University of Virginia Library

Compositors D and F of the Shakespeare First Folio
by
John O'Connor

Any analysis of the transmission of a Renaissance text from manuscript to printed book must include the role of the printing-house compositor, the man who sets the play in type. The potential significance of his job is overwhelming; his skill and conscientiousness greatly influence the difference between the manuscript copy and the printed text. But before the nature and degree of his influence can be evaluated, the compositor's share of the text must be identified through a systematic analysis of the physical traits of the text. The process of evaluation is the more significant part of compositorial analysis, but the process of identification must precede it.

A detailed knowledge of compositors' habits will have an immediate and significant effect on editions of Shakespeare in the treatment of accidentals (spelling, punctuation, capitalization, italics, and word-division), in lineation of verse or prose, and in the selection of emendations. This new information will also increase our understanding of the passage of copy into print. As we strip away the veneer of compositorial features, the characteristics of the no-longer-extant copy begin to appear. Thus, compositorial analysis not only can tell a bibliographer and editor what happened to the copy in the printing house, but also a little about the copy itself.[1]

Any study of compositors is cumulative: it builds upon previous research, and hopes that its addition is a solid foundation for future work. The early research in compositorial analysis of the Shakespeare First Folio is well known, and is briefly summarized by Charlton Hinman in his Preface to the Norton Facsimile of the First Folio of

Shakespeare.[2] Since then scholars who have attempted to identify and evaluate individual Folio compositors have achieved only mixed results. Monographs have been published on Compositors B and C, but neither has proved definitive.[3] During the past two years scholars have become especially interested in Compositors C and D, because they appeared to be more important than originally thought. However, although a major break-through has been made in establishing certain characteristics of Compositor C, his share in the Folio is still subject to some dispute; as for Compositors D and the newly discovered F[4], their roles in the printing of the Comedies have not been adequately defined nor have the characteristics been established by which they may be accurately identified. Both men are minor compositors and worked only on the Comedies; nonetheless, in the process of identifying their shares I have assigned some pages to Compositors B and C that were previously thought to have been set by Compositors D and F, and I have developed criteria that can be used with the other compositors.

Compared to the other Folio compositors, D and F are unknowns. They are shadowy figures, easily confused with Compositors A and C, and remarkably similar to each other in their compositorial habits. Not only do they spell doe | goe | here the same, but also most of the other words used to isolate Compositors A and C, like cosin, divell, griefe, indeede, mistris, and yong. Besides their superficial spelling similarities, two other circumstances make it difficult to separate Compositors D and F. First, they are minor compositors and therefore the amount of evidence from which standards can be established is slight. Often a distinctive spelling preference of Compositor D's does not appear on pages that might belong to Compositor F, and is therefore of little use in distinguishing who set those pages. The problem of limited evidence is intensified when we discover that Compositor F, unlike the other five compositors, never set a play from Quarto copy. Thus, for Compositor F, we are unable to develop any qualitative

evidence; that is, a list of changes he made from extant copy which clearly reveals his preferences. Instead we must rely upon spelling counts and other mass statistical data (quantitative evidence)[5] in order to describe Compositor F's preferences. As for Compositor D, although he did set portions of four Folio plays that used extant Quartos for copy (Ado, LLL, MND, and MV), he accepted the copy spellings so frequently that it is difficult to describe exactly what were his spelling preferences. For example, he changed copy adieu to adue four times (twice in long lines), but he also retained copy spellings adiew (2) and adew (in a long line), and changed adiew to adew once. Nevertheless, the information that the collation of Quartos and Folio provides will be crucial evidence later, and will demonstrate the value of qualitative over quantitative evidence.

The second difficulty in separating Compositors D and F is that they worked together on only one play, and even then in different quires (quires F and G in MM). Thus evidence, such as speech prefixes, which is unique to each play is of little value in distinguishing Compositor D from Compositor F. Case evidence is also of limited use, since Compositors D and F could set type from the same case at different times. In the one play on which they both worked, Hinman was unable to make sense of the type recurrence and could not identify the cases (quires F-H).[6] However, special evidence may prove invaluable in distinguishing these two compositors from the other workmen. For example, Compositor D prefers to italicize Jew in MV, and though this feature is of no use in separating him from Compositor F, who works in other plays, it very neatly distinguishes Compositor D from Compositors B and C, who prefer to set Jew in roman.

A more promising kind of evidence is psychomechanical,[7] such as has been developed to distinguish Folio compositors by the spacing of stage directions. It is more promising because it extends beyond the individual play, and is less affected by non-compositorial influences, like lack of space. Yet as with their spelling preferences, Compositors D and F are strikingly similar in their habits: both men use a similar form for their catchwords and stage directions. On the other hand, they do differ in their treatment of a verse line that is too long for their measure, and in their punctuation habits.

Before examining this evidence in detail, I shall consider articles by two men who have recently written on Compositors D and F. Dr. Andrew Cairncross, in two brief articles, attempted to identify the work of the Folio compositors through spelling and psychomechanical evidence.[8] Most of the habits which Cairncross identifies are common to two or more compositors; however, he attempts to present at least one unique habit for each workman. Compositor D, according to Cairncross, is "distinguished specially by his consistent weele forms; by his ie or aie endings, as in eie, saie, betraie, onelie; and by the dropping of final l, as in cal, fal, wil" (p. 44). The only distinctive features Cairncross lists for Compositor F are his general preference for shorter speech prefixes, and his preference for only the speech prefix (and not the prefix plus the first word of the speech) as a catchword. Compositor F, he states, elides the future tense like Compositor C (we'll), and his spelling preferences, doe, goe, and here, are the same as Compositor D's.

Although Cairncross isolates some significant features in the work of Compositors D and F, he is not sufficiently rigorous in his examination of all the evidence. He frequently misses important counter-examples to a 'habit' which he has identified and used as evidence. One habit of Compositor C's noticed by Cairncross is actually a spacing convention common to almost all of the Folio compositors. He asserts that Compositor C indents the initial speech prefix of the play, but the initial speech prefix is indented for every play in the Folio, except three where the prefix is missing.[9] One may notice the variable amount of indentation in the following two examples:

What is important, and what Cairncross fails to realize, is that each compositor differs from the others in the amount of indentation he prefers.

SPACING OF THE INITIAL SPEECH PREFIX

                               

Play	Page	Space (inches)	Compositor
Tmp	A1	20/32[*]	B
TGV	B4v	25/32	F
Wiv	D2	16/32	C
MM	F1	13/32[*]	C
Err	H1	16/32	C
Ado	I3	1&14/32	B
LLL	L1v	16/32	C
MND	N1	1&10/32	B
MV	04	1&14/32	B
AYL	Q3	18/32	C
Shr	S2v	1&18/32	B
AWW	V1v	1&10/32	B
TN	Y2	24/32	B?
WT	Aal	8/32	A
KJ	al	15/32	C

A clear pattern emerges: Compositor B centers the initial prefix, Compositor C indents approximately ½", and Compositor F indents approximately 3/4". The difference between Compositor C and Compositor F is slight but will be crucial evidence later when when I reject Cairncross' use of catchwords as a standard for separating Compositors D and F.

Similarly, his criteria for distinguishing Compositor D need to be sharpened. He examines the changes Compositor D makes from the Quarto copy on pages L2 and L4v-6v (LLL), but fails to check whether the features apparent there are consistent with other portions of Compositor D's work before generalizing about them. In LLL Compositor D shortens words with a final double ll to a single l seven times. On this evidence Cairncross asserts that Compositor D can be "distinguished specially . . . by the dropping of final l" (p. 44). However, on pages K2v and K5v-6v (Ado) Compositor D consistently changes his Quarto copy's single final l to double ll: still (1393), will (1504, 2197, 2382, 2473, 2488), all (2384, 2548), fall (2479), shall (2199, 2215, 2338, 2384), and well (2308). In addition, at least sixteen other instances of this change from single final l to double ll occur elsewhere in Ado in long lines. Finally, the same change from the copy's single

final l to double ll occurs, to a lesser degree, in LLL, MND, and MV. [10] Obviously the single final l is not a consistent feature of Compositor D's work and cannot be used to identify him.

On the other hand, Cairncross' other spelling standard for Compositor D, the -ie/aie endings, is at least partially accurate. Compositor D does seem to prefer the -ie/aie endings, but some important counter-examples such as day, way, stay, onely, try, and country make mass counts and simple generalizations about -ie/aie endings potentially misleading. In fact, T. H. Howard-Hill is misled into rejecting this feature as evidence for Compositor D by relying on mass counts. In reviewing Cairncross' attributions he asserts, "when one surveys 5 pages (O5v, O6, T5, V3 and V3v) which he attributes (correctly) to D, there are 86 -ie/aie to 70 -y/ay spellings" (pp. 73-74). This use of quantitative evidence is deceiving. First, since Compositor D prefers the -y/ay endings in some words which occur frequently in the text, a comparison of the total number of -y/ay endings with the total number of -ie/aie endings on a single page can be inconclusive or misleading. A single page could have more -y/ay endings than -ie/aie endings and still be set by Compositor D. Second, a comparison between the number of -ie/aie and -y/ay endings in Compositor D's pages needs a standard of reference. We should compare the number of -ie/aie and -y/ay endings in the five pages set by Compositor D listed above with five pages set by Compositor C or Compositor F. When I surveyed five pages (P4v, P5, T1, V4, and V4v) attributed to Compositor C, I counted 132 -y/ay endings and 34 -ie/aie endings.[11] Thus, a larger selection of pages can reveal a difference between compositors, but this generalization provides only limited help when one tries to identify the compositor of a particular page. Although Cairncross' use of -ie/aie endings as evidence for Compositor D must be qualified further, his method of noting the changes the compositor makes from Quarto copy is generally more revealing than simple quantitative, statistical evidence.

For Cairncross, once the work of Compositors C and D is isolated, the work of Compositor F is easy to distinguish. Some difficulties arise,

however, because Cairncross' standards for determining the work of Compositors C and D need additional support before many ambiguous pages can be identified, and because Compositor F shares a preference for doe, goe, here with Compositor D, and a preference for wee'll with Compositor C. According to Cairncross, the one feature that distinguishes Compositor F from both Compositor C and Compositor D is his preference for the short catchword, the speech prefix without the first word of dialogue.

Using this criterion, Cairncross argues that the catchword Sp. I on B4v is evidence for Compositor C (p. 397), but later he believes that Compositor F is "the only possible compositor" for D3v (p. 399)— despite its very similar catchword Qu. I. Actually Compositor F set both of these pages. Cairncross is led astray on B4v because of his reliance on the indented initial speech prefix (B4v is the head page of Two Gentlemen of Verona), which I have already shown to be evidence for Compositor F's hand because of its degree of indentation. Cairncross acknowledges, however, the force of the other reasons for believing B4v was set by Compositor F: "doe-goe-here spellings and short [speech] prefixes, e.g. Sp., and the x-case used" (p. 398).

Compositor F also uses the long catchword (the speech prefix and first word of text) on E3, F3, F5, and G4. These six instances of the long form spoil any generalization about Compositor F's preference for the short form, which he uses on five occasions (B1, C4, C4v, C5, and D3). Although Compositors C and D prefer the long form, the short catchword is of limited use as evidence for identifying Compositor F. Even when the short form appears, Compositor F did not necessarily set it, since Compositor C will set the short form on occasion, as on D2, K1v, and K2.

Since I have modified or challenged a number of the criteria which Cairncross proposes, let me summarize those criteria which I accept either in their original or modified form:

     

	Compositor D	Compositor F
future-tense contraction	-le	'll
-ie/aie or -y/ay	-ie/y/ay	-y/ay usually

     

catchword	long form	both forms
2nd line of a stage direction	centers	prefers to center, but will indent
initial speech prefix of a play	does not set a title page	indents 3/4"

Obviously these standards will leave many pages unassigned or doubtful, especially since they frequently coincide with Compositor C's preferences.

The other scholar who has studied these two compositors is T. H. Howard-Hill. In a long, thorough article, published before Cairncross' essays, he offered elaborate quantitative evidence for isolating the work of four of the Folio compositors. Despite the limitations in his method and occasional errors in his tabulation, Howard-Hill's results are a major accomplishment. In his area of research and his findings Cairncross occasionally overlaps Howard-Hill's earlier research, but Howard-Hill is more successful in distinguishing the new Compositor F of the Comedies from Compositor A of the Histories and Tragedies, and in isolating Compositor C from Compositors D and F. His criteria for separating Compositor D from Compositor F needs some revisions, but they provide a base from which other evidence may be constructed. His three general criteria for identification of the Comedies' compositors are: the future-tense elisions as later adopted by Cairncross, the spacing of commas at the end of a line and inside short lines, and the typographical arrangement of a verse too long for a single line.

His main standard, a spaced comma that is the result of setting a spacing quad at the end of a word before a comma, is usually easy to recognize at the end of a line.

In fact, the second example shows a spacing quad which was accidentally inked. This spacing of the comma at the end of a line is the distinctive feature of Compositor C's work, and one of the most important discoveries of the article, since it successfully isolates Compositor C, most of the time, from Compositors A, B, D, and F, who rarely leave a space before a comma at the end of a line (p. 68). As a result of this new criterion, Howard-Hill assigns Compositor C many previously undetermined pages. I have adopted this authoritative criterion for identifying Compositor C. The features of Compositor C's work that I have noticed and will describe later are merely supporting evidence, although they are important on prose pages where spacing evidence is usually invalid.

Howard-Hill also uses as evidence each compositor's supposedly consistent choice either to make a space after a comma that occurs within a short line or to follow the comma closely with the first letter of the next word. According to Howard-Hill, Compositor B prefers to leave a space after an internal comma; Compositor C also prefers to leave a space after an internal comma, but is not so consistent as Compositor B. Compositors D and F prefer not to leave a space after an internal comma, but Compositor F, Howard-Hill admits, is inconsistent in his spacing practices. In fact, I find the evidence for internal spacing varies so much for both Compositors D and F that it is worthless as an indicator of their work. The proportion of spaced internal commas to non-spaced commas in Compositor D's pages in LLL and MND (L2, L4v-6v, N4-5, N6-6v) is roughly one to four. But in pages from AWW and Shr (T5-6v and V3-3v) also set by Compositor D, the proportion is roughly one to one, with notable instances of three to one (T5 and V3). Internal spacing practices can offer, at best, a check for assigning pages to Compositors B and C. If the number of unspaced internal commas is greater than the number of spaced ones, Compositor B or Compositor C probably did not set the page. On the other hand, if the number of spaced internal commas is greater than the number of unspaced ones, any of the compositors could have set the page.

Compositor D's most distinctive feature for Howard-Hill is his habit of carrying over a verse line too long for his measure to the following line and indenting it two or three ems from the left-hand margin (p. 71).

Howard-Hill calls these lines indented turn-overs, but for reasons of clarity I will refer to them as indented flow-overs and thereby avoid confusion with the alternate method of setting an extra-long line, which Compositor F prefers. He sets an extra-long line at the right-hand margin above or below the end of a line; this is a turn-over or turn-under.

Howard-Hill charts (p. 71) the frequency of indented flow-overs in the pages that he personally assigns to the compositors:

INDENTED FLOW-OVERS

     

	B	C	D	F
pages	14	2	20	1
frequency	20	2	94	1

Seemingly Compositor D uses indented flow-overs on more pages and more frequently per page than the other compositors, and Howard-Hill concludes: "It is more likely than not that any page with more than one such indented turn-over [indented flow-over] was set by Compositor D, and it is very unlikely to have been set by A(F) or C" (p. 72). Unfortunately, this criterion, and the conclusion, must be modified upon closer examination of the evidence. The ratio of indented flow-overs per page implied in Howard-Hill's chart is misleading. Thirty-six of these indented flow-overs occur on H3v, which, as we will see shortly, was actually set by Compositor C. Another twenty-seven that Compositor D did set are on 21/2 pages (H4b, and L6-6v) of tumbling-verse dialogue like H3v in Err and LLL, where little or no room exists for a turn-over. The remaining thirty-one indented flow-overs that are set by Compositor D occur on 161/2 pages. Thus the statistical evidence in Howard-Hill's chart is badly distorted. (A similar distortion would be to say that Compositor C set thirty-eight indented flow-overs on three pages.) Compositor D uses the indented flow-over on significantly more pages than Compositors C and F, but the number of occurrences on a single page has little significance; i.e. if an indented flow-over occurs on a page, the page probably belongs to Compositor D, but the number of times it occurs is not important. Despite the fact that Compositor D sets indented flow-overs more frequently, additional evidence is necessary to assure us that the page was not actually set by Compositors B or C, who will use the indented flow-over occasionally. H3v is an example of how the use of only this standard can mislead. Howard-Hill assigned the page to Compositor D, reasoning that "[C] is unlikely, despite the spaced comma in column a, to have set H3v which has many indented turn-overs [flow-overs] of verse lines . . . I have assigned the remainder of Err., where verse turn-overs are frequent, to D" (pp. 82-83). But the indented flow-overs and the solitary spaced comma are not the only evidence on this page. It also contains an indented two-line stage direction, three instances of you'll, and y e, y u, and y t—all evidence for Compositor C.[12] No firm evidence exists for Compositor D. The traditional evidence of doe, goe, here/heere is inconclusive for this page because of the number of long lines requiring justification.    

	do	doe	go	goe	here	heere
H3v		:1[*]	1:2	2:1	1:6	:1

Hinman was unable to determine which type case was used for this page because of the generally irregular distribution of type in quires F through H. However, he did trace certain distinctive types through these quires, so that a clear pattern emerges here and there. Eight of Hinman's distinctive types appear on this page, but only one also appears in a column set by Compositor D before or after H3v, whereas the rest appear in columns associated with the y-case and Compositors B and C (II, 388). Also, only three -ie spellings occur on this page, despite indications that the Folio copy had -ie spellings.[13] Thus the evidence for Compositor C is considerable, even if no piece of evidence is conclusive by itself.

Although the evidence from indented flow-overs is not so clearcut as Howard-Hill thought, it is still a valuable start for separating the work of Compositors D and F. And despite my examples of Howard-Hill's misuse of quantitative evidence, his work is usually judicious and I will frequently refer to his discussion of particular quires in the Comedies.

My examination of the recent work in compositorial analysis has focused primarily on Compositor D, since both Cairncross and Howard-Hill have failed to develop sufficient standards to distinguish

him from the other compositors, especially Compositor F. A possible first step in determining which pages belong to Compositors D and F is to decide which pages were not set by them. This "negative" method is why I have evaluated the evidence offered by Howard-Hill for identifying Compositor C, who is most likely to be confused with Compositors D and F, and why I wish to add another criterion for Compositor C that is particularly useful on pages where the other distinctions are ineffective.

Although Howard-Hill's criterion of the spaced final comma is adequate for many pages in the Folio, a difficulty arises when one is confronted with a page containing mostly prose. Since most spelling tests are also less trustworthy if not invalid when the word is in a justified line as in prose, further evidence is needed to identify Compositor C on such pages. Fortunately, Compositor C has another spacing-spelling habit which he employs on these pages. He contracts yet, you, and occasionally we by placing the last letter above the first.

Although Compositor B also uses this technique, other evidence can easily distinguish him from Compositor C. In the Comedies I have counted thirty-eight instances of these contractions outside of Compositor B's pages.[14] Twenty-five of the contractions occur on pages every-one agrees belong to Compositor C. Since these contractions never appear on the undisputed pages of either Compositor D or Compositor F, despite frequent occasions for them to save space by contracting yet, you, and we, I believe the remaining thirteen contractions on eight pages (D2v, F4v, G4v, H2v, H3v, I4, K5, and M1) are sound evidence for Compositor C. On these eight pages other evidence also suggests Compositor C, although the contractions are especially important on pages D2v, F4v, G4v, H3v, and O2, where they confirm Howard-Hill's tentative assignments.

Determining which pages belong to Compositor C solves only part of the problem of identifying Compositors D and F: it is time to suggest further standards for identifying their work. When describing Compositor D, Hinman (I, 196) assigned him only pages set from the new case z, which was used for parts of quires K-L and N-Q. These

quires, with the exception of Q, contain the comedies that used Quartos for their copy (quire K: Ado; quire L; LLL; quire N: MND; quires O and P: MV). If we adopt as a temporary hypothesis that, at the least, Compositor D is responsible for the pages set from case z, we can make some generalizations about his distinctive features on these pages by noting the changes from the Quarto copy; then we should check back to see if any of the pages involved appear anomalous according to his identified characteristics. Later we will add pages from quires T and V set from a case made up de novo, which Hinman calls case x (II, 443, 452), and a few columns in quires F and H, where the case evidence is confused. This evidence suggests that Compositor D entered the Folio in quire F, after the printing of Wilson's Christian Dictionary was finished, that he completely replaced Compositor F after MM, and that he left during a "major interruption" in the printing of the Folio in quire V (II, 462). He was then replaced by Compositor A, who worked on the remainder of the Folio.

Most of the features that are characteristic of Compositor D are also features of either Compositor C or Compositor F. Thus, some of Compositor D's characteristics can be contrasted only with Compositor C's; some other evidence must separate Compositor D from Compositor F.

The spelling variant most apparent in the z-case pages is the frequent use of -ie endings. But the many -y endings also on these pages make any large generalizations dangerous and misleading. First, one must distinguish between Compositor D's choice of -y or -ie endings after a consonant or after a vowel. Cairncross has argued that Compositor D prefers final -aie over -ay (p. 74). But, with the exception of nineteen instances of praie (in contrast to thirty-three instances of pray), Compositor D consistently prefers final -y after all vowels. He will, it is true, often change -ay- to -ai- in the plural form of a word ending in -ay, as in waies and saies, but this variant is frequently used by all of the Folio compositors.

Second, although Compositor D often changes word endings after consonants from -y to -ie, he frequently retains the -y endings from the copy, and in any simple mass count their frequency would hide his actual preference for -ie. Some qualifying adjectives are a good example of his inconsistency and of the problem of using -ie endings as evidence. From his extant copy, the four Quartos, Compositor D changed any to anie four times in short lines and six times in long lines. He retained the any copy spelling seven times in short lines and eight times in long lines. With very he changed copy -y to -ie five times in short

lines and nine times in long lines, and retained three short-line copy -ie endings. He retained very in two short lines and eight long lines, and changed a copy -ie ending in one long line. With every his preference for -ie is not so strong: one change from -y to -ie in a short line and one in a long line; one copy -ie retained in a short line; four copy -y endings retained in short lines and five in long lines; and one copy -ie in a short line changed to -y. A chart reveals more clearly Compositor D's preferences, and the difficulty of generalizing from them:

The bottom line shows how mass counts may hide actual preferences. Nevertheless, occurrences of anie, verie, and everie will be used as evidence for Compositor D, since, as we will see later, Compositor F never uses the -ie ending for these words.

Along with the qualifying adjectives listed above, some other words reflect Compositor D's preference for -ie endings after a consonant despite his acceptance of many of the copy's -y endings. Included in this group of variants are: beautie, companie, giddie, heavie, heresie, Ladie, marrie, pittie, presentlie, prettie, and readie. After we examine quires T and V, more words will be added to this list, and indeed some of these new words occur as frequently as the adjectives listed above. The strength of Compositor D's preference for the -ie ending, however, is not consistent throughout the list. In the list below the variants can be used to distinguish Compositor D from Compositor F in the same way as the qualifying adjectives. Lady/Ladie is a particularly good example of how a word with a greater number of -y endings than -ie endings can still be an indicator for Compositor D. Despite his frequent use of the -y endings, Compositor D also uses the -ie for Lady, which Compositor F never uses. This list of words, properly employed, can be the best standard for separating Compositor D from Compositor F. When one of these words appears with an -ie ending, it is a clear sign of Compositor D's hand, since Compositor F never uses the -ie ending for these words. This criterion holds only for these fourteen words, since both compositors will set an -ie ending

for some other words. Compositor D sets many more -ie endings than Compositor F, but an individual instance of an -ie or -y ending (other than from the above list) can prove nothing.

A second spelling habit of Compositor D's that can be isolated by noting his changes from copy is eie for eye. Unfortunately, the analogous changes flie for flye, die for dye, and lie for lye are useless for distinguishing Compositor D from Compositor F. Although Compositor D generally prefers the -i- spellings, and Compositor F the -y- spellings, enough exceptions are present to make these variants useless as evidence.

As with the final -y/ie, Compositor D's preference is not absolute here either, but Compositor F's preference for eye is absolute.

With these two distinctive spelling habits of Compositor D's, we can go back and check to see if all the pages set from case z actually were the work of Compositor D. This check is essential since the above statistics cannot tell us if the -ie endings and eie spellings are clustered on only a few pages and missing on other groups of z-case pages, which would suggest that another compositor also used the z-case. In fact, both Hinman (in quires O and P) and Howard-Hill (in quire P) think some of the pages set with the z-case were done by Compositor F. The chart below shows on what pages the test words, with either an -ie or a -y ending, appear. An instance of an -ie ending of a test word is solid evidence of Compositor D; pages with only -y endings may have been set by either compositor.

Although this chart does not prove conclusively that case z was used only by Compositor D, the evidence is plentiful that he set at least the pages in quires K, N, and Q. L2 and O5v do not contain any decisive evidence, but the other pages in both of these quires clearly belong to Compositor D. Simple continuity is evidence against assigning these isolated pages to Compositor F. Portions of quire P show fewer signs of Compositor D's hand, although at least one -ie ending of a test word occurs on each page (except P1va). Later, additional evidence will support the assignment of Compositor D to all of these pages.

In the discussion of Compositor D's habits I have already mentioned several of Compositor F's traits that differentiate him from Compositor D. Before I examine the pages outside of case z that belong either to Compositor D or to Compositor F, two more habits of Compositor F's can be added to the identifying criteria. Neither of these traits appears in the z-case pages. Both Compositors D and F generally prefer a final -e when the word has a complex vowel.[15] Howard-Hill used indeed/indeede, for instance, to separate Compositors D and F from Compositor C, who frequently does not add the final -e after a complex vowel. With maid and meet, however, Compositor F prefers the shorter form, although he will use the long form of maide and meete on occasion. Compositor D, differing from his usual inconsistency in spelling, is absolutely consistent in using only the long form for both maide and meete. With sweet the situation is reversed: Compositor F uses only the short form, sweet, and Compositor D uses both forms. Compositor D, though preferring sweet on occasion, retains sweete so often that it can be used to distinguish him from Compositor F.

As evidence, the variant spellings of maid and meet must be applied separately from the variant spellings of sweet. The short form of either maid or meet appears only in Compositor F's pages, and the long form of sweete appears only in the work of Compositor D.

The second of Compositor F's habits that distinguish him from Compositor D is a minor punctuation habit—or lack of it. Compositor D never uses an apostrophe when a possessive proper noun is in italics, for example Lysanders (N5: 1098). Although Compositor F frequently fails to put in the apostrophe, he does use it often enough to aid in differentiating the two of them. This standard can also be applied to Compositor C, whose habit is similar to Compositor F's except that Compositor C does not always italicize the final s after an apostrophe (particularly in MV), as in Anthonio's (P4v: 1643), whereas Compositor F always italicizes the final s.

Most of the identification of Compositor F has been through a process of elimination. Compositor F is distinguished, for the most part, by the absence of Compositor D's traits on non-Compositor C pages. However, certain features, as the elision we'll or they'll or the two habits just described, are positive evidence. Also, Compositor F's preference for shorter speech prefixes, which Cairncross first noted, is a habit that can be important in the quire-by-quire examination, but this feature cannot be easily generalized, since it depends upon specific characters' names.

The first section of text to which we can apply the new standards for separating Compositors D and F are the pages in quires T and V (Shr and AWW) where an ad hoc case "x" was used. According to Hinman, cases x and z were "liquidated during the course of the work done for quire R" (II, 441). Hinman speculated that the Folio was going to become a single-case project, but for some reason in quires T and V a second case, made up de novo, appears and a second compositor shares the work with Compositor B, who continued to use case y (II, 443). Hinman thought the workman who shared these quires with Compositor B was Compositor A (F); Howard-Hill argues for Compositor D. In the chart below, those columns which offer strong evidence of Compositor D's work are italicized. (See chart on p. 99) We can add to this evidence in the chart two -le elisions, Weele (T5: 2239) and Youle (V3: 411), and at least one indented flow-over on each of the six pages. Because of these features plus no clear evidence for Compositor F, I support Howard-Hill's re-assignment of these pages to Compositor D.

From these pages we can add a few more words to the list of -ie spellings used to separate Compositor D from Compositor F. These words appeared too infrequently before to be of much value. When

they did occur in the z-case pages, Compositor D usually changed the copy -y ending to -ie (the first column below). As the figures in parentheses below show, these words appear more frequently in quires T and V, where the copy spellings are unknown.

For some of these words, as denie, dutie, and twentie, Compositor D shows a strong—potentially absolute—preference for the -ie ending.

Now that quires T and V have been added to Compositor D's canon, and more words have been added to the test list that identifies Compositor D, we can move on to an earlier group of quires linked together by a common case (x), quires A-E (Temp, TGV and Wiv). I have tacitly assumed these pages to be Compositor F's in my discussion of Compositor F's characteristics. Now we must test that assumption. With the exception of a few specific pages, all three scholars who

have worked with the Comedies' compositors have assigned these pages to F(A). I have again italicized the columns that would identify Compositor D if he were present.

The chart reveals almost no evidence for Compositor D. The single -ie ending on B3v (Companie, 2245) is obviously an anomaly. Although none of this evidence is overwhelming, enough differences exist in order to distinguish Compositor F from Compositor D, especially when the above evidence is coupled with the absence of any indented flow-overs and the presence of twenty-one -'ll contractions.[16]

The remaining section of the Comedies, quires F-H (MM and Err), which was set by three, possibly four, compositors is so confused in its type distribution that Hinman was unable to determine which case was used for each page. The main cause of this confusion was the completion of Wilson's Christian Dictionary, which permitted the Folio compositors to use a stock of type previously tied up; Hinman remarks it is "perfectly clear that the printers of the Folio found themselves in a position to make a fresh start, so to speak, with quire F" (II, 378). The fresh start included many new types and a second

skeleton, which suggests "two compositors simultaneously at work on the two pages of each successive form" (II, 379). But this new start upset the routine established in the earlier quires. Three or four compositors working with two cases and two skeletons tried to develop new practices, and the confusion in these three quires appears to be the result of trial and error arrangements. In order to identify who set which pages and columns, we must examine the quires closely, page by page, using criteria unique to these plays as well as the criteria previously developed.

Everyone who has attempted to determine the number of compositors—and which pages they set—in quires F and G (MM) has agreed that the frequent lack of solid and consistent evidence makes these quires the most vexing and challenging in the Comedies. Although scholars agree that determining each compositor's share in these quires is difficult, they differ significantly in their solutions, as the following chart shows:

The Compositors in Measure for Measure

                                                 

	Hinman	Howard-Hill	Cairncross	O'Connor
F1	A?	C/D	C	C
F1v	D	D/C	D	C
F2	D	C/D	F	C/D
F2v	C	C	F	D
F3	D	D	D	D
F3v	A	C	F	C
F4	A	C	D	C
F4v	A	C	D	C
F5	D	C/D	D	C/D
F5v	B	B	B	B
F6	B	B	B	B
F6v	D	D/C	C	C
G1	D	B/D	E?	B/C
G1v	D?	B	C	B
G2	A	D/C	C	F/C
G2v	B	B	E?	B
G3	B	B	E?	B
G3v	B	B	B/E?	B
G4	D?	C/F	C/?	C/F
G4v	D?	F/C	D?	F/C
G5	A	C	F	C
G5v	A	F	F	F
G6	A	F/C	C	F/C
G6v	A?	C	C	C

If Howard-Hill is correct in his assertion that many of the pages were split between compositors, as they experimented in search of an efficient method, the problem of proving who set what column becomes even more difficult.

As before, the first step in deciding which pages were set by Compositors D and F is to remove some pages—those set by Compositors B and C—from consideration. As would be expected, everyone agrees about most of Compositor B's stint: F5v-6 and G2v-3v. His role is less certain on G1ra and G1v, but I support Howard-Hill's assignment of these pages to Compositor B. Disagreement about G1v is not so great as may appear in the above chart, since Hinman, while assigning the page to Compositor D, felt G1v was "more suggestive of B than D, since both its 'doe-goe' spellings are in long lines (as its two 'do-go' spellings are not) and since Compositor D normally preferred 'here' to 'heere'" (I, 408). Hinman did not assign the page to Compositor B because he had not set either G1 or G2, but this reason is invalid when G1ra is assigned to Compositor B. Further evidence presented by Howard-Hill confirms Hinman's suspicion that Compositor B actually set G1v: three instances of Friar (Compositor B used it five other times in quire G) rather than Frier, which the other compositors preferred; and the lack of spaced final commas, despite a clear preference for spaced internal commas (p. 79).

As for G1ra, both Howard-Hill and Cairncross offer reasons for assigning it to Compositor B: four do/go spellings in short lines against one goe and one here's (Compositor B's habitual spelling of this contraction according to Howard-Hill) in long lines; a distinct preference for spaced internal commas, but no spaced final commas; and "spellings incompatible with the habits of the other compositors, including D" (Howard-Hill, p. 79). Previous scholars have not assigned the b-column of G1ra to Compositor B because of two instances of goe in a short line and doe in a long line. Another feature in column b that suggests a second workman is four instances of Frier; Compositor B always set Friar, including once in G1ra. The compositor who set column b preferred Duke. rather than Duk. as the speech heading (Duke. 8:3 against Duk. :1), and set the short catchword, Luc.. This evidence eliminates Compositor D, since he clearly preferred Duk. elsewhere and never set the shorter form of the catchword. Compositor F is also eliminated because of the speech prefix Duke., and eight instances of Duke in roman type in the text; Compositor F set Duke in italic all sixteen times it occurred in his pages of quire G. This same evidence supports the assignment of Compositor

C to this column. He preferred Duke. as a speech heading, and he seems indifferent to whether Duke in the text was in italic or roman. Although Compositor C usually set the longer form of the catchword, he does use the shorter form on occasion, as on D2, I4v, and K2. Since most of the column is in prose, spelling and spacing evidence are not entirely trustworthy, but they do not run counter to Compositor C's known preferences.

With Compositor B's pages settled (F5v-6, G1ra, G1v, G2v-3v), we can turn our attention to Compositor C. Unfortunately, Compositor C

does not inspire the same agreement among scholars as Compositor B. Howard-Hill assigns him the most columns, and offers the most convincing evidence to support his assignments. The following chart, adapted from his article, shows the criteria he used in quires F and G, and the columns that he assigns to Compositor C. After the sub-totals I have included those columns that I will later add to Compositor C's stint.[17]

As Howard-Hill shows, spaced final commas are often sufficient evidence to isolate Compositor C; more than two spaced final commas occur on F2ra, F4-5ra, G4ra, G5, and G6ra. And if pages F4 and F4v were set by Compositor C, then F3v was most likely set by him also. Hinman linked F3v-4v together through type recurrence (II, 381), and spelling and spacing features are similar on all three pages. The chart also reveals Compositor C's clear preference for doe/goe/heere on F1vb, F6vb, and G4vb. G2rb is assigned to Compositor C because of a combination of minor features: heere in a long line, Duke as the preferred speech heading, and the elision shee'll (Compositor F generally prefers a single e in she'll or he'll). Another occurrence of shee'll on G6v supports Howard-Hill's assignment of that short page to Compositor F.

The other two pages, F1ra and F2va, that Howard-Hill assigned to Compositor C need closer examination. On F1, Howard-Hill assigns Compositor C the first column and Compositor D the second, but he does not show why the page should be divided. Nor can I see a reason for the split. As I will suggest later when discussing Compositor D, the lack of significant contrasts in spelling and the similarity of the two columns in their spacing of commas makes me think the whole page was set by one man, Compositor C. With F2v it is a matter of subtracting columns from Compositor C's share rather than adding to it. The most striking feature on F2v, which everyone has noticed, is seventeen instances of Isa. (instead of Isab.) as the speech heading. Compositor B is the only workman who habitually uses this shorter form, but the frequency of doe/goe (5:1) precludes assigning him this page. Howard-Hill, assigning the page to Compositor C, asserted that "in this first page in which he encountered Isa. Compositor C followed copy" (p. 81). But F2v is not the first page where Compositor C would face a choice between Isa. and Isab. as speech heading. Since the innermost

forme of a quire was set first, Compositor C would already have set the longer form thirteen times on F4 and ten times on F4v; he does use the short form once in a long line on F4v. It is unlikely Compositor C would set the longer form on two pages, switch to the shorter form on F2v, and then return to the longer form on F5. Another contrast in spelling between F2v and Compositor C's preference elsewhere in the quire is praier/prayer. Compositor C sets prayer, his usual spelling, three times on F4v, but praier is set on F2v. Further support for re-assignment of F2v comes from Hinman's case evidence. Hinman managed to find occasional patterns of type recurrence in quire F, and he thought that F3 and F2v were set from the same case—a different case from the one used for F3v-4v (II, 381). Since Compositor D probably set F3 and Compositor C set F3v-4v, Compositor C would have had to switch cases along with his spelling of Isab. and prayer when he started F2v. Howard-Hill was led astray on this page by a crucial tabulation error in his spacing chart (p. 78). He lists twenty spaced medial commas and six non-spaced medial commas, but I find only thirteen spaced medial commas and thirty non-spaced medial commas, which would be strong evidence against Compositor C. Thus, F2v seems to be a non-Compositor C page, set by either Compositor D or Compositor F.

Now that we have finished a survey of Compositor B's and Compositor C's work in quires F and G, we can begin to distinguish between Compositors D and F in these two quires. The pages not assigned to Compositors B and C are: F1rb-1va, F2rb-3, F5rb, G2ra, G4rb, G4va, and G5v-6ra. As the following chart shows, the pages in quire G contain more distinctive features than the pages in quire F, and should be easier to assign. As previously, I have italicized those columns that would identify Compositor D if he were indeed present. The pages in quire G have a number of Compositor F's characteristic features: an apostrophe before the s in italicized proper nouns on G4va, G5v, and G6ra; maid on G2ra, G4rb, and G5v; she'll and we'll on G5v; Duke and Provost in italic on each page except G2ra; and no -ie endings in the twenty test words of the previous chapter. These features are particularly significant because they contrast completely with Compositor D's habits, thus suggesting that Compositor F alone is responsible for the seven columns in quire G.

In assigning Compositor F these columns in quire G, I disagree with Howard-Hill about only one column, G2ra, which he assigns to Compositor D. Compositor D, however, has been absent for five formes and does not reappear until quire H. Strong evidence would

be necessary in order to assign him this single column in the middle of quire G. In fact, the evidence suggests that Compositor F set this column: besides the short form of maid mentioned above, there is the contraction ith', which Compositor F sets frequently in other plays (Howard-Hill, p. 105), and which Compositor D never uses. Also, the -y endings of heavy (1806) and body (1786j) are counter to Compositor D's strong preference for an -ie ending for these words. Thus, I feel Compositor D is not involved in any of quire G, and Compositor F set G2ra, G4rb-4va, G5v, and G6ra.

In quire F, the question whether only Compositor D or only Compositor F, or both, set the 'left-over' pages is not so simply settled. The principal reason is the lack of distinctive features with which to work. The chart provides little evidence, and some of it potentially contradictory, from which to base an argument for or against Compositor D or Compositor F. Another part of the problem is that although we can describe Compositor F's preferences by referring to quire G, we are unsure of Compositor D's preferences on these features unique to MM. For example, Compositor F prefers Duk. rather than Duke. in the speech headings, but instances of Duk. in quire F cannot be

treated as evidence for Compositor F, since Compositor D may also prefer the shorter form. It would be tempting to argue that Compositor D does not begin work on the Folio until quire H—or even quire K and the appearance of case z—and that he replaces Compositor F, who worked through quire G. But this hypothesis is spoiled by certain key features on F3. The most important feature on this page is youle (589), which has led all scholars to assign this page to Compositor D. Although Compositor D is capable of setting they'll, as in Err, he never sets 'll after a vowel; plus, Compositor F never uses the -le form after any personal pronoun. Other features on this page also make it unlikely that Compositor F set it: two instances of do and two of go in short lines (Compositor F sets do/go only four times in short lines in all of his Folio pages), and Dukes (502) in roman type. Thus it appears that Compositor D set F3, and that his work in the Folio overlaps, rather than follows, Compositor F's. One problem with assigning Compositor D F3 is the notable absence of -ie endings, including such test words as deny and pitty.

Once we have established that F3 was set by Compositor D, we can begin to look on the other 'left-over' pages in quire F for features similar to those in F3. Earlier F3 was linked to F2v through case evidence; other features on this page strengthen this link and contrast neatly with features in the G-quire pages set by Compositor F. For example, Duke and Provost occur once each in roman type on F2v; Duke is in roman type on F3, and these settings in roman type contrast with Compositor F's clear preference for italic in quire G. A second contrast between F2v and Compositor F's practices in quire G is the spelling of Isa. (17 times) for the speech heading on F2v, which differs from Compositor F's choice of Isab. in quire G (Isab. 19 times, Isa. twice). Another feature that suggests that Compositor D set F2v is the increased number of -ie endings, notably unhappie (370), storie (380), libertie (414), and Abilitie's (430). A final, minor spelling variant on F2v, Praier (421), is frequently used by Compositor D, whereas Compositor F chooses prayer.

Sig. F5, the forme-mate of F2v, is split with Compositor C responsible for the first column. In column a, Compositor C spaces three final commas and clearly prefers to space internal commas (16 spaced, 6 not spaced); column b is just the opposite, with no spaced final commas and few spaced internal commas (4 spaced, 24 not spaced). Some slight spelling differences between the two columns support this division: die (994, 997, 1040) vs. dye (1091), and appear (1033) vs. appeare (1086j). In the second column two instances of do, a

number of -ie endings, and another instance of praier all suggest Compositor D set the second half of the page. However, some problems arise when Compositor D is assigned this column. First, some of the spellings are counter to strong preferences Compositor D exhibits later in the Folio; for example, suit (1077), dye (1091), body (1058), and beauty (1088). Second, Isab. is set nine times as a speech prefix in this column, but Compositor D used Isa. on F2v. One might expect some transitional fluctuation between Isa. and Isab., rather than this sudden change within a single forme. Despite these possible conflicts, I agree with Howard-Hill that this column belongs to Compositor D (the other compositors are even more unlikely).

Compositors C and D continued to split the work on the next page they set, F2. The evidence on F2 is similar to that on F5, only neater. Column a is assigned to Compositor C because of spacing evidence: three final spaced commas, twenty-one spaced internal commas and only one non-spaced internal comma. They'll in line 236 is supporting evidence. Column b was probably not set by Compositor C, since the spacing of commas is just the reverse: no final spaced commas, six spaced internal commas and twenty-five non-spaced internal commas. The spellings of do (2), here, and indeede, and the Duk. speech headings also conflict with Compositor C's preferences. The two do spellings are evidence against Compositor F, since he rarely set do in a short line (a total of three times in the Comedies). Thus, by a process of elimination Compositor D is assigned as Compositor C's partner. Unlike F5b, no features in F2b seem to conflict with Compositor D's usual practices.

Howard-Hill also splits F1v, F1, and F6v, but I do not find significant differences between columns on these pages, nor do I find evidence of Compositor D's involvement. Howard-Hill assigns column b of F1v to Compositor C for good reasons, but he is silent about why he assigns column a to Compositor D. Spacing and spelling evidence does not suggest that two compositors shared the page. The spacing evidence is slight because of the number of long lines in both columns, and the difference between columns in the spacing of commas in short lines is so small that an argument for splitting the page would have to be based on other evidence. Spelling also offers no significant contrasts. In fact, the occurrence of indeed in column a, even in a long line, is evidence against Compositors D and F, since both men strongly prefer indeede; both compositors set indeed only once each in all of their Folio pages. The one anomalous feature in column a is Duke (97) in italic. On previous pages in this quire Compositor C has set Duke in

italic once (F3v) and in roman twice (F2ra); Compositor D has set Duke in roman both times it occurred before (F3 and F2v); Compositor F consistently prefers to set Duke in italic in quire G. Although Compositor F may have set this column, I think the spellings of do, praies, and indeed make it unlikely. Moreover, this column would be his only work in the F quire. I find it considerably easier to assign the whole page to Compositor C, and argue that he is inconsistent in his setting of Duke and Provost in italic or roman.

As with F1v, Howard-Hill is silent about why he split F1, saying only that "there is not much useful evidence on F1 and F1v" (p. 81). But in fact, the spacing evidence on this page suggests that only Compositor C was involved, since one spaced final comma occurs in each column. Compositor D, the supposed partner, rarely sets a spaced final comma. Spelling evidence is slight, and the only possible variant that might suggest Compositor D was involved is here (66) in column b. But here is probably the copy spelling,[18] and Compositor C sets it at least ten times in short lines in MM. Finally, an indented flow-over is set in line 84; but as we saw earlier, Compositor C will use this device if there is no room for a turn-over or turn-under, which seems to be the problem here.

In short, I do not see any evidence that would justify splitting F1 and assigning column b to Compositor D.

Howard-Hill also split F1's forme-mate, F6v. Part of his reason for splitting F6v was the supposed division of F1: "F6vb is C's as also the other column may be, but the 'do' spellings there encourage me to maintain the division of quire F between C and D" (p. 82). But Compositor D seems not to have worked on the last two formes of quire F, and besides any argument from normalcy is of little value in this quire. The other reason for splitting the page, three instances of do in long lines, is not adequate evidence for distinguishing Compositor D from Compositor C. Compositor C sets do frequently on other pages in quire F—three times in short lines, six in long lines. Evidence against Compositor D is a number of non-Compositor D spellings in column b: ready (1391), company (1399), body (1404j), and beauty (1402j). I do not think that this page needs to be split; Compositor C set all of it.

In conclusion, Compositor D is involved in MM, but only briefly. He set F2rb, F2v, and F3. Compositor F worked in quire G—his last quire in the Folio—setting G2ra, G4rb-4va, G5va-6ra.

After quires F and G, the problems in quire H are minor. The task of determining the compositors is eased first, because Hinman was again able to trace some type recurrence, and second, because Howard-Hill was able to apply his spacing criterion effectively in a play that is 90% verse. Hinman matched case evidence to spelling variants and correctly assigned pages H4v-5 and H6-6v to Compositor B. These four pages were set from one case; the remaining eight pages were all set from a second case (II, 391). Hinman's choice of Compositor D as Compositor B's partner in this quire has been challenged by both Cairncross and Howard-Hill, who think Compositor C set most (all according to Cairncross) of the non-Compositor B pages.

Using spaced final commas as his primary evidence, Howard-Hill correctly assigns H1-1va, H2rb-3ra, and H4ra to Compositor C. The lack of spaced final commas and the contrasts in spelling between columns led him to assign the rest, H1vb-2ra, H3rb-3v, H4rb, and H5v, to a third compositor. As the following chart shows, the contrasts between some of the columns are so striking that only another compositor can account for them.

The two full pages Howard-Hill assigned Compositor D are discussed elsewhere: H3v has already been re-assigned to Compositor C (page 91), and H5v will be examined shortly.

Howard-Hill chose Compositor D rather than Compositor F, because of the frequent indented flow-overs on these pages.[19] The evidence I developed earlier supports his assignments. A look at the following chart shows that Compositor D—rather than Compositor F—had to be the third compositor in quire H. The italicized columns identify Compositor D.

Thus, the last quire that Compositor F helped set before quitting work on the Folio was quire G.

Although this chart eliminates Compositor F, it raises the major difficulty in assigning at least some of these columns to Compositor D. The difficulty is that the three elisions, occurring on H2ra, H3rb, and H5v, are counter to Compositor D's clear preference for -le. Everyone agrees that the -le elision is one of the most telling signs of Compositor D's hand; he used it on F3 and in later quires he changes Quarto copy 'll spellings to -le. Therefore, these anomalies should not be ignored or easily dismissed.

Before attempting to explain these anomalies, I wish to separate the two instances of they'll (H2ra and H3rb) from the one instance of

wee'll (H5v); I think that Compositor D set the they'll elisions, but that Compositor C set the wee'll elision. The argument that Compositor D prefers theyl rather than they'll is based on analogy. Scholars have assumed he prefers the -le form with all personal pronouns because he so clearly prefers youle, weele, heele, and sheele; but Compositor D does not actually contract "they will" anywhere else in the Folio. Unlike the other compositors, he might have distinguished between contracting a pronoun ending in a vowel and a pronoun ending in a consonant. When we look at the two columns where they'll occurs, it seems inescapable that D was the compositor involved. The evidence is overwhelming that a second workman shared H2 with Compositor C: five instances of mistris in column a versus six instances of mistresse in column b; eie and libertie (twice) in column a versus eye and liberty in column b; speech prefixes Luc. and Adr. in column a versus Luci. and Adri. in column b; heere in column b; and a centered (rather than indented) two-line stage direction in column a. Since this evidence eliminates Compositors B and F, and matches Compositor D's preferences, he must have set H2a, including they'll. Since Compositor D set they'll on H2 the occurrence of they'll on H3a, where the argument is the same, is anti-climactic.

The elision wee'll (1242) on H5vb is another matter, however, since Compositor D did set weele the other five times he contracted "we will." Yet almost all the other features on H5v suggest that Compositor D set the page. The eight instances of here and two of mistris, the lack of spaced final commas, and the recurrence of type from H3rb in column b of H5v are all evidence for Compositor D. However, another anomalous spelling occurs at the end of the page, near wee'll, that makes me question whether Compositor D set all of H5v. In line 1250 is the variant gon, which would be a unique spelling of gone in Compositor D's Folio pages. Compositor C, on the other hand, sets gon at least ten times in the Folio.[20] These two variants, wee'll and gon, that Compositor D never sets in the Folio but that Compositor C sets with some frequency, suggest that Compositor C took over from Compositor D somewhere in column b. The last line that is clearly Compositor D's is 1232, because of Mistris. Two lines later a short prose line might be the last line that Compositor D set in this column.

And (1235) could have fitted after the colon in line 1234, as an analogous situation in the same column shows:

Although this short line in a prose passage is not unique, it is unusual enough to be slight evidence that something irregular happened. If Compositor C took over at this point, the anomalous spellings wee'll and gon would be accounted for. Since the two men shared the forme-mate, F2, and split the next page to be set, F1v, Compositor C could have helped finish this page after being temporarily called away.

In short, I accept Howard-Hill's assignments in quire H with the exception of H3v and the last twenty lines of H5v, which I assign to Compositor C rather than Compositor D.

I began my examination of Compositor D by linking him to case z, which first appeared in quire K, and I saved for last quires F-H where the case evidence was very confused. In the process quire I (Ado), and one more page set by Compositor D, was skipped. Hinman thought "Compositor D evidently had nothing to do with quire I" (II, 396), and Howard-Hill and Cairncross agree. They assigned I1-3 and I5rb to Compositor B, and the rest to a second workman—Compositor A for Hinman, Compositor C for Howard-Hill and Cairncross. I agree that Compositor C set I4-5ra and I5v-6v, but I find so many features on I3v contrasting with those on the pages which follow that a third compositor must have set it. Howard-Hill lists, but seems to have overlooked, some of the evidence against Compositor C: indeede (211), here twice (145, 198) in long lines, don (198) in italic, and a strong preference for unspaced internal commas (12 to 1) in short lines. Another feature on I3v that conflicts with the pages set by Compositor C is the speech prefix Pedr., which is shortened four times (2:2) from the Quarto copy's Pedro. (Compositor C sets Pedro. twenty-nine times, and the shorter variant twice in long lines). Finally, the compositor of I3v preferred a number of -ie endings: worthie (214, 222), Ladie (115, 161, 182, 214), and beautie (186). In later pages in the same play, Compositor C chose the -y ending for each of these words: worthy (I6v), beauty (K3v), and Lady (I5ra, I5v-6v, K1, K2, and K3-4v). This evidence of a third compositor is also evidence for Compositor D. Each of the features listed above is characteristic of his work. Compositor D and Compositor C split the opening forme, I3v-I4,

of quire I; then Compositor B replaced Compositor D for the rest of the quire.

The final pages we must examine make up the first half of quire P (MV). These were included in the earlier survey of z-case pages, but they merit further consideration because the criteria used in that survey were inadequate for determining absolutely the compositor(s) of P1-3v and because of new case evidence. Although Compositor F's last appearance was in quire G, both Hinman and Howard-Hill feel he returned to the Folio to set part of quire P. Both men assign P2-3v to Compositor F, but acknowledge Compositor D as a possibility. Howard-Hill stated, "The spellings of MV are not so clear that it can be certain that every page has been correctly assigned to A[F] and D, and a closer analysis may find that all pages set from case z were set by compositor D" (p. 83). Compositor F has not been involved in the previous eight quires, and some very persuasive evidence would be essential in order to prove he returns to the Folio to set four pages in quire P and then to drop out for good. The feature in these pages on which Hinman and Howard-Hill support their assignments is the lack of do/go spellings, which Compositor D occasionally sets despite a preference for doe/goe. Against this rather weak evidence—especially considering doe/goe were the copy spellings—are spelling variants and case evidence that suggest Compositor D was the compositor of P2-3v.

As an earlier chart showed (p. 00), P3v contains the characteristic features of Compositor D, which are seen so clearly at the end of the O quire. On P3v the compositor changes eye to eie three times, and changes the -y copy endings of four of the test words to -ie (verie, 1379; happie, 1380; manie, 1429; and beautie, 1434, 1445). These distinctive spellings are supported by nine other -ie endings that have been changed from the -y ending of the Quarto copy.[21]

On the next page, P3, the evidence for Compositor D is almost as strong. The most telling piece of evidence for Compositor D is the elision youle (1354). Also on this page the compositor changes one -y ending of a test word to -ie in a long line, and retains one copy -ie ending of a test word in a short line; four other -y endings are changed to -ie.[22] Finally, the variants bloud (1251) and praier (1237), which occur on this page, run counter to Compositor F's preference for blood and prayer.

The evidence for Compositor D is not so plentiful on the remaining two pages, but the burden of proof has shifted against Compositor F. On P2v the compositor set sweete (1208) and heresie (1195), which were test words for Compositor D earlier. The general preference for -ie endings that characterized P3-3v is not so strong here, but three words besides heresie also end in -ie. On P2 go (1052) and Ladie (1001) are strong evidence for Compositor D. Although the evidence is slighter for Compositor D on these two pages, no feature on these four pages suggests that Compositor D did not set them.

As each page is assigned to Compositor D, the argument that only Compositor D used the z-case (which was used for P1ra, P1v-3v) becomes stronger. The z-case seemed to have been first used in quire N so that Compositor D could set type at the same time as two other workmen with two other cases. That situation arose again in quire P; Compositor B set P5vb with case y, while Compositor C set P5va with case x, and Compositor D set P2, the forme-mate, with case z. A similar arrangement occurred in the next two formes, P1v-6 and P1-6v. Hinman commented on the splitting of P6 and P6v, but does not mention the split of P1v nor has anyone previously assigned two compositors to P1v. The division of the page does not occur at the column break, but rather about a quarter of the way down column a, with Compositor D responsible for the final forty-seven lines in column a. Fortunately, a large number of distinctive types appear on this page. Seven distinctive types from case z, last seen in P3rb, are set in column a beginning at line 385. Earlier in the same column at lines 866 and 873 appear two distinctive types from case x, previously used in P4v. In column b only distinctive types from case x occur (nine times), most of them also coming from P4v.[23]

The division of the page according to case evidence is supported by significant contrasts in compositorial features. In column b, some of the non-Compositor D spellings which appear are: indeed (929), heere (934), pretty (938), choyse (974), and twenty (969), which was changed from the copy spelling twentie. Another non-Compositor D feature in column b is the apostrophe and the roman s in Bassanio's (949); the roman s is characteristic of only Compositor C. In column a, in the first line, is the contraction ith (862), which Compositor D never uses, but which Compositor C sets on occasion (Howard-Hill, 106). A few lines later two instances of mistris (876, 881) indicate

that Compositor D has taken over. The last piece of type from case x appeared in line 873; so the break must occur somewhere in these four lines.

The end of Shylock's speech (875) seems to be a good spot to pinpoint the change in compositors.

The reason Howard-Hill had assigned this whole page to Compositor D was five indented flow-overs. Four of these indented flow-overs occur in the z-case portion of column a and were set by Compositor D. The fifth occurs in column b (932), but an indented flow-over, by itself, is insufficient evidence of Compositor D. One final, interesting feature on this page is that the last line in column b is repeated at the top of F2. Although this repetition could be the fault of one compositor, it might be further evidence that the compositor of F1vb could be different from the compositor of F2.

Thus, a degree of normalcy is established in quire P. Compositors C and D set a page each in the first three formes, and then Compositor B joined in so that three compositors with three cases shared the work of setting each of the last three formes. Compositor D set P1-1va (after line 875), and P2-3v, and was the only person to use the z-case in the Folio.

Generally, I found these 'problem' quires to be more regular than Howard-Hill thought. Nevertheless, the compositors were clearly attempting to develop a method that could take advantage of the extra compositor and case that became available during the course of the printing of the Comedies. My new assignments diminish the role of Compositor F, and slightly increase the roles of Compositors C and D. In the Folio Compositor F set a total of 22 pages; Compositor D set 41 pages. Compositor F worked in quires A-E and G; Compositor D worked in quires F, H-L, N-Q, and T-V.

APPENDIX
PAGES SET BY COMPOSITORS D AND F

Compositor D

• F2rb (286-346)
• 2v (347-468)
• 3 (469-600)
• 5rb (1047-1112)
• H1vb (165-229)
• 2ra (230-288)
• 3rb (550-615)
• 4rb (809-874)
• [*] 5v (1128-1234)
• I3v (100-230)
• K2v (1392-1523)
• 5v (2168-2297)
• 6 (2298-2426)
• 6v (2427-2552)
• L2 (100-229)
• 4v (741-865)
• 5 (866-989)
• 5v (990-1113)
• 6 (1114-1241)
• 6v (1242-1369)
• N4 (728-850)
• 4v (851-981)
• 5 (982-1109)
• 6 (1238-1369)
• 6v (1370-1500)
• O5 (225-352)
• 5v (353-482)
• 6 (483-604)
• 6v (605-734)
• P1 (735-861)
• [*] 1va (875-923)
• 2 (983-1112)
• 2v (1113-1234)
• 3 (1235-1365)
• 3v (1366-1492)
• Q4v (358-482)
• 5 (483-608)
• 5vb (675-728)
• 6 (729-852)
• T5 (2176-2302)
• 5v (2303-2431)
• 6 (2432-2557)
• 6v (2558-2688)
• V3 (346-475)
• 3v (476-602)

Compositor F

• A1v (88-219)
• 5 (1007-1130)
• 5v (1131-1256)
• B1 (1513-1641)
• 1v (1642-1769)
• 3v (2160-2291)
• 4v (1-99)
• 5 (100-224)
• 5v (225-349)
• C4 (1359-1483)
• 4v (1484-1611)
• 5 (1612-1734)
• 5v (1735-1852)
• 6 (1853-1984)
• 6v (1985-2101)
• D3 (232-353)
• 3v (354-579)
• E3 (1765-1882)
• 5 (2257-2383)
• G2ra (1768-1827)
• 4rb (2330-2390)
• 4va (2391-2456)
• 5v (2654-2784)
• 6ra (2785-2850)

Notes