University of Virginia Library

The Bibliographical Significance of the Turned Letter
by
Robin Dix and Trudi Laura Darby [*]

Carelessness at one or more of the various stages of book production has caused bibliographers and editors many headaches over the years. And yet it is generally accepted that certain forms of carelessness can be very revealing—a leaf slit for cancellation but not actually removed, for example, is a valuable find. The humble turned letter can also be used for a wide variety of bibliographical purposes by those who are alive to its potential.

In normal circumstances, the production of an accurate text in the handpress period was the result of a team effort by author, scribe if there was one, compositor, in-house proof-corrector and, possibly, author as proofreader; and unless manuscripts and proofs have survived, it is not easy to distinguish their individual contributions to the accuracy of the finished product. In the absence of such documents, a turned letter can provide useful evidence about the compositor's working habits and the degree of care he used. It should after all have been more or less impossible to allow an inverted letter to stand: the nick in the front of each sort meant that by running his thumb along a line of type after he had set it, a compositor could check that each letter was the right way up. The existence of a turned letter in print thus implies that the compositor either did not check the nicks, or, having checked them and found an error, did not rectify it.

In fact, inversions of the grossest kind—that is, where the resulting symbol does not form a letter at all—are comparatively rare. Doubtless the nicks enabled compositors to find the vast majority of such errors themselves, and to correct them while the type was still in the stick; those that were missed at this stage would have been picked up easily enough in proof (assuming that the page in question was proofread). But there was one letter where the attitude of all concerned was likely to be very different: rounded 's'. Here, inversion did not result either in nonsense or in a different letter: the mark would be read correctly, even when inverted, by all who saw it; and while the upper and lower loops of the letter differed in virtually all old typefaces,[1]

the number of readers who noticed such minutiae would have been tiny. For these reasons, those in the printing-house who found an inverted 's' were unlikely to devote time and resources to correcting it.

Perhaps the most obvious use of the turned letter is in the identification of different states within a book. A glance at the bibliography of Locke's An Essay Concerning Humane Understanding (1690) provides a good example. The first edition of this seminal philosophical text exists in two issues. In what is now generally accepted as the later of the two, the title-page is a cancel, presumably introduced because considerable changes had to be made to the imprint; but the most immediately noticeable difference is that both 'S's of 'ESSAY' are inverted.[2]

The case is unusual only because of the prominence of the error: most editors of works from the hand-press period will find instances of turned 's' if the texts they are dealing with are substantial. In the works of Mark Akenside (1721-1770), for example, there are some twenty cases of inverted 's'[3]—and Akenside's printers were conscientious professionals such as Hughs and Bowyer. It has to be recognized that uneven inking or damaged type can mask the serifs and distort the proportions of the letter, and that in a small number of cases the inversion may thus be more apparent than real. Nevertheless, the evidence is normally unequivocal, and there are occasions when it can provide bibliographical information of significance beyond the existence of variant states. Thus, in 'Hymn to the Naiads', one of Akenside's contributions to Dodsley's A Collection of Poems by Several Hands (vol. vi, 1st edn., 1758), an inverted 's' not only led to the discovery of a previously unidentified variant state within the first gathering but also strongly suggested that a very late, major authorial change, hitherto unsuspected, had been made to the text. There are two typographical differences between the variant states: the final 's' of 'steams' (A3r, 15th line of text) is the right way up in the newly discovered state, and a hyphen which usually appears between 'ill' and 'repaid' (A3v, 23rd line) is omitted. Tests using a straight edge show that the type on both pages was largely reset between printings. This fact was in turn explained by Michael Suarez, S.J., of Campion Hall Oxford: in the newly discovered state without the hyphen or the turned letter, A3, which is usually a cancel, forms an integral part of gathering A.[4] The reason

for cancelling the original leaf remains obscure: these two tiny adjustments comprise the only textual differences. As such changes are clearly insufficient to warrant cancellation, the most likely explanation is that a late authorial change necessitated some resetting and the insertion of a cancel, and that at a suitable opportunity A3 was reset to incorporate the revision into the formes for A.[5] No manuscripts of the poem survive, and no copies of gathering A in any other state are known, however, so the theory cannot be proved.

These examples from Locke and Akenside show how the introduction or correction of a turned letter such as rounded 's' suggests resetting. Similarly, the recurrence of an inverted sort in supposedly different editions implies that not all the type was distributed. Take for instance The World Tost at Tennis, a masque written by Middleton and Rowley, and printed by George Purslowe. The British Library catalogue states that the book exists in two editions; but both contain two examples of turned 's': in 'names' on B3v and 'wrongs' on E4v. This strongly suggests that the type, in these areas at least, was not distributed between the printings. Further analysis shows that the different 'editions' are in fact merely different states: the title-page is a variant leaf, and the type is otherwise undisturbed.

A second example of recurring inversion, from Akenside's Odes on Several Subjects (1745; quarto edition), shows how a turned letter can provide important clues in bibliographical puzzles. Gathering B exists in three states: in one, it is printed normally; in another, B1v contains the text of p. 52 rather than p. 10; while in the third, B1 is a cancel, and the pagination runs correctly. It is possible to demonstrate that in the two variant states containing p. 10, the type is of the same setting; and a very useful piece of evidence contributing to that demonstration is the fact that in the penultimate line the 's' of 'is' is inverted. This in turn permits a credible explanation of the error. The book was printed normally, but it became necessary to print extra copies of B; the formes for B were reassembled while G was also being assembled at the imposing-stone; G2v was accidentally placed in the wrong chase; and when the error was discovered, it was rectified by cancellation.[6]

A third example of recurring inversion gives an interesting insight into

compositors' sharp practice in eighteenth-century print shops. Akenside's best-known poem, The Pleasures of Imagination, was brought out by Robert Dodsley in January 1744, and went through three further editions that year. The first two editions were quartos, the third and fourth octavos. A few of the later formes in the quartos are of the same setting of type; but the octavos have always been considered as quite distinct from each other. The third edition appeared in May, the fourth in November 1744.

In fact the octavos are not entirely distinct. The poem contains a number of long footnotes in the first half of Book III, and parts of these, along with some of the prelims, are of the same setting of type in both. The clearest signal to the editor is that at several points an inverted 's' appears in the same place in the same word in the two editions. A comparison of damaged sorts in the two editions, and tests using a straight edge, confirm that a considerable part of the long notes in gatherings G and H are indeed of the same setting of type.[7] The same tests, however, show that the poetry on these pages was reset. The implication seems to be that the compositors did not relish the prospect of resetting the long notes, which were in small type and which required justification, so they transferred type from the previous edition wherever possible for these; but they reset the verse, thus making the page look different, and so ensuring that they were paid in full. W. B. Todd identified and described a similar practice (which he called 'featherbedding') in his bibliography of Burke, and we venture to suggest that it is more frequent in the hand-press period than has been recognized.[8] It is obviously a habit that editors need to be aware of, and the recurrence of an inverted 's' at the same point in supposedly different editions is precisely the kind of feature which can serve to alert them to it.

It is possible that with rounded 's' a compositor might have made a conscious

decision not to emend: he may have identified the turned letter, seen that it was an 's', and ignored it on the basis that the inversion was unimportant. But even if some of the instances can be explained in this way, the implication is still that the compositor was prepared to make judgments about the weight to be attached to a mistake. From an editor's point of view, such an attitude must affect any assessment of the reliability of the text, and may justify a less conservative approach to troublesome readings than would normally be permissible.

The Akenside texts examined so far represent an extreme case, in that they were carefully set, and received attention from an author who was more than usually interested in typographical minutiae.[9] But the significance of the turned letter remains—and indeed in some respects increases—when texts at the other end of the spectrum are involved. Seventeenth-century drama provides good material to illustrate the point. Received wisdom has it that each newly-imposed forme was proofed;[10] but it is very common to find that, in quartos at least, printers' errors occur in groups, interspersed with relatively 'clean' formes. The obvious inference is that often only sample formes of the less prestigious books received proof correction. If the formes which were accorded this attention can be identified, the modern editor is able to assess the accuracy of each compositor involved in the production of a given text—and indeed the accuracy of the proof corrector. And a further conclusion is possible: when formes which did not receive the corrector's attention contain a significant number of turned letters, it is necessary to bear in mind the possibility of emendation wherever symmetrical letters can make a viable alternative reading by being rotated. Every 'jape' may in fact be a 'jade'.

The importance of turned letters—both those which were spotted by the corrector, and those which were not—is well illustrated by the surviving copies of A New Wonder, A Woman Never Vext, a play by William Rowley which Purslowe printed for Francis Constable in 1632. Eleven copies survive in the United Kingdom, one in Stockholm, and another twelve in the United States.[11] The press variants in sheets A-D are due mainly to loose or pulled type, or poor inking; but in the outer forme of E the nature of the changes

alters, and there are more corrections of obvious compositorial errors. For instance, 'solyd' is corrected to 'soyld' (E1r), 'handfell' to 'handsell' and a turned 'a' put right (E2v), 'some' altered to 'sum' (E3r), and an erroneous signature ('E5') corrected to 'E3'. The inner forme of F shows some thirteen press variants of a similar kind. Those on F4r are particularly interesting, as they imply a preferred spelling for certain words: these were amended, apparently to conform to house style.[12] The outer forme of G shows an added 'Sir' on G1r, and speech prefixes which were incorrectly assigned are changed on G2v.[13]

Equally important are the corrections which should have been made but were not—most obviously, in the inner forme of E, the signature 'E2' on E3 was not put right. The easiest inference is that E outer, F inner and possibly G inner were proof-read, but that the other formes were not. The alternative explanation—that a second, more conscientious compositor was involved with these formes—is unlikely: an analysis of running titles, italicised type, punctuation and spacing suggests that A-H were set by a single compositor. It was not until sheet I that an apparently more experienced workman was brought in, whose italic usage differed from house style.

The text of A New Wonder, A Woman Never Vext shows a relatively high incidence of turned letters. Moxon asserts in Mechanick Exercises that if a letter is turned, the corrector marks it and the compositor picks it out and puts it back correctly (p. 249). This seems to have happened on the corrected pages of E and F that we have noted; and on K2r an inverted 'm' in the running title was emended. On K4r, 'H' was set sideways below the line; this again was corrected in press. But other examples of inversion escaped correction: on F2v 'shade' is printed instead of 'shape'; on H2v we have 'barkiug' for 'barking'; on K3v 'disiuheritance' for 'disinheritance', and on K4v 'onght' for 'ought'. Of these, the n/u confusion may well be attributable to foul case, and is unlikely to cause serious problems for either editor or reader, however important it may be to a bibliographer concerned with the conscientiousness of the compositor. The story is rather different with the inverted 'p' on F2v, which produced 'shade': in context, the reading makes sense of a sort, and only the vigilant reader or editor will realize that the wrong word has been created. Close examination of the serifs shows that the 'd' is in fact a turned 'p'. But that in itself is insufficient to prove that 'shade' is wrong: foul case could explain the presence of a 'p' sort among the 'd's, and the compositor may have inverted the letter deliberately in order to obtain a 'correct' reading more quickly than he could by replacing the letter in its proper box and selecting another. Ultimately, the question of which reading to accept has to be decided by the editor, using the full range of his or her critical judgment.

Purslowe's shop printed another play in 1632: Changes by James Shirley. This text shares certain distinctive characteristics with A New Wonder, A Woman Never Vext, notably the setting of prose as verse. Purslowe had printed dramatic texts before—he printed ten in his nineteen years in business—but the last had been an edition of Mucedorus in 1626, and both Changes and A New Wonder, A Woman Never Vext seem to have been set mainly by a compositor unfamiliar with dramatic works. In fact Purslowe employed two apprentices at this time, both of whom joined him in 1629.[14] Changes shows turned letters on B2r (terminal 's' of 'spurnes'), D1v ('s' of 'appeares'), E3r ('Anr.' for 'Aur.' in the uncorrected state), and F4r ('s' of 'vertues'). An especially noteworthy turned 's' is found in the running title which appears on B3v, B4v, and then on 1v and 4v of C-F and G1v. The letter was corrected on G4v and recurs, the right way up, on H1v, H4v, I1v, I4v and K1v (K being a half-sheet). Sheets A-F are generally poorly set, with a number of obvious errors such as 'as' for 'at' (A3v), 'servantss' (C1v), 'athough' (C4r), 'Be' for 'By', 'ad' for 'and' (D1v), 'yon' for 'you' (D3r: probably foul case). Although G-L are not flawless, there is only one obvious error of this kind ('chsue' for 'chuse', K2r). It should also be noted that character names are set in roman in some stage directions, instead of in italic as in the remainder of the text, on F2v, F3r, F4v, G1v, G2r, G3r and G3v (there are no entrances on F3v, F4r or G1r; on G2v a marginal stage direction is entirely in italic). No collation of this text has been attempted;[15] but the alteration to the running title could suggest that at F outer a second, more accurate, workman was assigned to the project, and that it was he who corrected the turned letter, once he noticed it while composing G outer. This would accord with the practice in setting A New Wonder, A Woman Never Vext, for which a more experienced workman was apparently brought in at I.

An examination of turned letters in sample texts from the seventeenth and eighteenth centuries, then, suggests a number of points which are of interest to editors and bibliographers. First, inversions generally can help to prove that two supposedly different editions, or passages, are of the same setting of type; and where either the text was not proofread, or where the proofreader was unlikely to require a correction to be made (as with rounded 's'), inversions can also be useful in proving that the setting has been changed. Secondly, the frequency with which turned letters appear in uncorrected passages or texts gives an editor a quick and valuable indication of the compositor's reliability, and may also be of use in compositorial identification. Thirdly, and perhaps most radically, where inversion is frequent, editors need to beware of any characters which could form a different letter when rotated about a horizontal, vertical, or oblique axis. This remains true even

where the page was proofread and rounded 's' is the only unequivocal evidence of inversion to remain: since proofing was normally done orally, it would be easy enough for sounds such as 'shape' and 'shade' to be confused if the wrong word had been created. Nasal congestion, dialectal differences, and other factors apparently extraneous to a bibliographer's concerns, may play their part in determining the accuracy of the printed text.

Notes