University of Virginia Library

Reappearing Types as Bibliographical Evidence [*]
by
Robert K. Turner, Jr.

To illustrate the principle that one must have a rather good idea of what he is looking for before he can see it, Darwin told this anecdote: While he was still a student, he and a companion carefully explored a Welsh valley for fossil remains. Intent on their search, they did not notice that all about them were scored rocks, perched boulders, and moraines, all such plain indications of glaciation that had the valley been filled by a glacier at the time the signs could not have been more conspicuous. But Agassiz had not as yet announced his theory of a glacial period in the earth's history, and such matters were far from the minds of the two fossil-hunters. They no doubt found some fossils, but they passed by evidence for one of the great scientific discoveries of the nineteenth century because they were intent on other things.

Perhaps Darwin's story, with its implications that hypothesis is essential to observation, is known to bibliographers as well as to scientists. If not, it ought to be, for boulders and books are alike in that both exhibit phenomena that can be seen without being observed, that are meaningless, or virtually so, until we have a context in which they are meaningful. An instance is the reappearance within early printed books of recognizable rules, ornaments, and individual pieces of type. This phenomenon must have been seen quite often by bibliographers; McKerrow, for example, speaks of the reappearance of certain initials within the gatherings of a Bible printed by Barker in 1591 as proving the book to have been composed by formes.[1] But McKerrow thought that such reappearances could occur only in page-for-page reprints, as he evidently did not realize to what extent manuscript as well as printed matter could be cast off for setting by formes. From McKerrow's day until recently we have seen the use of evidence from recurring rules, ornaments, and types, but chiefly in connection with such matters as studies of skeleton formes, the identification of printers, or the determination of resetting. It was not, I think, until we learned something of the methods Professor Hinman had adopted for the examination of the Folger Folios and the results he was obtaining that bibliographers began to search for the reappearances within single books of type matter not a part of the skeleton formes and to use such reappearances as evidence

in rigorous bibliographical analysis. Although not as prominent as Darwin's moraines, this evidence can be obvious enough, and one of the many contributions of The Printing and Proof-Reading of the First Folio of Shakespeare is that it demonstrates both its meaning and its value so decisively.

When the influence of this book is fully felt, we will probably see a much more extensive use of evidence from recurring type matter, for from it, in conjunction with more conventional evidence, one can precisely reconstruct, as Hinman has shown, many details of printing-house procedure that affected the transmission of texts and that on other evidence would remain obscure. Some of these details, ones of great concern to Hinman, are the identity of compositors, the extent of their work, and the order of composition and printing of various parts of books. To supplement what we now know about this new kind of evidence from the study of the Shakespeare Folio, I wish to discuss three instances of its usefulness in other and somewhat different situations.

Let me first describe briefly the first quarto of Albumazar, an academic play by Thomas Tomkis, which was printed in 1615 by Nicholas Okes. The book collates A2 B-L4; the text begins on B1 and ends on L3v, a short epilogue being printed on L4 and L4v being blank. One gets the impression that Okes strove in this book for an appearance of higher quality than that of most 17th-century play quartos: an almost prodigal amount of white space was left around the scene heads, of which there are many; and nearly every scene begins with a two- or three-line initial, three of the acts with four-line initials, and the first act with a handsome decorative block. Speech prefixes were set in roman caps and small caps flush with the left margin, except on B1 where a couple were indented. Stage directions were generally set in the caps and small caps of a roman font slightly larger than that employed for the speech prefixes. On each full page there are thirty-seven lines of type or the equivalent, three leaves in each gathering are signed, and what small variation there is in speech-prefix abbreviation appears to be random. Throughout there is substitution of roman for certain italic letters that were in short supply. The book as a whole has a typographical consistency which suggests that it was set up by one compositor.

When spellings are examined, this impression, although perhaps not confirmed, is not altogether destroyed. Chart I shows the incidence of six variant spelling forms, those which appear to be significant. (For the moment please disregard the lines of the chart labelled "A types" and "B types.") The A and B spellings do tend to cluster in various parts of the book, but there are conflicts which undercut an assumption that two compositors were at work — the occurrence on B3 of the B-form wee'le amidst A-forms, the occurrence on C3v of the A-form wee'l and the B-form maister, the occurrence on H2 of both A-forms and B-forms, and so on.

Furthermore, the whole pattern is rather thin: there are a number of pages on which no significant spellings are found. I had some doubts about the matter, but in a study of this play made several years ago I spoke of the book as set up by one compositor, having decided, unwisely as it turns out, that the evidence was not strong enough to support a declaration for two.[2]

Please turn now to Charts IIa and IIb, which show the incidence throughout Albumazar Q1 of certain recognizable types. The numbers in the body of the chart are line numbers; thus one may see in Chart IIb that the "e" which is found in line 2 of B2v reappears in line 27 of D2, line 28 of E4, and so on. None of the types listed falls on any page not represented in the charts. Because the identifications were made from photostats, I should not want to trust them implicitly, but I think they are accurate enough for our present purposes. The point to be noticed is that the A-types and the B-types very definitely cluster in certain runs of pages: types found from D3 through E2v recur from F3 through G2v, from H3 through I2v, and from K3 through L2v; similarly, types found from C3 through D2v recur from E3 through F2v and so on. And when this information is taken together with the spelling evidence, that begins to make better sense. Looking again at Chart I (especially now the lines labelled "A types" and "B types"), we see that if we disregard Compositor A's use of wee'l (this spelling having been used by B in addition to wee'le) and Compositor A's use of sweet (this spelling too having been used by B in addition to sweete) we get perfect correlation except at B3. Thus we are able to assign the various parts of the book to their compositors with much greater confidence than we could on spelling evidence alone.

If, as with Albumazar Q1, we use types to differentiate between the work of two or more compositors, we should be aware of our assumption that the types set by each workman were distributed only into his type cases. I think in the instance of Albumazar Q1, because of the correlation of the typographical and spelling evidence, this assumption is in general justified. But it is not true of all the types in the book: the italic act head on B1, A's page, reappears with numerals changed on C4v, E4v, G4, and K1v, all B's pages; and I think the apparent conflict of spelling and typographical evidence on B3 can be explained on the supposition that all of Sheet B was set by Compositor A but that B2v and B3 were distributed into B's cases — that is, that the B types were not actually in use by Compositor B until he began work at C3.

In one book with which I am familiar the types were not regularly returned to the cases from which they originated, but even in it typographical evidence can be of use in compositor determination. Let us consider three quires in the second section of the Beaumont and Fletcher Folio of

1647, a folio in fours. The second section was printed by William Wilson,[3] and the quires in question — 2K, 2L, and 2M — contain the final page of The Captain, on which the prologue and epilogue are printed in italics (2K1), a blank (2K1v), and the complete text of Beggars' Bush (beginning on 2K2 and ending on 2M4v, a page containing only twenty-seven lines in each of the two columns of text). The pages which these quires comprise contain variant spellings suggesting that they were set by two compositors, but the assignment of some segments of the text to their proper compositor is difficult because of the paucity of significant spellings. However, by tracing type recurrences according to Hinman's method, it is possible to link parts of the text which lack significant spellings with those which do contain them and thus to make assignments on the assumption that two segments of text set from the same case would have been set by the same workman. Although this assumption cannot always be made, as one compositor could relieve another during the course of the work, it is probably safe if, after careful examination, one can detect no important discrepancies in spelling or details of typography between the two segments under scrutiny.

Charts IIIa and IIIb summarize the evidence relating to the three quires. The information given here is based on a preliminary survey and is probably incomplete; moreover, for the sake of clarity, I have suppressed not only the number of types being represented but also some conflicts in the evidence which I think are minor or capable of an explanation too involved to be presented here. I am concerned only to illustrate a principle, and the information given is, I believe, accurate enough for that purpose. On the charts themselves the horizontal headings represent parts of the text, usually columns, such as 2K3a and 2K3b, but entire pages in the case of 2K1 and 2M4v, which contain little text. The order in which these headings appear, incidentally, is the probable order of composition as suggested by evidence relating to the order of the formes through the press. Among the spellings indicated, "wee'l, etc." and "wee'll, etc." include other related variants such as shee'l/shee'll or shee'le and hee'l/hee'll or hee'le; "2 noun" and "2. noun" refer to any numeral in a speech prefix or stage direction followed by any noun (e.g., 3. Merchant), the presence of the point being the crucial matter; and "Florez/Floriz" and "Hemskirk(e) /Hemskirck" refer to characters' names. The notations for "Floriz" include a couple of occurrences of Floris and those for" Hemskirck," of Hemskirick. The rubric "odd s.d." among Compositor B's characteristics is used to designate stage directions whose internal elements are separated by periods rather than the more usual commas. Because both compositors spelled heart, hart only is included among B's characteristics. In the parts of the charts labelled "Type from" we find the source of the recognizable types which reappear in the designated parts of each quire. Thus in Chart IIIa one or more types from 2I2va are found in 2K3a, 2K3vb, and 2K4a. In

this section of the book the column seems always to have been the unit of distribution (omitting from account a few aberrant types which are not represented). This point is important, for if parts of columns had been distributed into A's cases and parts into B's, a summary of the kind we are now examining would convey quite the wrong impression.

In contrast with the practice of the compositors of Albumazar Q1, Wilson's workmen did not always maintain control of the same types. We see that although Compositor A set 2L2va, this column was distributed into B's case and that although Compositor B set 2L4a, this column was distributed into A's case. This feature of the compositors' work, illustrated in small here, is more prominent in other quires of this section of the Folio. It restricts the validity of the typographical evidence for compositor determination, but, as we see, does not render it useless, for one can still reason as follows: that as 2I2va types reappear in 2K3a, 2K3vb, and 2K4a, these columns were set from the same case, the one into which 2I2va was distributed; that as his spelling characteristics occur in 2K3a and 2K3vb, Compositor A was using that case; hence, that 2K4a, in which spelling evidence is lacking, was also set by Compositor A, evidence to the contrary being absent. We see how the columns in each compositor's stint are linked to one another by the reappearing types and how these correlate with the spelling evidence in a way which permits assignment of columns in which the spelling evidence is weak or non-existent.

Thus we understand that the types used in the texts of early books can sometimes repay close scrutiny because, among other things, they may help in compositor determination, but the types of the text are not all we have to work with. Some books, like the Shakespeare and Beaumont and Fletcher Folios, have on each page such typographical appurtenances as box and center rules which, as Schroeder and Hinman have shown,[4] can be put to good bibliographical use, and even lowly quartos ordinarily have running titles, which, as Professor Bowers has often reminded us, can be tracked through the book and made to tell some of the secrets of the printing. But even running titles, with which many of us are used to working, cannot be taken for granted, as the following example is meant to show.

Beaumont and Fletcher's Maid's Tragedy first appeared in a quarto of 1619, which collates A2 B-L4. The text, as Greg noted, was printed in two sections, B-G and H-L, the first, and the one with which we are now concerned, being the work of Nicholas Okes.[5] The running titles in this section (and indeed the second also) were set in rather large italic type and read The Maydes Tragedy., on both recto and verso. A glance reveals that the sheets were machined in two skeleton formes which alternately imposed the inner and outer formes of succeeding sheets, and the same glance suggests that there is no very great problem about the identification

of the running titles. Chart IV shows the eight titles included in the two skeletons as they appear in Sheets C and D.[6] In RT I on C1 and D2 we see an initial T with a broken right arm, d's that bend back at the same angle, and second T's that look much alike. Of course, there are some differences too: the arms of the initial T's do not form precisely the same angles with the stems and the serif on the left foot of the M has picked up a nick, but nothing is consistent in early printing and the small imperfection on the shoulder that takes ink over the period seems to make the identification conclusive. In the two appearances of RT II we may note the similarity of the M's the y's, and the d's but may dismiss the dissimilarities of the T's as accidental; in the two appearances of RT III we may note the similarity of the y's and the g's, and remark in passing that the initial T looks as though it was reset. And so for the rest of the running titles, not only between Sheets C and D but throughout Okes's section of the book: we may notice some points of likeness between titles, on the strength of these identify them as the same, and conclude that the transfer of the skeletons from forme to forme was quite regular.

Of course, as you have doubtless noticed, that is not exactly so. There seem to have been, it is true, lower-case types in sufficient numbers to set

the lower-case letters of all eight running titles, but, if my identifications are right, there were only five M's and ten T's available. Eight complete titles required eight M's and sixteen T's. The compositor was therefore obliged to shift some capitals from the running titles of each forme which was returned from the press to the next subsequent forme which was being imposed. In Chart IV partial results of this can be seen, for, to follow only one group of capitals for a short way, the capitals in RT VI at C2 (slightly trimmed in the copy used) move from there to identical positions in RT I at C1; from there the initial T and the M move to identical positions in RT VII at D4v and the second T becomes the second T in RT V at D2v. The progression of the various capitals through Okes's section of the book can be seen in Chart V, where I have identified each set of lower case types, each M, and each T by arbitrary numbers. A typical instance of the compositor's way of coping with the shortage may be seen if we look at the columns headed "C(i)" and "C(o)". Evidently before C(i) was returned from the press, the workman set into position for C(o) the four groups of lower case letters which previously had been used in B(i), LC6, LC5, LC8, and LC7, and he also used the three capitals which were then free, M1, T8, and T7. All the rest of the capitals had to be obtained from C(i); T9, for example, was taken from C1v to finish the C4v running title, three capitals were transferred from C2 to C1, and so on. Generally, this procedure was followed throughout the section.

What I am chiefly interested in showing here is that the types can be traced with some precision if one looks for them and that the information thus gathered is more than a mere curiosity, for it furnishes evidence of the order of the imposition of the formes. On three occasions capitals were moved from one page into precisely the same relative position in another page. The first two of these occasions, the transfers between C2 and C1 and D3 and D1v tell us little because either the inner or the outer forms of either sheet could have preceded the other forme of that sheet, but the reappearance of the E3 capitals in identical positions in F4v almost certainly means the precedence of F(o) over F(i) and this taken together with the alternation of the lower case groups indicates the order of formes shown on Chart V.

These three examples show how evidence from types may facilitate compositor determination and reveal other matters of more than cursory interest. Such advantages can often justify the patience and eyestrain necessary to identify and trace the types, but, in addition, on their testimony, inference about the relationship between composition and presswork, a matter not considered here, frequently becomes possible. Some precautions which should be observed have been mentioned, and we shall doubtless learn of others as more experience in the use of this kind of evidence is gained. It seems clear to me, however, that bibliographers and textual critics who want to know how their books were printed will have to search out broken types as one of the quite routine procedures of bibliographical investigation.

CHART V
The Maid's Tragedy (1619)
(Okes's Section)
Running Titles

                                 

B(o)

B(i)

C(i)

C(o)

D(o)

D(i)

E(i)

E(o)

F(o)

F(i)

G(i)

G(o)

1

HT

1v

LC5

1v

LC2

1

LC6

1

LC4

1v

LC5

1v

LC2

1

LC7

1

LC1

1v

LC6

1v

LC3

1

LC7

M4

M4

M2

M1

M5

M4

M2

M3

M4

M5

M1

T6

T5

T4

T1

T9

T3

T1

T8

T8

T3

T3

T9

T9

T6

T2

T3

T7

T3

T4

T10

T8

T7

2v

LC2

2

LC6

2

LC4

2v

LC5

2v

LC2

2

LC6

2

LC4

2v

LC6

2v

LC3

2

LC7

2

LC1

2v

LC6

M2

M2

M2

M5

M4

M1

M5

M4

M2

M1

M3

M4

T5

T2

T4

T10

T10

T2

T10

T4

T5

T3

T4

T8

T3

T10

T6

T2

T6

T7

T6

T8

T9

T7

T10

T10

3

LC3

3v

LC7

3v

LC3

3

LC8

3

LC1

3v

LC7

3v

LC3

3

LC8

3

LC4

3v

LC5

3v

LC2

3

LC8

M5

M1

M5

M3

M5

M2

M1

M1

M5

M5

M4

M2

T1

T3

T2

T3

T9

T5

T2

T2

T1

T6

T6

T2

T4

T8

T10

T7

T3

T4

T9

T10

T3

T4

T2

T9

4v

LC4

4

LC8

4

LC1

4v

LC7

4v

LC3

4

LC8

4

LC1

4v

LC5

4v

LC2

4

LC8

4

LC4

4v

LC5

M3

M3

M3

M1

M2

M3

M3

M5

M1

M2

M2

M5

T9

T4

T1

T9

T4

T10

T4

T6

T2

T2

T5

T6

T2

T7

T3

T8

T8

T6

T8

T9

T10

T9

T9

T4

Notes