University of Virginia Library

Setting By Formes in Quarto Printing
by
George Walton Williams [*]

IT HAS GENERALLY BEEN ASSUMED AS "VIRTUALLY axiomatic" that the pages of hand-set books of the sixteenth century were composed in seriatim order in first editions.[1] Moxon had suggested to be sure that casting-off manuscript copy and setting by formes were possibilities, but his tone intimated that the phenomenon was both troublesome and of infrequent occurrence.[2] Dr. William Bond in 1948 advanced the tentative hypothesis based on line-counts that setting by formes did occur in sixteenth-century quartos and octavos.[3] More recently Dr. Charlton Hinman has irrefutably demonstrated that the Shakespeare First Folio was regularly set by formes.[4] In this paper I shall argue that setting by formes in quarto printing was a more common occurrence than has been thought.

The evidence for this theory is derived from the sufficiency and deficiency of particular characters from a given font in a given text. It is based on two observations: (1) the count of type-pieces of a character as they appear in a sheet or a forme, and (2) the presence of substitutions for the character which indicate that the supply of the character has been exhausted in the course of the work.

A few difficulties must be admitted and a few cautions mentioned at the outset. Not all substitutions point perforce to the exhaustion of the proper character, and insignificant substitutions must be distinguished from significant ones. Studies of type shortage should depend on prior compositorial analysis. Differences in compositors or differences in their cases can obscure or invalidate the results of any findings.

Often the evidence is not really comprehensible; presumably the human element has intervened. Occasionally the compositor, knowing that he would run short of a character, seems to have saved his characters for use in prominent locations, mixed his proper and his improper sorts indiscriminately, or distributed his wrought-off pages in an unpredictable time sequence. Some substitutions may have been used to assist in justifying lines without regard to the supply or shortage of the proper character. Difficulties also exist in tracing shortages through the successive sheets of a quarto; characters in demand in one sheet are often not needed in another so that a continuing pattern is not attainable. But in spite of all these difficulties, the method is serviceable in some cases, as the following examples will make clear. Those presented first were all set from the same font of type by the same compositor working in the shop of Thomas Creede between 1593 and 1599.[5]

The first example of this investigation is found in the Epicedium: A Funeral Song upon the Death of the Lady Helen Branch (STC 12751), printed by Creede in 1594, one of the first works to be issued from his newly established shop. The Epicedium is a single-sheet quarto comprising 144 lines of verse, printed in roman type of a size approximating to modern pica (20 ll. 84 mm.). This particular font of type was apparently not supplied with the characters for lowercase and upper-case 'w' in sufficient quantity for this particular job, and the supplies of those characters were exhausted during the setting of the sheet. The place of the lower-case character was then filled with 'vv', and the place of the upper-case character with 'VV' — substitutions common enough in the period. The following table lists the incidence of 'w', 'vv' and 'W', 'VV' in the sheet:

                 

	A1	A1v	A2	A2v	A3	A3v	A4	A4v	Total
w		-	20	19	3	26	15	-	83
vv	1	-			30				31
									--
									114
W	-	-	8	4	3	4	2	-	21
VV					2				2
									--
									23

The lower-case character was required 114 times in the sheet. On 83 instances the 'w' was set in the proper fashion, but on 31 instances on pages A1 and A3 the substitute 'vv' was used in its place. The upper-case character was required 23 times in the sheet. On 21 instances the 'W' was set in the proper fashion, but on two instances on page A3 the substitute 'VV' was used in its place.

I have suggested that before drawing any conclusions from the appearance and disappearance of a character, we must be aware of the compositorial situation before us. But in any given problem we must also be satisfied as to the reactions of the compositor in one particular. In this example we must know the compositor's attitude toward 'w' and 'vv'. On signature A3 where both forms of both characters occur, the dividing lines between the proper and the substitute forms are distinct; the proper forms appear at the top of the page and the substitute forms follow them.[6] We may reasonably infer from this fact that the compositor was not indifferent to the w/vv distinction. He obviously used the proper character until he had no more. We may have some confidence, then, in believing that when this compositor set the upper- or lower-case 'vv' form, he did so because his supply of the proper sort was exhausted, and we may conclude that in the Epicedium the substitution is significant.[7]

It seems reasonable to suppose that the compositor began his type-setting with a supply of the 'w' character in his cases. It follows then if the compositor was careful in this matter — and we have just seen that he was — that the sufficiency of the character should appear at the beginning of the work and the shortage of the character and its substitution at the end of the work. From the evidence of the shortage of 'w' and 'W', we may conclude that pages A2, A2v, A3v, and A4, which exhibit a sufficient supply of the character were composed before pages A3 and A1, which by the use of the substitution exhibit an insufficient supply. An order of setting in which A3v and A4 precede A3 and A1 is comprehensible only if we assume that the compositor was setting by formes, the inner forme preceding the outer:

         

	A1v	A2	A3v	A4	A2v	A3	A4v	A1
w	-	20	26	15	19	3
vv						30	-	1
W	-	8	4	2	4	3
VV						2	-	-

This rearrangement by formes reveals a sufficiency of the 'w' and 'W' characters for all the pages of the inner forme, A1v (a blank), A2, A3v, and A4 — and for the first page of the outer forme, A2v. Three more 'w' forms and three more 'W' forms appear in the upper section of A3, but the substitutions for these characters occur throughout the rest of the page. The 'vv' substitution found on the title page, A1, argues that that page was set last in the forme (not unnaturally). Page A4v is a blank. The outer forme thus displays the shortage of 'w' in two of its pages and must therefore have been set after the inner.

Two objections to the theory of setting by formes for this quarto come at once to mind: the possibility of a distribution of type after page A3, and the difficulty of casting-off copy so as to set pages out of normal consecutive order. As to the first, a distribution of type in the midst of a work as small as the Epicedium would have been improbable.[8] Furthermore, such an argument cannot logically explain the presence of the 'vv' on the title page. As to the difficulty of casting-off copy, there is none. The verse is in twelve-line stanzas which are set three to a page. There could have been no difficulty in counting off the number of stanzas and setting them out of their seriatim order. Neither of these objections is of any genuine concern in the present example, and we may safely conclude on the basis of the evidence of type shortage in the pages of the outer forme that the quarto was set by formes, the inner preceding and the outer succeeding.

Another work printed by Creede which made heavy demands on this font was The First Part of the Contention betwixt the Two Famous Houses of Yorke and Lancaster (STC 26099), issued in 1594. In this quarto shortages occur frequently. Most conspicuous is the shortage of lower-case 'w', though there seems for this play to have been a sufficiency of the upper-case 'W'. Shortages in the associated italic font are also evident, but they seem to present no clear pattern. Random mixing of roman and italic forms of 'k', 'K', 'S', and 'Q' and the substitutions of 'VV' and 'W' for the proper 'W' (of which only three pieces seem to exist) are common in the quarto and are without

significance. The shortage of lower-case 'w', on the other hand, discloses a pattern throughout the quarto.

An ample supply of the character was available for most of the pages of sheet A:

     

	A1v	2	2v	3	3v	4	4v	1
w	-	8	19	17	25	35
vv						2	23	1

Pages A2, A2v, A3, and A3v were set with the proper character; on A4 the supply was exhausted at the bottom of the page, and A4v was set with the 'vv' substitution entirely. The division between the 'w' and the 'vv' forms is a clean one, the 'w' forms appearing at the top of the page, the 'vv' forms at the bottom. Such a pattern argues here for seriatim composition of sheet A beginning with A2 (A1v blank). At the same time the evidence of the single 'vv' on A1, the title page, suggests that that page was composed last, after A4v (as in the Epicedium).[9]

A seriatim arrangement for the pages of sheet B produces what would seem to be two distinct and contradictory instances of font depletion in adjoining pages, B3 and B3v. The arrangement by formes (the outer first)[10] offers a solution to the contradiction.

     

	B1	1v	2	2v	3	3v	4	4v	B1	2v	3	4v	1v	2	3v	4
w	23	21	29	24	15	25		28	23	24	15	28	21	29	25
vv					6	10	16				6				10	16

This rearrangement suggests that after the composition of B3 the type used in earlier press-work was distributed and the cases were restocked. Once more the divisions of 'w' and 'vv' on the two pages, B3 and B3v, on which both forms appear are clear-cut and distinct.[11]

Corroborating evidence for a distribution after B3 and for the font depletion on B4 exists in the appearance of substitutions for 'W' and for 'I'. Near the bottom of B3, just one line above the first of the 'vv' substitutions, occurs the unique example of 'VV' in the quarto. It indicates that the supply of 'W' — like the supply of 'w' — has been exhausted at that point. Similarly at the bottom of B4 occur eight italic 'I' substitutions for the correct roman 'I'. Since they occur in the last appearances of the character in the forme, they indicate that the font has been depleted and that the supply of the proper roman character has been exhausted.

Sheet C is likewise more regularly arranged by formes, the outer preceding. This rearrangement is based more on the relationship of the type-pieces in this sheet to adjoining sheets than on the equivocal evidence of the type shortage within the sheet.

     

	C1	1v	2	2v	3	3v	4	4v	C1	2v	3	4v	1v	2	3v	4
w	22	34	31	19	24	19	1	26	22	19	24	26	34	31	19	1
vv						5	24							5	24

Sheet C exhibits the shortage of 'w' in signatures 3v and 4, the last pages of the sheet to be composed. Sheets D and E exhibit the shortages in 4v. Hence we may conclude that 4v was the last page to be composed in sheets D and E and that the composition of these sheets was therefore probably seriatim. This conjecture involves certain assumptions, chief among which is this that the compositor was well in advance of his press at the beginning of sheet D. That this assumption is correct and that the compositor was not only ahead but was gaining on his press are indicated by the evidence for the distribution of type and the restocking of the cases. Thus it would seem that one forme was distributed and the supply of 'w' replenished before E1 was set (after depletion on D4v), one forme was distributed before F1v was set (after depletion on F1), and one forme was distributed before G2 was set (after depletion on G1v). The progression from one to one-verso to two suggests that the compositor was able to set the sheets of this quarto faster than his press could print them, for the delivery and the distribution of the wrought-off forme come increasingly later in the composition period.

                   

	D1	1v	2	2v	3	3v	4	4v	E1	1v	2	2v	3	3v	4	4v
w	18	13	30	27	17	23	16	15	30	25	25	24	33	33	19	20
vv								13								4
	F1	1v	2	2v	3	3v	4	4v	G1	1v	2	2v	3	3v	4	4v
w		23	17	17	29	29	20	26	24	13	27	26	28	25	32	9
vv	31									15						14
	H1	1v	2	2v	3	3v	4	4v
w	5	24	19	32	21	1	6	-
vv	18					6
w						13

Sheet H presents two difficulties in interpreting the w/vv pattern. The appearance of five 'w' forms on sig. H1 is not readily explicable. The pattern of deficiency and sufficiency for H3v is unique in the quarto. The page begins with a single 'w' form; this is followed by six 'vv' forms; and these by thirteen 'w' forms. It is arguable that the supply of 'w' was exhausted at the top of the page and that the compositor then turned to his substitution ('vv') for six settings. At this moment the type of a forme released by the press was distributed, and the cases were restocked. The compositor was thus able to finish sheet H with the proper 'w' character as he had begun it.

The fresh supplies of 'w' available for pages B1, B4v, C1, D1, E1, F1v, and G2 testify to a distribution of an earlier forme at the beginnings of these pages. The appearance of eight substitutions of 'I' for 'I' at the bottom of B4 and of two at the bottom of F1, and the resumption of the proper form on C1 and F1v respectively, confirm the fact that the font was depleted at those points.[12] The appearance of six substitutions of 'A' for 'A' in the middle of D2v suggests an exhaustion and a restocking on that page.[13]

Distribution after:

 

A1

B3

B4

C4v

C4

D2v

D4v

E3

F1

F3

G1v

G3

G4v

H1

H3v

             

Pieces utilized between points of distribution:
62	103	91	85	77	82	137	72	86	112	81	66	5	97	19
Pieces released by distribution of forme
62	68	36	90	75	91	85	77	82	107	102	72		89	87
	Ai	Ao	Bo	Bi	Co	Ci	Do	Di	Eo	Ei	Fo		Fi	Go
Increment:
	35	50				50

The preceding table indicates a high degree of correlation between the number of pieces released and the number of pieces utilized. Thus between pages E3 and F1, 72 'w' pieces are used; these are supplied by the distribution of D outer — containing 77 pieces — at the end of E3. Between G3 and G4v, 66 pieces are used; these are supplied by the distribution of F outer yielding 72 pieces. Sheets A and B derive their supply from initially full cases and from distribution of earlier press-work. Sheets C and E made extraordinary demands on the font. The fact that the shortages of 'w' do not appear in the earlier pages of these two sheets indicates that some extra pieces of the sort must have been added to the font during the composition of both sheets.[14]

The analysis of the shortage of 'w' in the Contention has presented an hypothesis for the order of the pages in the sheets and for the formes in the press. The appearance of a pattern of generally predictable depletion and distribution of the 'w' character in the play corroborates an argument for a single set of cases and for a single compositor and allows a conjecture as to the speed and the facility of the compositor.

The Menechmi (STC 20002), printed by Creede in 1595, offers an informative example of the compositor's solution to another problem of deficiency of type, this time in the italic font associated with the roman font under discussion. This English translation of the Plautine comedy exhibits a shortage of italic capital 'M'. This shortage is not occasioned by normal demands of the text on an inadequate font; rather is it the result of abnormal demands of the text on an adequate font. It is this compositor's practice to set speech prefixes and proper names in the text in italic, and as the play includes the characters of the Menechmi, Messenio, Mulier, and Medicus, and as mention is

made of Moschus, the need for the italic 'M' is frequent. The supply of the letter is exhausted early in the composition, and the roman capital is substituted for the italic.

The substituted pieces in this play are not divided from the proper pieces by the clear lines of demarcation that we have noticed in the preceding examples. On the contrary there appears after the first few pages a complete indifference to the distinctions between italic and roman capital 'M'. It is evident that the compositor's solution (a common one) has been to distribute italic and roman pieces of the capital 'M' into the same box indiscriminately and to use them thereafter without any regard to proper style. It is possible, however, by counting the number of italic pieces on each page to determine the proportion of italic to substituted roman. These ratios show clearly that in the pages of the outer formes there are more italic 'M' pieces than there are roman and in the pages of the inner formes there are fewer italic 'M' pieces than there are roman. The pages have consequently been arranged in the table by formes, the outer preceding as they exhibit the greater supply of the italic. (To conserve space A2, title page, and A2v, Argument, have been omitted; they contain no 'M' forms.)

               

	A3	3v	4	4v	B1	2v	3	4v	1v	2	3v	4	C1	2v	3	4v	1v	2	3v	4
M	-	1	2	3	15	8	3	8	9	3			15	9	5	8	5	1	4	2
M							6	3		4	14	13	4	1	8	12	6	3	5	11
Total: italic/roman		34/9	12/31	37/25	12/25
	D1	2v	3	4v	1v	2	3v	4	E1	2v	3	4v	1v	2	3v	4	F1	1v	2	2v
M	12	13	6	6	3	3	1	4	5	3	10	18	9	8	11	15	7	3	-	-
M	11	5	4	5	6	4	8	7	3	4	7	11	11	1	7	14	15	7
37/25	11/25	36/25	43/33	10/22

The last line of the table contains the ratio of pieces by formes. The figure to the left of the virgule represents the number of italic pieces in the forme, that to the right the number of roman pieces. Thus we have in sheets B, C, and D, a high proportion of italic in the outer and precedent formes, a low proportion of italic in the inner and subsequent formes. It would seem, furthermore, that the supply and the proportion of italic 'M' used for setting each forme were derived from the distribution of the comparable forme of the immediately preceding sheet. So we note that the twelve italic 'M' pieces used in sheet B inner are reused (and no others) in sheet C inner; the 37 italic pieces in sheet D outer are reused (and no others) in sheet E outer.

The outer forme of sheet E evidently parallels the outer formes of sheets B, C, and D, for it shares the same relatively high proportion of italic pieces. But the inner forme of sheet E offers figures that are entirely disproportionate to all other inner formes, and that resemble in fact much more the figures of the outer formes. On the other hand the pages of half-sheet F exhibit, where we should expect figures correlative to the outer forme, figures exactly correlative to the inner formes. The situation would be explained by supposing the composition of half-sheet F prior to the composition of the inner forme of sheet E. When this rearrangement is made, the resulting figures are found to be perfectly regular and in keeping with the rest of the quarto.

       

	D1	2v	3	4v	1v	2	3v	4	E1	2v	3	4v	F1	1v			E1v	2	3v	4
M	12	13	6	6	3	3	1	4	5	3	10	18	7	3	-	-	9	8	11	15
M	11	5	4	5	6	4	8	7	3	4	7	11	15	7			11	1	7	14
37/25	11/25	36/25	10/22	43/33

The hypothesis is a curious one and would hardly be tenable were it not that the analysis of the running titles proves that after E outer was imposed, half-sheet F was interjected before the imposition of E inner.[15]

There is an explanation for this phenomenon. The proportions of the mixed font of italic and roman 'M' reveal that the compositor was distributing his formes in a regular pattern and was managing to keep up with the press. But he was not gaining on the press as he was in the Contention, and any opportunity to save composition time would therefore have been welcomed. The composition of half-sheet F was such an opportunity. Half-sheet F consisted of two pages of type and two blank pages; these could have been prepared in just half the time required for the usual four-page forme. But these two pages required just as much time on the press as a full four-page forme. If the compositor was being hurried by the press, he could have gained time by setting the two-page forme out of order. Running-title evidence and type-shortage evidence combine to argue that this procedure was followed here.

But the question may well be posed why the compositor in the Contention was able to gain on his press and in the Menechmi was not. The answer lies in the nature of the texts. The Contention was a public play on a patriotic subject designed for the boards; the Menechmi was a private play, designed for the closet. It was the first translation of Plautus printed in English and was admitted to be a trial venture in the printer's address to the reader. There is hardly a need to look for further evidence that the edition was a small one. In a small edition press time would be briefer than composition time, and the compositor would always be concerned lest he fall behind and so delay his press.[16]

The Comicall Historie of Alphonsus, King of Aragon (STC 12233), printed by Creede in 1599, presents yet another solution made by the compositor to the problem of a deficiency of type in the italic font used also in the Menechmi. The normal practice of the compositor is to set all proper names in the text in contrasting italic type. In a violation of that practice, proper names of persons are set in this play in non-contrasting roman type in the inner formes of sheets B and C and in the outer formes of sheets D and E. A closer look discloses that the question is not simply one of the setting of personal names, but of the setting of personal names beginning with 'A'. The text includes the characters Alphonsus, Albinius, Amurack, and Arcastus, and references are frequent to Aragon, Asia, and Africa. It is difficult to avoid the conclusion that the shortage of capital italic 'A', induced by the excessive need for the character, is closely related to the failure to use italic for personal names, first for personal names beginning with 'A' and thence for all personal names. It is on sig. B2 that the shortage of italic 'A' is first apparent, and it is on the same page that personal names are first set in roman type. Thereafter the shortage of 'A' as it appears in the various pages of the sheets clearly argues for composition by formes, for the precedence in composition of the outer forme in sheets B and C, the inner forme in sheets D through H, and for distribution after each sheet.[17]

                   

			A3	3v	4	4v	B1	2v	3	4v	1v	2	3v	4
Italic 'A' found in	prefixes	1.					2	4	3	1	5		2	8
Italic 'A' found in	place names	2.	1			1	1	2	1	2				1
Italic 'A' found in	personal names	3.	3	2		2	3	3	6	1	2
Roman 'A' found in	prefixes	4.										6
Roman 'A' found in	italic personal names	5.										1	1
Roman 'A' found in	roman personal names	6.										8		1
Roman 'A' found in	italic place names	7.										1
Roman 'A' found in	roman place names	8.											1
Roman 'A' found in	Other personal names in roman	9.										1		4

                   

	C1	2v	3	4v	1v	2	3v	4	D1v	2	3v	4	1	2v	3	4v	E1v	2	3v	4	1	2v	3	4v
1.	1	2	4		3	2	1	2	1	1	3	1		1	2	1	1						1
2.		2		1	1	1			1	1	2	1			4	4				1	1	1
3.	3	1	5	1				1	3	4	4	5	1	1			4	5	1	6	1	7		6
4.
5.
6.					3	2	2	4					3	3	7	6					4		2
7.		1										1
8.
9.					3	1	4	3					7	5	3	6					3		4	1

                   

	F1v	2	3v	4	1	2v	3	4v	G1v	2	3v	4	1	2v	3	4v	H1v	2	3v	4	1	2v	3	4v
1.									3	2	3		2	1	4		3	1	2	2		1	3
2.	1			1	2		1					2
3.	4	1	2		3	2	1		3	2	3	2	2	1	6	7	4	5	2	5			4
4.																					1	1		2
5.							4	1					1								2	5		3
6.
7.							1	3
8.
9.

                   

	I1	2v	3	1v	2
1.	1	2	-	2	2
2.
3.	2	1		3	5
4.
5.	1	2			4
6.
7.
8.
9.

Confronted with the problem of an inadequacy of italic 'A', the compositor was obliged to find a solution. His solution was to set personal names in italic until the supply of italic 'A' began to diminish (i.e., after the completion of the first forme of the sheet), then to set such names completely in roman in order to avoid the mixture of italic and roman in the same word (the mixture is standard in the Menechmi). Such a decision provided for (1) uniformity of font in each word, (2) uniformity of pattern in setting (by including all personal names, not merely those beginning in 'A'), and (3) a supply of italic 'A' for use in place names and in speech prefixes, where the absence of the proper font would be more conspicuous.

Thus at the beginning of B2, having used 45 italic 'A' forms, the compositor was faced with a demand for sixteen more pieces of the character. These he did not have, and he was forced to substitute the roman 'A' to make up the deficiency. He set the prefixes on the page in italic type but with roman capitals, and he set one word in the text in the same "mixed-font" manner. At this point the decision was made to set all personal names in roman in the text as the supply of italic 'A' began to diminish, so that the lower-case letters would be uniform with the roman capitals, and to reserve the italic capitals for the prefixes. The few remaining italic pieces were used for the prefixes on B3v and B4, and personal names in the text were set in roman.

The pattern thus described for sheet B with a preceding forme containing personal names fully italicized, a succeeding forme showing signs of shortage of italic 'A' by the use of roman personal names is constant through sheet E. Sheet B required 66 italic 'A' pieces, and on the basis of this requirement the compositor adopted his method for setting the rest of the quarto. The number was above average, for no other sheet required so many. The compositor therefore began to abandon the policy in sheet E, setting personal names beginning with other letters in italic after E3, and in sheet F setting personal names beginning with 'A' in italic even though a roman capital had to be used. Sheet F used only eighteen pieces of italic 'A', and as a result sheet G was set with only one irregular roman form, doubtless an oversight. The 43 capitals required in sheet G induced the usual shortage in sheet H and in turn in sheet I.

The evidence from type shortage in Alphonsus tells us the order of composition of the formes and indirectly reveals information on a quite different matter, the setting of roman type in the place of italic.

The examples presented thus far have been from a small shop. Though the limited facilities of the small shop offer more opportunity

for evidence of type shortage, these examples are not unusual, and exceptional demands of certain texts will cause type shortages to appear in works from larger shops as well. One more example from another shop will be presented here.

In his discussion of the "bad" quarto of Romeo and Juliet (1597) (STC 22322), Dr. Harry R. Hoppe pointed out that sheets E through K were set by two compositors alternating in this manner:

   

X:

E1

E2vE3-4v

F3-4v

G3-4v

H3-4v

I3-4v

K3-3v

Y:

E1v-2

F1-2v

G1-2v

H1-2v

I1-2v

K1-2v

K4

Mr. Hoppe's argument was based on spelling characteristics of the two men.[18] The evidence from type shortage suggests that Compositor X was setting his half of the sheet by formes (the outer first) in the order 3 4v 3v 4, with a distribution at the beginning of each sheet (with the possible exception of sheet H).[19] On eleven of the thirteen pages where both forms of letters occur, the dividing lines are distinct; on F4v 'R' and 'R' are mixed, and on F3v one proper 'W' is misplaced.

                   

	E				F				G				H				I				K
	3	4v	3v	4	3	4v	3v	4	3	4v	3v	4	3	4v	3v	4	3	4v	3v	4	3	3	(4)[20]
R	4	5			3	3	8		3	1	4	1	1		1		3	1	2	4	4	1
R			4	8		5	1	8														5	4
I		5	6	2	5	3	5		5	2	3	5	4	4	1	5	2	2	1	3	2	3	3
I				3			2	2				6								3
N	5		1		4		5				2	3		3	1	2
N			7	2				1
W					4	2	4		2	3		8	1	6	2	3		2	4		7	1
VV							3	4		2			5						1	4		9	6

The pattern of type shortage in the work of Compositor Y is much less clear than that in the work of X, though Compositor Y too would

seem to have set by formes (indeed there would be no purpose in having X set by formes and Y set seriatim), because the cases used by Y are in several letters more adequately stocked than those used by X, and for other letters Y is less concerned to keep his proper type-pieces and his substitutes unmixed. The evidence is not so convincing as could be wished, and I hesitate to speculate further.

The examples that have been presented in this paper have demonstrated the utility of this particular sort of investigation in shedding light on mechanical and on compositorial practices that offer problems to the bibliographer. Type shortages can often reveal the order of composition of the pages of a sheet and the times of distribution; sometimes they can point to much more information. As the evidence can be shown to derive from peculiarities of the type and from the nonliterary reactions of the compositor and not from any oddity of the copy, these instances provide arguments from what is purely bibliographical information. Type-shortage evidence should provide a new tool for bibliographical investigation.

Notes