University of Virginia Library

II.

Whatever may be the significance of the figures to the pressmen of the eighteenth century, their importance in contemporary investigation cannot be underestimated. By ignoring these convenient indices to the formes and sheets of various books, bibliographers have devised descriptions so general as to encompass as a single entity works which actually exist in innumerable states, impressions and editions. Even such authorities as Kaye, Griffith, Gulick, Sale, and Worthington, all of whom have occasionally reported the figures as a means of differentiation, fail to report them consistently and completely. Hence their accounts of Mandeville, Pope, Chesterfield, Richardson, and Scott are, one and all, inadequate. Indeed my casual enquiry into the use of these figures as a positive sign of variation among copies allows the assertion that practically every scholarly bibliography and edition of eighteenth-century literature rests on undiscriminated texts.

The great value of these figures lies in the convenience with which they may be recorded and subsequently used to distinguish and classify the variants they disclose. For this purpose one need not fill reams of paper with the distinguishing characteristics of the headlines (a fruitless procedure, I might observe, for most editions in this period) or travel about the country with cases of microfilm. The figures are sufficient criteria.[20] In most cases they may be recorded, as a necessary part of the description of a book, by a simple reference to the page and number [as 39—7]. In some, however, it may be necessary, for the purpose of bibliographical analysis, to convert this reference to less simplified notations

indicating the forme in which the figure is located [C(i)7], the position of the figure within the forme [C7v(i)7], or, for half-sheet imposition, the sheet [C—7] or page [C4v—7] identified by the figure. The method of notation employed should, I think, be determined by the book, and not by any arbitrary principle requiring the presentation of useless information for some works and not enough of what would be essential for others. In the discussion which follows I have, therefore, adopted the most convenient system for the book considered.

The analysis of variants. Variation in the presence or kind of figures for a sheet in copies which are otherwise of a single impression indicates (1) a disrupted impression, as this may be occasioned by the substitution of one pressman for another at the same press, (2) a reimpression at other presses to compensate for a miscalculation at the time the tokens were set out for the original printing, or (3) a resetting for the same reason as that accounted for in (2).[21] Though an examination of copies which vary in this respect will usually not permit a discrimination between (1) and (2),[22] it will reveal (3) and allow the explanation suggested. Whether (2) or (3), a decision concerning the priority of variants may be offered on the basis suggested by Dr. Kaye: the later one will have a pattern of figures differing from that evident in the other sheets.[23]

Imagination can easily supply any number of incidents as the occasion for an interruption of the kind indicated under (1)—the failure of the first man to report for work after an evening at his favorite pub, or his departure for lunch, or his dismissal for one reason or another. And sometimes a reasonable hypothesis may be deduced from the aberration. In the previously undifferentiated second editions of Pope's Imitations of Horace, The

First Epistle of the Second Book and The Second Epistle (1737), the work was performed by four numbered men and one designated as "τ".[24] For both of these second editions numbers 2, 3, 4, and occasionally 1 complete the impression of their formes, but τ works infrequently and then only on the press operated by 1. I think we may presume from this circumstance that 1 is the master printer in Woodfall's shop (perhaps Woodfall himself), and has under his tutelage a young apprentice who is allowed, now and then, to try his hand with a few sheets. If this much is presumed, the conjecture follows that where both figures and symbols appear in the same book, the figures may designate the master and journeymen, who are assigned numbers according to seniority, and the symbols, the apprentices, printers' devils, or "smouters" who are only occasionally employed at press.

Consistent variation in the figures for all sheets indicates different impressions or editions. Since the figures are usually entered on or immediately below the direction-line, they may be tied up with that line and the adjacent letter-press when the latter is removed from the forme, and remain as a part of the type page until they are replaced by the figures of other pressmen assigned to

work a subsequent impression. When this is the practice the numbers, as always, will change between impressions, but their relative positions within the formes will remain unchanged. A typical example of this procedure is observed in the two "editions" of Samuel Johnson's Thoughts on the Late Transactions Respecting Falkland's Islands (1771), an octavo in half-sheets collating A2 B-K 4 L2, and bearing these figures: Here, as the figures imply and as inspection confirms, the second variant, though called a "Second Edition" on its title-page, is actually a reissue (reimpression).[25] Not infrequently, however, the figures will be dropped, along with other superfluous matter, before the pages are retied for storage, and thus reappear in a new impression as different numbers in different positions. Where this has occurred, it is impossible to ascertain, from the figures alone, whether the variant is of the same or another setting of type.[26]

During the course of an impression regularly processed according to the order of its signatures, it may be decided to overprint the remaining sheets in anticipation of a second edition. Such a decision will be reflected in the figures, which will vary in number and position throughout that portion of the text which has been reset or reimpressed, but will usually remain invariable in the latter portion, where a single expanded impression suffices for both editions. Thus the incidence of identical figures provides, for the books concerned, instances of a procedure for which Dr. Bowers found no evidence in books of an earlier period:[27] that of a deliberate increase in the number of very late sheets machined at one time to be divided between two editions. The other practice to which Dr. Bowers refers as normal for sixteenth and seventeenth century printing is that of laying aside and later reimpressing the pages to be overprinted. Of the many examples of both procedures, apparently only five are now on record for the eighteenth century: those involving the two editions, first volume, of Gibbon's Decline and Fall; volumes 2 and 4 of the second and third editions of Chesterfield's Letters; volumes 1 and 3 of the third and fifth editions of the Letters; volume 7 of the first and third editions of Richardson's History of Sir Charles Grandison; and two editions of The New Ministry.[28]

Printing according to the order of accessibility. Occasionally it may be inferred, with varying degrees of certainty, that the initial impression of a book has not proceeded according to the order of its signatures, but in some abnormal sequence determined by the availability of the formes. Irregularity is implied, but not demonstrated, whenever identical figures in a second variant reappear, not in the latter portion of that variant, as we should expect for a normal overprint, but scattered throughout several interior sheets. In the absence of indications to the contrary, one explanation for these reiterated figures—as they occur, for example, in

sheets D, F, and H of Johnson's Thoughts on the Late Transactions—is that the pressmen so identified accidentally returned to the same formes they had machined before. But for some works, certain considerations disallow this conjecture. For these, we may presume that the formes involved were overlooked in the process of the first impression, either by the pressman, who for one reason or another neglected to machine them until toward the end of his work, when they were then used for an overprinting, or by the compositor, who pushed them aside while he imposed or distributed the type for the others.

Before fixing the responsibility upon one or the other we should trace the progress of the forme from the compositor's bench to the press-room and back again in order to determine where and how it could be mislaid. Normally, of course, the proportion between compositors and pressmen will be such that the work proceeds without interruption. Once the intended date of publication has been advertised, however, the distribution of work must be so arranged that the pressmen are never allowed to be idle, for the schedule depends upon their machining a certain number of formes in a given time. It will be of advantage to the overseer, therefore, that he have one or more formes ready at each press so that there will be no delay and consequent disruption of the schedule. The practice of stacking the formes awaiting impression we may consider to be a custom, for a regulation covers the matter:

Since preservation of type is the primary concern, it is probable that the formes were laid, even with partitions, not one on top of the other, but against a wall where they would be convenient to all the pressmen, or, if they were previously assigned to one, against that man's press. Wherever their location, the formes last imposed would be those most accessible to the men.

Infrequently it might happen, particularly if the preceding work is running to a great number of copies, that all the formes of

a pamphlet would be stacked and ready for impression, in which case they would be machined throughout in the inverse order of their signatures.[30] So long as the pressman keeps the two formes of a sheet together, it is immaterial whether he selects those for sheets G or M or X, for he knows that sooner or later he and his companions will "work their way to the wall," where he may find some of the early formes—B, D, and F, for example. And if, before he gets that far, the issue has been increased, then B, D, and F will constitute a single impression for the two editions, while G, M, and X, though later in the alphabetical sequence, will be reset or reimpressed in the second.

Though there must necessarily be an accumulation of ready-formes in the press-room in order to avoid delay, any accumulation of used formes was considered a nuisance and subject to fine.

After it has been returned to the "wrought-off place" or directly to the compositor, he too must not be dilatory in distributing type. Usually, of course, the compositor would need no urging to perform this task, for it would often be essential that he distribute in order to keep himself provided with enough sorts to continue composition. And as it takes only one-sixth as long to break type as it does to set it up,[33] he not only has the compulsion of a penalty and the necessity for reusing the type as encouragements to distribute, but sufficient time for the job. It may be said, therefore, that whenever interspersed formes of the first impression reappear in the second impression or edition, these formes have been machined according to the accessibility of the type in the press-room.

It is one thing to construct a plausible hypothesis, quite another to prove it. The three examples cited below are arranged according to the degree of irregularity and the amount of corollary evidence supporting the argument advanced. In the first of these, Chesterfield's Case of the Hanover Forces (1743, 80 in half-sheets, A2 B-L 4 M2)[34] a comparison of the two early editions shows these similarities and differences:

     

	Sheets reset in "B"	reimpressed	continuously impressed
Edition "A"	C-2 D-1 E-3	B-- L-3 AM-1 G-1	F-3 H-3 I-1 K-2
Edition "B"	3 1 1	3 1 1 3	3 3 1 2

The appearance of an identically positioned figure in the sheet presumably imposed as A2+M2 would seem to indicate a single impression; but since the date on A1 of the "A" edition invariably reads "M.CDD.XLIII.", while that in "B" reads, correctly, "M.DCC.XLIII.", an interruption has occurred between the variants. The sheet in "B" must be considered, then, either as a reimpression, with the presence of the identical figure accounted for as the accidental return of the man to the forme he worked before, or as a continuous impression, with the corrected state appearing, again accidentally, only in the later edition of the copies examined. Though accidence may similarly account for the reiterated figures in F, H, I, and K, the possibility of this diminishes with each additional sheet and may be discounted altogether with reference to four. It would be a strange coincidence indeed that out of the eleven formes of this book, number 2 should return to one of the two formes he had previously impressed. More likely is the supposition that the sheets for "A" were impressed in the order of their accessibility, those later reset being the first, followed by those subsequently reimpressed and, as the stack was uncovered, those continued as a single impression for the planned overprint. The progress of these formes through the press may therefore approximately correspond to this formulary:        

Edition	A	AB	B	B
Sheets	C-E, B, L, AM, G	K, I, H, F	B, L, AM, G	C-E
Copies	1000	1500	500	500
Category	impressed		reimpressed	reset

The number of copies for this and later examples represents an arbitrary assignment.

In Cumberland's The Imposters (1789, 80, A2 B-F8 G6) an order of printing according to accessibility seems to be the only reasonable explanation of the facts. The status of the first two "editions" is this:[35] Since the formes for every sheet, except C and E, were simultaneously impressed in the one issue, but consecutively impressed in the other, it would appear that there was some reason for a shift in the printing arrangements. And that reason, we may confidently assert, can be found in the circumstances pertaining to the impression of C, a sheet which, unlike the others, is apparently of a single impression, and simultaneously machined throughout. What happened, presumably, is that the formes for C were covered by those for D-G, and were thus not available until inner and outer G had been picked up by 1 and 6. Then, after the completion of a run of—let us say—500 copies of G, but before the completion of the same run on C, the decision was made to double the issue, whereupon x and 4 together continued to impress about 900 copies of C in approximately the same amount of time that it took 8 and τ individually to print and perfect 500 copies

of B and E. Number 8 then completed the work on AG while 4 machined D and F. The facts of printing, therefore, were probably not too dissimilar to what is represented by the formulary:

       

Issue	1	1-2	2
Sheets	B, D-G	C	B, E, G, D, F
Copies	500	1000	500
Category	impressed		reimpressed

Combined in Matthew Lewis's Alfonso, King of Castile (1801, 80 in half-sheets, A-P4) are several kinds of evidence which together provide a convincing demonstration of the irregular printing of the first impression. In this, A4v carries a list of fifteen corrigenda present in eight of the fifteen gatherings. In the second issue (reimpression) of the play the list has been withdrawn and all errata corrected except those appearing in the three sheets having the same figures as before—sheets H, I, and N. Then in the second impression of these sheets (the third issue of the play, titled "The Second Edition") four of the five errata which they contained are finally corrected. From this it follows that H, I, and N were printed without interruption for both of the early issues before the corrigenda were made available for the second, and before they were removed from the press and corrected for the third. To understand these complications as they are now beginning to develop it will be convenient to have before us a tabulation of the data for the pertinent variants.[36]

       

Sheet	B	C	D	E	F	G	H	I	K	L	M	N	O	P	A	i	a	Q	R
1st issue	(3)	(3)	-	4	3	(4)	(3)	(4)	3	(3)	4	(3)	6	(6)	3
2d & 3d issues	6	-	6	6	4	6	(3)	(4)	-	4	6	(3)	-	-	6	-	4	6	4
4th issue	-	6	4	4	4	6	4	6	6	4	6	6	6	-		-	4	-	-

Since H, I, and N are both by figure and by readings confirmed as a single impression for the first and second issues, we may suppose that they were printed toward the end of the sequence of sheets for the first and in sufficient quantities to provide copies

for both issues. Yet they must precede the last two or three sheets, for sometime during the work on P, first issue, the press was stopped for the correction of "requium" to "requiem" (105.5), one of the fifteen errors noted by Lewis when he returned the proof for preliminary gathering A. If H, I, and N had followed A, then it is reasonable to assume that they too would have been corrected at press, especially since number 3 worked on A as well as H and N. And not only must P and A therefore come after the three but very probably the invariant O, a sheet worked by 6, who, we will observe, is lately assigned to this job, apparently for the purpose of hurrying it up.

Taking these several factors into account, we are obliged to assume that the first issue was overprinted for three sheets (H-I, N), then reduced to the original quota of copies for three more (O-P, A) in order to insure the publication of the original issue on the scheduled date. After this date there were a number of other improvisations in the making of subsequent issues, all of which may be represented in a formulary.

               

Issue	1	1-3	1	2-3	3	3-4
Sheets	B-G,K-M	H-I,N	O-P,A	B-E,G,K-L,O-P,A	i, Q-R	a,F,M
Copies	1000	2000	1000	1000	500	1000
Category	impressed			reimpressed	impressed
Issue	4		4
Sheets	i,B-E,G-L,N,O-R		HI
Copies	500		500
Category	revised and reimposed		reset

Should any more be needed, a measure of proof for the position of H-I, N in the sequence machined for the first issue can be inferred from the construction of subsequent impressions. For the third issue ("Second Edition") sheet A was discarded, four new sheets prepared (i, a, Q-R), and the remainder of the second issue combined with these to form a complete copy. As all the type for the first issue, excepting H and I, was still standing at the time the fourth issue (also titled "Second Edition") was ordered, it was corrected according to the author's latest desires, then machined by a new group of men. Why was it necessary to reset H and I?[37] If these two sheets had come in their proper sequence during the impression of the second issue, they would have been available for the fourth. But as they obviously were not available, the conclusion is inescapable that they were printed simultaneously with the first and at the beginning of the second issue, then distributed before the decision was made to retain the type of the second for a fourth reimpressed issue.

Cast-off copy. Normally an allocation of copy among several shops can be easily detected by variation in the headlines, font of type, type-measure, or paper. But occasionally these differentiae are so minute that they escape notice unless attention is directed to them by evidence of a more conspicuous nature, such as that provided by the figures. These may show a division whenever they appear in one portion of the text, but not in the second, or, less frequently, when they are of one kind in the first portion, and of another in the second. Why the former pattern should be almost invariably represented is inexplicable.[38] Some few copies might be considered as the joint product of two shops, one of which did not use figures; but the presence of numerous exemplars requires another explanation. Whatever the reason, the figures in the following editions signify what other evidence substantiates as a distribution of work.

                         

	Shop 1: figures	Shop 2: figures
Burnet, Reflections on the Relation of the
English Reformation (1688)
"A" edition (40, A-M4 N2) . . . .	A-K *	L-Nτ
"B" edition (40, A-H4) . . . . . .	A-Dτ	E-Hτ (different font)
Ellis, The Protestant Resolved (1688)
First Edition . . . . . . . . . .	B-F	G-N, Aτ *
Second edition . . . . . . . . . .	B-F	G-L, Aτ *
Johnson, An Account of the Life of Mr
Savage (1744) . . . . . . . . .	B-S 1, 2, 3	T-2B, A
Lyttelton, Dialogues of the Dead (1760)
Second edition . . . . . . . . . .	A-L [various]	M-X
Third edition . . . . . . . . . .	A-L [various]	M-X (N figured)

         

The Spectator, vol. V (cf. appendix) .	B-E §	F-2F
Tonson (ed.) Poetical Miscellanies: The
Sixth Part (1709)[39] . . . . . . .	A, E-M, Q-2S	B-D *
		N-P τ 3
	4A-3B *§

The figures alone, however, are not an infallible criterion. Even though they are grouped in one portion of the text, they do not indicate divided copy in Gay's Fables, as we have seen, nor do they suggest, in the absence of other and more reliable evidence, what has happened in Mason's Caractacus (1759). Here they appear in the last six gatherings only, but from their presence nothing can be deduced.[40]

Cancels. One obvious use for the figures which has been recommended in theory, though, so far as I know, never put into practice, relates to the detection of cancels. As Mr. Chapman has observed, if two figures are present in a forme, one is presumably upon a leaf supplied from another sheet.[41] In each of the following works cancellation may be suspected from the reduplicated figures, and is immediately confirmed by corroborative evidence.

       

	Figures (cancel in italics)	Collateral proof
Burke, An Appeal from the New to the Old Whigs
(1791) 80 [42] . . . . . .	C2v(o)3 C5 (0)1	chainlines, stub after C5
	G7 (o)8 G8v(0)1	chainlines, G8 (signed 'G7')

       

Burke, Speech on Conciliation
(1775) 40 . . . . . . .	C2v(0)5 C4v(o)5	chainlines
Cumberland, The Wheel of
Fortune (1795) 80 . . . .	A5(o)x A8 v(0) *	[wove paper] 'A8'

Other cancels in Burke's Appeal for which the figures provide no clue, however, are B5 (signed) and D8 (signed 'D7'). If there were no signature on the first of these to signify cancellation, it might be inferred, nevertheless, from the absence of a figure. All sheets except B have in their uncancelled state, presumably, two figures; B departs from the pattern by having only one. And as this is on the inner forme (B8-§), the other was very probably on the outer forme of B5 in its original state.

Where there is, as in the Appeal, an evident pattern of figures, any irregularity should be regarded with as much suspicion as duplicate figures. In Johnson's Thoughts on the Late Transactions, to cite again an oft-quoted example, the missing figure in sheet K of the first impression (see p. 183) is doubtless explained by the existence of a cancel at K2. Similarly in Edward Young's Centaur Not Fabulous (1755), an octavo, various deviations from customary practice are observed: the register of signatures through $4 for all sheets, except N2 and R2; and the insertion of a single figure for each sheet, except T and 2C, which have two, and E and U, which have none. As the copies examined have all been tightly rebound, corollary evidence is not easily obtained; but the occasional presence of stubs in some exemplars allows a provisional conclusion that this book contains at least five cancels—N1.2, R1.2, and T8 or U1—and probably has several more, or perhaps a complete resetting, in sheets E, T, and 2C.

Like the grouping of figures discussed in the previous section, the appearance or disappearance of figures is not always a certain sign of abnormality. To the several instances of this in which cancellation is confirmed, we may add several more which seem to indicate a cancellation that has not occurred. The first, Chesterfield's Vindication of the Case of the Hanover Troops (1743, 80 in half-sheets), has a figure 1 entered at D3v of the initial issue, the same 1 and another at D4v of the reissue (reimpression).[43] As this

pamphlet was printed by half-sheet imposition, requiring only one figure to a forme, the presence of a second calls for explanation. And the only plausible one is that upon the return of the forme for a reissue, the pressman, perhaps realizing that he was subject to a fine for working without figures,[44] looked about for the symbol of the previous man, and not finding it, inserted his own at D4v. Had he looked on the other side of the forme, we may presume, he would not only have found the figure, but discovered it to be his.

Another appears in a cheap reprint of Charles Shadwell's Fair Quaker of Deal (1769, 120) where, curiously enough, two figures, a 2 and a 4, are found on B11. Quite possibly—my conjecture rests on a single copy (NcD)—one man replaced another at press and failed to remove the figure belonging to his predecessor. A third example, evident in Fielding's Voyage to Lisbon, is more appropriately considered in the next section.

Imposition. When the format of a book is for any reason subject to various interpretations, the figures should provide a clue to the disposition of the pages within a given forme, and thus identify the process employed. For the normal method of half-sheet imposition, in which all the pages for a signature are placed within the same chase, the process, involving the use of a single machine, is revealed by a single figure in each gathering. For the alternate method, requiring an arrangement of two successive outer formes in one chase, the corresponding inner formes in another, the process, involving two presses or two distinct operations on the same press, would be occasionally signified by two figures, both of which will appear in one of the half sheets, none in the second, or one in a certain forme of the first, the other in the obverse forme of the second. Again, as with all of my remarks concerning the figures, these generalizations have exceptions. One, Chesterfield's Vindication, has already been mentioned as an instance of accidental duplication in the figures. Another, possibly, occurs in Fielding's second-printed, first-published edition of the Voyage to Lisbon (1755), a duodecimo with its final gathering

imposed as a half-sheet and figured N3v(i)7, N4v(o)2. Here we would expect, for one gathering, imposition according to the first method, but find, apparently, the arrangements required by the second. Perhaps, if the figures are a reliable indication, the second half-sheet produced was used in another book.

In duodecimoes the manner of imposition can be determined, in most instances, by the position of the figures. Normally these will be found only at the most convenient point of insertion, i.e., at the exposed foot of unsigned pages at either end of the forme. If the pages are, conveniently, of an outer rank, but so positioned that their headlines are exposed, the feet of these pages are as inaccessible as those for the inner rank, and thus remain unfigured. To illustrate the arrangement of the pages in the impositions usually described (a and d below), possible variations in these arrangements, and the probable location of the figures in all varieties, I present the following data for the outer formes: Practically all eighteenth-century books of duodecimo format were imposed as described under (a); some few, as for example those discussed below, have sheets accidentally imposed, it would seem, in the manner of (b) or (c); but none, so far as I have observed, were imposed as indicated by (d).

One instance of disarranged imposition appears in a reprint of

Thomas Southerne's Oroonoko (printed for C. Bathurst and others, 1776), in which sheet B is aberrantly signed through $6 (not $5, as normally) and figured 7v(i)3, 10v(o)2. The added signature as well as the location of the figures suggests (b) as the manner of imposition. Very probably the rank of leaves 5—8 was not discovered to be misplaced until the outer forme of B had been partially machined. Then, to avoid wastage, the inner was imposed in the same manner, and a signature added to $6 so that the binder would be certain, whatever forme was up, to sect the lower and not the upper third of the sheet.

Other instances of irregularity, but only in the location of the figures, are implicit in Mandeville's Fable of the Bees. Part II (1730),[45] where an asterisk is inserted at M10v, and in Smollett's Humphry Clinker (second edition, 1771), where there are the following discrepancies from the normal pattern: F11v(i)4, F10v(o)7; H8(i)4, H9(o)7; and M4v(o)7. All of these except F conform, again, to (b); and F, apparently, to (c). For these the evidence of aberration becomes less certain, of course, when it rests on the figures alone, and may be explicable on other grounds at present undisclosed. (Outer 11 and 12 would be conjoined whether imposition is [a] or [c].)

One example of the confusion that may result when the figures are abnormally placed in a normal imposition appears in the first-published edition of Fielding's Voyage to Lisbon (1755), a book of many bibliographical mysteries, not all of which have as yet been fathomed. In this all gatherings have two figures, one for each forme, except E, F, and G, each of which has three: If the figures in the inner rank were not duplicated in the outer, we might suppose, as for Humphry Clinker, that the first two sheets are instances of (b), the third of (c). Since the figures are reiterated, however, and since the chainlines for all leaves are contiguous with those for their normal counterparts, we must discount, not only this possibility, but another pertaining to

cancellation. A third conjecture, that these three gatherings were imposed by 4's and 8's to dispose of split sheet remainders, must also be dismissed; for if this were the procedure, then F would have two figures on the inner four-leaf segment to represent the work of a single press, and only one on the outer eight-leaf segment to represent the work of two. Thus we are perforce reduced to the comment that these figures must reflect some irregularity, not in the production of the book, but in the assignments for its production. Perhaps two agents were involved, the compositor, who accidentally inserts the figures in the middle rank of pages as he imposes them, and the pressman, who fails to discover them in the outer ranks, where they belong, and therefore figures again, number 1 at E12v, number 2 at F5v and G8.

Though the figures raise, in the Voyage to Lisbon, an intriguing little problem which may not be decided to the satisfaction of all, they do provide the means for solving one of much greater consequence. Excluding those in the inner rank of E, F, and G, which are demonstrably superfluous, and those appearing in I and K, which ambiguously identify either (a) or (d) imposition [I7v(i)3, I12v(o)7; K7v(i)3, K12v(o)2], the ones remaining are invariably disposed in locations accessible only for imposition with cutting (a). The watermarks, however, seem to indicate the alternate method without cutting (d). These are located, in each gathering, at the outer margin, half-way down the page, and overlapping leaves 9 and 10, their approximate position for the latter method. Between the two kinds of evidence we must choose, I am now convinced, that afforded by the figures, and explain the other as uncommon paper of foreign make, probably Genevan, with the watermark in the middle of the sheet.[46] In this unusual position it

appears exactly at the location described for the sheets of the Voyage to Lisbon and verifies the imposition which the figures certify as (a). Thus the information to be derived from the figures is completely justified and leads to a conclusion which might not have been forthcoming had it been ignored.

From these several demonstrations of the value of press figures in various phases of bibliographical analysis it is obvious that they are entitled to consideration whenever they appear in eighteenth-century books. In many instances they constitute the only convenient procedure for disclosing hidden variants, "presumed editions,"[47] and sophisticated or mixed copies,[48] the only practicable method for detecting and deciding problems at present unknown or unresolved, the only expedient means for selecting and organizing the material to be studied in any investigation, and for describing the exact nature of that material in the bibliographical record. Their presence in the record may not always be informative, but their omission must inevitably deny the scholar the opportunity of confirming the research of others and seriously inhibit his own.