University of Virginia Library

Compositor Identification in Romeo Q1 and Troilus
by
W. Craig Ferguson

The handling of spacing materials by Elizabethan compositors has recently been a subject of attention. Trevor Howard-Hill has demonstrated that compositors in the Shakespeare First Folio can be distinguished by the ways they space commas.[1] The point is made that this is a 'psychomechanical' habit on which the copy has little or no bearing, and thus is more useful as evidence than spellings, which may have been influenced by the copy. Greg's term 'accidental' might apply to the use of spaces, but perhaps we should take up the suggestion by Tom Davis, and refer to the handling of spaces and quads as part of a compositor's 'style.'[2]

I wish first to examine the use of spacing materials in the positioning of speech prefixes to see if compositorial style is evident. When a compositor sets a prefix, it is usual for him to distinguish it from the text. On rare occasions (as in the Ben Jonson quartos and 1616 folio) he will set it flush left, but by far the most common practice was to indent the prefix, and to do so he must begin by placing a blank space in his composing stick. But which space will he use? Is he indifferent to this and likely to use any spacing material that comes to hand, or will he reach habitually to one of the four boxes available to him? I believe that he will habitually reach for his preferred spacing, either an em quad, an en quad, or a 3-to-em space. The reach for a quad takes him to an entirely different part of his case from a reach for a space, and if it is done consistently one way or the other I think we have evidence of a stylistic habit which can help in compositor identification.

This is not all there is to the prefix; there remain the form of the prefix, the spacing, if any, between prefix and punctuation mark, the mark itself, and the spacing between completed prefix and the text. Does the compositor make an effort to place the same spacing after as before, thus setting the prefix neatly off on the page? Does he consider the spacing material as a primary or secondary variable when he has to justify a line? Does he have a habitual way with punctuation?

As a preliminary example, consider the prefixes in parts of the first quarto of Pericles. Here there are three compositors in two shops. One compositor, x, working in William White's shop, set sheets A and C-E. The remaining sheets, B and F-I, were set in Thomas Creede's shop by two compositors, y and z.[3] White's x and Creede's y uniformly indent the prefix by an em, but compositor z, who set B3-4v, F3v,4,G3-H2v, and I2v-3v can be very neatly identified by his habit of indenting speech prefixes two ems, which is rather excessive. This peculiarity has already been noted by David Hoeniger in the 1963 New Arden edition of the play (p. xxvii), where he states that Richard Hosley pointed out to him that one of the compositors of the text can be spotted by his habit of deep indentation of prefixes. I mention this case only to make the point that such seemingly small details can be decisive in compositor identification.

As a further example, let us look at the first quarto of Romeo and Juliet. The first four sheets (A-D) were printed by John Danter, and the remainder (E-K) by Edward Allde. There has been a view that the switch was the result of Danter's arrest during Lent of 1597, but recent scholarship suggests that it was a normal shared commercial venture.[4]

The presence of two compositors in Allde's section was detected many years ago by Harry R. Hoppe. He determined that one man, X, set E1, E2v-4v, F3-4v, G3-4v, H3-4v, I3-4v, and K3,3v, while the other, Y, set E1v,2, F1-2v, G1-2v, H1-2v, I1-2v, K1-2v and K4, basically an even division in which one took the first half of a gathering and his mate took the second. "From F3 until sig K the alternation is as regular as a pendulum," Hoppe remarked.

Hoppe's evidence was basically spellings, bolstered by apostrophe use. Compositor X preferred 'doo, frier, here, yong,' the speech prefix 'Capo.', '-ie' endings, and no apostrophes in contracted forms, whereas Y preferred 'doe, fryer, here, young,' the prefix 'Cap.', '-y' endings, and the use of apostrophes in contracted forms, such as 'wher's'.

There are two further preferences which match the two compositors and aid in identification. The first of these, which Hoppe noticed, was a tendency of X to use a swash 'I' in 'Iuliet', but this use of swash is actually part of a larger pattern, and holds true for all the letters for which there were swash sorts: 'A, B, C, E, I, M, N, P'. We find that X, setting a total of 358 italic caps, reached for swash sorts 131 times, a 37% use, while Y, setting 187 italic caps, used swash 28 times, only a 15% use.

There is a second preference in the choice of types, but this was not noticed by Hoppe. When selecting a colon to complete the italic prefix, X was somewhat more likely to continue to reach into his italic case, instead of making

the switch to his roman. He set 280 colons in his stint, and selected an italic sort 164 times, for a 59% preference. Compositor Y, on the other hand, preferred turning back to his roman case, and setting 177 colons, used italic 42 times, for a 24% usage.

I also wish to examine other evidence which Hoppe did not consider. Certainly as clear as spellings and swash or italics in identifying these two compositors are their different ways of setting speech prefixes. The prefix in each case consists of indentation by at least an en, the prefix in italic, and then a colon, followed by a space. However, on certain pages I found that the compositor frequently inserted a thin space between the prefix and the colon. Following these pages through the text one finds that this habit is seen only on the pages Hoppe assigned to compositor X, and also on F1v, F2, G1, and G1v assigned by Hoppe to Y. I would suggest that, as evidence, it is at least as strong as spelling preferences.

The additional evidence of spacing, swash, and colons essentially backs up Hoppe's assignments to the two compositors, but there are five doubtful pages which bear further examination. The first of these is E1. Hoppe was hesitant about assigning this page, as the evidence was flimsy, consisting of five '-ie' endings and a single spelling. The evidence, such as it was, pointed to X. This page, consistently, has the space before the colon, a marked preference (13 of 15) for italic colons, and the use of swash over a third of the time (6 of 15), suggesting that he was right in assuming X was the compositor.

I do not agree with Hoppe's attribution of F1v and F2 to compositor Y. The first of these pages has both spaced and unspaced colons after prefixes, seven spaced and five unspaced, of which three of the unspaced could be attributed to justification. The spelling evidence is slight, and Hoppe records two uses of the apostrophe in elisions (wher's and hee's), but he does not consider five occasions (pepperd, yfaith, heele, staind and tane) where an apostrophe could have been used and was not.

When one looks at italic colons and swash caps, the evidence for an attribution to X grows stronger, with an 8 to 4 preference for italic colons, and a 40% use of swash, 7 out of 17. I would suggest that Hoppe is wrong in assigning this page to compositor Y, and I assign it to X.

On F2 there is a slightly different pattern. The page begins with spaced colons, and then after two speeches and eight lines moves to the non-spaced pattern. The spelling evidence Hoppe advances is not definitive. He suggests that while X always used '-ie' endings in preference to '-y' endings, compositor Y will allow '-ie' endings on occasion. On this page, the only example, "beautie', occurs on the first line, and is associated with the spaced-colon prefixes. On the other hand, a decided Y preference, 'heere', is found in the non-spaced section of the page. Might one not suggest that one compositor, in this case X, began the page, and that his mate took over from him at line nine, ten, or eleven?[5]

The evidence on pages G1 and G1v suggests sharing again. Hoppe gave both pages to Y, which would fit the pattern of dividing gatherings in half. But on G1 there are 13 prefixes, 7 having spaces before the colon, which is

italic: an X form. There is a concentration of 3 non-spaced prefixes with roman colons in the last two lines and catchword which suggests to me that X set at least 22 lines, and Y came in only to finish the last few lines. The exact point of change cannot be determined, as 6 lines of speech come between the last X prefix and the first Y form, but Y set at least the last two lines and the direction line. There are 3 non-spaced prefixes, 'Fr:' on X's part of the page (one of which could be justified) which remain anomalies.

The Y pattern continues at the top of G1v, where the first line contains a Y form of prefix, and the six-line speech has two elision apostrophes and one Y spelling. This is followed at line 7 by an X-form prefix: swash cap, space, italic colon. In turn this is followed by two more X forms, then the Y pattern, suggesting that Y began the page, set 6 lines, was replaced by X who set 12 lines, and then back to Y at line 19, who finished the page.

I would suggest, then, that Hoppe was essentially right in his assignment of this quarto to two compositors, and that his evidence is backed up by the prefix composition and the use of swash, which strengthens his attribution of E1 to X. I would disagree with Hoppe on the assignment of F1v, F2, G1, and G1v, and note that he himself had doubts about the first four pages of sheet F.

The next text I wish to examine is the first quarto of Troilus and Cressida, printed in 1609 by George Eld. In this play it has been long understood that two compositors, A and B, set the play between them, with compositor A setting A2-B2v, C3-D2v, E3-G4v, H3-I2v, K1-3v, K4v, and L3-M1v, and B setting B3-C2v, D3-E2v, H1-2v, I3-4v, K4, and L1-2v. The evidence, first produced by Phillip Williams,[6] has been reviewed by Alice Walker in the 1957 New Cambridge edition of the play (p. 129).

One aspect not examined by Williams or Walker was the spacing of the speech prefixes, and this produces an interesting pattern. The prefix consists of an indentation, the prefix in italic (except where the compositors had run out of italic caps and had to resort to roman), a period or a colon, and a space. What varies, though, is the amount of indentation before the prefix, resulting from the use of either an en quad or the narrower three-to-em space. The wider en quad appears more often than the thinner three-to-em space. Two things about the thinner spaces are peculiar. They do not appear at random, but are grouped together on the page: either an entire page or a part of one. Where they are found on part of a page, the concentration is usually towards the bottom of the page, completing it. The second point is that they all occur on pages which have been assigned to A, who otherwise used en quads. Thus I wish to argue that the use of the thinner three-to-em space to indent the prefix is a habit of a third compositor, C, who assisted compositor A.

There should be other evidence to support the view that a third man was involved, and one useful test is to extract those lines of the text suspected to have been the work of C, and run a spelling check on them, together with a check on A and B's stints.

I used the Oxford Concordance Program to run my spelling checks.[7] The tapes of Troilus and several other plays were dumped into our mainframe,

and the computer asked to generate concordances by line numbers which corresponded to the lines I assigned to each compositor. As one cannot assume that C would take over from A at the beginning of a new speech, the lines of dialogue between the last A and the first C prefix were omitted from the concordances.

Assuming only two compositors, compositor A would have set 64% of the play, B would have set 36%. However, I suggest that B still set the 36% previously assigned to him, but that A set 40%, and C set 24%.

There are two problems with sorting out the compositors of Troilus. First, one must isolate B, who worked alone, and then separate A and C. Some spelling variations not previously noted are peculiar to B and confirm the assignments of Williams and Walker, while others are useful in separating A from C. The significant spellings are as follows:

                                       

A	B	C
Aga.(10)	Agam.(2)	Aga.(3)	Agam.(25)	Aga.(11)	Agam.(4)
Dio.(24)	Diom.(4)	Dio.(0)	Diom.(4)	Dio.(10)	Diom.(13)
Pan.(56)	Pand.(31)	Pan.(2)	Pand.(16)	Pan.(41)	Pand.(3)
Patr.(9)	Patro.(6)	Patr.(2)	Patro.(7)	Patr.(0)	Patro.(4)
Ther.(19)	Thers(i).(16)	Ther.(8)	Thers(i).(3)	Ther.(31)	Thers(i).(2)
Tro(y).(70)	Troyl.(0)	Tro(y).(11)	Troyl.(22)	Tro(y).(28)	Troyl.(2)
Ulis.(22)	Uliss.(12)	Ulis.(13)	Uliss.(4)	Ulis.(8)	Uliss.(11)
Aeneas(11)	Eneas(0)	Aeneas(10)	Eneas(0)	Aeneas(4)	Eneas(4)
Cressid(a)(21)	Cresseid(a)(4)	Cressid(a)(1)	Cresseid(a)(15)	Cressid(a)(8)	Cresseid(a)(0)
Helen(2)	Hellen(15)	Helen(6)	Hellen(1)	Helen(0)	Hellen(6)
blood(y)(3)	bloud(y)(6)	blood(y)(2)	bloud(y)(8)	blood(y)(0)	bloud(y)(13)
do(46)	doe(17)	do(36)	doe(6)	do(30)	doe(2)
go(18)	goe(10)	go(25)	goe(6)	go(10)	goe(9)
here(26)	heere(10)	here(13)	heere(21)	here(19)	heere(1)
lie(4)	lye(5)	lie(3)	lye(4)	lie(3)	lye(0)
O(30)	Oh(14)	O(28)	Oh(0)	O(5)	Oh(10)
sinews(1)	sinewes(1)	sinews(2)	sinewes(0)	sinews(0)	sinewes(3)
sweet(14)	sweete(15)	sweet(4)	sweete(4)	sweet(13)	sweete(3)
th'(elision)(13)		th'(elision)(6)		th'(elision)(0)

One notes in passing B's preferences for longer prefixes, such as 'Agam.', 'Troyl.', and 'Pand.' instead of 'Aga.', 'Troy.', and 'Pan.'. It is interesting also to note that B never used a colon as punctuation after a prefix: always (268 times) a period. The other two, between them, used colons about a fifth of the time; to be specific, 172 of their 816 prefixes were followed by colons.

We come now to separating A from C. The two men, I think it can safely be assumed, worked from the same type cases, as the unit for shortages of italic caps is the page, and not the stints of each compositor. Had they been using separate cases, then one would hardly expect them to run low in exactly the same places. The most distinctive characteristic distinguishing the two is their treatment of speech prefixes: C uses the narrower three-to-em space to indent the prefix, while A uses the thicker en quad. By glancing down the page one can see how C's prefixes are closer to the left-hand margin than A's.

There are two kinds of spelling evidence. In some cases A preferred one form and C preferred another, but in several instances one man would have a preference while his mate was indifferent. Take their approach to the spelling of prefixes, for example. C used the longer prefix, 'Diom.', 13 times, the shorter, 'Dio.', 10. More or less indifferent. Compositor A, on the other hand, had a decided preference for the shorter form, using it 24 times to the longer form's 4. A similar pattern emerges with the prefixes for Pandarus, only this time the evidence is the reverse. A is more or less indifferent, using the short form, 'Pan.', 56 times and the longer, 'Pand.', 31. C, on the other hand, used the short form almost exclusively, 41 to 3. A third pattern is seen in C's preference for the longer 'Vliss.' over 'Vlis.' (11 to 8) against A's 22 to 11 preference for the shorter form.

Common words show similar patterns. C never used the form 'blood(y)', using only 'bloud(y)', 13 times. A was less consistent, using 'blood(y)' a third of the time. C never used the form 'Cresseid(a)' and was almost invariant in preferring 'here' to 'heere', while A was less consistent. C preferred 'Oh' to 'O', two to one, while A was just the opposite. Compositor C had a decided preference for 'sweet' over 'sweete', while A was indifferent. C never used the contraction 'th'', while A used it 13 times. C was indifferent to 'go/goe', while A preferred the shorter form; but C was more consistent than A in preferring 'do' to 'doe'. Other spellings are less useful because of their infrequency.[8]

Therefore I wish to assign the following portions of compositor A's stint to the new workman, C: all of A3v, B2, G4v, H4, I2v, K1v,2, and L3-M1v. He shared the following pages with A, setting A2 (last 8 lines), A4 (last 24), B1 (last 9), B2v (last 18), C4 (last 18), C4v (last 17), D1 (last 32), D2 (lines 8 to 23), E3 (last 24), E3v (last 14), E4 (last 31), F2v (last 7), F3 (last 7), G2 (last 17), I1 (last 13), I1v (lines 12 to 24), I2 (lines 1 to 8, 14 to 38), and K4v (last 12).

We come finally to consider the reliability of these compositors. There are in the quarto 128 clear compositor errors.[9] Had the men been equally competent, B, setting 36% of the text, would have made 46 errors, A, setting 40%, 51, and C, setting 24%, 31. However, we find that B made 31 errors, A made 54, but C made 43, many more, proportionally, than the others. B was clearly the most accurate of the three, A fairly accurate, but C was by far the least accurate.

The three men made different kinds of errors. A omitted words 16 times, B only twice, and C 7 times. On the other hand, A had the best record of not substituting words of similar sound, doing so only 10 times, while B did it 8, and C 14 times. (One must keep in mind with these figures the percentages of text each man set.)

I do not think it would be reasonable to assert that A set all the pages previously assigned to him and that his accuracy deteriorated in the lines he set when he just happened to reach for a space instead of a quad to indent a prefix. Rather, the evidence of a habit of prefix setting, together with the evidence of spelling habits and relative accuracy, are not coincidental and point to the presence of a third, less competent, compositor.

Notes