University of Virginia Library

A set of variants preserved in a single early copy of The Roaring Girl (1611) has managed to elude editors and students of the play. These variants constitute the most significant of those recorded in surviving copies of the 1611 quarto. Although the copy now in the Robert H. Taylor Collection located at The Firestone Library, Princeton University, is included among those collated for F. Bowers's edition in The Dramatic Works of Thomas Dekker (Cambridge, 1958, rev. 1966), vol. iii, the variants unique to this copy are not noted. Moreover, close examination of copies of the 1611 quarto has revealed numerous instances of speech prefixes transferred whole and in part from one forme to another. This feature sheds new light on aspects of the printing of this play and may also have implications for other one-skeleton play texts printed by Nicholas Okes and possibly by fellow printers as well. In addition, a previously unrecorded copy has come to light: that in the Beinecke Rare Book and Manuscript Library, Yale University.[2] This copy contains no new variants, and press corrections conform to those evidenced in the other known copies. A note listing the disposition of its variants in relation to those in other copies is given below.

The unique variants in the Taylor copy are in the form of altered speech prefixes and direction line on sig. D4r. The changes to the speech headings are interestingly restricted to character identification only; the Mist. / Maist. sequence, including the arrangement of swash and regular italic capital Ms and ligatured sts remains exactly as in its original setting. Apart from these, the Taylor copy agrees with variants already noticed in other copies. For convenience, the following (with the addition of CtY=Yale) is designed to conform to the arrangement set out in Bowers's edition of The Dramatic Works of Thomas Dekker:

• SHEET D (inner forme)
• 1st stage corrected: BM 1, NLS, MB.
• Uncorrected: BM 2, Bodl, CCC, Dyce, CSmH, CtY, DFo, Pforz.
• sig. D4r
• II.i.367 Hogſ-/den] Hogſ-/dcn

---

160




• 2nd stage corrected: Taylor.
• II.i.364 Open.] Gal.
• 365 Gal.] Tilt.
• 366 Tilt.] Gal.
• 370 Tilt.] Gal.
• 371 Tilt.] Gal.
• catch word M.Gal.Come] Come

Possibly there is a missing stage in this sequence which would render the first stage corrected, the second stage corrected and so on. The instance provided by the Pied Bull Lear of a corrected comma may point to a similar circumstance here.

Several points helpfully establish the variant state of sig. D4r as the corrected and authoritative one. In the first change at l. 29 (II.i.364), the abnormally small space left between the speech prefix and the speech argues that Gal., has been replaced by the longer heading, Open., and a smaller space substituted for the quad between the new prefix Open. and the text. In addition, the variant arrangement splits between Tiltyard and Gallipot a speech formerly allocated to Gallipot alone (II.i.371 ff.; cf. the accompanying figure), and the variant catch words at the foot of sig. D4r explain this

alteration. In the original setting the catch word, 'Come', is the first word of the initial speech on sig. D4r, but this omits the speech prefix: presumably the compositor, under the misapprehension that the speaker was already Gallipot, saw no reason to repeat it. The failure of correspondence between catch word and the first words on sig. D4v, the prefix Maist. Gal. (II.i.372), which precedes this speech, has in the absence of an alternative reading remained a curious anomaly.[3] Moreover, in the uncorrected state, the last speaker on sig. D4r is Master Gallipot, apparently making redundant the sig. D4v speech prefix. The reallocation of the final sig. D4r speech (II.i.371) to Master Tiltyard in the corrected version makes sense of the change of speaker signalled by the speech heading, Maist. Gal., at the top of sig. D4v. Forty other instances of speech-heading catch words appear in the text and none of these includes the initial word of the character's speech: thus the corrected state appears to add the prefix to the previously existing catch word in an evident attempt to produce agreement between the direction line and the initial line, 'Maist. Gal. Come . . .', on the following sig. D4v. Also instructive is the case of sheet K of the Pied Bull Lear, in which a catch word was corrected in one forme to match an alteration in the other:[4] a correction to a forme first through the press which generated a disagreement would seem most unlikely. As Bowers has noted (Textual Introduction, vol. iii, p. 3), such corrections as do appear in the course of printing could have been made without reference to the manuscript. In the present instance, however, mere good sense could not have produced the prefix alterations; on this occasion correction may have been prompted by consultation of the printer's copy, or perhaps authorial intervention.

As for their relative dramatic merits, the corrected readings on sig. D4r, allowing for the literary slightness of the alterations, are superior in several ways. The Tiltyards get short shrift earlier in the scene: this is Master Tiltyard's first appearance in the play and Mistress Tiltyard has been allowed a scant seven lines.[5] Here both are given brief speeches and business which help to justify their presence. Mistress Tiltyard, who is silent in the uncorrected version, is given a line, presumably to do with the labour of shutting up the shop—and in this the pair may be intended as objects of amusement since Master Tiltyard in his thorough preoccupation with the dogs leaves all the work to his wife. This vignette does not really square with the Gallipots, the former possessors of these lines, since, on the strength of his behaviour later, the solicitous Master Gallipot would be more likely to spring to his spouse's aid. But the special appeal of the reallocation to Mistress Openwork of the speech at l. 29 (II.i.364) almost single-handedly confirms the authority of the altered speech headings. Her tiff with her husband over Moll earlier in the scene and her new involvement with Goshawk prepare the ground for a reply of dramatic interest to Tiltyard's innocent question which, if given to Mistress Gallipot, is barely capable of rising above the merely informational.[6]

A possible explanation for at least the first three of the errors may be found in the final word of Master Tiltyard's speech at l. 27 of sig. D4r where Openworke occurs as the sole occupant of a text line at the left side of the page among the speech prefixes. If, as the text was being set, the compositor mistakenly construed this as a speech heading in a manner similar to eye-slip (the italic headings would appear to have been set up separately in a set of quick exchanges such as this, as evidence adduced below suggests), the immediately following four prefixes may be seen to accord with those in the corrected version (i.e. the corrected sequence runs Open., Gal., Tilt., Gal., Gal.; the uncorrected sequence skips the first Open., and proceeds, Gal., Tilt., Gal., Gal.) Two lines given to the Gallipots then remain in their original setting (ll. 33 and 34; II.i.368-9). If this explanation is correct, the gender discrepancy between the two sequences at l. 29 may further indicate that the Maist. / Mist. sequence, which is correct in the earlier state, was set independently from the series of character abbreviations: again, possibly an error created by the kind of transfer noted below.

The explanation advanced for the first few lines fails to account, however, for the errors in the final two speeches on the page, also given in the uncorrected state to the Gallipots. Confusion of some sort clearly reigned, but in this instance it would appear to have been of a different order and its precise nature remains a matter for speculation. Possibly some doubled-up or otherwise disordered arrangement of these short lines in the manuscript may have produced the appearance of a stint of speeches between the Gallipots.

The set of errors significantly occurs on the final page of the inner forme where the discrepancy between the catch word and the first line of sig. D4v, the last page of the gathering, would not have been apparent if composition of the outer forme were incomplete when printing of the inner had begun. Such a circumstance would not be out of keeping with regular printing-shop practice, since the inner forme could be imposed and then printing begun after seven pages had been set. But the conditions of composing and printing The Roaring Girl appear to have been particularly tight, as reflected in procedures designed to spare type. The relatively small number of the early quartos which has survived does not encourage confident speculation, but the single instance of the corrected state among the twelve known copies suggests that correction occurred at a late stage. Just how late is problematic: evidence of one in twelve produces an unlikely figure of just over 8 per cent. of the white paper remaining. Writing of the Lear quarto, Greg observed, 'if the corrections came in when 80 or 90 per cent of the pulls had already been made, it would hardly be worth while altering the type' (op. cit., p. 192). But perhaps the errors were deemed serious enough to warrant correction,

albeit near the end of the run: these are without doubt the most significant corrections recorded in this text. Consonant with late correction, the disjunction between sigs. D4r and D4v could well have escaped notice through much of the printing since sig. D4v may not have been composed until almost all of inner D had been machined. A late discovery of the catchword discrepancy and the prefix errors would further explain how an earlier alteration on sig. D4r could have been made independently[7] —possibly as a follow-up to an earlier, and now lost, stage of correction, as suggested above.

Authorial involvement in the correction process cannot be entirely ruled out and could account for a late-stage correction, but no evidence for it exists elsewhere and the recognition of a failure in catch-word correspondence offers a plausible explanation which does not require either dramatist's presence. As a compositor's standard reference point, a catch-word error would naturally be more likely of notice than an internal error. Since no correction attended the beginning speech of sig. D4v, the sig. D4r catch-word alteration is not consequential, and hence does not exactly correspond to the instance of a corrected catch word on sig. K4r of the Pied Bull Lear. Further, the two-skeleton printing of Lear makes unavailable to the printing of The Roaring Girl the usual system of proof-correction which apparently brought that error to light, since in one-skeleton printing the formes would normally be corrected independent of each other. Whether the catch-word error and mistaken speech prefixes were found by accident or as a result of some regular procedure is a matter for speculation, but, barring authorial intervention, the mis-matched catch word offers the readiest logical explanation for its detection.

The use of recurrent type as bibliographical evidence is now well established; The Roaring Girl provides numerous instances not simply of recurrent individual letters but of repeated whole or partial speech headings. The clearest and in many respects the most interesting example touches also on some textually problematic matters—matters which were originally the chief interest of my investigation and which led to observation of the re-used speech prefixes. Contrary to normal expectations, evidence provided by sheet I demonstrates clearly that the inner forme had completed its printing run before even the third page of outer I had been fully set. With its succession of relatively short speeches by the Mistresses Gallipot and Openwork and later, Master Gallipot, sheet I takes a heavy toll on the stock of italic capital Ms. In inner I, which was first through the press, there are forty-five occurrences; this pattern characterizes much of outer I also. As a result, in the course of setting outer I, the stock was depleted and roman Ms substituted for italic. At l. 24 on sig. I3r, however, the italics resume: a condition which strongly suggests that a supply had just been released. That the run of inner I had been completed, and the forme, if not immediately distributed, was at least cannibalized for its store of italic Ms in the course of setting sig. I3r

of outer I, is confirmed by examination especially of faulty type in the speech headings after the recommencement of italic capitals. Of special interest, not only were the italic capitals transferred from inner to outer I, but in several instances the whole heading or a component comprising several letters was preserved intact. The manner of re-using previously set speech prefixes seen in these transfers and others noted below raises the possibility of a regular, though perhaps informal, procedure which could have figured in the run of sig. D4r errors. Certainly, a system which favoured retention of speech headings in constant use through several formes over immediate distribution makes perfect sense. As Bowers has noted, Moxon makes no mention of the practice of transferring running titles from forme to forme,[8] so his failure to mention re-use of speech prefixes in the specialized production of play texts is not surprising. Three speech headings on sig. I4r and two on sig. I1v which clearly recur on sigs. I3r and I4v respectively serve to demonstrate the procedure (numerous others are probably involved but the uniformity of their letter forms precludes certainty): at l. 24 on outer-forme sig. I3r, Mist. Gall., is a direct lift from l. 24 of inner-forme sig. I4r including the undotted i and the damaged G; at l. 26 on sig. I3r a nick in the right base of the initial M and the particular orientation of the double bent ll establish this as the same prefix as that at l. 32 on sig. I4r, but with the a removed to make Maist. into Mist.; and at l. 29 the damaged n and the st ligature correspond to those seen in Maist. Open. in l. 1 of sig. I4r; in addition, a clear transfer from sig. I1v, l. 13 of a complete heading with a defective M occurs at l. 28 of sig. I4v; another at l. 37 of sig. I4v is possibly a repeat (with an a added) of that at l. 30 of sig. I1v.[9] All instances of prefixes drawn from inner I naturally involve the reset state of that forme (Dekker, vol. iii, pp. 4-5). Significantly, depletion of the italics did not interrupt the process of composition.

Of additional interest in sheet I, five speech prefixes on the first page of the unreset outer forme have clearly been taken directly from the final page of the outer forme of H; in this case the sequence is identical in both and the prefixes have been preserved as integral units.[10] Numerous other recurrences

of traceable broken type similarly reveal a practice which corresponds to that noted by G. W. Williams in 'Setting by Formes in Quarto Printing' (SB, 9 [(1958)], 46 ff.), in which he investigated a like shortage of italic Ms in Creede's printing of Menechmi (1595). The important difference in the case of The Roaring Girl is that the exceptional frequency of italic Ms is not consistent throughout the quarto. The transfer from inner I is accordingly a recourse available to Okes's compositor (only one appears to have been involved in the setting of sheet I) which is not to Creede's.

Speech-prefix transfers are not unique to sheets H and I; other instances are observable in various parts of the text in which the essential condition of the continuance of a prefix into an adjacent forme is present and in which the fortunate repetition of defective type makes detection possible. With the exception of sheets D, K, and L, however, re-used headings are visible only in the formes of adjacent sheets rather than between the formes of a single sheet.[11]

Important in assessing the significance and implications of speech-prefix transfers is establishing their relation to compositorial stints. Although G. R. Price's attempt to differentiate compositors' shares is less than satisfactory, with some significant adjustments and modifications, it is likely to provide as serviceable a division as the text will allow.[12] Determining compositorial shares in this text is a vexed and almost unassailable problem. By way of illustration, among the more convincing traits isolated by Price is the appearance of 'I'le' with and without apostrophe. The Game at Chesse manuscript, though some dozen or so years later, shows that Middleton was himself inconsistent in the use and placement of apostrophes in such contractions. The apostrophe may be a genuine compositorial trait; on the other hand, it may represent a faithful rendering of the printer's copy of the play, inconsistency included: the likelihood of fossils of Dekker's or Middleton's presence in the manuscript (the question of who was responsible is far from clear; see my 'Some Textual Notes', pp. 338-339), not to mention a scribe's, makes the whole matter infirm. The absence of apostrophe may in a similar way reflect a compositor's generally successful suppression of the manuscript usage; but other explanations, though less satisfactory for the purpose of distinguishing shares, may be equally valid. If the various tidbits of evidence are to be used at all they must be handled with great caution in the interests of avoiding the temptation to regard hints of brick-like forms as foundation blocks.

In his initial plan Price sets out evidence for five compositors: as Bowers has remarked, an unlikely number for a one-press shop such as Okes's (Dekker, vol. iii, p. 3 note). Price's research is not thorough, however, and is at points confused. He omits from his trait allocations instances of trait 'Z' ('Elsewhere the Sir of speech-heads is in italic.' p. 182) in compositor E's stint.[13] As this is the sole distinction between compositors D and E in his analysis, D and E are one and the same. Further, in his proposed conflation of compositors A, B, and C to produce one (A), and D and E another (D), the assignment of pages in the stints of each on several occasions fails to agree with the data provided in the five-fold division given immediately above.[14] In preference to a reduction of the evidence to two compositors

(which may yet have some validity), Bowers favours three, retaining Price's A and joining B and C to make a second (B), and (like Price) D and E a third (D). According to this arrangement, A and B would appear to have set almost all of the early signatures, essentially from sheet B to part of sheet F, in a random sequence of alternation, and D almost all of the remainder, including most of the prelims.[15] According to this analysis, no occurrences of speech-heading transfers appear within A's share, although some which originate in his stint recur in another's: perhaps to be interpreted as further evidence for three compositors. Elsewhere recurrences of speech headings appear in the stints of both B and D, implying that the practice was not merely the idiosyncrasy of a single compositor. Although sheet I provides the only certain example, re-use may be generally related to type shortage or the danger of type shortage; but convenience and possibly speed would presumably figure also.

Sheet F provides a point of special interest since it strains the stock even further, with fifty-five Ms in the inner and forty-four in the outer forme, and does not resort to roman capitals. The case of italic Ms would appear to be capable of meeting the needs of both formes of F but be insufficient for I. Significantly, the two or three compositors' stints meet in sheet F, so the likelihood is that at least two cases were used in setting the sheet and met the heavy demand for italic capital Ms. A second case was clearly unavailable to sheet I which from all appearances was set by one compositor.[16] Type recurrences, whether single or comprising a whole or part speech heading, suggest that through most but not all of the text a single case was used

to compose The Roaring Girl: an unusual circumstance since compositors are generally thought to have set from their own cases.

Other instances of roman substitution in speech headings occur in the quarto but in stints too brief or too irregular to yield an identifiable cause or pattern.[17] Specially interesting, however, is a roman substitution for an italic I on sig. K3r, l. 14; here a distinctive k from the same Iack. Dap. speech heading is repeated from sig. K2r, l. 28, suggesting that the italic Is from inner K ought to have been free for use in outer K, but for some reason were not. The shortage continues into sheet L, so there is no reason to suppose that the outer forme was first through the press. Bowers has remarked on shortage of type as one possible reason for resort to one-skeleton printing:[18] a condition which clearly appears to have prevailed during the setting of this play. Yet since circumstances within the printed text of The Roaring Girl are unable to account for this and other shortage peculiarities, the likeliest explanation would seem to lie in some external draw on the type which helped to create depletions in certain letters: a proposition tentatively elaborated below.

The emergence of roman Ms on sig. I2v when none appear in inner I, coupled with the resumption of italic in sig. I3r in speech headings which have been manifestly drawn from inner I, points to setting by formes as opposed to the seriatim procedure more usual in single-skeleton printing—at least where setting from manuscript is concerned.[19] R. K. Turner remarks in connection with the exceptional case of Thierry and Theodoret (printed by Okes in 1621), however, that setting by formes probably continued after composition was scaled down from two skeletons to one.[20] Of special interest to The Roaring Girl, he also notes this procedure's frugal use of type in single-skeleton printing: seriatim setting would require three more type-pages in any given sheet than are needed in forme setting to prepare the first forme of that sheet for the press. Its partial resetting renders sheet I unusual, but additional evidence which similarly betrays early distribution and setting by formes is found in sheets D, K, and L (perhaps also F: see above, note 16). In each, speech headings incorporating broken type may be seen in the inner and outer formes. Occurrences significantly early in both formes of sheets D, K, and L, moreover, nullify the force of an argument based on a possible random reversal of the regular precedence of the inner

forme in the printing of this text. Instances of the same type in both formes of a sheet point additionally to delays in composition, which in turn appear to co-ordinate, in general at least, with a shortage rationale for setting by formes. Further, sheets which do not evince clear evidence of this kind are those in which a lengthy stretch of repeated speech headings does not run through both formes; so the absence of re-used italic type in headings or stage directions in no way demonstrates that those sheets were handled differently. Perhaps, bearing in mind D. F. McKenzie's cautions especially over the matter of setting by formes,[21] we should not leap too readily to the conclusion that this method was the norm throughout. But other evidence, admittedly of a less substantial nature, does agree with the notion: crowding in the form of doubled-up short speeches and generally less white space is almost exclusive to the first three pages of the outer forme in practically every sheet of the quarto. Other remarks by McKenzie in the same study are specially pertinent to the case of The Roaring Girl: "On the face of it, the most important reason for setting by formes in quarto is unlikely to have been urgency, nor even an unusually small fount, but a fount depleted because of concurrent printing—for if work overlapped on two or more books using the same fount of type, setting by formes would offer a method of making some progress with all." (p. 40).

Although in the absence of a companion text or texts proof is impossible, conditions of concurrent printing would very likely have given rise to some arrangement for dove-tailing composition, proofing, correction, and press-work on single-skeleton books with those operations on other productions of the same shop. Some variation of the conventional two-skeleton proofing procedure described by Bowers could perhaps apply to work on two single-skeleton books. Clearly, further work is needed to shed light on the matter of concurrent printing in shops of this period; but fuller investigation may merely reveal the kind of evidence needed to support such a procedure as I have sketched to be beyond recovery. We know, however, from studies of the Pied Bull Lear and other play texts[22] that Nicholas Okes certainly had sufficient type in his single-press establishment to sustain two-skeleton work, perhaps three or more, so it does not unduly strain plausibility to consider that the formes of The Roaring Girl shared press time and other procedures with one or more other productions. Some type shortage oddities, setting by formes (demonstrable at least in a persuasive number of cases), and the disposition of press corrections, with seven out of eight sheets invariant in one forme and an absence of gross errors in the invariant formes, at a minimum raise suspicions that a procedure and a proof-correction system may have been adopted which corresponds in some adapted form to that seen in two-skeleton

books. Further, if the formes of more than one book were being set up and wrought off in some pattern of alternation, the headlines and catch words would be the likeliest reference points when it came to perfecting for the avoidance of mismatching formes of one book with those of another and for alignment with the correct mate; and in this may lie a clue to the discovery of the sig. D4r errors.

The glimpse afforded by The Roaring Girl quarto of an apparent procedure in Okes's shop involving the preservation of certain speech headings whole from wrought-off formes for later re-use, if observed in other texts of the period may provide valuable bibliographical evidence touching on such matters as order of formes through the press, setting by formes, type shortages, and progress of composition, some of which have already received attention in relation to simple type recurrences: the frequent repetition of speech headings through neighbouring formes provides a natural occasion for the adoption of such a system in one-skeleton printing.