University of Virginia Library

Two Shakespeare Quartos: Richard III (1597) AND 1 Henry IV (1598)
by
MacD. P. Jackson

The two compositors originally identified as responsible for setting the Shakespeare First Folio have, with more intensive scrutiny of the volume, split and multiplied into at least nine men, whose stints have proved difficult to determine.[1] It seems likely that earlier bibliographical findings about some of the Shakespearian Quartos may also need revision in the light of evidence derived from new techniques of analysis. In this article I report on an examination of two Quartos—the "doubtful" Richard III (1597) and the "good" 1 Henry IV (1598)—which originated, in part, from the same printing house: 1 Henry IV was printed by Peter Short, as were sheets H-M of Richard III. No full investigation of the composition and printing of Richard III has hitherto been recorded, though Alan E. Craven has dealt in separate articles with aspects of the two sections into which the Quarto falls.[2] Craven can be proved wrong in claiming that Short's portion of Q1 Richard III was set by one man. And my work on 1 Henry IV corrects Charlton Hinman's conclusion, endorsed by Craven, that a single compositor set that particular Quarto.[3] In each case there are modest implications for editorial handling of the text.

The 1597 Quarto of Richard III was the first edition of the play. Five more Quartos followed, each printed from its immediate predecessor (except for Q5, which was printed from a combination of Q3 and Q4 copy), before the play's inclusion in the First Folio of 1623. The relationship between the Q and F texts, which differ in numerous substantive readings, and the relationship of each to Shakespeare's autograph, are still in dispute. There is general agreement that the manuscript behind the F text is one of considerable authority, possibly autograph, and since the publication of D. L.

Patrick's The Textual History of "Richard III" in 1936 the orthodox view has been that Q gives an inferior text, being based on a report of an abridged acting version of the play, that the manuscript which served as copy for Q was, in other words, reconstructed from memory by a common effort of the acting company, perhaps in order to replace a missing prompt-book while the actors were on tour in the provinces. This theory has been challenged by Kristian Smidt, who asserts that Q is largely a good text which, though showing some signs of memorial contamination, also derives in part from an authoritative manuscript closely related to the foul papers.[4] E. A. J. Honigmann has also offered evidence that Q is in some respects closer to the foul papers than F.[5]

The exact nature of the copy for F is also in doubt. It seems certain that F is in some way dependent upon Q3, at least for two sizeable passages, but dispute persists over whether it was printed from annotated Q3 copy throughout, from an interleaved mixture (corrected with reference to a manuscript) of Q3, Q6, and perhaps other Quartos, or directly from a manuscript which was itself a transcript conflating an authoritative source with one or more Quarto editions.[6] Bibliographical analysis of the First Quarto can scarcely hope to solve all these problems, but it does contribute a modicum of relevant information.

As Greg pointed out in his introduction to the Clarendon Press facsimile (1959), Richard III (1597) was obviously divided between two printing houses, sheets A-G being printed by Valentine Simmes, named as printer on the title page, sheets H-M, "in a slightly different type," having come from the press of Peter Short (p. v). The difference in type will be apparent to anyone who

studies the capitals in speech prefixes and stage directions. For example, a small roman 'G', instead of italic, is frequent in the first part but does not appear in the second, whereas a tailed italic capital 'G' is used sporadically in the second part but not in the first. Italic capital 'Q' in the first part is noticeably different from italic capital 'Q' in the second part (compare, for instance, B3v, which has two forms, with H3v). Then too, an italic capital 'T' first appears mixed in with the roman on H1v (ten lines from the bottom) and occurs occasionally thenceforth. Italic capital 'I' is also mixed in with the roman in sheets H-M (appearing at least once on two out of every three pages), but not in sheets A-G. The roman capitals generally (take 'T' and 'Y', for example) are larger in A-G than in H-M. Among lower-case letters 'g' is always useful for type identification: in the earlier sheets the lower counter is open; in the later sheets it is closed. The bowl of lower case 'e' is usually inked in within sheets A-G, seldom inked in within sheets H-M. The dot of lower case 'i' falls slightly to the left of the stem in A-G, but directly above it in H-M. Lower case 'k' is narrower in the later sheets.

The signatures differ in the two sections: nine stops are used in the signatures of H-M, none in those of A-G, where the numerals are larger and the spacing between letter and numeral is more variable. In A-G nineteen lines are turned up or down; in H-M there are only two such lines. Italic colons are quite common in sheets H-M but not in A-G.

More interesting are the several shifts in spelling and other accidentals from sheet G to sheet H. Detailing a few of the more conspicuous will give some idea of the effect of compositorial preference within the printing house upon presentation of the text. Apart from two instances of do on A3, doe is invariable (60 times) in sheets A-G; in sheets H-M do is preferred (20 times) to doe (15). Within A-G O occurs 26 times, Oh 29 times; within H-M O is almost invariable (27 times; Oh once). Been(e) is used in A-G (10 times), bene in H-M (6 times). A preference in A-G for heart (28; hart 3) gives place in H-M to a preference for hart (12; heart 4). Within Simmes's section of the play -y endings greatly predominate, whereas -ie endings are far more common thereafter. The abbreviation Lo: for Lord, very common in the early sheets (76 examples) is almost completely absent from the later sheets.[7]

Of more significance to an editor is the punctuation. Commas are used appreciably less liberally, colons far more liberally within sheets A-G, so that while there are 4-5 commas to every colon in the early sheets, the ratio in the later sheets is close to 20 to one. Usage in the earlier sheets is much more consistent than later on—and this brings us to the next point.

Valentine Simmes's compositors have been studied by W. Craig Ferguson and, more recently, by Alan E. Craven, who asserts that sheets A-G of Richard III Q1 "are clearly the work of a single compositor. Not only were they set from a single type-case, as demonstrated by the evidence of distinctively damaged types, but they also exhibit a uniformity of compositorial practices,

especially heavy spelling preferences."[8] He cites, among preferred spellings in A-G, heart, young, tongue, eie, -nes, here, go, and bloud, and attributes these sheets of Richard III to "Compositor A", who, in his judgement, set all of Much Ado (1600), 2 Henry IV (1600), Hamlet (1603)—the Bad Quarto— and most of Richard II (1597), to list only the Shakespearian play Quartos. I can find no good evidence to contradict Craven's claim that one compositor set sheets A-G of Richard III, and the use of a single skeleton throughout Simmes's section of the Quarto is consistent with it.

In Peter Short's section, on the other hand, two skeletons were used, one imposing H(i), H(o), K(i), K(o), and M(o), the other imposing I(i), I(o), L(i), L(o), and M(i). This pattern is suggestive of two presses and two compositors, and the text does in fact display conclusive evidence that typesetting was neatly divided between two men. Compositor N (as we shall call him) hardly ever (9 times out of 173) set a space after commas in short lines, whereas Compositor O, though erratic, set a space more often than not (241 times out of 464).[9] The stints are readily separable on this basis alone, as will be evident from Table 1.

Table 1: Spaced/unspaced commas in short lines within sheets H-M of Richard III (1597)

                 

	H	I	K	L	M
1	1/5	1/7	1/14	11/17	9/17
1v	0/6	0/13	0/11	11/15	11/12
2	0/5	0/7	0/12	9/20	13/14
2v	0/3	¼	0/4	12/9	1/9
3	19/4	9/8	17/9	0/9	1/19
3v	14/11	11/17	17/8	0/4	0/4
4	14/8	17/13	9/17	2/14
4v	14/5	15/6	9/13	1/14

Within each sheet the work was halved, Compositor N setting H1-2v, I1-2v, K1-2v, L3-4v, and M3-3v, Compositor O the other nineteen pages. This division is confirmed by an examination of the spacing of question marks. Compositor N never set a space before the 43 question marks in short lines on his pages; in 30 such instances Compositor O did not set a space, but in 21 instances he did. The figures for colons are similar. Compositor N set a space before only 2 of his 33 colons in short lines; O preceded 22 by a space and omitted the space from 24.

It will be noted that Compositor O set commas, regardless of spacing, far more readily than did Compositor N; O is in fact responsible for the increase

within Short's section of the play, N's rate remaining close to that found in Simmes's part. There are some differences in spelling between the two men: for example, Compositor N preferred do (16, doe 1) and Catesby (10), Compositor O doe (14, do 4) and Catesbie (6, Catesby 1); O was more careful to capitalize Lord in the appellation my Lord (N: my Lord 19, my lord 17; O: my Lord 18, my lord 2).[10]

The treatment of King speech prefixes, which are often unstopped, is variable within the stints of each compositor.[11] We must assume that both compositors were erratic in their treatment of copy which may well have been inconsistent; or perhaps copy had no stops after King prefixes, and the compositors sometimes remembered to add them and sometimes did not, making some slight effort to be consistent over a single page or stretch of text. Compositor O did not stop King in H and I, but thereafter he omitted the stop only 4 times, so that his overall figures are 12 King prefixes unstopped, 41 stopped (I ignore catchwords, and count only plain King, not King Ri). Compositor N's overall figures are 44 unstopped, 32 stopped. Despite the variability, the difference between N's and O's overall figures is highly significant statistically, with a less than one in 5000 probability of its occurring by chance (Yates's chi-square = 14.37, 1 df, p<0.0002).[12] If we were to decide that I1v, with its 11 unstopped King prefixes, must have been set by a different man from the one who set I1, I2, and I2v, with their 15 stopped King prefixes, then we would logically be driven to postulate (a) a non-spacing compositor who stopped King, (b) a non-spacing compositor who did not stop King, (c) a spacing compositor who stopped King, and (d) a spacing compositor who did not stop King—these four men working on stints that do not even correspond to pages, let alone feasible sequences of pages. I think that we may confidently accept the neat and rational division of labour implied by Table 1.

The implications for an editor are not perhaps of very great consequence. Of possible interest is a change in the positioning of entry directions which seems to occur at about the middle of sheet I.[13] All the entry directions which begin new (unnumbered) scenes in sheets H-M (10 in all) are centered. Otherwise from H1 through I2v 9 entries are set to the right and only one (the longish direction for Richard's entry aloft in the company of two bishops on H1v) is centered; yet from I3 through M3 only one entry is set to the right (on K3v) and 24, in addition to those which begin scenes, are centered. Entries

are sparse within I3-K4, and the only two centered entries are quite long, so it is not possible to determine the exact point of change, but it appears to be independent of the compositorial stints, since the figures I have cited in evidence of a shift divide up fairly evenly between the two men. The probable inference is that the change in the positioning of entry directions occurred within the manuscript copy for Q.

In this matter sheets A-G resemble sheet H, rather than sheets L-M. Within A-G the 15 entries which begin scenes are all centered; otherwise 21 entries are set to the right and only 5 are centered. So it is clear that whoever prepared the manuscript behind Q was fully conscious of scene divisions, and distinguished between entries which head scenes and entries within them. The distinction is preserved with some care until towards the end of the play. Of course most entries which open scenes are long and could not easily be set to the right, even with the abbreviations often resorted to for the other entries, but some entry directions which head scenes are short, and the practice with regard to centering is obviously not governed solely by length. Of the 5 anomalously centered entries in A-G, 4 appear in sheet F (F1, F2, F4, F4v), although 4 other entries in F are set to the right (F3, F3v, F4v twice).

Another noteworthy bibliographical feature of the Quarto is the tendency of the variants O or Oh within sheets A-G, where spelling of the exclamation is variable, to cluster beyond chance expectation. This will be clear from Table 2.

Table 2: Oh/O spellings in sheets A-G of Richard III (1597)

                     

O A2v	O C2v	O C4v	Oh E1v	O F2
Oh A3v	O C2v	O D1	Oh E1v	Oh F2v
Oh A4v cw	O C2v cw	O D1	Oh E1v	O F4
Oh B1	O C3	Oh D2	Oh E2	O G1
Oh B1	O C3	Oh D2v	Oh E2	Oh G1
Oh B1	O C3	Oh D3	Oh E2	Oh G1
Oh B1	O C3	O D3	Oh E3	Oh G2v
Oh B1v	O C3	Oh D3	O E4v	Oh G2v
Oh B3v	O C3v	Oh E1	O F2	O G2v
O C1v	Oh C4	Oh E1	Oh F2	O G3
O C2	O C4v	Oh E1	O F2	O G3

A simple statistical test of significance—a Runs Test for Randomness—reveals that the probability is less than one in a thousand of such clustering occurring purely by chance (z = 3.3, p<0.001). Now a single compositor faced with random O/Oh variation in his copy, or with one of the two spellings throughout, might conceivably tend to set a series of O spellings, then a series of Oh spellings, and so on, the clustering being due to the influence of each choice upon the succeeding one. But more natural explanations are that either two compositors, with contrasting preferences, were involved, or that Q's variations reflect copy. The contrast between C1v-D1 (O 14, Oh 1) and D2-E3 (O 1, Oh 14), for example, is, after all, striking. I have not been able to detect any clear compositorial differences between O and Oh sections of

Simmes's part of Q, and the spacing of punctuation marks is consistent throughout, so I would tentatively accept Craven's arguments for a single compositor in Simmes's sheets A-G, and suggest that the O/Oh variation points to possible heterogeneity in the manuscript copy. In the quality of the text they print, however, O and Oh sections of A-G seem much alike.

A point of interest is that while the O/Oh variation in Q1, sheets A-G, is preserved almost perfectly within successive Quarto reprints, it does not survive into F. Q3 misprinted Oh who at II.ii.34 as Wh who, and Q5-8 omitted the exclamation. O, O at III.v.21 was misprinted as G, O in Q3, but this obvious error was corrected to O in subsequent Quartos. Otherwise there is complete agreement between Q3 and Q1. Q6 changed Oh to O at II.ii.59, III.iii.9, and III.iii.18, but all its other spellings of the exclamation agree with those of Q1. (All the Quartos are in perfect agreement over the O spellings and single instance of Oh within sheets H-M, as they are in Q1.)

Now 12 times F has Ah where the Quartos have O/Oh—8 of the instances are within the section of F corresponding to Q1 sheets A-G; and larger differences in wording between the Quartos and F mean that several times F has O/Oh where the Quartos do not, and vice versa; but altogether F strongly favours O (66 times) in preference to Oh (10 times), and on all 12 occasions upon which the exclamation O/Oh in F does not occur at all at the corresponding point in Q, it is spelt O. To focus on the portion of the play covered by sheets A-G in Q1, only 5 of the 21 Oh spellings in Q1 and Q3, or 4 of the 18 Oh spellings in Q6, where F actually contains the exclamation, also appear in F; 2 of these 5 agreements occur within the first 164 lines of III.i, where F is generally acknowledged to have been printed from Q3. So outside this passage only 3 of Q1's and Q3's 19 Oh spellings (or 2 of Q6's 16), where F contains the exclamation, within the first half of the play, recur in F. The two Oh spellings within III.i.1-164 were set by Folio Compositor A. To break down the other figures by Folio compositors, in B's stints only one of 15 Quarto Oh spellings recurs, and in A's stints, ignoring III.i.1-164, 2 out of 4 Q1 and Q3 Oh spellings (or one out of 2 Q6 Oh spellings) recur. So if B was setting directly from Q copy, he had, by the time he worked on Richard III, developed a strong preference for the O spelling, not evident in his handling of straight reprints of Ado, LLL, MND, MV, 1H4, Tit, and Rom.[14] (Nowhere in the first part of Folio Richard III does B set Oh when the Quartos have O.)

But it is time to return to Peter Short's compositors, and to examine 1 Henry IV. The Quarto (Q1) was printed in 1598 from a Quarto, known as Qo, which originated in the same shop, and of which only a fragment survives. The copy for Qo is usually held to have been authorial papers of a tolerably "fair" kind, though the possibility of a non-Shakespearian transcript

has also been entertained: the present compositorial analysis will help illuminate the problem.

Hinman claims that "both spellings and various typographical peculiarities indicate that the whole of Q1 was set by the same compositor." (Clarendon facsimile, p. ix) Yet, as Peter Davison pointed out in his New Penguin edition (1968) of 1 Henry IV, the skeleton pattern of the Quarto suggests that two compositors set the text, perhaps for two presses (pp. 251-252). The pattern is similar to that of Short's section of Richard III: again we have the combination of a one-skeleton method of printing and the use of different skeletons in alternate sheets: one skeleton was used for both formes of sheets A, C, E, G, and I, and another for both formes of sheets B, D, F, H, and K; on D1v and D2, and on D3v and D4 the recto running-titles appear on the verso pages, and vice versa.[15]

Knowing the compositorial division of labour which accompanies this sort of skeleton pattern in Short's part of Richard III, and with some expectation that evidence from the spacing of punctuation will again prove valuable, we can readily distinguish the stints of two compositors. Table 3 records for each page the number of question marks which are and are not preceded by a space when set within short lines.

Table 3: Question marks preceded / not preceded by a space in 1 Henry IV (1598).

                 

	A	B	C	D	E	F	G	H	I	K
1		2/0	0/2	0/12	3/1	0/4	0/1	0/7	0/1	2/2
1v		0/1	0/1	0/9	4/2	0/2	0/3	0/0	0/0	3/3
2	0/0	0/0	0/0	0/2	0/2	0/1	0/0	0/3	0/3	1/1
2v	0/0	0/0	0/3	0/7	1/2	0/0	0/0	0/2	0/0	0/0
3	0/4	3/0	2/1	0/6	0/4	0/0	0/2	2/1	0/0	0/1
3v	0/3	0/4	1/1	1/4	0/4	0/1	4/3	1/1	0/0	1/3
4	0/7	0/4	3/0	3/1	0/2	0/1	4/1	0/0	2/2	0/3
4v	0/2	0/0	0/0	2/0	0/0	1/0	0/0	0/0	0/0	0/0

The outline of a regular compositorial division by half of a sheet already begins to emerge, and is clarified by differences in the signatures. The non-spacing compositor (Compositor X) regularly used arabic numerals in signatures; the spacing compositor (Compositor Y) mostly used roman numerals:

         

Compositor X					Compositor Y
A.2	D 1	E 3	H 1		B.i	E	H.iii	K 3
A.3	D 2	F 1	H 2		B.ii	F.iii	I 3
C.1	D 3	F 2	I 1.		B.iii	G.1.	K 1
C 2	E 2	G 2	I 2.		C.iii	G.iii.	K 2

If, as I shall give further reasons for doing, we attribute B1-B3, B4v, C3-C4v, D3v-E1v, E2v, F3-G1, G3-G4, H3-H4v, and I3-K3v to Compositor Y, and the rest of the text to Compositor X, then Compositor X's refusal to set a space before question marks within short lines was unremitting (there are 107 unspaced question marks within short lines in his stints), whereas Compositor Y inserted a space more often (46 times) than not (36 times):      

	Spaced ? marks	Unspaced ? marks
Compositor X	0	107
Compositor Y	46	36

The figures for justified lines display a similar pattern: Compositor Y spaced 33 question marks, and failed to space 31; Compositor X spaced 5 and failed to space 22. Compositor X was far more inclined to set a space before a colon in a long line than in a short line; Compositor Y's rate of spacing colons was unaffected by justification:        

	Colons in short lines		Colons in long lines
	spaced	unspaced	spaced	unspaced
Compositor X	7	64	26	7
Compositor Y	22	17	38	25

Neither compositor normally inserted a space after a comma within an unjustified line, but Compositor Y, who always spaced question marks, spaced twice as many commas as Compositor X. Poor inking and the use of commas of several different shapes and sizes make the evidence so ambiguous that it is impossible to reach exactly the same totals of spaced and unspaced commas in successive counts, but Compositor Y appears to have spaced some 16 per cent of the commas within short lines in his stints, Compositor X 8 per cent.[16]

Exit directions in the Quarto are set to the right. Most are stopped, but some are not. The distribution of stopped and unstopped exit directions is as follows:

     

	Stopped	Unstopped
Compositor X	5	10
Compositor Y	18	1

The difference between X's and Y's stints is highly significant statistically (Yates's chi-square = 11.771, 1 d.f., p<0.001).

There is ample spelling evidence to substantiate the proposed allotment of shares, which gives Compositor X 40 pages, Compositor Y 38. Most notably, Compositor Y had a decided preference for -tie endings, Compositor X a no less strong preference for -ty endings. The overall figures for unjustified lines, with additional figures for justified lines in parentheses, are as follows:

     

	-tie	endings	-ty	endings
Compositor X	1	(4)	30	(9)
Compositor Y	41	(16)	2	(4)

Compositor Y set -ie and -y endings generally in about equal numbers; Compositor X avoided -ie endings. Figures for -ie endings in unjustified lines, excluding monosyllables, names,[17] first elements in hyphenated compounds, and -sie and -fie endings (where a ligature is involved), are as follows:      

	-ie endings	-y endings
Compositor X	4	126
Compositor Y	73	82

Compositor Y strongly preferred the modern spelling of shall, whereas Compositor X was scarcely less willing to set shal. The figures for occurrences in unjustified lines are:      

	shall	shall
Compositor X	31	32
Compositor Y	4	24

The difference between the two men is statistically significant (Yates's chisquare = 8.57, 1 d.f., p<0.005). Compositor X was also much more tolerant than Compositor Y of al and wil. In short lines X set al 10 times and wil 14 times; Y set each spelling 5 times. Thus X's total for shal, wil, and al in unjustified lines is 55; Y's is 14.

Two further words over which the compositors display sharply contrasting spelling habits are blood and enough:

         

	Compositor X	Compositor Y
blood	0	9
bloud	10	5
enough	6	1
inough	0	6

Finally, one of the Quarto's "various typographical peculiarities" helps establish that two compositors set the text, and that the proposed division is essentially correct. The presence in the Quarto of the 'oo' ligature has been remarked by Hinman. There are, on my count, 134 instances; but the un-ligatured 'oo' occurs 235 times. If a single compositor set the text, we might expect a random distribution of the alternative forms; if my proposed allotment of compositorial shares is more or less right, it would be somewhat surprising to find each man setting the same proportion of 'oo' ligatures. The actual figures are as follows:      

	Ligature	Separate sorts	% ligatures
Compositor X	48	126	28%
Compositor Y	86	109	44%

The disparity here may not seem particularly remarkable, but it is statistically significant: there is less than a one in five hundred probability of its having occurred purely by chance (Yates's chi-squire = 10.16, 1 d.f., p<0.005).[18] However, availability of the ligature, as supplies were diminished through composition and replenished through distribution, is another possible factor in the use or non-use of the ligature.

Figure 1 presents the incidence of the main compositorial markers in graphical form. There the overall pattern is clear enough—a division of labour by half of a sheet appears to have been the intended norm—but for some pages any ascription must be very tentative. As Figure 1 has three X markers for every two Y markers, the latter must be given compensatory extra weight. On B1v the X markers are shal, wil (3 times), and a single unspaced question mark; these are among the less reliable discriminators—7 of Y's 9 shal and wil spellings in short lines are on his first three pages, B1-B2—and they are outweighed by the Y markers: 2 spaced colons and 2 -tie endings, one in a short line. D3v contains Y's only enough spelling, but if we were to attribute the page to X it would contain the sole spaced question mark in X's stints; the 2 spaced commas, out of 6, also suggest Y. On E2, where the ascription to X produces a break in a sequence of Y's pages, the markers are merely 2 unspaced question marks in short lines on a page with no spaced question marks; this is painfully slight evidence, but there are also 3 unspaced question marks in justified lines, and the spellings bloud and thril slightly suggest X, as does the absence of ligatured 'oo' on a page with 7 unligatured 'oo' forms. On G4v the evidence for X is a single -ty ending. On K2v the character of the 4 X markers does not quite warrant our assigning this page to X when the surrounding pages are by Y: the X markers are merely shal, al (twice), and one unspaced colon; the Y markers are one spaced colon and one stopped exit; this stopped exit would constitute strong evidence of Y were it not anomalously stopped with a colon after an abbreviation (Exit Ki:). Other uncertain ascriptions are of F3v, H3v, K3, and K3v to Y, and of I2v to X.

Some of the contrasting spelling habits of Compositors X and Y, deduced from Q1 as a whole, can be confirmed by examination of those pages for which their Qo copy survives. These Q1 pages are, for Compositor X: C1-C2v; and for Compositor Y: B4v (in part), C3-C4, and a few lines of C4v. Compositor X changed -tie to -ty once, and made 5 other alterations from -ie to -y endings, 3 of them, however, in long lines. Compositor Y changed -y to -ie 3 times; 2 of the changes occur in long lines, but Y's acceptance of 14 -ie endings from his copy, including 2 -tie endings in short lines, and 3 in long lines, sharply distinguishes him from Compositor X. Also significant are Compositor Y's 2 changes of enough to inough. Compositor X had accepted enough on C2v. The treatment by X and Y of -l and -ll spellings of shall, will, and all is more problematical, but if we confine our attention to short lines the evidence is at least consistent with our earlier deductions that X is much more tolerant of the short forms of these monosyllables than is Y: Compositor Y changed shall to shall twice in short lines; Compositor X's only changes in short lines were in the same direction—shal to shall twice, and wil to will twice—but he accepted his copy's shal twice and wil twice in short lines on C1v.

The fidelity with with Q1 reproduces the surviving portion of Qo has often been remarked.[19] Only three substantive variants have been noted:

whipt replaces whip at I.iii.239, C1 = X, mine owne replaces my owne at II.ii.37, C3v = Y, and fat is omitted from "How the fat rogue roard" at II.ii.118, C4 = Y. The first variant is an obvious and universally accepted correction, the second a plausible "improvement", rejected by the New Arden editor A. H. Humphreys (1960), and the third an error of omission. Compositor X also changed Saine in II.i.70, C2v, to Saint. The OED records Sayn as a 14th-16th century spelling of Saint, and Sayne and Sain as 15th century variants. So X's change can be regarded as a normalization of spelling.

One definite error and one unauthoritative correction over a whole sheet is a good record for Short's two compositors: X's work was verbally perfect. Both compositors were unusually faithful to the punctuation of Qo.

Much, therefore, depends upon the accuracy with which Qo itself was set. Can we detect the hand of either of Q1's compositors in the Qo fragment? The fact that five of the eight pages are in prose makes judgement difficult, because the evidence is distorted by the requirements of justification. The overall figures for spacing suggest Compositor Y: 7 question marks spaced, 6 not spaced, in short lines; 17% of commas spaced. The 6 -tie spellings (3 in unjustified lines) point in the same direction; there are no -ty spellings. And the roman numerals for signatures form another link with Y's stints in Q1. Moreover the exits (one on C2, one on C3, and 2 on C4v) are all stopped. The compositorial markers graphed in Figure 1, which, for the most part, reveal clear divisions in Q1, suggest that Compositor Y may have set Qo C2-C4v (21 Y markers, 8 X markers), and Compositor X C1-C1v (3 Y markers, 12 X markers).[20] Reading through the section one is struck by the prevalence of -l endings in C1-C1v, and their rarity thereafter, and whereas within C1-C1v shal occurs 6 times, shall not at all, on C2v, also a verse page, one example of shal is followed by 4 of shall. There is one further indication that X may have set C1-C1v and Y C2-C4v. In Q1, until we reach G3, Y's pages, 21 altogether, are consistently free from apostrophes except that B4v has taken over 2 (-'d and -c't) from Qo, whereas 14 of X's 27 pages contain at least one apostrophe, and the total number of apostrophes is 22—all but one (in ther's on C2v, where it has been taken over from Qo) in past participial -'d or -c't endings. On G3 Y sets one apostrophe, and thenceforth he is indistinguishable from X in his treatment of -'d and -c't endings. No other change in Y's habits occurs at G3, so there is no reason to suspect that from G3 onwards "Y" is really some new workman. Now in Qo, C1 has 2 apostrophes in -c't endings, and C1v 2 in -'d endings. Pages C2-C4v, in contrast, afford only the anomalous ther's on C3. So, in this feature also, pages C1-C1v

of Qo match Compositor X's pages in Q1, and pages C2-C4v match Compositor Y's.[21]

If X set C1-C1v and Y set C2-C4v, the only real anomalies are the presence of the roman signature on C1, for which I can offer no good explanation, and the 3 enough spellings on C2v, C3v, and C4. Y's preference for inough could well have developed as he set Q1; compositors sometimes accept copy spellings as they set a text for the first time, and harden in their preferences as they work on a reprint. Such a process might adequately explain the change which both compositors made from oy and ay to oi and ai spellings: X made 6 changes of this kind, Y 7, if we include 3 from Poynes to Poines. Y also made one change in the opposite direction, from faith to fayth on C3. Some oy and ay spellings still survive in Q1, though few of them are medial, and on C3v-C4v Compositor Y retained Poynes 7 times, and even expanded the speech prefix Po. to Poynes on C4v (as well as expanding it once to Poin. on the same page).

The evidence is not sufficient to enable us definitely to decide whether Compositors X and Y set the surviving fragment of Qo, but on the whole it seems to me more likely than not that they did so. Both formes of sheet C of Qo were evidently printed from a single skeleton, so one may speculate that the printing of Qo followed the same pattern as 1 Henry IV Q1 and the Richard III Quarto—a one skeleton method combining with alternation of two skeletons from sheet to sheet.

If Compositors X and Y did share the setting of Qo, as well as of Q1, we have good reason to suppose that Q1, though at two removes from the manuscript copy, is a fairly accurate representation of it. Was this manuscript holograph or a transcript? There is general agreement that copy cannot, in view of the Quarto stage directions, have been a prompt-book, but the formality of the orthography, where colloquialisms and contractions seem better suited to the tone or demanded by the metre, has led several scholars to suppose that "the copy for Qo was rather a scribal transcript of the foul papers than these papers themselves."[22] Of fourteen unelided forms which the New Arden editor lists in illustration of this feature of Q1, 6 appear in X's pages, 8 in Y's; 4 of them derive from sheet C of Qo, where they fall on

pages tentatively assigned to Y.[23] As Q1 follows Qo faithfully in these instances, it probably does so in the others. I think we must assume that absence of elisions was a feature of the manuscript underlying Qo, for the evidence which falls short of proving that Compositors X and Y set sheet C of Qo is at least clearly indicative of two workmen, and whether or not they were X and Y, they are unlikely to have shared a partiality for "sedate expansions."[24]

One small item of evidence, previously unnoticed, comes close to clinching the case for a non-Shakespearian transcript as copy for Qo. David J. Lake has shown, in The Canon of Thomas Middleton's Plays (1975), that playwrights had strong individual preferences with regard to variant forms of connectives, such as among and amongst, while and whilst, betwixt and between, preferences which may be consistently realized over many texts, set in diverse printing-houses. Now Shakespeare's plays exhibit 246 instances of between, to only 56 of betwixt.[25] In only three plays does betwixt occur more often than between—in Richard II, where the figures are between 1, betwixt 3; in Cymbeline, where the figures are between 3, betwixt 4; and in 1 Henry IV, where betwixt appears 7 times, between not at all. 1 Henry IV is thus easily the most anomalous play in its use of these forms. The contrast with 2 Henry IV, where the Quarto of 1600 has obviously been set from foul papers, is especially striking: there are 9 instances of between, none of betwixt. Again, we may argue that Compositors X and Y were not the sort of workmen to alter between to betwixt, and even if two other men set Qo it would be surprising if both were prejudiced against between, or if all instances of the connective fell into the stints of the one man who did have such a prejudice. Qo must have been set from a scribal transcript.

Among/amongst supports the same conclusion. Shakespeare's overall preference is for among (95 times, amongst 41), though in eight plays amongst outnumbers among. The figures for 1 Henry IV (among 1, amongst 3) contrast with the thoroughly typical figures for 2 Henry IV (among 6, amongst 1).

These same connectives furnish interesting evidence about the texts of Richard III. In F between occurs 8 times, betwixt once. F thus reflects Shakespeare's normal habits. In Q, in contrast, between appears 3 times, betwixt 7 times.[26] The lesser authority of Q is thus reflected in its atypical ratio of

betwixt to between—also, perhaps, in its treatment of among/amongst: F does not use amongst, but twice has Shakespeare's preferred among; Q once has nothing at the corresponding point, and once has amongst.

Can either of the two compositors who set 1 Henry IV be identified with either of those who set Richard III? It is impossible to be sure. The closest match is between Compositor X of 1 Henry IV and Compositor N of Richard III. Their spacing habits are identical: neither ever precedes a question mark by a space in an unjustified line, and neither spaces more than a tiny proportion of colons or commas. The spelling evidence is ambiguous: some of their preferences coincide (do and bloud, for example), some differ. N and X may be the same workman, influenced by different spelling patterns in copy.

There is less correspondence between Compositors Y and O. In their treatment of question marks and colons they are similar, setting spaces before about half of them, but though Y spaces more commas than X, he falls far short of O's actual preference for spaced commas. More significantly, in Richard III there is no trace of Y's liking for roman numerals in signatures. Spelling patterns bear some similarities, but many differences.

It is time to attempt a summary of my findings and a discussion of their significance. First, it has been shown that two compositors (N and O) set Richard III, Q1 (1597), sheets H-M, and that 1 Henry IV, Q1 (1598), another product of Peter Short's printing house, was also set by two men (X and Y), not one, as has been supposed. The surviving fragment of 1 Henry IV, Qo, was probably set by two compositors, who may plausibly be identified with X and Y. It is possible, even likely, that Compositors X and N are one and the same man.

Short's compositors divided typesetting evenly, setting half of a signature —exactly in Richard III, sometimes approximately in 1 Henry IV. This division of labour is associated with a skeleton pattern whereby a single skeleton is used for each forme of a sheet, but two skeletons alternate in successive sheets. A skeleton pattern of this kind is, in my experience, very common in sixteenth and seventeenth century Quartos, and equal division of typesetting by two men is a frequent accompaniment.

Sheets A-G of Richard III, printed by Valentine Simmes, appear to have been set by one man. Thus three compositors worked on the play, punctuating the text in widely different ways. There is some evidence of heterogeneity in the manuscript copy for the Quarto. Moreover, variations in the spelling of O / Oh, which persist into Quarto reprints, are not perpetuated in F—a fact which may be relevant to the question of copy for F Richard III.

There are excellent reasons for thinking that Q1 of 1 Henry IV was set with unusual fidelity to Qo, of which it was a reprint, and less substantial grounds for supposing that Qo itself fairly accurately reproduced its manuscript copy. This manuscript must, however, as a new item of evidence shows, have been a transcript of authorial papers, rather than actual holograph. For this reason, I believe that an editor should have no compunction about introducing colloquial and metrical elisions into his text, on his own authority and in deference to the obvious raciness of the dialogue—when, that is, he judges that Shakespeare himself is more likely to have written the contracted or elided than the full form. This is a contentious point, I know, but why should an editor act as if his own authority were less than that of some pedantic scribe?

Bibliography, Fredson Bowers once stated, rests on "the impersonal inductive interpretation of physical facts according to rigorous laws of evidence."[27] Ideally this may be so, but in practice to undertake the bibliographical analysis of an Elizabethan play Quarto is rather like undergoing a Rorschach test. The investigator is confronted with physical facts certainly, but what he notices will differ from what another bibliographer might have noticed, and the facts observed may be open to more than one interpretation. Bibliographers seldom re-do one another's work. When a scholar of Hinman's stature claims that the 1598 Quarto of 1 Henry IV was set by a single compositor, his verdict, even though he withholds his evidence, will, quite understandably, be regarded as final. But Hinman was mistaken—as, in the matter of compositor discrimination, the most competent among us will often prove to be. The bibliographer who finds no more than one compositor in any but the shortest sixteenth-century Quarto has probably not stared at it for long enough or done enough counting. Howard-Hill's evidence from the spacing of punctuation has gone a long way towards settling the question of compositorial shares in the Shakespeare First Folio.[28] The advantage of this type of evidence is that, unlike spelling, it ought to be unaffected by copy. Examination of spacing practices, though apt to dull both eye and mind (for "Light seeking light doth light of light beguile"), should form part of any bibliographical analysis of an Elizabethan Quarto. We cannot afford to ignore anything of possible consequence. For, as D. F. McKenzie insisted, "It cannot . . . be too frequently stressed how difficult the task of isolating a compositor's work may be."[29]

Notes