University of Virginia Library

Cupids Revenge (Q1) and its Compositors Part I: Composition and Printing
by
Hans Walter Gabler

The first quarto (1615) of Beaumont and Fletcher's Cupids Revenge is the only text of the play with claim to authority. But the extent of its reliability is less than assured. Because of the singular position of Q1 in the textual tradition, its degree of authority can be inferred only from a close bibliographical analysis of the quarto itself. Such an analysis is the purpose of this paper. Its intention is to provide the bibliographical evidence to which the editorial decision for a critical text can be referred.[1] Its concern is therefore not so much with the text as with the securing and testing of that evidence. The printing process has been investigated, and the data that shed light on it, such as the arrangement of the text on certain pages, or the stages of compositorial work as they are recognizable from the use and re-use of type, were considered in great detail. The working habits of compositors were distinguished. The analysis of the quarto of Cupids Revenge was at all critical stages controlled by reference to the closely parallel conditions of composition and printing in a second dramatic quarto, that of Wentworth Smith's Hector of Germany. Thus, the matter that is here most immediately presented for its relevance to the printed text of Beaumont and Fletcher's Cupids Revenge, is, in more general terms, also a contribution to the study of printing in an early seventeenth-century London printing-house.

On April 24, 1615, the two plays were entered in the Stationers'

Register for Josias Harrison.[2] Cupids Revenge was written before 1612. It was acted at Court in January, 1612, and again in January, 1613, by the 'Children of her Maiesties Reuels,' i.e. the second Children of the Queen's Revels who existed between 1610 and 1616.[3] The Hector of Germany, or The Palsgrave, a thoroughly fictional history play, was probably written in 1613, in connection with the marriage of Frederick V, Elector of the Rhenish Palatinate (i.e. a 'Palsgraue'), to Princess Elizabeth of England, and, as the title-page and the Prologue testify, it was acted on the stages of the Red Bull and the Curtain by a group of young amateur players.[4] Josias Harrison handed both plays over for printing to Thomas Creede. Both appeared in quarto[5] and with the date 1615, so that it can be assumed that they were produced at the printing-shop under similar working conditions and within a short time of each other.[6]

Cupids Revenge was set by two compositors, a Compositor X and a Compositor Y. The division of their work on the quarto[7] can be determined as follows:

Compositor X set the title-page, 48 pages and two half-pages, Compositor Y set 29 pages and two half-pages of Cupids Revenge.

The basic spelling tests for the identification of compositors X and Y in Cupids Revenge prove to be equally useful when applied to the quarto of Hector. They enable clear-cut attributions for a majority of its pages, and corollary evidence obtainable from the special word-material of this play (such as variations in its speech-prefixes, e.g. Clynton vs. Clinton, etc.) establishes fully its similar compositorial pattern:

Compositor X set the original title-page and 28 pages, Compositor Y set 36 pages and the cancel title-page of The Hector of Germany.

While there are irregularities in each case, some of which will call for explanation and interpretation, it seems safe to say that the general pattern represents a procedure of non-simultaneous seriatim composition, with the take-over of one compositor from the other ideally at $3, or the beginning of the second half, of each gathering. Thus, according

to what seems to have been their basic work plan, each compositor set the second half of one, followed by the first half of the next sheet. When one finished his stint, his partner was ready to proceed from the fifth page of text of the sheet in hand, having had ample time since ending composition at the bottom of page four of the previous gathering to fulfill the auxiliary duties required of him during the printing process, and to distribute type. Performed in each case by one compositor while the other was setting type, these tasks would not influence the progress of composition and printing of the text. Collaboration was thus apparently aimed at largely cutting out the 'non-productive' phases of compositorial work, allowing only the time of composition itself to count in the compositor:press relationship. It was therefore also limited, so it seems, to sections of a quarto where efficient distribution would directly affect the continuity of composition. At the beginning of a quarto, however, where work presumably started from a full case of type, and at the end, where a case could be allowed to run low, there was no immediate need for distribution, and, therefore, apparently also no need for two compositors. Thus, we find Compositor X alone setting a whole gathering and a half at the beginning of Cupids Revenge, and both X and Y are seen, at the end of Cupids Revenge and Hector, to compose on their own stretches of text considerably longer than the usual two consecutive half-gatherings.

But where collaboration was in effect, it allowed for great flexibility. Composing by two consecutive half-gatherings meant, in terms of the printing, a division by half-formes. Two of a compositor's four pages went into the inner, two into the outer forme of a sheet. Consequently, just two pages of wrought-off type at a time would be returned to him for distribution. Theoretically, and perhaps ideally, he would find four pages of type waiting to be distributed when he finished a stint of composition, and he received another four pages and distributed them, too, before beginning a new stint. This ideal situation may, of course, in practice never have occurred. Instead, what was of practical importance was that each two pages of type distributed replenished a compositor's case just sufficiently to allow him, in an emergency, to step in and resume composition early, and at short notice. The compositors were thus not strictly bound to work-units of the size they had set out to compose in — they would have been much more so bound had they originally settled for seriatim composition by sheets, or for setting by formes — but could alternate freely. If run, or improvised, to perfection, their alternation could presumably be made to mesh so smoothly as to equal the continuous setting of one

man from a 'bottomless' case. It would, basically, have required no casting-off, but simply the handing over of copy from one compositor to the other at the end of each stint, without a break in the continuity of composition. However, if one compositor happened to be ready to begin his prospective new stint before the other man had quite finished the preceding one, a little bit of ad hoc casting-off would enable him to proceed, thereby for a brief period actually setting simultaneously with his partner. If, on the other hand, he reappeared before his time to make up for a delay, he would, rather, be concerned with setting a relief page or two in his partner's stint. And, since a relief page could be the fourth page in a forme, the basic sequence of seriatim composition by two men was thus, when the need arose, apparently also open to an approximation of the principle of composition by formes.

In short: an initial survey of the pattern of compositorial stints in both Cupids Revenge and Hector shows two compositors at work in alternation, according to a system of non-simultaneous seriatim setting of type. The preceding observations indicate that it was a system with great flexibility in technical matters. This will have to be further substantiated, and it remains to be seen just how, as a mode of composition, it may also have been adaptable to the varying nature of printer's copy and thus have had an influence on the texts it produced.

The evidence of running-titles shows that both quartos were printed in two skeletons. The running-titles in Hector read 'The Palsgraue' on recto and verso, except for A3, C4v and D1 which have 'Palsgraue' without the article in what are two, or possibly even three distinct settings. The running-titles in Cupids Revenge read 'Cupids Revenge' on recto and verso throughout, except for 'Kupids Revenge' on C3 and C4v in the uncorrected state of C(o). While in Cupids Revenge the identification of running-titles and the tracing of the formes through the press is relatively uncomplicated, the running-titles in Hector are at times impossible to re-identify. Among the identifiable ones shifts of position within the skeleton-formes in consecutive printings are so common that one suspects many accidents may have befallen the others, with consequent re-lockings or resettings that prevent recognition. It is, however, possible to determine that, basically, the two skeletons which are found to be established in B(o) /C(i) and in C(o) print the subsequent formes of the quarto in regular alternation.

From the initial pattern of running-titles, and from the frequent shifts between compositors in the sequence of composition between C3 and D2v, an order of printing for the first three sheets of Hector can be reconstructed which explains the late consolidation of two alternating

skeleton-formes, and which not only suggests the order of composition of the early text and the pages of preliminary material, but also accounts for the deviation from the usual eight-page length of compositorial stints between C3 and D2v. From the manuscript of Hector it was certainly a simple matter to determine that the preliminaries, including the Prologue, would occupy the first half of gathering A, and that the text was to begin on A3. Compositor Y started here on his first full stint which took him through B2v. Compositor X followed with a full stint through C2v. While X was setting type, Y constructed two skeleton-formes which successively printed B(i) and B(o). Y resumed composition on C3, but he must have seen that he could not for long afford to have four pages of his type tied up in A, so he interrupted his stint, set the dedication for page A2, and imposed A(i). He used the skeleton from B(i) and, needing only two running-titles, he removed the other two. X, after setting C3v, followed suit in returning to sheet A and composed the play's original title-page and the Prologue, in order to prepare A(o) for the perfecting of A(i). It is possible that Y, before embarking on his four-page stint C4-D1v, stripped B(o) and placed the three pages of C(i) that had already been composed in its skeleton. For X constructed a separate skeleton-forme for A(o), with two new, and differing, running-titles in positions A3 and A4v. The printing of A(o) must then have progressed far enough by the time Y had composed D1v for him to interrupt his typesetting again in order to make C(i) ready for the press. X, meanwhile, set no more than D2 before he, presumably, was called upon to strip A(o) and impose C(o). C(o) contains the running-title from A4v in position C3, and, in all likelihood, its forme-mate from A3 in position C4v. The running-title in position C1 is taken from B3v and is one of the two that were removed when A(i) was imposed in the forme of B(i). The running-title in position C2v is disturbed in all copies of the quarto and cannot be identified, but it may be the other one that was removed from B(i). Compositor Y, meanwhile, set page D2v, being the last page of a stint that should have been consecutively his from C3 onwards. Compositor X, although he had twice relieved Y in its course, quite orderly took over at D3, and regular stints of two consecutive half-gatherings resume.

The skeleton-formes printing C(i) and C(o), then, are the ones that subsequently alternate in the quarto. The inference about the sequence of composition, imposition, and printing which establishes them draws additional support from the observation that the third of the three original skeleton-formes is demonstrably only used in A(i)

and B(i). The two running-titles in common to A(i) and B(i) do not reappear. The skeleton was thus presumably discarded in a state that contained these two running-titles only, and the likely order of the formes it printed was therefore B(i)-A(i). From this assumption, the rest of the argument follows, and it is not improbable that compositors X and Y worked in the manner described through page D2v of Hector, in the easy alternation and flexible timing that their mode of collaboration permitted. Further aspects of that collaboration will become evident from Cupids Revenge.

In Cupids Revenge, the stage at which half-sheet A was printed, cannot be deduced from the available evidence; but otherwise, there is no reason to doubt an order of printing that started regularly at B(i). The two skeleton-formes which in alternation printed the quarto are found established in B(i) (skeleton A, with running-titles I, II, III, IV), and C(o) (skeleton B, with running-titles V, VI, VII, VIII). Skeleton B had been used before to print B(o), but of the three running-titles in evidence there, only no. VI (B2v→C2v) survived into C(o). In skeleton A, running-title no. II (B2→C2) was reset for C(i), but from C(i) and C(o) onwards, the two skeletons move — with a few turnings, but only occasional resettings of running-titles, or shifts in their relative positions — smoothly through all the formes of the quarto but the last four. Of these, skeleton B prints K(o) only, and skeleton A prints K(i), L(i), and L(o), in that order, as L(i) takes all its three running-titles from K(i), whereas L(o) preserves only two of them and uses a reset third one.

As long as the skeletons alternate, skeleton A regularly prints inner formes, and skeleton B outer formes, except for an exchange of formes in sheet F. A double reversal of their order therefore took place for sheets F and G, and this may indicate a time lapse in the printing which in each case was sufficient to make both skeleton-formes available at the time of imposition of the first forme of each succeeding sheet. By tracing the order of composition from the beginning of the quarto to sheet G, in stints that, in spite of apparent difficulties incurred by Compositor Y in the course of setting E3-E4v, mesh smoothly and on two occasions actually overlap, it can be shown that the compositors may at that stage of the printing indeed successively have had both formes of type for sheets F and G available when the press was ready for them.

Compositor X set the first full sheet and a half of the play.[8] The

shifts at C3 and D3 were regular. Thereafter, Compositor's Y's second stint was due to begin at E3, which is indeed a page of sheet E he set. However, he could hardly have been fully occupied, at least not on work connected with Cupids Revenge, while X was working on D3-E2v; for only six pages at the very most (two each from C(i), C(o), and D(i)), though probably less, could during that interval have been returned to him for distribution. Thus, he would have been ready to set again before X had finished the last page or two of his stint. X may have held a part of the copy that contained the beginning of E3, and so it is likely that a beginning for E3v was determined, and that Y composed E3v and E4 before reverting to E3. E3v-E4 are comfortably set, but E3 is exceedingly crammed, and this seems to point not to a simple handing-down of copy, but rather to faulty casting-off as intervening between the stints of compositors X and Y in sheet E. There are signs on E3 not only of desperate shortage of space — two lines of dialogue are set in italics in the margin, without speech-prefixes, and there is no spacing whatsoever before the entry line of scene (II,5) — but even of possible textual corruption, perhaps cuts in the dialogue.[9] E4v must thereupon have been the last page of sheet E to be set by Y, as is apparent from its relation to F1. E4v is again crammed, though less so than E3. It prints verse as prose, and has 37, rather than 36, lines to the page. On F1, X reappears, and in effect takes over what would have been the second half of Y's stint through F2v. Therefore, Y must have been held up in his composition of sheet E, especially in E3 and E4v, and whatever cause prevented him from continuing into sheet F must have been foreseen, too, so that X got in ahead of him and started setting F1.

Y returned at F3 and worked into sheet G, X once more prematurely relieved him for two pages, G1v and G2, and Y completed sheet G. This irregularity of stints in sheet G may be related to the order of printing. If the press, due perhaps to a delay caused by Y's difficulties over E3-E4v, was still occupied with sheet E while composition progressed through sheet F, then both formes of sheet F would have been available for imposition into what happened to be a reversed order of skeleton-formes when the press was ready to print sheet F. The pattern of composition in sheet G (reinforced by the pattern of distribution of X's type, where type from F(i)(o) is not distributed until after H4, and before I2v) may, by analogy to the compositorial shifts in C3-D2v of Hector, indicate just where the compositors attended to the make-ready of the formes of sheet F. Again, the composition of G would have been completed by the time the press was ready for this sheet, and the order of the skeleton-formes could be once more reversed. Correspondingly, brief stints of composition once more alternate in sheet H. In addition, compositor X fails to distribute enough type before beginning to set H1; it will be shown that his cases are depleted by the end of page H2. These facts may indicate that efforts were concentrated on getting sheet G through the press, and, as compositor X entered Y's stint to set the single page H4, forme H(i) may have followed at the earliest opportunity.

That H4 was set by X for the express purpose of completing H(i) swiftly, can only be surmised. In sheet K, however, there is strong evidence that compositor Y entered X's stint for the length of one page to aid in the early completion of one forme of the sheet. Compositor Y ends his last full stint on Cupids Revenge in the middle of page I2v. The composition of the remaining 18½ pages of the quarto is left to X, except for page K4v. Two-skeleton printing, however, continues through sheet K. But it can be shown that skeleton B was abandoned early precisely because the full collaboration between the compositors had effectively ceased after I2v, and an order of printing in the last two sheets can be determined which resulted in a major disturbance of the text.

When Compositor X embarked upon his long final stint from I2v, line 15, to the end of the play, he did not have the advantage of a full case of type. He had, on the contrary, accumulated a considerable backlog of undistributed type, and, as is evident from the pattern of type-substitutions which will be discussed in greater detail below, he distributed all the wrought-off type available to him at ever shorter intervals: his four pages of sheet H between I4 and I4v, his two pages

of D(o) after K2 and into K3, and his two pages of I(i) before K3v. I(i) had thus already come off the press, and I(o) was presumably being printed, while the type-setting had advanced no further than K3. Faced with the need to distribute type before setting K3v, X must have recognized that he by himself would not soon have another forme ready. K(i) was only half set; but of K(o), three pages were in type, and this forme could next go to press if its fourth page, K4v, was set out of sequence, and ahead of time. For this purpose, compositor Y's assistance was once more secured, and the copy was cast off accordingly. K(o) was ready when I(o), printed in skeleton B, came off the press, and K(o) was imposed in the same skeleton. Skeleton A, after printing I(i), had apparently become immediately designated to print K(i); and it seems as if X, to retrieve his pages from I(i) for distribution, and to make room on his work-bench, had picked I3v and I4 out of skeleton A and dropped the two pages of K(i) he had finished, K1v and K2, in their places. For the skeleton turns between I(i) and K(i), and the running-title from I3v goes to K1v, that from I4 to K2.

But K(o) in skeleton B undoubtedly preceded K(i) through the press, and skeleton B, and thus two-skeleton printing, was thereupon abandoned for the last three formes of the quarto, K(i), L(i), and L(o). This assumption is confirmed by the typographical appearance of the text in X's pages K3v-K4. A comparative survey of X's and Y's pages in both Cupids Revenge and Hector shows that the pages of X generally accommodate less text and are the more liberal in their spacing.[10] But this distinction does not apply to pages K3v and K4. Both have 37 lines of text instead of the 36 lines standard in this quarto, and all but the first three lines of K3v (continuing correct verse from the preceding page) and two-thirds of K4 are set in solid blocks of lines to the full measure. Short exclamations by characters in dialogue are squeezed in where preceding paragraphs carry over only a word or two into their last lines, and even stage directions are not given lines by themselves. A relief to the eye comes only at the bottom of K4, where correct verse lineation resumes with the beginning of scene (V,3). But the entry of (V,3) is preceded by another entry, "Cornets. Cupid from aboue.", which, in analogy to previous entries of Cupid, should be followed by a speech. Cupids Speech which belongs here is in fact printed on L3v, following the final Exeunt. of the last scene. It was only by leaving it out of its correct position, therefore, that X could find

enough room on K4 for the thirteen lines of text of scene (V,3) that would join correctly with their continuation on K4v. Thus, K4v must have been previously set, following a casting-off of copy that was hasty and conspicuously faulty. The text for K4v must have been determined before X entered on K3v, for he begins to make his adjustments as soon as line 4 of that page. Therefore, it is also likely that K4v, in its forme K(o), was going to press early. For, as X discovered the faulty casting-off immediately on resuming composition, yet struggled for two pages to make ends meet, the alternative course of resetting K4v may not have been open.

From the evidence established thus far we conclude that the speed and intensity of production of Cupids Revenge varied appreciably through the sheets of the quarto. Beyond what has been said, any amount of statistical detail could be brought to bear on a strengthening or a modification of these preliminary findings. A survey of type reappearances, for example, would be suited to supplementing them most fully, by minutely tracing each phase of composition, printing and distribution, and would thus also do equal justice to both compositors. Such a survey has not been undertaken. We shall therefore have to forego the closer view of Compositor Y which only it would have given, and shall confine ourselves to tracing the work pattern of X by scrutinizing the type substitution he practiced in I's, P's and L's.

If a view of one compositor's work in isolation is to be meaningful, or even at all possible, it must be assumed that each of the two men set from, and distributed into, his own cases. At least one piece of evidence, the occurrence of a tailed italic m in Y's pages only, supports this assumption from the outset. Then, to arrive at standards for subsequent counts, it will have to be decided whether X started out with a full case, and whether his first sheet and a half, pages B1-C2v, were set with or without intermittent distribution. Here, the early occurrence of type substitutions for upper-case roman P provides the necessary evidence. There are five roman capital P's in sheet B, three in B(i) and two in B(o). In C1, C1v, and C2v there is one substitution each by an italic capital P for correct-fount roman, and, in addition, four small-cap roman P's or upper-case roman P's from another fount, in B2v, and one each in C1 and C1v, which are clearly also used as substitutes. The occurrence of a substitution as late as C2v, by which time distribution of the first wrought-off forme of B could theoretically have taken place, suggests that there was no distribution and re-use of type from sheet B before the end of C2v. This, in turn, implies that X must have set his first long stint of twelve pages from a full case. The

one surprise element in these observations is that the case from the outset should have contained no more than five upper-case P's of pica roman no. 6.[11] Subsequent counts in Cupids Revenge, however, and the recurrence of similar conditions in Hector, support this calculation.[12]

In his first twelve pages, X set a total of 113 I's (see table). If, before resuming composition at D3, he distributed B(i), he regained 29 I's, and he used 29 I's in D3-D4. As an irregular pattern of intermittent substitution starts at D3v, line 27, it is likely that the last of the pages of B(i), page B4 with 7 I's, was distributed piece-meal during the composition of D3v, line 27-D4. Altogether, the substitutions show that more pieces of roman I would have been needed than were supplied by the distribution of B(i). This suggests that the total of 113 I's in pages B1 through C2v had in fact drained X's supply to the last piece, and therefore constitutes the entire number of capital roman I's his case held.

An analysis of the numerical occurrence of correct-fount and wrong-fount I's suggests the following sequence for the progression of composition and distribution in the pages of Compositor X from the beginning of the text at B1 through K4:

                           

before	distributes	gains	uses	in	spare
D3 }	B1v, B2, B3v}	22}	22}	D3-D3v}
during D3v-D4}	B4}	7}	7}	D3v-D4}
D4v	B2v-B3	22	19	D4v	3
E1	B1	7(+3)	10	E1, to line 26
E1v	B4v, C1	10	10	E1v, to line 33
E2, line 17	C2v	18	13	E2, l. 17 thru E2v	5
F1	C(i)	27	27	F1-F2	5
F2v	E(i)	13(+5)	17	F2v	1
G1v	E(o)	20	14	G1v-G2	7
H1 }	}		{8	H1-H1v
before and during H2}	D3v}	11(+7)	{
	}		{	H2	1
H4	G(i)	14	11	H4	4

       

I2v	D4, F(i)(o)	48(+4)	56?	I2v, line 15 -- I4v, line 14
I4v, line 26	H1-H2, H4	28	28	I4v, line 26 -- K1v, line 8
K2v	D(o)	33	31	K2v-K3	2
K3v	I(i)	23(+2)	25	K3v-K4

The initial stages in this reconstructed sequence show composition and distribution in a but slightly precarious balance. Compositor X worked into his second stint D3-E2v almost the entire supply of I's he successively retrieved from formes B(i), B(o), and C(o), and there are five spare I's in his case when he ends E2v. As a wrong-fount I does not occur again until the last line of H1, the order of X's subsequent moves can only be hypothesized by assuming a postponement of the distribution of D(o), with 33 I's tied up in pages D3 and D4v, until some point after completion of sheet H. By evidence from the reappearance of very few pieces of roman P and the absence from the pages of the quarto for a very long time of the four P's tied up in D(o), this hypothesis will presently be substantiated. The counts of I derived from pages F1 to H1, and the peculiar substitution pattern found on pages H1v-H2, furthermore suggest that forme D(i), too, was left standing longer than might have been expected, and, with the distribution of D3v, only began to be drawn on during the composition of H1-H2. This leaves formes C(i) and E(i)(o) as sources of supply for the necessary type used in the two stints F1-F2v and G1v-G2, and the sequence of their distribution is inferred in a manner that allows for their successive return from the press.

Only minimal distribution, or none at all, can have preceded X's initial stint in sheet H. After setting 8 I's, seven of which were left over after G2, he runs out of the sort at the bottom of H1 and sets 16 substitute italic I's through H1v, and seven more, interspersed by 4+4 correct-fount I's, on H2. There is one solitary correct-fount roman I on H1v. The total count of roman I's in evidence on pages H1-H2 is just one short of the sum of his spare pieces left over after G2 plus the number he could have retrieved from D3v. The assumption that it was D3v he fell back on is strengthened by the appearance of two correct-fount roman P on H1v, one of which could have come from D3v, while the other had become available through distribution of E(o) before G1v and not used on G2, where there is, instead, a substitution. Judging from the substitution pattern, D3v must have been distributed piecemeal during H1-H2, although the precise course of distribution cannot be determined. If our analysis is correct, there is only one

roman I left in X's case at the end of H2. At the same time, the italic L's which are so much in demand for the speech-prefixes in this play are fourteen times out of fifteen substituted by roman L's in pages H1v and H2. It seems not unwarranted to believe that X was incurring a very acute shortage of type. Yet the beginning of H2v is free from all substitutions. If Compositor X did set the first fourteen lines of H2v, as the spelling evidence on the page seems to indicate, he can have done so not from his own, but from Y's cases, until Y was free to work from them in person, from line 15 onwards.

The next point of firm orientation is the complete absence of roman I's in K2, preceded by 12(+2) substitutions in K1v, and another six, flanked by three and six correct-fount I, on I4v. The number of I's from I4v, line 26 (last letter) to K1v is 28 — or exactly the number contained in the pages of sheet H which were set by X from his cases. The interval between the end of H2 and I4v can be accounted for, albeit with a minor uncertainty in relation to page I2v. Here, the spelling evidence once more suggests a division of the page between compositors Y and X. If X entered at line 15,[13] and if he set his section of the page from his own case, the number of correct-fount I's he used up to I4v, line 14, was 56, or four in excess of the number available after a distribution of D4 and F(i)(o). If, rather than taking over at line 15 as is suggested by the typographical change from prose-as-verse to genuine prose setting, he actually entered at line 20 where his own spelling characteristics truly resume, the number of correct-fount I's to be accounted for is still 54, or two more than our counts make available out of the 113 I's which until now have been repeatedly confirmed as the total number of I's X was using. Moreover, the possibility remains that, although divided between compositors, page I2v was yet composed from only one set of cases, those of Compositor Y, just as page H2v had been. If so, X would, by returning to his own cases with page I3, indeed have been using six pieces less than his total by the time the substitution of I's begins in page I4v. As forme I(o) is distributed only after the composition of K4, and as there are no substitutions of I's and P's in sheet L, and those of L's prove unreliable, there is no way by our present method of testing how page I2v was dealt with.[14]

It seems, however, that this uncertainty does not invalidate the subsequent analysis of composition and distribution in pages I4v, line 26, to K4. All of sheet H is distributed, as suggested, before composition on I4v is completed, and, although the supply of I's runs out again during K1v, the cases must otherwise have been full enough to sustain uninterrupted composition to the bottom of K2. Before K2v, D(o) is — at last — distributed, and of its 33 I's, 31 are immediately in demand on K2v and K3. To be more precise: the absence of correct-fount P on K2v, although four P's would have been released from D(o), and the subsequent appearance of two of them on K3, suggests that not all of D(o) was distributed before composition started on K2v. Rather, I believe that D3 and D4v were lying ready for distribution side by side on the compositor's work-bench; distributing the type successively on the whole width of the two pages, he got as far down as line 25 on each page before starting to compose K2v. These 2 x 25, or 50 lines of type would have yielded 9 plus 15, or precisely the 24 I's needed for K2v. The mixed substitution pattern of I's on K3 would thus be accounted for by a piece-meal distribution of the remainder of the two pages during the composition of K3, and the four P's from D(o) would indeed have been retrieved only during this latter operation, so that none had been available for K2v, but the two P's on K3 could be set correctly. By distributing I(i) before beginning to struggle with the composition of K3v-K4, X then gained another 23 I's to the two that were left over at the end of K3. 25 I's are used in K3v and K4, and the supply is once more drained to the last piece.

At this juncture, skeleton B is abandoned. It is now evident that at the end of X's stint in sheet K a considerable amount of distribution would necessarily have to intervene before the composition of pages L1-L3v could be begun and completed. It is likely, therefore, that X was at this point left to finish the setting of the text at his own

pace, and that no further effort was made to turn the entire resources of the shop to pulling the quarto through, as had apparently been attempted before, at least during the production of sheets H to K. Pages L1-L3v successively accumulate a total of 71 I's, which are easily within the range of Compositor X's case, if fresh supplies were always retrieved in time. This must have been possible, for there are no more substitutions in the last pages. But, consequently, no further statements can be made about stages of distribution and composition in L1-L3v.

The reconstruction of Compositor X's work pattern through Cupids Revenge hinges ultimately on the hypothesis of the delay in distribution of forme D(o). Our analysis will have some claim to representing the true order of events, if within the pattern of substitutions and reappearances of the letter P, proof for this delay can be furnished. It was argued before that X started out with only five roman capital P's in his case, which were used on B1v (P1,2), B3v (P3), and B4v (P4,5). If, according to our earlier assumptions, B(i) was distributed before D3, the two P's in D3 and the one in D3v were P1-3 retrieved from B1v and B3. But, with B4v according to the same assumptions not distributed earlier than between composition of E1 and E1v, the two P's appearing on D4v cannot be P4,5. They must therefore be two additional pieces, P6 (on D4v, line 29), and P7 (D4v, line 32). To prove that this is not mere speculation, two of the copies of Cupids Revenge in the British Museum[15] have been scrutinized closely in an attempt to trace the actual types. It has not been possible to identify beyond doubt each single one of them in its repeated occurrences. But, by good fortune, P6 is unmistakeable by a characteristic damage to the lower part of its loop. It can be positively stated that, while used twice in Hector (Hector B4v and G1v), P6 does not reappear in Cupids Revenge. When, therefore, on distribution of D (o) during the composition of K3, four P's were gained but only two were re-used, P6 cannot have been one of them. Furthermore, the P in K3, line 30, can with great confidence be identified as P1; it prints first in B1v, line 7, then in D3, line 27, but nowhere in the interval between D3 and K3. In combination, these observations seem sufficient proof that D(o) was indeed not distributed before K2v/K3. Of the remaining pieces, P4, too, seems sufficiently characteristic to be traced from B4v, line 21, to E1v, line 17, to F2v, line 1, to I2v, line 35. This would confirm that F (i)(o) were distributed before I2v, but not earlier, and B4v was distributed between E1 and E1v. From the evidence of the reappearance of P4 and P5 in E1 and E1v, the exact point where

distribution of B4v must have been undertaken can in fact be closely determined. If, as said in relation to the shortage of I, B1 was distributed before E1, and B4v together with C1 before E1v, no correct-fount P for E1, line 32, would have been available. However, as the one wrong-fount I on E1 occurs in line 26 of that page, the necessity of substitution at this point may have induced Compositor X to relieve immediately the impending shortage of I's and distribute B4v before even completing composition of E1. By distributing B4v somewhere between lines 26 and 32 of E1, he would have retrieved P4 and P5, using P5 on E1 and P4 on E1v.[16]

No further individual types can be identified and traced. But the general pattern of reappearances and substitutions of the letter P is such that, almost without fail, correct-fount P's reappear in the pages of Compositor X whenever, and as soon as, pages or formes which themselves contained correct-fount P's are supposed to have been distributed. One exception has been mentioned: the P which was regained by distribution of E(o) before G1v, and could therefore have been used in G2, does in fact not show again until H1v, where it joins the P which was retrieved from D3v. The second irregularity in the pattern occurs in K1/K1v, where there are three correct-fount P, although the preceding distribution of H(i)(o) provided only two. However, an examination shows quite clearly that the two pieces in K1, line 8, and K1v, line 11, must be the ones retrieved from H1v; the latter is followed by one substitution and then the third correct-fount P. This is an identifiable piece, battered and knocked about; it can confidently be called 8, and it appears nowhere else in X's pages in the quarto.

The hypothetical reconstruction of the phases of composition and distribution in the pages of Compositor X in Q1 of Cupids Revenge, which was based on the evidence of type shortages and the substitution pattern of italic I for roman I, has thus been confirmed and proved by a matching pattern of P for P, and by the identification of reappearing single exemplars of upper-case roman P. In conclusion, it must for the sake of completeness be added that the substitution pattern due to shortages of italic L in speech-prefixes is not fully consistent with the work-phases determined in the preceding analysis. However, the patterns are compatible for X's pages in sheets B-G, and in sheet H as far as H2. Thereafter, a consistently correct distribution ought at each stage to have regained sufficient numbers of correct-fount italic L to

make all those substitutions unnecessary which do in fact occur in H4, and in sheets I, K, and L, where the speech-prefixes in L- are so relatively few in number. We can account for this inconsistency only by assuming a disproportionate amount of foul-casing in the distribution of speech-prefix type-material.

Compositor X in his progression through the quarto, then, is surprisingly often seen to work from very low supplies of type, and even to distribute while composing. An inference of haste and undue pressure is tempting, and at certain times such factors perhaps contributed to setting the pattern for his work. However, they cannot be claimed to govern it, for, basically, there is nothing to be gained by distribution in small installments of a page or less and frequent alternation between composition and distribution, as against distribution by forme or half-forme and composition of correspondingly larger units. If one compares the situations in Cupids Revenge and in Hector, where the absence of type substitution suggests that at no time any acute shortage of type arose, they differ largely in only one respect: in Hector, neither X nor Y ever overreached himself as X did once in Cupids Revenge, when, by uninterruptedly composing the full length of the first twelve pages, he sacrificed from the outset the advantage his full cases had initially given him. After this beginning, what the pattern of type shortages and substitutions reveals may essentially be but a mildly modified 'normal' working process. The lack of corresponding evidence prevents an analysis of Y's work phases for comparison. But the expectation would be that they follow a very similar course, and that Y neither in Cupids Revenge nor in Hector incurs type shortages simply because he never drains his supplies to the same extent as X did in his first stint on Cupids Revenge. His manner of distribution might well be found to be very similar to that observed in X's stints, for it would ultimately be dictated by the system of composition adopted. This system was, of course, only by a certain degree of abstraction recognized as the underlying 'ideal pattern' of collaboration between compositors X and Y — but no more abstraction is required to see that distributing just two pages at a time was basically the habit that X adhered to. Indeed, as a habit it may have been so ingrained that perhaps it never occurred to him to make an early effort at retrieving more wrought-off type than usual, although he had in fact more type standing in the early formes of Cupids Revenge than was warranted within the limits of his total supplies. Thus, when getting into sheets F and G, he did not distribute in order, but turned to his most recent wrought-off type from sheet E. He simply skipped his pages in D and

temporarily left them standing until he had no choice but distributing them when only they could help him out of his shortages. Their distribution would, of course, then require additional time, and, moreover, there are signs that Y was not always available to do his expected share of the work on Cupids Revenge. The stretch of the quarto thereby affected is approximately sheets H-K, and here, the factors of 'haste' and 'pressure' can to some extent be supposed to have influenced the work. But, as we have seen, the situation was remedied by a temporary recall of the other compositor, and by a basic change of pace in the final reversion to one-skeleton printing.

On the whole, I believe that no strong argument can be advanced for haphazard or unsound compositorial workmanship in Cupids Revenge. This must be emphasized before conclusions are drawn from a comparison with Hector, which appears in so many ways as the superior quarto, and which certainly is a more successful printing job. It has been recorded above that the composition and the printing of Hector are performed in a manner very similar to that observed in Cupids Revenge. The cooperation of the two compositors is slightly more regular than in Cupids Revenge, and two-skeleton printing is in Hector consistent throughout. Beyond that, such irregularities and blemishes of production as have been observed in Cupids Revenge, provide only slight leverage for bibliographical analysis in Hector. Compositor X's setting of Hector, page E4, for example, soon after his previous stint had ended in E2v, may be another case of rushing the completion of a forme. And again, twice towards the ends of stints of both compositors — in B2v for Y, and in E2-E2v for X — there are signs of casting-off for overlapping composition. Among these signs are a few brief prose-settings of verse to save space. But in the entire quarto, there is no major mislineation, except in the last scene which stretches over H4v, line 13 - I1v, and which, in the hands of Y who is finishing the quarto, becomes thoroughly garbled in its lining. But the over-all picture is one of an ease of composition and typographical disposition in the pages of Hector which is truly superior to almost anything seen in Cupids Revenge.

In conclusion, the marked differences which exist between the quarto of Hector and that of Cupids Revenge may be largely due to a difference in the quality of the manuscripts. In Hector, there is evidence that both compositors are equally capable of giving a page a satisfactory lay-out, of calculating page-units with fair accuracy when the need for casting-off arises, and, above all, of consistently setting correct verse and prose. The copy for Hector which was responsible for

these results appears in all respects to have been clear and unambiguous in its textual arrangement. But the copy for Cupids Revenge seems to have had none of that clarity. Both compositors constantly show signs of struggling with the text and lineation. Prose is found to be set as verse, verse as prose, or verse to be mislined, on innumerable occasions. Both compositors apparently have much the same difficulties with the copy, although X is more resourceful and determined to solve them. His pages contain the larger percentage of what must be regarded as correct passages in either verse or prose; thus, for example, after Y has been setting all but the very end of an entire scene, (II, 6), in straightforward prose through pages E3v-E4v, X takes over on F1 in midcourse of a speech and ends it in verse.[17] But X does by no means avoid all the pitfalls of the text. He is almost as likely as Y to be thrown off by a half-line at a change in dialogue and garble the lineation of the subsequent verse passage. And in both his and Y's pages, there are, occasionally, very curious sections where what should clearly be prose is split up into irregular lines that have upper-case initial letters, just as if they were verse. Passages of this sort may show a compositor's bewilderment at its highest and reflect his retreat to a line-by-line reproduction of the copy; this, at least, is a possibility suggested by the appearance of a page such as I2v where X sets prose out of confidence that he is dealing with a prose scene, whereas Y at the top of the page may well be representing the original lining (and, as in the case of the abbreviation 'Gentl:', even some word forms) of his copy. If this were to indicate that Y, for better or for worse, was generally more inclined than X to take the copy at its face value, then, in assuming that he himself did the casting-off necessary to determine the beginning for his page K4v, one would recognize this casting-off as being not excessively faulty. If from the manuscript he did not gain a clear picture of which proportion of the text for K3v-K4 should be prose and which verse, the miscalculation would seem to be more or less within the limits of expectation for Compositor Y and his tendency to crowd more text into a page than X generally does.

These last few considerations taken together should act as a check on the inclination to accuse Compositor Y of unusual ineptitude in view of his work on Cupids Revenge. Within a larger context, they should be incorporated into the body of observations and conclusions

established in this paper to give foundation to a consistent hypothesis about the nature of the copy for Cupids Revenge. To argue such an hypothesis in every detail is inseparable from closest editorial attention to the play.[18] The object of this paper has been a more narrowly cirmcumscribed study of the composition and printing of Cupids Revenge Q1. To complete a full analysis of its compositors, it yet remains to turn the attention to X and Y as their spelling habits reveal them. By again setting off their usage in Cupids Revenge against that in Hector, and in other work where and when the need arises, I hope, in a second article, to consolidate their characteristics of orthography in a range sufficiently wide to allow conclusions to be drawn about copy spellings and, perhaps, even tentative statements to be made about authorial characteristics as they may be seen to survive into a quarto text.

Notes