University of Virginia Library

Search this document 


collapse section 
collapse section 
collapse section 
collapse section 
collapse section 
collapse section 



The word regularly used by printers is “composition” (hence the term “compositors” for those who “compose” the type); but I shall use “typesetting” here in order to avoid any confusion with another meaning of “composition,” authors' composing of their writings (since some descriptive bibliographies—such as Richard L. Purdy's 1954 Hardy, to cite a notable instance—do report on textual history and the manuscripts that precede the printed books). As for the persons who set type, I shall use “typesetters” and “compositors” interchangeably.


A similar comment appears in his discussion of running-titles, where he says, “This checking [of every running-title] is a necessary part of the analysis of the printing of a book which a scrupulous bibliographer should make before attempting its description” (p. 187). Paul Needham, too, lamenting the fact that the study of incunables often seems a separate field from the study of later books, has encouraged incunabulists to include in their descriptions—even the compressed ones in catalogues—more details that would link physical evidence in books with the analysis of their production; he emphasizes “the importance of organising bibliographical data in a way that responds to the reality of book production” (p. 52). See “ISTC as a Tool for Analytical Bibliography,” in Bibliography and the Study of 15th-Century Civilisation, ed. Lotte Hellinga and John Goldfinch (1987), pp. 39-54. On the role of analysis in descriptive bibliography and in its historical development, see also my “A Description of Descriptive Bibliography,” Studies in Bibliography, 45 (1992), 1-30 (esp. pp. 10-13), reprinted in Literature and Artifacts (1998), pp. 127-156 (esp. pp. 136-139); and my introduction to the 1994 paperback printing of Bowers's Principles.


In my Sandars Lectures of 1997 (not yet published), I attempted a general historical introduction to this field. Some historical overview is presently available, among other places, in my “Physical Bibliography in the Twentieth Century,” in Books, Manuscripts, and the History of Medicine: Essays on the Fiftieth Anniversary of the Osler Library, ed. Philip M. Teigen (1982), pp. 55-79, and “The Evolving Role of Bibliography, 1884-1984,” in Books and Prints, Past and Future: Papers Presented at the Grolier Club Centennial Convocation, 26-28 April 1984 (1984), pp. 15-31. The fullest theoretical underpinning of analytical bibliography is supplied by Fredson Bowers's Bibliography and Textual Criticism (1964), and the most detailed exposition of analytical techniques can be found in Charlton Hinman's The Printing and Proof-Reading of the First Folio of Shakespeare (1963)—which should be supplemented by Peter W. M. Blayney's The Texts of KING LEAR and Their Origins (1982), for its corrective commentary and its consideration of the printing of quartos. The most extensive listing of the writings on analytical bibliography is in my Introduction to Bibliography: Seminar Syllabus (Book Arts Press; 1996 revision), pp. 169-198.


Many of these characteristics were of course unintended by the producers of an edition, and some bibliographers have been misled by the term “ideal copy” (as the goal of a description) into thinking that unintended features, or “errors,” are not part of what a description aims to record. But the concept of “ideal copy” properly covers all features of books that existed at the time those books left their producers' hands; it excludes only the characteristics of individual copies of books that result from what has happened to those copies in their subsequent life. See my “The Concept of Ideal Copy,SB, 33 (1980), 18-53. (An earlier article of mine, “The Use of Type Damage as Evidence in Bibliographical Description,” Library, 5th ser., 23 (1968), 328-351—which is drawn on below and which I still regard as valid in general—was written before I had arrived at this understanding of “ideal copy,” and therefore I no longer support its comments on that concept.) The evidences of the post-publication life of individual copies are not without interest, of course, for the study of the reception and use of books, and such details can be recorded or referred to in the paragraph of a description that identifies the specific copies examined.


An example is Laurie E. Maguire's “The Rise of the New Bibliography,” in her Shakespearean Suspect Texts: The “Bad” Quartos and Their Contexts (1996), pp. 21-71, 343-358. She finds W. W. Greg a “xenophobic” scholar whose “author-centered approach stems not just from imperialist attitudes but from Victorian religious piety and sentimental family values” (p. 64). Whether or not these points are accurate—or whether Greg and his colleagues made excessive claims of scientific rigor—has nothing to do with an evaluation of the actual techniques developed by the New Bibliography for the examination of physical evidence.


The most famous example is D. F. McKenzie's “Printers of the Mind: Some Notes on Bibliographical Theories and Printing-House Practices,” SB, 22 (1969), 1-75. My response to this essay can be found, among other places, in “Bibliography and Science,” SB, 27 (1974), 55-89 (reprinted in Selected Studies in Bibliography [1979], pp. 1-36) and in “Issues in Bibliographical Studies since 1942,” in The Book Encompassed: Studies in Twentieth-Century Bibliography, ed. Peter Davison (1992), pp. 24-36 (esp. pp. 29-31). Fredson Bowers had in fact already given an answer in his 1964 book Bibliography and Textual Criticism.) Constructive discussion regarding responsible ways of proceeding with bibliographical analysis is also provided by Peter Davison in “Science, Method, and the Textual Critic,” SB, 25 (1972), 1-28, and “The Selection and Presentation of Bibliographic Evidence,” Analytical & Enumerative Bibliography, 1 (1977), 101- 136.


This University of Wisconsin dissertation, which is no doubt included (as it ought to be) in the microfilm collections of a number of research libraries, is readily available to anyone who wishes to order it, in either microfilm or xerographic form, from University Microfilms International of Ann Arbor (the order number is 81-20339). Some of Vander Meulen's later publications, growing out of the research for this bibliography and its projected revised and expanded form, show how descriptive bibliography, analytical bibliography, textual study, and literary history all intertwine; see, for example, “The Dunciad in Four Books and the Bibliography of Pope,” Papers of the Bibliographical Society of America [PBSA], 83 (1989), 293-310. I wish to record here my indebtedness to Vander Meulen for his invaluable suggestions on the present paper; but he bears no responsibility for any errors that remain—or for my comments on his work.


But see note 13 below. Another questionable use of recognizable types has to do with presswork analysis: bibliographers have sometimes used the reappearance of particular types as a guide to the elapsed time between the distribution of one forme and the setting of another and thus (given average speeds of typesetting and printing) to the number of copies in the edition. Such calculations obviously do not take into account other jobs simultaneously in progress, and therefore they are not reliable.


Before Hinman, W. H. Bond had suggested the possibility that Elizabethan printers might have set by formes, but he did not use evidence from damaged type (“Casting Off Copy by Elizabethan Printers: A Theory,” PBSA, 42 [1948], 281- 291). Hinman's basic article is “Cast-Off Copy for the First Folio of Shakespeare,” Shakespeare Quarterly, 6 (1955), 259-273; his fuller treatment of the use of identifiable types is in his Printing and Proof-Reading (see note 3 above), 1: 52-138. (See also Jesse C. Mills's summary of John Cook Wyllie's Rosenbach Lectures, “Detective in the Book World,” Graphic Arts Review, 23.5 [May 1960], 7-8, 46-48; and Robert K. Turner, “Reappearing Types as Bibliographical Evidence,” SB, 19 [1966], 198-209.) I have surveyed some of the ensuing work on setting by formes and have tried to show how it makes a contribution to printing history, in “Analytical Bibliography and Renaissance Printing History,” Printing History, 3.1 (1981), 24-33, reprinted as “Analytical Bibliography and Printing History” in Literature and Artifacts (1998), pp. 291-306. Sometimes bibliographers, including Hinman, having established that a book was set by formes, try to determine the order in which formes were set by noting the reappearances of particular types. But this use of identifiable types is risky because it is likely to involve assumptions (e.g., that the types used in one forme would not be distributed in time to be available for use in the one set next) that do not allow for concurrent work on different jobs. For an essential and salutary supplement to Hinman, taking up such matters as this, see Blayney's Texts of KING LEAR (see note 3 above), esp. pp. 57-58, 91-94, 176-177. Blayney has also provided, in The First Folio of Shakespeare (written to accompany a 1991 Folger Library exhibition), a very clear introduction to compositor analysis and the significance of setting by forms (pp. 9-14). (A list of the literature can be found in my syllabus [see note 3 above], pp. 144-145.) As Lotte Hellinga has pointed out, Henry Bradshaw had suggested, as early as 1877, the possibility of setting by formes in fifteenth-century printing, and Konrad Haebler discussed the matter in 1925; her own “Notes on the Order of Setting a Fifteenth-Century Book,” Quaerendo, 4 (1974), 64-69, depends not on type damage but on the correlation between the markings in a printer's-copy manuscript and the techniques for compressing and expanding text evident in the printed book.

Of course, determining the manner of setting (seriatim or by formes) does not in itself provide direct evidence of which side of the sheet was printed first (a matter taken up below in section c under “Presswork”). Similarly, the fact that the earliest printed books (before roughly 1475) were routinely printed one page at a time (not by imposed formes) does not theoretically rule out the possibility that setting could sometimes have proceeded by formes (i.e., sides of the sheet). But clearly the standard rules for using identifiable types to determine setting by formes (as outlined in the next paragraph) are not applicable to books printed one page at a time; for if types could be distributed and reused immediately after the printing of each page, identifiable types could reappear on any page, regardless of whether the order of setting the pages were seriatim or according to the side of the sheet (though if a compositor set a new page while printing was taking place, types could not appear on both pages of any pair that was set consecutively). (On single-page printing, see also notes 24 and 37 below.)


Although some bibliographers have proposed, as an additional motivation, that setting by formes minimized idle press time, the argument is not convincing because efficiency of press use would not have been dependent on the handling of a particular book but rather on the management of all the jobs simultaneously in progress in the shop. If, however, there was some cause for urgency in finishing a particular book as expeditiously as possible, setting by formes would be a help by decreasing the time needed for typesetting before the first forme could go on the press. (Concurrent printing on two or more presses would also be a help and, if accompanied by concurrent setting, would be another benefit of setting by formes.)


On the general question of the selection of evidence for publication (which is an issue relevant to all the categories of evidence to be considered below), see also Peter Davison's “The Selection and Presentation of Bibliographic Evidence” (see note 6 above), which in addition takes up the dangers of selectivity in examining the evidence in a single copy and in deciding on the number of copies to be examined. On the latter point, see David Shaw, “A Sampling Theory for Bibliographical Research,” Library, 5th ser., 27 (1972), 310-319.


“Font Analysis as a Bibliographical Method: The Elizabethan Play-Quarto Printers and Compositors,” SB, 43 (1990), 95-164; “Bibliographical Methods for Identifying Unknown Printers in Elizabethan / Jacobean Books,” SB, 44 (1991), 183- 228. Weiss shows his method in operation in “Shared Printing, Printer's Copy, and the Text(s) of Gascoigne's A Hundreth Sundrie Flowres,SB, 45 (1992), 71-104. He had earlier explained the inadequacy of facsimiles for certain kinds of detailed typographic research, such as demonstrating the use of standing type and identifying some running-titles, because in facsimiles one cannot see the actual type impressions but only the total ink smears and therefore cannot identify type damage with sufficient precision: “Reproductions of Early Dramatic Texts as a Source of Bibliographical Evidence,” Text, 4 (1988), 237-268. See also his important review of W. Craig Ferguson's Pica Roman Type in Elizabethan England (1989) in PBSA, 83 (1989), 539-546. As for the earlier period, Paul Needham, in “ISTC as a Tool of Analytical Bibliography” (see note 2 above), has pointed out that—despite the great amount of work done on fifteenth-century types—very little effort has been made to distinguish fonts “derived in full or in large part from a single set of punches, but used by more than one press” (p. 49).


Though Hinman, in Printing and Proof-Reading (see note 3 above), did find evidence for associating the First Folio compositors with type- cases, and Paul Werstine has continued this line of investigation in “Cases and Compositors in the Shakespeare First Folio Comedies,” SB, 35 (1982), 206-234.


Although, as Davison suggested in 1977 (see note 6 above), one cannot rule out the possibility that a compositor moved from one shop to another—or, for that matter, that a font may have so moved. Weiss's 1991 article (see note 12 above) makes some comments on the usefulness of font analysis in discovering shared printing between shops (pp. 188- 190).


David Vander Meulen makes this point, in “The Dunciad in Four Books and the Bibliography of Pope” (see note 7 above), when he says that bibliography serves us well “simply by filling out our understanding of the production of particular books and providing us with further evidence of the customs of a printer or a period” (p. 306).


There are, of course, other compositorial customs that reflect the conventions of particular times and places more than they do the habits of individuals—such as the choice and placement of signatures, the pattern of catchwords, the style of pagination, and the form of imprint dates. Such matters are discussed in R. A. Sayce's “Compositorial Practices and the Localization of Printed Books, 1530-1800,” Library, 5th ser., 21 (1966), 1-45 (reprinted, with additions and corrections, as a pamphlet in 1979); he suggests that “standard methods of bibliographical description should be modified to include more detailed and precise information about some of the features discussed here” (p. 45). An expansion of this approach to localization has been taken up by Frans A. Janssen, who also reviews the reception history of Sayce's work, in “Layout as Means of Identification,” Quaerendo, 25 (1995), 46-58.


D. F. McKenzie has shown that compositors sometimes placed blocks in their composing sticks to shorten the measure for verse lines, and Anthony Hammond has demonstrated that this practice caused compositors' habits in some verse lines to be uncharacteristic in the same ways as in justified prose lines; see McKenzie's “`Indenting in the Stick' in the First Quarto of King Lear (1608),” PBSA, 67 (1973), 125-130, and Hammond's “The White Devil in Nicholas Okes's Shop,” SB, 39 (1986), 135-176 (esp. pp. 137-142). Even in justified lines it is sometimes possible to recognize different compositors' habits by noting whether adjustments in the spaces between words are usually made in the first or second halves of lines; see Herman Doh, “Compositorial Responsibility in Fortune by Land and Sea, 1655,” Library, 5th ser., 29 (1974), 379-404.


One example of compositor analysis applied to a nineteenth-century book occurs in Fredson Bowers's edition (1965) of Hawthorne's The House of the Seven Gables (see pp. li-lv, lvii, 393-398); this example is not comparable, however, with the usual analyses of compositorial stints in sixteenth- and seventeenth-century books because here there is a printer's-copy manuscript with compositorial stints indicated, and the analysis is to see whether the printed pages corresponding to those stints exhibit any marked differences. One reason that compositor analysis is less fruitful for books of this period is that the increasing influence of publishers' house-styles (accompanying the rise of large publishers) makes the practices of individual compositors more difficult to detect.


Hinman, as usual, provides a basic introduction in Printing and Proof-Reading (see note 3 above), 1: 178-226; and Blayney, as usual, in his Texts of KING LEAR (see note 3 above), offers an incisive cautionary supplement, pp. 151-187. Blayney's discussion, indeed, is an important assessment of the whole undertaking and a useful demonstration of a helpful way of laying out the evidence. See also Alice Walker, “Compositor Determination and Other Problems in Shakespearian Texts,” SB, 7 (1955), 3-15; T. H. Hill [i.e., Howard-Hill], “Spelling and the Bibliographer,” Library, 5th ser., 18 (1963), 1-28; and the many articles listed in my syllabus (see note 3 above), pp. 137-143. Slight differences in the width of the type-page (or column) can serve as a preliminary (and, by itself, inconclusive) guide to sections of text set by different compositors; see Bowers, “Bibliographical Evidence from the Printer's Measure,” SB, 2 (1949-50), 153-167. Another kind of evidence that can contribute to compositor identification is turned types; see Robin Dix and Trudi Laura Darby, “The Bibliographical Significance of the Turned Letter,” SB, 46 (1993), 263-270.


Hinman's and Blayney's styles of tables for presenting spelling evidence can be seen, respectively, in Printing and Proof-Reading (see note 3 above), 1:369-424, and Texts of KING LEAR (see note 3 above), pp. 160-177. Another example of a series of tables reporting evidence for capitalization, punctuation, speech-prefixes, and spelling is the appendix to Susan Zimmerman's “The Uses of Headlines: Peter Short's Shakespearian Quartos I Henry IV and Richard III,Library, 6th ser., 7 (1985), 218-255. Anthony Hammond, in “The White Devil in Nicholas Okes's Shop” (see note 17 above), uses a compact table (pp. 153-156) in which each vertical column is headed by a number referring to a separate description of the discriminating feature being reported on in that column.


But they can occur later: see, for example, Paul Eggert, “A Cautionary Tale: Stop-Press Correction in The Recollections of Geoffry Hamlyn (1859),” Bibliographical Society of Australia and New Zealand Bulletin, 19 (1995), 267-269.


The first two of these are taken up below (the first involves evidence discussed in sections d, e, and g; the second in section h). As for discovering imposition, one of the techniques ingeniously worked out by Oliver L. Steele involves analyzing the patterns of the rough edges of leaves. See his “On the Imposition of the First Edition of Hawthorne's Scarlet Letter,Library, 5th ser., 17 (1962), 250-255; and “Half-Sheet Imposition of Eight-Leaf Quires in Formes of Thirty-Two and Sixty-Four Pages,” SB, 15 (1962), 274-278. Although the conclusion relates to presswork, it is likely to be reported elsewhere in a description than in a paragraph on presswork: it would probably appear as the format designation in the collation paragraph, and the evidence itself might well go into the paragraph on binding (since the characteristics of the edges of the leaves are an aspect of how the edges were treated in binding). (Other examples of bibliographical analysis of nineteenth- and twentieth- century books are listed in my syllabus [see note 3 above], pp. 157-161.) For similar reasons, I have not attempted in what follows to comment on all the techniques for identifying the format of hand-printed books, though some of the evidence discussed below can be of assistance in that process; a format statement is already a well-established element in a description, and such standard tests as noting the direction of chainlines and the position of watermarks relative to leaves are appropriately reported as part of the examination of paper.


Headlines without such obvious distinguishing features can sometimes be identified by the kind of microscopic analysis Adrian Weiss described in 1990 and 1991 (see note 12 above) or, somewhat less precisely, by the use of a collating machine. Another method, which requires going to the trouble of making transparent copies for superimposition, is explained by Randall McLeod in “A Technique of Headline Analysis, with Application to Shakespeares Sonnets, 1609,” SB, 32 (1979), 197-210.


The absence of recurrent headlines in a half-sheet obviously does not prove anything. See Fredson Bowers, “Running-Title Evidence for Determining Half-Sheet Imposition,” SB, 1 (1948-49), 199-202; and see note 38 below. In the case of early books printed one page at a time, the sheets were generally cut in half before printing (for formats smaller than folio); for some discussion of this point, see Paul Needham, “Res papirea: Sizes and Formats of the Late Medieval Book,” in Rationalisierung der Buchherstellung im Mittelalter und in der frühen Neuzeit, ed. Peter Rück and Martin Boghardt (1994), pp. 123-145 (esp. pp. 127-128). This process of printing on precut half-sheets is obviously very different from either of the two kinds of “half-sheet imposition” (described just above), in which full formes and full sheets of paper were used to produce printed sheets that were later cut in half. The recurrence of identical headlines on pages printed on the same side of a half-sheet is evidence of printing one page at a time—as are the recurrence of identifiable types on such pages (see note 9 above) and a pattern of type-indentations that displays both first- and second-side printing on the same side of the sheet (see note 37 below).


For the early announcements, see Fredson Bowers, “Notes on Running Titles as Bibliographical Evidence,” Library, 4th ser., 19 (1938-39), 315- 338; and two pieces in English Institute Annual 1941 (1942)—Bowers's “The Headline in Early Books,” pp. 185-205, and Charlton Hinman's “New Uses for Headlines as Bibliographical Evidence,” pp. 207- 222. Two classic examples of the advanced use of headlines are Bowers's “An Examination of the Method of Proof-Correction in Lear,Library, 5th ser., 2 (1947-48), 20-44, and Hinman's Printing and Proof-Reading (see note 3 above), 1: 171-178. Criticisms of this kind of analysis can be found in McKenzie's “Printers of the Mind” (see note 6 above) and Blayney's Texts of KING LEAR (see note 3 above), pp. 122-125. The necessity for recognizing that different jobs were simultaneously in progress in the printing shop (a basic element in these criticisms) has been confirmed by Adrian Weiss's demonstration that a pamphlet of Middleton's was a low-priority job produced in discontinuous intervals when it could be fitted into the work schedule for a high-priority job; see “A `Fill-In' Job: The Textual Crux and Interrupted Printing in Thomas Middleton's The Triumph of Honor and Virtue (1622),” PBSA, 93 (1999), 53-73. For a listing of the literature of headline analysis, see my syllabus (see note 3 above), pp. 149, 155; on the related topic of proofreading, see note 50 below. Since the box-rules around the pages of the First Folio (unlike the center rules) were reused as part of the skeleton, they can be analyzed along with the headlines; see John W. Shroeder, The Great Folio of 1623: Shakespeare's Plays in the Printing House (1956), esp. pp. 59-91, and Hinman's review of it in Shakespeare Quarterly, 8 (1957), 219-222. For a more general theory of the bibliographical use of identifiable rules, see Ernest W. Sullivan, II, “Marginal Rules as Evidence,” SB, 30 (1977), 171-180.


Running-titles enter bibliographical descriptions in other ways as well. It has become standard, following Bowers's recommendations in the Principles (pp. 186-192), to transcribe the running-titles and their variants in a separate paragraph (often just after the transcription of the head-title or the listing of the contents). In addition, a comprehensive paragraph on typography notes the style and size of the type in the running-titles, along with associated rules and the distance from the top line of text (see my “The Identification of Type Faces in Bibliographical Description,” PBSA, 60 [1966], 185-202). But these two treatments of running-titles are somewhat different in their focus from the identification of skeleton-formes. Indeed, Bowers's discussion of running-titles in the Principles struggles rather unsuccessfully with the problem of combining transcriptions of running-titles with a record of recurrent headlines, since some of the variations that make a headline identifiable are not readily transcribed and since a listing of recurrences that is subordinated to a record of variants (as in his “preferred form” on p. 192) does not facilitate the recognition of skeleton patterns.


In Blayney's system, a superscript figure indicates a variant of the identifiable headline, and an asterisk announces that the earlier state appears in some copies. When using his system—or, indeed, a numerical system—one should perhaps caution readers that the sequential labeling of headlines according to the order of their appearance in a book is not meant to carry any information about the order in which they were placed on the press.


The Disappointment: A Disappointment,” Library, 6th ser., 6 (1984), 50-60. There would be an argument—especially in dealing with long books containing many nonrecurrent headlines—for using a different kind of symbol (e.g., X) to stand for all nonrecurrent headlines, so that the lower-case letters would always designate headlines that reappear somewhere. At the opposite extreme, if a book shows only recurrent whole skeletons (and never any shifting combinations of individual headlines), there would be no need for designating the individual headlines; the letters (or other symbols) could then signify skeletons as wholes.


An alternative approach would be the reverse of this one: that is, making individual headlines rather than pages the dominant organizing principle. See, for example, Susan Zimmerman's “The Uses of Headlines” (see note 20 above), where a portion of one of her tables reads “Headline #2 A3v, C3v, E3v, G3v, I3v” (p. 250). It seems to me that this system does not make the pattern of headlines as immediately clear as Blayney's does and could not deal as well with long books or complex patterns.


Vander Meulen, who in his Dunciad bibliography (see note 7 above) uses roman numerals to designate different settings of headlines, capitalizes them for rectos, leaving lower case to indicate versos (pp. 82-86); the same record appears on p. 275 of his “The Printing of Pope's Dunciad, 1728,” SB, 35 (1982), 271-285. In the Blayney system, one could use capital and lower-case letters in the same way to show at a glance which headlines were not interchangeable, though if one has page designations at the top of each column this refinement might not be of much additional help. Vander Meulen's system, which involves other codes as well, is worth studying—especially in the form it took in his 1989 PBSA article (see note 7 above)—as an example of an attempt to convey symbolically several kinds of information (such as recurring sub-groups).


For comments on how points were used in some incunables, see Irvine Masson, “Digression on Pinholes and Their Interpretation,” in his The Mainz Psalters and Canon Missae, 1457-1459 (1954), pp. 16-22, supplemented by Kenneth Povey, “Pinholes in the 1457 Psalter,” Library, 5th ser., 11 (1956), 18-22, and Allan Stevenson, The Problem of the Missale speciale (1967), p. 342. See also Paul Needham, “Paul Schwenke and Gutenberg Scholarship: The German Contribution, 1885-1921,” PBSA, 84 (1990), 241-264, which on p. 258 comments on Schwenke's use of pin-hole evidence in his Untersuchungen zur Geschichte des ersten Buchdrucks (1900). Even when points were used to avoid paper slippage, type-page images could be affected by slippage of the frisket mask, and such “frisket-bite” has been commented on in connection with incunables by Walter J. Partridge in “The Type-Setting and Printing of the Mainz Catholicon,” Book Collector, 35 (1986), 21- 52 (see pp. 42-44), and Paul Needham in “Slipped Lines in the Mainz Catholicon: A Second Opinion,” Gutenberg Jahrbuch, 1993, pp. 25-29 (see p. 29). (See also the references in note 72 below.) In the machine-press period, the cylinder machine had grippers to hold the paper in place; the marks left by these grippers are analogous to point-holes for purposes of bibliographical analysis, but no work—as far as I am aware—has been devoted to studying them.


For formats that involved folding the paper so as to bisect its longer dimension, the points were placed in the middle of the two longer sides of the tympan but were set different distances in toward its center, thus ensuring that sheets could only be turned end for end, not side for side; for duodecimo, the points were placed one-third of the way down from the frisket end of the tympan and equidistant from the longer edges of the tympan, so that sheets had to be turned side for side, not end for end.


See Foxon, “The Printing of Lyrical Ballads 1798,” Library, 5th ser., 9 (1954), 221-241, and “On Printing at One Pull and Distinguishing Impressions by Point-Holes,” Library, 5th ser., 11 (1956), 284-285; Povey, “A Century of Press Figures,” Library, 5th ser., 14 (1959), 251-273 (see pp. 261-262); and Maslen, “Point-Holes as Bibliographical Evidence,” Library, 5th ser., 23 (1968), 240-241.


In such cases, only one leaf would be named (the one on which the hole appears), rather than two leaves as in the example above.


It may be that Vander Meulen's use of the percent sign in influential articles like his “The Identification of Paper without Watermarks: The Example of Pope's Dunciad” (SB, 37 [1984], 58-81) has made this symbol well enough established that one should not now propose alternatives to it. But Vander Meulen has written to me that he chose the percent sign only because it was present on typewriters and not otherwise used in bibliographical description. More symbols are now available in standard character sets on personal computers, including some that might possibly be more appropriate (like §) but not normally including the vertical wavy line suggested by John A. Lane in “Arthur Nicholls and His Greek Type for the King's Printing House,” Library, 6th ser., 13 (1991), 297-322 (see p. 318). Vander Meulen now regards Lane's choice as ideal.


“The Optical Identification of First Formes,” SB, 13 (1960), 189-190. He also reported on the incidence of outer and inner first-formes in a sampling of 201 books (totaling 5338 sheets) in “Working to Rule, 1600-1800: A Study of Pressmen's Practice,” Library, 5th ser., 20 (1965), 13-54.


Or each page in the case of those incunables printed page by page in page-number order. (Obviously the pattern of indentations resulting from printing single pages in page- number order could not occur in full-sheet-forme printing.) Paul Needham has noted that Heinrich Wallau as early as 1900 had used this test for identifying incunables printed page by page; see “Allan H. Stevenson and the Bibliographical Uses of Paper,” SB, 47 (1994), 23-64 (p. 27). But R. B. McKerrow, when discussing single-page printing in An Introduction to Bibliography for Literary Students (1927), referred first to the successive printing of individual pages of a given forme; in such a situation, of course, the indentations would be the same as if the whole forme had been printed together (p. 57).


Although not necessarily: a less likely possibility is that two half- sheets were worked together and that there was an interruption in the printing. Furthermore, it should be emphasized that half-sheet gatherings showing no variation in multiple copies cannot be assumed to have been produced by the method in which two half-sheets are worked together, simply because one may not have looked at enough copies to come across one showing the opposite pattern of indentation. See Kenneth Povey, “On the Diagnosis of Half-Sheet Impositions,” Library, 5th ser., 11 (1956), 268-272. See also note 24 above.


The copy numbers cited here (which could be used as needed in other paragraphs of the description as well) would be identified in the paragraph recording all the copies examined—as illustrated in my “A Sample Bibliographical Description with Commentary,” SB, 40 (1987), 1-30.


Other possible situations are obviously part-sheet gatherings and singletons (cancels or inserted leaves). The former require no special accommodation, though the “mixed” category would probably have a different significance, most likely signaling part-sheet imposition in which all type-pages are on the press at one time (work-and-turn). (The part- sheet gatherings resulting from impositions designed for separately gathered offcuts—as when a duodecimo is gathered in alternating 8s and 4s—could similarly be treated like full-sheet gatherings, and a mixed pattern would have the same possible explanations—even in the unlikely instance of the mixture occurring in a single copy of a book, where two companion part-sheet gatherings derive from two different full sheets. Vander Meulen, in an 8/4 situation, links the related gatherings with an oblique line: “I: C/D, G/H, P/Q, R/S; O: A/B; Mixed: E/F, I/K, L/M, N/O” [p. 222].) For singletons, first-forme impressions can be identified as “recto” and “verso,” as Vander Meulen suggests.


In “Working to Rule” (see note 36 above), this example from p. 35. On detecting felt and mould sides, see Allan H. Stevenson, “Chain-Indentations in Paper as Evidence,” SB, 6 (1954), 181-195; see also pp. 33-35 of Paul Needham's “Allan H. Stevenson and the Bibliographical Uses of Paper” (see note 37 above). For books with press figures, Povey also incorporated a record of figures into the same tables, to facilitate investigating any correlations, by adding two more horizontal lines, labeled “Figure (i)” and “Figure (o)” (and including page references, as in “5v4”). (On press figures, see section f below.)


Bowers, in the Principles, did say, “It is no part of the bibliographer's normal responsibility to collate the text of each copy of an edition that he examines in search of textual variants made in press”; however, he immediately added, “when departures from normal printing indicate the possibility of a corresponding upheaval in the text, it is his business to investigate the slightest piece of evidence, even if it be no more than a change in pagination, and to interpret his findings after a minute comparison of the text in the affected formes” (p. 30). The second part of his statement obviously undercuts the first; but logic here yielded to a hesitation to burden descriptive bibliographers with textual collation (as it has often done in other similar discussions).


That is, from the point of view of authorial intention, although there are times when authors consider certain visual features to be parts of their works, which then become mixed-media works. From the point of view of publishers and readers, on the other hand, many or all of the physical features of books may be regularly regarded as integral to the verbal works presented in them, since those features often play a role in readers' responses. For an elaboration of these points, see my A Rationale of Textual Criticism (1989) and the preface to (and many passages in) my Literature and Artifacts (1998).


I have given a fuller historical sketch of the uses of and responses to “points” on pp. 332-335 of my “The Use of Type Damage as Evidence in Bibliographical Description” (see note 4 above). Abuse of points has by no means died out, but an encouraging sign is the sensible approach taken in a recent article on Huckleberry Finn for a popular book-collecting magazine—especially since Huckleberry Finn is notorious for the number of “points” it contains and for the uses to which they have been put. See Kevin MacDonnell, “Collecting Mark Twain: Huck Finn among the Issue-Mongers,” Firsts, 8.9 (September 1998), 28-35.


Because the research underlying an edition and a descriptive bibliography is largely the same, it is sensible, when feasible, for both products to be goals of a single project. Then full reports of certain categories of information can be assigned, as part of a master plan, to whichever of the two they fit most appropriately, with condensed statements on these matters placed in the other volume. There may, of course, be some topics that require full treatment in both, if both volumes are to be reasonably self-contained.


For examples of bibliographies that report textual variants based on complete collations, see Craig S. Abbott, Marianne Moore (1977); James L. W. West III, William Styron (1977); and Stuart Wright, Randall Jarrell (1986) and Peter Taylor (1988).


If duplicate sets of plates were manufactured for a given book, one must keep in mind the possibility that such alterations might be made to the set in current use and not to the second set being held for future use. See Matthew J. Bruccoli, “Textual Variants in Sinclair Lewis's Babbitt,SB, 11 (1958), 263-268; and “A Mirror for Bibliographers: Duplicate Plates in Modern Printing,” PBSA, 54 (1960), 83-88. (See also note 55 below.)


In instances where the typographical variations are the only means of distinguishing the first and second (or any other two) impressions, the list of differences should normally appear in the entry for the first one, with a cross-reference to it in the entry for the second.


For fuller discussion of this point, see my “The Bibliographical Concepts of Issue and State,PBSA, 69 (1975), 17-66. Bibliographers should remember that intra-impression textual variants may have been produced in a manner that was not precisely “stop- press”; as Adrian Weiss has shown (see note 25 above), if the book or pamphlet in question was a fill-in” job, such alterations may have been made during an interruption in the printing of forme while a high-priority job took precedence.


They are also taken up in numerous articles, often in connection with analysis of the proofreading process; for a list of many of these articles, see my syllabus (see note 3 above), pp. 150-152; see also note 21 above. A recent article that discusses the selective recording of stop-press variants (drawing on Blayney's comments in his 1996 revision of Hinman's 1968 The Norton Facsimile: The First Folio of Shakespeare) is Anthony James West's “A Model for Describing Shakespeare First Folios, with Descriptions of Selected Copies,” Library, 6th ser., 21 (1999), 1-49 (see esp. pp. 12-13, p. 49, and the sample descriptions, as on p. 17).


One of the “Press-Variants” paragraphs in Tom Jones (1974) begins as follows (P. 1052): “Vol. I, sig. C10, p. 43.7-8 (55.33) uncorrected: `Reason|cry out' (ViU2, CtY); corrected: `Reason|to cry out' (BM, ViU1).” Although such paragraphs display a more compressed style of press-variant record in the context of bibliographical description, they still seem rather cumbersome (with their unnecessary “sig.,” “uncorrected,” “corrected,” and the like). An example of a still more compressed form is offered by the descriptions in Melvyn and Joan New's edition (1978) of Tristram Shandy, where a typical press-variant entry is “A8a, 6: disasterous|disasterous” (p. 912); one can accept the omission of a page number and the specification of whether the forme is outer or inner as not absolutely necessary, but a decision not to identify the copies that have each reading—while perhaps permissible in a description attached to an edition if the information is supplied in the textual apparatus—is not excusable in a separate descriptive bibliography.


This style is similar to that employed by Arthur Friedman in The Plays of William Wycherley (1979), except that he identifies the forme on a separate line and places the corrected reading first, as in “11.21 (23.78) Women] BL, CU, LU, N, QC; Weomen Bod” (p. 117). Blayney's system (in The Texts of KING LEAR [see note 3 above] pp. 559-629) can perhaps be seen as intermediate between Bowers's and this one, as far as compactness is concerned. Blayney gives the variants in columns, with each horizontal line labeled by a page-line citation (as “1v31”), but he uses a separate line for naming each forme and a separate table for showing which state of each forme is in each copy examined.


Another example of a selective record of shifting types placed within the paragraph on typography can be found in my “A Sample Bibliographical Description with Commentary” (see note 39 above).


Obviously any sheets not mentioned in this selective listing cannot be assumed to be without known variation, just as other variations besides the ones listed may be present in the sheets that are in fact referred to. The requirement stipulated above in regard to variations between printings—that every sheet be taken into account—would be pointless for intra-impression variations unless they were to be listed in their entirety (which, as stated above, does not seem sensible).


When dealing with copies printed from plates, one must also be aware of the possible use of duplicate plates, in which case copies showing damage at any given point were not necessarily printed later than copies without such damage, since an undamaged duplicate set of plates may have been brought into use precisely because the first set was damaged. (See also note 47 above.) As for the use of plate damage in determining imposition, see Oliver L. Steele, “Evidence of Plate Damage as Applied to the First Impression of Ellen Glasgow's The Wheel of Life,SB, 16 (1963), 223-231.


In my “The Use of Type Damage as Evidence in Bibliographical Description” (see note 4 above), I have offered some detailed suggestions for the collating of copies of the same edition (preferably on a collating machine), preparing a check-sheet for use in further checking, examining additional copies (beyond those that have been fully collated), and reporting the results (see pp. 337-351). An example of the usefulness of noting anomalies, even though they are identical in the copies examined, is offered by the original Murray edition of Melville's Typee. Copies with the 1846 title page may have either “Pomare” or “Pomarea” in the first line of page 19. The form that runs throughout later impressions is “Pomare,” and if all the 1846 copies selected for collation happened also to be “Pomare” copies, there might be no way of knowing to check for a possible variant at this point. But if the collator had noticed a nick in the headline rule just above and had recorded it as an abnormality, the discovery would not be a matter of chance but of routine and systematic investigation—for the rule was damaged when the “a” was removed, leaving permanent evidence that an earlier state existed. A photograph of this evidence, along with five other photographs of significant type damage in books by Melville, appears in the abridged reprinting of my article in Journal of Typographic Research, 3 (1969), 259-276 (see pp. 273-276).


Bearing types also sometimes were mistakenly inked and therefore left inked impressions: see Curt F. Bühler. “The First Edition of Ficino's De Christiana Religione: A Problem in Bibliographical Description,” SB, 18 (1965), 248-252 (reprinted in his Early Books and Manuscripts [1973], pp. 307-312), and the articles cited in his footnotes 13 and 14. See also, for good illustrations of inked bearer type, John R. Turner, “The Printing of Trissino's De la volgare eloquenzia, Library, 6th ser., 4 (1982), 307-313 (supplemented by Peter W. M. Blayney in 5 [1983], 175-176); and Susan M. Felch's edition of The Collected Works of Anne Vaughan Lock (1999), pp. lxx-lxxiv.

McLeod's lectures, entitled (when delivered) “Material Narratives,” have not yet been published, and I am grateful to him for letting me read them in their present form. His resourcefulness and ingenuity in discovering and using new kinds of bibliographical evidence have also been directed to other bookmaking activities in addition to presswork. For instance, his lectures make use of the evidence of folded corners of leaves (and the resulting impressions on adjacent leaves) in a copy of Estienne's quarto Hebrew Bible to show that, in the process of pressing and repressing folded gatherings in a standing press prior to binding, the gatherings were not always in their alphabetical sequence (and the turned corners were perhaps intended to signal the disjunctions). When deckle edges are visible in such turned-corner impressions, one may be able to use them to figure the dimensions of the sheet; when imperfections on the platen of the standing press leave marks on the paper, the varying positions of these identifiable marks can enable one to estimate the size of the platen, and the presence of those marks in different books can potentially indicate which ones were bound in the same bindery. McLeod deals with the rolling press for engravings in a published article, “Imagic” (Studies in the Literary Imagination, 32 [1999], 190-215), where he demonstrates, using an accidentally turned leaf-corner (and evidence of a plateedge burr) in a copy of Harington's translation of Orlando Furioso, that in some instances a whole printed folio-in-sixes gathering, folded inside out, was placed in the rolling press in order to add an engraving to one of the pages of its central opening.

Still other kinds of unintended impressions—from both the presswork and the postpresswork stages—can be found. For example, there are the impressions made by types that were accidentally pulled out of the forme (most probably by the inking ball) and landed horizontally on the forme, or by extraneous types or other items that got locked up in, or fell on, the forme: see the examples illustrated in Victor Scholderer's “The Shape of Early Type,” Gutenberg Jahrbuch, 1927, pp. 24-25 (reprinted in his Fifty Essays in Fifteenth- and Sixteenth-Century Bibliography, ed. Dennis E. Rhodes [1966], pp. 106-107); Curt F. Bühler's “Caxton Studies,” Gutenberg Jahrbuch, 1940, pp. 169-176 (see p. 174); Blayney's Texts of KING LEAR (see note 3 above), p. 140; and Roger E. Stoddard's Marks in Books (1985), item 7. And there are impressions made by objects left in copies by the binder—like the impressions of scissors reported by Blayney in The Shakespeare First Folio (1991), pp. 32-33, which would enter a description only in the record of examined copies.


These figures are not the only other marks besides signatures that can appear in the direction line: some late-seventeenth- and eighteenth-century books, for example, have asterisks, daggers, and other symbols on the first page of gatherings as a way of informing the binder of the particular type of paper used, in cases where two issues on different papers were produced. These “paper-quality marks,” as B. J. McMullin has named them, would probably be best recorded in a description in the paragraph on paper. See McMullin, “Paper-Quality Marks and the Oxford Bible Press 1682-1717,” Library, 6th ser., 6 (1984), 39-49. Even though such marks appear on the rectos of the first leaves of gatherings (where press figures rarely occur), they are still sometimes confused with press figures, since these same symbols (rather than numbers) were occasionally used as press figures, especially before the 1720s. (McMullin has also called attention to the occasional use of signing by page numbers relative to the gathering, in “Signing by the Page,” SB, 48 [1995], 259-268.) Another kind of number that might at first be confused with press figures is the “sheet number,” described by McMullin from some late eighteenth- and early nineteenth-century books (a series of consecutive numbers, one per sheet, in instances where each sheet furnishes more than one gathering, and thus more than one signature); see “Sheet Numbers, `Constable's Miscellany' and Irma,Bibliographical Society of Australia and New Zealand Bulletin, 17 (1993), 33-43. (These are to be distinguished from the numbers that W. A. Jackson also called “sheet numbers”— figures indicating the total number of sheets, or sometimes of quires, in a book, found on the title page or last page of some continental books of the sixteenth century; see “Printed Quire and Sheet Numbers,” Harvard Library Bulletin, 6 [1954], 96- 102, 363-374; cf. Library, 5th ser., 16 [1961], 197-201. See also Jeanne Veyrin-Forrer. “Fabriquer un livre au XVIe siècle,” in La lettre et le texte [1987], pp. 273-319 [see pp. 308-309]; and three articles by David J. Shaw: “Early Parisian Editions of the Works of Coquillart,” Gutenberg Jahrbuch, 1976, pp. 213-217 [see pp. 215-216]; “Use of Printed Quire and Sheet Letters in Sixteenth-Century France,” Library, 6th ser., 17 [1995], 311-320; and “Quire Numbers in Books Printed by Antoine Assourd,” Library, 6th ser., 20 [1998], 364-366.) Figures of yet another kind that accidentally survive in the Dent edition of Conrad's The Rescue (in the lower right corner of a fourth of the pages) are imposition figures identifying plate gangs (units of four pages cast together); see Matthew J. Bruccoli and Charles A. Rheault, Jr., “Imposition Figures and Plate Gangs in The Rescue,SB, 14 (1961), 258-262.


See R. H. Griffith, Alexander Pope: A Bibliography, 1 (1922): 155; 2 (1927): 362; and F. B. Kaye's edition of Bernard Mandeville's The Fable of the Bees (1924), 2: 394-395 (cf. McKerrow's review in Library, 4th ser., 6 [1925-26], 110). R. W. Chapman, in “Printing with Figures,” Library, 4th ser., 3 (1922-23), 175-176, was similarly not concerned with the reporting of all figures.


Knotts, “Press Numbers as a Bibliographical Tool: A Study of Gay's The Beggar's Opera, 1728,” Harvard Library Bulletin, 3 (1949), 198-212; Todd, Procedures for Determining the Identity and Order of Certain Eighteenth-Century Editions (University of Chicago dissertation, 1949), pp. 6-38. Todd concluded, near the end of his discussion of figures, that “they should become a part of the bibliographical record” (p. 38).


For an extensive listing of articles on press figures, grouped according to the stand they take on these questions, see footnotes 10 and 11 of my “Press Figures in America: Some Preliminary Observations,” SB, 19 (1966), 123-160; a more recent chronological list is in my syllabus (see note 3 above), pp. 191-192.


The latter point was made by Todd in his dissertation (see note 60 above), p. 10, and others since then have commented on the possible correlation between imposition and pressmen's choices of locations for figures: see, for example, Povey's “A Century of Press Figures” (see note 33 above), pp. 254-256, and my “Press Figures and the Cut-Off in Duodecimo,” Gutenberg Jahrbuch, 1966, pp. 242-246 (which also discusses the conceivable use of figures to shed some light on the physical arrangement of a shop). For a balanced treatment of the “impediments to the interpretation of press figures” and their potential usefulness when approached cautiously, see B. J. McMullin, “Further Observations on the Incidence and Interpretation of Press Figures,” in Writers, Books, and Trade, ed. O M Brack, Jr. (1994), pp. 177-200.


In “Eighteenth-Century Press Numbers: Their Use and Usefulness,” Library, 5th ser., 4 (1950), 249-261.


“Observations on the Incidence and Interpretation of Press Figures,” SB, 3 (1950-51), 171-205; “Press Figures and Book Reviews as Determinants of Priority,” PBSA, 45 (1951), 72-76; “Bibliography and the Editorial Problem in the Eighteenth Century,” SB, 4 (1951-52), 41-55; “Concurrent Printing: An Analysis of Dodsley's Collection of Poems by Several Hands,PBSA, 46 (1952), 45-57; “Patterns in Press Figures: A Study of Lyttelton's Dialogues of the Dead,SB, 8 (1956), 230-235.


In A Bibliography of Edmund Burke (1964), Todd sometimes uses tables giving the page numbers and press figures in vertical columns for several editions of one work (as on pp. 41-42) or for several works (pp. 62-65); these are more convenient for reference than the formidable solid blocks of figures in paragraph style (as on pp. 238-239) and in addition provide the signature letters in the left-hand column. In Todd and Ann Bowden's Sir Walter Scott: A Bibliographical History 1796-1832 (1998), long solid paragraphs of figures are again used but with signature letters attached to the page numbers, as in “T146- 3”; and commentary on the figures is occasionally provided, as in the table on p. 114 summarizing each pressman's assignments in the octavo and quarto impressions of the same edition.


“Variants in Johnson's Dictionary, 1755,” Book Collector, 14 (1965), 212-213. Two variations of Todd's basic notation omit the dash between the page and the figure. J. C. T. Oates runs the signature letters horizontally across the top of his chart, and for each copy of a book (horizontal lines) enters the leaf reference and figure, as “4a7” or “2b8”; see “Notes on the Bibliography of Sterne,” Transactions of the Cambridge Bibliographical Society, 2 (1954-58), 155-169 (see p. 161). Norma Russell, in A Bibliography of William Cowper to 1837 (1963), prints the record in paragraph form and so must include the signature letter, as “B2v10” or “P7v3”; but since she follows the unfortunate practice of letting rectos be inferred when versos are not specified, awkward spaces (no more economical than a superscript “r”) result in such notations as “H6 11,” especially when both formes of a sheet are figured and the signature is not repeated (as “F5 4, 8 5”). For item 84 (p. 64) she uses the basically different system of grouping according to figure (“6 on A3v B11v C12 D5v, etc., and 7 on C12v”).


“The Recording of Press Figures,” Library, 5th ser., 21 (1966), 318-325—from which I have adapted several paragraphs for use in the present discussion.


When the figures do not always appear in about the same locations on the pages, one could use oblique lines in the table for separating figures that appear in the left half from those that appear in the right half of the direction line. In this example the size of the figures has been ignored, but obviously separate rows could be made for 2 and *2, 4 and *4, and so forth. When it is known that half-sheets were produced by the method in which two half-sheets were worked together, the half-sheet section of the table could distinguish inner and outer formes; but it should remain a separate section and not be merged with the full-sheet section, since the figures would not be comparable (an unfigured half-sheet gathering may not have been printed from unfigured formes, for a figure may have been placed in the other end of one or both formes).


There are many summary tables in “Press Figures in America: Some Preliminary Observations” (see note 61 above); and some sequence tables as well appear in “Press Figures in America: The Shop of Thomas Dobson,” in Writers, Books, and Trade (see note 62 above), pp. 201-220.


“Concealed Printings in Hawthorne,” PBSA, 57 (1963), 42-49.


“Register Measurement as a Method of Detecting Hidden Printings,” PBSA, 73 (1979), 484-488.


Paul Needham has suggested that Gutenberg may have conceived the idea of casting two-line units of movable type into solid slugs. See “Johann Gutenberg and the Catholicon Press,” PBSA, 76 (1982), 395-456 (esp. pp. 429-432), and “Slipped Lines in the Mainz Catholicon: A Second Opinion” (see note 31 above); see also my comments in “Printing History and Other History,” SB, 48 (1995), 269-289 (pp. 284-285), reprinted in Literature and Artifacts (1998), pp. 307-327 (pp. 322-323), and the references cited there to the literature of the controversy surrounding this idea.


Such clues should be used even if there is a printer's imprint, since such imprints are frequently ambiguous (as when a printer is identified as “printer and stereotyper”) and since one should in any case try to test every production statement that appears in a book.


Shillingsburg, “Detecting the Use of Stereotype Plates,” Editorial Quarterly, 1.1 (1975), 2-3; Winship, “Printing with Plates in the Nineteenth Century United States,” Printing History, 10 (1983), 15-26 (see “Recognizing Books Printed from Plates,” pp. 22-23). Gaskell's chapter on “Plates” in A New Introduction to Bibliography (1972) is on pp. 201-206.


Or even lines containing missing letters: a type that drops out allows other types to shift, but if the absence of a letter is caused by plate damage the other letters do not shift.


They cannot generally be distinguished by reference to the categories of clues given above, which can show up in plates of all kinds. In photolithographic plates, for example, textual alterations are sometimes misaligned and have a lighter or darker appearance than the surrounding text.


“Offset Slur as Bibliographical Evidence,” PBSA, 70 (1976), 538-541.


Page 58