University of Virginia Library

REVISION IN BIBLIOGRAPHICAL CLASSICS:
“McKERROW” and “BOWERS”
by
David L. Vander Meulen

A major theme of twentieth-century bibliographical investigation and editorial procedure has been the need to locate and examine all forms of a work in order to understand its history. To commemorate the fiftieth anniversary of one landmark in that scholarly tradition, Fredson Bowers's Principles of Bibliographical Description, it seems fitting to turn some bibliographical classics on themselves by applying techniques their authors advocated. Principles itself is an obvious candidate for such consideration; so also is a book with which it was linked from the time of its publication, R. B. McKerrow's An Introduction to Bibliography for Literary Students. Bowers's volume, first published by Princeton University Press in late 1949, has been reprinted a number of times, without any acknowledgment that changes have occurred; McKerrow's, originally published by Oxford University Press in 1927, was reprinted the following year (and often since) with “a few corrections and small additions,” but without any specification of what was altered. Knowing what happened to these works over their lifetimes may at the least satisfy curiosity. If Bowers and McKerrow were right about the usefulness of bibliographical analysis, it may also help us understand better both the books and their authors.

I

The connection between the two books has been noted repeatedly. Bowers himself assumes his audience's familiarity with the earlier work and in his Foreword says that he will not touch on various materials “readily available in McKerrow's classic” (ix). Readers with access to the original dust jacket of Principles would find the connection reiterated. According to the flaps, “Principles of Bibliographical Description is intended to supplement and act as a companion piece to McKerrow's well-known Introduction to Bibliography; it covers a subject only briefly treated by McKerrow.” In reviewing Principles, F. C. Francis determined that “Professor Bowers's book should take its place beside McKerrow on every bibliographer's bookshelf,” and Curt Bühler envisioned similar cohabitation: “Bowers has done for technical description what McKerrow did for bibliographical investigation, and the `Principles of Bibliographical Description' will henceforth hold an honored place alongside `An Introduction to Bibliography for Literary Students.'” In

an account that indirectly pays tribute to McKerrow's lucidity and Bowers's thoroughness, Donald Wing opened his review with a slightly different characterization of the relationship between the books: “Let it be said at the outset that this beautifully printed volume of five hundred pages is not a substitute for McKerrow's classic Introduction to Bibliography. That remains the undergraduate approach to the material. Mr. Bowers has gone on to a graduate seminar where only the initiate can fully follow.”[1]

Given the care that had gone into the preparation of these books, one would anticipate little need for post-publication revision. Each book had received the scrutiny and advice of many consultants, including the greatest bibliographers of the day. McKerrow acknowledged his friends W. W. Greg and A. W. Pollard as having pride of place for assistance, including for reading the entire book in manuscript or proof (vii). In a similar vein, Bowers pointed out “my paramount indebtedness to Dr. W. W. Greg's searching criticisms of a large part of the manuscript” (x-xi), and he dedicated the book to Greg. Bowers also had had the advantage of seeing the 1947 Rosenbach Lectures—by Bühler, James McManaway, and Lawrence Wroth[2]— before they were published (as Standards of Bibliographical Description, in the same year as Principles). Both Bühler and McManaway, meanwhile, acknowledged having had access to Bowers's manuscript earlier, while preparing their own. Their familiarity with Principles and Bowers's acquaintance with their reactions to it enabled him to answer their replies before their responses were actually in print. This he did throughout Principles, but specifically in Appendix III, where he addressed Bühler's concerns.

Nonetheless, readers who have imbibed the lessons of McKerrow and Bowers may be dubious that these books attained either their desired or their final forms on first publication. Suspicions might be reinforced by recollection of alterations in other well-known bibliographical works. Philip Gaskell's A New Introduction to Bibliography was first published by Oxford University Press in 1972 but then reprinted with corrections in England in

1974 and in the United States in 1975—the latter in two different “second printings.”[3] G. Thomas Tanselle and David McKitterick have identified passages in some of Greg's best-known articles that changed between their original appearance and publication in his Collected Papers, and Tanselle has also observed how Bowers added footnotes (but marked them as newly inserted) when his essays were reprinted.[4]

Circumstances directly related to the McKerrow and Bowers books likewise encourage wariness in assuming that they attained their ultimate form on their first appearance. In a publication announcement for Principles, Princeton University Press acknowledged the challenge of getting the text right (and thereby called attention to the possibility that the challenge might not have been met adequately): it noted “the mechanical job of composition involving special type and other unusual problems.”[5] In reviewing the Introduction, George P. Winship noted McKerrow's willingness to modify what he had written earlier: “the quality in Mr. McKerrow's work which most surely inspires confidence,” wrote Winship, “is the matter-of-fact way in which he goes about demonstrating the mistakes in his own earlier Notes.”[6] The earlier article Winship speaks of is McKerrow's “Notes on Bibliographical Evidence for Literary Students and Editors of English Works of the Six-

teenth and Seventeenth Centuries,” written in the autumn of 1913 and published in 1914.[7] McKerrow is indeed forthright about changes he makes for the book: he acknowledges and corrects, for instance, “a bad blunder which I made in the original edition of these notes” (p. 40, n. 3).

Fine tuning after original publication is in fact a frequent characteristic of McKerrow's scholarship as well as of the treatment of his work by others. The “Notes” themselves acquired slight refinements between their periodical appearance and their issue as a separate. (Among the variants, the only clearly preferable reading occurs in the separate, suggesting that—for this instance at least—the separate contains the later state.[8]) McKerrow's first book, Eigo Hatsuongaku, was revised the year after its original publication in 1902.[9] The 1905 printing of The Gull's Hornbook, which he edited, identified itself as “in the main [only], a reprint of that issued last year”; and a cancel leaf was required for his 1907 edition of The History of Orlando Furioso 1594 to correct the spelling of his name (first given as “Robert”) in Greg's note on the back of the title page. When his five-volume edition of Thomas Nashe (1904-10) was reprinted “with corrections and supplementary notes” in 1958, the new editor, F. P. Wilson, noted that McKerrow himself had “printed corrections to his text on p. 484 of volume iv and p. 373 of volume v” and that “McKerrow also added explanatory notes as his edition was passing through the press” (5.373). Even the list by F. C. Francis that records most of these changes includes “revisions by the [new] compiler” when the record is reprinted with a collection of McKerrow's writings.[10]

Opportunities for intentional and inadvertent changes in Introduction and Principles multiplied as the books were reprinted. The accompanying chart lists the impressions that I have identified—fourteen of McKerrow and ten of Bowers. The printings are numbered consecutively at the left of each

column; superscripts indicate the sequence within a family of printings, or within what Bowers calls a “subsidiary edition” or “sub-edition.”[11] The pattern of these reimpressions reiterates the link between the two books that was noted immediately when Principles first appeared: reflecting but also propelling trends in scholarship, these books often have come forth in tandem over the past half century.

Printings of McKerrow's Introduction and Bowers's Principles

                                   

McKerrow, Introduction	Bowers, Principles
1 Oxford1, 1927
2 Oxford2, 1928
3 Oxford3, 1948
4 Oxford4, 1949	1 Princeton1, 1949
5 Oxford5, 1951
6 Oxford6, 1959
7 Oxford7, 1960
8 Oxford8, 1962	2 Russell & Russell1, 1962
9 Oxford9, 1964	3 Russell & Russell2, 1963-64?
10 Oxford10, 1965	4 Russell & Russell3, 1965-71?
11 Oxford11, 1967
12 Oxford12, 1972	5 Russell & Russell4, 1972-75?
13 Oxford13, 1977	6 Russell & Russell5, 1977
	7 St Paul's1, 1986
	8 St Paul's2, 1987
14 St Paul's1/Oak Knoll, 1994	9 St Paul's3/Oak Knoll, 1994
	10 St Paul's4/Oak Knoll, 1998

The publishing history of these two works has not always been clear even to those who have written about them. An article surveying the career of McKerrow half a century after the appearance of Introduction mentions only the uncorrected 1927 impression in its checklist of his works; the latest edition of a renowned guide to literary terms gives only the date of the tenth printing of McKerrow's “classic work on bookmaking and printing,” citing the book as “rev. 1965”; and the publishers of the recent tenth printing of Bowers's Principles identify it as the fifth.[12]

After the corrected impression of McKerrow's Introduction in 1928, that book was not reprinted again until two decades later—almost as if nudged back into public view by the appearance of Bowers's Principles (which was distributed in England by the Introduction's publisher, Oxford[13]). The Bowers book was listed by Books in Print (but identified there as Principles of Bibliographical Criticism) as “in press” in 1948 and 1949, and it was published on 30 December 1949.[14] In the same era (from 1948 to 1951), Introduction underwent three printings. The heyday of both books spanned the 1960s. From 1959 through 1972, Oxford University Press reprinted McKerrow seven times. When Oxford redesigned the jacket in 1959, Principles appeared on the back panel atop a newly added list of “Some OXFORD Books”, where it remained for the printings of 1960 and 1962 as well. Principles therefore seems to have been available in England as late as 1962, but in 1960 it dropped from the American Books in Print. As if now spurred by the popularity of McKerrow, the scholarly reprint firm Russell and Russell reprinted the Bowers book in New York four times in this period, beginning in 1962. Russell and Russell produced a fifth impression shortly thereafter—in the same year (1977) that Oxford released a thirteenth printing of Introduction. The Bowers book reappeared as a new subedition in printings from St Paul's Bibliographies in 1986 and 1987, but McKerrow's Introduction was not further reprinted (perhaps because it was incorrectly perceived to be superseded by Gaskell's New Introduction to Bibliography of 1972) until it and Principles were published as companion volumes by St Paul's and Oak Knoll Press in 1994. Principles was reprinted as part of this subedition in 1998, the same year that a Spanish translation of McKerrow's Introduction appeared.[15] The renewed interest in Principles reflected in the printings of the late 1980s and the 1990s was meanwhile mirrored by a 1991 Japanese book, Akira

Takano's Yōsho no hanashi (“A Story of European Books”). It was “Based on Bowers's Principles of Bibliographical Description,”[16] but by also drawing on McKerrow (directly adopting, for instance, his illustrations of pairs of leaves from pp. 167 and 169, which now appeared on pp. 19 and 21), it reemphasized the natural connection between the two works.

The exact dates of the five known Russell and Russell printings are hard to pin down, but details from the title and copyright pages provide useful clues. The two impressions that seem the earliest are those that have the publisher's city and name in separate lines on the title page and that print the copyright information, including the statement “Reissued, 1962,” at the top of the next page. In the other three impressions the imprint information appears in a single line and the copyright details move to the foot of the page, where they read largely the same except that Russell and Russell is then called “A Division of Atheneum Publishers, Inc.” Atheneum acquired Russell and Russell in 1965;[17] the copies without the ownership declaration therefore seem to have come before that, an inference supported by the fact that one of them has the date 1962 on the title page. This printing is in fact the only one in the subedition to have a date in the imprint, and it generates the surmise that, as the sole impression that can be linked indisputably with the year of the first reprint, it might indeed be that first reprint. Unless hitherto unrecognized printings exist, the other impression without the reference to Atheneum—and hence also implicitly published before 1965— would have been the second printing. At the other end of the series, the latest Russell and Russell printing is almost certainly the one bound in brown cloth (instead of the blue cloth with which the others are covered). Its copyright page again says “Reissued, 1962” but adds “Copyright renewed, 1977, by Princeton University Press”; it presumably appeared in 1977 or shortly thereafter. (I purchased my copy of this printing in November 1979.)

What then must be the intermediate printings of the subedition, the third and fourth of those that have been identified (both of them bound in blue, both with the copyright information at the foot of the page), can be distinguished by whether or not they include an ISBN number on the copyright page. That system for identifying books was approved by the International Organization for Standardization (ISO) in 1970,[18] and in 1973 Russell and Russell first added the number to its entry for Principles in Books in Print. My copy of the printing without an ISBN designation is signed and dated “1971” by a previous owner; the lack of other marks of ownership or

of use may be signs that he bought it new that year. The printing with the ISBN code, on the other hand, was available in December 1975, when Terry Belanger purchased a copy for the Book Arts Press. Those benchmarks make it plausible that the third printing of this subedition appeared between 1965 (the year Atheneum acquired Russell and Russell) and 1971, and the fourth during the years from 1972 through 1975.[19]

A full bibliographical analysis of these books would be interesting and productive, as even a cursory inspection suggests. For example, the volumes and dust jackets of McKerrow's Introduction offer a case study in the design and production methods of the Oxford University Press over half a century, and through their changing list of cities in which the Press had an office they even suggest the course of world events. The book in which McKerrow offers the general rule that “the handsomest edition of a book is the first” and in which he points out that the “tendency of reprints has at all times been toward the saving of expense in production” (p. 184) provides its own evidence for his observations as it gradually abandons paper with chain lines, removes the second color from its jacket, and reduces its height. The latter change—one that underscores McKerrow's point about the benefits that arise from “the comparison of several copies” (p. 230)—occurs during the binding of the fifth printing: the large copies (9″ high) have only their top edge trimmed, whereas the shorter (8⅝″) and presumably later ones, with a higher publisher's price, are smooth on all edges. The original size is partially restored beginning with the tenth printing, though the subsequent volumes never fully regain the height of the early ones. The pattern of dress is similar for Bowers's Principles, whose colored title-page cartouche becomes black in all reprintings and whose original dust jacket diminishes to brown wrapping paper on the Russell and Russell impressions.[20] (Currently both books are sold only in paper covers.) The overall consistency of these patterns makes variations from them suggestive; that the second Russell and Russell printing

of Principles has the thickest paper and finest-grained cloth of the subedition, for example, may hint that this impression rather than the one with the title-page date is the true first printing. The Bowers and McKerrow books also provide other fascinations to the bibliographical eye. It is interesting, for instance, to ponder the industrial procedures that resulted in alternating paper stocks in the 1986 Principles, or to watch the Oxford printers address the problem of how best to impose a book whose leaves in some impositions did not add up to full regular gatherings (the Oxford impressions of McKerrow have been variously gathered in fours, eights, and twelves).

The selective but extensive examination that follows is not intended as a full printing or publishing history; rather, it focuses on a particular physical feature of these books, their texts, in an attempt to answer some basic questions about changes that occurred.[21] Curiosity compels identification of McKerrow's unspecified corrections and additions, and a similar inquisitiveness prompts investigation of whether the printings of Principles that imply similarity with each other really do have the same texts. The incentive for these questions is practical as well: readers need to know whether some forms of the book should be consulted instead of others—and, indeed, whether certain ones should be used as the basis for reprints. Knowing what has been modified also helps in the actual reading of these books. The existence and nature of changes first of all provide assistance in understanding the relative care these authors exercised at various stages (and thereby contribute to assessing their overall credibility). The alterations may also reveal the development of the authors' thought, and those determined to be authorial also signal for the reader elements that likely reflect specially considered judgments. By reiterating to readers the possibility that a text may exist in different forms, this kind of analysis also provides a more general lesson in reading, for it reminds those who come to texts that the need to test all things includes the necessity of assessing critically even the typographical symbols that they have in front of them.

II

The most significant changes that appeared in either book occurred in the second printing of Introduction. The “few corrections and small addi-

tions” that McKerrow speaks of in his “Note to the Second Impression” (dated June 1928) do not approach in density or degree the ones introduced when a section of the book was reprinted in an anthology of typographic writings some years after his death,[22] but in the range of situations that call them forth they quietly exemplify major lessons taught in the book itself. The alterations are in matters of fact, grammar, and orthography, and occasionally in literary style and typographical design as well. Some are likely to have resulted from McKerrow's own contemplation, recognition of errors, and discovery of new information; others may have come from Oxford's printers and editors; and a number appear in response to the observations of reviewers.[23] With one exception (the repair of type damage), I have not found intentional amendments of the text after the second printing, though a number of accidental deteriorations do occur.

The largest and most important group of modifications consists of adjustments to make the information more accurate or precise, either by correcting errors or by providing new facts. In order to focus on this category, it is helpful to deal first with the smaller groups of changes, beginning with those that were not for the better. Several of them occur already in the second impression, which overall is printed from the same setting as the first. A comma vanishes at the end of the first line of note 3 on page 224, and two page numbers undergo alteration: “xi” drops from the preliminaries, and “113” is reset. For “113” the compositor happens to replace the second digit with the letter “r”, a character that remains in all subsequent printings. Throughout those later lithographic reprints, in turn, various changes result from degeneration of the type images. The barely detectable period that concludes the caption on page 149 in the second impression becomes invisible in the third; starting with the third printing, the closing period is absent after the final note on page 310; the letters “os” of “imposition” at 263.7 are missing in the sixth through twelfth printings and in the fourteenth; and beginning with the eighth printing the full stop that is presented as an example at the start of the fourth paragraph on page 316 is gone. The book also provides a rare example of type damage that seems intended. On page 145, the job number “3349” in the new direction line of the third printing is corrected to “3249” in the fourth (continuing through the twelfth), apparently by removing the lower portion of the second digit.

One other instance of the later reading being the inferior one—an example that incidentally serves as a reminder that graphic images are subject to the same vicissitudes as words and punctuation—occurs in Figure 9, on page 35. Two separate but related changes occur in the octavo imposition diagrams here: the inner forme is moved from the bottom to the top of the page, and it is rotated 180°. Whether both modifications were intended or whether one happened accidentally in the course of the other is unclear; either way, they may illustrate varieties of what McKerrow sees as a common problem, the “transformation of sense by the attempt to correct mistakes” (p. 198). The switching of the formes is puzzling, for in his other illustrations McKerrow consistently presents the outer forme first (he depicts the folio on pages 30 and 33, the quarto on 16-17, and the duodecimo on 170-171). Perhaps the exchange is an unintended consequence of revision, though it is not an alteration that obscures sense. The rotation, however, presents difficulties, and there are grounds for thinking that McKerrow himself is responsible.

Both the Introduction and the earlier “Notes” provide some suggestion that McKerrow does not have a firm grasp of the intricacies of imposition and that although his understanding develops it is not yet complete by the time of the book's publication. He increases the difficulty of comprehending these matters, moreover, by rendering the imposition patterns as pages of type lying on the bed of the press rather than as their images printed on a sheet. He recognizes the problem, for amid his repeated injunctions to read-

[Description: SB 52, Page 226]

[Description: SB 52, Page 227]

ers that the easiest way to understand these layouts is to take up and fold sheets of paper he reiterates that those aids to learning will be “looking-glass” images of his diagrams (as on pp. 18 and [beginning with the second printing] 37).[24]

Problems with imposition are first evident in the 1914 article. The folio (pp. 245-246) is portrayed properly, but the layout of the quarto (pp. 228-229) works only if the sheet is turned incorrectly—that is, pivoted on its long edge for the printing of the second forme. The octavo (p. 247) is again all right, with the sheet being turned on its short edge for the iteration. All three of these patterns—two correct, one incorrect—are repeated in the first printing of Introduction (with the diagram for the quarto matching the incorrect verbal description on page 18). The treatment of the duodecimo, however, undergoes changes from the article to the book and reflects development in McKerrow's understanding. In the “Notes” he points out that because “there are several ways in which the pages can be imposed” for small “sizes” such as duodecimo, and because these “sizes are seldom of any practical importance to those who are likely to read these pages,” therefore “It must... suffice here to give a single scheme for 12mo imposition, and to leave readers to work out others for themselves, if they wish to do so” (pp. 248-249). The imposition scheme he chooses to illustrate, duodecimo without cutting, was, to say the least, uncommon; according to Gaskell, “There seems to be no clear evidence” that it “was ever used in the hand-press period” (New Introduction, p. 107). By the time of the Introduction McKerrow's understanding and experience have progressed: though he again presents a diagram for folding without cutting (p. 171), he also includes one showing the more familiar pattern of “twelve-mo by cutting,” saying that “So far as my observation extends, this was the normal, and practically the only, method of producing a 12mo used by English printers of the late sixteenth and seventeenth centuries” (p. 170). McKerrow himself documents what happened between 1914 and 1927. In discussion after a Bibliographical Society talk by R. W. Chapman on 15 October 1923 that was itself a response to some points made in McKerrow's article, McKerrow acknowledges that “In these notes, being conscious that I did not know much about duodecimos but feeling it necessary to say something about them, I worked out the possible ways of folding a sheet in twelve and left it at that.” The “examination of a fair number of duodecimos since I compiled those notes,” he continues, “has led me to the belief that, round about the year 1600 at any rate, the method of cutting off the lower third of the sheet which Mr. Chapman has found practised in the eighteenth century was, if not the only method, at least the usual one.”[25] In the 1927 revamping of the material on duodecimos McKerrow

also makes one other significant change: for the first time he presents imposition diagrams based on “the printed sheet (not the form of type)” (p. 170).

McKerrow's uneasiness with some questions of bibliographical format was noted by at least one reviewer of the book. According to the TLS writer, “Dr. McKerrow illustrates these mysteries of imposition by diagrams, and, though with a momentary lack of courage he remarks `into the question of 24mo. I cannot enter,' he returns to the charge in an appendix.” McKerrow likewise falters on a related matter, the role of the pins or points that hold the sheet on the tympan during printing. His description on pages 22-23 is largely unchanged from the account on page 266 of the “Notes.” In it he fails to acknowledge a crucial feature of their positioning—that, to help the pressmen pivot the sheet in the right direction, they were placed asymmetrically, so that the holes they created would match only for a certain turning of the sheet. A passage he adds in the second printing of the book (p. 37) speaks to the turning of the sheet and does at least indicate his struggle with the complexities involved:

But the account of imposition remains incomplete, for this elaboration does not recognize that the different turnings produce different effects, nor does it specify when each is appropriate.

The uncertainty that characterizes McKerrow's dealing with format culminates in the change of the diagram for the inner forme in Figure 9. What was correct in the “Notes” and in the first printing of Introduction goes awry as the pattern is rearranged to show the sheet being turned across its long side instead of its short one. In various ways McKerrow warns his readers of the “strange muddle” caused when “sheets, after being printed on one side,” are “turned the wrong way round when being perfected” (p. 260; this wording also appears in “Notes,” p. 250). Though through his efforts McKerrow may save others from misunderstanding, on this point he does not save himself.

But most of the alterations in the second printing are appropriate. These involve first of all making consistent a number of typographical details, ones whose variability offers support to McKerrow's observation that “The human body is not a perfect machine” (p. 225). On the final page of the table of contents (xi, line 11), the concluding points are removed from “12mo.” and “24mo.” in order to match the text itself.[26] (This is the page on which the pagination also disappears—perhaps accidentally in the course of deleting

the points.) Excessive letter spacing is reduced between the first “i” and the “f” of “Bassifie” in the inset quotation on 242 and within “i.e.” in the note on 311. On page 340, short equals signs are replaced by ones matching the others in the list of Latin place-names (in the entries for Pisae, Regiomontium, Salmantica, Senae, Spira, and Venetiae).

This attention to typographical nuance finds its most surprising manifestation in two instances where passages have been reset simply for visual effect. In the first printing, the first line of the second full paragraph on page 57 has large gaps between words. To balance its appearance with the following line, a syllable from that one is moved up: “been| described” becomes “been de-|scribed”. In the second full paragraph on 94, lines 3 through 6 all start with a three-letter word, creating a river of white space to their right (three lines in a row begin with the word “the”), and line 3 also is set unusually loose. The revision moves a word from line 4 to line 3: the lineation changes from “below| the title” to “below the| title”, thereby addressing both problems. In a slightly different instance, where space instead is added, the structure of a page's content is clarified when the list of Latin abbreviations on page 324 is lowered to set it off typographically from the preceding prose.

The nature of some of the details needing correction makes it hard to tell whether they originated from the compositor's slips or from the faithful reproduction of the author's errors (which in turn might have resulted from momentary inattention or faulty understanding). Reviewers of the first printing pointed out how “proving” at 14.16 should be “proofing” (Byles, Modern Language Review) and, at 341.12-13, “Facsimilies” ought to be “Facsimiles” (Sisson, Library); both are corrected in 1928. At 161.9, the name “Moretius” becomes “Morelius”. To mark the beginning of a parenthetical statement, a comma is added after the first word of the note on 70, and a colon replaces a period at the close of the introduction to the inset quotation on 122.

McKerrow also uses the occasion of a second printing to make a number of stylistic changes. He converts the preposition in the phrase “puffing titles found on many Elizabethan books” to “in” (93.11), and to reduce repetition of forms of the word cause that appear three times in four lines he alters “which cause his work to fall below” to “which result in his work falling below” (240.8). Some verbal changes that might be considered stylistic are in fact necessary for the sense. In his Library review Charles Sisson pointed out that the phrase “may prevent the reproduction from falling short of our reasonable expectation” at the end of the opening paragraph on 239 made no sense; McKerrow changes the core of it to “from coming up to our”. Elsewhere a revision is necessary because the original syntax is incoherent. The opening of the note on 33 first reads:

Struggling to bend the sentence to sense using words that would occupy the same amount of space, he at least produces a pair of prepositional phrases, however awkwardly they match:

By far the largest category of textual changes for the second printing, however, comprises corrections and additions in the name of accuracy, bibliographical and otherwise. The reasons for certain kinds of modifications in the reprintings are obvious: the date on the title page must change, the information on the copyright page needs to be updated, or the colophon at the end of the book must report a new printer. Those alterations and ones in direction lines are not reported in the list that follows, which unless otherwise indicated cites changes between the first and second printings and assumes that all subsequent impressions follow the second. The adjustments in the preliminaries that were caused by introduction of the McKitterick essay in the St Paul's/Oak Knoll printing likewise are not included here. But all the other changes I have identified but have not yet mentioned are present. Each account begins with the page number on which the variation occurs, and it generally includes some explanation of why the text was altered.

viii. At the end of the Preface, the 1928 printing adds a note retained in all subsequent impressions:

xv. To the list of useful books McKerrow adds an entry after line 26 for one by his friend A. W. Pollard in order to account for a new reference in the note on 265: “— Fine Books. London, 1912.”

1. The opening word of chapter one is altered from “THIRTEEN” to “FOURTEEN”. Because McKerrow is speaking of the number of years since “I put together the `Notes on Bibliographical Evidence'” (“In the autumn of 1913,” according to a parallel sentence that begins the Preface) rather than the years since the article was actually published (in 1914), this change may be the correction of a misstatement rather than a recalculation based on the date of the revision, a year after the 1927 first printing.

34. The addition of “Note IV.” (quoted above) at the foot of 37 requires on 34 the insertion of a cross-referential footnote and the incorporation of a corresponding superscript “2” in the text seven lines from the bottom. The new note, six lines long, emphasizes that the arrangement of pages on the printed sheet is “a `looking-glass image' of the arrangement of type-pages in the forme itself.” The new footnote, inserted at the end of a short existing line of type, reads simply: “2See Note IV on page 37.”

38. Lines 2 through 5 of the second paragraph are reset for greater precision in dating (and better congruence with the date “1530” in the caption on the following page): the phrase “Badius's later block in 1520 (figure 10)” changes to “his later blocks in 1520 and c. 1529 (figure 10)”. In the next sentence, the dependent pronoun switches from “this” to “these”.

39. The following two-line footnote is appended to the caption of Figure 10:

To gain room for the new note, the final two lines of the opening paragraph on the previous page (38) are compressed into one (by dropping the word “also” after “apparently” and by tightening the spacing), thereby providing space for the top line of 39 to be moved to 38. The new note becomes the first one on 39, resulting in the change of the superscript of the page's previously solitary note both as it appears in the text (in the line immediately below the caption) and in the note itself. Stephen Tabor's account of McKerrow in the Dictionary of Literary Biography includes a reproduction (on p. 206) of the first state of page 39. (The on-line form of the article does not include any of its illustrations.[27])

72. In lines 18 and 19, McKerrow corrects information about the history of stereotyping, working the alteration into the existing space. In a change perhaps encouraged by someone's attentiveness to a local slight, “reinvented at Edinburgh in 1781, only to| be again abandoned” becomes “reinvented by A. Tilloch of Glasgow in 1781,| but again abandoned”.

74. The fourth and third lines from the bottom soften an earlier speculation, again adjusting the new wording to fit the available space: “was difficult, if not| impossible, for them” becomes “was naturally very| troublesome for them”.

80. McKerrow adjusts note 2 (which concludes on the following page) about the nineteenth-century American practice of signing plates for multiple imposition patterns. Though he does not fully understand the phenomenon, which lies outside the period of his greatest attention, he nonetheless provides useful evidence—and, in the revision, adds to it. The relevant part of the last three lines of the note on 80 and top five lines on 81 originally reads:

The revised portions are as follows:

93. Charles Sisson's identification in his Library review of earlier instances of advertisements citing authors as the writers of previous works prompts McKerrow to come up with examples still a century earlier. The revised footnote in which he incorporates them is two lines longer than the original one, requiring condensed setting of other passages on the page. The new information also entails slight revision of the text, albeit in situations that require the rewording to occupy the same amount of space as the first version. The initial change, a recasting of the opening two lines of the second full paragraph on the page, proceeds as follows:

A slight modification within lines 6 and 7 of the same paragraph deflects blame for the shortcoming away from Michael Sadleir, on whose helpful survey of his own collection McKerrow based his original remark: becomes Two passages are truncated in order to reduce the text by a line each time. The concluding lines of the second paragraph change as follows: The end of the final full paragraph on the page is likewise altered: The most important change occurs in the footnote. The early form reads: McKerrow's new investigations result in this version:

99. The addition of the two-line note 6 citing a 1906 article on the paper-maker John Spilman requires condensation of other passages in order to gain sufficient space on the page. The note reads: “6For a detailed account of Spilman, see G. H. Overend in Proc. of Huguenot Soc. of London, viii.180-96.” The new information also results in two corrections: of the spelling of Spilman's name (from “Spillman”) two lines from the bottom of the text (a superscript for the new note is also added here), and, two lines above that, of the date he was granted a license (from “1588” to “1588/9”). Verbal alterations and tighter word spacing twice effect the saving of a line of type: in note 3, “by an evident slip” is reduced to “by a slip”, and in note 5 “large quantities| of English rags” is cut down to “English rags”.

102. A newly available article, dated the same month as McKerrow's prefatory note to his second impression, prompts McKerrow to revise his estimate of the futility of studying the location of watermarks on the sheet. His duties as co-secretary (with A. W. Pollard) of the Bibliographical Society, which included “supervision of publications,” presumably would have acquainted him with the article before it was published.[28] The original and revised passages, which appear at the end of the first full paragraph on the page and involve four lines, read as follows:

Heawood's Library article, which begins “In his recently published `Introduction to Bibliography,'” takes as its occasion the 1927 appearance of McKerrow's book. In the course of discussing the position of watermarks in sheets of paper, Heawood also addresses another question McKerrow raises: whether early paper was ever made in sheets double the normal size.

131. The TLS review called attention to a 1909 article that McKerrow should have been aware of. He introduces the reference in the second printing by altering lines 2 through 4 of the first new paragraph on this page; the change results in the resetting of the subsequent twelve lines as well. The two forms of the passage are as follows:

206. The addition of a footnote on 218 citing a recent article by Percy Simpson requires shifting of the text on that page as well as inclusion of a three-line cross-reference at the end of the note on 206 (which begins on 205) and the subsequent adjustment of text in that area. This new material on 206 reads: “Cf. also the important article by Mr. Percy Simpson on `Proofreading by English Authors of the 16th and 17th Centuries' referred to on p. 218 note 1.” To gain the needed space, first the text of note 3 on 205 is squeezed from three lines into two by reducing the opening indentation of each of the three footnotes on 205 from 3.2 to 2.4 mm. (The single-line note 1 and the first line of note 3 are consequently reset, though they read the same.) One line of note 3 is then moved from 206 to 205, and the spacing above and below the quotation immediately above the footnote on 206 is reduced.

218. Information promised in the added note on 206 is provided as a new three-line note 1 on 218 (with a footnote superscript for it inserted at the end of the second full paragraph). That results first of all in the renumbering and repositioning of the single existing note on 218, which now becomes note 2 (with the superscript in the text changing accordingly). To provide the space required for the addition, the four lines at the end of the first paragraph on 217 are reset as three; the top line of 218 is moved to 217; the bottom line of 218 is moved to 219; and the final two lines of the paragraph concluding at the top of 219 are compacted into one. Perhaps to dilute the typographical impact of the insertion over several pages, the title of the article is given only in the new note on 206 and information about the journal in which it appeared only here: “1But see the article by Mr. Percy Simpson in Oxford Bibl. Soc. Proceedings, II. i (1928), pp. 5-14, for some important additions to this list.”

265. In the footnote which begins on 264, McKerrow reduces what amounts to a blurb for a British Museum exhibition catalog in order to mention a source he overlooked earlier, A. W. Pollard's Fine Books (1912). (This is the title McKerrow also added to the list on p. xv.) The original passage (from the end of line 8 through line 12 on 265) and the portions that are new are as follows:

297. Six lines from the bottom of the text, “black letter” is hyphenated in the second, third, and fourth printings. The hyphen may be obscured by accident in the fifth and subsequent printings, but the intentional intervention revealed by the word “black” simultaneously moving slightly left in this lithographic printing suggests that the hyphen was removed by design. On this and the following page, the phrase “black letter” (with and without a hyphen) appears nine times. All four occasions in which the words function as adjective and noun respectively appropriately lack the hyphen; of the four instances other than the current one where they function as a single adjective modifying another noun, three have a hyphen and one (the closest, seven lines later) does not. The present example therefore is inconsistent whether or not it includes the requisite hyphen.

319. McKerrow adds a footnote (and, after “practice” in line 2 of the second paragraph, a superscript “3”) that provides references to the use of contractions in manuscripts, the objects that lie behind the early printed books that are his chief concern:

To provide room for the three-line note, the printer reduces the space between the text and the first footnote, and McKerrow rephrases the end of the second paragraph to save one line of print, as follows:

324. The list of Latin abbreviations and contractions adds one entry and modifies three others, all in the first of the two columns. To accommodate the change, the final item in the longer first column is moved to the top of the second, thereby also equalizing their length. As noted earlier, the entire list is lowered slightly on the page to differentiate it more clearly from the introductory text. The new entry in this alphabetical roster occurs at the beginning of “g”: “g [with an overring] = ergo”. (Unlike in the other entries, “ergo” is not followed by a stop.) McKerrow also revises the abbreviations used for “apostolice,” “apostolorum,” and “igitur.”

353. In the second column of this index page McKerrow adds an entry recommended by Sisson in his Library review: “forel (a kind of parchment), 123”. He does not, however, incorporate the one for “measure” that Sisson also wanted. To accommodate “forel” the printer has reduced the white space between the entries beginning with “e” and “f”.

355. In the entry for “kern” in the first column of this index page, McKerrow corrects the page reference “318.” to “313.”, as Sisson advised in the

Library review. Sisson also hoped, unsuccessfully, that McKerrow would define the term earlier in the book than he does.

III

The text of Bowers's Principles of Bibliographical Description likewise changes over its lifetime. The modifications are significantly fewer than in McKerrow, though in one respect they are also more surprising (at least to those who have not been paying attention to the message of McKerrow and Bowers), for they are unannounced. I know of two occasions on which alterations in the book have been recognized. In his article “Textual Variation in Bowers's Principles: A Short Note” in the Swedish journal Text: Svensk Tidskrift för Bibliografi (5.2 [Feb. 1998], 118-119), Rolf E. Du Rietz identifies what prove to be the two most extensive amendments of the text, on pages 20 and 94; his hope that “the clue” he provides “will be followed up” has served as the immediate stimulus for the present exercise. Meanwhile, for several years Terry Belanger at the University of Virginia has distributed to his Rare Book School class in descriptive bibliography a sheet illustrating a variant passage on page 202 as it appears in four printings.

I have identified only one occurrence of a change for the worse, the result of accidental type damage. The other alterations all are improvements, designed to correct the work of either the printer or the author. Not all lapses in the original printing are rectified, however, including at least one that a reviewer spotted. The TLS pointed out that the middle initial in the name “Mr. A. E. Esdaile” (apparently the reference in note 27 on page 30) was wrong. It stays that way. (Arundell J. K. Esdaile himself did not call attention to the transgression in his own review of the book.) In another instance, the page from the heart of Principles chosen to illustrate Myers's 1979 article in Antiquarian Book Monthly Review (where Bowers's page 216 is pictured on page 363 of the journal) happens to be the one that in all printings lacks a superscript in the collation formula at the top.[29]

Such anomalies, in the book but also in reviews, sometimes prove useful in suggesting aspects of the pre-publication history of the work. Bowers's sample description of Washington Irving's Wolfert's Roost (pp. 480-484) first appeared as a mimeographed handout to accompany his talk “Some Problems and Practices in Bibliographical Description of Modern Authors” at the inaugural meeting of the Bibliographical Society of the University of Virginia on 26 February 1947. The published version differs in several ways (including in the elimination of a format designation at the start of the collation formula), but the most curious change is the addition of the phrase “stet 12 PARK” after the title-page transcription. (It also appears, in the same position, in an abbreviated “basic form” of the description that Bowers adds in the book version.) Although the title-page imprint itself contains the address `12 PARK PLACE', later paragraphs of the description report that pub-

lisher's advertisements incorporated at the end of the volume are headed `10 PPARK PLACE' and that the third through sixth impressions have the street number 10 instead of 12 on the title page. What seems to have happened is that a copy-editor of Principles regularized the street numbers and possibly the typography of the street name, that Bowers or someone else corrected that mistake by writing “stet” and the original form of the address on the printer's copy, and that the compositor then mistook that note for text that was to be set in type.

In a somewhat different vein, F. C. Francis's Library review (cited in note 1) may also reveal something about the pre-history of the book. Francis makes two statements that are puzzling: he questions the appropriateness of the identification on page 6 of an inserted leaf as “χ1”, and he cites a statement that he thinks requires rewording: “`The symbol χ was devised to stand for an unsigned prefixed leaf or gathering', p. 20; italics mine [i.e. Francis's]” (p. 211). The first reference, however, occurs on page 7 instead of 6, and χ is not mentioned at all on page 20. Nor have I been able to find the sentence anywhere in the section Bowers devotes to “The symbols π and χ” (pp. 213- 222)—where he specifically disagrees with the statement Francis cites. It is possible that Francis simply got both page numbers wrong, but the quotation of a passage apparently not in the published book may indicate that he was working from a pre-publication text and may also point to what that text actually said.

As far as the published forms of the book are concerned, alterations certainly did occur. The ones I have identified follow.

6. In the fourth line from the bottom of the final footnote, the word “of” (in the phrase “possibility of textual”) is lightly printed in the first Russell and Russell impression and almost invisible. It is inked normally in the next three printings but disappears completely in 1977, where the “v” in “variant” immediately above is slightly damaged as well. Conceivably a copy of the first Russell and Russell printing was photographed to produce the 1977 one, or possibly the same lithographic plates were used for both impressions.

12. In the first printing (and in all St Paul's reprints) the page number is improperly printed in the left margin, as a hanging indent. It is properly placed within the measure in all Russell and Russell impressions.

20. Another apparent printing error results in a duplicate of the fourth line from the foot of the page (“with some exceptions limit a book only to editions of quartos up to”) being inserted as the eighth line up. Du Rietz says that “I remember at least one messed-up passage in the original edition being mentioned in one of the contemporary reviews, thereby inviting some action to be taken in forthcoming impressions” (p. 118). He probably had in mind Francis's Library reference to this passage (p. 211). Because the text originally intended between the ninth and seventh lines up was not the same length as

the nonsensical words that did appear, and because adjustments here could entail resetting the remaining fifteen lines of the paragraph, lines around the problematic passage were instead rewritten to fit the surrounding text in order that only four lines—the tenth through the seventh from the foot— had to be reset. The revision appears in all Russell and Russell printings; the original and new portions of the sentence that was changed are as follows:

85. The date in line 12 is corrected from 1640 to 1670 in the Russell and Russell printings. Lines 12, 13, and 14 are then reset, with a change of lineation: “one| recorded... title| advertising” becomes “one re-| corded ... title adver-| tising”. In the course of this replacement, lines 10 and 11 move slightly with respect to the text above them on the page.

94. The change in line 13 from “the title and preliminaries” to “the preliminaries” in the Russell and Russell impressions results in the resetting of the rest of the paragraph (lines 13-23). The distance between lines 12 and 13 is about half a millimeter greater in the revision; partly as a result, tips of some ascenders from line 13 are visible from the earlier text, over which the new setting has been pasted. Because the revision makes the new text one line shorter than the original, the paragraph is now followed by a full line space. Du Rietz observes that the new type is a bit thinner and lighter than the adjacent passages.

202. In the first collation formula (line 10 on the page), the inappropriate superscript “4” after the opening asterisk has been removed (apparently by opaquing) in the Russell and Russell printings. The deletion leaves a space between the asterisk and the subsequent hyphen.

275. In the first of the group of three examples at mid-page (line 20), “4 leaves” is corrected to “40 leaves” in the Russell and Russell copies. The entire line is simultaneously reset, in slightly smaller type.

375. In line 18 of the Russell and Russell impressions, the end of the title is corrected from “In the Key of the Blue” to “of Blue”—as William R. Parker had suggested in his review article on the original printing.[30] The final five lines of the paragraph (the correction appears in the first of these) are reset and apparently pasted onto a copy of the original printing; they are

lighter than the surrounding text and tilt slightly to the lower right. Fragments of ascenders from the original setting are visible above the last four words in the first line of the new text.

433. Lines 4 and 5 are reset to correct the final word in “The Marble Fawn” to “Faun” in line 5. Again Parker noted the error (p. 222, n. 2, of the 1950 PBSA); again the new type is lighter than the rest; and again some slight misalignment is evident (the new lines are indented minutely—by a third of a millimeter—from the left margin, though the right margin is satisfactory).

IV

Among the interesting conclusions that arise from this examination is that the latest reprints of “McKerrow” and “Bowers” are not based on the most correct texts. Principles was revised for the first Russell and Russell printing of 1962, with the corrections retained throughout that subedition. The St Paul's printings of the 1980s and 1990s, however, reproduce the original 1949 Princeton impression, with its various errors. The most reliable reprint would be based on one of the first four Russell and Russell printings, the ones bound in blue (the fifth suffers from a disappearing word on page 6). Puzzlingly, the St Paul's publishers show no awareness of the reprintings that come before their own. Each of their impressions provides on the copyright page some indication of previous ones, in a manner that increasingly suggests it is meant to be inclusive. The most recent statement, from 1998, reports the 1949 original and the St Paul's printings of 1986, 1987, and 1994. It also explicitly calls itself the fifth—rather than the tenth, as in fact it is. This preliminary analysis of Principles also points to kinds of information relevant to a fuller investigation of its printing history. For instance, it is clear from aberrant line spacing and residual text from the photographed copy that the 1962 Russell and Russell printing was reproduced lithographically, with reset text pasted across the original before it was photographed. That knowledge in turn may provide insight into the physical processes by which the original 1949 Princeton impression was manufactured. Machine collation of the 1949 and 1962 printings reveals horizontal movement of lines 1 and 2 against 3 through 6 of the note on page 48. This variation is unlikely to have resulted from the lithographic method used to produce the 1962 printing; it may instead appear because the copy of 1949 photographed for 1962 differs from the copies of 1949 that I have collated. Such a variation in turn probably would result only if the original impression was printed from type or Linotype slugs rather than from plates (either relief or planographic).[31]

The latest impression of McKerrow's Introduction is not based on the best available text either, though at least here the reprint derives from a copy that incorporates the changes publicly introduced in 1928. According to statements on successive copyright pages, the latest impression is reprinted either “photographically” (printings 3 and 4) or “lithographically” (printings 5 through 12) from “sheets of the second impression.” (The thirteenth printing says simply “First published 1927| Thirteenth impression 1977”; it too descends from the 1928 revision, without acknowledging its ancestry.) Despite that account of later printings radiating from the 1928 one, a cursory identification of type damage throughout the series suggests that most new impressions derive from the immediately preceding one.

Future bibliographers will need to test that proposition (recognizing that different sections of a book may have been photographed from different copies), but at the least it is clear that, whatever their assertions, some printings are not prepared directly from that of 1928. In the partial word “pensate” that begins the final line on page 11, for instance, the upper half of the stroke of “s” has a bulge from the fourth printing on; in the pagination on page 79, the stem of the “7” is broken just below the bar in all printings beginning with the sixth; and in the first line of note 2 on page 223, the “ce” of “cancellation” is partly absent from the eighth printing on. It is conceivable that incredible coincidences have made anomalies arise identically and independently in a number of printings, but it is more likely—indeed, virtually certain—that instead of being based directly on the 1928 revised printing, subsequent impressions stem from the first or later ones in each of these series, either because an early one was photographed for later impressions or because negatives or lithographic plates were reused from one to the other. In at least one instance economic factors confirm the likelihood of such a genealogy. Beginning with the third printing and continuing through the twelfth, the book was bound in gatherings of eight leaves rather than four, and the printed signatures had to be changed from the first two impressions. It then made sense to base subsequent printings with the new binding structure on a printing that had the revised set of signatures, rather than to modify a copy from 1928 every time the book was reprinted.

But while not all later printings descend directly from the 1928 impression, it is also the case that not all of them derive (at least for all pages) from the immediately preceding one. The fourth printing does not repeat the third printing's “a” with the depressed shoulder in the “pensate” example (from the foot of page 11), for instance, nor are the fifth printing's anomalous instances of the number “7” (with large serifs at the upper left in the pagination for pages 78 and 79) or the extensive damage of the bottom of the “a” in “especially” (in the seventh line from the foot of page 118) duplicated in the impression that follows. Nor, for example, does the thirteenth printing follow the twelfth on page 115 in obliterating significant parts of “th” in “the”

four lines from the bottom of the notes and of “Th” in “There” in the line below that.

Both of these observations—that later printings are not derived directly from that of 1928, and that they also do not always descend from the immediately antecedent impression—are manifested by the 1994 St Paul's printing. The description of its text on the copyright page is on the model of that in the Oxford impressions: “St Paul's Bibliographies... reissued the 1928 text... in 1994.” It is first of all clear that this printing, the fourteenth, was not produced from the preceding one. The thirteenth, the first to have gatherings of sixteen leaves, has no signatures; the fourteenth, also in sixteens, seems likewise intended to have no signature markings, but vestiges of them remain from an earlier printing. The letter “H” appears at the foot of page 97, and in some copies a smudge, the slight evidence of the signature “F,” is visible under the word “for” at the foot of page 65. (As McKerrow pointed out, “we sometimes find that the compositor in setting up the reprint forgets that he has to alter the signature” [p. 194]). The fourteenth printing therefore comes from one that has those signatures—that is, from the third through the twelfth impressions. The range of possible sources is further reduced by the recognition that at line 7 of page 263 the St Paul's printing follows the sixth through twelfth impressions in omitting the “os” of “imposition” (the thirteenth printing restores these letters, albeit in a different font). The exact source of St Paul's is pinpointed by a particular page number that inexplicably appears in four different positions throughout the various printings. The “2” of “291” sometimes falls under the “e” of “printer” in the previous line—4.2 mm below it (from base line to base line) in the first seven printings, and 5.8 mm down in the eighth. After that, it occurs under the final “r”—5.6 mm below in the ninth and tenth printings, and 5.0 mm below in the eleventh through thirteenth. In the St Paul's impression the numeral is 5.8 mm below the “e”. This fourteenth printing clearly was produced from the eighth impression, an inference confirmed by the great number of “hickeys” or dirt spots that are present uniquely and identically in the two printings.[32] The best existing form of the book to use for a reprint would indeed have been the second impression (modified by inclusion of two elements from the first printing: the original Figure 9 from page 35, and, less importantly, the pagination from page 113). Progressive type damage, resulting in the disappearance of certain characters, means that impressions starting with the third are less adequate, those beginning with the sixth less acceptable yet, and those from the eighth on least desirable of all.

In his review of Principles in the journal that McKerrow founded, Herbert Davis avouches that for “the student of textual problems,” bibliographical concern with late reprints “with which the author can have had nothing

to do” is “sheer waste of time.”[33] Davis's assertion begs many questions even for the audience he speaks of—including how one might determine that the author in fact had nothing to do with the reprints. For readers of McKerrow's Introduction and Bowers's Principles his generalization is especially problematic, for it would seem to preclude close examination and resulting insight into the only forms of the books that most of them will encounter. What case studies like the current ones can do is to help readers of these works see particular authors and printers going about their tasks and understand better what the books are and why they are so. But such examinations also have application for all readers. Besides suggesting ways in which books might be read and studied, they provide reminders of the nature of verbal works and of how they develop, and they underscore the need to read critically at the most fundamental level, assessing the very symbols on the page.

POSTSCRIPT

After this article was set in type I was able to locate and consult correspondence about Principles between Bowers and the Princeton University Press.[34] These letters provide additional evidence for a number of points established by analysis of the book itself, and they cast light on several related topics as well. They also reveal that the idea behind the current paper— analyzing Principles by its own standards—is not original. On 27 March 1948 Bowers proposed to a Princeton editor, Mariam Brokow, that the book include a bibliographical note describing its gatherings and typography according to the methods the volume espoused. (Such a note was not included.)

The correspondence illuminates the conceptual genesis of the book, in particular its relationship to McKerrow's Introduction. Many of Bowers's exchanges were with Datus C. Smith, Jr., director of the Press and the person who on a visit to Charlottesville in April 1946 had urged Bowers to submit his developing manuscript for the Press's consideration, even though it was only (as they then underestimated) seven-eights complete. On 3 November of that year Bowers wrote Smith,

As publication of Principles approached and the complexity of the undertaking became clearer, Bowers reiterated his plans: “I've something in mind in the distant future about rewriting McKerrow for Princeton if Datus doesn't lose his shirt on me this time” (Bowers to Brokow, 26 June 1949). Bowers never did get to that project, though his vision of what such a book would require seems to lie behind his disappointment with Philip Gaskell's A New Introduction to Bibliography. In “McKerrow Revisited” (PBSA 67.2 [Second Quarter 1973], 109-124), Bowers concludes his review of Gaskell by pointing out that “it is deceptive to call it a New Introduction as if it were intended to act as a replacement for McKerrow.”

Much of the correspondence points to the special typographical difficulties posed by the volume. The book apparently was the responsibility— and the achievement—of the Press's designer, P. J. Conkwright, who, according to a letter of 7 November 1946, preferred dropping the definite article from the proposed title The Principles of Descriptive Bibliography. A couple of weeks after publication Smith wrote Bowers, “The printing of this book was surely a challenge to all of the people engaged in the enterprise, and although we certainly would not want something of that sort to be going on around here every week, it was a stimulating experience for all of us” (13 January 1950). Smith noted that charges for Author's Alterations exceeded the amount to which Bowers was entitled without charge ($263.25) by $174.68 —and that the Press would absorb the cost: “we feel that in the case of this book many of the alterations were a function of the typographic improvisation, and a natural part of the whole job of producing a book of this sort.” Bowers insisted on paying anyway, eliciting this response from Smith on 19 January:

Bowers's petition was denied.

One of the most important revelations of the correspondence is that Bowers himself was responsible for the alterations in Principles. Corrections had been on his mind almost from the start: on 5 December 1950 he wrote Smith, “In case you plated the book and on some future occasion ever run off more copies, for God's sake give me a chance to mend a few places, especially one where a line of type dropped out and another got repeated after I'd OK'd page proof!” When Russell and Russell decided to reprint the volume, Bowers responded to the invitation of Gordon Hubel at Princeton to make changes; in a letter of 12 December 1961 he listed the points I have identified above, except for the mispositioned page number on 12.

The file also casts light on a number of other matters raised in the present article. Lawrence Wroth had identified himself to Bowers as a reader of the manuscript for the Press; the file reveals that the other reader was René Wellek. According to letters of 14 September 1946, each was paid $25. Bowers had recommended Francis R. Johnson and James M. Osborn as evaluators; according to a letter from Smith to Harold Munger at the Press on

3 April 1946, Bowers seems to have advised against asking William A. Jackson. The Press nonetheless approached Jackson and Henry Bartlett Van Hoesen, both of whom declined the undertaking. Jackson's letter of 3 May 1946 added a skeptical note: “I cannot refrain... from saying that I have my doubts as to whether any system of descriptive bibliography can be made standard even were it demonstrably the best.”

The comment about Principles by Victor Hugo Paltsits (cited in note 1) was indeed not from a review but from a solicited statement Paltsits submitted on 1 August 1949 after reading galley proof; another Paltsits quotation appeared in a full-page advertisement for Principles on the back cover of Antiquarian Bookman for 21 January 1950. On 25 September 1949 Bowers encouraged the Press to send advance review copies to two major English publications, The Library and Times Literary Supplement. Because three days later the Press mailed page proof to Earle P. Walbridge, editor of PBSA (in the hope that his journal might report on the book along with or even before its review of Bühler, McManaway, and Wroth's published Rosenbach lectures), it would seem likely both that advance copies were in fact sent to the two London periodicals and that these copies too were in the form of page proof. The discrepancy between the references in F. C. Francis's Library review and the text of the published book might then be explained by the fact that he was using a pre-publication form of the work.

On 21 June 1949 Miriam Brokow wrote Bowers about the “linotype schedule,” thus corroborating the method of production that seems clear from the physical evidence in the book. A few weeks after publication Datus Smith noted to Harold Munger (on 13 January 1950) that two thousand copies of Principles had been printed, at a cost of about $5900.