The combined examination of printed book and account ledger unravels secrets about The History of Pendennis also. The first edition exists in several impressions and substates.45 A superficial comparison of any copy in parts with any copy bound in two volumes shows that the book issue is composed of sheets printed from the same typesetting used for the parts issue. The two forms of the book, however, incorporate many differences. While the particular changes themselves are of little critical significance (and seem to be editorial in origin), the evidence they provide concerning printing practices is significant to bibliographers both textual and descriptive, to book collectors, and to printing and publishing historians.
In brief the situation is this: Pendennis in parts was first printed from type. Stereotype plates were subsequently cast from the same setting of type in anticipation of the printing for the issue in two volumes. During this process intentional changes were introduced (see Appendix E, table 2) and unintentional damage occurred. Finally, in 1865 - ten years after Bradbury and Evans had brought out a revised, cheap edition of Pendennis and nearly two years after Thackeray's death - Smith, Elder and Company acquired the plates of the first edition, repaired a few damaged pieces of type, furthered the unintentional damage, and printed the novel several more times.46 These last impressions are distinguishable from other printings by the substitution of the Smith, Elder imprint for the Bradbury and [186/187] Evans imprint at the foot of the last printed page in each volume. The title page and dates remained unaltered in all printings.
That summary glosses a multitude of complexities. Because the idiosyncrasies of bindings-up often resulted in mixtures of impressions within single copies of the novel and because new impressions of single numbers, as well as new general impressions, could occur, the printing history of Pendennis must be approached on the level of sheets.47 One cannot say categorically that a copy of Pendennis in parts represents the first printing or that a copy in two Volumes with Bradbury and Evans imprints represents the second or third. Furthermore, in dealing with the printing history of the novel, it is not possible to take the publisher's records at face value, for they do not tell the whole story. Stereotyping is entered with the initial cost of printing Pendennis, bur machine collations show beyond doubt that the first printing was from type. In fact, variants in copies of numbers 2 and 5 printed from type seem to indicate separate printings from the same setting of type before stereotyping. In addition, while I have seen one copy of the issue in book form composed partially of sheets printed from type (definitely not issued previously as parts), I have also seen four copies issued in parts with scattered sheets printed from plates.48 To complicate matters, I have noted four copies of the book issue of volume 1 composed of sheets printed from plates except for number 12 - [187/188] the last two sheets in the volume - which was printed from type.49 The problem that remains is to reconcile the records with the evidence of collation and to determine when the stereotyping occurred.
The one copy of the first book-form issue containing some sheets printed from type (other than those in number 12) is the most crucial in settling these problems. That this book belongs to the first issue in book form is not absolutely certain from external evidence though it is a presentation copy from Thackeray to Peter Rackham. The distribution records of the publisher show Thackeray receiving presentation copies early in 1851; in fact, Thackeray's presentation copy to Dr. Elliotson is dated 1 January 1851. However, it is certain that the Rackham copy is not a bound set of parts, for there are no telltale stab holes in the gutters. The records and the evidence of this book suggest an answer to the question of when stereotyping took place.
By 31 December 1849 the total number of copies printed of part 1 was 10,000 (in one printing); part 2, 9,500 (in two printings); parts 3-4, 10,000 (in two printings); and parts 5-12, 9,000 (in one printing)50. The initial 8,000-copy printing of part 2 was exhausted by May 1849, and the initial 8,000-copy printings of parts 3 and 4 were exhausted by August 1849, for reprints were required on those dates. It can be assumed on this basis that an average of 8,000 copies per number had been issued in parts before volume 1 was ready to be bound in book form. If the assumption is correct, only 2,000 copies of the first printing of part 1, none of parts 2-4, and only 1,000 of numbers 5-12 could have remained to be issued in book form. One might expect, then, that about 1,000 copies of the book issue were composed of sheets printed from type with the exception of numbers 2-4 and that about 1,000 more contain number 1 as well as number 12 printed from type.
The presentation copy to Peter Rackham shows that, like all early issues in book form which I have examined, part 12 is printed from type. But, more significantly, unlike any other copy issued in book form which I have seen, volume 1 is composed mostly from sheets printed from type - the exceptions being numbers 2-4, which were reprinted before the first issue in book form. The evidence of the Rackham copy indicates that the initial printings entered in the publisher's records (10,000 for part 1, 8,000 [188/189] each for parts 2-4, and 9,000 each for numbers 5-12) actually equal the number of copies printed from type before stereotypes were cast. Copies of volume 1 issued as parts sometimes contain numbers other than 2-4 printed from plates. Of those numbers printed from plates, the one most often noted in copies of the parts is number 9 (signatures S and T). There is the possibility, of course, either that sets of parts with sheets printed from plates were issued after the novel became available in book form or, and this is more likely, that incomplete sets of parts were made up with later printed sheets.51
As the novel established its popularity, which it did at a faster rate than did Vanity Fair, it apparently became necessary to print more copies of the early numbers to meet the demands of new readers picked up midway through the publication of the first volume, and these demands were met by reprints from the standing type. Variants in numbers 2 and 5 seem to indicate this. Then, perhaps anticipating the necessary reprinting for issue as a book but too rushed for time between Thackeray's submission of printer's copy and the publication deadline of each new part to allow for plating before any printing was done, the printer started plating immediately after the initial printing for the parts issue.52 And the early parts [189/190] until then kept standing in type also were plated. This scenario would account for the entry in the extant records for stereotyping in the initial
cost of publication.
When in September 1849 Thackeray's illness forced the suspension of publication until January 1850, only one number, the twelfth, was needed for the completion of volume 1. By this time plates of the first eleven numbers had been cast, and demands for back numbers may have been such that sheets printed from plates were sold as parts as early as 1849. (This is at least one explanation for the four sets of parts of volume 1 I have examined which contain sheets printed from plates.) Finally, when Thackeray had recovered sufficiently to write the twelfth number, sheets printed from type went into both the parts issue and the book issue. This accounts for the many book-form copies with only number 12, the last in volume 1, printed from type. It is impossible to hold that there were simultaneous printings from type and plates because in that case the plates would have had to have been cast from the type before the printings from type were
run off. Collation shows that the printings from type were made before the plates were cast, for damage occurring in late copies printed from type is carried over into the plates, And corrections made for the plates do not occur in copies printed from type.
A given set of parts of volume 1, then, instead of being the first printing, may be composed of the second impression (from type) of early sheets, the second or third impression (from plates) of the middle sheets, and the first impression of the last number. On the other hand, the first book issue of volume 1 may be composed of the third impression of some early sheets, the second or third impression of the rest through number 11, and the first impression of the last number.
Volume 2 is not so complicated. The publisher's records indicate a single 9,000-copy printing of each number from 13 to 24. There is no reason to believe otherwise. If the sale of parts of volume 2 matched that of volume 1 (that is, about 8,000 copies distributed, or at least bound, as parts before issue in book form), there should have been about 1,000 copies of the first book-form issue of volurne 2 made up of sheets printed from type. Three presentation copies in book form have been noted in each of which [190/191] volume 2 is composed entirely from sheets printed from type.53 The publisher's records of printings and distribution also tend to corroborate this hypothesis. First, only one entry is given for the 9,000-copy printings of numbers 13-24, and no reprint was required until 30 June 1853, two and a half years after the book-form issue. The distribution records are significant, even though they omit the record of sales at the date of issue in book form, December 1850, which would have made their use here much more precise. They show that by 31 December 1851 the publisher had disposed of 215,050 parts out of 228,500 printed. The number printed includes 11,000 copies of number 1, 10,500 of number 2, 10,000 each of numbers 3-12, and 9,000 each of numbers 13-24. If the 215,050 parts distributed were made up into as many copies as possible of the whole novel (not a likely occurrence), then about 8,950 copies of the book were out of the publisher's hands within one year after issue in book form. This figure, with a slight downward adjustment to allow for readers who may have discontinued the novel after a few numbers, fits in well with the 8,000-copy figure arrived at on the basis of the required reprints for numbers 2-4 mentioned above. Later copies of volume 2, of course, are composed entirely of sheets printed from plates.
As with sets of parts of volume 1, copies of volume 2 in parts have been noted time and again with sheets printed from plates. Again, I know of only two possible explanations - both of which may hold true. The first is that readers with incomplete sets of parts made up their copies after reprints from plates became available, and the other is that the publisher continued to promote the sale of the novel in parts even after the book-form issue.
The identification of impressions combined with the current practice of examining wrappers, bindings, and advertisements may make it possible to know with some certainty just what copy of Pendennis one has. But to elaborate Michael Sadleir's hypothetical case [Sadler, p. xiv], it is not only possible for the last 500 copies of sheets from one printing to remain for some time in the storeroom to come forth at a later date in an alternate binding or with new advertisements but also for, say, the last 200 copies of sheets of one impression to come forth dressed identically with, and mixed with, the first 200 copies of sheets from a new impression. In fact, as my [191/192] examination of Pendennis seems to show, the combination of sheets from different printings bound together could be almost endless. Even the one set of parts I have examined which is composed entirely of sheets printed from type is not a first printing, for in the case of at least three of the sheets, earlier printings (or at least earlier printed copies within an impression) are bound into other sets containing some sheets printed from plates.
The reprinting from type of isolated sheets of early numbers produced several variants. Like the readings which are defective in all examined copies but were apparently correct at one time (see Appendix E, table 1), some of these variants may be attributed to type batter. However, gathering D of number 2 does exist in three states, if not printings. Early printings read where, in the at page 42.27 and have a clear semicolon after ear at 43.46. The second state reads where, i h e at 4.2.27 but still has the clear semicolon at 43.46. The third state restores the correct reading at 42.27, but all copies of this state so far examined have a chip off the top of the semicolon at 43.46. This evidence may indicate a stop-press correction rather than separate printings, but gathering K of number 5 seems definitely to exist in two printings before plating. The lines at 143.2-6 and at 144.2-6 have been reset for no apparent reason; there are no textual variants in the reset lines, only respacing. On page 143 the printings are most readily identified by the position of the L of London at line 6 in relation to the space between the words and frets at line 5. In the first printing the L is under the space; in the second and all subsequent (including plated) printings the L is to the right under the f of frets. On page 144 the printings are identified by the position of the y in they at line 6 in relation to the space between the words were by in line 5. In the first printing the y is slightly to the left, half under the second e of were; in the second and all subsequent printings the y is to the right under the space. The distinction is an important one. By indicating the necessity for reprinting early numbers, it forms a partial basis for concluding that the readership of Pendennis was growing. One reason for this may have been pressure from readers made unpatient by the three-month delay in the publication of volume 1 as a whole, imposed by Thackeray's illness.
Of less importance, perhaps, but of interest is the evidence the first edition of Pendennis presents for the fact that plating does not fix the text of a book. In addition to the damage sustained through reprinting and plating, there is evidence of activity in gathering H (pp. 97-112) of Volume 1. Just before or in the process of the second printing (from plates), resetting became necessary for the portion of page too below the illustration. At 100.1 (above the illustration) the first impression (from [192/193] type) reads Digby; the second (plated) and all subsequent impressions read the correct Derby. In some copies printed from plates, the lower part of page 100 (below the illustration) exists reset (with respacing) and printed from type with no textual changes. Copies also exist reset a second time and plated, introducing three textual changes. The first impression of H, then, is from type and reads Digby at 100.1. The second impression, printed from plates and reading Derby at 100.1, exists in two states: (1) with the lower part of 100 plated without change: (2) with the lower part of 100 reset (respaced) without textual variants but printed from type.55 A third impression of gathering H is from plates with the lower part of 100 reset again, plated, and introducing "champion" for "champion," (100.10), "young" for "young," (100.12), and Milly (turned semicolon) for Milly;(100.13).
Only one oddity breaks the textual simplicity of volume 2. The two conjugate leaves at the center of gathering U in number 22 (pp. 295-98) are plated in one copy examined. The rest of the signature is from type. Type damage at 296.19b (defective r on nor) persists in the Smith, Elder impression, indicating that the same plates were used. This copy may be part of a late binding-up and issue in parts. And possibly it represents an anomaly created by an owner who replaced a torn page in this copy (printed from type) with a conjugate leaf taken from another, perhaps otherwise defective, copy (printed from plates).
While the foregoing description of impressions is necessarily imprecise, enough is known to avoid an editor's uncomfortable feeling "that in his bibliographical ignorance it is quite possible for him to base his text on ... an unrecognized late impression containing possible alterations of dubious authority." [Bowers, p. 359]
Esmond in Three Volumes
efore discussing the production and cost-sharing arrangement for Henry Esmond, it would be good to review what is meant by the terms impression, edition, and issue. In modern bibliographical parlance, edition refers to all copies of a book printed from a single setting of type; impression refers to all copies of a book consisting of sheets printed in a single pressrun; and [193/194] issue means all copies of a book marketed as part of a single publishing or marketing drive and (usually) bearing some distinguishing mark of that publishing drive, such as tipped-in title pages, different covers, or significantly different advertisements bound into the book itself. The terms were much more fluid in the nineteenth century. "Edition" was used sometimes to denote a new impression from the same typesetting, sometimes to mean a new impression from reset type, and sometimes to mean part of an old impression reissued with a new title page. "Impression" was used sometimes to mean the total number of copies specified in a contract, regardless of how many printings were required to produce them, and sometimes to mean any of the three things mentioned above as the meanings for edition, impression, and issue.
The Esmond contract specifies that the proceeds from the entire "first impression" of 2,500 or 2,750 copies (at George Smith's option) of the three-volume work were to belong to Smith, Elder and that "such impression may be published in one edition or divided into two or three editions as he may consider expedient" (NLS; see Appendix A). Confident but not knowing in advance how well the book would do, Smith was hedging his bets. By specifying the number of copies he could print and call his own, he seemed to commit himself to print at least 2,500 copies, but he left himself with a little cushion in case the sales went well. By allowing himself up to three "editions" in which to dispose of the "first impression," he allowed for the possibility that it would take years and several marketing efforts to dispose of the first printing.
The publication records for Esmond are sketchier than for any other of Thackeray's major works. The ledgers in which Smith, Elder kept accounts up until December 1852 have been misplaced or lost. The first entry for Esmond in the new ledger is for May 1853. Fortunately, one of the few surviving financial reports from Smith, Elder to Thackeray is for Esmond, but, of course, it begins with the "second edition " because Thackeray had parted with all rights to the "first impression." (Smith called it a second edition, and in this case it seems, in fact, to have been a true second edition, entirely reset, though at least 500 copies of it seem to have been issued with title pages identical to the first edition.)
Although there is no surviving ledger record for the first printing of Henry Esmond, the contract agreement indicates that Smith, Elder paid Thackeray £1,200 in three installments. The contract further specifies that Smith, Elder and Company was to have the total income on a "first impression" of 2,500 or 2,750 copies according to the publisher's discretion. If the first impression was of 2,750 and sold out within eighteen [194/195] months, Smith was, by contract, to pay Thackeray an additional £100. The first impression, published in October 1852, just as Thackeray was headed to America on his first lecture tour, was completely sold out within six weeks. Thackeray wrote to James Field on 29 November, "Smith writes to me from London that the whole of the first edition of Esmond is disposed of, an edition of 3000 57 at a guinea & a half!" [Huntington] Thackeray wrote at the same time to friends in England that Smith had indicated a second edition was being undertaken [Letters 3:135]. There was no need for a special agreement about the new edition because the original contract specified a straight profit-sharing system for subsequent editions. Since there are no extant records of the first-edition production or of any of the payments to Thackeray for the first edition, there is no way to determine whether Smith printed 2,500 or 2,750 or whether Thackeray got the extra £100.58 Whatever the case may be, if Smith sold 2,250 copies of the first impression (leaving 250-500 as gratis, promotion and review, and discounted copies) at 22s, 6d. each (the price commanded by the first 600 copies of the second edition), he would have grossed about £2,500. Mrs. Proctor wrote to Thackeray on 23 November that "Mudie says. 400 Copies of Esmond not being sufficient 100 more are added." [Letters 3: 216]59 It is likely that Mudie's Circulating Library did not pay 22s. 6d. per copy; hence, the generous allowance for discounts in my calculations. According to the terms of the contract, £1,200 and possibly £1,300 of that £2,500 went to Thackeray; at least £700 went into production costs (judging from the cost of the second edition, which involved a resetting of all the type, and by a rough comparison with the production costs of Charles Dickens's comparable Great Expectations in three volumes). John Sutherland, remarking on the frequency and prominence of contemporary advertisements and notice of Esmond, suggested that an expensive promotional effort was mounted for the book [Sutherland, Novelists, p. 113]. The relevant ledger evidence is not available. No special advertising effort is evident in the ledger accounts for the second edition. Smith, Elder could easily have made a profit of about £500 on the [195/196] first edition - perhaps more if a greater number of the 250-500 other copies were actually sold.
Smith's financial statement to Thackeray begins in December 1852 with production and promotion costs for the "second edition," which consisted of 1,000 copies. Production costs included composing type, paper, pressing, and a charge for 250 "third edition" title pages, all coming to £379.2.6. A note by the entry for the third-edition title pages says "not used," and in May 1853 the production of 250 "second edition" title pages is recorded. No other record of title-page production or reprinting of this edition exists in the ledgers; however, copies of the book with second edition sheets have been noted with first-edition titles, second-edition titles, and third-edition titles. In addition, a few copies with second-edition titles have first-edition sheets — which indicates that Smith may have projected the sellout of the "first edition" before all copies were out of the bindery and that "second edition" titles were produced before all "first edition" sheets were bound. In any case, it seems clear that of the 3, 500-3,750 copies of Esmond in three volumes that were produced, no more than 250 have "second edition" title pages, and no more than 250 others have "third edition" titles.
Be that as it may, by 30 June 1853, 668 copies of the new edition were reported sold for £662.5, from which Smith deducted £33.2.3 as his 5 percent commission on gross income. That left a credit balance of £250.0.3, from which £46.18 was spent in September 1853 to purchase back 67 copies "returned from America" in order "to protect stock in this country" (100 es had been sold to Appleton in New York, but Harper had produced a much cheaper authorized edition for which it had paid Thackeray $1,000). At this point, September 1853, Smith divided the remaining balance of £203.2.3 with Thackeray, the author getting the odd penny. In 1854, 53 more copies were sold for £57.7.6; Smith deducted the firm's 5 percent commission (£2.17. 4) and £10. 16 for promotion costs, etc., and divided the remaining £43.14.2 evenly with Thackeray. In February 1855 author and publisher cheerfully (or solemnly) divided the proceeds from the sale of two more copies, each receiving 8s. 9d. after Smith had deducted the company's 2s. 3d. commission. And so the story goes, until by 1864 only 124 copies were left and the selling price had fallen to 10s. 6d.
From the approximately 3,750 copies printed of Esmond in three volumes, Thackeray made about £1,360, Smith made an estimated £640, and production costs equaled around £1,100. Gross sales brought in, then, about £3,100. By comparison, ten years later Charles Dickens's Great Expectations in three volumes cost just over £1,000 to produce the same [196/197] number of copies (3,750), grossed £3,250, of which about £1,960 went to Dickens and about £250, including a commission of 7.5 percent and profits, went to the publisher, Chapman and Hall [Patten, p. 385].
Most of Thackeray's books fell off in sales relatively quickly after the first year of publication; sales of Esmond in three volumes dipped to an average of ten copies a year from 1855 to 1863. Smith had gambled safely on the first printing, knowing that he could rely on the circulating libraries, and he no doubt banked on Thackeray's newly established reputation as author of Vanity Fair and Pendennis [see Sutherland, Novelists, pp. 12-30, esp. 15-17, for an account of the stabilizing influence of circulating libraries]. When the first edition sold out in one and a half months, Esmond proved to be a best-seller, but a second printing of 1,000 copies was still a shrewd decision by Smith. True, he had not anticipated a second printing and, therefore, had not asked the printer, Bradbury and Evans, to prepare stereotyped plates; consequently, he had to pay for recomposition for the second edition. Because the ledgers show that stereotyping cost nearly as much as composition, however, it is clear that except for the savings resulting from smaller print runs, stereotyping did not become economical until a third printing was required. No third printing was at all likely for a novel in three volumes, selling at a guinea and a half and readily available from the circulating libraries - and none was required for Esmond. (Copies of the book with title pages indicating a third edition are merely new title pages attached to second-edition sheets.) Smith might have done better to order a second edition of 750 copies (many three-deckers had first printings no larger), but that could have been too conservative a figure. Within two months 600 copies of the new edition were sold, and already there was a small profit to share between author and publisher. Renewed sales, spurred by the news of Thackeray's death in December 1863, nearly carried off the remaining copies in 1864.
Although Thackeray complained several years after the publication of Esmond that a review by Samuel Phillips in the Times in 1853 had stopped its sales (Letters 4: 125), the three-volume edition had been a very good venture, exceeding Smith's original expectations and far outstripping the sales and profits of most three-decker novels.63 The few hundred copies remaining and slow sales match the residual sales of all Thackeray's [197/198] first editions and is remarkably like the sales record of Charles Dickens's Great Expectations in three volumes, which sold 3,461 copies in the first year, 39 the next, and then stopped dead with 247 copies remaindered at 2s. 3d. apiece three years later [Patten; p. 385].
Esmond in One Volumen 1857 Smith knew he was not making the money he could for himself or his author with this valuable dormant property. Bradbury and Evans was reprinting Thackeray's early magazine pieces in the Miscellanies; successful cheap editions of Vanity Fair and Pendennis were out, and The Newcomes had already proven to be a substantial money-maker. Smith was ready to work the property and took his cue from Bradbury and Evans.
The cheap edition of Esmond in one volume has 464 pages of text and 16 pages of preliminaries. It is composed of thirty octavo gatherings (eight leaves, sixteen pages per gathering). The book was printed on paper large enough to accommodate two octavo sheets (sixteen pages) at a time. It took 300 reams of long primer paper or double octavo (both terms appear in the record) to produce 10,000 copies of the book. Either of two methods of printing could have been used: work and turn with both forms of one gathering or with the inner forms of two gatherings imposed together and the outer forms together. In either case, each sheet would have been cut in two before binding. It should be noted that Smith, Elder acquired its own printing shop in 1857, having previously depended on other printers, notably Bradbury and Evans, for book production. Esmond was reprinted on Smith's own presses. Publication was announced on 17 October 1857 [Harden, Esmond, p. 87] though the ledger accounts are dated November and the book itself bears the date 1858.
It might be said that Smith overestimated the demand for a cheap edition of Esmond, since in June 1863 the firm's warehouse still held 3,320 copies out of a first printing of 10,000. Yet there was every reason to bet high on Thackeray in 1857. Although Bradbury and Evans lived to rue the euphoria of that year, claiming to have lost thousands on The Virginians, the amazing thing is that Smith did not lose a penny. In the first month and a half the company had sold 3,528 copies, enough to pay all its costs for the cheap edition of Esmond and divide £6.17 with Thackeray. In the next year author and publisher divided a profit of £88.9, and in the next they shared an even £100, after which they still had half the edition in hand. Far from [198/199] being a drug on the market, the remaining stock was like money in reserve, for the author was Thackeray and the publisher had a long-term commitment to the property. Thackeray's unexpected death sent sales soaring, and by January 1866 Smith was preparing a new edition for a new printing of 1,000 copies. By hindsight one could say that Smith did not bet high enough in 1857, for he failed to order stereotyped plates, which he could have used for the January 1866 edition and for a second printing of 1,000 more copies in September of the same year. By 1866 Smith had completed his acquisition of the Thackeray literary empire by purchasing copyrights from all other owners, so he did not have to share profits with anyone. In eighteen months, by June 1867, he had recovered costs on the 1866 edition of Esmond and made another £57.12, plus the publisher's 5 percent commission figured on total income.
The available records are incomplete, but one can estimate that the cheap edition of Esmond made between £800 and £1,200 of profit from 12,000 copies produced in three printings over an eight-year period on an investment of under £700, all of which was recovered within two months. By comparison, the profits for Bradbury and Evans's cheap edition of Vanity Fair over a thirteen-year period came to about £1,300 on 22,000 copies produced in eleven printings and a remainder in 1865 of 681 copies.
Last modified: 8 April 2001