University of Virginia Library

Search this document 


expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
The Variant Sheets in John Banks's Cyrus the Great, 1696 Fredson Bowers
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
collapse section 
collapse section2. 
expand section2.1. 
expand section2.2. 

expand section 

The Variant Sheets in John Banks's Cyrus the Great, 1696 [*]
Fredson Bowers

COPIES of John Banks's play Cyrus the Great (1696), the first edition, exist with sheets B, C, and D in two completely different typesettings but without change in text. In the setting which I shall identify as (*), the second line of the opening stage-direction on sig. B1 ends with the word "Battel" and B2 is not signed. On C3v the reading in line 3 is "express'd" and in line 29 "bred"; on D1 the catchword is "Adorn'd" and on D4 the catchword is "Thy". In the other setting, identified as (§), the direction on B1 ends "Battel where-" and B2 is signed; the readings on C3v are "Express'd" and "bread"; on D1 the catchword is, correctly, "Let" and on D4, incorrectly, "They".

The central problem, since it is a matter of prime textual importance, is to determine the order of setting and printing these variant sheets: because one is a straight reprint of the other, only that setting made directly from the manuscript can have authority. Ordinarily, in an examination of this sort, one seeks to determine priority by analytical bibliography, which, if one is fortunate, may decide the issue on the laws of mechanical evidence without calling on the less demonstrable inferences of textual criticism based on the variant readings.[1] In Cyrus the Great the bibliographical evidence, in its salient features, conforms so strongly to a certain well-established pattern of printing that an assumption can be made about priority on this evidence alone. However, the case is not an open and shut one, for there remain some difficulties, and at least two alternative hypotheses can be


Page 175
evolved to fit the facts. As a consequence, it will be more appropriate to survey the textual evidence first to see whether its findings will coincide with what seems to be the preferred bibliographical explanation.

Textual evidence is of two varieties. The first is concerned with the relative "goodness" or "badness" of the readings in an attempt to demonstrate that one version of a text is more correct than another. The second, with less concern for correctness, tries to determine whether one set of variants would evolve more naturally from the other than in the reverse direction. Taking only the substantive variants between these three sheets, we see that of the total of nine, a more correct reading is found six times in (*) against three times in (§). When these are viewed more narrowly, however, it is apparent that there is a real difference between the kinds of faults corrected. Thus (§) properly corrects the simple misprint of "Doys" to "Days" (B1v, l. 38), "no" to "not" (B4v, l. 31), and "thy" to "the" (C1v, l. 12); but when serious variants occur, such as the dropping in (§) of "to" in "to Cyrus" (B2, l. 15), and the omission of "Was" in "Was Hystaspes" (B4v, l. 29), it is (*) which is always the correct text, in addition to such variants as the correct "adorn" (B1, l. 44) in (*) against the "adore" of (§) or the correct "all my" against "my all" (D2v, l. 29). In the matter of substantive readings there is little question that (§) is a comparatively degenerate text, and this deterioration (an almost invariable accompaniment of reprints) is customarily taken by textual critics as proof of later typesetting.

The degree of correctness in the accidentals of a text is less good evidence, for in reprints one may often find substantive degeneration accompanied by somewhat superior punctuation, usually in a fuller system. For the record, however, it may be stated that of the fourteen clear cases where there can be no question of superiority in the punctuation by Restoration standards, (*) is the more correct eleven times, and (§) only three. Of five cases of elision for metrical reasons against non-elision, (*) is correct in all.

This evidence for the priority of (*) conforms to our general experience with texts, but it comes far short of actual demonstration, since the argument is always plausible that the earlier setting was particularly careless and has been corrected in the later. And, in fact, so far as one can determine from the readings, there is only one (§) variant—the omission of "Was" in "Was Hystaspes" (B4v, l. 29) —which would be really difficult for a compositor to mend if (*) were, instead, the later setting.

Other evidence may be adduced, however. For example, on D2v the catchword in (*) is "Have" but in (§) it is correctly "Cyr." for the first words on D3 which are "Cyr. Have. . . ." The setting of a catchword omitting a speech-heading is a not uncommon characteristic of dramatic texts set from manuscript but I have not observed it in reprints, in which a compositor may check on the correctness of a catchword but seldom alters it in favor of an incorrect one. Somewhat similar is the incorrect catchword "Adorn'd" on D1 of (*), which is right for the first word of the second


Page 176
line on D1v but is properly changed to "Let", the first word of the first line, in (§). Again, this is an error found in texts set from manuscript, in which it may probably be assigned to wrong marking of the stopping point, or to a skip of the eye, but it is not characteristic of reprints.[2]

Finally, certain inferences may be drawn from the variant spellings in the two settings of the sheets, but this can be applied only in connection with the interpretation of the bibliographical evidence, and will thus appear later.

When we move to the bibliographical evidence for the printing of this book, the following facts appear. Cyrus is a quarto, collating A-H4 I2, the preliminaries occupying sheet A and the text beginning with a head-title on B1. The text concludes on I1, followed by an epilogue on I1v and I2, and advertisements on I2v. Although there is no break in the text, clear differences occur in the printing of sheets B-E (including both settings of B-D) and of sheets F-I. Most prominent is the fact that though the font of type remains the same, and the vertical measurements of the type-page are almost identical, the printer's measure (i. e., the length of the printer's composing stick as evidenced by the measurement of a full line of type) switches from 119 mm. in B-E (including both settings of B-D) to 123 mm. in F-I. Also, sheet A is set to this 123 mm. measure. To complete the evidence that different skeleton-formes were used for the two sections, this alteration in the typographical layout is accompanied by different settings of the running-titles in each part.

Sheet E, then, is the only sheet which was not reset in the first section composed with the 119 mm. stick, and this fact becomes of considerable importance. In the (*) setting of B-D, each sheet was printed with two skeleton-formes, one for the inner and a different one for the outer forme; on the evidence of the running-titles, these two skeleton-formes were also employed to impose the type-pages of the respective formes of sheet E, thus establishing a firm link between the printing of invariant sheet E and the (*) sheets of B-D. In setting (§), on the contrary, a completely different group of running-titles is found, not elsewhere repeated in the book. This group is found in only one skeleton-forme, however, which was used to impose both the inner and outer formes of B-D. The running-titles of F-H are in different settings, and of a different font, from any in either setting of B-D and of course from E. The group used in the skeleton for inner F also printed outer G, and that for outer F appears again in inner H. Inner G and outer H were each printed from skeleton formes differing from each


Page 177
other as well as from those in the rest of the book. The single running-title in half-sheet I, that on I1, does not seem to be found elsewhere either. The same typesetting of the words "Cyrus the Great" found in the head-title on sig. B1 of setting (*) —but differing in (§) —appears as the first line of type on the title-page on A1. Finally, of the eighteen copies of this book I have examined to date, six contain the (*) B-D sheets, and twelve the (§) setting.

If we were acquainted with this book only in the (*) state, there would be little hesitation in interpreting the evidence as indicating a book in which the manuscript was halved and the two parts B-E and F-I plus A were simultaneously set and printed on two presses. This was a common way of printing Restoration plays, the division of pages is nearly equal, and point by point the facts correspond to the pattern for such printing established in other books.[3] The evidence of this printing is in all essentials so strong that it must, I think, be taken as the basis for any plausible hypothesis about the book. If we accept it, at least for the moment, then under any normal conditions the use of the same skeleton-formes in sheet E in the same positions as in (*) B-D would indicate seriatim printing and thus that the (*) setting is the original one.

This was the tentative conclusion of the textual evidence, and it may be strengthened by the spelling, which demonstrates that the compositor of (*) B-D also set sheet E. The strongest single piece of evidence to this effect is the spelling "Battels" which occurs three times in (*) B-D and once in l. 8 of E2v. Interestingly, we find that the (§) compositor followed the "Battels" form in its first appearance in the stage-direction on B1, but in the other occurrences spells it "Battles." He would not, therefore, seem to be the compositor of E2v. Correspondingly, we find that the single occurrence of the word "sence" in l.3 of D2v is matched in spelling by "Sence" in line 23 of E3v although the (§) compositor changed the D2v spelling to "sense." There is probably a parallel here to the spelling in l.12 of B1 in (*) as "ascent" but in (§) as "assent." In B1, B2 (twice), and C4v we find the (*) spelling "o'er" which also appears on E2, E3, E3v, E4, and E4v. In the four occurrences noted in B and C, the (§) compositor changed the spelling to "o're," although he followed "o'er" under the influence of his copy once on B3v and once on D3. I can make nothing of the differences in capitalization habits between the two compositors, but it is worth mention that such characteristic spellings as disclous'd, Genious, and uncontroull'd from (§) where (*) is conventional, find no place in sheet E. Against this cumulatively invariable evidence that E was set by the compositor of (*) B-D, the one scrap of negative evidence is the spelling Stroak on E3v matching up with Shoar on (§) B2 as against Shore in (*).

So far it would seem that the building of the hypothesis has followed Sir Walter Greg's (and Whitehead's) advice to "seek simplicity." The book


Page 178
conforms to the standard patterns for two-section simultaneous printing. Under such conditions, the evidence of the running-titles and of the spelling characteristics indicates that sheet E not only was set by the compositor of (*) B-D but also followed these sheets on the press, and is, therefore, the original and only authoritative setting. All that remains is to evolve a satisfactory theory to account for the resetting of B-D. Here the pertinence of Greg's full quotation, "Seek simplicity, and then distrust it," becomes apparent, for no thoroughly satisfactory theory occurs to me. There are two standard reasons for such resetting: (1) short printing of the original sheets either as a result of a miscalculation or of a decision to enlarge the edition after a certain number of gatherings had already been machined; (2) an accident happening to the sheets.

There are various reasons for being distrustful of the first. If we take it that short printing of B-D occurred by a miscalculation, we are faced with the uncomfortable situation of imagining that in a book simultaneously composed and printed on two presses in the same shop,[4] a major misunderstanding about the size of the edition-sheet could go undetected for several days although the paper was being brought out from the warehouse and given to each press and the printed sheets taken away to be dried. Moreover, the number of preserved copies does not encourage this theory. It is very odd indeed that of the eighteen I have seen, the proportion has been two to one in favor of what seems to be the later setting. With all allowances for the operations of chance, it would seem that at a minimum the supplementary printing would need to have been larger than the original, and this would have meant a remarkably small edition-sheet for the miscalculating press.

Finally, we should need to assume that sheet E was machined to the full number of sheets. If so, then the point at which the change in estimate was made becomes of crucial importance, for it is a very curious fact that no standing type appears in reset sheet D. For example, if the error was discovered during the printing of the first forme of E to go on the press, it is not very likely that there would have been time to distribute the type from both formes of D.[5] If the decision was not made until sheet E was perfecting, it must have been very early, for neither forme of E shows any signs of having been unlocked. Under these circumstances, the first forme would not have been broken into but would have been held intact to perfect the additional white paper printed by what had been the perfecting forme. This is


Page 179
possible, of course, but it still assumes a rapid though not incredibly expeditious distribution of all of sheet D. Nevertheless, if it occurred, we may legitimately enquire why the skeletons of E were not used to impose reset B-D, for we should expect that the additional sheets of B-D would be printed immediately. On the other hand, if the error was discovered early in printing the first forme, we could believe that the initial forme of B could be quickly set by casting off copy and would be ready for the press before any skeleton of E could be released. This theory would explain the new skeleton, but it does not explain why none of the type of D was left standing if the change was made so early. However, one forme of B might still be set if the decision were made early in the perfecting, so that even the perfecting skeleton might not be available when reset B was prepared for its press. Still, it is odd that—on the evidence of the spelling—a new compositor was assigned for the (§) setting of B-D, although the original compositor should have been free. In my opinion the lack of standing type remaining from D, the construction of a new skeleton-forme for the reset printing, and the change in compositors all combine to argue against the immediate printing of reset B-D. It is true that a delay in this printing, though abnormal, would not destroy the theory of short printing, but a delay is better explained by some other hypothesis. To complete the arguments, one should mention that it is very odd to find two skeleton-formes employed in setting (*) if the original edition-sheet was as small as it would need to have been. On the evidence of the spelling, only one compositor set (*) B-D, but with about half a normal edition-sheet, he could not have kept up with the press and therefore would not have imposed with two skeleton-formes.[6]

The same arguments operate with even added force against a hypothesis that a decision was made to enlarge the edition-sheet sometime during the printing of E, for in such a case there would have been every incentive to machine reset B-D immediately, and to save as much standing type from D as possible. In addition, there is the uncomfortable matter of the proportion of the preserved copies. But the really crushing argument against this theory is that it will not conform to the hypothesis of two-section simultaneous printing, for by the time E was printed, the workmen assigned to F-I and A must have finished their job, or been fairly close to the end.[7] On the evidence, a sufficient number of sheets was printed for all of the second section.

The alternative, an accident to a large number of sheets of B-D, fits the


Page 180
case perfectly for a delay between the printing of sheet E and the reset sheets, since it allows for D to be distributed as well as the skeletons of E, and for a different compositor to appear on the scene. Moreover, there need be nothing odd about the proportion of copies preserved. The one really disturbing matter is its vagueness. What kind of an accident could affect what seems to have been an equal number of sheets[8] in three successive gatherings and not affect the fourth? Moreover, this accident would need to have occurred after sheet E was printed, but before all the sheets were folded and collated for delivery to the publisher.[9] No satisfactory explanation occurs to me save to conjecture that originally only about a third of the edition, represented by the (*) setting, was collated for binding in order to satisfy the initial demand, the remainder being stored packaged by sheets. If, before these sheets were opened and gathered, an accident happened to the packages containing B, C and D, they would need to be reset, and we should find the consistent copies exemplified by the eighteen examined, with three sheets either in the (*) or the (§) setting.

It seems necessary, therefore, even at the risk of unduly extending this examination, to survey as briefly as possible other theories which might fit the facts to see if any other is preferable.

If we adhere to the initial hypothesis for two-section simultaneous printing, then it would seem impossible to reverse the order of the settings of B-D, even though the proportion of copies I have seen suits (§) as the original better than (*). Until sheet E is printed, the book cannot be bound. If we suppose that the resetting (*) was not decided on until E was at press, we cannot explain the construction of new skeleton-formes for E which are thereupon used in the resetting. Moreover, there is the fairly clear fact that E was set by the compositor of (*) and not of (§). Since any tortuous explanation of the use of the skeleton-formes must necessarily involve the immediate resetting of B-D, we cannot explain why no standing type from D was preserved. Finally, we cannot escape these various objections by supposing that the decision to enlarge or to overprint was made while D was on the press. In such a case there would have been an over-run of at least one forme of D, and preservation of standing type. Also, if D were on the press, the compositor would necessarily have already set some of E, but on the spelling evidence he did not.[10]

It will be clear that the original hypothesis advanced must hold (whatever the explanation) provided we continue to accept the interpretation


Page 181
of the evidence that the book was simultaneously set and printed in two sections. It is legitimate to enquire, therefore, whether a reasonable alternative can be adduced. In fact, a somewhat plausible set of inferences, of a quite different cast, can be contrived involving seriatim typesetting for the whole, at least as the original plan.

For example, if a decision were made to enlarge the size of the edition while sheet E was on the press, two-section printing might have been introduced at this point consequent upon the necessity to reprint B-D. If this were so, we should have a situation differing materially from the first hypothesis. Since the book would not have been planned from the start as a two-section job, sheets F-I would not have been printed simultaneously with the first setting of B-D but, instead, at the same time as the second setting. In this manner the facts of two-section printing would be reconciled with the theory of an enlarged edition, a theory which must be rejected if two-section printing had been planned from the start. There is much that is attractive about this hypothesis. The use of two or more presses in the second section is explained by the necessity to print four and a half full sheets while a single press is reprinting three sheets in smaller quantities. This hypothesis can be maintained, however, only if we discard the theory that press I (for the first section) printed (§) B after completing the last forme of E, for otherwise the compositor would have used the first wrought-off skeleton from E for his imposition. If, therefore, (§) B did not follow E, the alternative is that E (at least as a whole) did not follow (*) D on the press.[11] The hypothesis, then, must be altered to place the decision to enlarge the edition late in the machining of the second forme of D. By this time the compositor should have set up to E4 and possibly even E4v. If, thereupon, before press I had finished machining (*) D, its compositor began the resetting which resulted in (§) B, it would have been possible for press I to have printed the first forme of E in order to keep busy and then to have swung over to (§) B, leaving the perfecting of E to press II or III while its compositors were setting F and G. The timing would apparently be right for such a schedule, and the construction of a new skeleton for the B-D resetting would be explained, together with the absence of compositor I's setting in F. That F-I did not thereupon employ the skeletonformes of E would doubtless result from their changed measure which would have necessitated the cutting of new furniture reglets.

The prime difficulty in accepting this theory is the fact that no standing type from (*) D was preserved, although it would have been available, to save time and money in printing reset (§) D. This is so inexplicable, indeed, that it should cause us to be more than ordinarily suspicious, the


Page 182
more especially since it explains on largely fortuitous grounds the equal division of a book between two or more presses, which we are accustomed to view as the result of the simultaneous printing of the two sections. Moreover, on the spelling evidence, the compositor of reset B-D was not the same workman as that of the original sheets, whereas the theory almost positively requires him to be one and the same. Finally, the proportion of the respective settings of B-D in the preserved copies argues strongly against a hypothesis based on the enlargement of an edition-sheet. These are such strong reasons that I do not believe we can accept this alternative.

If, in a last desperate attempt, we reverse the order of the B-D settings, we are up against the reasonably clear textual, spelling, and semi-bibliographical evidence that (*) was the first; we should need to assume that E (on the evidence of its (*) running-titles) was not machined until the last operation although it must have been set earlier; we are faced with the strong probability that E, nevertheless, was set by the (*) compositor; and we still cannot explain the lack of standing type in D.

It would seem that the most prominent alternatives involving any probability in normal printing practice fail to satisfy the prime requirements for any satisfactory hypothesis, which must explain (a)- the proportion of variant B-D sheets in the observed copies; (b) the lack of standing type in D; (c) the different skeleton-forme in reset B-D; (d) the equal division of the book between two or more presses; (e) the change of compositors between the two settings of B-D.

The only theory which will fit every one of these facts is the original one, which introduces the plan for two-section printing from the beginning, and posits a delay before the reprinting of the reset B-D sheets. The only difficulty in full acceptance of this hypothesis is the vagueness of the explanation, involving some peculiar sort of accident to three successive sheets. But until a better explanation occurs to someone, I believe this is the theory we must hold. Regardless of the explanation for the how and why, however, it would seem sufficiently clear that the (*) setting of B-D, the details of which were described at the start, is the original and only authoritative form of the text in these sheets.



This note is part of an investigation of the Bibliography of Restoration Drama under grants from the Research Council of the Richmond Area University Center and the Research Committee of the University of Virginia.


A classic example, which I shall always remember, is my wrong "critical" assignment of priority to a resetting in Dekker and Middleton's Roaring Girl (1611), properly reversed on strict bibliographical evidence by J. G. McManaway. See my "Thomas Dekker: Two Textual Notes," The Library, 4th ser., XVII (1937), 338-340; and McManaway, "Thomas Dekker: Further Textual Notes," The Library, 4th ser., XIX (1938), 176-179. Another example, which I can mention with pleasanter associations, is my bibliographical investigation of the order of the two 1669 editions of Dryden's Wild Gallant in The Library, 5th ser., V (1950), 51-54; see also Studies in Bibliography, III (1950), 52-57.


This catchword correction is of some real interest. Under ordinary circumstances a compositor of a reprint follows regularly the catchword on the recto page of his printed copy, since he must turn the page to know whether it is right or wrong, and thus original variants between catchwords on the recto pages of successive editions and following words on the versos often become perpetuated. On the other hand, for a reprint to make such an error would be most difficult to explain. Simple misprints, of course, do occur, as the misprint "They" in the D4 catchword of (§) for the correct "Thy" of (*).


For various examples, see my "Bibliographical Evidence from the Printer's Measure," Studies in Bibliography, II (1949), 156-162.


It would be much more plausible to find a miscalculation about the size of the edition-sheet existing if the two sections of the book had been printed in different shops. But the fact that the standing type from the (*) head-title was used on the title-page printed by the workmen of the second section is sufficient evidence that the two parts were manufactured in one house.


Since type was not distributed until its skeleton-forme had been stripped, and this stripping was not accomplished until the next forme was ready to be imposed, we should also need to suppose that the compositor had made an almost immediate imposition of the second forme of E from the skeleton of the last forme of D, although the press would not require the forme for some time. However, this is not impossible.


The relation of the compositor to methods of imposition according to the speed of the press governed by the size of the edition-sheet is a technical matter carefully studied by C. J. K. Hinman, "New Uses for Headlines as Bibliographical Evidence," English Institute Annual, 1941 (1942), pp. 207-212.


The pattern of the four (probably five) sets of running-titles in F-I is clearly not that associated with one-press printing and instead indicates that two or more presses were assigned to the section. It would, therefore, be printed more quickly than B-E if both had started at the same moment. But there is nothing against a delay in beginning the second section that would cause both sections to be finished at about the same time.


So far I have seen no mixed copies.


According to Moxon, at this date it was the printer's duty to fold the sheets and collate them into separate copies, ready for the binder, before delivering them to the stationer. Of course we do not know that this was the invariable practice, but evidence exists to suggest that it may well have been the usual one.


If one wants to lean the maximum weight on the evidence of he spelling of "Battels" versus "Battles," the appearance of "Battels" on B1 in both settings is most readily explained as inadvertent following of the (*) copy by the (§) compositor, whose normal spelling habit thereafter asserted itself. If so, (§) would necessarily be the resetting.


The difficulties in the order of (*) D followed by E have already been sketched in another connection. They consist chiefly in the fact that the compositor of E, on the evidence of the measure, did not compose any part of F, although shortly after the first forme of E was delivered to the press he should have begun this setting. The lack of standing type from D is also a difficulty which has been discussed.