Book Excerpt



New Castle, Delaware & Winchester: Oak Knoll Press & St. Paul's Bibliographies, 2009. 6 x 9 inches. hardcover. 462 pages. Reprint of the 1995 Oak Knoll edition. Ronald B. McKerrow's An Introduction to Bibliography for Literary Students has been the classic manual on bibliography, showing how the transmission of texts might be affected by the processes of printing, but he concentrated almost exclusively on "Elizabethan" printing - the period from 1560 to 1660. However, in... READ MORE

Price: $65.00  other currencies  Order nr. 60423



Before discussing the techniques of bibliographical identification we must deal with the technical terms which are used to define the relationship between different copies of the same work.1 An edition, first of all, is all the copies of a book printed at any time (or times) from substantially the same setting of type, and includes all the various impressions, issues, and states which may have derived from that setting. As to the meaning of ‘substantially the same setting of type’, there are bound to be ambiguous cases, but we may take it as a simple rule of thumb that there is a new edition when more than half the type has been reset, but that if less than half the type has been reset we are probably dealing with another impression, issue, or state.

Editions of the hand-press period are usually easy to identify. Resetting by hand, even when the compositor follows the spellings and abbreviations of printed copy word for word and line for line, always results in identifiable differences of spacing between the words, and the random pattern of damaged types is likewise different. With practice, two very similar settings can be told apart at a glance when copies are laid side by side, something that is usually easy to arrange with the aid of photocopies. If it is not practicable to compare copies or photocopies directly, resetting will normally be revealed by comparing notes of the precise positions of signature letters relative to words in the lines immediately above them.

Editions of the machine-press period, however, are not so readily defined. Resetting by hand is still easily identified, although resetting of Linotype by Linotype, where the spacing is mechanical, may be very difficult to detect. But the problem of defining editions of the machine-press period is not so much that of identifying new settings of type as of classifying reprints made indirectly from one setting, one act of composition.

Let us start with plates. Stereos and electros are three-dimensional models of the original setting, which may not itself have been used for printing at all. Copies printed from plates must therefore be regarded as part of the original edition. This is true, moreover, even if there are long pauses in a book’s printing history. If a book is reprinted from standing type, the reprint is part of the original edition, even if it is not part of the original series of publishing units; and if a book is reprinted from an old set of plates, the result is again part of the original edition.

Which leads us on to the much more difficult question of photographic reprinting. It is evident that the photolithographic version of a current book—made for the American issue, say, of a work printed from type or plates in England—has to be considered part of the original edition. But once this is granted, and bearing in mind that a long pause in a book’s printing history does not in itself make a new edition, it is difficult to avoid the conclusion that any photographic reprint is part of the edition from which it derives, and that this includes modern photographic facsimiles of early books and even individual photocopies. While this conclusion makes for some absurdity (for one cannot seriously consider a xerox of Q1 Hamlet to be part of the first edition of Hamlet), it is necessary for dealing with commercial photolithographic reprints in multiple copies (which will normally bear an additional imprint, and where parts of the original may have been suppressed or added to), and with such modern possibilities as the transmission of settings by line for photolitho printing at a distance.

We see, in fact, that in the machine-press period the edition may be defined by the image of the setting as well as by the setting itself. Similarly with Monotype, the edition is defined not by any particular setting of type but by the programme for a setting punched on a paper spool. The Monotype spool can be run through the caster several times to produce several duplicate settings of type which, although not physically identical, are in practice indistinguishable from each other; and copies from any of these duplicate settings must be thought of as part of the one edition defined by the spool since they derive from a single act of composition. What is more, if the Monotype caster is fited with a different matrix case for one of these settings, so producing it in type of a different design from the rest, copies printed from this different type are still part of the same edition, deriving from one spool, one act of composition.1a Similar considerations will apply to programmes of composition recorded in other ways, for instance on magnetic tape.

For books of the machine-press period, therefore, an edition may include not only all the copies of a book printed at any time (or times) from substantially the same setting of type, but also copies printed from relief plates made from that setting; copies reproduced photographically; and all the copies deriving from a particular Monotype spool, computer tape, or other programme of composition, however reproduced.

Impression, which means all the copies of an edition printed at any one time, is as a concept less ambiguous than edition, but impressions can be very difficult to identify. In the hand-press period it was normal to distribute and re-use the type from each sheet as it was printed off, so that at that time the edition and the impression were generally the same thing; even then there were exceptions, as when the type for a small or successful book was kept standing for reprinting later. One of the chief advantages of stereo and electro plates, however, was that they could be kept and used to print further impressions after the first, and it was not unusual in the nineteenth century for stereos to be used for ten successive impressions, and for electros to be used for as many as thirty; while, if a set of plates was kept as a ‘mother’ from which further sets could be made, the number of successive impressions of an edition that could be printed from plates was virtually unlimited.

New impressions were sometimes marked by special symbols, if not by new signatures; and in the eighteenth century press figures were normally altered. Less obviously the relative position of the pages or plates in the forme may have been altered, or the lines of a typeset page may have shifted slightly if it was re-imposed with new furniture. In the absence of any typographical differentiation, the best clue is likely to be found in the paper used.

Before considering issue and state, the bibliographical concept of ‘ideal copy’ must be introduced. Since different copies of an edition may vary from each other in a number of ways, the bibliographer examines as many copies as possible in order to construct a notional ideal copy of the edition he is studying. A description of this ideal copy would note all the blank leaves intended to be part of its gatherings, and all excisions, insertions, and cancellantia which belonged to the most perfect copy of the work as originally completed by its printer and first put on sale by its publisher. This is the basic ideal form; and the description of ideal copy is completed by the addition of notes of any subsequent changes made by the printer or publisher to improve the book or to modify the conditions of its sale, and of any unintentional alterations to its form. Roughly speaking, different publishing units within the one edition are called issues, while parts of an edition embodying changes to the text, whether intentional or not, are known as variant states. Sometimes these differences were introduced during the course of a single printing, sometimes they were connected with the production of a new impression.

An issue is all the copies of that part of an edition which is identifiable as a consciously planned printed unit distinct from the basic form of the ideal copy. The criteria are that the book must differ in some typographical way from copies of the edition first put on the market, yet be composed largely of sheets deriving from the original setting; and that the copies forming another issue must be a purposeful publishing unit removed from the original issue either in form (separate issue) or in time (reissue).

Cases of separate issue would be: the alteration of title-pages to suit the issue of a book simultaneously in two or more different forms; the reimposition of the type pages to produce copies in different formats; impressions on special paper distinguished from ordinary copies by added, deleted, or substituted material. (Special-paper copies not distinguished typographically from those on ordinary paper—or distinguished only by altered margins—are said to be of a different state, not of a different issue, even though they represent a purposeful publishing unit.) Reissue normally involves a new or altered title-page, and includes cases such as: the cancellation of the title-page to bring old sheets up to date; a new impression with a new title-page; and collections of separate pieces with a new general title.

The term state is used to cover all other variants from the basic form of the ideal copy. There are five major classes of variant state. (1) Alterations not affecting the make-up of the pages, made intentionally or unintentionally during printing, such as: stop-press corrections; resetting as the result of accidental damage to the type; resetting of distributed matter following a decision during printing to enlarge the edition quantity. (2) The addition, deletion, or substitution of matter, affecting the make-up of the pages, but carried out during printing. (3) Alterations made after some copies have been sold (not involving a new title-page) such as the insertion or cancellation of preliminaries or text pages, or the addition of errata leaves, advertisements, etc. (4) Errors of imposition, or of machining (e.g. sheets perfected the wrong way round; but not errors of folding). (5) Special-paper copies not distinguished typographically from those on ordinary paper. Apart from class (5), differences of state are generally the attributes of individual formes, or sometimes of individual sheets.

These terms are used with reference only to the printed sheets of a book, and are not affected by binding variants. An exception might be made of printed wrappers which are peculiar to an edition, for they sometimes did duty as supplementary title-pages and were printed along with the sheets. Variant binding cases of the machine-press period may suggest re-issue, especially when they are related to particular inserted advertisements, but they cannot normally prove it.


Turning, then, to the techniques of bibliographical identification, when an editor or librarian is confronted with a more or less mysterious volume he will want to know the name of the author and of the work, the names of printer and publisher, the places and dates of printing and publication, and the place of the volume in the printing history both of the work as a whole and of the particular edition to which it belongs. Usually he will already know, or can easily look up, most of these facts, but occasionally one or more of them may prove difficult to establish, and the following notes suggest ways in which such questions may be answered.

Most books have title-pages, and sometimes imprints and colophons as well, which supply the author’s name, the title of the book, the name of the printer or publisher, and the place and date of publication. Sometimes this information is partly or wholly lacking, however, and sometimes it is given wrongly, when it is the bibliographer’s task to supply it as best he can. He does so by weighing the information that is given in the book against its physical make-up, referring at the same time to any external evidence that is available. As far as author and title are concerned the search must normally be made externally, both in contemporary sources such as the Stationers’ Register, the Term Catalogues, and the trade-sale and publishers’ catalogues of the eighteenth and nineteenth centuries, and in modern bibliographies and reference books.2 It is also worth remembering that the author’s name or initials may be found in the body of the book (commonly at the end of a foreword or preface) and that the title may be used for running headlines.

Edition statements should be received with caution. The term ‘edition’ has always been used in the trade for ‘impression’ or ‘issue’ as well as for edition in the bibliographical sense; a book that is advertised as a ‘new edition’ may indeed represent a new setting of type, but it may be a reimpression from standing type with or without correction, an impression from plates, or simply a reissue of the original sheets with a new title-page. To take a modern example, Methuen ordered two sets of electrotype plates of A. A. Milne’s Winnie-the-Pooh, and had twenty-seven impressions printed from them in the period 1926–41. Although all twenty-seven impressions, deriving from a single setting of type, were part of a single edition, the publishers advertised each one as another edition, so that, when a new set of electros was made in 1942 from a new setting of type, what was then issued as the ‘twenty-eighth edition’ of Winnie-the-Pooh was in fact the first impression of the second edition.3

The position of an edition or issue in a series can often be established by comparison with other editions or issues of the same book of which the dates or positions are known. This generally involves detailed textual bibliography (see pp. 336–60), but there may also be typographical clues in altered signature series and press figures, and in the deterioration or alteration of blocks and plates.

Printers or publishers were sometimes shy of giving their real names—usually because a book was treasonable, or libellous, or a piracy—and for similar reasons they might give a false place of publication and a false date. Books might also be wrongly dated by accident—roman numerals were often mis-handled—or because of the trade custom of pre-dating: ‘The Rule in general observed among Printers,’ wrote Nichols concerning eighteenth-century practice, ‘is, that when a Book happens not to be ready for publication before November, the date of the ensuing year is used.’4 Confusion is also possible between old and new style dating.4a There has always been a proportion of undated books, especially in the machine-press period when cheap reprints were left undated so that they could be sold over periods of years without advertising themselves as old stock.

The most usual problems of identification, indeed, concern imprint information—printers’ and publishers’ names, places and dates of publication, which are all interconnected—and their solution is best approached by means of a systematic general investigation. The procedure is to date and place the problem book approximately—it is assumed that the investigator has some experience of the sort of books he is working with—and to follow this with a detailed assessment of printed clues, typography, paper, binding, and provenance.

First then printed clues, beginning with anything that the imprint or colophon does reveal. The omission of the names or initials of printer and publisher itself suggests secretiveness, and the practice was often combined with printing a false place of publication. A good many French books of the eighteenth century, printed in a plainly French typographical style, have ‘À Londres’ or ‘À Amsterdam’ in their imprints; they were printed in France, but their printers and publishers used this doubtless transparent device to evade the displeasure of the authorities. Similarly, a search should be made for any initials or dates appearing elsewhere in the book. Next the positions, conventions, and typographical style of the signatures, catch-words, pagination, press figures, and imprint dates should be noted, any of which could be strongly characteristic of particular places or periods.5

Turning now to typographical evidence, we have seen that the sum of a printer’s typographical equipment—all his founts of type in their various stages of wear, his blocks, initials, and plates—was always in practice unique to him, and if the typography of the book under investigation can be matched exactly with the work of a known printer—which is easy if mixed founts and woodcuts are involved—it is virtually certain that he printed it.5a Even if this cannot be done, a study of the types will often reveal when they were commonly used, and where. This can also be a powerful tool in the machine-press period—the famous Carter and Pollard investigation depended partly upon the identification of a mixed fount of type6— but the widespread use of stereos and electros, which could produce an impression far removed in place and time from the original setting, should be kept constantly in mind. Type stocks become much more difficult to identify in the period of mechanical composition, when the majority of printers were using closely similar sets of Linotype and Monotype matrixes. On the other hand the numerous different processes used for reproducing illustrations in the machine-press period may be used in dating.

The evidence of the paper is chiefly valuable for placing and dating books of the hand-press period. The watermarks of hand-made paper, or certain combinations of marks, can usually be assigned to their country of origin and approximate period, sometimes (as when individual pairs of moulds can be identified) to within fine limits. There is much evidence to suggest that printers regularly bought paper from particular wholesalers, and that apart from scraps they rarely used paper that was more than about two years old. It could of course happen that a printer used up a substantial batch of an old or unusual paper, but such cases were rare. Occasionally moulds were dated, though there was a tendency for dates to be repeated in later moulds: a French regulation required moulds to be dated in 1742, and French mould-makers continued to date their products ‘1742’ for decades; for similar reasons much English paper made from 1795 to 1800 was dated ‘1794’.7 It is much more difficult to date the machine-made papers of the nineteenth and twentieth centuries, which were rarely water-marked and which looked much alike wherever they were made, so that a different batch of paper may identify a new impression, but is unlikely to date or place it. A date of earliest manufacture can sometimes be established by a fibre-analysis carried out at a specialist laboratory.

The evidence of the binding cannot be accorded so much weight as that of the physical make-up of the book itself, since it may have been added long afterwards and in quite a different place. Nevertheless a binding of the hand-press period is more likely than not to be approximately contemporary with the book it covers—say within a decade or two—while, if it appears to be a trade rather than a bespoke binding, it is also quite likely to derive from the same part of the world. It will of course be remembered that a book may have been rebound at any time; and there is also the possibility, though a remote one, that it may have been rebound in a binding taken from another, possibly earlier, book. Edition bindings of the machine-press period are more useful in placing and dating than hand bindings, since they were carried out for the publisher in large batches, usually near the place of printing and often within a few years of publication.

Finally, early marks of ownership may provide useful dates, the owner often being identifiable even if his inscription is undated. They are poor guides to placing, however, since there has always been a brisk international trade in new books.


The identification of a facsimile, usually of a few leaves in a printed book but occasionally of a whole book, is really a special case of the problem of placing and dating. Facsimiles can be hand drawn, with or without the use of some form of mechanical reproduction, or they can involve photographic copying, or they can be printed from type. Mostly they are made without dishonest intent, although some have certainly been intended to deceive, and the ease with which they can be identified varies with the process used.

Surprisingly enough, expert hand-drawn facsimiles are amongst the hardest to spot. There were three nineteenth-century facsimilists called John Harris, of whom the second specialized in supplying missing leaves for early printed books in astonishingly faithful pen facsimiles which can be difficult to tell from the ordinary printed leaves, especially if the rest of the book has been washed and pressed. But no facsimile can resist identification once it is suspected. Examination under a glass will show the hand-drawn letters to have cleaner edges than the printed ones; more important, the paper will almost certainly show some suspicious feature. Even though the facsimilist’s paper is of the same period as that of the rest of the book, he is most unlikely to be able to match it precisely in all its characteristics: thickness, texture, colour, chain-lines, watermark, and the propinquity of worm-holes and stains.

A word should be interpolated here about ‘made-up’ copies. Although it is less common than it used to be, booksellers and collectors sometimes complete an imperfect copy of a valuable book by supplying missing leaves from another, yet more imperfect, copy. In this case, too, any worm-holes and stains in the inserted leaves are sure to be out of line; and although the paper of the inserted leaves will be of the correct thickness, texture, and colour, the chain-line and watermark patterns of the gatherings will be interrupted. (Cancellation, of course, can have a similar effect.)

The other forms of facsimile are easier to identify. Hand-drawn lithographs, anastatic prints, and photolithographs all show the flat, rather lifeless, texture of lithography, and they were seldom printed on early paper. Photo-etched blocks, which became available in the 1870s, give the colour and impression of letterpress printing but they lose definition in the process of reproduction, so that line-block facsimiles combine rough edges with an improbably even colour. Type facsimiles, which have been attempted at various times since the very early days of printing, can be identified immediately if they can be laid beside a copy of the original; and, even if they cannot, they are likely to be given away by anachronistic type and paper.

1 See Bowers, F., T., Principles of bibliographical description, Princeton 1949 (and later impressions), chs. 2, II.

1a For instance the same spool can be used for casting either series 101 Imprint or series 110 Plantin.

2 See the bibliography, pp. 392, 401–2, 410–11, for a survey of some of these sources.

3 Payne, J. R., ‘Four children’s books by A. A. Milne’, Studies in bibliography, xxiii, 1970, pp. 127–39.

4 Nichols, J., Literary anecdotes, London 1812, iii, p. 249 n.

4a See Greg, W. W., Collected papers, Oxford 1966, p. 372.

5 Sayce, R. A., ‘Compositorial practices and the localisation of printed books’, The Library, xxi, 1966, pp. 1–45.

5a But see p. 39, n. 36a.

6 Carter, J. W., and Pollard, H. G., An enquiry into the nature of certain nineteenth-century pamphlets, London and New York 1934.

7 Arrêt du Conseil du 18 Sept. 1741; the Excise Act 1794, 34 Geo. III, c. 20.