University of Virginia Library

Bibliography and the Editorial Problem in the Eighteenth Century [*]
by
William B. Todd

OF THE MANY PROBLEMS THAT HARASS THE bibliographer of eighteenth-century literature, perhaps the most vexing are those involving the proper discrimination and ordering of multiple editions. In this period undenominated reprints are to be suspected everywhere, among elaborate editions running to several volumes, among popular books of any length, even among casual works of lesser size and consequence. Though it would be futile to suggest an occasion for all of this duplication, an explanation for some may be found in a reported conversation between Andrew Millar, a prominent bookseller of the day, and several members of the clergy at St. James. When asked what he thought of David Hume's essays, Millar replied, as he later informed the philosopher, "I said I considered yr Works as Classicks; that I never numbered ye Editions as I did in Books We wished to puff."[1] From this we may presume that, for Millar and others of similar intent, multiple editions may appear as the result of a deliberate policy among the booksellers to disguise reprints of standard works. The greater the importance of a book, they must have reasoned, the less the necessity for pushing its sale, and the more the inducement for duplicating the original issue. A close facsimile would be readily accepted by those who wished to buy the first, authoritative printing. A close facsimile was, then, what they oftentimes attempted to produce.

If Millar's statement can be inferred as expressing a general policy, then we have not only been forewarned as to what we may

expect in Hume's works and advised as to what actually happened in those of Fielding's which Millar published, but given some reason for the presence of two or three concealed editions in practically every major production in the eighteenth century, and some anticipation of further discoveries and other difficulties in this neglected province of bibliography.[1a]

For the presence of other reprints, however, Millar's observation will not suffice, since it excludes all works not considered—in the eighteenth-century view—as classics. But where there are two or three hidden editions of the major works, there seem to be among certain kinds of minor productions as many as four, five, or six, all very much alike and often undiscriminated in bibliographies. The occasion for multiplicity of this order arises, perhaps, from economic rather than literary considerations. For obvious reasons, a publisher would be reluctant to invest in such ephemera as topical poems, plays, and political tracts, and would accordingly prefer to underprint, and print again as necessary, than to overproduce a lot of sheets which would then remain unsold. Paper, in these days, was an expensive item, amounting to forty per cent of the cost of production for small issues, over sixty per cent for larger ones,[2] and therefore not to be squandered on casual effusions of uncertain reception. Hence we may see why Lawton Gilliver printed four limited editions of Bramston's Art of Politicks (1729), any one of which might be considered as a "first";[3] why Thomas

Cooper, a prolific purveyor of literary trifles, found it necessary to print five editions of Lord Lyttelton's Court Secret (1741-42) before a lagging demand required a "puff" with one called the second; and why this same publisher, a year later, printed no less than six editions of Edmund Waller's and Lord Chesterfield's Case of the Hanover Forces (1742-43) prior to the publication of a "second."[4] Given the cause in economic necessity, all of this reprinting becomes credible, and should become a matter of record.

In addition to the literary and economic reasons, already presented, another and more significant occasion for reproduction in the eighteenth century is the removal of all legal restrictions upon the size and frequency of issue and the use and reuse of type. With the expiration of the Licensing Act in 1695 and the gradual abrogation of the powers exercised by the Stationers' Company, the esoteric "arte and mysterie" of printing, as it was called in the earlier period, became in this a common trade, subject to few prohibitions, and producing on a scale of greater extent and complexity than anything witnessed before. Despite the laws of copyright, or perhaps because of them, the publishers duplicated their own issues, not infrequently copied those of their competitors, and, with notable exceptions, resorted to the term "edition" for successive issues only as an expedient for moving the book. Thus for one reason or another, or for no reason at all, reprinting became a custom, evident, as I have indicated, among first editions of any sort, and, as well, in several other categories: (1) among subsequent editions properly described, as in the two fourth editions of Gray's Elegy (1751), the two second editions of Pope's Use of Riches (1733), and, if I may again refer to variants undifferentiated in the present accounts, the three third editions of Pope's Epistle to Burlington (1731);[5] (2) among authorized but undesignated reprints, as in the double editions of the 1723 Hamlet, the 1724 Othello, the 1729 Macbeth,[6] and the 1770 versions of Goldsmith's

Traveller; and (3) in outright piracies, as in the three undiscriminated 1797 editions of Burke's Letter to Portland [7] and the four misrepresented 1770 editions of Goldsmith's Deserted Village. And so the reprinting continues, to be reported for relatively few books, to be observed in many, to be suspected in all.

If multiple editions in different settings of type represent a problem too often ignored in contemporary accounts, even more of a problem, but one accorded even less recognition, is that of multiple impressions from the same setting of type. Where the seventeenth-century compositor, with a limited amount of type at his disposal, usually had to break down a setting after every sheet in order to recover sorts for further use, the eighteenth-century compositor, with apparently unlimited quantities of type at hand, could on occasion set as many as 350 pages and allow this enormous aggregate of metal to remain intact for innumerable impressions. Some of these are labeled second and third editions, some are not so dignified. Some appear with substantial textual alterations, some without the alteration of a comma. In one printing house no less than 94,900 pieces of type were composed for a certain book and then retained for a series of impressions, all published within seventeen days. The paper which went over the presses in these seventeen days I conservatively estimate at 137,500 sheets.[8]

For such extensive reimpression as this there is no precedent in the sixteenth or seventeenth century, hitherto no exposition of an expedient means for detecting it in the eighteenth, and consequently, no reliable method for interpreting the complexities in printing which have developed over the years. To illustrate the resulting confusion, let us consider the voluminous bibliographical literature on Samuel Johnson, and particularly the commentary on those of his works published by Thomas Cadell. Since Cadell was an apprentice to Andrew Millar, and later his successor, we should expect some irregularity. And something is amiss, both in the bibliographies and in the books. Of four works published between 1770 and 1775 a "second edition," not otherwise described, is recognized from its designation in the books I will call A, C, and D,

and presumed from the existence of a "third edition" in B. Actually all, including B, have at least one "second edition," but of a kind quite different from that which the term implies. In A and the unknown B it consists of a partial resetting and a partial reimpression; in C, a complete reimpression; and in D, no less than three distinct reimpressions. With the evidence now at our disposal, and shortly to be divulged, all of this is easily perceived. Without it, the second edition of B remains "suppressed," as we are now informed,[9] the three second editions (i.e., impressions) of D undiscriminated, and the problems inherent in all of these entirely undisclosed.

From this example of misinformation we now turn to bibliographies in which, though more may be expected, occasionally less is fulfilled. In S. L. Gulick's account of Chesterfield—recommended, I might add, by no less an authority than R. W. Chapman as a model of accuracy [10]—an examination of seven works discloses all manner of confusion: two impressions erroneously classified as states, two sets of two editions, each described as one, and a third set entirely overlooked. Similarly, in R. H. Griffith's bibliography of Pope, a survey, again confined to seven works, reveals two states classified as impressions, three sets of two editions, each described as one, and one set of three, likewise represented as a single entity. This production has also been approved in the reviews, including, for instance, those by George Sherburn and A. E. Case,[11] where it is reported in one (and implied in the other) that the copies at Chicago and Yale are identical with the ones described, when, as a matter of fact, they differ from these, from others accessible to the reviewers, and indeed from others in the same collection. If this is the situation in the bibliographies of Johnson, Chesterfield, and Pope, all of which are considered as definitive, what may we expect to find in those not so recognized? And what, we may inquire, is the status of scholarly commentary based on the premises of these bibliographies? To these questions we may individually provide the answers which seem appropriate.

Having the temerity to proceed thus far in what may be taken

as a general indictment, I must go a step further. No one, least of all the present writer, should presume to say what has happened without suggesting the cause and the cure. As any insinuation of incompetence would, of course, be irrelevant for bibliographies regarded as authoritative, this may be discounted and two other reasons advanced. For one, the standards of examination so readily accepted ten or twenty years ago, and so often in evidence today, have been superseded by those which require, among other things, some facility in the interpretation of various technicalities, the comparison of such differentiæ as headlines, the use of control copies or something that will serve as a control, and—whatever the inconvenience to the person concerned—the inspection of numerous copies in addition to the several represented in special collections. For the other, the books to be examined, in the eighteenth century, are, as I have intimated, the products of conditions of greater complexity than those which apply in the earlier periods, and therefore occasionally require supplemental techniques for their analysis. It has not been sufficiently realized that printing, in this century, has progressed beyond the era of the simple handicraft and now represents one of mass production, where not a few but hundreds of pages of type may be retained and repeatedly returned to press, where not one or two individuals but batteries of pressmen and compositors may produce, in a matter of hours, editions running into thousands of copies, where not one but several books may be put to press and worked concurrently by the same personnel. These practices, though extraordinary in the seventeenth century, have become commonplace in the eighteenth, and now demand a consideration frequently evaded or ignored in our present studies.

The only procedure that fully accounts for the altered circumstances of book-production in this period is one that utilizes the evidence peculiar to these circumstances. Evidence of this sort, we may be assured, is usually not to be found in the materials of production—in obvious irregularities in type, paper, and ink—all of which tend to be standardized and uniform, but in the actual process of production, as this is revealed by the press figures. These little symbols, appearing now and then in the lower margin of the page in many eighteenth-century books, were first reported, I

believe, by Griffith in his Pope bibliography, and have since become involved in several suppositions as to their original use to the printer and their present use to the bibliographer. Contrary to McKerrow's prediction that they would prove to be of little importance, recent investigation has shown that they may be interpreted as signs of cancellation, variant states, half- or fullpress operation (indicating the employment of one or two men at the machine), type pages arranged within the forme in some irregular pattern, sheets impressed in some abnormal order, an impression of the formes for each sheet by one man working both formes in succession, or two men working both simultaneously, impressions interrupted for one reason or another, reimpressions or resettings of the book, in whole or in part, copy distributed among several shops, overprints involving an increase in the number of sheets machined for certain gatherings in order to meet an unanticipated demand for copies, and underprints consisting of a decrease in the number of sheets in order to reduce the issue and speed its publication.[12] Now that they are displayed in the figures, these various ramifications in composition and presswork may not be overlooked by any editor cognizant of his responsibilities, for here and there in a process so involved he should expect to find an unnoticed reading, an altered passage, or an entirely different edition from that on record.

Since the figures disclose so much of possible textual significance which might otherwise remain unrevealed, they are, notwithstanding McKerrow, of some importance, not only in the discrimination and analysis of concealed or known editions, but in the bibliographical description of the books examined. If the figures are included in the record, any variant unknown to the compiler can be immediately detected and appropriate measures taken to establish its relationship to others in the series. If they are not included, then, of all the variants already mentioned, the most elaborate description will permit the identification of only a few of the editions and none of the impressions.

An interesting example of the relative value of the figures and of the more conventional means of discrimination is afforded by

Burke's Reflections on the Revolution in France (1790). According to Gathorne-Hardy's account, which is duly offered on the basis of differentiae in catchwords, ornaments, the spacing of type, and the length of a rule preceding the text, the first two editions of this book consist of sheets in two settings of type separate or mixed in several combinations.[13] As the figures intimate, however, and as an inspection of the discrepant copies will disclose, the type for this book exists in one to four settings, each occurring in one to four impressions, and all combined in a strange and wonderful manner to produce five distinct variants. Among these we may identify two editions—A, comprising a single impression throughout, and B, comprising four impressions, of which two (B1-B2) are undesignated and two (B3-B4) called a "Second Edition." Now, with all of this decided, unequivocally, the editor can proceed, in the manner that Sir Walter Greg suggests, to select as his copytext what is identified as the first edition, and then to add to this the revisions which Burke introduced in the edition called the "Third," an edition disregarded in the previous commentary, and also, in part, a reimpression. Though this is the required procedure, it has not been followed for the Reflections or for numerous other works of a similar nature simply because the problems incident to their manufacture have been ignored or misconstrued.

Even when the variants are known, however, their relationship may be undetermined, again for want of appropriate bibliographical techniques. In some cases, as with the Reflections, the several editions and impressions can be ordered on the evidence of the figures alone, without reference to copy. In others, the later edition may be identified by the usual criteria, such as the inclusion of textual matter in the preliminary gathering of the Deserted Village piracies—those which are so highly esteemed as "private issues," or the gradation in the size and format of the Bramston, Chesterfield, and Lyttelton pamphlets to which I have referred, or the correction of errata in Tom Jones and in Gibbon's Decline and Fall, or the absence of cancels in Johnson's edition of Shakespeare, or finally, in watermarks, imprints, and advertisements, all of which materially assist in deciding the priority of the Monk editions.

Other works not amenable to these several approaches can be subjected to a mechanical demonstration, such as the one Fredson Bowers exemplified for Dryden's Wild Gallant in his address before this Institute last year.[14]

Nevertheless, with all of these techniques at our command, we still must contend with many eighteenth-century editions which continue to be bibliographically intransigent. These bugaboos infest what one writer has called "The Bibliographical Jungle"[15]— a wild locale visited by a goodly number of adventurers, all of whom have hacked about in the bushes, but produced nothing to show for their labors except a covey of unappealing conjectures and misconceptions. To produce the results desired, we should, once again, abandon the conventional methods of attack, all of which would seem to be of little avail, and devise for our purpose some unconventional tool.

The one I would propose concerns the use of a third text of known date and provenance, deriving independently from and therefore identifying the first of the several editions under consideration, yet, paradoxically, extraneous to these editions, to their printer, and to the author, and thus, in a strict sense, of no editorial or bibliographical significance whatever. This anomalous text is that provided by the literary reviews which, beginning with those to be found in the early numbers of the Gentleman's Magazine, [15a] usually appear shortly after the publication of the first edition and occasionally cite passages of considerable length "for the interest of their readers," and now, we may say, for our interest too. When the readings in the abstracts are collated with those in the indeterminate editions they will, of course, collate only with those in the text from which they were copied, and thus establish the

priority of that edition in the sequence. It is rather surprising that evidence so abundantly available in the periodicals should be so rarely employed. In fact, so far as I can recall, only one scholar, Austin Dobson, has ever resorted to the reviews for the purpose I recommend, and on that occasion—some thirty years before the discovery of the printer's ledger confirmed his decision—he was able to demonstrate by a simple collation that, contrary to normal expectation, the expurgated version of Fielding's Voyage to Lisbon (1755) was issued prior to the one with the original text.[16]

Of the many books which can be ordered through reference to the journal readings, several may be mentioned here. The first, Smollett's Humphry Clinker (1771), has long confounded every attempt at analysis. For the last twenty-five years a succession of bibliographers have undertaken the disheartening task of collating the 800 pages in this novel, amassed reams of worthless notes, and eventually confessed their failure to establish the precedence of the editions.[17] For this, as for numerous other works, a mixture of corrupted and corrected readings precludes a demonstration of priority.

Undoubtedly, any approach to a solution in the case of Humphry Clinker has been somewhat impeded by an uncertainty as to the number of existing variants. Once the true editions are disengaged from sophisticated copies, variant states, and "ghosts," all of which have crept into the discussion of this riddle, it will be noted that there are, essentially, four distinct settings:

Volume I

         

Edition	Title-page		Press figures [18]
A	1671		Throughout
B	1671 or 1771		Only in gathering M
C	1772	2d ed	None
D	1771	2d ed	None

Now between June 18, 1771, the date of original publication, either for A or for B, and September 1, the approximate date that work was completed on the alternate "first edition," a total of seventy-four pages from the novel were reprinted in four journals, The London Magazine, The Gentleman's Magazine, The Critical Review, and The Weekly Magazine, or Edinburgh Amusement.[19] As these reprinted pages represent 138 readings, all differing from those in B, and all agreeing with those in A, the conclusion follows that A is unmistakably the first of the two editions. With the order of these determined, the relationship of all texts, complete and abstracted, may be represented in the stemma:

Thus after years of misdirected activity, a little effort, very little, has here provided a certain decision.

No less certain are the results to be obtained by the use of figures and reviews, independently or together, in the reconsideration of problems already known and decided. Two instances may be cited, one reported within the past several months, one on record for the last twenty years, both entertaining versions of the kind, number, and order of variants, and both entirely contrary to fact. The first, appearing in the latest catalogue issued by Peter Murray Hill, is a brief notice of The Life of David Hume, Esq. (1777) — a book about which Millar has given us ample warning. Concerning

this, Hill reports two "issues," with the first containing a reading "himself" and the second containing the reading supposedly corrected to "myself." Actually these are not "issues," as they are called, or states, as we might infer from the description, but separate editions, of which the figures indicate different settings of type and the reviews a different order of publication.[20]

For the second instance, Edward Moore's periodical, The World (1753-56), the current explanation is as simple as the one I have just refuted, but the circumstances of printing somewhat more complex. In J. H. Caskey's account we are informed that there were, apparently, two simultaneous "issues" of every one of the 209 numbers.[21] First, it would seem, came the issue bearing a vignette, and then, set from proof-sheets of this, the one containing a headpiece of printers' ornaments. Though the reasons for this facile hypothesis are not made explicit, they seem to rest on a reference to the fact that 2500 copies were required of some numbers. Perhaps it was thought that Robert Dodsley, the publisher, needed to have two settings for an impression running over the limit established by regulation, or that two were necessary in order to shorten the time at press. From the point of view of those dealing with the literature of the sixteenth or seventeenth centuries, either explanation would appear to be plausible. From our vantage ground in the eighteenth, both must be considered irrelevant.

The size of an impression was certainly not determined, at this time, by directives of the Stationers' Company. That these had long since become inoperative is attested by the printers' own accounts: William Woodfall cast off 4500 copies of Edward and Eleanora (1739); Samuel Richardson, 2500 of The Centaur Not Fabulous, third edition (1755); William Strahan, 3000 of Tom Jones, third edition (1749), 3500 of the fourth edition (1750), 5000 of Amelia (1752);[22] and each of these editions, I am satisfied, is of a single impression. Nor can it be argued that there was insufficient time to

produce an issue of The World from one setting of type. If two presses were used for the full sheet, one for printing, the other for perfecting, and a third press for the half-sheet, even as many as 2500 copies could be easily completed within three days, well within the schedule for a weekly paper. With the liberty and the time to do what he pleased, the publisher would hardly put himself to the expense of a second, unnecessary setting at the time of publication. The only reasonable explanation therefore seems to be that the variant "issue" was set at some time after publication to meet an unexpected demand for certain numbers.

If the variants are, thus, nothing more than occasional reprints, they should occur irregularly in the sequence of numbers and not, as the previous hypothesis would require, in every number. What, then, is the situation, and to what extent may it be explained by the usual method of analysis? Examination discloses that, among the copies I have examined, the variants are not precisely two to a number, nor can they be considered as "issues," nor were they printed in the sequence described. From the 19th to the 209th and last, all numbers bear the vignette. Among the first eighteen numbers one encounters the usual disorder in what is assumed to be entirely regular: three editions of the first and second numbers, two from the fourth through the eleventh, one from the twelfth to the seventeenth, and two of the eighteenth. A distinguishing characteristic of some of these editions is the headpiece, of which there are, not two, but three variants:

Whenever one copy differs from another in any of these respects it also differs in setting. And whenever the settings B and C are found for the same number of the periodical, we may assume from the evidence of the fractured block that C was printed at a later date than B. This date is fixed by the progressive deterioration of the vignette in the numbers which exist as a single edition. From the 19th to the 23rd number the block is sound; from the 24th to the 87th it is broken in one place, as in variant C; and beginning with some copies of the 88th number a second break appears. Thus the C edition of the earlier numbers, though dated January, 1753, must have been printed after June 7 of that year, the date of the 23rd number, and before September 5, 1754, the date of the 88th.

Whereas B and C have the vignette in common, a collation of the readings indicates that they are separated by an intermediate text A without vignette. The precise relationship of the three editions, wherever they occur in the periodical, therefore corresponds to one of these three stemma:

Further than this we cannot proceed by the critical method, for the potentialities of that method have now been exhausted. But if we choose to apply collateral evidence, a second series of texts is made available, and the analysis can then continue to a conclusion.

These texts, present in The London Magazine, The Monthly Review, The Scots Magazine, and The Universal Magazine,[23] abstract readings from the first two numbers of the periodical. Among the readings in the several editions of the first number the variants are accidental, amounting to a few changes in capitalization. Among those of the second, however, the variation is substantial:

             

	Reading	Editions		Reviews
7,8	shine, but	A	C	MR	SM
	shine but	B
8,29	nothing to say	A	C	MR	SM
	not a word to say	B
10,32	absence;	A	B	MR	SM	LM	UM
	absence:	C

A collation of the seven texts now at our disposal demonstrates the literal sequence of the three editions to be of the order A-B for the one, A-C for the other. The chronological sequence, where C with the broken vignette follows the unbroken B, is therefore:

This determination of the priority and relationship among the three editions of one number of The World establishes a reasonable presumption for a similar arrangement of the editions in others. Without reference to the reviews any determination, even for one number, is impossible.

From the evidence adduced in this and the preceding examples I would suggest that by one method or another, and sometimes by these methods alone, we are now in a position to discover and decide problems presently unknown or unresolved. To such a declaration it is hardly necessary to add that these several techniques do not constitute a panacea for all of the duplications and disorders to be encountered in eighteenth-century literature. Some books requiring differentiation are unfigured, figured in uninformative patterns, or, in rare instances, exactly reiterated in their figures. Some requiring arrangement are unreviewed, reviewed by a synopsis not subject to collation, or cited in portions where there are only a few or no discrepancies. For these exceptional cases other procedures must be found, or the project abandoned as insolvable. For all except these, however, the methodology now available will facilitate a solution to any problem concerning the discrimination, description, and classification of variants. Even in these elementary, though fundamental preliminaries to editorial endeavor there is much to do in the eighteenth century, and much of what has been attempted must be done again.

Notes