University of Virginia Library

The Printing of Calderon's Tercera Parte
by
D. W. Cruickshank [*]

This paper was originally a chapter in the introduction to a critical edition of Calderón de la Barca's play En la vida todo es verdad y todo mentira, but it became too long and too wide in its scope to fulfill its original role. It is in some respects a continuation of a study by Professor E. M. Wilson, "On the Tercera Parte of Calderón".[1] I shall therefore be brief in my introduction.

The autograph manuscript of En la vida survives. The play was first printed in 1664, when it appeared as the first of the twelve plays in the first edition of Calderón's Tercera Parte (Excelmo). In this edition the play was, at best, printed from a very bad copy of the autograph manuscript, and the editor, a certain Vergara Salcedo, neglected his duties. Nine copies of the edition survive: I was able to collate the En la vida texts of eight of them on the British Museum's Hinman machine.[2] I also examined the apparently unique copy of Excelentissimo, the second edition, which also bears the date 1664.[3] The name of the printer of both editions is Domingo García Morrás. I shall refer to the play by the autograph manuscript title of En la vida, although all printed versions call it En esta vida.

The format of the first edition is quarto in eights. There are thus four formes in each gathering. Until the order of printing is established, I shall refer to formes by gathering, sheet, and side of sheet, in that order (by "side of sheet" I mean inner or outer). Thus the outer forme of the inner sheet of the first gathering is A (i,o).

Four formes can be printed in twenty-four different orders, i.e. factorial four. However, it is possible to tell by optical means whether the inner or outer formes were printed first. The first forme side of a sheet is composed of indentations where the type from the other side does not coincide with that of the first; where the type does coincide, the print of the first forme

is raised and pushed out by that of the second. One must beware of irregularities in the paper.

Many of these tell-tale indentations would be lost if a book was pressed. I suspect that this happened to the B.M. and E.M.W. copies, which were rebound last century. The U.L.C. copy, however, is still in the original limp vellum binding and the indentations have not disappeared. Some of these can be seen without artificial aids. For example, there are three printer's ornaments on 26v, at the end of En la vida, the first play; 26v is in the outer forme of the outer sheet. If one looks at the recto of the leaf, one can see protrusions caused by the ornaments coming through. These protrusions include the print on 26r. Where 26v is blank, e.g. between text and ornaments, 26r has only indentations. Clearly 26r, belonging to the inner forme, was printed first.

Some identifications can be made with the naked eye, but others have to be made with a lens and a collimated light, which throws the embossed and indented portions into sharper relief. Examinations made in this way showed that over 94 per cent of the inner formes in the book were printed before the corresponding outers.[4] The outer sheets of gatherings I and LI had their outers printed first, as did the inner sheet of gathering X. I suspect that the outer sheet of Hh also had its outer printed first, although the evidence is not certain. I shall return to these exceptions: the main thing is that they do not affect the principle. The printing of the inner formes first cuts the possible orders to six. In an attempt to narrow the possibilities further, I examined individual pieces of type.

Since the printers would normally be machining one forme while the compositors were preparing the following one, the same type cannot recur in consecutive formes, unless the compositors and pressmen interrupted their work to set up and print something else. This happened in the Shakespeare First Folio, as Prof. Hinman has shown.[5] In my search for distinctive pieces of type I used the Hinman collating machine. This had obvious advantages in that two copies could be examined simultaneously. The Tercera Parte is carelessly printed on poor quality paper, so that the same types often look different in different copies of the book: so it is essential to use at least two copies of the book when looking for distinctive types.

En la vida occupies only the first 26 leaves of the Parte. I examined all of the first four gatherings, i.e. 32 leaves, which include part of the second play, El maestro de danzar. I found over 100 distinctive types recurring in these pages, but rejected over 40 because they were not sufficiently distinctive or because they occurred very rarely. By "distinctive types" I mean damaged ones, marked by being dropped and trodden on, or by having a

hard object squeezed into them by the platen during printing. The type used to print the Parte was not new. Few pieces were seriously damaged when printing began, but many were nicked and cracked. Most of the capital 'Q's had lost part of their long tails.

As I said above, there are six ways of printing a gathering in which the inner formes are done before the outers: four of these involve doing the two inner formes first [1: A(i,i), A(o,i), A(i,o), A(o,o); 2: A(i,i), A(o,i), A(o,o), A(i,o); 3: A(o,i), A(i,i), A(i,o), A(o,o); 4: A(o,i), A(i,i), A(o,o), A(i,o)]. There are two ways of printing the inner formes alternately with the outer formes [5: A(i,i), A(i,o), A(o,i), A(o,o); 6: A(o,i), A(o,o), A(i,i), A(i,o)]. I found ten damaged types appearing in formes A(i,i) and A(o,i), i.e., in both the inner formes. Thus the first four possible orders must be rejected (ignoring for the moment the possibility that work was interrupted). I also found that eight types occurred in formes A(o,o) and A(i,i), so that order number 6 should be rejected. I tested the remaining possibility in all 136 formes of the book, and on only one occasion found the same piece of type occurring in consecutive formes. This was in formes 60 and 61, and indicates an interruption or a temporary change in the order of printing. En la vida is not involved.

When the book was printed, therefore, the inner sheet was perfected first, and in both sheets the inner forme was printed before the outer; i.e., the compositors began in the middle of the gathering and worked outwards. Obviously a good deal of casting off was necessary, since 3v was the earliest page that could have been set up. The compositors had to cast off five pages in every gathering before they could begin it. This was not as hard as it sounds because they were dealing with lines of verse. The method of working out from the centre of a gathering was that used by the compositors of the First Folio.

According to Prof. Hinman, one compositor could reasonably be expected to set 11,000 ens in a twelve-hour day.[6] Since one forme of the First Folio comprises about 10,600 ens, Hinman suggested that one compositor could comfortably set a forme of it in a day. However, the accounts of the working of the Cambridge University Press in the years 1696-1712 show that 10,600 was an exceptional figure: only one compositor attained it for any length of time.[7] Dr. McKenzie found that 6000 ens a day was a good average: see the table in Vol. I, p. 120. The most striking thing shown by this table is the great inconsistency in speeds of setting. Now one forme of the Tercera Parte contains 9000 ens. One compositor working at 6000 ens a day would take a day and a half to set this. On the other hand, there are very few problems of justification and the like in the Parte. The compositor might well have managed to set one forme a day.

Lack of information about the printing-house of García Morrás prevents

any further calculation here. McKenzie found that "it was unusual for a compositor to work for any long period on one book to the exclusion of all others — normally he would be setting type for two or three concurrently".[8]

Unless we discover account-books for García Morrás's printing-house, guesses about the size of the Tercera Parte edition will remain guesses.

However, we can discover something about the manner in which the Parte was printed, particularly about the number of compositors involved. Some evidence can be obtained from an examination of damaged types, and so I turned to these again. Since one compositor is essential, and there is no evidence for any more at present, I shall refer to "the compositor".

The first forme to be set up was A(i,i), pages 3v, 4r, 5v, 6r, but not necessarily in that order; there is no way of discovering anything about page order from a forme taken in isolation. No doubt a single compositor would usually set up 3v first and 6r last (see below, however), but if two compositors were working simultaneously, 3v and 4r might have been set up first together. Obviously it would help if we could distinguish a 3v and 4r set up consecutively from a 3v and 4r set up simultaneously.

The second forme to be set up was A(i,o), or pages 3r, 4v, 5r, 6v, which perfects the first sheet. A(i,o) was no doubt set up while A(i,i) was printing or waiting to be printed: the two formes certainly have no distinguishable types in common. The third forme was A(o,i), or pages 1v, 2r, 7v, 8r. As I said above, ten damaged types occurring in A(o,i) are also found in A(i,i).

At this stage I began to divide formes into pages, to see if the recurring types always recurred in the same place in the forme. Pages are described as top left (tl), top right (tr), bottom left (bl) and bottom right (br). In these terms it is assumed that the forme is sitting with the signature in the bottom left-hand corner, that is, in the position for printing the signature at the bottom right-hand corner of the sheet. When I refer to pages, I do not here mean the impression on the sheet of paper but the metal type of the forme.

The order of the formes having been established, I shall refer to them as 1, 2, 3, etc. In this way folio 1v, which occupies the top left-hand quarter of forme 3, is referred to as 3 tl. Of these two symbols, the forme number 3 is the most important; the tl will, I hope, help the reader to visualize the position of the page in the forme. I shall usually refer to pages by their folio numbers as well.

Forme 3 — A(o,i), 1v, 2r, 7v, 8r — contains one identifiable type from 1 tl (3v), four identifiable types from 1 bl (4r), one from 1 br (5v), and four from 1 tr (6r). These totals do not include running-heads. I have listed identifiable running-heads, but do not count them among the types. This is because they ought, in theory, to remain standing in the chase, or skeleton.

When I refer to "recto-head", I mean the De Don Pedro Calderon de la Barca which appears on the recto of the folios. By "verso-head" I mean the titles of the individual plays, which appear on the verso. The list of types and running-heads is on pp. 245-247.

Pages 7v and 8r have no identifiable types from forme 1, although these two pages have eight identifiable types between them, which appear in later formes. There is a simple explanation: 7v and 8r were set first. In the course of setting up formes 1 and 2, the compositor had cast off as far as, and including, 6v. Thus he had to cast off only 7r before he could set up 7v and 8r. Since he had probably turned to the equivalent of 6v (the last page of forme 2) in his manuscript, I imagine he went straight on from there, and set 1v and 2r last. By the time he had done two pages, forme 1 was back from the press, and he used type from it to finish forme 3: the ten types from forme 1 all occur in 1v and 2r. There was nothing to be lost from this procedure, or to be gained.

Verso-head III remained in the top left-hand page of the forme, and the two other identifiable running-heads also remained in their positions. The recto-heads of 6r and 8r cannot be identified positively, but no doubt the same one was used. This means that two chases were being employed at this stage of the work: if only one chase had been employed, we should expect to find the running-heads also used in forme 2. Thus the four pages of forme 3, two of them composed of type from forme 1, were put into the chase previously occupied by forme 1; while this was being done, forme 2, in a different chase, was being machined. All this is quite straightforward.

We might have expected forme 4 (A[o,o], 1r, 2v, 7r, 8v) to be made up of the types of forme 2: it is not. There are few distinctive types in forme 2, and those there are do not appear until forme 5. We could reasonably have expected to find left-overs from forme 1, which was distributed for only half of forme 3: these are all we find.

None of forme 1's positively identifiable types occurs in folio 1r, but I do not think any conclusions can be derived from this fact: this page is half occupied by the title of the play and the list of characters. In any case, I have noticed one type which may be the same: an 'n' with a nick at the top right-hand corner, which occurs in 1r (col. 2, line 1, 3rd letter) and in 4r (col. 2, line 24, 3rd letter). There may be others I have missed: type-watching is not an exact science.

The absence in forme 4 of types from forme 2 suggests that forme 2 had not been distributed when 4 was set. This is confirmed by the running-heads. The chase and running-heads of forme 1 had been used for forme 3, of course, so that they were not available. Instead of using those of forme 2, however, the compositor made up three new running-heads (1r does not need one) and so presumably used a third chase. Such running-heads in forme 2 as can be identified do not appear until forme 5. Thus forme 2 may have been in the press while forme 4 was being set up; otherwise the

compositor would surely have taken its chase and running-heads, leaving the pages to be distributed at his convenience. Thus the compositor had set four formes and distributed one before the press had machined two.

Had the press been fully manned, this would have produced an edition of several thousands, unless other matter was being machined or the Parte was being set by two or more men working simultaneously. Dr McKenzie points out that "if two compositors were working simultaneously on the same sheet, each setting one forme, they would have had to set from separate cases;"[9] Both the formes of the second sheet of the Parte (3 and 4) contain type from forme 1, therefore they were not set by different compositors, at least not in units of a forme.

Forme 5 (B[i,i], 11v, 12r, 13v, 14r) is set from formes 2 and 3, but only part of it from the latter. I do not think there is any significance in the scarcity of types from three of the pages of forme 2 (4v, 5r, 6v); two of the pages of forme 5 have only one column of type. There are ten identifiable types in forme 2, one of which is on 4v and does not reappear until forme 8; it is 'q I', quite distinctive, and I have looked carefully and in vain for it in forme 5. Forme 8 is composed mainly of types from forme 5, and this jump from 2 to 8 is almost the only piece of evidence I have to suggest that more than one case of type was used. However, one type is not enough. The movement of the other types so far examined suggests that only one case was used.

I saw no types from forme 2 in 14r; yet there are four from forme 3. The most likely explanation is that the compositor distributed forme 2 in order to set forme 5, of which he then did three pages. He became short of type and distributed 8r and at least some of 1v, from forme 3, in order to set 14r, the highest-numbered page in forme 5. An examination of the make-up of forme 6 supports this. Forme 6 contains thirteen identifiable types from forme 3, and only one of these is from 8r.

In forme 5, then, the last page to be set up was the one with the highest number, that in the top right quarter of the forme. This is of little significance: what is important is that the page was almost certainly set up last on its own. Evidence from the first four formes is against simultaneous setting by formes. The evidence here shows that two compositors did not share a single forme by setting two pages simultaneously; if they had done so, using the same type, we would have found types from forme 3 in another of forme 5's pages. Thus if two compositors were working together, they were doing one column of a page each: even more inconvenient, since it would entail casting off by columns as well as by pages. A glance at the running-heads shows that the recto-heads kept their places, whereas the verso-heads changed round (br to tl, and tl to br).

There is nothing unusual about forme 6 (B[i,o], 11r, 12v, 13r, 14v), of which all the identifiable types come from forme 3. I tried to establish

patterns in this forme, without success. 11r, at the bottom left, has one distinguishable type from 2r, also at the bottom left; 12v, at the top left, has three types from its top-left counterpart, 1v. But 11r has one type from 8r (top right) and one from 7v (bottom right); 12v has one from 7v also, and so on with the other pages.

I tried next to see if there was any consistency in the division of types between columns. 'ç' moved from column one on 1v to the single column part of 12v; 'g I' moved from the same place to the first column of the dual-columned part of 12v; and although 'g III' also came from a first column, it entered the second column of 12v. 'C' moved from a first column to a second, as did 'd III' and 'g II'. 'm II', 'q III' and 'v II' all remained in first or second columns. Of the three types in the only column of 11r, one (N II) came from a first column, the others from a second column. Thus we have something of everything — and evidence that if two compositors (or more) were working, they did not trouble to keep their type separate, whether they worked by the page or by the column. This would have been an abnormal procedure.

As for the running-heads, those from forme 3 were used, in the same position in the chase. I cannot say for sure that those of 8r and 13r are the same, but I do not see why they should not be so.

There are more complications with the type of forme 7 (B[o,i], 9v, 10r, 15v, 16r); it seems certain that the whole of forme 4 was distributed first, however. The compositor presumably ran out of certain kinds of letter after setting up the first two pages, and distributed half of forme 5 (11v and 12r). Six types from these pages occur in forme 7, and none at all from the other two pages (13v and 14r). In forme 8, on the other hand, there are no less than twelve types from 13v and 14r, and none from 11v and 12r. This seems to confirm the likelihood that only half of forme 5 was distributed for forme 7.

The distribution of two pages in order to set two more might indicate two compositors working simultaneously, were it not that the types are scattered: at least one type from each source-page appears in the two pages set. And when the six types reach forme 10, they are again scattered without any pattern.

The peculiarities of forme 7 involve the running-heads. The compositor could obviously choose between the chase and heads of forme 4 and the chase and heads of forme 5. He used two heads from the chase of forme 5 (13v and 14r to 9v and 16r), but he certainly did not use the other two, for they were used in 9r and 10v of forme 8.

Forme 4, it will be remembered, was the one for which the three new running-heads were made, and which was put in a new chase; obviously this chase could not be used as it was, for it lacked a running-head on the bottom left page (1r). The bottom left page in forme 7 is 10r, but the recto-head from forme 5 does not fill this empty space. It appears instead in the

top right corner (16r), where there already was a running-head. The running-head which was there (i.e. 7r) has no identifiable types in it, but it appears to have gone to the bottom left, i.e. to 10r. As for the two verso-heads of forme 4, neither appears to have been used for the second verso-head of 7 (15v) — unless the one fom 8v was re-set such that the 'v's of vida and verdad changed places. I am at a loss to explain this. Professor Bowers records a case of two running-heads vanishing when a change was made from one chase to two,[10] but this cannot explain the present problem.

Forme 8 (B[o,o], 9r, 10v, 15r, 16v) has types from four different formes, although its main source is forme 5. The case of 'q I' has already been mentioned. In page 15r of this forme we have a similar example, 'N I'. Page 7v (3 br) was not distributed to set up forme 5, but it was certainly used for forme 6. 'N I' missed this forme, which is not really unusual; it is a little more unusual that it should have missed forme 7 as well. However, 'N I' cannot be said to suggest that more than one case was being used, as might have been implied had it not re-appeared until forme 9.

The 'v I' and 'u I' which were previously used in forme 4 are almost certainly left-overs from the distribution made for forme 7. Their appearance in forme 8 instead of in 10 suggests that there was only one case of type. The lack of types from 11v and 12r (5 tl and bl) suggests that these two pages had already been distributed and used, as I said above.

The running-heads of forme 8 follow the now established pattern: two from forme 5, two from forme 6, although I cannot identify positively the running-heads of 13r and 15r. However, the p of Pedro appears to have a small defect where the bottom of the loop joins the stem. Probably the chase previously containing forme 5 was used, since the running-heads from 5 do not change their positions. Those from 6 do. The whole procedure seems very inefficient.

There appears to be nothing unusual about forme 9 (C[i,i], 19v, 20r, 21v, 22r). A left-over 'n' from forme 5 provides still more evidence for the use of only one case of type. 'n' could not have arrived at forme 9 via forme 7, for the page it occupied in forme 5 (14r) was not distributed for 7. The movement of the other types does not seem to require comment.

The two running-heads from the chase of forme 6 have been reversed: perhaps the chase was turned round. This time it appears that only one running-head was taken from forme 7: verso-head IV. The recto-head on 16r found its way into forme 10, and in those of 10r (7) and 21r (10) the l of la prints badly, so that they are probably the same. Where the recto-head of 20r comes from I do not know.

The make-up of forme 10 (C[i,o], 19r, 20v, 21r, 22v) is apparently quite straightforward: the forme contains twenty types from forme 7. There are fewest from 9v and 10r, no doubt because all of 10r and most of 9v have

only one column of type. Most of forme 10 goes to make up forme 13, but two of the types, 'd II' and 'Q I', find their way into formes 14 and 15 respectively; the latter is definitely not in forme 12. Thus the case which was used to set 10 was also used for 13, 14 and 15.

The strange movement of the running-heads goes on, with the origins of those of 21r and 22v obscure. Perhaps, as I suggested above, 21r comes from 10r (forme 7). That on 22v has an inverted m, a sign that it was a fresh composition, or proof that it was at some time out of its chase. I think it may be the one previously found in forme 7 (15v). Fortunately nothing important depends on this information.

The first column of 20v was corrected; in line 31 the B.N.P. copy and one of the Vatican copies (V.K.) give "Focas viua." The last word is an adjective, and all the other copies examined give viuo. The other pages of the forme have no corrections. One wonders, in view of the many variants from the autograph manuscript in these pages, if the corrector looked at them; if he did, and passed them, his version of the text was very corrupt. Folio 19r, for example, has nearly thirty variants from the autograph manuscript, at least two of which could have been corrected without an original: the abbreviation Ssb. for Sabañón in column 1, line 21; and the stage direction Vase., twice attributed to Astolfo (col. 1, lines 32 and 37) when it should have been attributed once each to him and to Lisipo. Errors such as these suggest that not all four pages of the forme were corrected, or at least that they were corrected very carelessly.

Forme 11 (C[o,i], 17v, 18r, 23v, 24r) was also corrected, but this time the uncorrected version is found in B.N.P. and E.M.W. Formes 10 and 11 are on separate sheets, i.e. forme 11 does not perfect 10. Thus the fact that E.M.W. and V.K. have only one of the two uncorrected formes indicates that the sheets were probably not bound in the order in which they were machined: i.e., that the tenth book (say) to be bound was not composed of the tenth impression of every sheet. (This assumes that when forme 9 was printed, the pile of sheets was turned, so that they were perfected in the order in which they were first impressed: this would be normal). A forme was probably corrected early, while few copies had been printed from it, but formes would probably not always be corrected at the same stage, e.g., after ten copies had been printed. If, say, there were only ten uncorrected copies of forme 10 and fifteen of forme 11, and the sheets were bound in the order in which they were printed, there would be five volumes containing only the uncorrected forme 11, but ten containing both uncorrected formes. This did not happen: so no one book will contain all uncorrected formes. This probably means that other tell-tale uncorrected formes were scattered through long-since lost copies of the Tercera Parte. If, on the other hand, Vergara Salcedo decided to fulfill some of his obligations as editor, and paid occasional visits to the press to supervise correcting, many incorrect copies of a forme might have been printed, and corrected formes might be few and far between.

Only three pages (17v, 18r, 24r) of forme 11 were altered in the correcting; there was one change in 17v, five changes in 18r and one in 24r, a total of eight for the two formes. In 23v the verso-head was moved an eighth of an inch — to the left in the forme, to the right in the printed sheet; this was probably done accidentally when the wedges were removed for correction. Once again, to judge from the number of errors remaining, the corrector's copy was very corrupt; yet correction was done from a copy, for one of the corrections could not otherwise have been made. This is the changing of decidido to dicidido (fol. 18r, 1, 25). Calderón wrote dicidido, but decidido was an equally correct and possibly more common form in the mid-seventeenth century.

The verso-head from 8 tl (10v) was used in forme 10. No doubt this was why the compositor used the verso-head from 9 tl (19v) in forme 11. He was still going back three formes to get the bulk of the type for the one he was setting, but the press was not falling behind, for it was completing the machining of the forme two earlier than the one he was working on.

Forme 12 (C[o,o], 17r, 18v, 23r, 24v) appears to be composed of forme 9, some residue from 8, and at least one running-head from 10: where the others come from I do not know. No types from the top two pages of forme 9 (19v and 22r) were observed in the two lowest-numbered pages of 12. This could mean that the bottom pages were distributed first and that the top pages were not distributed until two pages of the new forme had been set. I am a little troubled by 'N III', which went from 19r (10 bl) to 23r. Most of the type from 19r seems to appear in forme 13. However, 'N III' occurs at the end of page 23r (column 2, line 29); the compositor may have been running short of 'N's, and may have taken it from forme 10 without actually distributing any type.

The same has happened in forme 13 (D[i,i], 27v, 28r, 29v, 30r) as in forme 5: in his first three pages the compositor used seventeen identifiable types from forme 10, quite evenly scattered (7, 4, 6), and in his fourth page only one. The fourth page also has two types (O II, ſt I) from a page of a new forme, forme 11: clearly it was set last, on its own, after the other three. Simultaneous composition of two pages is ruled out. Indeed, since both of forme 11's types occur in the second column of 30r, the evidence is against simultaneous composition of two columns.

I do not know what movement took place among the running-heads, but old ones were being distributed and new ones set up. Thus some of the identifiable ones disappear and the new ones cannot be traced.

There are not many recognizable types in forme 14 (D[i,o], 27r, 28v, 29r, 30v), so that reliable information cannot be obtained. The shortage of types in 14 from pages 17v and 24r of forme 11 may indicate that part of these pages was used for forme 13: at least some of the former must have been used. The left-over 'd' from forme 10 has already been mentioned.

Little comment on the last two formes is necessary; 15 (D[o,i], 25v, 26r,

31v, 32r) does not appear to use any types from 13, nor 16 (D[o,o], 25r, 26v, 31r, 32v) any from 14. The identifiable running-heads used in the earlier formes have now almost completely vanished, although I have traced recto-head I from 15 to 17, 19, 22, 24, 26, 28. Running-heads are a problem in themselves, and the problem is probably aggravated in En la vida by re-setting after accidental breaking-up.

There is no doubt, however, that the whole book was printed with the same fount of type. Many of the distinctive types occur all the way through and even in the preliminaries. The heading FAMOSA COMEDIA is used for all twelve plays, as we can tell from the badly-scarred E. The M of COMEDIA is the same M as in the COMEDIAS of the title-page — witness the left half of the serif at the foot of the left leg. There is no pattern in the movement of type which might indicate the use of several cases of type kept separate. If more than one case was used, no attempt was made to keep their types separate. There is no evidence of simultaneous composition of two formes, pages or columns in the first sixteen formes of the book; on the contrary, the evidence does not admit of simultaneous composition. In any case, simultaneous composition of a long book from manuscripts is not common: it is much more common in a page-for-page reprint.

During my examination of the broken types, I also looked for differences in spelling, punctuation, setting-up, etc. I found none, either in the first play or in the whole Parte. Instead I found definite indications that the same man was at work throughout the book.

I attempted to confirm these early findings from data on page-widths. The measure used in two-columned pages was of approximately 59 mm, and the columns were separated by a vertical spacer about 4 mm broad, making a total width of 122 mm, the same as in the prose pages of the preliminaries.

I say approximately 59 mm: I am satisfied that a 59 mm opening was used throughout gatherings A — D, those I have been dealing with in this study. Measurements made in later gatherings showed that 59 mm was used for most of the book. Towards the end, however, the opening appeared to widen to 60 and then to 61 mm. The spelling remains the same, and the same damaged types occur in these gatherings. This kind of evidence is not helpful, as McKenzie has shown: "Not only do the widths of pages set by the same compositor vary considerably, but different compositors are often found setting to an identical measure . . . it is evident that differences of even 1 — 2 mm cannot be relied upon to indicate the end of one compositorial stint and the beginning of another".[11] There is certainly no evidence in Parte page-widths for two or more compositors, whether they worked simultaneously or consecutively.

I next tried to obtain evidence from the dates of the book's various preliminaries. Vergara Salcedo's letter to Calderón is dated August 2, 1664. The plays, if not the preliminaries, had presumably been printed by then. They had certainly been printed a week later, for the Fee de Erratas and the Tassa are both dated August 9 (The tassa is simply the price, which was based on the number of sheets in the book). Calderón's letter to the Marqués de Astorga has no date. The aprobación of Dr. Manuel Mollinedo y Angulo, who judges that the book is very worthy of being published and given to the press, is dated June 15. The licencia of the Licentiate García de Velasco, who gives the permission to print, is dated June 17. Finally, the long-winded aprobación of the Licentiate Tomás de Oña, who describes the book as very worthy of being printed, is dated July 2.

These dates appear to indicate that the book was printed between July 2 and August 2. July 2, 1664, was a Wednesday, so that the period from July 2 to August 2, inclusive, contains 28 working days, without making allowances for fiestas or the heat in Madrid at that time of year. The Tercera Parte has 136 formes without the preliminaries, which our single compositor, even working at the rate of one forme a day, would take four and a half months to set up. Thus to set up the book in 28 days would have needed four compositors working more than twelve hours a day, or five compositors working at a more human speed.

I can see only two possible explanations for this: first, that the book was approved after it was printed, whether from the manuscript or from a printed copy. Secondly, that the manuscript was divided into four or five lots, as might be done for a page-for-page reprint, and that these were set simultaneously. This was the only possibility I had not fully investigated, since it involved more than the first sixteen formes. For example, one compositor might have begun at forme 1, another at forme 34, another at forme 68 and a fourth at forme 102. This would have entailed casting-off three-fourths of the book before starting: unlikely but not impossible, since a large percentage would have had to be cast off anyway.

Four or five compositors dividing a large book between them would use separate cases of type, so I examined all 136 formes of the book for the distinctive types I had found in the first 16. I found at least one recognizable type in every forme. For example, I noticed the very distinctive 'g II' in 99 (Bb1v, 2, 19), 123 (Hh2r, 1, 28), 126 (Ii5r, 2, 30), and 134 (Ll5r, 1, 30). Although distinctive, 'g II' is not easily seen. Other distinctive types were seen to occur regularly, every two or three formes. Perhaps printing stopped temporarily after forme 60 and all the type was distributed, so that type from 60 occurred in 61. The more noticeably deformed letters vanished gradually, although others, no doubt freshly damaged, kept appearing. Obviously attempts were made to keep the stock of type in good condition.

This evidence confirms that the same type was used throughout the book, including the gatherings where the measure was more than 59 mm.

During the examination I noticed two lines which reminded me of others in En la vida. One was in También hay duelo en las damas, the last play: Quien las noches de vn Ibierno (Ii1v, 1, 38). This is an uncommon (though not unique) spelling for invierno or ymbierno, unless it is misprint. I do not think it is a misprint, for it also occurs in En la vida: Ibiernos la elada yerua (A2r, 2, 14).

The second line reads, in En la vida: para que buelta la aguja, (C8r, 1, 11). We know from the autograph manuscript that aguja is an error for quilla, and that Calderón is therefore remembering a passage in Góngora's Soledad Segunda (line 548). In Ni amor se libra de amor, the eighth play, we find the following line: al cielo la aguja buelta, (Z1r, 1, 25). The sense here calls for quilla, not aguja, as later editors realized. (They did not realize it in En la vida, interestingly enough). I am not quite sure what to make of the two aguja errors. If they are the result of a coincidence, the coincidence is a surprising one. Perhaps they are the result of a compositor making the same error twice, or making the second because the first had already been made; possibly Vergara Salcedo was responsible. I am more certain about the Ibierno; this, I think, suggests that one compositor was involved in setting the two plays concerned.

Thus the editor and printers anticipated that official approval would be given, and printed either all or part of the book before permission to print was granted. There was little chance that twelve old Calderón plays which had already been granted performing licenses would be refused printing permission.[12] I do not see what could be gained by this, unless García Morrás simply began printing when it was most convenient for him. Possibly the censors actually approved the printed book. Thus we are left with evidence that admits of only one compositor.

Had the press been printing only the work of this compositor, the edition would have been quite large: as many impressions as could be made in a day, assuming that the press was not under-manned (i.e., half-press working), and allowing for the fact that the compositor might have worked more slowly than one forme a day: say 1500-2500. This guess seems too many, although it bears comparison with the size of other Calderón editions printed in Spain about this time. These are the 1750 copies of the auto Las espigas de Ruth, seized by the authorities in 1703 and never published,[13] and the 3025 sets of Pando y Mier autos of 1717-18.[14]

Calderón's autos were tremendously popular, however, and we are dealing here with editions published forty and fifty years after the Tercera Parte. Besides, the Tercera Parte appears to have gone into a second edition in the same year as the first. There is also an alternative printing method to that which might have produced 1500-2500 copies.

Perhaps while our compositor was setting the Parte, other compositors were setting another book or books; these would be machined at the same time. If this happened, as it probably did, we cannot make an accurate guess at the size of the Parte edition. If there were two compositors, it would naturally be unnecessary to machine the work of both in equal numbers: the important thing would be to machine the two formes for as long as it took the compositors to set one each. Let us take an example.

Suppose that B, our compositor, and A, a work-mate, start together and each finish setting one forme. The press takes A's work and machines it while A and B set half their second forme. Then the press machines B's first forme until A and B have completed their second forme. Thus B is just beginning his third forme by the time the press has finished his first. And if A is a faster worker, or began ahead of B, or if the press does more copies of B's work, B will be well into his third forme before his first comes back from the press: which is what seems to have happened, as we can see from the make-up of forme 3 (see above, p. 230). After all, although a modest firm would probably have only one press, it would surely have more than one compositor; and if only one compositor set the Parte, the others must have been working on something else. There was no need to rush out the first edition by putting several men to work on it, for several of the plays were at least a decade old, and none was brand new.

The way to prove this would be to find a book that García Morrás printed at exactly the same time as the Parte. If it was printed with different type and perhaps had different spelling, we could conclude that another compositor worked on it from another case, and the hypothesis would be proved. There is another book printed by García Morrás in the summer of 1664: but examination showed that it was set by the compositor of the Parte, using the same type. I shall discuss this and other books presently. I also examined the second edition of the Parte, which purports to have been printed by García Morrás in the same year.

Professor Wilson had already remarked that some of the gatherings of this book, viz., the preliminaries, A-F and Cc-Ff are printed with double-dotted 'j's.[15] I was able to confirm this difference by noticing that the 'p' used in these gatherings has a small bulb set rather high on the stem. This is a total of 11 gatherings out of 35, counting the preliminaries in each total.

While I was looking through the rest of the book, I noticed that there were two other distinguishable founts. In one of them the 'p's have large

bulbs and the 'g's have their loops further apart than usual. This fount was used to print the twelve gatherings following the first six. The third fount also has 'p's with large bulbs, but these bulbs and the bulbs of the 'q' are too high above the line — in fact, the letters as a whole are too high on the body of the type. In addition, the 'g' of this fount is different from the 'g's in the other two founts.

The whole picture is as follows: ¶, A—F: fount 1; G — S: fount 2; T — Bb: fount 3; Cc — Ff: fount 1; Gg — Ll: fount 3. Thus there are eleven gatherings in fount 1 and twelve each in founts 2 and 3. One cannot divide 35 gatherings into three parts any more accurately.

There seem to be two possible explanations for this state of affairs. These have to take into account the fact that none of the founts is the same as that used in the first edition. Thus none of the first edition's distinctive types is to be found in the second edition. Furthermore, the spelling used in the second edition, although varying within the book, is not the same as that used by the compositor of the first.

The first possibility is that the date of 1664 is genuine and that García Morrás farmed out the whole book. The second is that the date is false and that although he printed some of the book himself, he had got rid of the old type he was using in 1664. I was able to eliminate a third possibility: that the date is genuine and that the book was set up by three of García Morrás's compositors, but not the one (nor his type) who set up the first edition on his own.

If this third possibility had been fact, we would have expected running-heads and other recurring titles such as FAMOSA COMEDIA to be the same throughout the book. They are not, although certain headings recur in the various parts. In the fount 2 section (G — S), in which four plays begin, all that was common to the four headings was left in standing type and used in these four, but not in the other eight. Thus the whole book was farmed out between three printers, or García Morrás did one third and farmed out the rest. If he did do one third, he might have done the third which contains his wood-block (the coat-of-arms on the title-page), except that this third contains the double-dotted 'j's (see below).

The division of work between three printers implies haste. There would be no need to rush out a second edition in 1670, say, but if the first edition sold out soon after printing, it might be expedient to get some more copies on the market quickly.

One of the printers was possibly Joseph Fernández de Buendía, according to Prof. Wilson; he bases this suggestion on Fernández de Buendía's frequent use of double-dotted 'j's. This seems to indicate that García Morrás lent Fernández his wood-block and that he printed none of the book himself.[16] One thing troubles me: if the book was to be divided evenly between

three printers, why was it not split into three instead of five parts of 7, 12, 7, 4 and 5 gatherings? This curious division would not make the printers' task any more difficult, of course, for they were printing a page-for-page reprint.

Fernández de Buendía printed María de Zayas's Novelas (1664) and several other books between 1669 and 1672, all with double-dotted 'j's.[17] I have found a few such 'j's in his printing of Martínez de la Puente's Historia de Carlos V (1675). All the 'j's in his 1666 printing of Escogidas XXIII are normal, however. My attempts to find the same wood-blocks used in different books (there is a P in the preliminaries of the Tercera Parte) have failed completely. However, none of this evidence conflicts with the suggestion that Fernández de Buendía printed part of the Tercera Parte in 1664, although I have another name to add to Prof. Wilson's list of those who used double-dotted 'j's: Lucas Antonio de Bedmar, in Escogidas XXXV (Madrid 1670). Every gathering has some of the 'j's, and the typographical style is different from that of Fernández de Buendía.

I have not tried to discover anything about the compositors of the various sections; that is, I do not know whether any one section had one or more compositors. The book would not have taken so long to set up the second time. Only one thing appears to be certain, unfortunately: neither the compositor of the first edition nor his type took part in the printing of the second.

Too little is known of García Morrás and his printing-house. I know of forty-nine items bearing his imprint, and since these carry dates from 1648 to 1678, his firm must have possessed some economic stability. The Parte was "done on the cheap," nevertheless. The paper is rough and lumpy (no "fine Genoa paper" here), the obvious compositorial errors frequent, and evidence of correction scarce. Vergara Salcedo, as editor, must also be blamed for this. We are unlikely to discover much more about Vergara (a few poems of his survive), but perhaps examination of other books printed by García Morrás would give us more information.

I examined seven other books printed by García Morrás and a fragment of an eighth. The earliest of these was Fabricio Pons de Castelví's Gustavo Adolfo (1648). The size of the type is different from that of the princeps of the Parte, although the spelling is apparently the same. The standard of workmanship is poor.

Next is Francisco de Vergara y Alava's Orden de Santiago (1655), a well-printed folio; the spelling here is quite different. The fragment consisted of one volume of a Latin-Spanish Virgil (actually part of the Georgics); this has the date 1660, but the Fee, Tassa and foreword all bear dates in February 1661. The spelling is the same as in the Parte: in particular, I saw two examples of the word invierno spelt with no nasal and with a capital: Ivierno on p. 251, line 7, and the same on p. 253, line 15. The type seemed new, in good condition, possibly the same fount as was later used for the Parte. The capital 'T's almost certainly came from the same matrix, for all the 'T's of the Parte have the right arm higher and thicker than the left, just as they do in the Virgil.

Next comes Juan de la Portilla's España restaurada por la Cruz (1661). Here the spelling is different from that of the Parte, and the type appears to be different from the Parte and from the Virgil: this implies another compositor with another case of type. In 1663 there is Mariana Caravajal's Navidad de Madrid: the spelling is apparently the same as in the Parte but the type is not the same size.

In 1664 García Morrás printed Desiderio del Final's Viage de la famosa villa de Madrid a la ciudad de Roma. The aprobaciones and licencias cover the period from June 24 to August 21, whereas the Fee and Tassa are dated August 22; perhaps this is another example of an already printed book being approved. The spelling is the same as that of the Parte, and (more conclusive) so is the type. In a brief examination I found 'n' (18v, l.26), 'p IV' (18v, l.27), 'E II' (20r, l.11), and the unmistakable 'g II' (26v, l.10; see lists on pp. 245-247). The compositor may have worked on this at the same time as on the Parte, or immediately afterwards: it is a short little book, only seven octavo gatherings including the preliminaries (¶8 A-E8 F8—2). If the Parte was printed by August 2, as seems possible, the Viage could have been set and printed between then and August 22.

The two remaining books are less interesting. One is the second edition of Francisco Santos's El no importa de España (1668). Spelling and type are different. The colophon describes García Morrás as "Impressor de el Estado Eclesiastico de la Corona de Castilla, y Leon." The other is Santos's La tarasca de parto (1672). It told me nothing.[18]

Perhaps it would be helpful to examine more of these books, but there is still much more information to be obtained from the Parte itself, particularly on the subject of compositorial accuracy. In my introduction to En la vida I have examined the press-corrections at more length. All of them occur in the first columns of their respective pages, so that there is no proof that the second columns were also corrected. There were 34 variants from the autograph manuscript in these columns before correction and 26 after. Thus perhaps 8 in 34, or nearly 25 per cent of Excelmo's En la vida errors occurred in the printing-house and could have been removed by presscorrection of all the formes (the many typographical errors in the other formes suggest that they were not corrected).

Until comparatively recently it was thought that the Tercera Parte represented the first edition of all twelve plays contained in it. This is not so. The second part of La hija del aire appeared in 1650 in Comedias nuevas de diferentes autores, parte XLII (Zaragoza), ascribed to Antonio Enríquez Gómez. I have compared the British Museum copy of this with the U.L.C. Excelmo text, and find that they differ in only 170 places, of which 115 can fairly be described as corrections, or attempted corrections, made in Excelmo, of the Zaragoza text.

I am quite certain that Diferentes XLII was used as the copy-text for Excelmo, for reasons which I need not give here in full: suffice it to say that the line-arrangement of the two texts is identical for the last twenty pages. The 115 "corrections" are sometimes quite ambitious but also, on occasion, obviously wrong. For example, Diferentes reads thus on p. 16, col. 2, 25-6: por deziros que el ausencia, / en mi no es madre el oluido. The second line does not make sense. The obvious correction is the simple "madre de oluido", but Excelmo reads: en mi nunca engendra el oluido. This correction cannot be what Calderón originally wrote, for it has nine syllables: nor would we expect Vergara Salcedo, himself a minor writer of verses, to make such an erroneous correction. However, a compositor working from a marked-up copy of Diferentes might easily have included the el which an editor had not properly deleted. en mi nunca engendra oluido, although probably not what Calderón wrote, is at least an apparently correct line: and the little we know of Vergara suggests that he would be likely to prefer the more pretentious version.

Some efforts were made, therefore, to correct the copy-text of this play (there are more examples which do not read like the compositor's work). Thus all twelve Parte plays may be affected by editorial mis-corrections such as the one just quoted. I have seen some possible examples in En la vida, although thankfully few. For example:

• Autograph MS, I, 15-18. Excelmo.
• venga y en ora dichosa Venga en hora dichosa,
• tanto que halle a su ovediençia tanto, que allà su obediencia,
• con siempre Rendido afecto con siempre rendido afecto,
• su patria a sus plantas puesta su patria a sus plantas vea.

The dropping of the y in the first line could have happened at any stage. The changing of halle a to allà is typical of the "close approximation" kind of error we associate with the copying of semi-legible handwriting; it suggests an intermediate stage where halle was spelt alle. The change from puesta to vea, however, is almost certainly a deliberate attempt to restore sense to the lines after the allà error had been made.

Calderón, let it be emphasized, had no part in these emendations: his letter to the patron of the book indicates that he knew nothing about it until after the printing. So we have to contend with an edition printed from copy-texts that were corrupt not only because of bad copying but because of tampering by an editor who did not consult the author.

More could have been said about La hija del aire, but Dr. G. Edwards is preparing an edition of this play. He shares my views about the Diferentes text and is using it as a basis. Prof. Wilson has already shown that would-be editors of La púrpura de la rosa, the tenth play, must consult the manuscript copy in the Bodleian Library.[19] I hope to have shown that an editor must use the Tercera Parte with great caution: it has no Calderonian authority, and some apparently correct readings may be due to tampering. No doubt the degree of corruption was different in all twelve copy-texts, but all the evidence points to a single compositor who was relatively accurate while working from a printed text (only 55 of the variants in La hija del aire II appear to be accidental), but who was probably much less accurate when working from manuscript, although this would vary with the legibility of the handwriting.

The compositor set up the book as the First Folio was set up, from the middle of the gathering outwards. If he had set in order of pages, he would have needed to set eleven pages to complete the first forme: he did not do this because sixteen different types occur twice in the first eleven pages of the book. The rare occasions in the Parte when a sheet is impressed with the outer forme first are not important. They may well have arisen because the compositor was well ahead of the press, and had set a second and third formes before the press had finished with a first; the press would then print in the order 3,2, instead of in the order 2,3, in which they had been set.[20]

We cannot make a reliable estimate of the size of the edition. Some copies of the Parte were certainly shipped to the Indies. Investigators who have examined cargo-lists in the Archivo de Indias in Seville report that "incredible numbers" of plays were shipped out in the seventeenth century,

especially plays of Calderón.[21] The rate of survival of such copies as escaped destruction on the journey was practically nil, but it would certainly be helpful to know how many copies were loaded in Seville. (To take only one example: the loading register of the bark Espíritu Santo, dated March 22 1605, included 100 copies of Don Quijote). We know, at least, that the Marqués de Astorga, to whom the book is dedicated, received a copy bound in gilt vellum; one hopes it was printed on decent paper.[22]

Certain external difficulties stand in the way of a more complete answer to the problem: it took me a week just to examine four gatherings of six copies in the collating machine. And only three copies of the Parte are in Britain, the best-furnished country. Collation of the other plays would probably reveal more press-corrections which might indicate the reliability of the compositor's copy; and the make-up of the rest of the book could be established.

The list of distinctive types follows. Here I used the same method as Hinman, i.e., a free-hand drawing on a card on which were then marked the places where the letter was seen. I refer to folio-numbers, since the first 32 folios are all correctly numbered.

Recurring Types in the Tercera Parte.

The number in brackets is the forme number; 1v, 1, 25 means folio 1 verso, column 1, line 25. If the middle number is lacking it means there is only one column on the page or part of page where the type is.

• a I: gap in stem just above spot where lower part of bulb joins it. 1r, 2, 15 (4); 16r, 1, 41 (7); 22v, 2, 37 (10); 27v, 1, 7 (13); 26v, 1, 3 (16).
• a II: notch in right of stem just above spot where lower part of bulb joins it. 14r, 1, 25 (5); 9r, 2, 15 (8); 17v, 1, 1 (11); 28v, 2, 3 (14).
• a III: lacks lower part of loop. 12r, 26 (5); 16r, 2, 12 (7); 19r, 2, 26 (10); 29v, 2, 22 (13); 25r, 1, 6 (16).
• a IV: small mark between end of hook and bulb. 21r, 2, 6 (10); 27v, 2, 19 (13); 32v, 2, 42 (16).
• b: nick in left of stem, one-third up. 11v, 33 (5); 15v, 2, 8 (7); 20v, 1, 30 (10).
• C: top arm slightly flattened. 2r, 1, 12 (3); 13r, 2, 33 (6); 19v, 1, 34 (9); 24v, 2, 28 (12); 32r, 1, 7 (15).
• c: flattened at top. 4r, 2, 34 (1); 1v, 1, 25 (3).
• ct: mark half-way up RH side of t's stem. 4v, 2, 5 (2); 11v, 7 (5); 15v, 1, 7 (7); 19r, 1, 33 (10).
• ç: two nicks at top. 4r, 1, 28 (1); 1v, 1, 16 (3); 12v, 12 (6); 20r, 2, 33 (9); 18v, 2, 16 (12).
• D: piece missing from top left corner: 16r, 2, 19 (7); 19r, 1, 16 (10); 28r, 2, 5 (13); 31r, 2, 8 (16); seen as late as forme 37.
• d I: large nick in middle left of bulb. 3v, 1, 22 (1); 2v, 2, 42 (4); 10r, 29 (7); 19r, 1, 21 (10); 27v, 1, 13 (13); 26v, 2, 30 (16); seen as late as forme 120.
• d II: dent in middle left of bulb. 3r, 2, 34 (2); 11v, 21 (5); 16r, 1, 1 (7); 19r, 1, 32 (10); 27r, 2, 1 (14); seen as late as forme 135.

---

250




• d III: two nicks in bulb, another in stem. 3v, 1, 22 (1); 2r, 1, 40 (3); 14v, 2, 22 (6); 22r, 1, 12 (9).
• d IV: dent at foot of leg. 21r, 2, 7 (10); 29v, 1, 4 (13).
• E I: end of centre-piece bent down. 6r, 1, 34 (1); 2r, 2, 12 (3); 11r, 16 (6); 20r, 2, 1 (9); 18v, 2, 40 (12); 26r, 2, 32 (15).
• E II: bottom leg bent upwards. 29v, 1, 39 (13); 32v, 2, 19 (16).
• e I: small nick just below centre left. 3v, 1, 24 (1); 2v, 2, 25 (4); 10r, 20 (7); 19r, 1, 21 (10); 27v, 2, 6 (13).
• e II: large dent in middle left. 5v, 1, 16 (1); 2v, 1, 12 (4); 10r, 28 (7); 19r, 1, 9 (10).
• e III: double bend at end of tail. 16r, 2, 37 (7); 21r, 1, 30 (10); 29v, 1, 10 (13).
• F I: small crack bottom left of stem. 3r, 2, 37 (2); 13v, 1, 28 (5); 10v, 29 (8).
• F II: notch in left of stem just above cross-piece. 13v, 2, 21 (5); 9r, 1, 7 (8); 23r, 1, 1 (12); 25v, 1, 16 (15).
• f: end of cross-piece bent up. 6v, 1, 38 (2); 13v, 1, 30 (5); 9r, 1, 21 (8); 18r, 1, 31 (11); 30v, 2, 32 (14).
• g I: piece missing from top right of upper bulb. 4r, 1, 40 (1); 1v, 1, 20 (3); 12v, 1, 18 (6); 20r, 1, 30 (9); 18v, 1, 11 (12).
• g II: two small nicks close together in top of bottom loop. 6r, 1, 39 (1); 2r, 1, 21 (3); 14v, 2, 8 (6); 19v, 1, 16 (9); 24v, 2, 32 (12); 31v, 1, 11 (15); seen as late as forme 134.
• g III: bottom loop broken upwards. 7v, 1, 17 (3); 12v, 2, 15 (6); 20r, 2, 24 (9); 26v, 1, 9 (16); seen as late as forme 65.
• g IV: notch and dent in top left of upper bulb. 12r, 16 (5); 15v, 1, 12 (7); 22v, 1, 30 (10); 29v, 2, 35 (13); 31r, 2, 4 (16).
• H: upturned serifs at tops of legs. 1r, 2, 12 (4); 15v, 2, 23 (7); 19r, 1, 25 (10); 30r, 2, 12 (13); also in forme 17 and up to forme 44.
• h: left leg bent in. 19r, 2, 23 (10); 27v, 2, 14 (13); 25r, 1, 14 (16).
• l I: horizontal nick about one-third up. 20r, 1, 38 (9); 17r, 2, 14 (12); 31v, 1, 32 (15).
• l II: slanting nick about one-third up. 16r, 1, 35 (7); 21r, 2, 14 (10); 28r, 2, 24 (13); 32v, 2, 11 (16).
• m I: left leg bent to right. 6r, 2, 33 (1); 1v, 2, 40 (3); 14r, 2, 21 (5); 9r, 2, 7 (8); 18r, 2, 15 (11); 28v, 1, 41 (14); seen as late as forme 28.
• m II: only the three legs left. 6r, 2, 15 (1); 2r, 1, 39 (3); 14v, 1, 3 (6); 19v, 2, 12 (9); 23r, 1, 3 (12); 25v, 1, 7 (15); seen as late as forme 43.
• m III: no connection between 1st and 2nd legs. 24r, 1, 18 (11); 27r, 2, 3 (14).
• N I: bent left leg. 7v, 1, 8 (3); 15r, 1, 7 (8); 18r, 2, 38 (11); 29r, 2, 22 (14); seen as late as forme 58, but also prelims., ¶ 5r, 30.
• N II: left leg broken inwards. 7v, 1, 38 (3); 11r, 33 (6); 22r, 2, 36 (9); 23r, 1, 26 (12); 31v, 2, 23 (15); seen as late as forme 117.
• N III: left leg broken outwards. 19r, 2, 5 (10); 23r, 2, 29 (12); 31v, 2, 3 (15); seen as late as forme 73.
• n: large dent in left leg. 8r, 2, 7 (3); 14r, 2, 6 (5); 22r, 2, 13 (9); 24v, 1, 26 (12); 31v, 1, 8 (15); seen as late as forme 90.
• O I: projection, inside, bottom left. 20v, 1, 32 (10); 29v, 2, 1 (13).
• O II: projection, inside, bottom left, slightly higher than in O I. 9r, 1, 40 (8); 17v, 1, 7 (11); 30r, 2, 9 (13); seen as late as forme 65.
• P: two nicks on left side of stem; lower part of loop does not meet stem. 9v, 23 (7); 21r, 1, 19 (10).
• p I: cone-shaped nick in top third of stem. 5v, 2, 24 (1); 1v, 1, 20 (3); 12v, 12 (6); 21v, 2, 32 (9); 18v, 1, 11 (12); 26r, 2, 40 (15); very distinctive, also found in 18, 21, 24, 27, 30 and as late as 100.
• p II: piece missing from stem inside bulb. 6r, 1, 21 (1); 2v, 2, 26 (4); 9v, 18 (7); 20v, 1, 19 (10); 29v, 2, 30 (13); found as late as forme 95.
• p III: damaged where bottom of bulb meets stem. 11v, 24 (5); 15v, 1, 18 (7); 20v, 1, 6 (10).
• p IV: nick on top right outside of bulb. 3r, 2, 29 (2); 13v, 2, 26 (5); 10v, 11 (8); 23v, 2, 16 (11).
• Q I: nick just to left of where tail joins circle. 6r, 1, 6 (1); 8v, 1, 29 (4); 16r, 2, 25 (7); 19r, 1, 36 (10); 32r, 2, 25 (15).
• Q II: tail slightly bent in two places. 4v, 2, 8 (2); 12r, 24 (5).

---

251




• q I: nick in stem just above meeting-point with bottom part of loop. 4v, 1, 42 (2); 16v, 1, 26 (8); 23v, 2, 35 (11); 29r, 1, 4 (14); seen as late as forme 66.
• q II: foot of stem bent well to left. 8r, 2, 9 (3); 14r, 1, 41 (5); 9r, 1, 40 (8); 18r, 2, 24 (11); 28v, 2, 33 (14); seen as late as forme 71, also in prelims., ¶ 4v, catchword.
• q III: kink in serif at bottom of stem. 7v, 2, 32 (3); 14v, 2, 14 (6); 19v, 1, 29 (9); 23r, 1, 5 (12); 26r, 1, 34 (15).
• r: small piece missing in angle between stem and horizontal part. 8r, 2, 18 (3); 14r, 1, 22 (5); 9r, 1, 8 (8); 18r, 1, 13 (11); 29r, 2, 18 (14).
• ſ: large slanting nick just above cross-piece. 15r, 2, 29 (8); 23v, 1, 22 (11); 29r, 2, 25 (14); seen as late as forme 123.
• ſſ: top end of second letter bent up. 29v, 2, 27 (13); 31r, 1, 30 (16).
• ſt I: stem of ſ damaged between foot and cross-piece. 14r, 1, 8 (5); 16v, 1, 5 (8); 17v, 2, 7 (11); 30r, 2, 42 (13); 31r, 1, 20 (16); seen as late as forme 52.
• ſt II: ligature broken and bent to left. 18v, 2, 8 (12); 26r, 2, 28 (15); seen as late as forme 68.
• ſt III: mark above left-hand stroke of t. 1r, 2, 11 (4); 15v, 2, 31 (7); 21r, 2, 23 (10); 27v, 1, 27 (13).
• st: s cracked and bent at bottom. 8r, 2, 14 (3); 11r, 24 (6).
• t: bottom of tail bent down. 3r, 2, 32 (2); 13v, 2, 41 (5); 15r, 2, 20 (8); 17v, 2, 17 (11); 29r, 2, 7 (14).
• u I: dent in left-hand leg. 4r, 1, 22 (1); 2v, 1, 28 (4); 15r, 2, 17 (8).
• u II: nick near top of left-hand leg. 3v, 1, 9 (1); 7r, 1, 5 (4); 9v, 17 (7); 19r, 2, 17 (10); 28r, 2, 5 (13).
• v I: piece of serif missing from left leg. 3v, 1, 10 (1); 2v, 1, 26 (4); 16v, 1, 38 (8); 23r, 2, 14 (12); 31v, 2, 11 (15); seen as late as forme 47.
• v II: only serif remains of right leg. 4r, 1, 29 (1); 2r, 2, 36 (3); 14v, 2, 14 (6); 19v, 2, 23 (9); 24v, 2, 12 (12); seen as late as forme 43.
• v III: an italic V with serif of right leg bent into a loop. 3r, 2, 38 (2); 13v, catchword (5); 9r, 1, 39 (8).
• z I: serif of top horizontal touches middle of diagonal. 1r, 1, 9 (4); 15v, 2, 35 (7); 22v, 1, 20 (10); 27v, 2, 31 (13); 26v, 1, 2 (16);.
• z II: notch at bottom right of lower horizontal. 5r, 1, 8 (2); 13v, 2, 36 (5); 15r, 1, 17 (8); 24r, 1, 35 (11); 30v, 1, 4 (14); seen as late as forme 80.

These descriptions are not intended to be minute, but they should be enough to help the reader pick out each letter in the line in which it occurs; anyone checking the list would be well advised to use more than one copy of the book. A list of running-heads follows.

• Recto-head I has a B with a dot in the lower bulb: 2r, 4r, 11r, 22r, 23r, 26r.
• Recto-head II has a damaged o in Don: 3r, 9r, 12r, 24r. 25r, 30r.
• Recto-head III also has a damaged o in Don, but is different from II: 15r, 18r.
• Recto-head IV: small nick in e of Pedro: 5r, 14r, 16r, 19r.
• Verso-head I: s of ligature st nicked near bottom: 5v, 7v, 14v, 19v, 23v.
• Verso-head II: E has dot in middle of lower half: 6v, 10v, 11v, 18v, 20v.
• Verso-head III has a distinctive st ligature: 1v, 3v, 12v, 16v, 17v.
• Verso-head IV has a small a in verdad: 4v, 9v, 13v, 21v.

Notes