University of Virginia Library

A Tale of Two Printings: Don Quixote, Part II
by
R. M. Flores

It has been tacitly accepted that all copies of the first edition of Don Quixote, Part II (Juan de la Cuesta, Madrid, 1615) were made up with sheets belonging to the same printing.[1] This long held, though hitherto untested assumption is incorrect. A thorough collation of copies Serís 12 of the Hispanic Society of America and Arch.B.e 7/3 of the Bodleian Library reveals a substantial number of textual and typographical variants between these two copies.

No two copies of the first edition of Don Quixote, Part II, are identical,

of course, a few textual variants are the result of impromptu changes made by the pressmen during the printing process, but the great majority of textual and typographical differences one finds between most copies arose when types got damaged, displaced, or fell during the running off of the formes.[2] However, the textual variants with which I am concerned here do not fall into any of these categories, because they were not incurred accidentally, nor are they offhand corrections done in press.

The majority of the variants listed in Table I are spelling variants, addition, deletion, or switching of words, and obvious typographical errors. The nature of the variants, and the fact that they are clustered together in only twenty-four pages (out of a total of five hundred eighty-four pages) belonging to exactly three full sheets of the job imply that two sets of these sheets must have been set and printed at different times. What in all probability happened is that when the Cuesta-men were gathering the various sheets that make up the copies of the first edition of Don Quixote, Part II, they found themselves missing some copies of the inner sheet of gathering A (A3-A6v), the inner sheet of gathering G (G3-G6v), and the outer sheet of gathering Q (Q1-Q2v and Q7-Q8v). Then, instead of disposing of several hundred sheets of paper which were already printed and folded (a complete copy of the first edition of Don Quixote, Part II, consists of seventy-three sheets), Cuesta must have decided to have the twenty-four pages re-set and the missing number of sheets run off to make up for the shortage; paper, rather than skilled labour, being the dearer part of most printing jobs during the hand-press period.

We can confidently state, too, that pages A3-A6v, G3-G6v, Q1-Q2v, and Q7-Q8v of copy Serís 12 belong to the first printing of these sheets because their running titles are the same as those used throughout the first edition, and, hence, that the corresponding pages of copy Arch.B.e. 7/3, which have their own characteristic set of running titles, belong to the second printing.

Furthermore, because seventeen errors which occur on the pages of copy Serís 12 recur on the re-set pages of copy Arch.B.e. 7/3 we can safely conclude that the compositors who re-set them must have used the printed pages rather than the authorial manuscript as printer's copy.[3]

Table 1 shows the corrections made and the substantive changes introduced by the compositors of the re-set pages. Table 2 shows the distribution of accidentals (contractions, expansions, additions and deletions of hyphens, and displacements of line endings), including some punctuation changes not listed in Table 1 (changes from colon to semicolon, or viceversa, and from comma to colon or semicolon).

Three spelling variants between copies Serís 12 and Arch.B.e 7/3, and one correction have not been listed in Table 1: Serís 12, "mãdo" (A3, 16), "ſe" (A5v, 10), "Quizà" (A6v, 24), and "tra |baſſe" (G5, 25-26); Arch.B.e 7/3, "mãdô", "ſê", "Quiza", and "tra-|baſſe", because they are copy variants of the first printing (He53/16 and Harvard; "mãdô", "ſê", "Quiza", and "tra-| baſſe"). The variant concerning the missing hyphen is unimportant for my purposes here, but the fact that the three spelling variants of the second printing coincide with the readings of the Yale and Harvard copies suggests that the state of the printer's copy used by the compositors of the re-set pages was similar to the state of the Yale and Harvard pages.

Tables 1 and 2 have the twenty-four re-set pages already grouped into six distinct compositorial stints: A3, A3v, A6, and A6v; A4, A4v, A5, and A5v; G3, G3v, G6, and G6v; G4, G4v, G5, and G5v; Q1, Q1v, Q8, and Q8v; and Q2, Q2v, Q7, and Q7v. Below I explain how I arrived at this page distribution, but I found it convenient to separate the pages into compositorial stints here in order to make the orthographic preferences of the compositors who re-set them stand out, expediting the presentation of my theories concerning the division of labour.

Firstly, it is necessary to keep in mind that the Cuesta-men who set the printer's copy used to re-set these twenty-four pages were compositors F (A3-A6v) and I (G3-G6v, Q1-Q2v and Q7-Q8v).[4] Then, knowing, as we now know, that compositor F had strong spelling and punctuation preferences, and given the large number of spelling variants introduced by the compositors who set the pages for the second printing of the inner sheet of gathering A (eighteen spelling variants in each group of pages), we can safely conclude that he is unlikely to have had any hand at re-setting these pages. He certainly did not re-set pages A3, A3v, A6, and A6v, because whoever re-set them capitalized the two occurrences of the word "andante" one finds in these pages (variants 17 and 22), a form which goes counter to the preferred spelling of compositor F. He was the only compositor of the first edition of Don Quixote, Part II, who set this word invariably with a lower-case initial.[5]

Just as in this instance one dominant spelling variant has helped us clear compositor F of having re-set pages A3, A3v, A6, and A6v, two other dominant spellings point to the identity of the compositor who did, in all probability, re-set them. The only compositor of the first edition of Don Quixote, Part II, who seldom accentuated the preposition "a" or the disjunctive conjunction "o" was compositor I (variants 2, 8, 19, 23, 24, and 29; the only exception is variant 15, which is probably a foul-case error).[6] If we now compare the other spelling variants introduced in these four pages with the preferred spellings of compositor I, it comes as no surprise that most variants (variants 4, 12, 17, 20, 21, 22, 25, and 27) correspond with the spellings of the majority of occurrences of these words as they appear in the gatherings of the first edition of Don Quixote, Part II, assigned to him. In fact, the only form not set by compositor I in the first edition, or by any other compositor, is the word "cerebro" which always occurs as "celebro". There is, however, one puzzling feature of the overall changes introduced in this first group of pages; the relatively small number of non-substantive variants made by the compositor of the re-set pages (see Table 2). One would expect a compositor re-setting pages originally set by a different workman to introduce far more than forty-one non-substantive variants in a four-page stint of Don Quixote. There is no contradiction in this fact, however. Of the ten Cuesta-men who worked on the first editions of Don Quixote compositor I was the workman whose punctuation habits were more inconsistent and unpredictable than those of any other compositor; and this lacking of punctuation preferences on the part of compositor I led him into accepting and retaining as a matter of course the punctuation marks set by the other compositors.

We do not find this go-along attitude in the approach of the compositor who re-set pages A4, A4v, A5, and A5v. The workman who re-set these pages made one hundred fifty-five non-substantive changes, almost four times as many as compositor I made in the same number of pages. This discrepancy between the two groups of pages and the substantial number of non-substantive changes introduced automatically eliminates both compositor F and compositor I from having re-set pages A4 to A5v, especially because the majority of the spelling variants introduced in these pages do not correspond with the spelling preferences of either compositor. They correspond, instead, with those of compositor J. For example, the only occurrence of the construction

"ê yo" in the first edition of Don Quixote, Part II, appears in a gathering set by compositor J (signature M7, 17; cf. variants 35, 36, 37, 51, and 59). Also, of the five men who set type for this edition, compositor J was the workman who showed more consistency in accentuating some verbal forms (see variants 34, 43, 53, and 54; variant 42 is probably either a compositorial inconsistency or, more likely, a foul-case error).

It is not too difficult to guess the identity of the compositor who re-set the next group of pages (G3, G3v, G6, and G6v). None of the other compositors but F would have changed the three occurrences of the word "Andante" that appear in these pages to "andante" (variants 89, 98, and 103; see above). It is doubtful, also, that any of the other workmen would have introduced so many spelling variants (thirty-three) in only four pages of text. Moreover, all the spelling variants correspond with the forms preferred by compositor F.[7]

But of the six compositorial stints entered in the Tables, the fourth (G4, G4v, G5, and G5v) and the last stints (Q2, Q2v, Q7, and Q7v) are the least difficult to unravel. The total number of changes made in these eight pages is so small (one hundred two; including sixty-five accidentals and seventeen corrections!) and the spelling variants introduced so few (four!), that only the compositor who set these pages in the first place could be the workman who set them anew, e.g., compositor I.[8]

Lastly, all textual evidence points to compositor J as the Cuesta-man who re-set pages Q1, Q1v, Q8, and Q8v. The overall nature and numbers of the punctuation changes made by the compositor of the re-set pages tally with those made by compositor J in pages A4 to A5v closely, even though these two groups of pages were originally set by two different compositors (see Table 2). The orthographic changes made from unaccented o's to accentuated forms (variants 127, 128, 129, 130, 135, 155, and 157) and from accentuated a's to unaccented forms (variants 44, 45, 46, 47, 132, and 133); the accentuation of future-tense forms (variants 134 and 137); the spelling of the word asolver with double s (variant 148; see "aſſoluerle", Don Quixote, Part II, Bb3, 1); the capitalization of the word "Norte" (variant 136; see Don Quixote, Part II, Gg4, 30); and the typographical change from the ending -ſſô to -ſsô (variant

140), are all spellings and setting habits characteristic of compositor J's. The only departure from his well-established, though not entirely consistent, spelling habits is the change from "Gouierno" to "gouierno" (variant 138). (No occurrence of the word "irreparable"—variant 131—appears in the first edition of Don Quixote, Part II.)

Three compositors, then, re-set the pages of the second printing. Compositor I re-set pages A3, A3v, A6, A6v, G4-G5v, Q2, Q2v, Q7, and Q7v; compositor F re-set pages G3, G3v, G6, and G6v; and compositor J re-set pages A4-A5v, Q1, Q1v, Q8, and Q8v.

Of the four copies of the first edition of Don Quixote, Part II, which I have examined (see footnote 2), only one has the re-set pages, copy Arch.B.e 7/3 of the Bodleian Library. In this copy alone I have found no sufficient typographical evidence to pinpoint the exact sequence in which the pages were re-set and the formes run off, but five typographical features give us a general, overall picture of what might have happened.[9]

• 1. The running title used in one of the re-set pages, signature Q2, belongs to a set of running titles used in some of the other gatherings of the edition (see signatures Bb3, Ee4, Gg8, Ii6, Ll5, and Ll7).
• 2. A set of eight running titles was set up especially for the re-set pages, and was used in all six compositorial stints. The running titles appear as follows: (i) A3v, G5v, Q7v, (ii) A4, G6, Q8, (iii) A4v, G6v, Q8v, (iv) A5, G3, Q1, (v) A5v, G3v, Q1v, (vi) A6, G4, Q7, (vii) A6v, G4v, Q2v, and (viii) A3, G5 [?]. Some of the types used in the main sets of running titles were distributed and then re-used in some of the new running titles; for instance, the lower-case n used in the running title of signature Mm6v reappears in running title iv, the upper-case Q used in the running title of signature Mm1 reappears in running title vi.
• 3. The state of wear of running title i suggests that the first forme run off of the three which have this running title was the inner forme of the inner sheet of gathering G, then the inner forme of the outer sheet of gathering Q, and last the inner forme of the inner sheet of gathering A.
• 4. The compositorial stints do not correspond with the pages that make up formes, but rather with the pages that would appear either on the upper or lower half of the printed sheet. This distribution implies that three sheets of the first printing were cut in half and the half pages were then distributed amongst the compositors to be used as printer's copy.
• 5. Only pages G4-G5v were imposed with an upper margin of a similar width to that of the average margin used elsewhere in the book. The other five

  ---

  294

  
  
  
  groups of re-set pages were imposed in the chase with narrower head spacing pieces, so that when the book is open, let us say at pages A2v and A3, the lines of page A2v (first printing) are not aligned with the corresponding lines of the opposite page (A3, second printing), the upper margin of page A2v being some 9.5 mm wider than that of page A3.

Taken together, this typographical evidence suggests that the shortage was discovered shortly after the last formes of the edition had been run off, when the sheets had already been folded and were being gathered to make up the books. At least one running title from the original sets was still standing, but the majority had already been distributed because the compositors had to make up a new set, in which they re-used some of the types used in the old sets.

The use of only one set of running titles for all the stints indicates that the pages were all part of one job. That is to say, the pages were re-set, and the formes were imposed and run off not as the individual shortages were discovered, but only after all the shortages had been accounted for.

The compositors were given copy in a fashion intended to speed up the completion of formes. Two compositors setting type for one forme concurrently (compositor I was setting pages A3 and A6v probably at the same time that compositor J was setting pages A4v and A5 to complete the outer forme of the inner sheet of gathering A) would finish the four pages of a forme in half the time it would take one compositor to do the job alone setting by formes, which was the more common practice for re-setting at the Madrigal-Cuesta press. This distribution of pages indicates that the compositors wanted to have the formes ready for the pressmen one immediately after the other, probably because only a few copies of each sheet were needed and it did not take the pressmen long to run them off.

The anomalous width of the upper margin of the majority of the re-set pages implies that the Cuesta-man who imposed these pages was not aware that compositors I and J had used wider spacing pieces throughout the other gatherings. Given that pages G4, G4v, G5, and G5v show an average upper margin, we can deduce that these pages were not only set, but also imposed by compositor I who used spacing pieces with the appropriate width. Compositor F, on the other hand, who had set only one gathering of the first edition (gathering A) and who might not have imposed even those pages himself, is likely to have assumed, incorrectly, that his coworkers were using narrower pieces and simply imposed his own pages with the wrong wooden furniture at their head. It is possible that when compositor F finished setting his stint (G3, G3v, G6, G6v), he imposed his pages in chases already containing those set by compositor I, which already had a head spacing piece in place. He might have then decided, or been asked, to take over the task of imposing the other formes himself, thus leaving compositors I and J to continue the setting of type without unnecessary interruptions. This theory would support my contention that the first forme run with running title i in it was the inner

forme of the inner sheet of gathering G, and would explain (1) why two formes have two pages with narrower than average upper margins and two pages with margins of an average width, (2) why the other four formes were all imposed with only narrower than average head spacing pieces, and (3) why compositor F set only one stint of the job.

The reconstruction of events I have just outlined, and the theories I have advanced seem to me a plausible interpretation of the meaning of the scant bibliographical evidence we have at our disposal in this instance. But whether or not they describe accurately what did indeed take place at the Madrigal-Cuesta press during the re-setting of these twenty-four pages from the first edition of Don Quixote, Part II, has little bearing on the rest of the story. What is of the greatest importance is the fact that we have extant samples of the printer's copy used to re-set these pages, and, thus, that we can bear witness to the manner in which compositors F, I, and J approached their copy.

The compositors corrected forty-five errors, true, but they missed seventeen other obvious errors (see footnote 3) which should have been set right. They also introduced one hundred thirty-four errors and non-authorial readings (thirty-three typographical errors, eighty-seven spelling variants, and fourteen alterations that affect the text), and made one hundred seventy-three important changes to the punctuation of their copy. The number and nature of the punctuation and substantive changes made by these three compositors may seem small and unimportant to cause much concern amongst specialists, but when the average number of changes per page (thirteen) is multiplied by the number of pages with text (six hundred fifty-eight) that make up the first editions of Don Quixote, Parts I and II, the apparently innocuous thirteen changes per page attain their true proportions in relation to what the compositors who set type for these editions must have done to the text of the authorial manuscript. Over eight thousand possible, important changes to the original text is not a number which could be dismissed lightly. Such a massive number of changes would distort the style, vocabulary, and spelling and punctuation characteristics of the printer's copy. It is becoming more and more evident that this is exactly what has happened in the case of Cervantes's works. We perceive Cervantes's style, vocabulary, and orthography only through the irregularities, imperfections, and impurities of the bottle-glass panels placed between us and Cervantes's writings by the compositors of the first editions of these works. One example:

The earlier version of this passage evinces a clearly stylistic purpose in the repetition and alternation of two phonetically different forms of the same word (meſmo/miſmo). After the ill-advised and insensitive regularization and modernization imposed by compositor F to his copy, nought is left of this delicate balance but the monotonous repetition of the same form.

If the compositors made these sorts and this number of substantive changes to a perfectly legible copy, even though the text had already, no doubt, been put through a process of regularization and modernization by the compositors who set the pages of the first printing (compositors F! and I), and even though they had to make an effort to follow the text of the printer's copy closely because of the restrictions imposed on them by the line-by-line, page-by-page setting, one can only shudder at the changes they must have made to the not-so-legible, not at all restrictive format of the authorial manuscript from which they set the first editions of Don Quixote.

The tale these two printings tell is not a new one, nor does it have the sort of ending upon which one could build pleasant thoughts; but tales such as this one will never lose their fascination, because they are never identical, and their differences make the telling worthwhile.

Notes