University of Virginia Library

Evidence of Plate Damage as Applied to the First Impressions of Ellen Glasgow's The Wheel of Life (1906)
by
Oliver L. Steele

Plate damage is one of the most useful kinds of evidence available to bibliographers for distinguishing the impressions of modern books.[1] However, as this note will suggest, such evidence is potentially treacherous and must often be correlated with another kind of evidence if it is to be used correctly. Secondarily, but perhaps more valuably, this note will suggest a tentative solution to the problem of imposition in modern books, a problem which bibliographers have had little success in solving.[2] It should be remembered that the solution offered here is, as stated, tentative and needs further and more rigorous testing.

In a sample of 150 copies of what, for the present, will be called the first impression of The Wheel of Life, all have in common the misspelling "forefeiture" in line 17 of page 449.[3] Present in all copies is an extraordinary

amount of plate damage, the most significant of which is listed below:

• Inner forme, 6th gathering (beginning p. 75), p. 84, last line (so sincere).
• Outer forme, 6th gathering, p. 86, running title.
• Inner forme, 28th gathering (beginning p. 427), p. 432, last line (to).
• Outer forme, 28th gathering, p. 434, last line (humiliated).
• Inner forme, 29th gathering (beginning p. 443), p. 444, last line (meant).
• Outer forme, 29th gathering, p. 446, last line (ideal).[4]

Damage to these three gatherings is distributed throughout the 150 copies in the following patterns:[5]                                                  

		84	86	432	434	444	446
1.	Vassar C	X
2.	British Museum		X
3.	Iowa State U	X			X		X
4.	Purdue U	X		X			X
5.	Howard C	X			X	X
6.	Smith C		X	X			X
7.	Wyoming U		X		X	X
8.	Minnesota U		X		X		X
9.	Georgetown U		X	X		X
10.	Oberlin C	X		X		X	X
11.	Wisconsin U	X		X	X		X
12.	Fresno P L		X	X	X	X
13.	San Diego P L		X	X	X	X
14.	Virginia U	X		X	X		X
15.	Boston U	X		X	X	X
16.	Pittsburgh U		X	X		X	X
17.	Howard U	X	X		X		X
18.	Birmingham P L	X	X	X			X
19.	Toronto U	X		X	X	X	X
20.	Georgia U		X	X	X	X	X
21.	Swarthmore C	X	X		X	X	X
22.	Arizona U	X	X	X	X		X
23.	Princton U	X	X	X		X	X
24.	Wellesley C[6]	X	X	X	X	X	X

Although there is a perceptible progression of damage from the Vassar copy to the Wellesley copy, apparently there was some mixing of early and late states or impressions of gatherings within copies. But the makeup of various copies is best approached by analyzing the printing of the sheets

of the book, and since the kind of damage found in gathering 6 is characteristic of all three gatherings, the analysis will be centered there. One might hypothesize three possible ways of printing gathering 6: 1) simultaneous impressions from two sets of plates,[7] 2) different impressions from one set of plates, 3) a single impression from one set of plates. Of these, only the third is at all probable. The second can be ruled out immediately. For if a first impression produced the damage to page 84 found in the Vassar, Boston University, and University of Toronto copies; then any subsequent impression would of necessity also produce it, and such copies as the British Museum and University of Georgia copies could not possibly result. The first possibility is also untenable. For if one set of plates produced the damage to page 84 found in the Vassar copy and another set the damage to page 86 found in the British Museum copy, then such copies as the Wellesley copy could only have been the result either of plate damage which developed in both sets of plates at exactly the same places, or of a third impression printed from formes made up of the damaged page plates of both sets. Neither of these explanations, depending so heavily on unlikely coincidence, is convincing.

The 150 copies listed, then, contain gathering 6 in only one impression. Furthermore, the damage found there must be in-press damage, plate damage which develops during printing rather than between impressions; for the supposition that some of the damage occurred to the type before casting the plates or to the plates before printing can be rejected on the same grounds as different impressions from one set of plates. The fact that all of the damage considered in this note is, like that of gathering 6, in-press damage casts some doubt upon the use of this kind of evidence as a means of distinguishing separate impressions of modern books. The damage found reflected in these gatherings occurs at the top, bottom, and side edges of pages, just where page plates are most likely to be damaged during storage between impressions. For that reason, given several copies like the Iowa State copy and several more like the Wellesley copy, one might have distinguished two impressions of the novel without misgivings. The only cure for this problem, in the absence of other evidence, is a reasonably large sampling of impressions and, often, a close analysis of damaged gatherings in an attempt to relate the damage to possible methods of printing.

With the fact established that all 150 copies of gathering 6 are of the same impression, one can reconstruct with near certainty the method by which the impression sheet was imposed. The reconstruction requires only the assumption that the 150 copies examined are an adequate sampling of the impression sheet.[8]

In an impression, say, of 3,000 copies, the ratio of damage variants in gathering 6 would be:

• 63 copies damaged on p. 84 only 1,260 copies
• 44 copies damaged on p. 86 only 880 copies
• 43 copies damaged on pp. 84 & 86 860 copies

Two methods of imposition could have produced the required variants in the correct ratio. The less likely method, printing the white paper with two inner formes imposed together (gatherings 5 and 6) and perfecting with the two complementary outer formes, would result in

• 1. 880 sheets printed without damage
• 2. 2,120 sheets printed with damage to p. 84
• 2. a. 1,250 sheets perfected with damage to p. 84
  b. 860 sheets perfected with damage to pp. 84 and 86
• 1. 880 sheets perfected with damage to p. 86

or 3,000 copies of gathering 6. Clearly this method could produce the ratio and variants of the sample, but it could do so only if the order of perfecting outlined above were followed exactly. The introduction of an undamaged inner into the press before page plate 86 had become battered during perfecting would have produced an undamaged sheet. Further, this explanation would require that, in the printing stage, the sheets were delivered to the table printed-side-down; for if they had been delivered printed-side-up, then the undamaged inners would have been the first sheets perfected and undamaged sheets would have resulted.[9]

It is much more likely that another method of imposition — work-and-turn of the inner and outer formes of the gathering — was used. In this method the white paper is printed with the inner and outer formes of the same gathering imposed together; the sheets are then flipped over the short axis and perfected by the same inner and outer formes. Each sheet, when cut, produces two identical half-sheets. This is the more probable method because it requires no definite order of perfecting to have produced the

variants and ratio of the sample and it could not produce an undamaged gathering:

• 63 copies 1,260 half-sheets 630 sheets
• 44 copies 880 half-sheets 440 sheets
• 43 copies 860 half-sheets 430 sheets

• 1. 880 sheets printed undamaged
• 2. 380 sheets printed damaged to p. 84
• 3. 240 sheets printed damaged to pp. 84 and 86

• 1. perfected 880 half-sheets (gatherings) with p. 84 damaged and 880 half-sheets with p. 86 damaged
• 2. perfected 380 half-sheets with pp. 84 and 86 damaged and 380 halfsheets with p. 84 damaged
• 3. perfected 480 half-sheets with pp. 84 and 86 damaged

or 3,000 copies of gathering 6. Because plate damage developed during printing (rather than during perfecting), the order of perfecting is of no importance since any order would have produced the same variants and ratio.[10]

That all of the damage did occur during printing, assuming that the ratio of damaged states (63/44/43) in the sample is correct for the impression, can be demonstrated by testing the idea that the damage to page 84 (or to page 86) occurred instead during perfecting. Had that been the case it would have taken 860 full sheets to produce the gatherings with damage to both pages 84 and 86, and 1,260 full sheets to produce the gatherings with damage to page 84 alone. But this is a total of 2,120 sheets — over 600 too many. Furthermore, the ratio of gatherings would have been 63/212/43.[11] If the sample used is an adequate reflection of the impression, then gathering 6 was printed by the work-and-turn method and damage occurred before perfecting.

This conclusion, however, appears to be contradicted by the evidence of gatherings 28 and 29. For if the work-and-turn method does not produce undamaged gatherings when damage occurs during printing, then there seems to be no explanation for the Vassar and British Museum copies. One might return for explanation to the theory of printing with two outers and perfecting with two inners. However, disregarding the inherent objections

to this method, an analysis of the printing of gathering 28 by this method will show that such an explanation is an unlikely one:

• 6 copies with no damage 120 sheets
• 40 copies with damage to p. 432 800 sheets
• 29 copies with damage to p. 434 580 sheets
• 75 copies with damage to pp. 432 and 434 1,500 sheets

The following order of printing the sheets, or one similar to it, is necessary to produce the variety and number of copies found in the sample:

• 1. 600 sheets printed without damage to p. 432
• 2. 2,300 sheets printed with damage to p. 432
• 1. a. 120 undamaged sheets perfected without damage to p. 434
• 2. a. 800 sheets with p. 432 damaged perfected without damage to p. 434
• 1. b. 580 undamaged sheets perfected with damage to p. 434
• 2. b. 1,500 sheets with p. 432 damaged perfected with damage to p. 434 also

or 3,000 sheets.

Although this method will explain the states of gathering 28 found in the sample, the circumstances which would have led to such a peculiar sequence of damaged and undamaged sheets during perfecting are hard to imagine.

If this method of printing may be rejected as unlikely, the only alternative is to suppose that two impressions of gathering 28 (and gathering 29) are represented in the sample. The first impression is represented by copies like the Vassar and British Museum copies and the second by all those copies which show some damage in gathering 28.

That this is the correct explanation of the states of gatherings 28 and 29 is strongly supported by some striking physical differences between copies which contain these gatherings in an undamaged state and all other copies presumably.[12] All of the former are cut and trimmed, with a uniform page length of 19.2 cm; all of the latter which I have seen or had report of are partially uncut and untrimmed and have page lengths varying within copies from 19.7 cm to 20.3 cm.

The difficulty with this explanation is to account for the small number of copies containing the first impression of gatherings 28 and 29. This problem is perhaps best resolved by supposing that these copies were advance copies made up for reviewers. A possible explanation is that the printer began printing the gatherings by work-and-turn, 3,000 at a time, and that as the publication date approached, he decided to interrupt this process in order to print a limited number of the later gatherings of the

book to make up advance copies.[13] Thus all copies of what has been called the first impression of The Wheel of Life contain the first six gatherings, at least, in the first impression. Copies like the Vassar and British Museum copies contain the last three gatherings, at least, in the first impression also; and all other copies contain the last three gatherings in a second impression. Finally, since the second impression of gatherings 28 and 29 shows all the characteristic damage of gathering 6, the analysis made of the printing and imposition of gathering 6 is applicable to them as well.

At this point, it might be valuable to state some of the implications of this analysis. The most important of these is that it is possible to use plate damage to determine the method of imposition in modern books if certain conditions obtain: 1) damage must be present in both inner and outer formes of gatherings, 2) the sample on which analysis is based must be random and large enough to reflect a statistically sound ratio of damaged states within the impression.[14] The problem is obviously one of determining the size of the sample needed to give a statistically sound ratio. If the size of the impression can be found out, then the size of the necessary sample can be determined by conventional statistical methods. When, as in the case of The Wheel of Life, records are not available, a guess has to serve, and conclusions must, consequently, be tentative.[15]

So far, primarily for clarity, the assumption has been that the gatherings of The Wheel of Life were printed from both the inner and outer formes of a single gathering laid on the press together. That is, the bed of the press contained sixteen page plates. However, this assumption needs further testing. The cylinder presses of the time could certainly print thirty-two and even sixty-four standard novel-size pages at once, and it is a reasonable assumption that any shop which had such a press would have taken the obvious economic advantage of printing the larger sheet. Here too the evidence from damaged plates may explain the number of plates imposed together and their arrangement.

In the normal operation of a cylinder press, the edge of the type which bears the initial impact of the cylinder as it rolls over the bed of the press receives more stress and is, thus, more likely to be damaged than any other area of type. This edge is called the leading edge. After the leading edge the opposite (following) edge of the type receives most stress, and the side edges somewhat less. The type inside the four edges of a forme receives little stress and, barring accidents, is not likely to be damaged during printing.[16] In the following analysis of the damage in gatherings 28 and

29, it will be assumed that plates were damaged in the normal course of printing and that, therefore, the damage occurred at one of the vulnerable edges of the forme. The problem will be to associate the observed damage with one of the known methods of work-and-turn imposition so that it lies on one or more of those edges.

If the pages of gathering 28 are fitted into the usual scheme of imposition for work-and-turn of a single octavo gathering, the damage to the bottoms of pages 432 and 434 will appear internally within the edges of the forme. And if the same is done to the pages of gathering 29, the damage to pages 444 and 446 will appear on the side edges of the forme.[17] By reversing the positions of the inner and outer formes of gathering 28, the damage to pages 432 and 434 can be made to appear on the side edges also; but, of course, if the formes of gathering 29 are similarly reversed, the damage to pages 444 and 446 will appear inside the edges. If the assumption that the damage to the two gatherings occurred in the normal course of printing, and thus on the edges of the formes, is correct; then it is most doubtful that these gatherings were printed from the inner and outer formes of a single gathering imposed together. For to say that the damage to both gatherings occurred at the edges of the formes is to argue, in this case, that one gathering was imposed in a scheme different from the other. It is unbelievable that a printer would vary his imposition scheme from gathering to gathering.[18]

If the sixteen page plates of a single gathering were not imposed together in printing The Wheel of Life, it must be the case that the plates of two or more gatherings were imposed together. Again if the assumption about the location of plate damage is correct, the same objections which have been made to imposition of sixteen page plates in two octavo formes may be made to the imposition of sixty-four page plates in eight octavo formes to print four gatherings.[19] For in such a scheme the damage to gathering 28 would still appear inside the edges of the formes, and no arrangement, short of the kind argued against in regard to the sixteen page imposition, could bring it out to one or more of the edges.

DeVinne illustrates only one scheme of imposition for thirty-two pages containing the four formes of two octavo gatherings.[20] But this scheme explains the damage in gatherings 28 and 29 and, at the same time, allows the assumption that the damage occurred at the edges of the type in the normal course of printing. In this scheme, the damage at pages 444 and 446 of gathering 29 and at pages 432 and 434 of gathering 28, all, appears either

on the leading or the following edge — at those places where damage is most likely to occur during printing. I conclude, therefore, that this was the scheme used to print The Wheel of Life.[21]

Notes