University of Virginia Library

Heading (D). The approach to reference numbers and letters reflected here is set forth in Tanselle 1984, pp. 28-38. Underlying it is the idea that bibliographers should think independently about what will be most efficient in a given situation and not automatically assume, for example, that all the separate publications by an author must go into a section labeled "A"—a system that produces unnecessarily cumbersome references, since a typical "A" section is likely to require more complex reference numbers than the rest of the bibliography anyway, and attaching "A" to every one of them only complicates matters. One solution, as here, is to use letters to stand for an author's successive separate publications: Redburn, Melville's fourth book, is therefore taken up in a section labeled "D." This system works well with Melville, for the treatment of his book-length works would require the letters (skipping I) through about T, leaving the remaining letters to be employed in the more conventional way, designating sections recording multiple works (in this case, such categories as shorter works, collected sets, collected volumes, and appearances in anthologies). Of course, an author who wrote more books than Melville did would have to be handled differently, either by going into a doubled alphabet or (probably better) using numbers instead of letters (and then using letters instead of numbers for the next digit—e.g., "4A" instead of "D1"). In the heading illustrated here, the presence of the year 1849 is not meant to suggest that the section deals only with the first edition; this heading stands at the beginning of the entire section on Redburn, and the year is present only to aid the reader by showing the date of first publication and therefore the basis for the ordering of the sections. (In the present instance, one might wish to use "September 1849" instead, since Melville's preceding book, Mardi, had also appeared in 1849.) See also the comments on the D1 heading below.

Introduction to the section. Because each section dealing with the entire publishing history—from original publication to the present day—of one of Melville's books is fairly extensive, it seems useful to have an introduction to each section, providing the reader with a concise summary of that publishing history. This introduction may, but need not, contain information not present in the actual description; its function is to draw together in one place the salient facts established elsewhere and thus to assist readers by providing them, in effect, with a general outline of what is to follow. By combining facts from individual descriptions, it can give an overview that will help readers to place particular editions and printings in a larger context. Textual relationships among editions are an important aspect of publishing history and should be covered, along with key physical points and facts drawn from external sources. (Identification of sources used is unnecessary here if the descriptions that follow do, in one way or another, provide it.)

Heading (D1). Following the system in Tanselle 1984, esp. p. 37, the arabic number following the letter designates the edition number. The rest of the heading is meant to offer a convenient shorthand designation in words for the edition under discussion. To say "First Edition" here would add no information to what is already conveyed by "D1," and it seems sensible to convey additional information in such headings whenever possible. Moreover, the phrases "original British" here and "original American" later emphasize a more important point than the simple priority in publication, by a few weeks, of the former over the latter.

Introductory discussion (following the D1 heading). For each edition treated in detail, one may wish to begin with some introductory commentary, partly to serve the same function for the edition that the earlier paragraphs did for the work, but also to assemble into a coherent narrative such evidence as there is regarding contracts, publication and copyright dates, the author's relations with the publisher, and the like. (On copyright, see Tanselle 1969.) In effect, all the kinds of details that come from sources outside the copies of the edition themselves can be gathered here, leaving the descriptions that follow to concentrate on the physical evidence. Many bibliographical descriptions in the past have placed such information in one or more sections called "Notes," following the treatment of physical details, and sometimes these sections are not narrative in form but are simply enumerations of relevant facts. This approach remains one possibility, of course, but I show here a different approach, which reflects more directly the concept of a bibliography as a work of historical scholarship. The crucial point is not the manner of presentation of the material but the underlying conception of the work: a bibliography is not a listing of points for identification but a historical account, partly biography and partly a segment of publishing history (which itself encompasses the history of the arts of book-making as well as economic and intellectual history). This view of bibliography underlies Bowers's Principles and his 1953, 1966, and 1969 essays; see also Tanselle 1984 (passim, but esp. pp. 2-3 and the references in footnote 3) and Tanselle 1985 (esp. pp. 46-61), as well as Vander Meulen 1985. Providing narrative introductions of the kind I illustrate here does serve to drive home the point—both to authors of bibliographies and to their readers—that bibliographies are histories and that, like all histories, they attempt to offer coherent readings of the past.

Different bibliographers will make different selections of details; what I show here seems to me one reasonable selection, but there are always more details that can be included. One point naturally to be kept in mind is what kind of scholarly coverage already exists. In the case of Melville, the existence of the Northwestern-Newberry Edition, with its historical and textual essays and records of variant readings, would make redundant in a bibliography an extensive account of the composition and reception of Melville's books or lists of variant readings. The research underlying a scholarly edition and that underlying a scholarly bibliography are complementary, and the published results can be complementary, too. Thus when, in the present introduction, I refer to the 79 substantive variants between the Harper and Bentley editions of Redburn, there is no need to record them, for they are already available in the Northwestern-Newberry volume, and readers of the bibliography would have been told in the general introduction to the whole work that for complete lists of variants they should turn to the Northwestern-Newberry Edition. I am free to be selective in my use of variants, and I have decided here that only one deserves mention, given the scale on which the bibliography is being written. (However, variants between copies of a single printing—such as those created by stop-press alterations or by type or plate wear—demand fuller treatment somewhere in the description, even if they have been reported in another work; see the paragraph on typography below.) Similarly, documentation of frequently used sources need not be repeated in each of these introductions; instead, the general introduction can explain, for example, that quotations from Melville's letters always come (unless otherwise specified) from Merrell R. Davis and William H. Gilman's edition (1960) or that certain kinds of detail always come (unless otherwise specified) from the Northwestern-Newberry Edition. Documentation would then be supplied only for unusual sources or for sources inconvenient to refer to without specific references (hence the citations here of particular pages in the volumes of Bentley Papers at the British Library). For those details repeated in this introduction from one of the later descriptions, documentation—if needed—would come at that later point, where the matter is likely to be more fully treated. I do not find any awkwardness in the shift from narrative in these introductions to the more formulaic style of the physical descriptions: the two parts deal with very different material. A formulaic rendering of dates, quotations, sums of money, and the like seems to resemble undigested raw material and is not the most effective way of conveying what is in fact a narrative. On the other hand, the formulaic style is generally more efficient in laying physical details out clearly and more convenient for readers who wish to check copies against the description. (Cf. Tanselle 1984, footnote 6.) I ought to add, however, that one should never hesitate to use sequential prose within a physical description when it seems to offer a clearer way to make certain statements.

Heading for first Bentley issue. No entry number is assigned here, in the belief that descriptions of issues should not be regarded as separate entries but rather as subsections of the entry for the printing of which they are a part. It should also be recognized that furnishing entry numbers for issues would complicate any system considerably and thus perhaps make it less efficient: this Bentley issue would have to be numbered—in the system I am using, which is essentially a very simple one—D1.1a. It could not be labeled "D1a," for lower-case letters attached to edition numbers ought to refer to subeditions, so that numbers could be attached, following a decimal point, to signify printings. Clearly an issue designator has to be associated with the number referring to printing, but since this Bentley edition consists of only a single printing, it seems unnecessary (indeed, illogical) to call it a first printing. On this general matter, see Tanselle 1984, esp. p. 37 and footnote 64. The concepts lying behind the terms "edition," "subedition," "printing" (or "impression"), "issue," and "state" are set forth in Bowers's Principles, pp. 37-113, 371-426, supplemented (and somewhat modified) in Tanselle 1975. When separate platings (e.g., duplicate relief plates made from the type, or offset plates made photographically from a copy of a printed book) can be identified they should be noted (perhaps in the heading); but, contrary to the argument of James L. W. West III ("The Bibliographical Concept of Plating," Studies in Bibliography, 36 [1983], 252-266), it seems advisable to use subedition (i.e., any printing or sequence of printings with a different publisher's or series imprint) rather than plating as the classification between edition and printing (for an explanation, see Tanselle 1984, pp. 10-20).

Title-page transcription. The dominant system of quasi-facsimile transcription, as developed over the years, is the one codified, extended, and thoroughly explained in Bowers's Principles, pp. 135-184, 427-429. It attempts to report more typographic features (such as line endings) than does ordinary quoting, but it does not suggest relative sizes of type, except when both large and small capitals appear in the same line. (However, David Vander Meulen's 1981 dissertation, a descriptive bibliography of Pope's Dunciad to 1751, shows that a type-face measurement can be easily and unobtrusively added in brackets at the end of each line of the transcription.) That such transcriptions are not superseded by photographic facsimiles is touched on in the Principles, pp. 135-137, and in Bowers 1953, esp. pp. 9-15; and the argument is most fully set forth in Tanselle 1985, pp. 46-61. One should assume that the description of Redburn presented here would in the actual bibliography be accompanied by appropriate reproductions of title-pages, but their presence would not affect the transcriptions. Photographic facsimiles are illustrations that supplement the verbal account but do not substitute for it. It is also imporant to remember that title-pages are not the only parts of books that may usefully be transcribed in quasi-facsimile (and not necessarily the most important ones).

Collation. The system of signature collation that has evolved in the English-speaking world since the late nineteenth century received its classic treatment (superseding several earlier landmark discussions) in Bowers's Principles, pp. 196-268, 332-339, 431-434, 487-499 (with "A Digest of the Formulary" on pp. 457-461), now supplemented and somewhat revised in Tanselle 1985, pp. 61-80. (See also Greg 1959, Stevenson 1961, and Margadant 1968.) The principal function of the collation is to show the physical structure of books, and the collation is therefore the indispensable heart of any description (more important than the title-page transcription), though some bibliographers of twentieth-century books have attempted to argue otherwise. (Redburn contains no inserted plates; but when they are present, they obviously complicate the task of recording structure—see the Principles, pp. 287-289, Stevenson 1961, Bridson 1976, and Tanselle 1982.) A record of where the signatures appear and in what form is a separate (and lesser) matter, but it should be provided as a supplementary statement to the collation. The signature statement in this sample follows the form in Tanselle 1985, pp. 78-80, which only slightly modifies the form suggested by Bowers (Principles, pp. 269-271). (On the use of the dollar sign to stand for a signature and thus to refer to every gathering, see the Principles, pp. 266-268.) In the collation given here, I have assumed the 23-letter alphabet (without J, either U or V, and W) traditionally used by printers, and this fact would be explained in the introduction to the bibliography; but for books of this period some bibliographers might prefer to be more explicit and write "B-I12 K-O12" instead. Indication of format (here "12°") must be included when it can be ascertained, since the regular number of leaves in gatherings does not necessarily reveal the format (the number of times each sheet was folded); when (as in many twentieth-century books) the format cannot be determined, measurement of the leaf size can be substituted, as Bowers suggests. A helpful assemblage of imposition diagrams for various formats is found in Gaskell's New Introduction, pp. 84-107. On the rules for recording pagination (and the crucial one of not combining unnumbered and numbered pages with a dash, as in "i-viii"), see the Principles, pp. 271-284, 435-438, 462 (but numbered and unnumbered pages can be combined in the contents paragraph or elsewhere, the facts having been put on record here).

Contents. The basic rules are in Bowers's Principles, pp. 289-299, 438-444. In a book made up of numerous short pieces, such as stories or poems, one should not simply have an entry for "text" but should enumerate the contents (either in this paragraph or in a separate paragraph below, where references to their original appearances in periodicals, for example, could also be included, perhaps in columnar form). Any words that one wishes to quote might as well be quoted in quasi-facsimile, since it conveys more information and scarcely takes up any more space; and one should wish to quote instead of paraphrase whenever feasible (e.g., quoting the form of the heading to the table of contents instead of simply referring to "table of contents"). I show here the convention of using single quotation marks to signal quasi-facsimile transcription and double quotation marks to indicate ordinary quoting. Punctuation supplied by the bibliographer must therefore go outside the single quotation marks but may—if the American system of punctuation is being followed—go inside the double ones.

Paper. The system followed here and the considerations underlying it are set forth in Tanselle 1971(1), where references to much of the relevant literature also appear. Anyone attempting to describe paper should be acquainted with at least as much historical background as is provided in Gaskell's New Introduction, pp. 57-77, 214-230, before beginning to examine the relevant specialized works. Those dealing with pre-nineteenth-century paper should certainly know the great series of essays by Allan Stevenson (1949, 1952, 1954, 1962) and Vander Meulen's important 1984 article (a worthy successor to that series). Vander Meulen's own descriptions of paper in his bibliography of The Dunciad (his 1981 dissertation) are exemplary, showing an orderly way to set forth mixtures of papers (e.g., pp. 72-77). Measurements here (and throughout) are in millimeters, though one need not use the metric system; it is extremely convenient, however, and Vander Meulen notes its particular convenience for paper measurement (pp. 60-61). My measurements of paper thickness (made by micrometer) are to the nearest thousandth of a millimeter, and the leaf and total-bulk measurements to the nearest third of a millimeter. (The bulking thickness of each sheet is of course arrived at by dividing the total bulk by the number of leaves—and rounding off to the nearest thousandth of a millimeter.) The tolerances I have set for myself would be specified in the general introduction to the bibliography. On the importance of stating tolerances, see Tanselle 1968(1). The present description represents a moderate level of detail, which I have tried to adhere to in other paragraphs of the description as well. More details could be given, but more would probably then have to be given in other paragraphs, so that the proportions of the whole description would remain balanced.

Typography. The approach illustrated here is essentially that advocated in Tanselle 1966 (1), where the value of recording these details in a description is explained. Bowers's Principles includes a brief treatment of typography (pp. 300-306, 344-347, 444-446); and, as with paper, Gaskell's New Introduction gives the basic background (pp. 9-56, 201-213; but see Vander Meulen 1985, p. 215). The goal in this sample, with a moderate amount of detail, is to suggest the principal characteristics both of the type faces used and of the typographic layout. Even at this modest level of detail, the identification of the text type might well be carried further: for a rare example of meticulous attention to typography in post-1800 books, see Nicolas Barker's notes on types in David Gilson's A Bibliography of Jane Austen (1982). The paragraph on typography is the place to draw together comments on all aspects of what is printed on the sheets, and when there is a great deal to record regarding running titles and catchwords, as often in pre-1800 books, it may be necessary for ease of reference to provide special subsections, separated by spaces from the main paragraph, to cover these matters. A splendid treatment of such problems (but one that is unfortunately not very accessible) is Vander Meulen's 1981 dissertation, his bibliography of The Dunciad to 1751, where the sections on typography are sometimes the longest in a description because they include subsections recording running titles, catchwords, signatures (which may be regarded as an aspect of typography and need not be recorded in the collation paragraph), and ornaments (sometimes with tables to show patterns of recurrence). He also sensibly offers a section on presswork, where he records first-forme impressions (whether the inner or the outer of each sheet), point holes, and press figures (the latter in the form suggested in Tanselle 1966(2)). What I have here labeled "Typographical variation" might better be placed in a paragraph on presswork, for these variations are certainly not part of the planned typography of the book. On the other hand, they are a characteristic feature of the typographic impressions that appear on the sheets being described. In the case of Redburn these variations consist only of unimportant shiftings of type during the course of printing; the principal examples are noted (by page and line, with empty brackets to show where space exists as a result of shifting type and with each instance documented by reference to individual copies—see the paragraph on copies examined below); but they do not need to be reported in full or given the prominence of a separate paragraph. (On the general use of type damage, see Tanselle 1968(2).) However, stop-press textual alterations are indeed important and deserve to be reported in full (or perhaps selectively, if already reported in an edition) and keyed to individual copies; therefore they may, if extensive, require a paragraph of their own. The important theoretical point here concerns the treatment of states. Typographical variations create states of individual sheets; but such states—and states of all kinds—refer only to particular parts of books, not copies of books as wholes (cf. Tanselle 1975). Thus states do not call for separate entries (indeed, could not be handled clearly by separate entries) but instead should be described under whatever heading (paper, typography, binding) encompasses the particular variation involved.

Binding. The term "binding" is used here to refer to any covering of the sheets that was supplied by the publisher, whether paper wrappers or a casing of cloth over boards. But a more purist usage would not allow "binding" to be used when "casing" is meant. The overall form of the description is based on Tanselle 1970 (the illustrations in which are referred to by the cloth-grain citation numbers), with colors specified in reference to the Centroid charts, as recommended in Tanselle 1967. Publishers' binding is discussed briefly in Bowers's Principles, pp. 446-450, and in Gaskell's New Introduction, pp. 231-250. (Dust jackets, when they exist, would be described either in the binding paragraph or, more appropriately, in a separate paragraph following it—see Tanselle 1971(2), where several samples are provided on p. 114. Descriptions of the endpapers and of the treatment of the edges of the leaves, along with an account of any advertisements or other leaves inserted during the binding process, do clearly belong in the paragraph on binding.) The reference here to "Bentley 2" illustrates how one can incorporate citations of illustrations into verbal descriptions. The actual Melville bibliography would include a section of photographs showing the various stampings used on the cloth casings of Melville's books; the citation here of the second such Bentley pattern does not lessen the bibliographer's obligation to describe it in words. Variations here in spine stamping and in endpapers again illustrate the treatment of states. The concept of state does not refer to whole copies of books but only to their parts, and these variations only result in states of the binding and endpapers; they are therefore properly taken up in this paragraph and do not necessitate separate entries. Every variant state must be documented by citation of specific copies from among those listed in the record of copies examined (see the next paragraph of the description).

Copies examined. The importance of a paragraph identifying the copies that provided the evidence for the bibliographical account that precedes it cannot be overestimated. But I am tempted to belabor its importance because so often it has been disregarded. A description that does not record copies examined is simply incomplete and unscholarly. Readers must know what copies were looked at, so that they will know when they are looking at additional copies and so that they can locate, when they wish, copies that display any of the features described. As set up here, the record of copies examined is organized by country, with copies arranged alphabetically under each country according to library abbreviations (those of the National Union Catalog for the United States and Canada and those of the British Union Catalogue of Periodicals for Britain, with additional ones constructed on the same pattern). Following each library identification is the shelfmark or call number, followed in turn by a reference to any post-publication alterations to the copy (such as rebinding, the excision of leaves, or the insertion of a bookplate or inscription). The much misunderstood concept of "ideal copy" simply refers to all those characteristics that copies possessed when they left the control of the publisher, as opposed to those characteristics that individual copies have subsequently acquired at the hands of their owners. The bibliographer's task is to record the former: "ideal copy" does not necessarily mean a single copy but rather encompasses all the variations of the book as published. Bowers has commented on the concept in the Principles, pp. 113-123, 404-406; the fullest analysis is in Tanselle 1980 (and cf. Tanselle 1977). The record of copies examined must note post-publication alterations of copies, at least when they result in the elimination of parts of the originals, so that readers will know exactly which copies did, and which did not, contribute to the body of evidence on which each element of the description was based. If the roster of copies is sequentially numbered, individual copies can then be referred to concisely in the body of the description. References to copy numbers in the paragraphs on typography and binding above show how variant features of the copies as published can be conveniently documented. The division should be clear between variations that are part of a book's publication history (they are taken up in the description, with references made to particular copies) and variations that occurred subsequently (they form no part of the description but are noted in the record of copies examined as peculiarities of particular copies).

Second Bentley issue. This section shows how subsequent descriptions under the heading of a single edition can be made to relate to the first description. The cancel title of the 1853 issue exhibits enough differences from the 1849 title that it seems easier to provide a new transcription, whereas the differences in collation and contents are slight enough that they can be best handled by simple statements setting forth what the differences are.

Newby issue. Sometimes an extensive search for copies fails to produce an example of some form that has been previously referred to in print. In many instances such references can be dismissed as bibliographical ghosts (though they should normally be mentioned and explicitly dismissed). At other times one may judge the reference to be trustworthy and deserving of a place in the bibliography. Michael Sadleir's assertion that he had seen a copy of the Newby issue seems to me to fall into the latter class. (His Excursions in Victorian Bibliography of 1922 and his 1923 bibliography in the Constable edition, referred to here in short form, are among the basic works that would be identified in the general introduction to the bibliography.)

D2. Original American Edition. Beginning here I shall abridge the sample (using the three-dot ellipsis mark), to avoid repetition of procedures already displayed, and shall print only those parts that illustrate something new.

Collation. This collation is shown to illustrate a problem posed by the pagination here that did not arise with the Bentley edition: there are no running titles or page numbers on the pages that begin new chapters. This fact would be reported in the paragraph on typography, but it also needs to be taken into account in the register of pagination here. For this book, it seems sufficient to insert a parenthetical explanation rather than to enumerate the 61 affected pages, but for another book (particularly from an earlier period and with irregular pagination) the decision might be different. In any case, the sequence "14-390" should not be interrupted 61 times—a procedure that would require the reader to look carefully through the entire series of numbers. If the numbers are to be recorded, they should be brought together in the appended parenthesis.

Contents. Only the end of the contents paragraph is shown here to make the point that advertisements on integral leaves must be reported in this paragraph, because they constitute part of the contents of the sheets. (But advertisements that form a separate gathering and were inserted by the binder are not included in the collation or contents paragraphs but only in the binding description.)

Typography. The section on typographical variation is given here because it raises a different problem from the shifting type in the Bentley edition. Here there is some question whether two 1849 printings in fact occurred, and the uncertainty must be explained (with a cross-reference to the fuller comment at D2.2) so that readers have a basis for judging for themselves the status of copies they examine.

Bindings. Here the heading is in the plural and the paragraph divided into two subsections, since the Harpers made Redburn available in both cloth and paper. The description of the wrappers would conform to the same plan as that of cloth bindings: description of the material (whether paper or some other material, and its style, thickness, and color) would be followed by an account of the front cover, the spine, the back cover, and the edges (and, when there is printing on them, the inside covers). Because wrappers are generally of paper and contain a considerable amount of printed text, the problems they pose for description are analogous to those offered by dust jackets—see Tanselle 1971(2), which includes sample descriptions (p. 114). The subsection on wrappers would of course identify (by copy number) which copies examined are in wrappers—just as any other feature not identical in all copies is documented by reference to specific copies. I have treated the wrappers here as a variant binding, not as a feature indicative of a separate issue, although one could argue that they represent a discrete (if simultaneous) marketing effort. In an instance of this kind (as opposed to the Bentley situation above), it is a matter of little moment whether or not one introduces the issue classification: to do so would make the arrangement of material under D2.1 somewhat less convenient and would produce no compensating advantage (conceptual or practical). One should in general be conservative in the designation of issues, and when there is no compelling need for the classification it probably should be avoided. For an exposition of the view that such decisions ought to be a matter of reasoned judgment, not of mechanical rule, see Tanselle 1975, esp. pp. 42-56, which also develops a working definition of issue.

Later Harper printings. The entries for D2.2 through D2.6 show how descriptions of later printings can build on the full description of the first printing and on each other, variously using prose paragraphs or more formulaic statements as the occasion warrants. In D2.2 a discussion of the problematical status of the second printing must come first, and since it explains the difference in collation and contents and draws on the typographical evidence, no separate paragraphs on those matters are called for. Bindings, however, are a separable matter and are best described here in a conventional binding paragraph. In the succeeding entries, D2.3 through D2.6, typographical variations play no role in identifying printings, and they are not taken up in the opening statements, which in these cases only need to be concerned with questions of title-page wording and dates, collation, and contents; typography as well as binding therefore has its own paragraph in these entries. And all entries must have a record of copies examined. The entries for these later printings, by drawing on prior entries, can be relatively concise, and yet the later printings are described no less fully than the first printing. (On the question of when to give less full descriptions, see the comments below on D4a.)

D3. "British Library" edition. Because the Galignani-Baudry edition was pirated, it is of no textual significance for the scholars wishing to establish Melville's intended text. But its text was the one most readily available to readers on the continent in the 1850s, and its very existence is an important fact of Melville's nineteenth-century publishing history. It was the only edition of Redburn in English, other than the two authorized ones, to appear in the nineteenth century, and it was one of only five unauthorized editions in English of Melville's works during the century. I would give it full bibliographical treatment—as I would all editions and printings of Melville's books (in the original language, that is) during his lifetime. Others might wish to draw the line differently (this question is further discussed below, at D4a).

D4. "Standard Edition." Whether to describe a collected set at a single location, or to take up each volume separately at the point where it would have been described if it had not been connected with a set, is a question that should be decided individually for each bibliography. More important is the recognition that the individual volumes of sets must be entered at two places, once with a primary entry and once with a cross-reference. Because some economy, and perhaps clarity as well, results from treating an entire set at one place, I have chosen here simply to make a reference to the place—in section V, devoted to collected sets—where the Constable edition is described. The same treatment is accorded two subeditions (i.e., printings bearing another publisher's imprint or series designation) of the entire set (D4c, D4d). But the two subeditions of Redburn alone (D4a, D4b) are best treated here. On the arrangement and numbering of entries for subeditions, see Tanselle 1984, esp. pp. 9-20, 37. (Some might argue—and they would have a point—that the original printing, with the Constable imprint, should be numbered "D4a," allowing "D4" to refer unambiguously to the entire edition, encompassing all its subeditions. On the other side is the awkwardness of making the originating publisher's printings appear to be a subedition. Neither solution is entirely happy, and both of them can be defended.)

D4a. "Constable's Miscellany" subedition. This entry is an example of a condensed form of description that I suggest might be used for all editions of Melville's works after 1891, the year of his death. The idea of shifting, at some point in a bibliography, to a less detailed form of description—a concept sometimes called, following Falconer Madan, "the degressive principle"—has been much debated. By far the most thoughtful treatment of it is to be found in Bowers 1969 (supplemented by Tanselle 1984, pp. 20-28). The decision where to curtail description (if full description throughout is not deemed feasible) involves thinking through the primary purpose of the bibliography: the line would be drawn at one point if the focus is on textually important editions, at another if the aim is to write a biography in bibliographical form, and at still another if the emphasis is on how printers and publishers have handled the work of a classic author. But wherever the line is drawn, one must understand that shorter descriptions do not necessarily mean less research: one is still trying to establish and report certain facts, and the space finally given to them does not necessarily reflect the amount of time required to establish them. Furthermore, the condensed descriptions should still have logic and balance of their own. In the form proposed here some attention is paid to paper and to binding (nothing need be said about typography in an entry for a subedition, unless the typography is altered photographically, electronically, or by means of a Monotype roll); the recording of contents is selective, but the signature and pagination collations, being indispensable, are treated no less fully than they would be for a nineteenth-century edition. No precise form for condensed descriptions can, or should, be prescribed; the description I offer here is intended only as one example of a highly condensed and yet relatively well-rounded bibliographical account.

D4b. Richard R. Smith subedition. This entry, on the transatlantic counterpart to the preceding subedition, illustrates how one condensed description builds on another. (If the Smith publication were an issue of the "Constable's Miscellany" printing and not a separate printing, it would not receive an entry number and would be described in a subsection of D4a—following the principle of treating printings as entities, explained in Tanselle 1984, p. 18.)

D5. L. C. Page edition. A month is included in the heading here to show why the Page edition is discussed before the Boni edition, both of which were published in 1924. The Page parent edition and some of its subeditions went through more than one printing; in conformity with the style of the condensed descriptions, an enumeration of the printings of each subedition (with identifying numbers, such as "D5a.2," attached and with significant differences noted, including an indication of whether title-page dates are altered and whether printings are identified on title-leaf versos or somewhere in the books themselves) could be run on in paragraph form (constituting a second paragraph of the entry). Because the entry for the later Cape "New Library" edition is some distance from this entry, a cross-reference linking the two Cape imprints is useful (they would also be linked in the index, of course).

D7. Pickwick edition. A collected volume poses some of the same questions as a collected set (see the discussion of D4 above), but the matter is more easily decided. Wherever this volume is described, there will have to be seven cross-references to it, and it makes more sense to have the main entry in the section devoted to collected volumes than in the section dealing with any one of the works included in it. The presence of a cross-reference here is important, even though this edition of Redburn does not constitute an independent volume, for the goal is not simply to record separate volumes but to convey the publication history of the work. The inclusion of an edition of Redburn within the Romances volume is a significant part of that history (as is the inclusion of another edition within a "Library of America" volume, recorded here at D12).