University of Virginia Library

Search this document 


expand section 
expand section 
expand section 
expand section 
collapse section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 

expand section 

This article is one in a series devoted to describing the printing of the Beaumont and Fletcher folio of 1647. Robert K. Turner, Jr., who initiated the project, laid the groundwork for the study in his description of section 2 of the folio published in Studies in Bibliography, XX (1967). There he wrote generally about the efforts of Humphrey Moseley to publish thirty-three plays and one masque previously unprinted belonging to the "Beaumont and Fletcher" canon, and specifically about the kinds of evidence used in his analysis. Section 4 ($A-I4) and part of section 8 ($8D-F4), printed in the shop of Ruth Raworth, contain four plays: The Maid in the Mill, The Prophetess, and Bonduca in section 4, and Four Plays, or Moral Representations, in One, in section 8.

The model for any such study as this is Charlton Hinman's monument to invention and perseverance, The Printing and Proof-Reading of the First Folio of Shakespeare (1963), to which both Turner and I refer for more elaborate descriptions of certain techniques used by us. Owing to some technical differences in the printing of the two folios, however, Professor Turner devised modifications in the display of evidence, chiefly in the graphs (Hinman's term) showing the sequence followed by the compositors in the distribution and setting of each quire. Rather than repeat his detailed explanation of the collection and analysis of evidence, I have thought it best in this digest to refer the reader to his article, though I include here a "reading" of Quire G as a vade mecum for that part of the folio printed by Mrs. Raworth, as well as some remarks on the special features to be found therein. Like Professor Turner, I have deposited a typescript of the entire study with University Microfilms, Inc., Ann Arbor, Michigan, whence


Page 166
it may be obtained for closer examination. We will follow this method of double publication for the remaining sections.[1]

Almost nothing is known about Ruth Raworth. Her husband John became free of the Company of Stationers on 6 February 1632 by patrimony; his father may have been that Robert Raworth who, with his partner John Monger, got into trouble for printing Venus and Adonis (SR, III, 703, 704). John printed, according to Morrison's Indexes to the STC and Wing, 84 items between 1638 and his death in July, 1645. His wife carried on the press in her name until 1648 when she married her husband's apprentice, Thomas Newcombe; in the interval of her widowhood her shop printed about twenty books, not counting her work on the Beaumont and Fletcher folio. This is obviously not a very big output; there is no reason, external or internal, to suppose that she had inherited more than one press from her husband. To look at the size of the Raworth shop from a different angle, only one boy, John Cottrell, had been apprenticed to John Raworth before 1640.[2]

Her biggest job in 1647 seems to have been the printing of Enrico Davila's The History of the Civil Wars in France (Wing D 413), a fat folio printed lavishly in large type with considerable white space. That folio plus a quarto (Fanshawe's translation of Il Pastor Fido), an octavo on the Creed, and a broadsheet comprise, with her share of the Beaumont and Fletcher folio, her known production for 1647. I mention these facts because of the bearing they might have on any discussion devoted to whether there was a lapse of time between the setting of the plays in section 4, or between the completion of section 4 and the setting of her part of section 8. I may as well state at once that, so far as I can tell, there were no appreciable delays; if there were, analysis of the other known 1647 books will not help since they are not printed in the type used for the Beaumont and Fletcher folio.


Page 167

The most singular anomaly, if only because it is so easily observed, in the printing of the folio is the use of lower-case letters at the beginning of verse lines in three of the Raworth plays, The Prophetess, Bonduca, and Four Plays in One. The first play in section 4, The Maid in the Mill, is treated conventionally. W. W. Greg judged in 1922 that the non-capitalization "betray[ed] the individuality" of a new printer called upon by Moseley to help speed section 4 to its completion,[3] but R. C. Bald recognized from the recurrence of the skeleton rules that one printer was responsible for all four plays.[4] And we may add that one of the compositors who capitalized in The Maid in the Mill did not capitalize in the succeeding plays. The answer must lie elsewhere.

The Prophetess begins on D1. Quire D exhibits a number of odd characteristics, including the use of one skeleton for three of the four formes, but it is like the preceding quires in that type used to set the earliest pages appears in the last pages, which means that each compositor — they were setting by formes — had only about three pages in type at any given moment up to the end of Quire D. But in Quire E, and thereafter in all but two columns each in Quires H, 8E and 8F, setting stayed one full quire ahead of distribution, so that instead of there being about five pages in type at one time, eight pages became the rule. In Turner's analysis of section 2, I have found no instance of distribution being that far behind composition. It seems proper to infer, then, that a decision to cease capitalizing verse lines was made to keep composition going as fast as possible instead of having to stop to replenish the supply of capitals, of which there were proportionately fewer than of the lower-case letters. That the supply of capitals was relatively low is difficult to prove, but two items of evidence point to it: the number of italic M's seems to have been as low as 25 at one time, and never higher than about 45; and the odd fact that in the instances of types appearing in the work of a compositor who apparently did not "own" them, the anomalous types tend to be capitals, leading to the possibility that the capitals were, if not held in common, at least frequently borrowed.

It has long been supposed that Ruth Raworth was slow in printing the plays assigned to her (Bald, pp. 35-36; Gerritsen, p. 254). Bald speaks of the "break in the printing between The Maid in the Mill and The Prophetess"; he refers to the non-capitalization, of course, but implies that there was a long delay associated with that change in


Page 168
procedure. Actually there is less evidence of delay between Quires C and D than there is in the printing of Quire D itself, where the triple use of one skeleton and the migration of a running-title from one skeleton to another mark obvious delays. The decision not to capitalize, a radical departure from convention, most likely came about as the result of some pressure to speed work up.

But we would expect to find two kinds of evidence if the delay had been very long. Had the skeletons been disassembled to get at the furniture and the rules, for example, the chances of their being reassembled for Quire D in precisely the same order that they were in in Quires A-C would be very small indeed. And we could not expect the type in Quire D to reflect the orderly distribution of Quire C if printing had ceased for a long enough time to necessitate the distribution of all of the quire.[5] Thus, although the wary bibliographer remembers that nihil alienum est was apparently the printers' motto, there is not sufficient bibliographical evidence to point to an extended break in the printing between The Maid in the Mill and The Prophetess, nor, by an extension of these same arguments, between Bonduca and Four Plays in One.

Returning to the fact that distribution was frequently a full quire behind composition, I want to point out a major implication involved in this distance or spread, namely that it often makes judgments about the order of the formes through the press very tentative. When they are present, progressive deformations of the skeleton-rules tell with some certainty which of two formes imposed in the same skeleton was printed first, for example that 2v:3 preceded 1v:4, and 2:3v preceded 1:4v. But the deformities say nothing about whether Skeleton I or Skeleton II of a given quire was the forme first printed.

The analyst starts with two assumptions as he approaches a quire. The first is that the skeletons will alternate, since imposition of alternate skeletons around type-pages is the most economical method of printing. The briefest experience, however, warns him that this first expectation is not one to count on. The second is that there will be an orderly relationship between distribution and setting, that the two will be reasonably close together in time, and specifically that type from the forme first printed will appear later in the same quire, giving him


Page 169
pretty certain information about the first- and last-set pages. Thus if deformations in the rules show that X2v:3 was printed before X1v:4, and if types are common to X2v and X1v or X1, he will properly assume that X2v was the first-set page of the quire. Other considerations might make him modify or even scrap the assumption, but he would start there.

Now if, on the other hand, he finds no type common to any page of Quire X, he loses that initial confirmation of predictability. If, further, he discovers that all eight pages of the preceding Quire W were still in type just before the setting of Quire X began, he must reckon with the possibility that the compositors had a rich reserve of six — perhaps all eight — type-pages available to choose from when it became necessary to distribute in order to set Quire X, i.e. W1:4v in the press, W2v:3, W2:3v and W1v:4 all wrought off and waiting undistributed when the compositors turned to set X2v:3. Under such conditions it is impossible to tell much about the relationship between composition and presswork: the press could be hot on the compositors' heels, or a quire behind them. In section 2, Professor Turner found no full quires which did not have types in common between two or more pages; and in his eighteen quires he shows only two occasions when, by the consecutive use of the same skeleton, there was a delay in printing. In sections 4 and 8 I have been forced to postulate such delays nine times in the printing of eleven quires (see Appendix A, Table I, Skeleton Formes and Center Rules), beside three or four shorter delays involving center or horizontal rules. I doubt very much whether all those delays in fact took place, but I have no way of proving the matter one way or the other.

A particularly aggravated and aggravating instance of delay is Quire H, where Skeleton II imposes three of the formes, the last two, according to the deformation of the rules, being 1:4v and 1v:4 (itself a problematic reversal of the expected order). Not only was there a delay between these two formes, there was also one between H1v:4 and I2v:3 which although it was imposed by Skeleton I, contains type from H4, i.e. H1:4v had to be printed and distributed before I2v:3 could be set. Actually this latter kind of delay is the more rational one since it might easily mark a moment when the end of a day coincided with the completion of a perfected sheet.

It is easy enough to see from the number of delays, even if several are more apparent than real owing to the distance between distribution and composition, that printing proceeded much more slowly than was theoretically necessary. How much time was in fact lost is not easy to calculate, however. The compositors' job included imposing formes;


Page 170
whether they alternated skeletons efficiently or sometimes used one skeleton for consecutive formes, their time spent on imposition would remain the same. The pressmen who had to make register and otherwise look to the often tedious business of make-ready were the ones who might chafe the most at having to wait. Cumulatively the time lost must have added up to a considerable amount, leading always to the possibility noted earlier that Mrs. Raworth fell far behind in the printing of the plays assigned to her. And yet it is difficult to equate the no doubt desultory technique of her shop with any crucial delay involving, as Gerritsen argues, the reassignment of three of her six plays to another printer.

Graphs and Supporting lists. Though it seems most economical to refer the reader to Robert Turner's analysis of section 2 for a full explanation of the kinds of data used by us both, I here give a reading of the graph for Quire G.

The kinds of evidence used in our study divide into three categories, presswork, composition and distribution, and the identity of compositors. Under the first are gathered (I use Turner's words):

  • 1. Evidence from the components of the skeleton forme, such as running-titles and rules, and from center rules, taking into account the testimony of priority given by disfigurations, dislocations, and other peculiarities;
  • 2. Evidence from type matter reappearing within the quire;
  • 3. Evidence from embossing.
Under the second come:
  • 1. Evidence from reappearing types;
  • 2. Evidence from reappearing rules, display types, heads, and other typographical matter not a part of the skeleton forme;
  • 3. Evidence from type shortages.
The identity of the compositors, established on the basis of their spelling preferences, serves chiefly to corroborate and refine assignment of their stints.

One always starts with the evidence relating to the presswork before constructing the graph. Skeleton I, which had imposed the last-printed forme of Quire F, imposes G2:3v and G1v4; Skeleton II imposes G2v:3 and G1:4v. There are no deformities in the rules of either skeleton to help establish the order in which the formes were printed. (If, for example, Rule 10 had a bend in it in F4v and G3v, but not in G1v, the assumption would be that G3v had been printed before G1v.) Horizontal Rule T is common to G1a and G4b in Skeletons I and II. There is no type common to any two pages of the quire.


Page 171

Now the difficulty here is that the use of the skeletons works against the grain of economical alternation. Ideally Skeleton II would impose both inner formes (2v:3, 1v:4) and Skeleton I would impose the outer formes. As it stands in Quire G, there will either have to be a delay between the printing of G2:3v while Skeleton I is transferred to the waiting forme, or else the graph will have to demonstrate that G1:4v was set before G1v:4, or — and this is a possibility which comes as a result of the fact mentioned earlier, that composition was a full quire ahead of distribution — there were two formes (G1v:4, G1:4v) waiting when G2v:3 (Skeleton II) was wrought off, and that skeleton was arbitrarily imposed around G1:4v. Unless there is evidence to the contrary, it is best to assume that printing occurred in the same order as composition. The contrary evidence here is the delay caused by the transference of skeletons, but such delays occur with considerable regularity in sections 4 and 8, so that I have somewhat cautiously shown the probable order of printing as:

Skeleton:  II  II 
Forme:  2v:3  2:3v   1v:4  1:4v  
Note that Rule T, common to G1a and G4b, also represents a delay.

Turning to the graph, we find in the vertical column the pages undistributed when the setting of Quire G began. Line 1 shows that E1a, still undistributed after the composition of Quire F, was distributed into Compositor A's case. Five pieces of type from E1a appeared in G2va and another in G2vb. Line 2 shows that E1b was distributed after the setting of G2va was completed and before or during the setting of G2vb. Then Compositor A apparently distributed two full pages, F3-F3v (lines 3-6) before setting G2, and so on until he completed his stint at the end of G1b. F4b was the last column distributed because type from it appears nowhere in Compositor A's work save in G1. The single type (line 6) from F3vb may seem too frail to support any evidential weight at all; but F3v has only 24 lines of text in each column, and since F3va produced four types found in G2a, it seems best to assume that both columns were distributed at the same time. Compositor B set his four pages in seriatim order to match the work of Compositor A who set in reverse seriatim order; they thus produced formes by their synchronized labors.

"The order in which the columns of Quire G are listed at the head of the graph is the order of composition from each case as indicated by the order of printing and the order of distribution. It is to the second of these matters that the numerals entered in each line to the right of the distributed-column designation have relevance. A '1' indicates


Page 172


Page 173
that the particular column was a part of the first forme of its quire to be machined, a '2' that it was part of the second forme, and so on" (Turner, pp. 49-50).

Lines 19-27 represent Turner's category of latent type. These are types which come from formes distributed for the setting of pages of the preceding quire (here, Quire F), but which appear in Quire G either because they were not needed, or because they filtered down deeply into the sort, or because they were in fact used in Quire F but were unobserved there. Their value as evidence comes in the degree to which they remain in their "owners'" cases. A large quantity of latent type from Case A (that is, the case used by Compositor A) appearing in the work of Compositor B will indicate that something worthy of analysis has taken place. Almost always, however, latent type corroborates "ownership" as it does in this graph except for line 21, where E2a types, last noted in the possession of Compositor A, appear in Compositor B's work. The incidence is not high — two pieces — but a check of the Quire F graph shows a single E2a type in Compositor B's F2vb. Now both this piece and the one in G4a are capitals, and as I remarked earlier, it seems likely that capitals were often borrowed, or perhaps held in common. Still one wants to pursue the possibility that Compositor A might have set a small portion of F4a. There is no doubt that the bulk of the type in G3-4v came from Case B; furthermore, lines 22-27 show latent Case B type in the same pages. What evidence does the tabulation of spelling preferences, the last segment of the graph, provide? The preferences of both compositors cluster comfortingly in the pages assigned to them by the type evidence, except for Compositor A's "devil" in G4a, and since devil-divel are strongly marked preferences of the two compositors, the investigator has further ambiguous evidence to deal with. Turning to tables not included in this digest, he learns that the anomalous G4a types are found in lines 50 and 52, suspiciously close together. Other G4a types (lines 13 and 14 of the graph) appear between lines 7 and 38 of column a, not mingled with the anomalous types. "Devil," however, is in line 26, not near the anomalous types, and since Compositor B has elsewhere shown himself susceptible to copy-spellings, we may properly decide here that none of G4a was set by Compositor A.

After the first fine blush of optimism in the early 1950's regarding the certainty with which compositors could be identified and so be assigned to definite sections of a text, experience modified the expectation of reliability which might be attached to compositor determination. Now, however, taken in conjunction with evidence provided by the patterned repetition of type, the evidence of spelling tests has


Page 174
gained a new security and function as a corroborator of the patterns, and even as a method for refining inferences drawn from the type evidence.

As we might expect from a shop so apparently small as Mrs. Raworth's, few compositors worked on the folio. It is relatively easy to distinguish two, Compositors A and B, who set the bulk of the plays; it is much less easy, and finally perhaps wrong, to identify a third, Compositor C, who set part of The Maid in the Mill only.[6] The spelling preferences may be set out in tabular form, followed by an explanation.

Compositor A   Compositor B   Compositor C  
Romane (tolerates Roman)  Roman 
devil  divel  devil 
Emperour  Emperor (tolerates Emperour) 
-inde (tolerates -ind)  -ind (tolerates -inde) 
2 noun  2. noun (tolerates 2 noun)  2 noun 
long pronoun odd S.D.  (tolerates long pro.) 
Of the three, Compositor A's preferences are the most stable. After he began work, in Quire D, he tolerated Compositor B's preferences many fewer times than Compositor B tolerated his. By "tolerate" I mean the use of the other's preference at least twice in the course of the man's whole share of the folio; Compositor A for example prefers -inde (as in winde, finde, binde), but tolerates -ind a total of eight times in five of the nine half-quires he set. "2 noun" means that in speech-prefixes such as "1 Lord" or "2 Soldier" he does not use a period after the numeral; "2. noun", Compositor B's preference, indicates the use of the period. This preference seems to be significant only in speech-prefixes and not significant in such uses as the designation of songs ("3 Song") or stage directions in C1v and C2.

Compositor B's preferences are less stable, though they are quite clearly marked. Because he tolerates most of Compositor A's spellings, the impression grows that he was more easily influenced by the spelling of his copy than was Compositor A. Here is where the category of "odd S.D." makes itself felt. The term refers to the use of roman proper names in italic stage-directions, e.g. "Enter Julio and Franio" (B1b), whereas the more frequent form puts all words in italic type. It is hard to decide whether Compositor B's use of this convention reflects a preference or the appearance of his copy. The occurrences of it cluster in The Maid in the Mill (Quires A-C), diminish in Quires D and


Page 175
E, and then die out altogether except for one instance in 8E4. They usually appear on the same page with other all-italic directions. But most significantly, the odd S.D.'s are always associated with those parts of The Maid in the Mill written by Rowley, frequently associated with the Massinger parts of The Prophetess, though mixed with Fletcher's parts (three Fletcher to five Massinger), appear once in 8E4 in a Fletcher part of Four Plays in One,[7] but never in the play wholly by Fletcher, Bonduca. What I suspect happened is that The Maid in the Mill was a composite manuscript written in the hands of Fletcher and Rowley, that Rowley used the roman-name variation and was followed in it by Compositor B who let it affect his practice sporadically thereafter for two quires. This is the more likely if, as I believe, Compositor C set the odd S.D. in C1v, for that would mean that two compositors adopted the same convention.

Unhappily, the existence of Compositor C is not certain. He resembles Compositor A so thoroughly that I did not distinguish between them until quite late in the study when a refinement of the earliest spelling tests showed Compositor B's willingness to use long forms of pronouns (yee, shee, hee, wee). Compositor A never used the long pronoun form in those parts of the quires already confidently assigned to him on the basis of type ownership, that is from Quire D onward, while in Quire B a mild toleration for -ee appeared in conjunction with two of his preferences, devil and 2 noun. Since the type used to set Mrs. Raworth's plays did not begin to be divided into two cases until the distribution of Quire B (and was not completed until the distribution of Quire D), it was not clear to me how many compositors had set Quires A-C. I had been put off by the great rarity of -inde words (only four in those three quires) which later on became strong identifiers of Compositor A. Eventually it seemed best to suppose that a third man had helped to set The Maid in the Mill, a compositor who very much resembled Compositor A except for the absence of the -inde identifier.

Type substitutions occur whenever there is an especially heavy run on a sort, as often happens as the result of repeated speech-prefixes in plays where the italic capitals cannot meet the pressure put on them. As the compositor nears the end of his supply of a letter, but usually before the sort is exhausted, he begins to substitute roman for italic letters so that Caratach (to take an example from Bonduca in Quire G) becomes Caratach, the speech-prefix Car. becomes Car., and so on. A rather simple computation often allows the analyst to show that the


Page 176
substitution coincides with the pattern of distribution and composition already established, or, in cases of doubt, to bolster one possible solution against another. For example, in Quire G Compositor A set 52 P's in G2v-1v before he began substituting P's on G1, the last page of his stint. Since Quire F used no P's and hence could not act as a source of replenishment as formes from it were distributed, the calculation is easy. More complicated is what happened in Compositor B's stint. He substituted 11 I's in G4vb after setting 65 J's in G3-4va and part of G4vb (Iudas for Judas), which again corroborates the order of setting shown in the graph. But he substituted D's and C's in G4 and then got a supply of italic letters for G4v from some source other than the distribution of F1v which used none. This may be shown in the following table, where the bracketed pages are those which were distributed before the setting of the Quire G pages, and the bracketed numbers indicate the C's released for use in those pages — in this example, none.
The number of C's in use when substitution began is very small, 23 at most; indeed the total used in G3-4v is only 35, which hardly compares with the 52 P's or 65 J's noted above, but I do not know how to account for the inequities. They come up with sufficient frequency to remind us that we are not dealing in an exact science.

Finally, the graph may be summarized thus:


The charts set out certain information in tabular form. Table I shows in the vertical columns the order of printing elaborated to include the relationships among the rules and running-titles, which are given arbitrary arabic and roman numerals respectively. Other tables not included here describe each rule and running-title fully. The Summary of Printing shows which compositor set each page, or column when the page was divided, of the plays set in Mrs. Raworth's shop. In the Summary, a (b) indicates a blank page.


Page 177



Page 178