University of Virginia Library


Page 97

Observations on Two Caxton Variants

THE TWO CAXTON VARIANTS WHICH PROVIDE the topic of discussion for the present investigation are by no means recent discoveries; quite to the contrary, they have been noted for many years in the standard books of reference. It is most strange, therefore, that these well-known variant states have been subjected neither to critical study nor to thorough technical analyses, especially since, as distinct from many other Caxton variants, it is absolutely certain in each instance which is the earlier state and which the later. This circumstance further permits the investigator to ascertain specific details as to the workings of Caxton's printing office, with special reference to the resetting of copy previously printed at the same press.[1] The two variant states to be discussed here are found in: Lydgate, The Pilgrimage of the Soul, Westminster, 6 June 1483 (Duff 267) and Christine de Pisan, The Book of the Fayttes of Armes and of Chyvalrye, [Westminster], 14 July 1489 (Duff 96). The separate states are identified by the fact that one sheet in each book appears in two different settings of type.


In E. Gordon Duff's Fifteenth Century English Books (Bibliographical Society, Illustrated Monograph No. XVIII, 1917), one finds the following note appended to his description of No. 267:

There are two issues of this book: in the original issue (B.M.) the two inner pages of sheet f3 have been imposed wrongly, so that what should be on f3b is


Page 98
on f6a and what should be on f6a is on f3b, and the whole book is in type 4. In the second issue (Britwell) this whole sheet has been reprinted in type 4*, so as to read correctly.[2]

Of the six copies that have survived to our day,[3] the British Museum is the only one to have sheet f3.6 in the original state,[4] the remaining copies all belonging to the later "issue" (contrary to the opinion expressed in STC 6473).

Since it is absolutely certain that the Museum copy belongs to the earlier state, one may well ask how it compares with the "corrected" later form of sheet f3.6? The ready answer to this query is that the resetting (though corrected so far as the imposition is concerned) is much the more inferior text. Not only does the reprint contain eleven misprints to seven found in the original setting[5] but it also omits nineteen words found in the original— a fact of much greater significance. To be exact, twenty words are omitted by the reprint but, by way of partial compensation, one word[6] not found in the first setting is added by the second.

Almost half the words wanting in the later sheet are clearly omitted because of careless type-setting. The compositor of the reprint, although he was setting type from printed copy, did not follow his original line for line.[7] On the verso of signature f3 (of the reprint) he departed so far from his copy that when he reached the bottom of the page he found that he had less than half a line to accommodate a full line of the original text. By making radical omissions, the compositor succeeded in having his page end at the proper place but not without doing violence to the text. In the original state (f6) the passage in question reads:


Page 99
. . . And leue || it wel yt though the passiō of crist profite not these innocēts to their || ful saluaciō yet it profiteth them so moch yt sathanas lyeth loken in || the depthe of helle / so that he ne may not ne none of his mynystres || [f4] annoyen ne tormenten none Innocent / as their malyce wold / ne || harmen none persone / but by his owne assent . . .
In the reprint, the same text (f3 verso) appears as:
. . . & || leue hit wel that thong (sic) the passion of Jhesu crist prouffite not || these innocentes to their sauacion / yet it prouffiteth them soo || moche yt sathanas lyeth lokē in helle yt none of his mynystres || [f4] annoyen ne tormenten none Innocent / as their malyce wold / ne || harmen none persone / but by his owne assent . . .
In making these omissions the compositor succeeded in having his text end with the correct word, but in order to achieve this both sense and grammar were sacrificed to expediency.

Apart from this instance, the textual differences are slight and occasionally reflect no more than the spelling habits of two different compositors;[8] for example, the British Museum state has "nought" eight times where the reprint has "not." In those cases where a choice can be made, however, the British Museum setting is always the better. Thus where the soul asks the body "how hast thou lost al thy queyntyse" (BM, f3v, l. 33), the reprint offers "how thou hast lost al queyntyse." Clearly then, instead of seizing upon the opportunity to improve the text, a careless compositor has permitted the text of the reprint to deteriorate.

Curiously enough, the reprint exhibits a technical—as well as a textual—deterioration. Concerning the use of "guide-letters" or "directors," Konrad Haebler[9] observes that in order "to lighten the rubricator's work, the custom was gradually adopted of printing in small type (usually in lower-case), in the space left to be filled in by hand, the initial which the rubricator was to add in colors." This had been common practice on the Continent from as early as 1471-2, and directors appear in books printed at Caxton's


Page 100
press even in the days of its activity at Bruges.[10] In the Pilgrimage, guide-letters were used throughout the volume including the first setting of sheet f3.6, but directors are not present in the reprint of this sheet. Thus the reprint demonstrates the return to a more primitive practice of type-setting where such reminders were not considered necessary for the benefit of the rubricator.


To its description of this edition (No. 6648), the Gesamtkatalog der Wiegendrucke adds the note: "Das Doppelblatt 140 u. 143 ist neu gesetzt worden." This appears to be the first notification that sheet S2.5 is known in two different settings, a fact that (apparently) escaped the attention of the editor of this work for the Early English Text Society's edition.[11] Of the twenty-one copies of this book known to me, only four have the variant sheet specified by the Gesamtkatalog—Göttingen, Bibliothèque Nationale, Bodleian (S. Selden d 13) and the Grenville Kane copy now at Princeton University Library.[12] Thirteen copies have the text as printed in the Early English Text Society series, while three copies want (among others) the pertinent leaves—Windsor, Bodleian (Auct. QQ supra l. 25) and the York Minster copy now in the collection of Phyllis Goodhart Gordan and Howard L. Goodhart of New York. Seymour de Ricci informs us that the Sion College copy lacks S5, but I have been unsuccessful in my attempts to obtain information as to the state of S2 in this copy.

Although the Gesamtkatalog simply assumes without further proof that the Göttingen state is that which was "neu gesetzt," there is plenty of evidence to prove that this is certainly the case.[13]


Page 101
In this instance, positive proof is supplied by the different systems of punctuation employed by the two compositors. As will be seen from a perusal of the modern reprint, the enormously preponderant punctuation mark used in the Fayttes is the virgule (/). For example, on signatures S1, S3, and S4 (six pages common to all copies), there are 164 punctuation marks of which 163 are virgules, the remaining one being a semi-colon. In the first setting as represented by the copy in The Pierpont Morgan Library (PML 781), there are 118 instances of punctuation on S2 and S5; of these 117 are virgules and the remaining one is a semi-colon.

For the other setting, the Kane-Princeton copy shows (on sheet S2.5) 105 cases of punctuation, of which 96 are periods, 6 are semi-colons and only 3 are virgules. Obviously the text of this sheet was set by a different compositor—one following his own rules of punctuation—than the one who had composed all the rest of the volume. Thus it is quite certain that sheet S2.5 of the Morgan copy belongs to the original setting and that the Princeton one is the reprinted sheet.

Again we may enquire what a comparison of the two states reveals and again we will note that the reprint shows a deterioration of text. The state of the Morgan copy has six misprints, while the total in the Princeton setting amounts to some seventeen such errors; two misprints are common to both states.[14] Characteristic of the better text of the first is the reading "Consules of Mountpellyer" (S2, l. 4) where the reprint specifies "Consules Mountpollyer." Conversely, however, the second setting has the more correct last lines in the colophon which read: ". . . he may || atteyne to euerlastyng lyf in heuen. whiche god graunte to || hym and to alle his lyege peple. AMEN. || Per Caxton ||". The first contained the misprints "euerlastpng" and "gaunte." However, since the far greater number of misprints is found in the form as represented by the Princeton copy, we must consider this second setting as the more inferior text.


Page 102

A most singular peculiarity of the Princeton volume requires special mention. The blank last leaf (S6 recto) in this copy shows a distinct off-set of the text of S5 verso, a condition often encountered among Caxton imprints. But the startling fact here is that the Princeton leaf shows an off-set of the original setting, not of the resetting which now faces it in the volume. There seem to be only two logical explanations for this phenomenon, either that the copy was "made up," perhaps when it was rebound by Bedford,[15] or that the off-setting took place in Caxton's workshop. If the latter assumption be the correct one, we would have certain evidence that the copies of the Fayttes were not bound up as soon as printed but that the sheets were stored unbound.[16] Thus it would have to be argued that the Princeton sheet S1.6 came into contact soon after printing (possibly by being gathered with it) with a sheet of the earlier setting of S2.5, but that, in the long run, it was not bound up with this particular sheet.

Turning to the other alternative, it seems highly improbable that the blank leaf was added to this copy,[17] since no one would ever have considered it necessary to supply such a leaf in order to create a "perfect and complete" example. Again one cannot assume that only leaf S5 was supplied, since S2 is its proper conjugate; therefore, if anything was added to the Princeton volume, it must have been the whole sheet (S2.5). Furthermore, if the blank be the original one, as we have good reason to believe, it seems unlikely that the important printed leaf just ahead of the useless blank could have been missing from this copy at any time; surely, if S5 was lost, S6 would have disappeared too. One must believe, then, that if the volume is not in the same condition as it was when sold by Caxton (at least, as far as the identity of leaves is concerned), one must assume that the Princeton book was so "made up" that S1.6 was supplied from one copy and S2.5 from another. This seems to be so highly improbable a hypothesis that


Page 103
the first explanation for the presence of the off-set in the Princeton Fayttes appears the more acceptable.

In conclusion, a word should (I think) be said on the subject of the origin of these two variant settings. In the case of the Pilgrimage, of course, the sheet was reprinted in order to correct an error of imposition;[18] a similar technical error[19] occurs in the very first book credited to Caxton's press—the Recuyell of the Histories of Troy, [Bruges, circa 1475]. The cause for the reprint of sheet S2.5 in the Fayttes is less obvious. It illustrates, however, a phenomenon previously noticed in several volumes printed by England's prototypographer. Since the second setting contains more misprints than the first, it was probably not called into being through a desire to improve the text. True enough, two glaring errors in the colophon were corrected by the second setting, but this improvement could have been achieved by simple stop-press corrections and certainly would not have required the resetting of four entire pages.[20] It seems likely that the reprinting was necessitated either because something happened in Caxton's workshop which required the resetting of sheet S2.5, or that Caxton failed to print the correct number of sheets to complete the copies in hand and was thus obliged to make good the deficiency after the original formes had been distributed.


Page 104

One must, then, assume that the reprinting of sheet S2.5 was the result either of an accident in the printing office or of a short count when the sheet was being machined. If an accident necessitated the reprint, it is scarcely probable that this took place during the printing of the sheet since it would be an incredible coincidence for both inner and outer formes to "pie" at approximately the same stage in the course of production.[21] If sheet S2.5 was not reprinted because a short count made this necessary, one must believe that some accident took place at the press after the full number of sheets had been printed and the formes distributed, for the loss of a sufficient number of sheets would have compelled Caxton to reset the text and supply the necessary number in a new setting. While this would be a perfectly reasonable explanation for the reprinting of sheet S2.5 in the Fayttes, one can only with difficulty credit a theory which assumes that Caxton permitted such accidents to happen again and again. There are reprinted sheets of precisely this sort in the Dictes and Sayings of the Philosophers (Duff 123), the Morte d'Arthur (Duff 283), the House of Fame (Duff 86), and probably elsewhere, for aught I know.[22] Surely these could not all be the results of physical accidents. On the other hand, contemporary accounts prove beyond question that short printing was a common enough occurrence to be a source of annoyance and trouble to both printers and publishers. Neither of these explanations for the presence of the variant setting in the Book of the Fayttes of Armes and of Chyvalrye is completely convincing, but if there be a more satisfactory one, it is not apparent to me.



It is a great pleasure to record once again my hearty thanks for the kind help of Professor Fredson Bowers. In this case, as always whenever he has been consulted, he has been ever-ready to offer advice and criticism.


The Britwell copy is now in The Pierpont Morgan Library (Accession number 20892; Check List 1778). Under no. 6474, the STC states: "Sheet f reprinted" and cites only the Morgan copy.


Five copies are cited in Seymour de Ricci's Census of Caxtons (Bibliographical Society, Illustrated Monograph no. XV, 1909, pp. 78-79). The sixth copy was formerly in the possession of the firm of William H. Robinson Ltd., and was described by W. Loftus Hare in his study: "A Newly Discovered Volume Printed by William Caxton," Apollo, October, 1931 (11 page reprint published by the firm). Details as to this copy I have as the courtesy of the present owner.


Leaf "fiij" appears to be signed "fij" in the Morgan copy, but this may be the result of improper inking.


On f6, line 27 of the BM copy, is found the word "lauement" ("lauament" in the reprint); the earliest occurrence of this word noted by NED is 1597.


See below, where the reprint has "Jhesu crist" for the "crist" of the first setting.


On f6 verso, 27 lines of the first setting have been expanded to 28 lines in the second.


For discussions of this problem, see Charlton Hinman, "Principles Governing the Use of Variant Spellings as Evidence of Alternate Setting by Two Compositors," The Library, 4th ser., XXI (1940), 78-94, and Philip Williams, "The Compositor of the 'Pied-Bull' Lear," Papers Bibl. Soc. Univ. Virginia, I (1948), 61-68.


The Study of Incunabula, trans. by Lucy E. Osborne (New York, 1933), p. 112.


Guide-letters are found regularly in the Recueil des histoires de Troie (Duff 243) and occasionally in the English version of this work (Duff 242). They appear regularly in the History of Jason (Duff 245), which both Blades and Proctor considered the first large English book printed in England. Blades (no. 9) dates it as belonging to the "early part of 1477."


Original Series 189 (1932), edited by A. T. P. Byles. In the re-issue with corrections (1937), Mr. Byles takes note of the two states (p. xxxi) but no variant readings are given in his text. Volume VI of the Gesamtkatalog, which contains the description of the Fayttes, bears the date 1934.


This is the former Huth copy, listed by Seymour de Ricci under no. 28.2.


The first line on folio B7 recto normally reads: "swīmyīg ouer a gret ryuer / and thurghe thees waye of swī-|| mīg . . ."; in the Yale copy the same line reads: "swīmyn oer a gret ryuer / and thurghe thees waye of swīs-|| mīg . . .". This may represent a variant setting caused by a desire to improve the text or it may be no more than a mechanical variant brought about by having the type pulled out by the ink-ball and incorrectly restored. Another minor variant indicating stop-press corrections may be found on N6; here the "Capytulo" should be numbered 16 but no copy seems to have this number. Some copies (for example: Morgan, Columbia, Yale, Queens [Oxford], etc.) have "Capytulo xiiij" while others (Huntington, Princeton, Bodleian, University Library Cambridge, etc.) have the number "xv." It is not clear which is the earlier setting in this case.


On S2, l. 22, both settings have "theunto" for "therunto" and on S5v, l. 6, "hyeues" for "hyenes."


Neither Seymour de Ricci nor the Huth Catalogue (I, 310) indicate that any leaves have ever been added to the copy. According to the sale catalogue (p. 444, lot 1570), the copy is "perfect and large" and was described as being in a binding by Riviere.


For a discussion of when Caxton volumes were presumably bound, see my "The Binding of Books Printed by William Caxton," Papers Bibl. Soc. America, XXXVIII (1944), 1-8.


The modern binding makes it impossible to be entirely certain, but S1 and S6 appear to be conjugates. Whether conjugacy can be proved or not is immaterial, since it is shown above that no one would supply a blank leaf anyway. Accordingly it is reasonable to assume that S6 is the conjugate of S1.


A similar error of imposition occurs in the Fayttes where O3 verso contains the text which should be found on O6 recto and vice versa. All the copies seen by the writer contain this error. An example of an error of imposition which led to the suppression of the sheet is cited in my note "Caxton's Blanchardin and Eglantine: Notes on the Leaf Preserved in the British Museum," BSA, XXXIX (1945), 156-61.


For a discussion of the classification of variants, see my paper "Variants in English Incunabula," Bookmen's Holiday (New York, 1943), pp. 459-74.


It has been suggested to me that Caxton might have suppressed the first setting of S2.5 because of the errors in the colophon (representing the printer's official statement) and that the incorrect sheets were withdrawn from the made-up copies and the new sheets substituted for those suppressed. Thus the off-set in the Princeton copy could have come from the first sheet S2.5 before the corrected one was substituted for it. However, this seems to me an untenable theory in view of the fact that only four copies of the second setting are known against the thirteen of the first. If Caxton reprinted and substituted sheet S2.5 in order to correct the colophon, one would (I should judge) expect the majority of copies to have the second (official) setting and only the occasional (overlooked) example to preserve the original reading. If, however, extra sheets of the second setting of S2.5 were printed because of an original short count, it may be argued that these were substituted (because of the superior colophon) for sheets of the first setting to the extent that these correct sheets were available; this also would explain the presence of the "wrong" off-set in the Princeton copy. But one should observe that errors in the colophon did not seem to disturb Caxton unduly; note the many instances of such misprints cited by W. J. B. Crotch, The Prologues and Epilogues of William Caxton (Early English Text Society, Original Series 176, 1928). Other Caxton editions with errors in the colophons include: Ars moriendi (Duff 33), two Chronicles of England (Duff 97-8), Doctrinal of Sapience (Duff 127), etc.


Furthermore one would then be obliged to assume that Caxton employed two presses, one to print one forme and the other to perfect the sheets. In that case it is easy enough to believe that one forme might go to pie but that both should do so about the same time passes the bounds of credibility.


Such reprintings are not peculiar to Caxton. See, for example, the case cited in my article "Notes on Two Incunabula Printed by Aldus Manutius," BSA, XXXVI (1942), 18-26.