From: Cervantes: Bulletin of the Cervantes Society of America 8.1 (1988): 55-60.
Copyright © 1988, The Cervantes Society of America

IRTUALLY EVERY LINE of the first editions of Parts I and II of Don Quixote (Juan de la Cuesta, Madrid, 1605 and 1615) invites close scrutiny and careful consideration from a specialist preparing an old-spelling edition of this work of Cervantes's. Seemingly unimportant readings cannot be glossed over. Generalizations unsupported by textual evidence are not infrequently incorrect. For example, the compositors of the first editions of Don Quixote set a hyphen at the end of a line to denote that a word had been split and set in two lines only when space allowed it; hence, one could assume that the first-edition readings “mandarnos | lo” (I, 01^v, 12-13) and “vereiƒ- | lo”^* (II, L8^r, 6-7) should be edited to read “mandarnoslo ” and “vereiƒlo.”¹ In the following

     ^* Another version of this article is available which uses tiny .gif files to produce ,  , and rather than ƒ ,  ƒl, and ~u. -F.J.
     ¹ Unless otherwise noted, all quotations are from the first editions; references are by Part, page, and line. For the workmen who set these editions see my monographs The Compositors of the First and Second Madrid Editions of “Don Quixote,” Part I, The Modern Humanities Research Association, London, 1975; “The Compositors of the First Edition of Don Quixote, Part II,” Journal of Hispanic Philology, 6 (1981), 3-44; and “A Tale of Two Printings: Don Quixote, Part II,” Studies in Bibliography, 39 (1986), 281-96.

pages I shall demonstrate that these forms are not necessarily the most appropriate solution; see Table 1, items 10 and 45.²

     ² Three of the times listed in Table 1 are occurrences of the noun oíslo (mi oíslo ‘my wife’), a compound word made up of the verbal form oís and the direct object pronoun lo; see items 4, 39, and 64. Line divisions have been noted only when they affect the cluster sl; cf., for instance, items 3 (first-edition reading, “moƒtrad- | nosla”) and 10. The first-edition readings for items 41 and 66 are “en comendemos lo” and “darem osles.” The letter entered after the bibliographical data given for each item identifies the compositor who set the reading. Contrary to what I stated in my article dealing with the compositors of Part II, the workman who set the outer and inner formes of the outer sheet of gathering M was not compositor I, but rather compositor H; see my article “More on the Compositors of the First Edition of Don Quixote, Part II,” submitted for publication. See Table 1, items 46, 47, and 48.

     Table 1 lists the sixty-six occurrences of compound words made up of a verbal form ending in s and a third person object pronoun that appear in the first editions of Don Quixote. Confronted with the three different fashions in which the occurrences of this grammatical construction were set (one word with ƒl, one word with sl, and two or three words), an editor must first decide whether to retain all three forms, only two, or regularize all readings to agree with the form most commonly used by the compositors.
     There are only thirty-three readings containing the ligature ƒ1 in Don Quixote, and they appear scattered throughout the work;³ therefore, it is unlikely that the use of types from this sort could be the result of a temporary shortage of type.⁴ The variant ƒl | sl is not a spelling variant, but rather a typographical variant resulting in exactly the same consonant cluster, but set with types from different sorts (ƒl, s and 1). Hence, this variant must reflect a compositorial setting habit which should not be regularized offhand.
     The nine two- and three-word occurrences (items 22, 35, 41, 47, 48, 51, 54, 57, and 66; items 10 and 32 are words split and set in two lines) could be instances of the compositors' common practice of interpolating separation quads to justify their lines of type, in which case the elements of each occurrence should be united. But, because in addition to the nine two- and three-word occurrences entered in Table 1, one finds sixty-four other occurrences of similar readings set in the same fashion, and because we can be absolutely certain that at least eight of these sixty-four readings were set as two- or threeword forms purposely,⁵ we can safely conclude that setting

     ³ The cluster sl is not too common in Spanish. Only one hundred fifty-four readings containing this cluster, including two compositorial errors “ceƒƒô” (II, N5^r, 25) in lieu of “ceffô” (FCE, line 27796), and “desl~ubrada” (II, Q4^r, 31) in lieu of “deƒal~ubrada” (FCE, 29340)— occur in Don Quixote. The abbreviation FCE stands for the control edition of my old-spelling concordance of Don Quixote, University of British Columbia Press, Vancouver, 1988. The thirty-three readings containing the ligature ƒl appear in FCE in the following lines: 54, 190, 613, 5309, 6101, 7501, 7576, 7665, 7696, 7698, 7923, 9370, 9418, 9603, 9608, 9703, 13077, 13096 (2 occurrences), 15193, 15262, 15674, 15728, 16784, 17714, 18303, 21109, 21696, 21903, 22912, 27089, 27796 (error), and 36646. Note that items 45 and 63 of Table 1 have no ligature.
     ⁴ For a sort shortage of the ligature ƒ see FCE, Volume 1, at page xxvi.
     ⁵ FCE: 1786, 5815, 6791, 7741, 7848, 7972, 9428, 9671, 12335, 12539, 12608, 13398, 13806, 14228, 14369, 14568, 17583, 17635, 20710, 20994, 21003, 21019, 22209, 22297, 22426, 23208, 23246, 24662, 24763, 24899, 25036, 25743, 25796, 26264, 26532, 27921, 28598, 28616, 29002, 29518, 29665, 30655, 31799, 31843, 31859, 31904, 32425, 32740, 32747, 33423, 33899, 34437, 34497, 34712, 34906, 34931, 34980, 35093, 35233, 35470, 37202, 38313, 38596, and 39468. The compositors set two n's together only for the words “innumerable” and “annis,” and never set a medial lower-case s before an m, an o, an l,or a t; therefore, the readings “Quitamos ƒe | le” (I, Pl^v, 1-2), “has me” (I, Cc3^v, 7), “Viƒtes os” (1, Ee6^r, 5), “Prometimosƒelo” (I, Ff7^v, 6), “es me” (II, GB^r, 10), “quieres te” (II, Pl^r, 16), “dexen nos” (II, Y7^v, 3), and “acometierõ nos” (II, Dd4^v, 9) are clearly two- or three-word forms.

these grammatical constructions in this fashion was an accepted, though perhaps not a common, orthographic convention amongst some compositors. All two- and three-word readings of these grammatical constructions should, therefore, be retained as they occur in the first editions.
     With three differing but perfectly valid ways of setting the same word from which to choose, it would be too sanguine to expect one-hundred-per-cent consistency in how each compositor set the occurrences that appear in his gatherings. Table 2 shows that compositors did, none the less, have a strong preference for one or another form.

     Some of the seeming inconsistencies found in Table 2, disappear when one takes into account other setting habits of the compositors. When compositor I split a word between an s and the following letter at the end of a line, he always set an ƒ instead of the usual s (see, for instance, “deƒ- | hazia,” II, Aa8^r, 34-Aa8^v, 1; “tranƒ | formaron,” II, F8^v, 8). Also, four readings set by compositors E and H with ƒ's (items 2,

21, 39, and 46) contain the cluster iƒl / yƒl, which these compositors customarily set with a ligature (see: “legiƒlador,” I, ¶¶1^v, 22; “yƒla,” I, K3^r, 15; “Iƒla,” I, Ff7^v, 31; “aillados,” II, D4^v, 4). Therefore, the two occurrences with ƒ's entered in the last column of Table 2 under compositor I, and the four occurrences with the cluster iƒl / yƒl entered under compositors E and H are not inconsistencies. On the contrary, the use of ligatures and ƒ's in these six instances merely re-affirms well-established compositorial practices.
     It is obvious, from Table 2, that compositors C, D, E, I, and J preferred the one-word form with sl; that compositor H preferred the two- or three-word form, but when a reading had the cluster iƒl he set it with a ligature; and, lastly, that the only workman who preferred the one-word form with the ligature ƒl was compositor F.⁶
     Given the distinct preferences of the compositors for one form over another, one probably should, when producing an old-spelling edition of Don Quixote, edit items 10, 32, 45, 63, and four other similarly made-up readings (one of the readings was run together and three appear set in two lines), as follows: “mandarnos | lo” > “mandarnoflo” (compositor F; FCE, 7247), “Quitamos ƒe | le” > “Quitamos ƒe le” (compositor F; FCE, 7741), “contaros | lo” > “contaroslo” (compositor C; FCE, 16717), “vereiƒ- | lo” > “vereislo” (compositor I; FCE, 26897), “Prometimosƒelo” > “Prometimos ƒe lo ” (compositor E; FCE, 15168), “ƒeñalen | ƒe le” > “ƒeñalen ƒe le” (compositor I; FCE, 27921), “traeƒ- | los” > “traeslos” (compositor I; FCE, 38350), and “boluio | ƒe la” > “boluio ƒe la” (compositor I; FCE, 38596).
     It may seem unnecessary to spend so much time and effort deciding whether to enter the first edition reading “mandarnos | lo” as “mandarnoƒlo,” “mandarnoslo,” or “mandarnos lo,” but what is

     ⁶ For the possible identity of compositor F see my article “The Compositors of the First Edition of Don Quixote, Part II,” footnote 27, at page 43. Compositor F's use of the ligature ƒl over the two-type sl form is a strong compositorial preference. Only two of the seventeen readings containing the cluster sl set by him appear without the ligature. But even these two departures from his common practice cannot be considered inconsistencies. Both readings occur in Part II, and both are capitalized occurrences of the word “Isla” (II, A2^r, 4 and A6^r, 15). The case compositor F was using had the ƒl sort, because he set the reading “legiƒlador” (II, Al^v, 19) with a ligature; hence, we can conclude that he either had changed his setting habit or, more likely, that he did not like the looks of the cluster Iƒl (the only other occurrence of the same word set by compositor F appears with a lower-case initial and an ƒl-ligature; “iƒla,” I, Ggl^r, 4). It is worth noting, too, that whilst setting the second Madrid edition of Don Quixote, Part I (1605), compositor F changed some occurrences of sl to ƒl; see The Compositors, item 179 of Table 14, at page 84.

really in question here is, of course, not what the most appropriate form of one isolated reading is. The stakes are much higher.
     The printer's copies from which the first editions of Don Quixote were set, Cervantes's own manuscripts, are no longer extant. The characteristics of the authorial orthography and style are unknown, and lie hidden underneath the particularities of the differing spelling preferences and setting habits of several compositors. In other words, because we can reach at Cervantes only through the compositors, no bibliographical, typographical, or textual peculiarity of the first editions can be glossed over or dismissed without careful consideration. No error, variant, or puzzling reading is unimportant. Before attempting to establish Cervantes's intended text, one must familiarize onself with the spelling preferences and setting habits of the compositors, and produce an edition which is consistent with these characteristics.
     The deductive process I have followed for deciding how to edit the reading “mandarnos | lo” may, perhaps, be of some interest, but what one learns about the setting habits of the compositors, and the realization that only one compositor had a marked preference for using the ligature ƒl rather than two separate types, take us beyond Don Quixote and Cervantes.
     The Madrigal-Cuesta press was an active and important printing house during the first quarter of the seventeenth century. Compositor F's setting idiosyncracy, like compositor E's acceptance of the authorial spelling “vuo” (The Compositors) or compositor H's characteristic spelling of the word “sê” (“More on the Compositors”), could be used as the means to unravel the distribution of labour in many another work printed at the Madrigal-Cuesta press in those auspicious years. Only when seen under this light can one do full justice to the relevance of otherwise apparently unimportant loose pieces of biblio-analytical evidence.
     The importance of having complete, accurate, and accessible records of all occurrences of all words found in Don Quixote cannot be overstated. This study, like every one of my previous monographs on the first editions of Cervantes's works, would not have been possible without my having these records at hand.