:: :: University of Virginia Library

For students of modern bibliography Sinclair Lewis's Babbitt offers worthwhile evidence about textual variation, and its problems, in present day printing from plates. The facts of variation are in themselves of no particular literary significance (save that they occur); but the implications that can be drawn from a study of Babbitt's printing history as revealed by these facts are of real moment to critics of contemporary texts and to scholars who must quote from texts produced by 20th-century printing shops. Of more specialized interest is the appearance of two bibliographical problems, for one of which, at least, no certain solution is available.

The variants in question appear in a table at the end of this note. They were discovered by examining copies of different printings on the Hinman Collating Machine owned by the University of Virginia Library.[1]

In all, five states of variation may be found between the original state of the plates and the state represented by current reprints in the Modern Library and Harbrace Modern Classics editions (both printed from the same plates). The first stage of correction occurs during the initial printing published on September 14, 1922, by Harcourt, Brace & Co. Here the substitution of Lyte for Purdy, and of any for my on page 49 is of some seriousness since the original reading had misnamed the character. The problem is in two parts, of course: (1) to determine whether the error was corrected by stop press or by cancellation within a single impression, or whether the error was confined to the whole first impression, and the corrected state of page 49 to a different impression of the book; and (2) who caught the error and when, and how it was corrected.

Because stop-press alteration of plates is expensive in lost press-time when undertaken by modern printing establishments, absolutely attested cases have not, apparently, found their way into published bibliographical investigation. Hence one's first impulse might be to conjecture that there were two pre-publication printings of Babbitt and that the changes on page 49 were made between the two impressions. There were indeed two prepublication printings. The Harcourt, Brace records[2] show that the first

264

printing bill for 80,500 copies of Babbitt was dated July 27, 1922; and the second press bill for 20,000 copies September 23, 1922. The first plate correction bill was September 23, 1922. Since publication date was September 14, it is clear that the second impression had been ordered before publication and may even have been run off the presses by that date although (see below) not released until October.

However, this impression can be identified, and it is not the same as the impression in which the only variants are those on page 49, for in this second printing the Hinman Collator discloses some evidence of slight plate damage that would occur through storage of plates between impressions. Moreover, additional plate corrections were made, on pages 75, 85, and 271, not present in the various copies containing only the page 49 altered readings. Doubtless because the order for the second printing of 20,000 copies was placed before publication as a supplement to the original printing, no change from the first-impression date was made on the verso of the title-page for these copies; hence they can be distinguished from first-printing copies only by the presence of the corrected readings on pages 75, 85, and 271. However, on the title-page verso of the third impression this second printing is given belated recognition as Second Printing, October, 1922, followed by the line Third Printing, October, 1922.[3]

To sum up. The publisher's records list between the first and third printings a second of 20,000 billed on September 23, 1922. By some slight plate damage and by alterations to the plates of pages 75, 85, and 271 (accounting for the September 23 bill for plate corrections) this impression differentiates itself from the impression with the original states of the plates and (on the evidence of the Hinman Collator in the lack of plate damage) from the same first impression though with the two alterations on page 49. This second printing is subsequently numerated on the title-page verso of the third and fourth impressions, although not on its own title-page verso. The Lyte-any variants on page 49, therefore, were not made in the interval between two impressions but instead are present within the original first impression.

As to the exact circumstances of these page 49 alterations within the first printing we have no information and at this late date are unlikely to secure any. On some occasions a modern pressman at the start of printing will scan an early copy of a sheet for mechanical errors; but that such a workman would be so intimately concerned with the sense as to catch the mistake in a name, as here, is scarcely credible. It would be possible to conceive on other occasions that delay in the delivery of author's corrected

265

proof for final revises, or some error in failing to make final corrections, might inadvertently extend even to the early stages of printing. No precise decision seems possible here, although conjecture may be allowed. It would seem most probable that by some accident, the details of which cannot be recovered, these two page 49 corrections ordered for the text in linotype form had not been made before casting of the plate; and the failure to correct was then discovering by comparison of such proof with the foundry proof (pulled from the plates) only after printing of the forme had started. We know that Lewis had partly relied upon the professional proofreaders in the Quinn & Boden printing establishment. Hence, as an alternative theory we might guess that someone, perhaps interested in the story, took it upon himself to read the whole collection of foundry proofs and in this process came upon the page 49 errors.[4]

Thus we are faced with the simple proposition, either (1) the case is one of stop-press correction, or (2) the case is one of sheet cancellation. That the Lyte-any sheets of the first printing represent completely reimposed and newly imprinted sheets at a later time is possible but not probable on the evidence. If a cancellans sheet had been manufactured before binding had started, the Purdy-my sheets would need to represent faulty copies which had been ordered destroyed but by some error found their way to the bindery along with corrected sheets. Such a theory cannot be maintained in view of the evidence that substantial numbers of the original state are in existence (6 of the original against 8 of the corrected in the 14 copies I have checked).

As an alternative we should need to suppose that binding had started before the error was discovered and the cancellans sheets provided; in such a case the Purdy-my copies would represent copies initially bound before the cancellans sheets arrived. There is no positive evidence for this theory, and some rather slight evidence against it. In the first place, no records of a cancellans sheet exist. In the second, I am inclined to lean some weight on the fact that at the time the alterations were made in the original set of plates, a duplicate set of shells (unbacked plates) that had also been cast were similarly altered. The argument is not strong, but on the evidence that later plate alterations were not (except for two anomalous pages) simultaneously transferred to the shells, there may be reason to associate the page 49 corrections as near as possible to the original casting and checking of both original plate and shell, since changes made in the one seem to have been made simultaneously in the other. The odds are, therefore, that in the page 49 variants we have an example of comparatively rare modern stoppress correction.[5]

266

In the unnumbered second printing (not distinguished from the first but subsequently listed as the second, and as distributed in October), the page 49 alterations are found, of course, and in addition errors on pages 75, 85, and 271 were corrected in the plates; and in the numbered third printing (October) a further necessary change was made, on page 397, from against the Open Shop to for the Open Shop. In large part as the result of Mr. Louis Feipel's suggestion,[6] twelve further corrections were made in the fourth printing distributed in November of the same year. This state of the plates also machined the Grosset & Dunlap movie reprint edition of 1924.

An immediate anomaly appears when the series of printings are collated as represented by the Harbrace Modern Classics, post-1942 cheap edition Grosset & Dunlap editions, and the Modern Library reprints, for with the exception of the two corrections on page 49 stemming from the plate alteration to this single page during the first impression, and the variants on pages 188 and 196 first found in the fourth printing, the remaining fourteen alterations revert to their original faulty readings as present in the first-impression plates. Duplicate plates provide the explanation. The Harcourt, Brace & Co. records show that on July 1, 1922, one set of plates and a duplicate set of shells (unbacked plates) were manufactured (information

267

again by the courtesy of Mr. Gerald Gross). On December 20, 1941, the original set of plates was melted, and on January 7, 1942, the duplicates, or shells, were backed up to make them into printing plates. From this second set, therefore, were printed all the Harbrace Modern Classics and Modern Library editions.

That page 49 in these duplicate plates appears in the corrected state is explicable in the normal course provided the stop-press alterations were made from foundry proof and thus in both sets at the same time, plates and shells. However, the seemingly arbitrary selection only of pages 188 and 196 to bring into conformity with the first-set alterations in the fourth impression is very puzzling. Both pages are in the outer forme of the same sheet, but the significance of this fact is difficult to apply.[7]

From this history of the plates we learn that the Lewis who sneered at an English Rotarian for speaking of "Bertrand Shaw" was not himself so scrupulous as he might have been about the accuracy of his references to the fraternal organizations that he was satirizing. The B. in B. P. O. E. he twice expanded in error as Brotherly (pp. 9, 165, corrected in the fourth printing) although once correctly as Benevolent (p. 55). The forms Oddfellows (p. 203) and Redmen (p. 188 but correctly on p. 203) might seem as ignorant to members of these orders as "Bertrand" Shaw to a literary man.

But the really interesting conclusion from this plate study, outside of the two bibliographical problems involved, is the evidence that in modern textual transmission one cannot necessarily trust the latest editions of plated books to be the most correct. Before the institution of the duplicate plates the history of Babbitt was one of steady improvement in the state of the text through the fourth printing, although no evidence is present to suggest authorial correction at any stage. But beginning with 1942 editions, the almost wholly uncorrected second set of plates reverted to a state of textual error that wiped out substantially all the results of the improvements in the second, third, and fourth impressions. As a result, the only reading texts now being printed are more corrupt than any impression after the original; and a textual critic of the future will need to search out either the fourth impression or some reprint before 1942 since only these represent the most highly corrected state of Babbitt's readings.[8]

268

Table

	September 1922. Harcourt, Brace.	September 1922. Harcourt, Brace.	"September" (i.e., October) 1922 Harcourt, Brace.	"Third Printing, October, 1922". Harcourt, Brace.	"Fourth Printing, November, 1922". Harcourt, Brace. Also Grosset & Dunlap movie reprint.	Harbrace Modern Classics, Grosset & Dunlap, and Modern Library reprints.
9.l.27	Brotherly				Benevolent	Brotherly
15,l.30	tread				trod	tread
44,l.18	raise cain				raise Cain	raise cain
49,l.4	Purdy	Lyte	Lyte	Lyte	Lyte	Lyte
49,l.5	my	any	any	any	any	any
75,l.34	plain		plane	plane	plane	plain
	geometry		geometry	geometry	geometry	geometry
85,l.5	I means		I mean	I mean	I mean	I means
121,l.4	benny				kelly	benny
165,l.22	Brotherly				Benevolent	Brotherly
188,l.15	Redmen				Red Men	Red Men
196,l.24	tawney				tawny	tawny
203,l.11	Oddfellows				Odd Fellows	Oddfellows
235,l.34	principles				principals	principles
240,l.20	Offut				Offutt	Offut
271,l.11	Three		Three	Three	Three	Three
	Black		Black	Black	Black	Black
	Pennies		Pennys	Pennys	Pennys	Pennies
349,l.13	raising				raising	raising
	cain				Cain	cain
392,l.1	against			for	for	against
397,l.18	from				with	from

Notes

269

[1]

The two variants on page 49 were first discovered by Mr. Jacob Blanck (see Sinclair Lewis: A Biographical Sketch by Carl van Doren, with a Bibliography by Harvey Taylor [1953], pp. 102-103; the rest, I believe, have not been recognized. I am grateful to Mr. John Wyllie, University of Virginia Librarian, for suggesting this study to me and for helping me to work out the evidence, and to Professor Fredson Bowers for assistance in the preparation of this account of my findings.

[2]

Teste Mr. Gerald Gross of Harcourt, Brace & Co., to whom I am particularly indebted for much valuable information. I am also grateful to Mr. James T. Quinn of Quinn & Boden Co. (the original printers of the book), and to Mr. William Simon, Jr., of H. Wolff Co (printers of the Modern Library edition), for answering questions about the printing of Babbitt.

[3]

Extended search has failed to turn up any copy listing itself as the Second Printing on the title-page verso. It is not likely that one will be found since the time interval between this recorded second printing and the third printing is very likely too narrow for still another to exist.

[4]

The possibility that this stage of the plate alteration resulted from some error in the printing plant might be argued since it would seem that these particular plate alterations were never billed to Harcourt, Brace.

[5]

A collector concerned with 'points' will, of course, choose the Purdy-my state as the earlier. There is no validity whatever to the arguments of the bookseller Robert K. Black that since the Lyte-any readings are found in a copy autographed by Lewis on the day after publication (the copy is now in the University of Virginia Library), they must represent the state found in the author's advance copies and thus must be considered as "gathered and bound before any of the so-called 'first state' copies, so that in terms of priority of issue, though not of printing, it actually precedes them and is therefore more desirable" (see Mr. Black's List Number Forty-Five issued c. March 1956 from Upper Montclair, N.J.). In the first place, the copy cannot be proved to be from the advance copies sent Lewis. In the second place, the assumption that an author's advance copies represent the first copies off the binding machines is ordinarily wrong: it supposes that copies are shipped to the publisher in driblets from the bindery and that the publisher immediately breaks open a case and sends out the author's copies. In the third place, the order of pre-publication distribution of copies can have no bearing on questions of 'issue,' especially when mere press variants are involved. For a full discussion of this matter, see Bowers, Principles of Bibliographical Description (1949), pp. 409-411.

[6]

Mr. Feipel, who makes a hobby of proofreading published books to note their errors and inconsistencies, wrote Harcourt, Brace on November 6, 1922, enclosing a list of over seventy readings that in his view required alteration for consistency of usage or because of real error. In response to the question whether the spelling areoplane on pp. 19 and 85 was intentional, Mr. Donald C. Brace of Harcourt, Brace, acknowledging on November 9 Mr. Feipel's list, replied that the spelling was intentional, and added, "It may interest you to know that it was an effort, each time the proof was read, to keep the proof reader from changing it." Lewis's response to the list was in character: "J. Henry! This man Feipel is a wonder — to catch all these after rather unusually careful proofreading not only by myself and my wife but also by two or three professionals at Quinn's! . . . (Gawd, Feipel has me nervous about hyphens!)" (Lewis to Donald Brace, November 10, 1922, quoted from Harrison Smith, ed., From Main Street to Stockholm [1952], pp. 113-114. The change from benny (i.e., overcoat) to kelly (i.e., hat) was suggested by Keith Preston in his column "Hit or Miss" in the Chicago Daily News for Nov. 4, 1922, on p.8, in connection with a note, signed 'XYLOID,' about the glossary in the British edition of Babbitt.

[7]

The Hinman Collator discloses that type wear on pp. 188 and 196 present in the fourth impression does not appear in the later printings that were made from the duplicate plates of these pages. Hence one is barred from conjecturing that for some reason the original plates for these pages were held over for use in later impressions: it is certain that pp. 188 and 196 were printed from the duplicate plates, which must therefore have been altered. But why other plates were not also altered at the same time to conform to the fourth-impression corrections cannot be explained.

[8]

As a pendant to this commentary on modern textual transmission, one might point out that the single really substantive alteration that corrects something not clearly a misprint (except for the Purdy-Lyte alteration) was in fact made on the suggestion of an outsider (i.e., kelly for benny on p. 121).