University of Virginia Library

The Textual Relation of Q2 to Q1 Hamlet (I)
by
Fredson Bowers

I

With the publication in 1934 of J. Dover Wilson's detailed monograph on Hamlet,[1] the general textual problem appeared to be solved in all important matters, and a scholar could well have believed that no new information was likely to be recovered which could basically affect editorial procedure. It was soon generally accepted, following Wilson, that the Folio manuscript was a copy of a copy of the manuscript which stood directly behind the Second Quarto, and that this earliest manuscript was Shakespeare's own autograph 'foul papers.' Though seriously corrupted from time to time by an inexperienced and blundering compositor, Q2 was therefore elevated as the primary substantive text, and the Folio print lowered in esteem as an actor-sophisticated and unreliable witness at two removes. Some 'bibliographical links' between the bad Q1 and the good Q2 were identified in the first act, but the number of these was minimized. Professor Wilson gave it as his impression that the relation was one only of occasional consultation when the Q2 compositor was puzzled by Shakespeare's handwriting; that "it would be going much too far to suppose that act I of Q2 was printed from a corrected copy of Q1"; and that the compositor, "knowing, as he probably did, a good deal about its origin," would no doubt have consulted Q1 cautiously.[2]

Professor Wilson's plausible arguments for the generally supreme authority of Q2 had a heady effect on editors, as may be seen in the Craig and Parrott edition of 1938 which attempted the first scientific old-spelling text of Hamlet according to Wilson's principles. Since the Folio was taken to be an independent authority, concurrence of F and of Q2 was almost automatic proof of the authenticity of the reading; otherwise, such agreement

in a wrong reading could be explained only with difficulty either as fortuitous or else as a common error in reading the original by the Q2 compositor and Wilson's Scribe P, the copyist of the promptbook. Agreements between Q1 and F against Q2 when they did not suit editorial taste could usually be brushed off as reflecting actors' changes introduced into the promptbook. On the other hand, agreement between Q1 and Q2 was the double-eagle guarantee of authenticity in all cases where contamination was not suspected, for here, it was assumed, one had the acting version supporting the foul papers at an earlier stage than the special manuscript made by Scribe C to serve as printer's copy for F.

In this reconstructed textual history the problem of corruption in Q2 deriving from Q1 ought to have been highlighted as of singular importance, since agreement between the two quartos as against the Folio was ordinarily taken as the strongest demonstration of Folio sophistication; and hence it is rather odd that in the succeeding years no further enquiry was made into the exact physical relation of the quartos. That editors seemed content to approve of the Wilson 'impression' is evident and may in part be explained, perhaps, by the new optimism induced by Wilson's rehabilitation of the Second Quarto, combined with the theory that the Folio was an independent witness which could be counted on to vary and to produce the right reading in those few and surely obvious cases when the Q2 compositor had consulted Q1. Editors, indeed, found great comfort in the Q1 contamination theory. It neatly explained without recourse to the always troublesome common-error hypothesis such a manifest double mistake as cost for cast in I.i.73 of Q1 and Q2, for example; and for those unwilling to believe with Wilson that sallied flesh in Q1 and Q2 was merely a misprint for correct sullied, fortification was given for preferring the traditional F solid.

Wilson's hypothesis that the consultation of Q1 by the Q2 compositor was cautious and occasional was buttressed by his listing (I, 159) of only twenty-five readings in which Q2 had felt Q1's influence and from these his segregation of only five in which he felt that Q2 had departed from its manuscript to follow Q1 in error.

• I.i.44 horrowes (Q1 horrors)
• I.i.73 cost
• I.ii.129 sallied (for sullied)
• I.iii.74 of a most select (ditto F1)
• I.iv.49 interr'd

Such a list could not give rise to serious doubts about substantive error, for, of the five, three were really only spelling errors. This list of five is not, however, very accurate, for the evidence either of MSH, I, 161, or of the New Cambridge text shows that Wilson also preferred the Folio reading in the following cases when F diverged from Q1-2 agreement:

• I.ii.40 or 42 [no exit]
• I.ii.205 gelly, with . . . feare ^
• I.iv.71 bettles (Q1 beckles)
• I.v.33 rootes
• I.v.64 leaprous
• I.v.113 Heauens
• I.v.132 I will goe pray (F Looke you, . . .)

In each reading the Q2 'error' ought reasonably to have arisen from Q1 contamination.

It is interesting to observe that this list, conservative for its number of five (really twelve) assumed errors, did not meet with universal approval by subsequent editors. More disagreement might, of course, have been expected from the Old Cambridge editors. However, of Wilson's assigned five errors they had anticipated him in four and differed only about I.ii.129 sallied. This general concord contrasts strongly with the opinions of six post-Wilson editors (Kittredge, Craig-Parrott, Harrison, Campbell, Alexander, and Sisson[3] ) who as a group concur that Q2 has indubitably been contaminated by Q1 in only two readings of the five. And since these two readings are horrowes and cost, we do not have much of a bag. Moreover, of the seven additional contaminations in which Wilson believed, as shown by his text, the Old Cambridge editors had anticipated him in all but one (I.v.33 rootes); on the other hand, the six later editors agree as a unit with Wilson in only two trifling readings (bettles as a misprint for beetles, and the supplying of an exit at I.ii.40 or 42).

The textual situation in the view of recent editors may be summarized thus. Of the 70 significant variations of F from Q1-2 agreement in sheets B-D (for the list, see the Appendix A), Wilson and all the listed editors after him concur that the Folio is in error and that there has been no contamination in Q2 from Q1 in respect to these readings in 48 cases. If, thereupon, we deduct from the remaining 22 readings the 4 cases in which all editors agree that Q1-Q2 are in error, we are left with 18 disputed readings in which some one of seven recent editors of Shakespeare feels (as shown tacitly by his preference for F over Q1-2) that Q1 has contaminated Q2. (It is interesting to see that of these disputed 18 the Old Cambridge edition

preferred F in 17 and in no case is unique. The New Cambridge, on the other hand, prefers F in only six of the disputes.) If we allow these editors either a moment of extraordinary insight, or else one of temporary aberration, and deduct the four cases in which only one prefers F and thereby disagrees with the others, who go along with Q1-2 (Campbell at I.i.17 and I.v.91; Kittredge at I.i.34; Craig-Parrott at I.iv.9), and the two times one editor holds out for Q1-2 as against the preference of the six others for F (Craig-Parrott at I.ii.205 and I.v.64), we find reasonably serious disagreement in 13 readings. Of these there is substantial agreement at I.iii.74 (except for Wilson) that something is wrong with Q2, but three editors are content to repeat the Folio while the other three make different attempts to emend (Kittredge, Alexander, Sisson), of which Kittredge's may be the most successful perhaps.

From the remaining dozen disputed readings, therefore, we have 5 in which two editors, or a minority, think Q1-2 corrupt and choose F:

• Q2 I.ii.177 prethee (i.e. prithee) Harrison, Campbell
• I.ii.185 Where Kittredge, Harrison
• I.v.33 rootes Wilson, Kittredge
• I.v.62 Hebona Harrison, Campbell
• I.v.126 in the [right] Harrison, Campbell

Of these, Q1-2 roots versus F rots is perhaps the most important, followed by Q1-2 Where versus F Oh where. The others represent forms of words or, in one case, a presumed metrical consideration.

Of the remaining 7 readings only a bare majority of four editors retain faith in Q1-2, whereas three indicate their belief in corruption by choosing F:

• Q2 I.i.164 that time Kittredge, Harrison, Campbell
• I.iii.65 new hatcht vnfledg'd courage Kittredge, Harrison, Campbell (who emend both Q and F)
• I.v.150 Ha, ha Kittredge, Harrison, Campbell

In these three readings, of which only Q1-2 courage versus F comrade is of especial importance, it is interesting to see that the three dissenting editors follow the Old Cambridge text.

We begin to arrive at more general agreement about Q1-2 corruption in 3 readings in which four editors, or a majority, take the F variant:

• Q2 I.ii.226,227,228 All. Kittredge, Harrison, Campbell, Sisson
• I.iv.49 interr'd Wilson, Kittredge, Harrison, Alexander
• I.v.113 Heauens Wilson, Kittredge, Harrison, Campbell

Of these interr'd versus F enurn'd is the most interesting.

In the remaining two readings there is a preponderance of opinion, five to two, in favor of F:

• Q2 I.ii.129 sallied (i.e. sullied) all but Wilson and Craig-Parrott
• I.v.132 omits Looke you, all but Craig-Parrott and Sisson, though Wilson and Alexander choose Q1-2 I will instead of F Ile

As given the acid test of editorial treatment, therefore, in the twenty years after Wilson's monograph, which publicized the 'bibliographical links' between Q1 and Q2, only some one of seven important editors has considered that in the first act there were as many as 22 possibilities for corrupt readings in Q2 derived from Q1, although the most any individual editor has been willing to assign as corrupt and to reject is Harrison's and Campbell's 20. Only two editors of the seven have agreed there were as many as 19 corruptions, and three that there were as many as 14. When we finally come to a majority opinion, 11 errors are assigned (as against the 11 or 12 in which Wilson seems actually to have believed: he is, of course, one of this majority). Five editors (a stronger majority since Wilson does not now tip the scales) can agree on only 8 contaminated readings; six editors on 6 such readings; and all seven on only 4. These figures very likely indicate the opinion of the editors concerned about the general accuracy of the 'foul-papers' Q2 text and cannot be taken to result from any searching individual enquiries into the Q1 contamination problem.[4]

Given the new view of the authoritative origin of the Q2 text which Wilson so largely introduced, the acceptance of his impression that the consultation of Q1 was cautious and occasional, and then of his belief that F provided an independent witness, it follows that the editorial treatment of the more 'scientific' of the editors is about what might have been expected. Certainly there is no indication of any fear that contamination from Q1 offered a serious problem. And in this connection it is interesting to observe that the only reading of any importance in which all three early texts agree and yet which editors have viewed with some suspicion is of the most select; but even this is not a clearcut case. No stronger evidence could obtain for the belief in F as an independent witness.

Recently, Dr. Alice Walker has attacked two of the three components of the Wilson hypothesis upon which this particular editorial reconstruction

rests.[5] Her main thesis is that F is not an independent witness but instead that the printer's copy was an example of Q2 annotated by reference to the playhouse manuscript. Although the precise details of her reconstruction have not gone unchallenged and the question has been raised either of an intermediate manuscript copying an annotated-quarto promptbook or else of some slightly indeterminate use of Q2 short of annotation,[6] that Q2 was in some manner and to some degree an ancestor of the mixed F text seems demonstrated. If this is so, not only is the total textual picture changed but there are important applications to the specific problem of the Q1-Q2 relationship: given an inadvertent failure in F to select the manuscript rather than the Q2 reading in cases when the Q2 reading has been contaminated, it is quite possible for all three texts to agree in a reading which is non-Shakespearian.[7] Operating under the Wilson hypothesis, editors could reasonably believe that, with luck, in all cases of Q1-Q2 contamination the Folio would vary and would normally provide the correct reading. The only problem, therefore, was to assess F's variants from Q1-Q2 agreement either as typical F corruptions and sophistications where Q2 was correct, or else as true readings in contrast to Q2 errors. Strongly armed with a high respect for Q2 as set directly from Shakespeare's autograph papers, the more textually 'scientific' the editor the more he favored Q2 and depreciated the F variants as unauthoritative.[8] The statistics presented above tell the story. And they would presumably justify the position taken by such a middle-ground editor as Alexander, say, if the corruption in Q2 from Q1 were no more serious than was exposed by F variation, even accepting the new status of F according to Walker (which none of the four editors had the opportunity to consider). The argument would be appealing that if no more corruption than was suspected could be found by F variation, it was unlikely that much if any by bad fortune would have escaped the annotator or comparer of the F printer's copy, since as a general rule it would seem that he had introduced more annotations than he had failed to make. Thus the soundness of the Q2 text would be reaffirmed.

However, the second pillar attacked by Walker was the hypothesis for only occasional consultation of Q1 by the Q2 compositor. In her view the evidence is convincing that Act I of Q2 Hamlet had in turn been set from an annotated copy of Q1. If this is so, the textual picture changes radically again. Under the former view, except for between 5 and 22 readings taken over from Q1 in error from a total of 25 (or up to about 40) specific consultations, Q2 was an independent witness set from manuscript. Under Dr. Walker's theory, if Q2 were set from a copy of Q1 annotated in the printing house by reference to the manuscript to serve as easier printer's copy, the incidence of error would be markedly higher. Thus even with the new theory about the formation of F's printer's copy, instead of a comparatively narrow base for Q2 corruption from Q1 to be passed on to F, there would instead be a comparatively wide base. Moreover, since neither of the two editions which had some transcriptional link with authority could in the area of Act I be counted on to have a full and uninterrupted link (for neither had been set directly from a manuscript), the number of Q1 'bad' readings which could be transmitted directly to F through Q2 must pyramid sharply, and this increase in possible corruption would be accompanied proportionally by a decrease in the evidence of textual variation as a means for detecting it.

Each part of Dr. Walker's hypothesis, therefore, has its own importance. If only one of the two texts had as a basic ancestor an earlier print, the textual case for Act I is not so serious; but if in this same area both Q2 and F cannot be called substantially independent witnesses, then the editorial problem is extremely serious and the amount of corruption in the early part of the Hamlet text might be really considerable. It is of importance, therefore, to test her hypotheses with some scrupulousness in order that we may know where we stand. In this connection I must point out that any estimate about the nature of the corruption, as well as the amount to be anticipated by an editor, is dependent on more than the simple establishment of some immediate dependence of one printed text on another, either of F on Q2, or of Q2 on Q1. That Q2 in some manner used Q1 is self-evident; and I am willing to accept Dr. Walker's basic arguments and to believe at least that in some manner Q2 is an ancestor of F, even though I am inclined to query whether an annotated copy of Q2 served directly as printer's copy. My concern here is not with the Q2-F relationship, however, but with that between Q2 and Q1. And in this matter (as with Q2-F) we cannot estimate either the kind of contamination, or the amount of it, until we can establish a working hypothesis

based on the best evidence that can be secured as to whether Act I of Q2 was set from an annotated copy of Q1, or whether it was set instead from manuscript with consultation of Q1; and if this latter is the case, whether the consultation was only occasional as has been traditionally accepted or whether it was more extensive and peculiar.

In such an investigation, the basis must be bibliographical in so far as such techniques can properly be employed. We must recognize, however, that as bibliographical investigation turns from the analytical to the textual the evidence on which it operates becomes less certain and therefore the results grow more speculative and tend more to argue for probability than to offer practical demonstration. This is especially true when we come to study compositorial characteristics, an extremely variable matter from which evidence can be adduced only with caution.

What we may determine about the relation of Q2 to Q1 starts with what is known about the printing of Q2. This subject has recently been investigated both in relation to the presswork and to the composition, with mutually supporting results. The analysis of the running-titles reveals that two presses printed Q2, as follows: Press I machined sheets B, C, D, F, I, N and O+A (in that order); Press II machined sheets E, G, H, K, L, M.[9] An analysis of the variant spellings of certain significant words indicates that two compositors, whom we may call X and Y, typeset the quarto according to the sheets assigned to the two presses, except that compositor X assisted his fellow by setting sig. L4 verso, and sig. L1 recto, in sheet L. Thus to compositor X we may assign sheets B, C, D, F, I, N, O+A, and sigs. L1,4v. To compositor Y go sheets E, G, H, K, M, and sigs. L1v-4 of sheet L. Moreover, these two workmen can be identified by their spelling habits as the compositors who set The Merchant of Venice Q1 (1600) and Titus Andronicus Q2 (1600), as well as various other books which came from James Roberts' shop.[10]

If compositor X had begun to set this book and was subsequently assisted by compositor Y and Press II after he had set three sheets B, C, and D, the above facts would not have a great deal to do with the problem of the relations between the quartos. But, on the contrary, compositor X (who may well have been a somewhat faster workman) was very likely still engaged on sheet B or at the most had no more than started sheet C when

his fellow Y joined him on the book. The running-title and spelling evidence indicate something like the following schedule for the typesetting and printing.[11]

When this schedule is surveyed, it may be seen that the irregularity of assignment of sheets in the quarto to the two compositors had no physical reason for its existence in the time relationship, and that there was no reason why Q2 Hamlet could not have been printed as was The Merchant of Venice in regularly alternating sheets:

Here, it seems clear, X started The Merchant at a point very close in time to that at which Y was to start Q2 Hamlet. But in Hamlet copy was cast off as far ahead as sig. E1 for compositor Y's beginning, although at the very least sig. C1 would have served in case X had already commenced on sheet B, or sig. D1 if X had started sheet C.

A comparison of the printing of these two quartos re-inforces normal bibliographical interpretation of the evidence from Hamlet that some special reason must have existed for starting compositor Y so deep in the book as sheet E, although he was beginning his work no later than the time that compositor X was starting sheet C. Since this reason does not inhere to any part of the normal mechanical process, we may raise the question of the printer's copy. Apparently the circumstances of printing Q2 are not unrelated to the fact that Act I, in which the major if not entire share of influence from Q1 can be detected, is contained exclusively in sheets B, C, and D, except for the last seven lines which run-over on sig. E1. And it is seemingly not fortuitous that this area of influence coincides with the bibliographical oddity that compositor X set these three sheets in sequence, though by normal printing practice compositor Y should have set one of the sheets. It is reasonable to conjecture, therefore, that the use of Q1—evidence for which can be detected in each of the three sheets—must have dictated the abnormal sequestration of these sheets for X to work on. And if this is so, it follows that there was some physical circumstance in the use of Q1 which made it impossible for both compositors simultaneously to typeset those Q2 sheets in which Q1 was being

employed. This circumstance must have been either (1) the fact that, as Dr. Walker believes, Q1 had been annotated to serve as printer's copy for Act I (sheets B-D); or (2) Q1 was not annotated but was being closely consulted by the typesetter. If (2) is to have any credibility, we must suppose (a) only one copy was available,[12] and (b) the consultation was not occasional, as has been supposed, but instead so constant, or general, as to prevent two workmen from using the single copy at the same time. Finally, if we are to believe in this modified consultation theory, we must conjecture that the usefulness of Q1 had been determined and the method by which it would be employed had been worked out before compositor Y came on the job and was forced to cast off copy to begin his stint so late as sheet E.

The bibliographical evidence for an unusual division of the compositorial responsibility joins with the textual evidence, therefore, to suggest that Q1 was utilized in a rather important manner in the typesetting of sheets B-D of Q2 and that this fact must serve as the basis for any hypothesis about the method of its employment. If this is so, the compositor could not merely have consulted it with caution on the 25 occasions posited by Dover Wilson but must instead have utilized it almost continually. The inadequacies of the occasional-consultation theory have already been pointed out on the evidence of the text itself by Dr. Walker.[13] I am in full agreement with her that the evidence of the outsetting of the Q2 speech-prefixes at the start of the first scene (of which more later) combines with the four curious similar spellings she remarks in scenes iv and v to force us to discard the consultation theory in the form in which it has previously been presented. The question, then, enforces itself: was an example of Q1 annotated in the printing house to provide the copy for Q2 sheets B-D, as Dr. Walker argues, or was Q1 employed in some other manner but one that prevented two compositors from making simultaneous use of what must have been only a single example. As remarked above, this question is more than an academic one, and on it hinges extremely important estimates as to the nature and extent of Q1's corrupting influence.

In her investigation of Hamlet, Dr. Walker's main purpose was to deal with the Q2-Folio relationship, and hence she did not offer in support of her hypothesis for an annotated Q1 the same detailed evidence as was attached to her Q2-F argument. Her brief statement is in two parts.

(1) On sig. B1 of Q1, as first noticed by Greg,[16] the compositor set his speech-prefixes sans indentation and full out to the left margin of his text. On sig. B1v, for whatever reason, he reverted to the conventional indentation of his prefixes. Correspondingly, on sig. B1 of Q2 the speech-prefixes are unconventionally treated, being outset beyond the text left margin. The text is expanded from that of Q1 so that when Q2 turns the page to sig. B1v the material is still that on B1r of Q1; and on Q2 B1v the prefixes are set flush to the left margin (as in Q1) precisely to that point where Q1 passed over to normal usage after turning the page to its B1v. Here, a third of the way down on Q2 B1v, the prefixes immediately forsake the flush setting and are indented like Q1's. The change in procedure midway on Q2's page coinciding with the change in procedure as Q1 began a new page is so manifestly the result of Q2's imitation of Q1 that Dr. Walker argues, "the change . . . can only have been due to the Q2 compositor's working from Q1 and mechanically following his 'copy'.

Corroborative evidence that the Q2 compositor was working from a copy of Q1 is found (as Greg also noted, loc. cit.) in the commas which appear at the end of speeches at I.i.6 and I.i.15."

(2) After noting that the Q2 title-page was obviously set up from that in Q1, she continues:

Dr. Walker then rests her case by pointing out that (with the possible exception of Valtemand's speech in II.ii) the agreements in anomalous spellings cease after Act I and thereafter Q2 continues by introducing anomalies of its own. No common typographical errors appear after Act I, and various of Q2's blunders could scarcely have been made from Q1 printed copy. After the end of Act I, therefore, it is concluded, the use of Q1 was abandoned, and it was not even consulted occasionally since finding the relevant matter would have wasted much time, often to no purpose. "In the absence of any reading common to Q1 and Q2 which is certainly erroneous, it would seem legitimate to suppose that Q2 is an independent witness from the end of Act I."

It is perhaps unnecessary to point out (as stated above, this was not Dr. Walker's main case) that rather far-reaching inferences are here drawn from evidence which may be susceptible of quite another interpretation, and that the evidence itself is suggestive rather than demonstrative. For example, it is assumed that the only purpose for which the Q2 compositor would consult Q1 on sig. B1 recto and verso would be to determine how to arrange his prefixes, and since that proposition is patently

absurd, the inference is drawn that the similarity in treatment of the prefixes could have resulted only from the use of Q1 as actual copy. On this assumption it is further assumed that since the far from easy annotation of the text on B1 recto and verso was therefore practicable, it would be no more difficult elsewhere in the first act, and that the anomalous spellings demonstrate it. On the other hand, if we start with a different set of inferences, and conjecture that the Q2 compositor was closely consulting Q1 in conjunction with his manuscript, it is by no means improbable that he would have his eye so fixed on Q1 that he would be led to follow its typographical treatment of the speech-prefixes. And roughly similar close consultation would thereupon explain, just as well, the further evidence for agreement between the quartos. Neither logic nor the available evidence enforces us to believe one or other theory without the presentation of more detailed and certain evidence than has hitherto been published.

Before the presentation of such evidence is attempted, we may try to clear the ground of the 'occasional consultation' theory. Dr. Walker is certainly correct in her arguments that the speech-prefix typography on B1r-v goes far beyond anything that would have resulted from occasional and cautious reference to Q1 by the Q2 compositor puzzled about a reading in his manuscript. Moreover, two of the three irregular commas on sig. B1 come direct from Q1:

Q1
• 1. O you come most carefully vpon your watch,
• Mar. And leegemen to the Dane,
• O farewell honest souldier, . . .
Q2
• Fran. You come most carefully vpon your houre, (I.i.6)[18]
• Mar. And Leedgemen to the Dane,
• Fran. Giue you good night. (I.i.15-16)

In spite of the fact that an anomalous end-of-the-speech comma (without reference to Q1) appears in Q2 after Bernardo's "Long liue the King" (I.i.3), the first comma at the end of both the above speeches is imitative; and the second is decisive, for a comma in Q1 at the end of a line that was continued has become a comma to end a speech in Q2. Occasional consultation to puzzle out a difficult word cannot include such close imitation as this. Nor is it possible to assume here, as has been done in Q2 of Romeo and Juliet,[19] that once the compositor referred to his printed bad quarto to confirm a difficult manuscript reading, he thereupon for a few lines set from the quarto rather than his manuscript. What reading would have proved difficult enough by line 6 of the play to call in use the printed Q1 is certainly questionable, but even so the second comma cannot result from continuing to set from Q1, for the two texts diverge so greatly that only the manuscript would have provided the intervening readings. One would not wish, perhaps, for another consultation to have taken place by line 14 or 15 in order to decipher a manuscript word.

However, the case cannot rest on only one page, positive as its evidence is, for one might argue, on the evidence of the speech-prefixes and these commas, that the Q2 compositor gave a special treatment, which he abandoned later, to the first page or so. For a typical passage we may turn to sig. C2:

Q1
• Hor. Two nights together had these Gentlemen,
• Marcellus and Bernardo, on their watch,
• In the dead vast and middle of the night.
• Beene thus incountered by a figure like your father,
• Armed to poynt, exactly Capapea
• Appeeres before them thrise, he walkes
• Before their weake and feare oppressed eies.
• Within his tronchions length,
• While they distilled almost to gelly.
• With the act of feare stands dumbe,
• And speake not to him. . . .
Q2
• Hora. Two nights together had these gentlemen
• Marcellus, and Barnardo, on their watch
• In the dead wast and middle of the night
• Beene thus incountred, a figure like your father
• Armed at poynt, exactly Capapea
• Appeares before them, and with solemne march,
• Goes slowe and stately by them; thrice he walkt
• By their opprest and feare surprised eyes
• Within his tronchions length, whil'st they distil'd
• Almost to gelly, with the act of feare
• Stand dumbe and speake not to him. . . . (I.ii.196-206)

Here Capapea is one of the usually noted coincidences offered as evidence of consultation for a difficult reading. Wilson also lists tronchions and gelly as other words in which the spelling has been drawn from Q1. If so, we have rapid consultation, for the text between Capapea and tronchions could not be set from Q1 alone. If, on the other hand, Q1 was again consulted, this time from tronchions, then we should be forced to believe that the compositor did in fact continue from Q1 without reference to his manuscript, not only to produce gelly but also, possibly, the stop after gelly instead of after feare, which affects the modification in a manner rejected by most editors.

In the same scene, studient is another of Wilson's concurrences.

Q1
• Ham. O I pre thee do not mocke mee fellow studient,
• I thinke it was to see my mothers wedding.
• Hor. Indeede my Lord, it followed hard vpon.
• Ham. Thrift, thrift, Horatio, the funerall bak't meates
• Did coldly furnish forth the marriage tables,
Q2
• Ham. I prethee doe not mocke me fellowe studient,
• I thinke it was to my mothers wedding.
• Hora. Indeede my Lord it followed hard vppon.
• Ham. Thrift, thrift, Horatio, the funerall bak't meates
• Did coldly furnish forth the marriage tables, (I.ii.177-181)

The anomalous apostrophe in bak't is in fact as imitative in Q2 as is the spelling studient, and demonstrates that if the Q2 compositor consulted Q1 to read studient (granted there could be the necessity), then he certainly kept on with Q1, setting Q2 from the printed text, on the evidence of the form bak't. As suggested before, this goes far beyond occasional and cautious consultation if it were true.

Many other passages of similar nature could be offered to the same result. In this connection one may compare the lines in the familiar I.i.70-79 containing the much noticed common error cost as well as strikt, forraine marte, and ship-writes listed by Wilson. But these are no stronger as evidence than the inconspicuous

Q1
• How look't he, frowningly?
• His beard was grisleld, no.
• Keepe wassel, and the swaggering vpspring reeles,
• And as he dreames, his draughts of renish downe,
• The kettle, drumme, and trumpet, thus bray out,
• That beckles ore his bace, into the sea
• I sent the mornings ayre,
• Hic & vbique
• Then are Dream't of, in your philosophie,
Q2
• What look't he frowningly? (I.ii.231)
• His beard was grissl'd, no. (I.ii.240)
• Keepes wassell and the swaggring vpspring reeles:
• And as he draines his drafts of Rennish downe,
• The kettle drumme, and trumpet, thus bray out, (I.iii.8-11)
• That bettles ore his base into the sea, (I.iv.71)
• I sent the morning ayre, (I.v.58)
• Hic, & vbique (I.v.156)
• Then are dream't of in your philosophie, (I.v.167)

A careful collation establishes, in my opinion, evidence not for sporadic consultation for the definite purpose of deciphering a single word, as has formerly been argued, but instead a relatively constant and steady influence on Q2 from Q1 in small as well as in the more obvious coincidences. Moreover, this influence is steadier than would have occurred if there had been occasional consultation followed by setting a few lines direct from Q1, as convenient, in the manner conjectured for Q2 Romeo and Juliet. Not only this, but most of the passages of any extent where such a procedure might have been practicable show that, in fact, it was not employed, for always readings intrude which must have come from manuscript. Lines I.ii.177-181 above, which exactly mirror Q1, are most unusual if not unique. Ordinarily we have mixed passages like I.ii.196-206 and I.iii.8-11, also quoted above. Occasional consultation to decipher some special word, followed by setting for a few lines from Q1, cannot be established in Hamlet.

It was undoubtedly the observance of this steady stream of influence which led Dr. Walker, quite justly, to reject the Wilson occasional-consultation theory. Whether the alternative must be the hypothesis that Q1 was annotated for Act I to serve as printer's copy may now become the point of enquiry. Only recently has the basic material been made available to conduct such an investigation. Results for which any validity might be claimed could be reached only if the text from which the evidence must be drawn could be established as uniform in its typesetting, or diverse; that is, whether one identifiable compositor set the text so that we can count on uniformity of treatment, or whether two were engaged, who would need to be distinguished and their work isolated for separate examination. Moreover, some control material set by the compositor in question would need to be studied to offer independent evidence about his characteristics when setting from manuscript and from printed text. From the researches of Dr. Walker, of Mr. J. R. Brown, and of Messrs. Cantrell and Williams, we now know that only one compositor, the man we call X, set sheets B-D containing Act I; and we also know that among other books for Roberts this workman had a hand in setting The Merchant of Venice from manuscript and the second quarto of Titus Andronicus from printed text. We thus have not an ideal but at least some amount of contemporary dramatic text of two sorts to act as a control for comparison.

The method of investigation will be to analyze the characteristics of compositor X chiefly when setting parts of Titus Andronicus Q2 from printed copy and then to compare these with his characteristics in Q2 Hamlet, the purpose being to attempt to determine whether in Hamlet he was using an annotated quarto as printer's copy. Such an investigation could be readily managed if the purpose were merely to demonstrate that Q1 had exerted an influence on compositor X. For example, one could point to his usual practice in Hamlet of spelling gracious and malicious either when Q1 is not present or when Q1, as usual, spells gratious and malitious. One could also point to these characteristic spellings in the Merchant and show in Titus that t spellings in Q1 were similarly altered

to c. Thus when in Hamlet on B3v, I.i.164, we find—for the only time—compositor X setting gratious, we are justified no doubt in pointing to the Q1 t spelling at that place as the reason. Similarly, the concurrence of the two quartos at I.v.17 (D2v) in stars, whereas elsewhere compositor X sets starres, is also evidence for influence, and there are other cases of the same sort of thing. For instance, though X has some predilection for odd apostrophes, as in tain't (I.v.85), low'd (III.iv.52), temp't (III.iv.182), yet even the similar op't at I.iv.50 and pop't at V.ii.65 cannot destroy the significance of Q2 agreeing with Q1 in bak't and look't (I.ii.180, 231).

On the other hand, it is freely admitted that the Q2 compositor of sheets B-D must have consulted Q1 frequently and closely, either in annotated or in unannotated form. Hence there is no reason to suppose that concurrence of the two quartos in spellings uncharacteristic of compositor X can be used in any profitable way as evidence for the precise method by which he employed Q1. His own spelling characteristics were so strong for some words and forms (as in heere) that both in Titus and in Hamlet he habitually altered his copy for these words. Even though for other words his habits were less firmly fixed, it appears to me to be impossible to show that in line after line his concurrence with Q1 was so extreme that no form of consultation could explain the agreements satisfactorily, and hence one would be forced into the assumption that annotated Q1 must have been the printer's copy. The extreme concurrence in all details necessary to convince us of an annotated quarto is not, in my opinion, found in Q2 Hamlet. Moreover, as will be illustrated below, there is such a comparative lack of influence from Q1 in some details that should have affected Q2 if Q1 had actually been the printer's copy that the case against annotated Q1 copy can, in effect, be demonstrated by negative evidence.

Other evidence must first be surveyed, however. Despite the undoubted indications that Q1 influenced Q2 spelling with some frequency, it is still possible to argue that when Q1 contains a characteristic X spelling, and when in Q2 compositor X diverges into an uncharacteristic form, the uncharacteristic form came from manuscript copy and hence we cannot have an annotated quarto. An example might be the Q2 spelling of eies at I.i.58 where Q1 reads eyes. From Titus Q2 we know that compositor X preferred the y spelling in such forms: in the Titus pages assigned by Cantrell and Williams[20] to X, eyes appears thrice (E2, E3, I3), and each

time with y despite the fact that the Q1 copy was spelled eies. Better evidence for X's preference is found in The Merchant of Venice: here in the seven times that the word appears in X's share the spelling is invariably eyes. Of the total of fifteen appearances in X's pages in Hamlet, fourteen times the spelling is eyes despite the occasional Q1 eies. Thus when in I.i.58 for the only time in three plays we meet the anomalous spelling eies in X's typesetting, given the Q1 spelling eyes at that place it is possible to argue that eies was produced by manuscript influence and that an annotated Q1 was therefore not the printer's copy. To take another example, we find at I.i.23 that Q2 has saies though Q1 reads sayes. In the Merchant X's spelling is invariably sayes, which is the form for the only three occurrences in Titus (all on H1) although Q1 read saies. Elsewhere in Hamlet X sets the word only three times and in each place in text that must have had manuscript copy. Twice it appeares as saies (I.ii.57, I.iii.24) and once as sayes (II.ii.198). By analogy with y forms that X prefers in words like dayes and wayes, as well as from the evidence for y in the Merchant and Titus Q2, it would appear that ordinarily X preferred sayes but could be influenced by his copy to set the form saies.

The difficulty of this method is to secure a sufficient number of reasonably invariant or strongly preferential spellings from which to argue for the influence of manuscript copy when Q2 veers to an uncharacteristic spelling in places where the Q1 form represents his normal usage. The preferential spellings listed by Mr. Brown to distinguish X from Y provide insufficient data since X is so very conservative in departing from them. For example, of the eight times in sheets B-D of Hamlet that X uses the uncharacteristic ending -ow or -ew instead of his normal -owe and -ewe, three must have been set from manuscript even though the remaining five agree with the very characteristic -ow or -ew endings in Q1 for the same words. The evidence is a stand-off, it would seem, for the appearance of the short form in parts set from manuscript suggests that the concurrences with Q1 may be fortuitous and X merely indulging in a minority spelling. Similarly, in the one time that X sets uncharacteristic said, and in the two occurrences of maddam, the manuscript must have been his copy. Hence the single occurrence of uncharacteristic honourd following Q1 may be fortuitous. Of all these distinguishing words, perhaps the strongest case is reuenge. Elsewhere in Hamlet X sets reuendge invariably, but the first three times the word appears (all on one page, D2v) it takes the conventional form of reuenge found in Q1. Some argument might be made for Q1 influence here; but even so there are difficulties, for on this Q2 page the first occurrence of reuenge (18 lines above the second) must have been set from manuscript. Moreover, although in Titus and the Merchant X exhibits a general liking for dg spellings, he

never sets reuendge in either play. It is unfortunate that in the B-D pages the word does not appear again, for the evidence of sig. D2v is far from conclusive. Clearly, such evidence as the above that rests in this manner on the distinguishing spellings is so mixed as to be of dubious value either as showing the influence of Q1 or of manuscript.[21]

On the other hand, Mr. Brown points out that various of the significant spellings which he chooses as characteristic of each compositor in Hamlet do not always hold from book to book set by the same two men, and therefore some of the seemingly significant spellings may do no more than reflect the compositors' general faithfulness to the characteristics of their copy. In his opinion, this appears to be especially true of the -owe and -ewe endings which are one of the best means of indicating the two compositors in Hamlet. If this is so, the evidence that in sheets B-D compositor X spelled such words just as uniformly as he did in the other sheets he set, which must have been at least predominantly uninfluenced by Q1 and utilizing manuscript as printer's copy, might well have a significance. For example, in Titus Q2 in the pages assigned to X of at least 23 occurrences X follows the Q1 -ow and -ew forms 21 times and only twice alters -ow to -owe (F2v and I2v, the latter time possibly for justification). This is so slavish a following of copy as quite to reverse the characteristics X exhibits in Hamlet. That these heavily majority -owe and -ewe endings in Hamlet represent the characteristics of the copy behind Hamlet and not a hardening of X's habits in the interval between 1600 and 1604 is an opinion rather firmly held by Mr. Brown on the evidence of the mixed characteristics of other books he has examined between these years, and after Hamlet, in which X's typesetting appears. In Hamlet Act I, of the 22 times that Q2 has a corresponding word in Q1, compositor X concurs in the -ow and -ew form with Q1 only five times, and would have changed Q1's -ow or -ew to -owe and -ewe seventeen times. This contrasts so markedly with Titus and yet is so consistent with the remaining Hamlet sheets as to lead to a presumption that Q1 was not the copy from which X was basically setting his type.

More marked spelling differences in which Hamlet departs from the Q1 form like twelfe for twelue (three times), hundreth for hundred (hundred on sig. F2v) and perticuler or particuler for particular fail for lack of evidence in the Merchant or Titus. The form a leauen for eleuen is also found as a leuen in X's part of the Merchant but whether this is

compositorial spelling or the influence of copy is not clear. However, in these spellings, as also in staukes for Q1 stalkes, brazon for brazen, forfait for forfeit, a maz'd for amazed, a bord for aboord, familier for familiar, a doption for adoption, springs for springes, beckins for beckons, horrable for horrible, pray for prey, barckt for barked, oremastret for or'emaister it, it would appear that we have compositor X departing more widely from his copy spelling than he did in Titus, provided Q1 of Hamlet had been his copy.

There are some readings which it is difficult to derive from Q1. Perhaps the most prominent of these comes in I.v.55-57. Q2 reads:

The Angle linckt here has placed this passage on all lists as influenced by Q1, which has a version supported by Folio Lust and sate, though F corrects the misprint fate (f for ſ) and supports Q2 in reading in for from a celestial bed. Although an argument might be raised in favor of reading sort it selfe from, I have no doubt that the editorial choice of the Folio version is correct, since Q2 in shows that sate must be the reading; and otherwise we should need to believe that Q1-F sate was a prompt-copy corruption. If sate is correct, then sort is scarcely a memorial error for Q1 fate, but instead—as has been remarked—a handwriting misreading as or of an a open at the top of a kind found elsewhere in Shakespeare. I have no opinion about Q2 but for lust except that it is more probably a simple handwriting misreading than the product of Wilson's ubiquitous miscorrector. The nature of these two errors does not suggest by any means (despite the Angle linckt) that when he was setting these specific words the compositor had his eye on Q1. An annotated quarto corrector would, one would suppose, merely substitute an s for the Q1 misprint f in fate, and not write out the whole word. If, on the other hand, we exonerate the compositor and blame the annotator for the misreading, thus still arguing for an annotated quarto, we must presumably blame him also for substituting nonsense not only here but also in but for Lust. On the evidence of Titus, compositor X was quite reasonably careful when setting from printed copy and his mistakes are few, at least of such a serious nature. On the other hand, how he could get mixed up with manuscript copy may be shown by such lines as Hamlet I.iii.129-131: Here a misprint appears in three successive lines (there is not much doubt, I think, that bonds should be bauds). Such typesetting is unknown in Titus (or in the Merchant where copy seems to have been clean), but it is found in Hamlet Act I, where it provides evidence, or strong presumption, for composition from manuscript. It is perhaps fortunate that the two misreadings but and sort occur in the same and the next line to the Angle linckt phrase, commonly taken as one of the strongest proofs of Q1 influence. On the evidence of but and sort, if Angle linckt came from Q1, as I agree it must, it was not set from an annotated quarto.

Memorial error in a compositor is difficult to analyse sometimes, but I should believe that the following errors or oddities in Q2 were less likely to have come from Q1 copy than from setting from a manuscript: brazon for Q1 brazen (I.i.73), your spirits for Q1 you Spirites (I.i.38), Or euer for Q1 Ere euer (I.ii.183), what someuer for Q1 whatsoeuer (I.ii.249), fonde deedes for Q1 foule deeds (I.ii.257), Or of a most select for Q1 Are of a most select (I.iii.74), Withall for Q1 With all (I.v.79), Oremastret [22] for Q1 Or'emaistret (I.v.140), and perhaps even horrowes for Q1 horrors.[23]

Except for the Lust-but and fate-sort readings, which I am inclined to take seriously, the import of the above may perhaps be a matter of opinion. However, there are two other orders of evidence which must be considered. On the evidence of X's pages in Titus, he followed his printed copy in the matter of elision and its forms with some scrupulousness. It is true, that he showed some tendency to introduce an apostrophe: in these Titus pages he reproduced 4 Q1 apostrophes, ignored none, and added 7 to Q1 elisions without apostrophe. On the other hand, at least 41 times he followed the exact form from Q1, with especial reference to final d or t in preterite or participle, as in skipt, possest, and distild among a number of others. He did not elide any word for metrical reasons, although various might have been so treated. The treatment in Hamlet differs. In Act I compositor X could have followed 12 Q1 forms with apostrophe, but 11 times he omitted

an apostrophe (though maintaining the elision) when one was present in Q1. Only 3 apostrophes were added to Q1 forms elided without apostrophe; he followed elided forms without apostrophe 10 times with the same -d or -t ending; he elided 6 Q1 full forms (of a sort less likely to be an annotator's) without apostrophe, and 11 with apostrophe. The additional freedom with which the Q1 text would have been treated, if it had been copy, is marked compared to Titus, especially for the freedom with which in Hamlet X dropped the apostrophe, a habit not observed in Titus. All in all, such variation would seem to point to manuscript rather than to printed copy, for we cannot suppose an annotator would busy himself about such minutiae, and X (on the evidence of Titus) was a compositor much influenced in such matters by printed copy.[24]

It is interesting, moreover, to see that in Act I of Hamlet whenever he is setting from manuscript he uses the spelling then for than invariably, and that his practice does not change in lines which could have been set from Q1, although Q1 normally has the than form. On the other hand, setting from printed copy in Titus he follows Q1 than four times, and changes than to then only once. Since in Titus he shows only a slight tendency towards then, it might be argued that in the intervening years his habitual spelling hardened and hence the difference developed in Hamlet; but this will not hold water, for in the Merchant, set the same year as Titus Q2, then for than is as invariable as it is in Hamlet.[25] General experience with compositorial characteristics shows a tendency for the workman to be more influenced by copy-spellings when setting from printed copy than from manuscript. It would seem, then, either that in Hamlet and The Merchant of Venice compositor X's usual habits showed much more strongly than in Titus because in both he was working from manuscript, or else, as Mr. Brown has argued from other evidence, the copy for Hamlet and the Merchant was basically similar in its spelling characteristics and that compositor X followed these with some faithfulness in his setting of the two plays. In either case, the evidence is against the use of an annotated quarto for Hamlet Act I. Despite the fact that various Q1 spelling and other forms are repeated in Q2, there is more essential divergence than similarity. The comparison with Titus, supported for some words by the Merchant, does not encourage the hypothesis that annotated Q1 was X's immediate printer's copy: his divergences from Q1 characteristics are more likely to agree with the Merchant set from manuscript than they are with Titus set from printed copy.

The treatment given the punctuation by compositor X in Act I of Hamlet constitutes the next range of evidence. Mr. Brown, who went through various Roberts books for his investigation, believes that the light punctuation found in the Merchant and in Hamlet is not especially characteristic of the two compositors. Lumping their work together, he points out that in the Titus reprint they made some 240 changes towards heavier punctuation in comparison with some 120 in which the Q1 punctuation was made lighter. He concludes, "Titus Andronicus therefore suggests that the compositors of Hamlet and The Merchant tended to add to the punctuation, not reduce it."[26]

A study of X's specific treatment of his copy punctuation in Titus has various points of interest for the Hamlet investigation. In the Titus pages assigned to X by Messrs. Cantrell and Williams, I count 18 commas added where Q1 had no punctuation, but only 8 of Q1's commas omitted. One Q1 comma is wrongly made into a period, very likely a misprint; a wrong Q1 comma is correctly altered to a full stop; five Q1 clause-ending commas are changed to periods; and a period is correctly added at the end of a sentence. In an obvious query a question mark is substituted for a Q1 period. Two Q1 colons are reduced to commas as slight improvements, and one colon is correctly removed. Only one semi-colon appears, where it is copied from Q1. X thrice strengthens a Q1 comma to a colon, once a comma to a query, and once a colon to a period. The evidence shows that he was usually content to follow printed copy with remarkable fidelity, and that his alterations were only sporadic and far from consistently introduced. The increased heaviness of the Titus Q2 punctuation over that for Q1, therefore, seldom lies in the substitution of heavier for lighter stops but instead in the addition of commas. Ten of these added commas are inserted between independent clauses, ordinarily when no conjunction is present; two separate parts of a complex sentence; and one each marks apposition, direct address, a vocative, an ejaculation, and a phrase. Once an unnecessary line-ending comma is added. Of the eight Q1 commas omitted, five delete not wholly necessary line-ending pauses, one a caesural pause; and two omissions are in error.

The single copying of a semi-colon in Titus compared to the fairly frequent use of this stop in the Merchant and Act I of Hamlet could indicate similarity of copy in the latter but might also indicate that when setting from manuscript compositor X felt free to follow his own proclivities.

In sheets B-D of Hamlet in all places when Q2 and Q1 verbal readings coincide, there are 44 changes strengthening the punctuation as against 114 making it lighter. The largest number of changes comes (exactly contrary to Titus) when 75 Q1 commas are removed and nothing

substituted. In this lightening of punctuation the next in number are 10 semi-colons reduced to commas, 8 colons reduced to commas, 5 question marks reduced to commas, 4 periods altered to commas, 3 sets of parentheses removed, 2 each of queries reduced to periods, colons to semicolons, periods to nothing, semi-colons to nothing, and one exclamation reduced to nothing. On the heavier side, we find 15 additions of commas where no punctuation had appeared, followed by 8 commas made into semi-colons, 7 commas made into periods, 3 sets of parentheses added, 2 each of added periods from nothing, colons from nothing, colons changed to queries, colons changed to periods, and 1 each of a semicolon added, a query added, an exclamation substituted for a comma, and a query for an exclamation.

Obviously, the treatment in Titus Q2 of the punctuation in the Q1 printed text differs from that found in sheets B-D of Q2 Hamlet if annotated Q1 had been the printer's copy. We cannot say that Q1 Hamlet is an overpunctuated text; but in contrast to the 18 commas added in Titus and 8 reduced, we have 14 added as against 75 reduced. These statistics, moreover, must be viewed against the fact that as a general proposition the punctuation of Q2 Hamlet (with special reference to sheets B-D) approximates that of the Merchant. Finally, although such a judgment is bound to be subjective, perhaps, I am not impressed by any differences in the punctuation method between the parts of Hamlet after sheet D which compositor X set from manuscript and the system he employed in the area of contamination. Since in Titus he had shown himself to be a close follower of his printed copy's punctuation, the natural inference in Hamlet is that—whether he was following his manuscript copy fairly closely or else largely punctuating for himself—compositor X was not setting type directly from printed copy. There are agreements in oddities of punctuation, just as there are agreements in spelling, but the results will not permit us to conjecture that conditions were the same (plus annotation) in Hamlet as in Titus.

One final variety of evidence may be mentioned. It is perhaps speculative to assert that an annotator of Q1 would not be likely to prove so scrupulous to make such minor changes as he to colloquial a and similar small matters. But speech-prefixes and their forms are something else. In reprints, in so far as the compositor can—consistent with any problems of justification—there is a general tendency to follow the forms of speech-prefixes in printed copy, and indeed this fact has been used as bibliographical evidence for the reliance of one edition on another.[27] One should notice

that in Titus Q2 compositor X alters the forms of Q1 prefixes only seven times, and thus shows that he was a conservative follower of printed copy. Hence it is more than a trifle odd in Hamlet sheets B-D that though the Q1 prefix for Horatio is invariably Hor., we find with but three exceptions on one page, sig. B3,[28] that Q2 invariably prefixes the form Hora., and this even on the first page, B1, when certain Q1 characteristics are being followed so closely that the speech-prefixes are outset and two faulty speech-ending commas are faithfully reproduced. This alone is suspicious if an annotated quarto is in question. But there are also other grounds for suspicion. In some respects the prefixes of the two compositors differ sufficiently to show that they did not (as we should scarcely suspect) make up an agreed-upon list. For example, in sheets C and D compositor X sets Ophe. (Q1 Ofel.) but in sheet G and H compositor Y almost invariably sets Oph. Compositor Y's treatment of Horatio is interesting. On G4, the first time the character appears in Y's stint, he is initially Hora. but then Hor. once on the same page and once on the verso. On H3, next, he is Hora. three times and Hor. once. On L2v he is Hora. twice and Hor. twice; on L3 Hor. once. In sigs. M2v-3v he is Hora. seven times and Hor. twice. Yet when X takes over with sheet N he is Hora. without exception to the end of the play. The fixity of X's use, though it runs counter to the Q1 form which in an annotated quarto he would constantly have been looking at in sheets B-D, may well have been taken (at least originally) from manuscript; and the manuscript may well be reflected in Y's variable but majority use of the longer form, also, although elsewhere Y almost invariably sets prefixes of only one syllable, as Oph. for X's Ophe. In X's persistence against Q1 in the long form Hora., I believe we have additional evidence of some especial weight to oppose the theory that an annotated quarto was used as printer's copy.

As a postscript to this presentation of the evidence against an annotated quarto as printer's copy one matter of opinion may be mentioned. If a copy of Act I of Q1 had been annotated in the printer's shop, we should presume this had been done before copy was cast off for Y's entrance. If so, it is almost inexplicable that copy was cast off with such exactness as to bring the last seven lines, only, of this annotated material onto the first page of Y's stint rather than assigning it to X and starting Y off with the first scene of Act II. On the other hand, if the example of Q1 had not been annotated, the division which runs over the last seven lines of Act I onto sig. E1 set by compositor Y is the more understandable. Incidentally, it is clear that Y did not have access to a copy of Q1 when he was setting sheet E.

[to be concluded]

Notes