a pagination scheme until later in the process, when a new order begins to emerge. Césaire’s texts, as those who study his bibliography know, have a tendency to migrate within and without themselves without end.

The page numbers belonging to each stage are written immediately to the left of the colored columns. To signal pages where we can tell a pagination is offered at a later stage of development, we use an “x.” The numbers in gray correspond to either missing pages or page numbers marked in an earlier pagination scheme and implied in the later stage. Missing pages are devoid of color, and are indicated by dashed borders. A pagination scheme is written at the top-most cell where that pagination occurs for the first time on a given stage. Places where the text suffers a split are marked by dotted lines at the cut and splice. A few cells are marked with additional texture to help you locate their destination on a later stage. Long curly brackets in the far left contain the final act divisions. As opposed to table d, the order of the pages from top to bottom is , rather than rational, in the sense that the vertical alignment is determined by design choices meant to help you appreciate the transposition in Act 1, the shifting elements around a structural core in Act 2 and the intercalated doubling of Act 3.

My rationale for this reverse-engineering follows the results.5

nine stages

figure p | nine stages (click for full image)

The fact that both P1 and P2 cohere around separate and distinct paper support, and that their paginations stand unaccompanied by any other pagination until the text receives a global sequence in S6 through the combination of P6 and P7, makes it difficult for us to choose between them with empirical certainty for the title of earliest extant segment in the typescript. My analysis bends me in the direction of the 18 pages marked P1 (23-40), all using coco paper.6 I indicate this choice in figure p as S1.

Let me illustrate the way the graphic works using this example: Under the label S1, the matte-colored cells represent only a total of 18 pages, the pages of P1, starting with page P1.1. Pages P1.3 and P1.4 have gray numbers next to an empty space surrounded by dashed lines, indicating in this case that they’re missing, but implied. Since I estimate that these 18/20 pages were the first to be written, all other cells in this column are empty below the colored cells. Because P1.1, by definition has the number 1 written on it at the top, and because I want to emphasize the drastic forward shift of these pages at a later stage of composition, I placed the column right at the top, in the starting position of our left-to-right reading order.

The handwriting used for this pagination is different from Césaire’s, and it can be found on the top left of its pages, very close to the corner. To reiterate, the pagination of the extant pages is indicated on the table by the black numbers under the P1 column. At this early stage, we have no evidence that Césaire envisioned his work in progress as a three-act drama—but despite his fib to Beloux, the early pages do suggest Césaire began writing a play meant for a theatrical stage (Beloux 28). The segment begins without the heading “Acte” at the top of page 1, as all other pages marked 1 do in the other numeration schemes. coco, the thin white paper, is also not present in any other pagination scheme. Only in one other case does a specific kind of paper correspond to a specific pagination scheme: batata and P2—the failed candidate for S1.

Five features of the text suggest that P1 was the earliest segment: a) Page 1 begins with stage directions that set the scene “à Saint-Domingue,” announcing the general setting for Act 1 and Act 2; b) The episodes in the segment roughly align with the early period of the Haitian Revolution; and, c) Césaire provides and subsequently eliminates several specific passages in P2 that refer to the mid-stage of the revolution; d) The verb tense in P2 changes from Passé simple to Futur simple on several occassions; and finally, e) The segment contains the least amount of poetic interventions by Toussaint (Le Rebelle of the published versions), the chorus or the reciters of all the drama, providing the most straightforward episodes of the whole. If indeed P1 was the first batch composed, this last feature suggests that the original orientation of the drama leaned towards a popular stage—even if it remains to be seen whether the drama is suitable for an actual theater or not.

If I’m correct, all the pages with P1 (23-40) shifted down in the overall structure of the text after the re-arrangement of those pages marked P2. The strongest evidence for this shift comes from the fact that, though both start their sequences close to or at 1, the pages of P2 where the only ones to land in the final ordering close to where its own numeration indicated they should be. If we were to claim that P2 was composed prior to P1, then we would have to conclude that P1 was not meant to be the beginning of Act 1. The evidence does not support this conclusion. The reader will remember that P1 begins with the number 1, and at no point until P6 does this change. Why would Césaire begin at number 1 what will be page 23 if he had meant for this new text to be a continuation of P2?

One possible response to this line of reasoning is that he wrote both in tandem, undecided about their relationship to each other. This perhaps would explain the use of different paper stacks. Another possible response is that P1 was written as an alternative beginning to P2. After all, both retain their low numbers until P6. Yet another is that P1 was meant to be the beginning of a putative Act 2. The problem with these hypotheses is that the elements of the story and the verse tense that were revised within P2, mentioned above, do not make much sense outside of the shift. A much simpler explanation is that P2 was written after P1, and received its pagination in the next stage, S3. Furthermore, as we shall see in the next chapter when we take a closer look at the text of P2, after its historical references and the tense fall prey to the mighty pencil, P2 surrenders the clavis aurea for the typescript and then some.

We can neither prove with absolute empirical certainty that the P2 pages were composed before any of the next segments before P6, but the order of composition in figure p remains likely. Present in Act 1 only, the numbers of P2 were written in pencil on the top-left corner, going sequentially and uninterrupted from 3 to 23 (2-22).7 Just as in P1, and as I said above, the pages of P2 correspond to one unique type of paper. In this case, the unapologetic pink batata. The next pagination in the series, P3, provides further reason to believe that P2 should come second in the series. In indicate this in figure p as S2. The numbers of P3 were typewritten on the top-left corner of the page. This pagination can be found exclusively in segments belonging to Act 2 and Act 3. If P2 was written after P3, I don’t see why it wouldn’t have followed suit and marked its pagination in type as well. It seems probable that the pages of P2 were written as those of P1, before the structure started congealing, leaving the pagination blank during the typing, and adding it later in pencil, after the edifice started to take shape.

P3, the typewritten pagination, will eventually be replaced by P4 and P5, with page numbers now written on the top-right corner. Since we can be fairly certain all other paginations inhabiting the same pages as P3 are posterior to it, we can be confident that P5 and P4, which overlap for the most part with P3, mark close, subsequent stages of development in the structure of the text, giving us S3 to S5. Top-right pagination for P2 would have aligned it more with this middle period. The case being the opposite, the top-left pagination groups it with the earlier stages. Since P3 was also marked in the top-left corner, albeit with type, the empty corners ofP1 and P2 could be filled in on the same spot as the type.

The extant P3 pages begin at P3.3(43), with P3.1 and P3.2 for Act 2 missing, and in Act 3 with P3.2(74), with the ever-intriguing page 41 possibly in the role of P3.1, leaving the number 1 out of the page because it begins the act.8 This numeration scheme, restarting in each act from 1, will soon be replaced by global continuous pagination. In that sense S3 marks the stage of development when we know with certainty that the text takes the current three-act structure.

Regardless of the placement of the pagination, if P2 was written after P3, (still a possibility) this means that it was not written with a clear sense of where it belonged, even after the overall structure was becoming clear to Césaire. It would also entail that Césaire started using atemoya to compose P3, then switched to batata to compose this strange section without a home, to return to atemoya for P4 and P5, leaving pink paper behind. This seems unlikely. Putting aside the lingering suspicion that Suzanne was at the typewriter, if Aimée was indeed using different paper types to divide possible segments/acts, why does Act 2 and 3 use mostly the same paper? Furthermore, the first page of the current witness, typed in atemoya paper, replaces a now missing page of P2, suggesting a timeline for the support materials. Based on this evidence and a dash of Occam, in the most likely scenario, the pages of P2 were written before the composition of Act 2 and Act 3, and therefore before P3, P4 and P5.

Thus far, while arguing for the position of P2 in the sequence, we managed to place P3, P4 and P5 as well, and with each a new stage. As we approach S6—when Césaire unites all acts under one sequential numeration for the first time—we should note that all three acts follow a different genetic pattern (figure p). Act 1 and Act 3 are different from each other in that the first joined two blocks of text sequentially, like stacking a deck of cards, while the last intercalated text in the crevices of an earlier structure, like shuffling it. Act 2 evidences a much more complex genetic process. While Act 1 and Act 3 attain their stability soon enough, after P6 the middle act continued to be reworked, prompting changes in what was clearly meant to be a final sequence. Notwithstanding, by the time we arrive at S6, the text has clearly achieved a structural landmark.

Although the changes to Act 2 could be considered minor in relation to the whole text, they are substantial relative to Act 2 itself. As you can see from figure p, the changes after P5 comprise about 1/4th of the final text of Act 2. Several of these changes are also attempts at reframing the act. Starting with P6.68, and following with P8.41'-41''' we see an effort to change the way the act begins and ends.

Determining which came first, P6, P7 or P8, has not proved to be an easy task. Two of the segments were revised with a writing instrument, P7 with red pencil, and P8 with dark ink. However, they both use the same kind of paper that comes into play after P6, durian. Moreover, while both have blocks of pages that relocate before the final organization, the transpositions of the former take place in the middle of the text, while the transpositions of the latter take place around the frame, making it difficult for us to determine their relative order of composition.

Complicating us further, the P8.41'-41''' set includes characters not present in the rest of the drama, Les bonnes gens and the Déterreurs des pierres, and P7.47 includes an all too common poetic exchange between the Reciters and Toussaint, making both the set and the single page relatively portable. All of these may have been written without a clear resting place in mind. The exeption to this mobility are both P7.48 and P7.56-57, which fall under the episode of “des parlementaires.”

I end up placing P8 in S8 after all because of one basic inference: P6.42 was written during S5 as the beginning of Act 2, and takes on the number 42 at the top of the page as part of the P6 sequence, and it does so with the same stroke of the pen that erases the words “Acte II” from the header. That means that at this point, page 41 already had its number and current title, “Acte II.” Since P8 is nothing other than a three-page insert between these two, that gives us some reason to suspect they were written at a later stage. In the end we can be flexible ourselves and consider these late stages to be part of the same final push to get the pages in order and declare the drama done.

Finally, the Page pagination adopts most of the numbering of P6, with some from P7 and P8. The process continuously shifts the overall numeration down to accommodate all extra pages added after P6, right up until the last moment, when the transposition of P8.41-41' to the end of Act 2 tilts the sequence up a bit, and is the only major change of this stage. The final sequence in Page is how the witness is ordered today in a folder belonging to the Fonds Claire et Yvan Goll at the Saint Dié des Vosges bibliothèque, near the German border.

works cited


  1. If you would like to verify my results or reconstruct my observations, here’s the complete set of codicological tables as comma separated values↩︎

  2. I have not been able to establish the identity of the typist, despite inquiries to more senior scholars who knew the poet, or analysis of the typos. I suspect, though, that Suzanne Césaire is our mysterious typist. ↩︎

  3. As we will see shortly, Page inherits most of P8 until page 55. What I am calling Page is a combination of page numbers from P6, a few pages of P7 and P8, and page numbers written at the last stage of pagination. In Table A, in the column named Page, the page numbers which are inherited are written in gray in order to distinguish them from those written at the time when the text received its final organization. ↩︎

  4. Other numbering systems will be indicated as pagination.page. Sometimes the Page number will follow in parenthesis when appropriate, or to disambiguate paginations by act. Ex. P6.55 (56) is the same as ‘the page with the number 55 in the number set labeled P6, with the number 56 in the set labeled Page.’ ↩︎

  5. Instead of presenting the full visualization here, I cropped the original to reduce load sizes and to help you get back to the text quicker, in case you want to skip it altogether. Design decisions like these are part of a certain minimalist bent in my work. ↩︎

  6. As I indicated before, the numbers in parentheses represent the Page pagination. In this case, these pages were originally numbered 1, 2…20. That would be the P1 numbers. In the final ordering they became 23, 24…40. These would be the Page numbers. ↩︎

  7. We assume the existence at some point of P2.1 and P2.2, but these have disappeared without a trace. ↩︎

  8. Page 41 leans two ways. On the one hand the page is clearly marked as the beginning of Act 3 before a correction makes it the beginning of Act 2. The page also flows into P3.2 through a simple bibliographic accident: Page 41 ends with the speaker “La Recitante”, but no speech, close to the bottom of the page, having run out of space; P3.2 begins with a speech close to the top of the page, but no speaker. One objection to this bibliographic evidence, would be the speech itself, which places in the mouth of the Recitress the idea that Toussaint’s punishment was part of “une justice suprème,” a supreme justice. An attentive reading of the ethical ambivalence of CRR dismisses this quibble. A possibility is that the page was indeed written posterior to P3 as a replacement for the original P3.1 and the speaker without a speech at the bottom of the page was done to accomodate the speech on the next page.

    Much stronger is the bibliographical evidence pulling page 41 in the other direction. The page is written on durian paper, with the page number 41 written with dark ink, doubly grouping the page with P8, which follow it immediately as P8.41'-41'''. One thing is clear, the page was written before the P6.49 which replaces it as the introduction to Act 3. ↩︎