University of Virginia Library

More on the Compositors of the First Edition of Don Quixote, Part II
by
R. M. Flores

In my article "The Compositors of the First Edition of Don Quixote, Part II", Journal of Hispanic Philology, 6 (1981), 3-44, I concluded that five workmen, compositors F. H, I, and J, and apprentice Z, had set the text of the first edition of Cervantes's Don Quixote, Part II (Juan de la Cuesta, Madrid, 1615). This overall conclusion still stands. I allotted the setting of gatherings as follows:

• Compositor F: gathering A
• Compositor H: gatherings B and D
• Compositor I: gatherings ¶, C, E-Aa, Cc, Dd, Ff, Hh, Kk, Mm, and 2N (apprentice Z: pages ¶7r, I4v, O1r, O1v, O2r, O2v, O7r, O7v, O8r, O8v, S1r, S1v, S2r, S2v, S7v, V1r, V1v, V2r, V2v, V7r, V7v, V8r, V8v, Dd6r, and Hh1v)
• Compositor J: gatherings Bb, Ee, Gg, Ii, and Ll

I noted, however, that apprentice Z had probably set more pages than I was able to account for with the textual and typographical evidence at hand, and that another apprentice might have helped compositor J with his stints. The reasons for these caveats were the close similarities between the setting habits and spelling preferences of compositor I and apprentice Z, and the many puzzling typographical inconsistencies one finds throughout the book. Then I discovered that both formes of the inner sheet of gatherings A and G, and both formes of the outer sheet of gathering Q had been set anew and machined for a second printing. The copies made up entirely of sheets from the first printing were classified as belonging to family group A, and those having the three sheets from the second printing as belonging to family group B.[1]

In this article I shall limit myself to re-examining the setting patterns and the distribution of labour in four gatherings of copies belonging to family group A of Don Quixote, Part II: gatherings B, L, M, and Y. All quotations are from copy Serís 12 of the Hispanic Society of America; references are by page and line.

The different widths of the leading separating lines of type in formal breaks of the text set in pages belonging to the same gathering, and the countless irregularities in the numbers of quads, contractions, characters, and abbreviations used by the compositors who worked on the first edition of Don Quixote, Part II strongly suggest that many formes of this job were set from cast-off copy.[2] Non-sequential setting took place throughout most of the work and can be detected already in the first gatherings of the alphabetical sequence, some gatherings evincing this process more clearly than others.

When a compositor thought that he was not going to be able to fit the text of the cast-off copy in his stint he began taking immediate, and not infrequently drastic, steps to forestall this mishap. Setting extra contractions, running words together, failing to set leading between text and titles, making pages with more lines of text than the normal number of lines, and widening the opening of the composing stick are all common compositorial practices designed to fit more text per line and per page than a stint would usually take. These measures are relatively easy to detect and explain. A practice which is difficult to uncover is when a compositor either was given or purposely took fewer pages of cast-off copy than those needed, thus altogether avoiding the possibility of having any text left over at the end of his stint. In such a case the compositor knew from the start that he did not have to worry about overrunning his stint and could, in fact, begin setting more quads, fewer abbreviations and contractions, and wider leadings than was his usual practice. All these irregularities may, if one is not thoroughly familiar with the compositor's normal setting habits, be misleadingly regarded as characteristic of the compositor.

In order to illustrate the complexities of the setting patterns followed by the compositors during the setting of the first edition of Don Quixote, Part II, I shall first examine a gathering set entirely by one compositor. Gathering B, set by compositor H, shows clear-cut signs of having been set in an unconventional fashion; see Table 1.[3]

At least five possible explanations for the irregular per-page distribution of quads, contractions, and characters, and the different widths of the leadings used in these pages could be advanced, namely: (1) inconsistent compositorial habits, (2) type shortages, (3) setting by more than one compositor, (4) setting by formes from cast-off copy, or (5) a combination of any of these possibilities.

In my opinion, explanations 1 to 3 should be ruled out. I have found no textual or typographical evidence to suggest that more than one compositor might have set type for gathering B. Type shortages and compositorial inconsistencies were not uncommon phenomena during the handpress period, but one would expect such shortages and inconsistencies to occur at random and, consequently, to appear in the printed text independent of page format. In fact, only possibility 4 can fully explain all the typographical irregularities shown on Table 1 without resorting to any of the other possible explanations for further support.

When one separates the figures entered in Table 1 into two groups, according to whether they occur in the inner or in the outer sheet of the gathering, and adds them up, the contrast between the numbers of quads and the numbers of contractions and characters set in each sheet becomes too striking to be coincidental. The majority of contractions and characters was set in the pages belonging to the inner sheet of the gathering, whereas the majority of quads appears on the pages of the outer sheet (the inner sheet has 81 quads, 181 contractions, and 12860 characters; the outer sheet has 127 quads, 48 contractions, and 12709 characters). The inverse relationship between the numbers of quads, and the numbers of contractions and characters set per sheet (more quads / few contractions and fewer characters; fewer quads / more contractions and more characters) suggests that compositor H altered his setting stride whilst working on this gathering.

What could have happened is that the pressmen ran out of formes and asked compositor H to finish another forme for them. Compositor H might have obliged by casting off copy and begun setting in a non-sequential fashion. One can deduce from the contrast between the number of contractions, characters, and quads set in pages B1r to B3v and those set in page B4r (the first line of this page has three contractions!), that compositor H began setting from cast-off copy soon after he had finished setting type for page B3v.

Because compositor H had four pages of the outer sheet already set (B1r, B1v, B2r, and B2v) and only two of the inner sheet (each page belonging to a different forme: B3r, outer forme, B3v, inner forme) he apparently decided to

cast off copy for pages B7v and B8r and set these pages first in order to complete and impose a forme for the pressmen to run (inner forme of the outer sheet). The printer's copy had a chapter division in the cast-off copy allotted to these pages; therefore he did not bother to alter his setting habits in any substantial way, and when he finished setting pages B7v and B8r he found himself with an extra line of type. Because the last line of Chapter 4 (B8r, 25) consists of only one word the compositor could have pulled out the quads he had set on the upper section of the page, justified these lines of type anew, and moved up the last word of the chapter to save one entire line; but he must have been in a hurry to meet the pressmen's needs because instead he merely substituted a narrower than normal leading for the one he had initially set between the text and the chapter-title for Chapter 5, thus saving both time and a lot of work for himself.[4]

If these premises are correct we could further speculate that compositor H could have begun setting type for pages B7r and B8v using more contractions and fewer quads than was his usual practice, in an attempt to avoid finishing with more lines than he needed for these pages. The result probably was that he ended one-half line of type short, which he could have filled out by developing some contractions and increasing the number of quads already set. Instead compositor H apparently resorted, once again, to a more practical though less appropriate manner of dealing with the problem. He introduced a paragraph division where in all likelihood none occurred in his copy (B8v, between lines 2 and 3), even though he had to readjust thirty-two lines of type.[5]

The remaining pages of the stint were apparently set in a similar fashion, pages B4v, B5r, and B6v first (for the outer forme of the inner sheet) and pages B4r, B5v, and B6r last (for the inner forme of the inner sheet) because of the progressive increase of contractions and decrease of quads set per page in these groups of pages. As before, the text of the cast-off copy did not fit in the assigned pages exactly, and compositor H found himself with an overrun for each set of pages. To absorb the overruns, compositor H set a narrower leading in page B5v, and the considerable number of contractions set in this page and in pages B4r, B6r, and B6v suggest that he must have pulled out a substantial number of types (vowels, m's, n's, q's) replacing them with the appropriate contractions (vowels with a tilde, q͂'s; see note 4). But the adjustments he made to the pages of the inner forme did not, apparently, suffice to absorb the entire overrun (approximately four lines of text), because there seems to have been some tampering with the text of the printer's copy.

A passage from pages B5v and B6r reads: "Eſſo es coſa facil, y no acontecimiento | nueuo, que lo meſmo le ſucedio a Sacripãte, quando eſtã | do en el cerco de Albraca, con eſſa miſma inuencion le ſa-|cô el cauallo de entre las piernas aquel famoſo ladron lla-|mado Brunelo" (B5v-B6r). In the printer's copy this excerpt must have had a reference identifying the speaker; e.g., dixo or interrumpió don Quixote.[6] But compositor H left out this textual reference in order, presumably, to make the text of his copy fit in the pages

he had not as yet imposed. By omitting this clause compositor H saved an additional space equal to at least eighteen types.

This typographical and textual evidence leaves no room for doubt. I am virtually certain that pages B4r to B8v were set from cast-off copy and that the progressive decrease in the numbers of quads and the progressive increase in the numbers of contractions and characters set per page as well as the use of narrower than normal leadings can only lead to the setting order shown in Table 2.

If one accepts the non-seriatim order of setting proposed in Table 2, the pieces of this typographical and textual puzzle all fall neatly into place. More important, the unconventional manner in which pages B4r to B8v were set is a clear warning to the investigator that other gatherings of the first edition of Part II might also have been set in a similar fashion, as indeed they were.

In "The Compositors of the First Edition of Don Quixote, Part II", I stated that because the "typographical evidence supplied by gatherings B and D" was inadequate, I could not tell "exactly whether or not [compositor H] was setting from cast-off copy", and that only six of the twenty-nine gatherings assigned to compositor I had been set from cast-off copy (p. 39). I was wrong. The two gatherings assigned to compositor H were both set in this fashion, as was the majority of gatherings assigned to compositor I. It would be somewhat repetitious and, I think, unnecessary for my purposes here to study every gathering of the first edition of Part II in the same depth as I have re-examined gathering B, but three other gatherings of this edition do beg for attention. The majority of pages belonging to gatherings L, M, and Y show, like pages B4r to B8v, clear typographical signs of having been set in a non-sequential order; but in the case of these three gatherings I am concerned not so much with the order in which the pages were set as with whether or not the compositors speeded the output of formes by each working on a different section of the same gathering concurrently.

Striking variations in the number of quads, contractions, and characters set per page, and on the thickness of leadings used, are useful typographical data to determine whether or not a gathering, a sheet, or a forme has been set from cast-off copy, provided that the entire text of the section being examined has been set by one compositor. This kind of typographical evidence should, however, be approached with caution when it is being used to delimit the stints of different compositors each working from cast-off copy

on different pages belonging to the same gathering. True, different compositors usually took different precautionary measures when setting from cast-off copy, absorbing overruns and distributing shortages of text in different ways; but one frequently finds the same compositor resorting to different solutions to solve similar typographical problems. Spelling preferences are more reliable indicators than typographical evidence because they are not last-minute responses to unforeseen typographical complications and they can, in most instances, be followed throughout a job.

As I was revising my findings concerning the setting of gathering B, I noticed that compositor H had set all occurrences of the word sé (I know) appearing on gatherings B and D with a short s and an accented ê (see pages B3v, 24; B4v, 11; B6r, 9 and 16; B6v, 28; B7v, 5 and 17; B8v, 8 and 10; D2v, 8; D3v, 3; and D7r, 21) vis-à-vis ſe/ſê, which were the forms preferred by the other compositors. When I referred to the concordance of my control edition of Don Quixote I discovered that six other occurrences of the form sê appear in a gathering which I had originally assigned to compositor I (see pages L6v, 22; M1r, 7 and 10; M7r, 29; M7v, 23; and M8v, 5) and that these gatherings also have five occurrences of the forms ſe/ſê (see pages L2r, 15; L7r, 15; L8v, 29; and M6r, 2 and 3). Table 3 shows the distribution of these variants.

The six occurrences of the variant sê appear on pages belonging to the inner sheet of gathering L and to the outer sheet of gathering M. The five occurrences of the variant ſe/ſê appear on pages belonging to the outer sheet of gathering L and to the inner sheet of gathering M. This discriminating distribution of variants cannot be coincidental. Two workmen, compositors H and I, must have shared the setting of gatherings L and M, and they were in all likelihood setting by formes and concurrently. That this highly practical manner of distribution of labour between different compositors or between a compositor and an apprentice did in effect take place during the setting of the first edition of Don Quixote, Part II, has already been shown in the above mentioned article (pp. 48 to 50); thus, these new findings should not strike one as unduly surprising. Nine spelling variants occurring in gatherings L and M support this theory; see Table 4.

The nine spelling variants listed on the left-hand column of Table 4 clearly show that pages L5v (variants 1 and 7), L6r (variant 2), L6v (variant 9), M1r (variants 3 and 9), M1v (variants 3 and 4), M2r (variants 5 and 7), M2v (variant 6), M7r (variant 9), M7v (variants 7 and 9), M8r (variant 7), and M8v (variants 7, 8, and 9) were not set by compositor I, whose characteristic corresponding spelling variants appear on the right-hand column. These pages must, therefore, have been set by compositor H, whose characteristic spellings they bear. But what about the other pages of the inner sheet of gathering L (pages L3r, L3v, L4r, L4v, and L5r) which show no variant characteristic of this compositor? Considering that pages L6r, L6v, M2v, M7r, and M8r have only one variant each it is not too daring to conclude that pages L3r to L5r were also set by compositor H even though they have no spelling variant to

back up this assumption; but the fact that the five pages showing no compositorial variant characteristic of compositor H follow one another is suspicious because one would expect to find such pages not as a group but rather interspersed amongst the other pages of the inner sheet of gathering L. The deliberate omission of some words occurring in this section of the gathering is also suspicious.

The first reading of the first line of page L5v does not tally with the catchword of the preceding page: (L5r, 34) . . . ojos. | Con | cuchauan . . .(L5v, 1; the printer's copy probably read, ". . . ojos. Con grandiſima atencion eſcuchauan . . .", see FCE, lines 26725 and 26726). It is unlikely that a compositor setting two consecutive pages would have made such an error. If one now refers back to Table 4 one soon realizes that the readings "eſcuſara" (L4v) and "deſpertara" (L5r) which were entered under compositor H as the only exceptions to variant 7 are, in fact, not exceptions at all but rather readings characteristic of compositor I. In other words, compositor I set thirteen pages of gathering L (L1r—L5r and L7r—L8v), whereas compositor H set only three pages of this gathering (L5v—L6v). And the presence of some outstanding typographical irregularities clearly implies that when the compositors were working on gatherings L and M they were setting concurrently and from cast-off copy.[7] Furthermore, the number and distribution of the quads set before commas and contractions appearing on page L5r (seven of the ten quads set before commas appear on lines 4 to 8, and the sixteen contractions—seven of which are q͂'s—appear on the last twenty-one lines of the page) strongly suggest (1) that compositor I decided or, more likely, was asked to cast off copy when he was in the middle of setting page L5r for which he had already set from thirteen to fifteen lines of type, (2) that he probably kept one entire page of the printer's copy to complete his page,[8] and (3) that he immediately began setting fewer quads and more contractions than was his usual practice. Lastly, from the type of error occurring between pages L5r and L5v one can deduce that despite the cautionary measures taken by compositor I he was unable to fit into the number of lines he had available the entire text of the manuscript page, whereupon he omitted the text he could not fit in his stint with no concern whatsoever for the textual gap he was creating.[9]

Another compositorial error that can be explained once one realizes that casting off copy and setting concurrently were common practices at the Madrigal-Cuesta press when a job had to be set and run in a hurry, as no doubt had the first edition of Don Quixote, Part II, occurs in gathering Y.

On page Y2r the text reads: "Si la ſente͂cia paſſada de la bolſa del ganadero mouio â admiraciõ a los circũſtantes, [la del labrador y el ſaſtre] les prouocô a riſa" (lines 14 and 15). The problem lies in the fact that Sancho's judgment concerning "la bolſa del ganadero", which the text announces as having already taken place, does not in fact begin until the middle of page Y3r (line 13). This textual error cannot be authorial. In Cervantes's manuscript Sancho's judgment concerning the cattleman's pouch must have preceded that of the farmer and the tailor. Such a substantial transposition is most likely the result of an accidental shifting of manuscript pages. If this

theory is correct one should be able to reconstruct a hypothetical printer's copy of the entire episode of the cattleman's pouch by assigning nineteen lines of type set for each manuscript page (see note 8) and adjusting the number of lines of type set according to the abnormal increases in the number of types saved by the use of contractions (approximately forty en types make a full line of type). Originally the episode of the cattleman's pouch must have begun and ended in a slightly different fashion than it now does in the printed text, but the words "vna | muger" (Y3r, 13-14) and "admirados" (Y4r, 20) encompass the entire episode without leaving any textual gaps; see Table 5.

It is crystal clear from Table 5 that the episode of the cattleman's pouch fits, as was to be expected, in exactly four manuscript pages or, more pointedly, in exactly two sheets of paper written on both sides. The theory propounding the shifting of sheets can thus be taken one step further.

Like gatherings B, L, and M, gathering Y shows some of the most common typographical irregularities one finds in gatherings set from cast-off copy: pages Y1v and Y2r each has thirty-five lines instead of the normal thirty-four lines per page; no leading was set between a poem and the text on page Y5v nor between the text and a letter on page Y8r; two narrower than normal leadings appear on page Y4r; a leading was set between the running title and a chapter-title on page Y6v; an unnecessary paragraph division occurs on page Y1r; and the number of contractions set per page varies from 4 (Y8v; six types saved) to 69 (Y1v; ninety types saved). There is no question in my mind that the great majority of pages belonging to this gathering were set from cast-off copy. Moreover, the per page distribution of contractions and quads and the location of the pages where the other typographical anomalies occur strongly suggest that gathering Y was set by formes; see Table 6.

It is no coincidence that the pages belonging to the outer formes (usually the first to be set for each sheet) have fewer contractions and more quads per page than those belonging to the inner formes.[10] It is no coincidence that

the typographical anomalies that suggest that the compositor had more text than was needed occur in the inner formes (see pages Y1v, Y2r, Y4r, Y5v, and Y8r) and those showing that the compositor had less text than was needed occur in the outer formes (see pages Y1r and Y6v). One should now be able to reconstruct a hypothetical printer's copy for the first seven pages of gathering Y; see Table 7.

In addition to the displaced episode's having to fit in an even number of manuscript pages, three other conditions must be met to make the aforementioned theories watertight:

Furthermore, because these essential conditions have been clearly met one can be virtually certain that manuscript pages 2 and 3 had to have been the recto and verso pages of one sheet of paper (see also pages 4 and 5, 6 and 7, 8 and 9, 10 and 11, and 12 and 13; sheets iii, iv, v, vi, and vii respectively).

Once the page order of the manuscript pages used as printer's copy for this section of the book has been restored one can deduce why, when, and how the shifting of manuscript pages 4, 5, 6, and 7 must have occurred. Textual and typographical evidence suggests that the compositor who set gathering Y (compositor I) cast off copy immediately after he had finished setting the first page of the gathering. He probably tried to finish his page with the last words of manuscript page 2 to cast off copy by complete manuscript pages (see footnote 10), but he apparently found himself short of text. Rather than turning to the following manuscript page for the needed text (manuscript page 3; "eſtâ eſcrito y notado el dia", Y1v, 1), he decided to introduce an unnecessary paragraph division (Y1r, 6-7) even though he had to go to the trouble of rearranging the type for the remaining lines of the page. He then cast off copy for pages Y1v to Y8v and began setting type for the other three pages belonging to the outer forme of the outer sheet (Y2v, Y7r, and Y8v). After having finished setting the outer forme of the outer sheet compositor I turned his attention to the pages of the inner forme of the outer sheet (Y1v, Y2r, Y7v, and Y8r). Because compositor I was setting in a nonsequential order he was still unaware that two sheets of the printer's copy (sheets iii and iv) had been accidently displaced from position A (see Table 7) to a position somewhere past manuscript sheet vii. Hence he had no reason to query the misplaced reference to the episode of the cattleman's pouch (page Y2r). Had he noticed this textual error in time he would have in all probability deleted the misplaced reference and reworded the sentence. The excerpt "Si la ſente͂cia paſſada de la bolſa del ganadero mouio â admiraciõ a los circũſtantes, eſta les prouocô a riſa" could have been easily changed to read, Esta sentencia provocó a risa a los circunstantes. Had the compositor made this amendment readers would have never known that the order of one of Sancho's judgments had been tampered with.

Compositor I must have realized that something was amiss whilst setting type for page Y3r (first page of the sequence for the outer forme of the inner sheet), when after having set the last line of manuscript page 13 (verso page of sheet vii) he began setting the first line of manuscript page 4 (recto page of sheet iii).[12] Because the outer forme of the outer sheet was in all probability already being run and the inner forme of the outer sheet had already been, or was being run, compositor I had no other recourse than to interpolate Sancho's judgment concerning the cattleman's pouch between the judgment dealing with the two old men and the ten escudos and the closing of Chapter 45 (pages Y3r, 13; Y4r, 20). Given the textual corner into which compositor I put himself, his solution to the problem of the shifted manuscript pages is more than adequate. True, editors must still deal with the misplaced reference on page Y2r, but this editorial hurdle is a minor obstacle easy to overcome.

In conclusion, one can now ascertain that the page patterns followed by the compositors during the setting of the gatherings that make up the first editions of Parts I and II of Don Quixote are far more complex than appears on the surface and that they are the result of specific circumstances which arose during the setting and running off of formes.

I have studied here in some detail the differing sequences in which the pages of gatherings B and Y were set (B1r, B1v, B2r, B2v, B3r, B3v, B7v, B8r, B7r, B8v, B4v, B5r, B6v, B4r, B5v, B6r; Y1r, Y2v, Y7r, Y8v, Y1v, Y2r, Y7v, Y8r, Y3r, Y4v, Y5r, Y6v, Y3v, Y4r, Y5v, Y6r) and the distribution of labour in gatherings L and M (compositor I: L1r, L1v, L2r, L2v, L3r, L3v, L4r, L4v, L5r, L7r, L7v, L8r, L8v; compositor H: L5v, L6r, L6v; compositor I, M3r, M3v, M4r, M4v, M5r, M5v, M6r, M6v; compositor H: M1r, M1v, M2r, M2v, M7r, M7v, M8r, M8v). It is evident that concurrent setting and setting from cast-off copy were, at least in the case of the first edition of Don Quixote, Part II, the norm rather than the exception. Cuesta must have wanted to finish the job promptly and must have asked the compositors to begin setting concurrently in order to maintain a steady supply of formes to keep pace with the pressmen.

The importance that typographical and textual evidence has in disentangling the different setting habits and spelling preferences of the compositors and apprentices who set type for this work of Cervantes's and, thus, in solving and explaining the countless textual queries that face editors on every page of the first editions of Don Quixote cannot be overstressed. Press conditions may in some instances be a matter of chance. Typographical and textual evidence is not. It is a permanent and reliable record of when and how a work was set and printed. When examined closely and thoroughly understood, textual and typographical data are demonstrably trustworthy and helpful sources of information.

Notes