University of Virginia Library

The Printing of Pope's Dunciad, 1728
by
David L. Vander Meulen

For more than a century the most puzzling bibliographical problem of Alexander Pope's satire The Dunciad, and the one which has stimulated the most spirited but not always the most pleasant debate, has been the question of which of the 1728 impressions was printed first. The issue first rose in the pages of Notes and Queries in the 1850's; the discussion reached its greatest intensity in the ripostes between R. H. Griffith and T. J. Wise early in this

century; and in the 1950's the argument arrived at a stasis after a leading modern bibliographer, David Foxon, confirmed some of the earlier findings. Though a belief in the precedence of the 12° impression over the 8° has now become orthodox, it is an acceptance based mainly on the faith that a few scattered typographical variations are sufficient to reveal the printing order. In the pages that follow I shall attempt to establish the order of the impressions more conclusively, relying not on newly available but rather on freshly examined evidence, particularly that of running titles. This evidence, combined with the insights provided by resettings of portions of the text, also clarifies the printing of the three additional impressions in 1728 by the Dunciad's first printer, James Bettenham.

I

William Thoms, the editor of Notes and Queries, made future discussion possible when he compiled the first checklist of versions of The Dunciad.[1] He cited three variants of the 1728 "first edition": A, a 12° with the first line reading "Book and the man . . ."; B, an 8° with that same first line; and C, a 12° beginning "Books and the man. . . ." He decided that, because of the size of the type page, A (a 12°) probably preceded B (the 8°), and that because of various "corrections" C (the second 12°) was the latest of the three.

Thoms's assessment of the order of those early printings remained authoritative for over half a century. Then, in 1915, Reginald H. Griffith presented the initial results of his consideration of the problem. He challenged the existence of Thoms's A, an objection which subsequently has not been overturned, and he hesitantly suggested that variant C, the 12°, may have been printed before the 8°, item B. [2] By the time the first volume of his Alexander Pope: A Bibliography appeared in 1922, he had considered the matter more thoroughly. He admitted there that the evidence remained ambiguous, but after a lengthy examination of the variations between the two impressions he decided that the 12° probably came first and that the 8° was intended as a large-paper issue.

Griffith's opinions were soon challenged by one of the most respected bibliographers of the day, Thomas J. Wise, who also had been studying the 1728 Dunciads. In the fourth volume (1923) of The Ashley Library he reinterpreted the ambiguous evidence that Griffith had cited and argued that the 8° came first and that it appeared in both large- and small-paper issues.[3]

Griffith responded in an appendix to the second part of his bibliography (1927); while remaining tentative about his own hypotheses, he was unequivocal in rejecting Wise's conclusions. In 1931 Wise published the Pope section of his Ashley catalog as A Pope Library. Most of the material was reprinted verbatim, though Wise made numerous minor changes, especially where Griffith had challenged him, and he expanded his defense of the existence of a small-paper 8°—an argument based, ironically, on a copy which by then had passed to the Wrenn Library under the curatorship of Griffith in Austin, Texas.[4] But overall Wise merely reiterated his earlier arguments and ignored Griffith's specific objections. Griffith had the last word: drawing on some points raised in an article by W. K. Chandler,[5] he provided his definitive statement on the problem in 1938, the year after Wise's death. His tentativeness now disappeared: "the princeps of the Dunciad was the duodecimo."[6]

The controversy then lay dormant for another twenty years until David Foxon revivified it. In a 1958 TLS essay[7] he examined several of the variations between the issues and decided that, for the parts of the book where they were found, the 12° had come first. Foxon's analysis has been influential as the persuasive argument of one of the foremost modern bibliographers of Pope, though Foxon himself noted that his insights did not yet solve all of the problems. The most valuable contribution of his study has proven to be not his conclusions but two features of his methodology. He was the first to use the Hinman Collator to examine copies of the book, and he was the first to acknowledge that priority for one part of a copy does not entail priority for all parts of that copy.

II

The relationships among the impressions of The Dunciad printed by James Bettenham can be seen most clearly on the accompanying chart, which lists the appearances of running titles and indicates resettings of the text. Both the determination of resettings and the identification of recurring running titles were made by comparison of copies on the Hinman Collator. The chart employs conventional methods of presenting the information whenever possible,

but because no conventions exist for some of the features, an explanation of the chart is in order before proceeding to an analysis of the patterns.

The four main columns of the chart represent the different impressions. As a way of condensing and clarifying the arrangement, the first column combines the 12° and 8° issues of the first edition; those issues are from the same setting of type throughout, and the running titles and their positions with respect to the text are identical. Combining them also eliminates the need to decide their order before the priority of one has been demonstrated. On the far left are lists of the page numbers and the signatures for the volumes. The signature letters in parentheses are those of the 8° issue of the first edition; those without the parentheses are for the remaining copies, all 12°. The pagination is the same in both formats.

Across the top are the "edition" numbers according to the books themselves and also the designations given to these Dunciads in Foxon's English Verse 1701-1750 (1975). The poem was first printed in both duodecimo and octavo formats (Foxon P764-765), with the octavo apparently intended as a large-paper issue. A 12° reimpression with the words 'The Second Edition' on the title page appeared next (P766); about one-and-a-half gatherings were reset. An additional gathering was reset in each of the following two reimpressions, both in 12° and calling themselves 'The Third Edition' (P767-768). The contents of the five impressions are virtually identical; in both formats the poetic text begins at the start of gathering B and continues to the end of the book. The collation formula for the first 12° impression also applies to those which follow: 12°: A-E6 F2. That of the lone 8° is: 8°: a 4 b2 B-G4 H2. This chart of running titles ignores the preliminaries ($A, a, and b) which contain only a few running titles, all of which remain unaltered through the series of impressions.

Inner and outer formes are not distinguished, for the half-sheets almost certainly were imposed for printing by the ordinary work-and-turn method. Thus, all pages of a gathering were on the press at the same time, and distinctions of inner and outer skeletons are meaningless. That method of imposition is difficult to prove equally well for each gathering, both because much of the evidence is available only in rare uncut copies and because by the nature of the evidence conclusive proof is available only for the other well-known way of printing half-sheets, the working of two different gatherings together. Nonetheless, over various copies of the 12° impressions at least two and sometimes more tests can be successfully applied to each of the half-sheets, and that evidence is always consistent with imposition for work-and-turn. Evidence for the lone 8° is scant, but it too indicates work-and-turn.

Briefly, the evidence in the 12° is this. When two half-sheets are worked together, there ordinarily will be an approximately equal number of cut and deckled fore-edges on $1 and 6 and on 3 and 4; in work-and-turn, the fore-edges of $1 and 3 will always be deckled and $4 and 6 cut. In the sheets in the Dunciads, all examples are of the second group. Secondly, paper machined by the first method will, on the average, have an equal number of

watermarks on $1, 3, 4, and 6; work-and-turn sheets will have them only on $3 and 4. All watermarks in these Dunciads are in fact on $3 and 4. Point holes follow a similar pattern. They appear with nearly equal frequency on $2, 3, 4, and 5 when two half-sheets are worked together but only on $2 and 3 in work-and-turn. In Dunciads, they occur only on $2 and 3. (In gathering A of the second edition they appear on $1 and 2; apparently the opposite end of the sheet, here containing the frontispiece, was intended as the cut-off. The position of the holes is nevertheless consistent with work-and-turn.) When two half-sheets are worked together, the first-forme impression will be consistent, unless the sheet was produced by concurrent printing; in work-and-turn, the first-forme impression will be evenly divided (in a sufficient number of examples) between the outer and inner formes of the gathering. The first-forme impressions in the Dunciads vary; that feature is again consistent with work-and-turn printing. The accumulated agreement of these indicators strengthens the suggestions of them individually; moreover, the absence of any counter evidence, in situations where half of the examples might be expected to vary, makes persuasive the claim that the 12° Dunciads were printed by the work-and-turn method of half-sheet imposition. Most of these tests cannot be applied to the 8°. Nonetheless, the first-forme impressions here also vary, suggesting that the 8° was also printed this way.

The running titles themselves are represented on the chart by roman numerals—capitals for those on versos (here, 'The Dunciad.'), and lower case for rectos ('Book the First.', 'Book the Second.', or 'Book the Third.'). An asterisk before the running title number indicates that that running title is being used for the first time. Because the chart represents a completed assessment, it has not been possible to avoid indicating which of the two examples of the same running title in a particular copy appears first. But in order not to exclude evidence which conceivably could sustain a different interpretation, I have also noted second examples of running titles in a particular impression; I have enclosed the asterisks for them in parentheses.

The number in parentheses following a roman numeral identifies another page on which that running title is used in that impression (on the chart, another page in the same column). An equals sign before a roman numeral indicates that that running title is identical with the one on that page in the previous impression. Under special circumstances, only part of the recto running title 'Book the. . . .' may be the same from one impression to the next. Sometimes the last word of 'Book the First', for example, is changed to 'Third' when the running title is reused later in the book. When the running title is subsequently restored to the beginning of the book in the following impression, a different setting of 'First' may be used. As a result, the running titles on a given page of two impressions may be identical only in part. To indicate this partial agreement, I have added the fraction ⅔ in parentheses following the equals sign. Furthermore, the presence of fractions incidentally indicates that some interruption occurred between the printing of those

pages in the two impressions. I have not indicated the normal replacement of the last word in the running title within a single impression; such a shift can be assumed, and a record of it would merely clutter the chart. In a couple of cases I have been unable to determine with absolute certainty that certain running titles in the same impression are identical; I have then added a question mark behind the running title number or the cross-reference pagination.

The chart also indicates resettings of the text. Although such changes are not inevitably linked with the running title pattern, the chart proves to be a useful place to record them, for every page of the poetic text is listed here. Moreover, there is in fact a frequent connection between textual resetting and the running titles, and the juxtaposition of the two patterns makes clear those links. The symbol used to indicate resettings is the number sign, #. Although it is a familiar sign, it is not used elsewhere in bibliographical description. Furthermore, its form may suggest its function; this symbol which indicates non-agreement of texts is similar to the "not equals" figure. The addition of a footnote is not here considered a resetting.[8]

III

Earlier efforts to establish the order of the impressions have often depended on attempting to trace textual changes through the various printings. The argument which follows also depends on evidence generated by the juxtaposition of varieties of the text, but now it is primarily evidence of resettings and of the recurrence of running titles. For ease in reference I have used the edition designations indicated on the title pages of the various impressions. The term edition is not strictly accurate, for although each impression has some material reset, the new portion is never enough to qualify the book as a new "edition" by customary standards. Despite that slow change, however, most of the poetic text has been reset once or more by the end of the five impressions.

The core of the argument for establishing the priority of one of the impressions of the "first edition" is quite simple. Two features of these impressions are basic: that the settings of their headlines and type pages are the same, and that a basic set of running titles which appears in these impressions is reused elsewhere in them as well as in the other printings of 1728. The identity of the text and of the sequence and positions of the running titles throughout the 12° and 8° indicates that individual sections of the two impressions were printed without any intervening dismantlement other than that required for imposing the pages in a new format. But because some of the headlines eventually were removed so that the running titles could be

reused elsewhere in the book, it is evident that the pattern of printing was not all of one of the formats followed by all of the other. Rather, there had to be some alternation between the formats.

The argument proceeds with the establishment of the temporal relationship between the printing of the sections of the first edition which use the same set of running titles. For the moment the distinction between the two formats is not important, for we know that, whatever the order of the formats with respect to each other, the corresponding sections in the two were printed consecutively. In the two impressions of the first edition, the group of running titles used for pages 1-16 is also used for pages 37-48. In the second edition, pages 1-16 have been reset and their running titles changed from those in the first, but pages 37-48 are identical both in their text and in the order and position of the running titles. Variant press figures confirm that pages 37-48 are in fact from a separate impression rather than from an overprinting of the first edition. It is virtually certain that of the two parts of the first edition in which the same set of running titles appears, the part which corresponds precisely with the parallel section of the second edition was printed closer in time to the second edition than the part which is not identical. Thus, pages 37-48 were printed later than pages 1-16. If the opposite had been true, that is, if pages 37-48 were printed first and then their running titles transferred to pages 1-16, where they appear in a different sequence, then only a mind-boggling coincidence could explain how the running titles were rearranged in exactly the same order and position with respect to the text when they were transferred to pages 37-48 of the second edition.

Pages 1 through 36 of the first edition contain the original appearances of the running titles which are used throughout the 1728 Dunciads. Some of these—the ones in the first sixteen pages—are, as we have just seen, reused later in the first edition, but the rest are not. In normal printing house practice, the compositor set running titles for each page as he went along when existing ones were not available.[9] Because the running titles as well as the other parts of the skeleton were subsequently reused by being transferred to other formes, it was logical to stop the series of new running titles at the end of a gathering. In the first Dunciad, the break between new and reused running titles occurs after page 36—midway through gathering F in the 8°, but at the end of D in the 12°. If the compositor had prepared the gathering which includes page 36 for initial imposition in 8°, he would have set new running titles for the first four pages and then used four from earlier gatherings to complete the forme. A simpler explanation, and one which fits better the ordinary pattern of compositorial work, is that he composed the running titles for a 12° imposition, preparing sufficient new ones to complete the gatherings he was working on. By that reasoning, pages 25-36 would have first been imposed and printed as gathering D of the 12°. Likewise, because

the presence of reused running titles matches exactly the extent of gathering E in the 12°, those pages, 37-48, were most likely first imposed for that 12° format instead of for part of F and all of G in the 8°. 8° E (pp. 25-32) would thus have been printed after 12° D, and 8° G (pp. 41-48) after 12° E. 8° F (pp. 33-40), which includes pages from both 12° D and E, would have followed both. Given that set of relationships, two types of sequences are possible: either these two 12° formes were printed before any of the three 8° ones, or else after the first of these two 12° gatherings was printed, the one complete 8° gathering within it was machined before the other 12° forme. I am inclined to think that all of the 12° preceded all of the 8° here; in light of the press figures in these sections, the other sequences would mean that at some point the printers chose to keep a press waiting while the 12° pages were reimposed for the 8° rather than to print the remaining 12° forme which could have been imposed while the first was printing. Although oscillation between the formats is possible, the succession of all of one by all of the other is more probable for this section. There is, moreover, some basis for determining the relative order of 12° D and E. It is reasonable that the pages which continue the series of new running titles were printed before those with reused ones; if so, 12° D preceded 12° E.

Gathering E of the 12° employs running titles first used on pages 1-16— that is, B and part of C in the 12°, or all of B and C in the 8°. The running titles in 12° E would have been collected from the later of the two impressions of those early pages. Because they correspond exactly with an integral division in the 8° volume, they seem to have been drawn from the 8°. In contrast, those running titles of 12° E match all of B but only the first four pages of C in the 12°. It would have been another amazing coincidence for those particular running titles from 12° C to be selected for use in 12° E. It is apparent, therefore, that because the running titles of 8° B and C were transferred to 12° E, all gatherings of the 12° of which 8° B and C form a part must have been printed before the 8° ones; in other words, both 12° B and C preceded 8° B and C. Because pages 1-24 (12° B and C) were thus first imposed for 12°, gathering 12° C was almost certainly printed before 8° D (pp. 17-24) as well. Furthermore, it is likely that the running titles reused for pages 37-48 were the first ones that became available. If so, then 8° B and C were run off before 8° D.

The running titles in the closing pages of the book, 12° F or 8° H (pp. 49 ff.), remain the same throughout the 1728 impressions and offer no clues for determining priority. The imposition order for those final pages is conclusive from earlier investigations, however. Foxon observes that in the 12° the direction line on page 49 lies immediately below the footnote, while in the 8° and in the second edition a lead has been inserted between them. Because these pages of the second edition were not reset, page 49 apparently is closer in time to the 8° than the 12° ("Two Cruces," p. 52). Chandler made two useful discoveries about the same gathering ("First Edition," pp. 68-69). On page 51, line 273 in the 12° and in most copies of the 8° has the phrase

'B--- longsole judge'. But a copy of the 8° in the Wrenn Library at Texas as well as all copies of the second edition have 'B—n longsole judge'. Later on the page, line 277 in the 12° has the word 'townlongs-mens'; both the 8° and the second edition drop the long s in favor of 'towns-mens'. For this gathering as well, therefore, the 12° preceded the 8°.

Traditional printing house procedures as well as patterns within this particular work provide some basis for speculating that gatherings b2 and H2 of the 8° were printed as a single half-sheet. It obviously would have been easiest for the printer to combine them and end up with a whole number of half-sheets for the book. The other gatherings of the poetic text were printed in alphabetical sequence; that pattern, coupled with the usual practice of preliminaries being printed last, would support a chronological link between the two quarter-sheets as well as suggest that they were printed after the section that ends with page 48 had been completed. The fact that new running titles were produced for pages 49-51 does not weaken the theory; it would have been logical to compose these three separately rather than to plunder the set of a full gathering which could otherwise be transferred as a unit to another forme. Paper evidence also suggests some connection between the beginning and end of the book. Of the four varieties of "BF"-marked paper here, two occur almost exclusively in gatherings B-E, and two in F-H and a-b. Finally, the evidence of point holes, deckle edges, tranchefiles, and watermarks reveals that the quarter-sheets were machined as they would be for joint imposition in a half-sheet: the first leaf of one gathering (H) always falls at the outer edge of the full sheet, and the first leaf of the other (b) on the inside.

The priority of the imposition of the rest of the preliminaries remains problematical. The running titles for this section, "The Publisher to the Reader," are invariant throughout the five 1728 impressions, and no typographical variations appear in the text. Certain speculations in light of what is known about the rest of the book are nonetheless possible. The pattern in the other gatherings immediately suggests that here too the 12° preceded the 8°. It is true, as McKerrow points out, that "an edition in which the signatures are all of one alphabet, beginning with A and proceeding regularly, is likely to be later than an edition in which the preliminary leaves have a separate signature." That would argue for the priority of the 8°. But he also says that "A" was often saved for the preliminaries and that a preliminary sheet signed A when the text begins on B tells us "nothing at all."[10] According to the signature pattern, then, the 12° could have been the first impression, and the preliminaries could have been composed and printed at the customary time, after the rest of the book. If the pages that were distributed first were the ones that had been printed first, we have another hint that the preliminaries were not worked at the start of the press run; they remained intact, for use in the second edition, while the equivalent of 8° B, C, and some of D were distributed and then reset.

With the actual sequence of these impressions now in mind, it is interesting to take another look at earlier arguments about the priority of editions. The spirited debate to a large extent turned on the question of whether a clearly corrupt or a correct form of the text came first. An important group of variations consisted of those that either could have been caused by type movement when the format was changed or else might have been corrected at that point. All but one of the variations, it turns out, were from good to bad. The extensive type movement on the opening pages of the text, along with the dropping of the 'S' of 'BOOKS'; the two raised letters in the footnote on page 5; the removal of the exclamation point after the catchword on page 8; and the dropping of the final letter in the catchwood 'As' on page 9 all resulted when the poem was reimposed for octavo. Only one error of this kind, the omission of the final letter in the catchword 'This' on page 29, seems to have resulted from a mistake in composition (or from the dislodging of the letter early in the press run) and to have been corrected for the second impression. Yet while the octavo text is inferior at these points, it contains seven corrections of material whose states are independent of type movement caused by the opening of the formes. The pattern of changes thus suggests that sheets of the 12° impression served as the proof sheets for the text of the 8°, with the appropriate revisions made between impressions. The 8° seems to have received no separate proofing; that omission allowed the errors caused by type movement to enter.

IV

Features which are instrumental in determining the printing order of the first two impressions of the 1728 Dunciad, particularly resettings of the text and the recurrence of running titles, also provide helpful insights into the production of the following three impressions. The most important principle involved in interpreting such evidence is this: when a group of running titles appears twice within the same work, the group which matches the corresponding part of the preceding impression was printed first. In these particular cases that principle is corroborated by two probabilities involving resettings of the text: the type distributed first is likely to have been the type which was used first, and pages from standing type are likely to have been printed before those which were reset.[11] Thus, sections for which the setting is identical with the corresponding part of the previous impressions are likely to have been printed before the sections that are reset.

The remaining impressions of the 1728 Dunciad—the "Second Edition" and the two impressions bearing the designation "Third Edition"—are all in duodecimo. The discussion which follows deals only with the bulk of their poetic text, gatherings B through E. The preliminaries (gathering A) and the final three pages of text (in F2) retain the identical running titles and text

and therefore, as in the first two impressions, provide little evidence for priority.

As the chart shows, pages 1-18 of the second edition are newly set, with running titles which do not match those on the corresponding pages of the first edition. Pages 25-48, or gatherings D and E, are usually from standing type and identical in their running titles. The only exception is D4v (p. 32), where the text is reset and the running title is taken from elsewhere in the book. Running title XVI, which appeared on that page in the previous edition, never shows up again. Although line 325 is now wrongly identified as 525, there are no other textual variations between page 32 here and in the first edition; the page was probably pied between impressions. The pattern of running titles and resettings clearly indicates that gatherings D and E were closest in time to the first edition, and thus before B and C, where the running titles reappear. Moreover, E came before D; the running title used for the apparently pied D4v was taken from E4v, where it also occurred in the first edition.

The running titles from D were then transferred to B, retaining their sequence for the most part. B1r required no running title, so the one from D1r could not be used in the parallel spot in B. Instead the compositor placed it on the adjacent page in the forme, B6v. That was wrong. B6v should have had a running title from another verso, one which indicated the book number. The error remained in edition 3a, where gathering B employed the same running titles and text, though it was corrected in 3b, where the headlines occur in a different sequence and where the text is reset. The appearance of the anomalous running title on B6v of the second edition meant that there was no place for the one that had occupied a parallel spot in the group being transferred to B. As a result, running title XVIII (from D6v) was apparently discarded; it does not recur in this or any subsequent edition.

Gathering C is the most complicated one of the second edition. Some of the text was reset, and some imposed from standing type. Some running titles were newly composed, others were transferred from elsewhere in the book, and still others were received with the standing type of the previous edition. Gathering C in the first 12° edition contained the type used for part of 8° C and all of 8° D. Whereas the running titles of 8° B and C were transferred to a point later in the first edition and those pages of text distributed, the running titles of 8° D were not needed elsewhere. As a result, 8° D and its running titles seem to have remained standing longer, with consequences for the second edition. Pages 21, 22, and 24 (that is, C5 and C6v) with their running titles were transferred intact. The running title and much of the text of page 20 were also transferred, though the irregular correspondence of the text to that of the first edition suggests that it may have been salvaged from partially pied type. To make room for a footnote on page 22 in the second edition, the compositor moved the final two lines of text from 22 to the top of 23 (C6r). Those lines, as well as possibly a few other segments of the text, are thus from previously prepared text, but most of the page is reset. The running

title is changed; the fact that, unlike its original companions, the running title used on this page in the first edition never appears again suggests that it—and the rest of the page—were pied. One opportunity for the introduction of such disorder would have been, of course, the point at which the two lines were transferred.

In contrast to those pages from 8° D, the ones that correspond to 8° C (pp. 13-16, or C1-2 in the second edition) are all reset, as are pages 17 and 18 (C3). Running titles from E of the second edition are used on most of the newly set pages in C. Perhaps distribution of E had already started when those running titles were needed, for while some new ones occur here (pp. 18 and 19; 3v and 4r), not all of those which should have been available from E are used. Those unused ones, moreover, never again appear in the series of Dunciads. In at least one case, however, a "new" running title in C is similar enough to one in E (XX on p. 18 resembles VIII on p. 48) to suggest that the earlier example persists, albeit with a couple of sorts or the internal spacing changed. Page 15 (B2r) is the beginning of Book II and does not contain an actual running title, but the new setting of part of the division head ('Book the Second.') here and in the following impression is later used as one. Therefore it is included in the chart, but it is placed in brackets.

The pattern in the first of the two third editions is less complicated. The running titles and text of gatherings B and C are identical to the second edition. E, which was one of the first parts of the second edition printed and very likely also distributed, is completely reset in edition 3a. The running titles in E are those from C, though in a different sequence. The division heading from C2r now assumes the status of a full-fledged running title on E1r. The running titles in D are from B; here the order is the same, except that the running title on 1r, a leaf that had none in B, comes from 6v, a leaf that appears without a running title in D. D6v could do without one because it contains a division heading, the start of Book III. Moving the running title from B6v also removed the anomaly of the incorrect one which had been introduced in the second edition.

Although the running titles in D are the same in the second and third editions, that parallel is fortuitous. Those from D of the second edition had been moved to B of that edition; that arrangement of text and headlines was reimpressed for the third edition, and then the running titles were moved to gathering D of that edition. Apart from that trail of the movement of the running titles, the clearest evidence that D in the third edition was not printed directly from the setting of D in the second is that in recto running titles the book number is sometimes from different settings of type. When the running titles were first moved to an earlier place in the second edition, the word 'Third' was replaced by 'First'; when they were again moved to D in the third edition, 'Third' was reinserted. It thus became possible for the bulk of one running title to be identical with another one but for the final word to be from a different setting. As additional evidence that some interruption has taken place between the printing of the second and third editions of D,

the spacing between the running title and text on a given page often varies from one edition to the other.

The features of running titles and settings of the text in the final Bettenham printing, the second one to be called "Third Edition," are similar to those of the previous impressions. Gatherings D and E have the identical text and running titles as the comparable sections of edition 3a, in which they were the last gatherings printed. Because running titles from both D and E appear elsewhere in 3b in positions that do not match 3a, D and E were evidently the first sections of 3b to be printed.

The running titles of B, which was newly composed, and C, printed from standing type, are for the most part from D. In C, the running titles on leaves 1, 2v, 3-5, and 6r follow the sequence of D. C2r has no running title, so title xii from D2r is not used. On the other hand, D6v had no running title to provide for C6v, so the compositor instead took the one from E6v. The running titles from C were then moved to B. Because in B page 1r rather than 2r has no running title, the one from C1r is moved to B2r. Most of those in B are the same as those in the previous edition; the set of 3a B was moved to 3a D, it recurred as 3b D, and then moved to 3b C and up to 3b B. But as with gathering D in editions 2 and 3a, the book number in recto running titles is a different setting.

It is possible that the movement of running titles in 3b might have been from D to B to C rather than D to C to B. That would mean that as the compositor transferred the running titles from D to B, he departed from his careful practice of transferring running titles to identical positions in B in the single case of B2r, where he would have inserted the one from D1r rather than the expected D2r. The consistency of the rest of the pattern makes that seem improbable. Moreover, the order D-C-B fits with what might be expected concerning the printing order of a forme of standing type and a newly set one: it is likely that the standing C was worked before the reset B. The resetting of B also provides a clue about the printing of the early gatherings of 3a. If the one of two formes which was distributed first had also been run off first, then 3a B was printed before 3a C. But in 3b, in any case, it seems that the actual order was indeed D-C-B.

The neat identities between gatherings in adjacent impressions raise the possibility that portions of the books were produced not in the way described here but by overprinting: as the presswork for one of the "editions" was nearing completion and the need for another one became evident, extra copies of the first could have been machined at once instead of being separately impressed for the next issue of the poem. But evidence from changed ornaments, press figures, and paper indicates that such was not the case for any gathering in any of these Bettenham impressions.

In summary, then, the printing of The Dunciad of 1728 most likely proceeded in the following order:

The sequence of the remaining sections of the first edition is not as clear. Certainly 12° F was printed before 8° H, and very likely 8° H and b were produced together. It is also possible that both the final gathering and the preliminaries were printed after the rest of the book. What is evident, however, is that the printing alternated between the 12° and the 8°, and that each section of the poetic text of the 12° was run off before the corresponding part of the 8°.

The point at which gatherings A and F were printed in the subsequent impressions is not apparent either, apart from inferences generated by the pattern within the rest of each impression. But the sequence for B-E, or the bulk of the poetic text, seems to be as follows:

• Edition 2: E; then D; then C and B (with C probably first).
• Edition 3a: B and C (with B probably first); then D and E (with D probably first).
• Edition 3b: D and E; then C; then B.

In each of those later impressions at least one gathering was reset, according to the following pattern:

• Edition 2: B, most of C (1-3, some of 4v, and most of 6r), and D4v.
• Edition 3a: E.
• Edition 3b: B.[12]

Notes