University of Virginia Library

Roberts' Compositors in Titus Andronicus Q2
by
Paul L. Cantrell and George Walton Williams

The problem of distinguishing the work of James Roberts' two compositors, X and Y, in Titus Andronicus Q2 (1600) provides an interesting compositorial analysis. John Russell Brown first revealed the orthographical personalities of these two men and identified their work in two other Roberts' quartos, Hamlet Q2 (1604/5) and The Merchant of Venice (1600).[1]

The evidence for the respective stints of the workmen in the two latter plays is quite definite and clear-cut. Titus Andronicus, however, offers a different problem for two main reasons: first, because it is a fairly accurate reprint of a previous edition; second, because the two men did not set their respective shares of copy in well-defined and obvious bibliographical units.

The printer's copy for Titus Andronicus Q2 was an exemplar of the first edition printed in 1594 by John Danter. Hamlet Q2 and The Merchant of Venice were set from manuscript. It is ordinarily accepted that the identification of different compositors is more difficult in a reprint than in an original edition set from manuscript. In some cases it may be impossible. The reasons for this are easy to see. Considering the somewhat casual nature of Elizabethan manuscript copy, it is only natural that when a compositor comes to set up this manuscript into type, to some extent he will—almost unwittingly—impose his own personal orthographical and typographical characteristics upon it in his ordering it for the press. From previous studies, it appears that no two compositors are likely to do this in the same way.[2] The transition is a real one, a matter of transferring comparatively unorganized handwritten material into organized material suitable for printing.

The case for a reprint is quite different. Here the copy is printed material

with an orthographical and structural order which has already been imposed upon it by previous compositors. The normal tendency is for the reprint compositors conveniently to follow the original print in many matters of form, spelling, and typographical detail. In a recent article on compositorial determination, Dr. Alice Walker writes: "Reprints are, of course, a trickier problem than first editions, since we must expect the trail to be confused by the spellings of the print used as copy. Roberts's reprint of Titus Andronicus certainly gives a much distorted picture of the habits of the compositors who set it and suggests that it may be little use to try to arrive at a compositor's normal practice from a reprint."[3]

The problem of Titus Q2 is further complicated by the fact that in the reprint X and Y did not combine to set their material in a normal pattern for two-compositor work in which each man serves a different press. In fact, the peculiar feature of Titus is that there should be a second compositor at all. The running-title pattern indicates no such second workman. The book was printed throughout with one skeleton-forme, and so necessarily on one press, with the one set of four running-titles transferred from inner forme to outer forme in orderly succession. The only aberration occurs in sheet K where the recto quarters were exchanged diagonally in the forme as they were shifted from outer I to inner K, and then are maintained in that reversed position into outer K. There is nothing in this minor shift, however, that has anything to do with the question of a second compositor. And this second compositor, Roberts' X, is at work long before K.

In Hamlet and The Merchant, each compositor's work-unit was the sheet, with the sole exception of sheet L in Hamlet, where X set two pages, L1r and L4v, and Y the remainder, L1v-L4r. In Hamlet, X set sheets B-D, F, I, N, and O + A; Y set sheets E, G, H, M, K, and L (with the exception of two pages as noted above[4] ). In The Merchant, Y set sheets

A, B, D, F, and H; X set sheets C, E, G, I, and K. Thus in these two plays an equable pattern of composing is observed which is not to be seen in Titus, where Compositor Y seems to have been responsible for the bulk of the reprint, but we find X assisting him at unpredictable intervals and for differing lengths of time. Such irregularity, along with the fact that Titus is a reprint, makes it harder to distinguish the work-pattern of the compositors.

Mr. Brown's work on Titus Andronicus naturally stemmed from the data for his successful spelling tests in Hamlet and The Merchant, but many of the significant factors did not apply since Titus was a reprint. Of the twenty-odd paired spellings for X and Y which he found in Hamlet and The Merchant,[5] only nine have any significance for Titus, and these in varying degrees. Those that he did find, however, are meaningful because they definitely identify X and Y as being the same two compositors in Roberts' shop.

On the strength of all of the evidence he could garner which separated X and Y in Hamlet and The Merchant, Mr. Brown tentatively assigned 38 of the 79 pages of type (A1v is blank) in Titus to Y, and 14 pages to X, leaving 27 pages unassigned. His allocation is as follows:

• Y: A1, 2, 3v, 4, 4v, B2v, 4, C1, 1v, 2, 3v, D1, 1v, 2, 4, E1, 1v, 2v, 3v, 4, F1, 1v, 2, 3v, 4, 4v, G1v, 2, 4v, H2, 3, I1, 1v, 4v, K1, 1v, 2, 2v.
• X: B1, 1v, 2, 4v, C4v, D2v, 3, 3v, E3, 4v, F3, I2v, 3, 3v.

It is the purpose of this present paper to indicate the results of another examination of the compositorial problem in Titus, an examination which will amend and augment Mr. Brown's findings and which will allow of a new and more certain allocation of pages. This further examination confirms how many of the characteristics of Hamlet and The Merchant are not applicable to Titus, owing to the retention of many uncharacteristic forms by the two compositors in the reprint. Re-examination also indicates that at least four of the spelling factors which Mr. Brown used for tentative

attribution to X and Y in Titus are without supported relevance and must be omitted from the analysis. These are the -ow,-ew/-owe,-ewe distinctions, and the spellings honor, etc./honour, etc., sweete/sweet, maddam(e)/madam (e), which Mr. Brown notes are "not entirely trustworthy" (p. 34).

This leaves him a small core of nine very strong differentiating factors. These invariable factors, which we may designate as Class I of the evidence, are the more significant because their Q2 form in Titus is a change from that in the Q1 copy. These words and the pages on which they occur in Q2 are:

                       

	Q1	Q2	Q2 Pages
(Y-FORMS)	their	theyr	A1, A2, A4(3), A4v, D1(2), E1(2), E1v, F1v(2), F2, F4v, G1v(2), I1, I1v(3), I4v(3), K1, K1v(3), K2v.
	mooued	moued	F1v
	deare, -er	deere, -er	A4(2), C1, D2, E2v, E3v(2), E4, F1, F1v, I1v, K1, K2v.
	being	beeing	D1v
	ile	Ile	I1
	choise	choyse	F3v
	noise	noyse[6]	C4
(X-FORMS)	Ile	ile	D3, E3, E4v, I3(2), I3v
	moude, proue	moou'd, prooue	B4v, D2v
	perceiue, deceiude	perceaue, deceau'd	E3, I2v
	howres	houres	C4v

Two facts may be instantly noted from this array: seven of Y's pages and one for X are confirmed by the appearance of two factors, and no pages contain conflicting X and Y evidence.

Further, since the Ile/ile distinction is so strong in The Merchant of Venice (printed the same year as Titus), we may with some confidence admit as oblique but suggestive evidence those instances where Ile is copied from Q1 to Q2 as being Y's normal practice; similarly, the one instance where ile is copied from Q1 to Q2 we may attribute to X. This attribution allows further confirmation of seven pages already allocated to Y, tentatively adds four new pages for him, and adds one new page for X. These are the occurrences of the straight copying:[7]

• Y: Ile B2v(2), D1v, D4, E2v(2), F1(3), F4v(3), G3v, H2v, I4v, K1, K1v.
• X: ile H4v.

Still no conflicts occur. This fact, along with the fact that seven previously-established pages were confirmed for Y, is almost the sole justification for the admission of such evidence. Proof of this sort is tenuous and inferential at best, for it may very well be that the compositor was faithfully adhering to his printed copy. This faithfulness is a practice which X, unfortunately, appears all too prone to follow. The point is that in analyzing a reprint, inferential evidence of a negative character is sometimes the only evidence available because of the nature of the work. It should only be used, ideally, when at least three of the following four conditions are met: (1) when no conflicts develop; (2) when confirmations for previously-allocated pages are obtained;[8] (3) when the factor is a very strong one, verified by reference to other works set by the same man; (4) when there is an almost invariable positive factor (a definite change of the same word in the copy-spelling) by the other compositor to balance the permissive, negative factor. Of such a nature is X's spelling of deare, and Y's spelling of howrely and here, which will be discussed later.

The re-application of a spelling test to the three Roberts' plays also uncovered several pieces of fresh evidence which fell into three general categories:

• (1) New, differentiating characteristics found in a reappraisal of Hamlet and The Merchant, unnoted by Brown, which also have significance for Titus.
• (2) Identifying factors, very few in number, which serve to differentiate the two men in Titus, but which do not appear in either Hamlet or The Merchant.
• (3) Pertinent copy-spellings which are retained from Q1, but which, under the limitations just discussed above, it seems safe to admit as permissive factors of negative evidence.

These three categories of fresh evidence we may designate Class II. Obviously, these categories are not of equal weight, and are valuable in a descending scale of importance. The last two are considerably weaker than the first, which, because not invariable, is itself weaker than those invariable factors in Class I with which we began.

In the first category of Class II—those factors which are applicable to all three plays—five more determining characteristics for Y were uncovered; none, regrettably, for X. These five are: (1) Y's preference for the ampersand, which he uses about three times as much as X, a practice particularly revealing in The Merchant;[9] (2) Y's exclusive use of the tilde,

usually for justification;[10] (3) his preference for the oo-form in doone, dooing, and in doost;[11] (4) his occasional spelling of do;[12] (5) and his spelling of greefe(-s, -ues).[13]

The following assignments in Titus are predicated on these fresh distinctions:

               

(Y-FORMS)	Q1	Q2	Q2 Pages
	and	&[14]	B1v, C3v, D2, E1
	done	doone	I1v, K2
	doing	dooing	I1
	dost	doost	C2v, D3v, D4, E1v, F2, G2(2), I4v
	doe	do	F3, I4v
	them	thē	I4v
	griefe, etc.	greefe, etc.	C1(2), E4, I2

In addition to confirming ten Y-pages, this last group of spellings common to all three plays has added eight new pages to Y's span, B1v, C2v, C3v, D3v, F3, G2, I2, and K2, pages which had not as yet been assigned.

In the second category of new evidence, four identifying factors unique to Titus were found, one of which serves to separate X and Y, but the other three to add to our pages for Y. It appears certain that Compositor Y altered the Q1 speech prefix Moore to Aron in Q2 on C2v, D1, F1, and G1v. In a typographical change, he altered the alignment of stage directions on E1, where a two-line reverse pyramid is converted to one in which the lines are equally indented, and on C4, G3v, K1, and K1v, where paragraphed stage directions are identically altered to this indented form. Sig. K1v offers a typical example:

• Q1: Sound Trumpets. Enter Emperour and Empresse with Tribunes and others.
• Q2: Sound Trumpets. Enter Emperour and Empresse with Tribunes and others.

It is also fairly certain, as Brown assumed, that Compositor Y changed the Q1 spelling of this to thys on C1v(2), C2(2), C3v, F4(2), and I1v. In addition, pages for both X and Y may be allocated by their differing spelling, both in the text and in the speech prefixes, of the name of Saturninus' brother, as indicated by the following divergences from the copy:      

	Q1	Q2	Q2 Pages
Y:	Bascianus	Bassianus[15]	B2v(2), C1, D1, D1v
X:	Bassianus	Bascianus	A3, B4v, D4v(6)

All of these four practices (for they can hardly be called characteristics on the basis of one play) have the virtue of confirming an array of Y-pages already allocated to him on other evidence. The three new pages which are added for him—C2, C4, and F4—have no X evidence in them. Conversely, of the three pages attributed to X on the basis of his spelling of Bascianus, one of them, B4v, is now confirmed, and two new ones—A3 and D4v—are added, with no conflict with any Y evidence. In fact, the more evidence that accumulates, the more meaningful the lack of any Y characteristics becomes for any given page.

The last category of new evidence in Class II, three examples of negative copying, may be cited with less certainty. The stipulations that must be observed before this sort of evidence can be admitted have already been set out above. The spellings are:[16]

       

	Q1	Q2	Q2 Pages
Y:	here	here	A4v(5), B1v, D1v, D2(3), E1, E3v, F1v(2), F2, F3v(2), G1v(2), G1(2), G2(2), G4v(3), H1v(3), H2, K1.
	howrely	howrely	G2v
X:	deare(-er, -ly)	deare(-er, -ly)[17]	B4, F2v, F3, H4.

Compositor Y's copying of here in Q2 as it is found in Q1 is a case in point for permissive evidence. It is again inferential, but the attribution is formulated on sound support. Compositor X is almost invariably a heere-speller;

only twice (once for obvious justification) does he use the simpler spelling in The Merchant (G2,K1), and at no time in Hamlet. Y uses the short form nine times in The Merchant (A2v, D2(2), H1(2), H2(2), H2v, H3v), and twice in Hamlet (M1v(2)). Of the 31 times this short spelling is used in Titus, on no occasion does a conflict develop with X evidence.[18] One copy-spelling of deare conflicts with Y's spelling of do on F3 (see below).

Finally, in Class III, there are two other rather general typographical practices which seem indicative of differing treatment by the two men, not always distinctive, but more correlative than not. Each practice involves a choice on the part of the compositors: the first is the placement of one sort of stage direction; the second is the spelling of the speech prefixes in the transfer from the original to the reprint.

In the matter of certain stage directions, the point of difference is their comparative centering in the line, whether it is precise or only approximate. Excluded from consideration are those stage directions which are not amenable to centering (or which appeared so to X and Y)—those either obviously flush right or nearly right in Q1 (such as Exit, Exeunt, etc.), or those (like the indented forms already noted as set by Y) which, whatever their indentation or paragraphing, continue to the right margin, or nearly so, and take up the bulk of the line. Within a tolerance of two millimeters, Compositor X as a general proposition made sure that his medial stage directions were precisely centered in the line, or to put it another way, were centered in his long stick. This involved setting the stage direction first, then placing quads or blank spaces equally on either side of the setting until the line was of the proper length. Compositor Y, however, while he obviously intended these stage directions to be approximately in the middle of the line, made no such precise effort to center them, and apparently guessed at the spacing on either side; hence, his medial stage directions usually deviate from the midpoint of the line in varying degrees. Into his stick, which was the same length as X's, Y would evidently put first the amount of quads or spaces which he felt would place the stage direction in the approximate center of the line, set the stage direction, then fill the line out with the number of spaces required for justification. This procedure most often yielded an off-center setting, but, depending on the accuracy of the guess, sometimes accidentally resulted in a centered setting.

The 22 pages in which these precisely-centered stage directions occur may be divided as follows: (1) nine pages where X's presence is confirmed: A3, B4v, C4v, D4v, E3, E4v, H1, H4, and I2v; (2) six hitherto neutral pages on which no prior evidence has been established either way: A3v, B3, B3v, E2, H3v, and I4; (3) seven pages attributed to Y with which this evidence conflicts: A2v, D3v, D4, F1v, F3, G4v, and H1v.

Non-centered stage directions which we have postulated as the result of Y's practice occur on no X-pages, a fact certainly meaningful, and thus serve to confirm the following 13 pages already suggested for Y: A4, A4v, C2, C4, D1, D2, E1v, E2v, F1, H2, H2v, I2, and K2.

While not always decisive, the above evidence does show a high degree of correlation with what has already been established with antecedent material.[19]

In the matter of speech prefixes, from what is observable in the pages already assigned to the two men it would appear that X's general tendency is to copy the speech prefix exactly as it stands in Q1; on the other hand, it seems to have been Y's general tendency to alter the original settings several times on a page, more often by contraction than by expansion. Such a broad division tends to support many pages previously allocated. This general correlation can be seen by examining what happened to the speech prefixes on the pages which, either on strong evidence or weak, have been already allocated in this study. In the following table, the numerator of the fraction represents the number of changes made in the speech prefixes, the denominator the number of speech prefixes on the page, and therefore the number of opportunities for change of copy.[20]

• (X-PAGES)[21] B4 0/13; B4v 0/8, C4v 0/9; D2v 2/12; D3 0/10; D4v 0/9; E3 0/8; E4v 0/7; F2v 0/1; H1 0/12; H4 0/3; H4v 2/9; I2v 0/6; I3 1/7; I3v 2/7; K3v 0/4; K4 0/5.
• (Y-PAGES) A4 2/7; A4v 0/3; B1v 3/11; B2v 4/18; C1 1/4; C1v 3/7; C2 1/3; C2v 5/9; C3v 3/9; C4 0/4; D1 1/2; D1v 1/7; D2 3/7; D3v 2/10; D4 3/8; E1 5/13; E1v 3/3; E2v 0/4; E3v 0/5; E4 2/7; F1 1/10; F1v 1/6; F2 2/8; F3 3/7; F3v 4/11; F4 2/4; F4v 0/7; G1 3/8; G1v 7/14; G2 6/20; G2v 2/8; G3v 2/4; G4v 0/6; H1v 1/7; H2 0/2; H2v 2/9; I1 2/8; I2 2/9; I4v 2/15; K1 0/1; K1v 2/6; K2 2/13; K2v 2/11.

Analysis of this array is tabulated as follows:

     

Compositor	no. of pages	no. of pages on which changes occur	no. of pages on which no changes occur	total no. of prefix changes	total no. of prefixes	% of change
X	17	4	13	7	130	5.3
Y	43	35	8	90	335	26.8

There is no doubt that the difference in the percentage of change is statistically valid.

Lack of speech-prefix change is the only evidence appearing on six pages, B1, C3[22], G3, G4, H3, and K3, where 4, 7, 9, 5, 6, and 1 speech prefixes, respectively, are unchanged; but since the practice is not invariable, any attribution to X must still be indecisive. On the other hand, four pages contain only speech-prefix evidence combined with comparative-centering evidence: A3v, B3v, E3, and H3v, where both practices combine to suggest X.[23] Two pages contain no other evidence except that of speech-prefix changes: B2 3/5; and K4v 1/2.

The final allocation of pages in Titus is therefore predicated on material of various degrees of assurance. In Class I are placed those pages for which there is strong evidence:[24]

• Y: A1, A2, A4, A4v, C1, C4, D1, D1v, D2, E1, E1v, E2v, E3v, E4, F1, F1v, F2, F3v, F4v, G1v, I1, I1v, I4v, K1, K1v, K2v.
• X: B4v, C4v, D2v, D3, E3, E4v, I2v, I3, I3v.

In Class II are those assignments for which there is fresh evidence based on similar occurrences in Hamlet and The Merchant:

• Y: B1v, C2v, C3v, D3v, D4, F3, G2, I2, K2;

for which there is fresh evidence unique for Titus:

• Y: B2v, C1v, C2, F4, G3v
• X: A3, D4v;

and for which there is only negative permissive evidence:

• Y: A2v, G1, G2v, G4v, H1v, H2, H2v
• X: B4, F2v, H1, H4, H4v, K3v, K4.

In Class III are those assignments for which the evidence is purely tentative, based on a correlation of speech-prefix and comparative-centering evidence only:

• X: A3v, B3v, E2, H3v.

If we may bring together all ranges of evidence, the pattern of assignment is:

There remains only briefly to consider why Titus was composed in such an irregular pattern, or why Compositor X came to assist on the book at all. One hypothesis, consonant with what we know went on in Elizabethan printing shops, would in general explain X's irregular share of the work and also account for the fact that most of the X-pages occur late in the sheet. This hypothesis, based on the assumption that Titus was composed in continuous but not simultaneous setting by the two men, is simply that X was called in to help while Y was distributing the type from a forme that already had been printed. We know that distribution was a necessary and regular operation throughout the printing of a book. If, at the times Y was distributing, X were available, or could easily be made available, this postulation could be feasible. Since, theoretically, Y would be ready for distribution at about the same time in relation to the composing schedule of each sheet—distribution which would not of course take exactly the same amount of time for each forme—Compositor X could then be called in approximately at the same time in the schedule for each sheet. While Y was distributing, it is reasonable to suppose that X continued with the setting of new pages. If X was not in precise control of the exact moment when he could become available, this delay would account for his irregular entrances into the sheets, perhaps in the middle or lower part of a page, since, presumably, Y would leave off composing only

when X appeared. This could easily have happened when, in a stint of two or three consecutive X-pages, an X-factor occurs only in the lower section of the first page of the stint. Similarly, on the conclusion of his distribution, Y would not always return to relieve X at the precise moment of a bibliographical break within the forme, and could take over at some place within a page which X had begun.

Since, as has been estimated, distribution time is about one-third of composing time, this division of the work would explain the shorter share of X, which on the basis of the allocations posited in this study is about one-third that of Y's.

In conclusion, any complete division of Titus Andronicus by compositors is at this moment impracticable. The larger picture is reasonably clear, but beyond the allocation of those pages established in this study, it is now difficult to proceed without an undue amount of conjecture. Further progress must await two desiderata, the accumulation of additional knowledge of the compositorial habits of X and Y by examination of other works, preferably plays, set by these two men, either together or singly; and subsequent research from Titus itself, or Hamlet, or The Merchant of Venice, which will reveal significant materials unnoticed so far.

The present results are not completely inutile: enough definite pages have been noted for the two men to be of use in analyzing what they did to printed copy, information which has its use elsewhere.

Notes