The Early Editions of Rob Roy

Abstract

According to their title pages, all dated 1818, there are four numbered editions of Sir Walter Scott’s three-volume novel Rob Roy.¹ Despite the title-page claims, however, there are in fact not four editions but only two, one comprehending editions 1–3, the other, edition 4. My concern here is with editions 1–3. (I continue to use ‘edition’ as a matter of convenience, though in bibliographical terms editions 1–3 are clearly three issues of the one edition—i.e. all are substantially from the one setting of type.)

The practice of issuing factitious editions derived from a single setting of type may be not without precedent, but it is one seemingly introduced in James Ballantyne’s printing house in Edinburgh in late 1817 in the course of working off Rob Roy. This article is an attempt to clarify the processes involved in the printing and publication of the first three editions of the novel. Much the most exhaustive account of the production of these editions, on which I have drawn extensively, is presented by Professor David Hewitt in the ‘Essay on the Text’ included in the apparatus to the Edinburgh edition.² That account is complemented by the formal descriptions (including commentary) to be found in the bibliography of Scott by W. B. Todd and Ann Bowden.³

All twelve sets⁴ on which the present article is based are from the Poynton Collection, Special Collections, Baillieu Library, University of Melbourne. Since the total edition number (i.e. the sum of the three editions) amounts to 10,000, the sample represents only 0.12 per cent, so that some of the conclusions reached here may need to be modified should a larger sample become available.

The three editions (T/B 112Aa, Ab, Ac) were printed by Ballantyne for Archibald Constable in Edinburgh and Longman, Hurst, Rees, Orme, and Brown in London. According to contemporary newspaper advertisements, recorded by T/B, they were published in Edinburgh: 1st, 30 December 1817; 2nd, 7 January 1818; 3rd, 10 January 1818. Hence, at first sight, there is a conundrum: How could two further editions be set and worked off in the span of less than the two weeks that followed publication of the first?

Truth to tell, as already indicated, the difficulty is illusory: both Hewitt and T/B recognize that the three editions were (with a smattering of re-set gatherings) printed from a single setting—i.e. ‘SECOND EDITION.’ and ‘THIRD EDITION.’ were inserted in the title pages at press, thereby creating the three editions, in these tranches: [1st]5000:[2nd]2500: [3rd]2500.⁵ If the normal processes of printing in the late hand-press period had been followed, involving perfecting consecutively (i.e. printing in turn, in its total edition number, one forme of a sheet followed by the other forme), all 10,000 sets, together comprising the three editions, would have to have been printed by 30 December in order to be available for dispatch to agents on that day, in sheets. In other words the brevity of the intervals in publication between editions would not be material, since all 10,000 would have been available simultaneously. However, if the normal processes were to have been followed it is doubtful that initial publication would in fact have been possible on 30 December, because Ballantyne did not receive author’s marked-up proofs of the last four gatherings of volume 3 (M–O¹² P⁶) until 9 pm on 26 December (a Friday),⁶ with publication advertised for the following Tuesday.⁷ Little correction was required, so that within a quarter of an hour ‘they were all at press’, implying that printing then began and, in view of the lateness of the hour, that it continued through the night.⁸ But, even with printing continuing overnight, would it have been possible to complete the printing of 10,000 copies of seven formes⁹ in two further working days (Saturday and Monday), given that the text was not set in duplicate—and no matter how many presses were available, no matter by how many hours the day was lengthened through the pressmen working overtime and even overnight? The answer is, almost certainly not, at least not if following normal processes.

That the projected date of initial publication, 30 December, was met can be confirmed: on that day 1,666 sets—a third of the first edition—were sent by sea to Longman in London, and on the same evening and the following day an unspecified number were sent by coach or wagon to various destinations—whether 3,333 (the residue of the first-edition total of 5,000 remaining after the dispatch of the Longman consignment) or some other quantity is not known.¹⁰

Since an extraordinary effort to meet the publication date would have been required, undoubtedly extraordinary means would have had to be resorted to. The one mechanism that would certainly have made the task possible, I suggest, is concurrent perfecting—a suggestion I advance in spite of Philip Gaskell’s caveat that ‘concurrent perfecting would always have been difficult to fit into the normal complexity of work flow in a printing house with two or more presses, and if it happened at all it is unlikely to have been more than an exceptional resort in cases of urgency when the speed of two-press operation was necessary’—precisely, I submit, the case with Rob Roy.¹¹

In concurrent perfecting, the edition quantity of a particular sheet (here 10,000) is divided equally between two presses, one printing the outer forme, one the inner; the two heaps are then exchanged and printing is completed, thus producing 10,000 perfected sheets (i.e. printed on both sides) in half the time that it would have taken to print them by consecutive perfecting; similarly, if so desired, 5,000 perfected sheets could have been produced by concurrent perfecting in a quarter of the time that it would have taken to print the full 10,000 by consecutive perfecting.

The availability of seven presses can be regarded as a given: in 1817/18 Ballantyne had eleven presses,¹² nine of them employed, though not exclusively, on Rob Roy. The actual rate of production cannot be known, but if, for the sake of illustration, we take the accepted nominal rate of a token¹³ of 250 impressions an hour over a productive working-day of, say, ten hours we can arrive at what seems to me to be a likely maximum output:

1. Printing x copies by consecutive perfecting: 125 per hour.

2. Printing x copies by concurrent perfecting: 250 per hour.

3. Printing 10,000 copies by consecutive perfecting: 80 hours.

4. Printing 10,000 copies of the white-paper forme, 5,000 of them then perfected: 60 hours.

5. Printing 10,000 copies by concurrent perfecting: 40 hours.

6. Printing 5,000 copies by consecutive perfecting: 40 hours.

7. Printing 5,000 copies by concurrent perfecting: 20 hours.

On the basis of these simple calculations the attraction of printing 5,000 copies (or a lesser number) by concurrent perfecting is obvious. Fewer than 5,000 sets may well have been available on 30 December, either by design or by circumstance: all we know is that it could not have been fewer than 1,666.

Since P is a half-sheet we would expect it to have been printed by halfsheet imposition from a total of 5,000 sheets, producing 10,000 copies—i.e. both formes imposed together and printed in one impression, and each sheet then, when perfected, cut in half to produce two copies of the gathering, exhibiting the same characteristics as M N O (see below); imposed in this way the half-sheet P would have been printed in half the time taken to print a full sheet, whatever the procedure.

If concurrent perfecting of the final sheets of volume 3 was resorted to in order to meet the 30 December deadline for the publication of the first edition, Ballantyne would have needed initially to print only 5,000 copies of those perfected sheets—or fewer, but at least a sufficient number to enable a substantial proportion of the edition to be available on the announced date. The formes would then have been left on the presses—as suggested by the unchanged press figures¹⁴—and the remaining copies (5,000 or more, in however many lots) printed on the following days, whether by consecutive perfecting or by concurrent, thereby meeting the dates announced for publication of the second and third editions.

That perfecting was not carried out consecutively for the total edition ahead of publication of the first edition is confirmed by the fact that ‘sales were so fast that in the first week of the new year copies were not arriving quickly enough from the printer’s’¹⁵—i.e. printing continued beyond 30 December.

In the same context Hewitt quotes Cadell, writing 3 January 1818, on the need to ‘take down’ several sheets. He assumes that ‘taking down’ involves breaking up formes and distributing the type so as to use it for completing the setting of Rob Roy, but at this date nothing of the novel remained to be set. In fact Cadell’s reference is to the act of using a peel to take down printed sheets from the racks on which they have been hung to dry,¹⁶ being then ready for gathering; presumably these were copies of the final sheets of volume 3, sheets printed after the dispatch of copies of the first edition.

There can thus be little doubt but that the printing of the final sheets of volume 3 continued beyond 30 December; whether the sheets were perfected consecutively or concurrently remains to be considered.

T/B were well aware of the true nature of the three editions, but their comments are confusing, in that they use terms that can convey a misleading view of the relationships among them. My aim here is not to question their categorisation of 112Ab, the second edition, as ‘First Edition, Second Issue’, and 112Ac, the third edition, as ‘First Edition, Third Issue’. It is simply to attempt to clarify the relationships by rehearsing the evidence provided by the twelve Poynton copies and T/B’s forty-six.¹⁷

The confusion arises first when T/B describe the second edition as ‘an overrun of first edition sheets’, ‘a remainder’ and ‘the same impression’. The confusion is then compounded by the Notes section appended to volume 3 of the third edition:

Essentially in its first issue (1) four sheets (A-C, E) are further overruns of the first edition, four others (D, K-L, O) overrun from the ‘second edition’, the remainder finally of a new setting. Eventually, however, (2) another overrun sheet from the second edition is used ([with press figures] C61-10 62-5) and, in some copies, a further sheet reset (E100-1). In this further E resetting a variant reading occurs at page 97 line 23, previously ‘imminent’, now ‘eminent’.

The implication of the general tenor of their comments, particularly in the repeated reference to ‘impressions’ and ‘reimpressions’, is that the three editions are separated temporally not just in publication but also in part in printing, whereas that cannot be: with the exception of the last seven formes of volume 3 and the small number of sheets that were re-set (to be considered below), the formes of all 10,000 copies were worked off in one run, in the normal manner—i.e. by consecutive perfecting.

In containing press figures in every forme the three volumes are unexceptionable in Ballantyne’s output at this period. Possibly significant, however, is the disposition of the figures in the seven formes constituting M– P: no figure is represented more than once, thus lending support to the assumption that, as would admittedly have been anticipated, the seven formes were being worked off simultaneously, by seven presses—i.e. by concurrent perfecting.¹⁸

A record of the press figures appearing in a publication may serve various purposes, but notably—when figures in particular formes differ among copies—by drawing attention to the possibility that variation in figure may accompany variation within the page/forme, whether in an alternative, textually-invariant, setting or in one incorporating change.¹⁹ It is worth noting here that in Rob Roy there is an instance of the reverse, of a forme existing in two settings but bearing the same press figure on the same page (see below)—hence two inevitable observations: (a) the considerable value of the Poynton Collection in containing multiple copies of all the novels, thus allowing comparisons of text and setting to be made in situ;²⁰ and (b) the possible limitations inherent in depending on a record of press figures in isolation. In Rob Roy, with one exception,²¹ variation in figure accompanies a textually-invariant re-setting, undoubtedly occasioned by the need to make good a shortage due to miscalculation or accident that must have arisen when the number printed in the first setting ran out.²² The few re-set sheets occur predominantly in volume 3, as would be expected, since the shortages would have been apparent only after the 5,000 (or fewer) sets of the first edition had been worked off and dispatched. Nonetheless, given that the remaining sheets of volumes 1 and 2, along with all up to L of volume 3 (5,000 or more), had already been worked off—though not necessarily, as would at some stage have become apparent, in the full edition total of 10,000—in theory any particular copy of any of these thirty-eight sheets²³ would have been available for binding up as part of a set of sheets preceded by a preliminary gathering (π²) containing the title page of the second or third edition and, as will be shown, occasionally by a preliminary gathering of the first. They are not mixed editions, pace T/B, who categorize the set of the first edition in the Sterling Collection at the University of London as mixed or sophisticated (‘text for volumes 1,3 corresponding to the “Third Edition”’).²⁴

The evidence of pulled-press figures would seem to confirm the order of printing of particular formes and the proposition that a sheet from any stage of the printing process was available for gathering in a random assemblage making up a bound volume. O4^v (p. 320) of volume 2 is figured 11 in T/B and in eleven of the Poynton copies, but in Poynton copy 4,²⁵ of the first edition, both digits have been pulled—i.e. at least the outer forme of a sheet appearing in this copy of the first edition must have been printed after the corresponding forme in those second and third editions represented by the Poynton copies and, therefore, was available for binding, as here, in a first edition, but, presumably, was also available for binding in copies of the second and third editions. Similarly T/B record press figure 3 in B5^r (p. 37) of volume 2, and it occurs there in the second edition and one copy of the third in the Poynton Collection, but in Poynton copy 1²⁶ of the first edition and the other copy of the third the figure is 11 (representing a re-setting). Further examples might be cited, all illustrating the proposition, already advanced, that any sheet exhibiting an identifiable variant (here in setting, manifest in press figure) in a Poynton third edition has the potential to be found elsewhere in the second or the first, and any in a Poynton second edition to be found elsewhere in the first or the third. In other words, on the evidence of the twelve Poynton sets, such variant settings as exist are not peculiar to an edition: any copy of any volume from any of the three editions will be a random collection of sheets. That said, however, there is an understandable tendency for the few re-set sheets to cluster in the second and third editions, since copies of the first edition had, with the odd known exception, already been dispatched to the trade before the shortage was realized.

The last three-and-a-half sheets of course constitute a separate category, in that their printing could not start until the marked-up proofs had been received from the author and the few required corrections made in the readied formes. In theory all 10,000 copies of an individual forme should have the same press figure, but it is notable that the inner forme of O is recorded by T/B as figured 4/- (i.e. either figure 4 or without a figure) on O9^v (p. 304) in the first edition—in other words, a ‘late’ state, with the press figure pulled, had been seen by them in a copy of the first edition though not exhibited by any Poynton copy, regardless of edition.

As far as formal description is concerned, I suggest that in situations such as that exemplified by the first three editions of Rob Roy the most lucid form of presentation would be to treat the bibliographical edition as the unit, rather than the title-page statements, supplemented by a series of notes.

In summary, it can be said that the first three editions of Rob Roy are essentially from a common setting. Assuming that Ballantyne was unable to print all 10,000 copies of the final three-and-a-half sheets of volume 3 by 30 December 1817, the date announced for publication, those sheets are likely to have been printed, in at least the first batch, by concurrent perfecting, in a number sufficient to enable their dispatch on that day and the next of a substantial proportion, but not necessarily all, of the 5,000 intended as the first edition. Late in the run of a small number of sheets it was found that an insufficient number of copies had been printed; these sheets were re-set and bound indiscriminately in volumes designated [first], second or third edition. Though the re-set sheets are clustered in the third edition they are not peculiar to it. How the remainder of the final sheets of volume 3 beyond the first batch were printed cannot for the present be known.

As already noted, there is no set of the fourth edition (T/B 112Ad) in the Poynton Collection, but since it is from a new setting it requires only passing notice here.²⁷ It too was printed by Ballantyne for Constable and Longman. T/B report publication in Edinburgh 31 January 1818, but Hewitt believes that the advertisement in the Edinburgh Weekly Journal of 1 July 1818 is likely to refer to this edition.²⁸ None the less, the January date is plausible, in that a new edition, which had already been mooted by 2 January and ordered on 21 January, numbered only 3,000 and would have been set from printed copy.

Note: In advancing the proposition that the final sheets of volume 3 are likely to have been printed by concurrent perfecting I have been tentative. There are two, intertwined, difficulties encountered in any attempt to prove that that process was employed in Rob Roy. One is that we cannot know at what stage a particular copy of any one of those sheets was printed: as part of the initial batch (5,000 or fewer) or later, and, if later, by what process. The second is the capacity to identify a sheet as having been printed by consecutive or concurrent perfecting. A consequence of either method of perfecting is that where print coincides on the two sides of a leaf/forme the forme printed second will leave a series of hillocks on the other; demonstration of the order depends on the ability to view, in an unpressed copy, this embossing effect. In concurrent perfecting the proportion of the edition total of a sheet printed by that process—assume here that it is the first 5,000—will have the resulting hillocks prominent on $1^v in 2,500, demonstrating that the outer forme (containing $1^r) was printed second, and vice versa. The same division will apply to P if it was printed by half-sheet imposition.

Being able to identify which forme of a sheet was printed first may be difficult. Kenneth Povey devised a lamp to aid this determination,²⁹ but I have been unable to use my version of it with any confidence—whether from incompetence or the lack of a suitable environment. In the Poynton Collection there are three sets of Rob Roy, all from the first edition, that have not been pressed, so that the determination should be feasible, disregarding the first difficulty (of knowing whether the final sheets formed part of the initial run). My attempt has been limited to viewing and feeling the leaves, an exercise undertaken on two occasions, a year apart. For what it is worth, my results were identical: M in copy (i) inner forme printed first; N in copy (i) inner forme printed first; O in copies (i) and (iii) inner forme printed first (in the remaining sheets, outer forme printed first). However, even if my observations are tenable and the relevant sheets were printed as part of the first batch, the sample is too small to produce conclusive results. The scarcity of unpressed copies may well remain a constraint; none the less, concurrent perfecting represents a plausible explanation for the printing, in the allocated time, of at least the first batch of the final sheets of Rob Roy.

Footnotes