University of Virginia Library

The Bibliographical Description of Patterns
by
G. Thomas Tanselle

The basic technical problem of bibliographical description arises from the difficulty of expressing the visual in verbal terms. Parts of a description like the collation and contents paragraphs, which are condensed statements of sequential arrangement, can be handled easily enough in words or formulas; but the title-page transcription and the paragraphs on binding, paper, and typography present the same challenge that one meets in the attempt to frame an exact (not impressionistic) description of any physical object, whether it be a tree or a sculpture. Books are not exempt from the human urge to decorate empty spaces, and the descriptive bibliographer is faced with a wide array of patterns and designs (as in binding cloths and endpapers) which he must somehow record in a fashion precise enough to serve as an identification.

Among the decorative elements of a book, patterns and illustrations can be usefully distinguished. A pattern results when a figure (or combination of figures) is repeated (either exactly or approximately) at regular (or irregular) intervals or in a systematic arrangement; if a figure (or combination of figures) appears only once, the result is an illustration. While illustrations are frequently representational and patterns abstract, these qualities do not serve to distinguish the two, since a representational figure can be repeated as the motif in a pattern and an abstract figure can be used by itself as a single decoration. Repetition — whether precisely detailed or suggestively vague — of a basic unit is the essential feature of a pattern, and it provides a means for classifying the pattern. Although the number of possible patterns is infinite, individual patterns bear structural relationships to one another and can be grouped into a limited number of families. Whenever such a framework can be established as a standard of reference, verbal descriptions can become both more concise and more exact. Illustrations, by their nature, are less readily amenable to identification on the basis of a structural classification (though they can be

classified by technique of reproduction) and therefore present a separate problem. Pragmatically, from the bibliographer's point of view, illustrations can best be handled, as they come up, on an individual basis; but patterns, which occur in some form in nearly every book, can often be treated most meaningfully and efficiently in terms of a classification scheme.

There are three ways in which patterns can be recorded in a bibliography — in pictures, in words, or in a combination of the two. The first is the most straightforward and explicit method and at the same time the most objectionable — and not simply on the grounds of expense. To rely exclusively on photographs of patterns is to abandon description in favor of reproduction. The task of a description is to provide a verbal identification which can be quoted in contexts where pictures are inappropriate. Pictures may of course be useful supplements to a description, but they are not substitutes for any part of it. The opposite extreme, of describing patterns exclusively in words, can be successful and precise only if an adequate vocabulary has evolved. In heraldry, for example, the technique of blazoning utilizes a special vocabulary and syntax which make the resulting descriptions both concise and unambiguous. The bibliographer has no such established terminology to draw on in describing patterns, except perhaps in the case of marbled papers. As a result, bibliographical descriptions of patterns must use some kind of combination of words and pictures. That is, the description itself will contain only words (preferably a standardized wording which will convey the same meaning to a large number of readers), but part of that verbal description will be a reference to a readily accessible illustration; in this way any reader who cannot visualize the pattern from the verbal description clearly enough for his purposes can look up the illustration which serves as a standard of reference.

The two essentials, then, in the bibliographical description of patterns are a standard terminology and a visual standard of reference. In a few areas, bibliographers already have such references at their disposal — R. B. McKerrow and F. S. Ferguson's Title-page Borders Used in England and Scotland, 1485-1640 (1932) is a good example — but for the most part no accepted standards exist. Enough research has taken place in certain areas, however, that it would not be premature to attempt to codify a standard of reference for those areas, particularly if it is set up in an expandable fashion, so that additions can be made in the future without affecting the basic classification. Patterns in books generally appear in one of three media: in binding cloths, both

as a grain and as a stamping upon the grain; in decorated papers, whether used as endpapers or as a covering for boards; and in sheets of letterpress, either as borders or as divisional indicators. Each of these areas involves special problems and should be considered separately.

I. Binding Cloths

Before the introduction of edition binding in the 1820's, the binding of a book generally had no connection with the publisher and was not part of the book as it was issued to the public. The description of such bindings, executed individually for owners of books, is thus outside the scope of the descriptive bibliographer's task and constitutes a separate field of investigation.[1] The only exceptions are the temporary coverings in which books were sometimes issued to serve as a protection until the books could be properly bound;[2] these covers usually consisted of plain boards with printed paper labels, and they offer little difficulty for the descriptive bibliographer. From the 1820's on, however, most books have been issued in bindings or casings by their publishers, and these coverings must be described in bibliographical descriptions. The most common material for publishers' bindings has been cloth, and the history of publishers' cloth has been traced by several bibliographers — notably Michael Sadleir, John Carter, and Joseph W. Rogers.[3]

The historical investigation of publishers' cloth, while it provides a perspective for viewing specific cloths, does not in itself furnish the framework for classifying them. Sadleir's book of 1930 took the first step by including four photographs of cloths, showing their distinctive textures or "grains"; Carter's of 1932 (which displayed twelve photographs) discussed the problem in "A Note on Terminology" (pp. xvxviii) and worked out the equivalences between Sadleir's descriptive terms and the letter designations used by the Winterbottom Book Cloth Co. Ltd., the chief manufacturer of book cloth; and Rogers's 1941 essay furnished illustrations of eleven cloths, labeled with the Sadleir-Carter terms (Plates 30-40). It was not until 1951, however, that a collection of photographs of cloth grains was published which could serve as a comprehensive standard of reference. In that year Michael Sadleir included, at the end of the first volume of his XIX Century Fiction, illustrations of twenty-four cloth grains, labeled with descriptive names such as "sand grain," "hexagon grain," and "dotted-line-ribbed." Although the photographs were based on the Sadleir collection, that collection was extensive, and the photographs could be taken to represent most of the grains in common use in the nineteenth century;[4] but the photographs were not arranged in terms of any overall system of classification, and their physical location in a large two-volume reference work, which could not always be at hand when a bibliographer needed to identify a cloth, limited their influence as a standard.[5] Four years later, in 1955, appeared a second large collection of photographs. Jacob Blanck, at the beginning of the first volume of his Bibliography of American Literature,[6] provided illustrations of

twenty-eight grains, identified by letter designations (based on the system used in the cloth trade), such as "T" (for ribbed cloth) and "Z" (for what Sadleir calls "honeycomb"). Blanck's selection, with four more photographs than Sadleir's, covered nine grains not present in Sadleir and omitted five which Sadleir included.[7] The fact that a large and important reference work like the BAL keyed binding descriptions to photographs by means of letters was bound to be influential, and some bibliographers now refer to "T-cloth," for instance, without finding it necessary to explain the term or provide a reference; yet the system has the same disadvantages as Sadleir's — and in addition utilizes terms which convey no meaning except to the initiated.

Both systems have had a beneficial effect to the extent that they have called attention to the necessity of photographic samples as standards for verbal descriptions. At the same time, the two systems have dramatized the essential split between the two kinds of verbal descriptions: those which employ ordinary words and convey a meaning to every reader, and those which use technical symbols and convey a meaning only to those who have been introduced to the symbols. Both are based on photographs, but the selections are different, and neither suggests a systematic classification. Here the matter rested until 1967, when Martha Hartzog worked out, for the first time, an outline which shows relationships among types of cloth grains.[8] What was needed, as she recognized, was "an overall organizing principle which is consistent and logical, involving a symbol system which is concise and meaningful" (p. 115). She set up seven basic categories of grains, using terms largely derived from Sadleir — Morocco, Pebbly, Beaded, Geometric, Rippled, Striped, Woven — and designated the initial letter of each category as its symbol. Distinct patterns falling within these groups were then numbered — "Sandy," for example, as the second style in the "Pebbly" category, was "P2." Varieties of patterns could be indicated by modifying adjectives, so that "Coarse sandy" became "coarse P2." In Miss Hartzog's words, the system indicates "differences of degree by an adjective and quality by a separate variant number" (p. 118). Although she did not provide photographs, her chart indicated the correspondences between her terms and the illustrations of Sadleir, Blanck, and Carter.[9]

The Hartzog system is important as the first attempt to provide "a coherent organizational framework within which all cloth bindings can be fitted" (p. 118), and the nature of the system is a revealing indication of the kinds of problems involved. The choice of the seven main divisions, for example, is not self-evident. Since "Geometric" is an inclusive term which subsumes many patterns of geometric regularity, perhaps "Beaded" would more properly be classified as a variety of "Geometric" than as a separate division parallel to it. On the other hand, if the main divisions are to have more specific headings like "Rippled" and "Striped," then perhaps some of the varieties of "Geometric" — like "bead & line" or "herringbone" — deserve to be elevated to the status of generic divisions. The difficulty in formulating a classification of patterns is that there is no natural spectrum on which to base it. In the study of color, one begins with the concept of a spectrum or a solid representing the range of all possible color, and each of the infinite varieties of color has its unalterable place in the scheme. But cloth patterns are artistic, not natural, products, and no natural continuum encompasses them. Is "Striped" a development of "Rippled"? Is "Woven" closer to "Striped" than it is to "Pebbly"? One could construct an argument for various kinds of relationships, but each would be subjective and none would be definitive.[10]

The symbols employed in the Hartzog system further reveal this problem, for such symbols as "P2" and "G7" combine two principles of notation. The letters are intended to be mnemonic — "P" for "Pebbly" and "G" for "Geometric" — and not to show relationships, for alphabetical order cannot be expected to coincide with structural evolution. The numbers, on the other hand, have no mnemonic value but imply a fixed order within each lettered division; yet there is no reason that the patterns must be arranged in this order. Thus the numbers are merely arbitrary index figures, while the letters are suggestive; and the modifying adjectives — as in "coarse P2" — carry the symbols farther in the direction of rational content.[11] Another comparison with the field of color terminology may be helpful. In the ISCC-NBS Centroid

Color Charts (1965) the letter symbols — such as "d.yG" for "dark yellowish green" — are intended to serve only as easily recognizable abbreviations (to be used when it is inconvenient to spell out complete words), not as reference notations. For purposes of reference the centroid colors are numbered consecutively from 1 to 267, and any given centroid color may be referred to by its number ("d.yG" is 137). A single consecutive sequence of figures is feasible in the case of color because the spectrum of possible colors defines the scope of the classification. If the entire range of color is divided into 267 segments, those segments can be definitively numbered, because it is logically impossible for anyone to discover a new segment which should be inserted, say, between segments 137 and 138. In the classification of patterns, however, subdivisions can never be definitively established, since (in the absence of a spectrum) the possibility will always remain that another pattern may be discovered or a new pattern devised. An outline, with provision for expansion at any point, is therefore the most workable form of classification for material which, like patterns, does not fall into a continuous series. The Hartzog system recognizes this fact; while the numbered series of species under each major division (or genus) does not represent any necessary sequence, it at least provides for the indefinite expansion of that division, and the modifying adjectives allow for variations within the species.

The outline and illustrations offered below as a standard for the classification and nomenclature of binding-cloth grains are therefore derived from the Hartzog system, but with a number of modifications based on the rationale just presented. In the first place, it is imperative that the divisions in a classification in outline form be not only parallel but mutually exclusive. The seven divisions in the Hartzog system do not fulfill this condition, for such terms as "Pebbly" and "Beaded" are too precise to be parallel to inclusive terms like "Geometric" and "Woven"; neither are they mutually exclusive, for "bead & line," included under "Geometric" (G7), could have been placed under "Beaded," since "pebble & line" is listed under "Pebbly" (P5). The first mutually exclusive division which suggests itself is one which distinguishes those patterns that are regular from those that are irregular: regular patterns reproduce themselves precisely, while irregular ones repeat themselves only in general effect but not in exact detail. Regular patterns, in turn, may be divided into those which are lineal in their symmetry and those which are radial; both "Rippled" and "Striped," among the Hartzog terms, are lineal, while both "Beaded" and "Geometric" are radial. Irregular patterns divide themselves into

those which resemble fabric threads and those which resemble leather grains; the Hartzog heading "Woven" corresponds roughly to the former, but the latter comprises both "Morocco" and "Pebbly." Beyond these four divisions, mutually exclusive categories are increasingly difficult to formulate and cumbersome to employ, and it seems most sensible simply to list the basic patterns under them. In the numbering system suggested here, the four divisions are assigned to the hundreds-digit, and the two succeeding digits signify the particular pattern.[12] Since any pattern can be made more coarse or fine, and since most patterns can be given a watered effect or a diagonal arrangement, these variations do not properly constitute separate patterns and are indicated here by letters which can be used to form subdivisions of any of the numbered patterns.

CLOTH PATTERNS
REGULAR

• 100 Lineal
• 2 Rib
• 4 Ripple
• 6 Wave
• 8 Dotted-line
• 10 Dot-and-line
• 12 Dot-and-ribbon
• 14 Beaded-line
• 16 Weave
• 18 Net
• 20 Crisscross
• 22 Checkerboard
• 24 Diaper

• 200 Radial
• 2 Bead
• 4 Bubble
• 6 Hexagon
• 8 Honeycomb
• 10 Pansy

IRREGULAR

• 300 Fibrous
• 2 Calico
• 4 Linen
• 6 Cord

• 400 Coriaceous
• 2 Morocco
• 4 Straight-grain morocco
• 6 Pebble
• 8 Sand
• 10 Patterned sand
• 12 Whorl

Modifiers

• (a) (regular)
• b fine
• c coarse
• d moiré
• e diagonal
• f moiré diagonal

Thus "rib-cloth" is 102, "fine rib-cloth" 102b, and "diagonal fine rib-cloth" 102be. Additional patterns can be accommodated by inserting numbers between those given (only even numbers are used here to allow for this possibility) or by extending any of the series of numbers.

It should be understood that these numbers and letters are only reference figures and not substitutes for pattern names. They have no significance in themselves, but they serve two functions: they provide index figures for referring to photographs; and they make possible a meaningful arrangement of photographs, in which related patterns are grouped together. In a bibliographical description, both the name of the cloth pattern and its reference figure should always be given (followed by a designation of the color of the cloth), as in these examples:

The verbal description enables the reader to visualize the cloth — exactly enough for many purposes — without reference to any illustration; but if he needs to know more precisely the details of the pattern described, he can use the reference number to locate the illustration in the accompanying set of photographs (or in any future set based on this system). When he does so, he finds related patterns close together; but if the cloth on the copy he is checking turns out to be a variety not illustrated, this system provides for interpolation. If, for example, it is bead-cloth of a coarser texture than that illustrated as 202, he can label it "coarse bead-cloth (202c)," even though no illustration for 202c is available. The importance of a standardized terminology cannot be overemphasized: a name, when defined by reference to an illustration, can be precise only if it is used exclusively for that pattern and if the pattern is always referred to by that name. Since the terms employed by Sadleir and Carter have achieved fairly wide acceptance and utilize commonly understood words, they form the natural basis for any standard terminology. The descriptive names assigned here are essentially theirs, though for parallelism all are listed in noun forms; for bibliographical descriptions, the names of the patterns should probably be attached with a hyphen to the word "cloth" (as "bead-cloth"). Modifying adjectives are reserved for indicating variations of a pattern: "fine" or "coarse" immediately precede the name of the pattern, but, when a pattern corresponds to the norm between the two, no adjective is required; adjectives for other qualities (like "moiré"), when needed, precede the indication of coarseness ("moiré fine rib-cloth," not "fine moiré rib-cloth"). When the letters standing for these adjectives are appended to the reference figures in alphabetical order,[14] the resulting arrangement of photographs preserves the basic grouping of variations according to coarseness.[15]

Since the illustrations in Blanck's Bibliography of American Literature and in Sadleir's XIX Century Fiction have been consulted by many bibliographers in recent years and since some bibliographical descriptions refer to one or the other, it may be convenient to have a table of equivalences. The list below records the reference figure in the present system which corresponds to each of the photographs in BAL and Sadleir (and in the three previous showings of grains):[16]

EQUIVALENCES
BAL

             

A	306	CM	408c	LG	404b	T	102
AA	102bd	EC	122	LI	402	TB	118
AR	306c	FL	108	P	406	TR	106
B	304	H	124b	PD	112ae	TZ	106ae
BD	202	HC	206	PR	412	V	302
BF	202b	HT	110	RH	210	YR	304c
C	408	L	404	S	102be	Z	208

Sadleir (1951)

           

i	402b	vii	102be	xiii	106ae	xix	118
ii	402	viii	102bd	xiv	104	xx	116
iii	124b	ix	202	xv	108c	xxi	406
iv	124c	x	202b	xvi	108	xxii	208
v	408	xi	106	xvii	110	xxiii	210
vi	410	xii	204	xviii	120	xxiv	206

Sadleir (1930)

 

IVa

402

IVb

406

IVc

102

IVd

202

Carter

           

a	302	g	106be
b	402b	h	106
c	102bf	i	202
d	124	j	114ce
e	406	k	408
f	102	l	110

Rogers

           

30	402	36	404
31	102af	37	202
32	124c	38	106
33	406	39	408
34	102	40	110
35	102b

This table facilitates cross-references among the systems. If a bibliographer, checking a previously published description which utilizes Blanck's letters, finds it more convenient to look at the photograph of the cloth in the chart presented here, he can easily make the conversion to the proper reference figure. Similarly, if a bibliographer is examining a book but has only Sadleir readily available, he can match the cloth with a Sadleir photograph and later convert the description to the name and figure suggested here. What is important is not the particular photograph used but its relation to one central system with fixed terminology.

How exact the match between a cloth and a photograph should be is part of the general question of tolerances[17] and must necessarily vary with different circumstances. The dividing lines between "fine," "regular," and "coarse" — like those between "condensed" and "expanded" in regard to type faces — are not precise, and the decision to call a cloth "coarse" rather than "regular," when it falls between the illustrations of the two, will sometimes be subjective. Whether or not this imprecision takes on practical importance depends on the degree of accuracy required in a particular instance. If two states of the binding of a given book are too similar to be distinguished by means of a standard set of photographs, the bibliographer may find it necessary to include in his bibliography special illustrations of the two varieties. The standard provides a frame of reference but cannot eliminate entirely the need for individual photographs; in most cases, however, such fine distinctions in the specification of cloth grains are not necessary. Only the bibliographer who has examined a great number of copies of a book is in a position to decide the degree of precision desirable in any instance.

Since various levels of accuracy and detail are appropriate in different situations, it is helpful to think in terms of a standard series of

graduated levels.[18] On the first and least detailed level, only the distinction among basic binding materials is made — leather, vellum, cloth, or paper. Many early bibliographies, even those which give careful attention to such details as line-endings in title-page transcriptions, represent this level; but it is now generally agreed that publishers' bindings require fuller description than the simple designation of "cloth" or "paper." The next step, if the material is cloth, is to classify the design in terms of one of the four major divisions (e.g., "a lineal-pattern cloth," "a fibrous-pattern cloth," and so on). This level, like the previous one, is not sufficiently detailed for most bibliographical work. In general, descriptive bibliographies can be expected to conform to one of the next two levels: the third, involving the designation of specific patterns ("rib-cloth," "pebble-cloth," etc.); or the fourth, adding the adjectives which distinguish varieties of each pattern (as "coarse diaper-cloth" or "diagonal rib-cloth"). If further refinement is necessary in a particular case, one can move to a fifth level and specify divergences within the standard modified terms. Instead of piling on additional adjectives, one can use the phrase "a variety of"; the reference is then either to a special photograph (if an exact representation is required) or to the standard photograph, with the index figure preceded by "cf." (if an approximate match is adequate): It should be understood that "cf." is used only to indicate variations between cloths covered by the same name — between, for example, "fine bead-cloth (202b)" and "a variety of fine bead-cloth (cf. 202b)." Generally speaking, if such distinctions are required, it is also necessary to provide individual photographs of the specific cloths involved. The absence of "cf." in a given case does not imply that the match is exact but only that the fourth level is adequate. Finally, one can attain a sixth level in some instances, if the proper documents survive and the effort is worth making, by tracing a cloth to a specimen book of a particular manufacturer; but the standard pattern names employed on the preceding levels are still required to complete the verbal description.

Once the specification of the cloth grain on one of these levels is completed, there remains the problem of describing the pattern or

decoration stamped into the cloth for the edition casing of a particular book. Sometimes such patterns resemble those found on leather bindings and could be referred to by the names employed for them by the historians of the art of bookbinding decoration — such names as "Harleian" or "Grolier," "fanfare" or "cottage." But the historical associations of such terms, as well as their established use in connection with hand-tooled leather bindings, make unwise any attempt to apply them to the mass-produced designs of publishers' cloth casings. Instead, terms more immediately descriptive, like those adopted for cloth grains, seem more appropriate. Some of the phrases used in the description of leather bindings, of course, are of this sort — such as "center and corner piece" or "interlacing strapwork" — and they can equally well be applied to the stampings in cloth bindings; but normally terms must be devised to meet specific situations. Any such ad hoc description should be kept as simple as possible and be accompanied by a reference to a photograph of the design. The number of binding designs employed by publishers over the years is naturally so much greater than the number of cloth grains that it would not be feasible to attempt the compilation of a comprehensive and classified set of photographs; rather, those designs which need to be referred to in a given bibliography should be illustrated by plates within that bibliography.[19] Such a requirement does not necessitate showing every individual binding, for the principle of illustration can still be generic rather than specific: one design, used on several different books, need be illustrated only once. Since publishers frequently have employed the same binding design, as a kind of trade-mark, on successive volumes by the same author, the number of patterns which require reproduction in any single author-bibliography should not be excessively large. This system combines — as in the designation of cloth grains — a readily understood general description with a specific reference to an illustration and at the same time emphasizes, by the use of generic illustrations, the relationships among binding designs.

As an example of a complete binding (or casing) description, the

following represents the paragraph on casing from a description of the 1855 Harper impression of Melville's Typee: Such a description is understandable without reference to any illustrations and allows any reader to visualize the casing. If a reader is in doubt about the meaning of a patricular term or if he thinks he has discovered a variant but is not sure on the basis of the verbal description, he can turn to the references indicated and find visual specimens. For the cloth grain and the colors, he is referred to a comprehensive generic standard; for the ornamentation of the cloth he is provided with a plate in this particular bibliography. But in every case a commonly understood term is reinforced by the citation of a specific standard. Only in this way can the terms become standardized and take on an exact significance.

II. Decorated Papers

Paper, decorated by various processes, was used in connection with bookbinding long before the introduction of publishers' casings, and it remains today an important resource for the designers of bindings. In both hand- and machine-bound volumes, endpapers are often decorated in one fashion or another, and the boards forming the front and back covers are frequently adorned with decorated papers (in conjunction with cloth or leather spines, and sometimes corners as well).[20] Despite the widespread use of decorated paper as a binding material, little historical study of it has been made, and virtually no discussion treats it from the viewpoint of the descriptive bibliographer; the literature of the subject, though extensive, concentrates more on the methods of producing the patterns than on the history of the art. One exception is Rosamond B. Loring's classic Decorated Book Papers (1942, 1952),[21]

a pioneer work based on the study and classification of the Loring collection (now at Harvard); it should be the starting point for any bibliographical investigation of the subject.

What the bibliographer requires, in this area as in any other involving patterns, is a system of classification and a standardized terminology. Since the characteristics of patterns are determined, to a large extent, by the methods used to produce the patterns, the most meaningful basic scheme for classifying decorated papers groups them according to the processes of production. The conventional system, as it emerges from the work of Loring and other writers,[22] has four divisions into which the majority of decorated papers fall:

In most cases a bibliographer will wish to include somewhat more detail than the simple indication of "decorated paper"; his next step, then, is to place the paper in one of these categories. Marbled papers are produced by bringing the paper into contact with colors which are floating on the surface of a gelatinous solution; the designs frequently resemble the veined patterns of marble but, if the colors are combed, may take on more regular forms. The term "printed paper" encompasses all papers produced by the transfer of a design from a carved wood-block, an engraved plate, and the like. Embossed papers have raised patterns; and paste papers, produced by manipulating a flour-and-water mixture on their surfaces, can be recognized by the blurred and cloudy effect of their designs. Technological developments have tended to make these divisions less distinct by introducing different methods for reproducing characteristic designs. Marbled and paste patterns, for example, can be reproduced and printed photolithographically, and it would be inexact to refer to the result as "marbled paper" or "paste paper." What the bibliographer should do in such cases is to speak of "marbled-pattern paper" or "paste-pattern paper"; if he can determine the method of reproduction he can add that information: "paste-pattern paper, photolithographically printed." This usage avoids the ambiguity which arises when the basic terms are allowed to refer at times to the patterns as distinct from the processes and at other times to the patterns as the result of the processes.

Classifying a paper under one of these headings does not yet provide the reader of a bibliography with a very precise idea of the pattern, and most bibliographers will move to a third level of detail, which involves describing the main features of the particular pattern. A standard set of photographs would of course be helpful for this purpose, if one could be produced; but the last three divisions listed above — "Printed," "Embossed," and "Paste" — offer the same difficulties as those presented by the patterns stamped into cloth for the casings of particular books. So many variations are possible in these categories that no chart could be devised which would do more than illustrate characteristic specimens; it could not, in other words, be used as a standard for matching and identification. Under such circumstances, the most efficient and exact method is to frame in words a simple description of the prominent features of the pattern and then provide a reference to an accompanying plate in which the particular pattern is illustrated. Indication of the principal colors is also a necessary part of the verbal description, as in these examples:

Standardization of terminology, beyond the names of the basic kinds of paper and the colors, is not possible in this situation; what can be achieved is standardization of approach.

The remaining category, "Marbled," can be handled somewhat differently. Because of the nature of the marbling process, the number of types of pattern is more restricted; every specimen, though unique, corresponds in its general outlines to one of a relatively small number

of kinds of pattern. Specifying marbled papers[23] is thus analogous to identifying binding-cloth grains, and, as with grains, a set of photographs showing the principal patterns is feasible. Although much has been written on the subject of marbling,[24] there has been almost no attention given to the problem of classifying the various patterns. The two major books in English on marbling — those by Woolnough and Halfer[25] — are comprehensive enough, however, both in their discussions of methods and in their inclusion of actual samples, to provide a useful starting point for surveying the range of possible patterns. But they would not be practical choices as standards of reference because they are not at present widely available and because their emphasis is on the techniques of production rather than on classification. What is needed, from the point of view of the descriptive bibliographer, is a set of photographs of basic patterns, each illustration accompanied by a standard name and reference figure.

The outline below is an attempt to extract from the literature of marbling a meaningful and comprehensive scheme of classification. All marbled patterns fall into one of two categories, according to whether or not the colors are drawn after they are dropped onto the solution. If they are not touched, the resulting patterns (here called "Whisked") consist of irregular spots separated by veins and resemble real marble; if the colors are manipulated by a stylus, comb, or other utensil, the resulting patterns (here called "Combed") contain more regular lines, swirls, or loops and generally do not suggest actual marble. Traditional names for the individual patterns have emerged over the years (marbling dates from the sixteenth century); even though some of these names are not in themselves descriptive, they have become so well

established that it would be impractical at this point to introduce a new set of terms. Alternate names exist for several patterns, and they are recorded in parentheses in this outline, with the preference in each case given to the more descriptive one.

MARBLED PATTERNS

• 1100 Whisked
• 2 German
• 4 Hair-Vein (Italian)
• 6 Stormont
• 8 Gloster
• 10 Shell (French)
• 12 Smooth Body
• (1112d = Spanish)
• 1200 Combed
• 2 Nonpareil
• 4 Dutch
• 6 Antique Spot
• 8 Curl (Snail)
• 10 Peacock
• 12 Bouquet
• Modifiers
• (a) (regular)
• b fine
• c coarse
• d moiré
• g drag
• h moiré drag

This classification follows the same plan as the one for cloth grains and allows for expansion in the same way. The index numbers facilitate reference to standard illustrations; and the accompanying set of photographs, showing the basic patterns, can perhaps serve as the first step toward such a standard. It is printed in black and white because its function is to display patterns, not colors. Although some of the patterns are traditionally associated with certain colors, there is no necessary connection, and it is best to keep the specification of color a distinct process from the specification of pattern.

Given the nature of these names, one can learn their significance most readily by studying the photographs, but it may be helpful to point out verbally some of the characteristics of the patterns. "German" consists simply of small spots and, unlike the other whisked patterns, does not have veins. "Hair-Vein" (often called "Italian"), has, as the name suggests, a fine network of thin hair-like veins. In "Stormont" the presence of turpentine creates many small dots which give the pattern a lacy effect; "Gloster" also has fine dots on the body color, but its veins are thick and multi-colored, in contrast to the thin veins of "Stormont." The "Shell," or "French Shell," pattern displays light shell-like rings on the body color, produced by the addition of olive oil. If a pattern has veins of medium thickness and a body color not mottled with dots or rings, it may be given the name "Smooth Body," as here. The most common variety of this pattern, called "Spanish," is

characterized by diagonal streaks traversing the basic pattern, giving it a moiré effect.[26] Among the combed patterns, "Nonpareil," easily recognizable by its horizontal parallel lines, has been perhaps the most widely used. "Dutch" is similar, except that the colors follow one another in an exact sequence and can come together in an occasional curl. If the colors are drawn along an irregular path, rather than in straight lines, and spots of a lighter color are then dropped on, the result is another pattern encountered frequently, "Antique" or "Antique Spot." "Curl," "French Curl," or "Snail," as the name implies, involves the use of a frame which can produce a series of rotary movements and thus a pattern consisting of rows of coiled colors. Similarly, "Peacock" and "Bouquet" require equipment which can produce fanshaped designs reminiscent of peacock feathers or sprays of flowers.[27]

If the bibliographer finds it desirable to move one step farther, he can add modifying adjectives to these names, as in the designation of cloth grains. The relative terms "fine" and "coarse" are more frequently applied to the combed patterns to indicate the nature of the comb, but they can also be used with the whisked patterns to suggest the relative distances between the veins. When it is necessary for the bibliographer to make distinctions on this basis among examples of the same basic pattern, he can take the illustrations presented here as "regular." The other adjectives, "moiré" and "drag," are more commonly used for the whisked patterns. In this connection, "drag" refers to the elongated spots produced when the paper is dragged along the surface of the size.

Whenever the addition of these adjectives does not provide a

precise enough reference, the bibliographer can go on to a fifth level (as with cloth) and note variations within patterns by the phrase "a variety of" (and "cf." before the index figure); he may also find it necessary to include specific photographs of the papers involved. The fact that only a small number of adjectives are listed in the outline should not prevent the bibliographer working on this level from employing additional ones when the occasion arises. Those listed represent the standard ways of modifying a pattern and are theoretically applicable to all patterns; but, particularly in the case of the combed patterns, the unlimited variations cannot always be accurately designated without recourse to other descriptive words or phrases (such as "crosswise nonpareil," meaning nonpareil produced by drawing the comb crosswise). When such additional terms are employed, they should be accompanied by a reference to an illustration of the particular pattern.

Specification of the colors in marbled patterns is best handled in two different ways, depending on the nature of the pattern. For whisked patterns the most prominent color (the "body color") should be given first, followed by the colors of the veins; for combed patterns it is generally sufficient simply to list the colors included:

In binding descriptions, "paper" should be taken to mean "papercovered boards," in the same way that "cloth" means "cloth-covered boards"; if the paper is used by itself, the term "wrappers" should be included or the paragraph should be headed "Wrappers." The following examples of openings of possible paragraphs on publishers' casings will illustrate the usage:

Decorated papers can thus be dealt with in the same fashion as publishers' cloth, by making reference to standard illustrations for a limited number of basic patterns and to special photographs for the numerous other patterns which may arise.

III. Letterpress Sheets

Patterns may also appear in the sheets of a book, on the same pages as letterpress. The most common location for such decoration is the title page, where there may be a border or a design separating the major elements of the page; but patterns occur frequently in other places, such as the beginnings and endings of chapters or principal divisions. They may be reproduced by means of wood-blocks, typographical ornaments, or various other processes.[28] The nature of these patterns, therefore, is not similar to that of cloth grains or marbled papers, with their relatively small number of standard designs; obviously, from the point of view of the difficulty of classification, patterns in letterpress sheets are analogous to those of cloth ornamentation or of printed, embossed, or paste papers. A publisher can use a different border on every title page if he chooses, and no limited selection of illustrations of these borders could do more than offer characteristic examples; it could not serve as a guide to the identification and description of any given border. Generally, then, verbal descriptions of patterns in letterpress sheets must be keyed to specific illustrations of individual patterns.

There are exceptions, however. Whereas binding cloths and decorated papers are of interest to the descriptive bibliographer chiefly in the period since the advent of publishers' bindings, when the output of books has been enormous, the patterns in letterpress sheets are the descriptive bibliographer's concern in all books since the beginning of printing. For the earlier periods, when the number of books was smaller, the supply of types and woodcuts more restricted, and the technology less advanced, the task of cataloguing all the patterns in particular categories, though not an easy one, is at least feasible. As a result, a few excellent reference works of this kind exist, particularly those issued by the Bibliographical Society in its series of "Illustrated Monographs" and "Facsimiles and Illustrations." R. B. McKerrow and F. S. Ferguson's Title-page Borders Used in England and Scotland,

1485-1640 (1932)[29] provides a model of a pattern-reference which bibliographers can hope may eventually be repeated in other areas. Within its self-imposed limits (borders designed specifically as borders, not those made up of typographical ornaments), it illustrates each border and records the books in which each appeared. When a bibliographer is describing a book that includes one of these borders, he can identify the border precisely by simply appending to his brief verbal description the appropriate McKerrow-Ferguson number. Besides serving as a standard of visual reference, the McKerrow-Ferguson work, in listing all the occurrences of a border, provides in effect a history of each pattern; this information about the course of a pattern through the hands of several printers may be important for the bibliographer's analysis of the dating and printing of a particular work. A. F. Johnson's German Renaissance Title-Borders (1929) and A Catalogue of Engraved and Etched English Title-Pages down to . . . 1691 (1934) are other examples of title-page reference works for the same period.

Woodcut illustrations are frequently not "patterns" in the sense employed here, but some patterns are woodcuts, and any reference book which comprehensively catalogues the woodcuts of a period should be consulted in connection with the identification of patterns; for early English books, the chief work is Edward Hodnett's English Woodcuts, 1480-1535 (1935). Publishers' devices constitute a similar category: not patterns themselves, they can form the basis of patterns and are repeated from one book to another. Verbal descriptions of devices in early English books should be keyed to R. B. McKerrow's Printers' & Publishers' Devices in England & Scotland, 1485-1640 (1913).[30] While it is possible to survey all the borders, woodcuts, and devices used in particular countries during the first two or three centuries of printing and produce catalogues for precise identification, it is clearly impossible to attain such comprehensiveness for later periods, when the output of the press reached staggering proportions.

A workable approach for later periods is to organize the study of patterns by publisher. The task of recording the borders and other patterns (and devices) used by a particular publisher, even a prolific one of the nineteenth or twentieth century, is manageable; indeed, it should be a regular part of any full-scale study of a publishing house. Publishing histories could then become the standard reference works

for the descriptive bibliographer in his specification of the patterns in letterpress sheets. Some histories of publishers already exist which serve this function in regard to devices. Geoffrey Keynes's William Pickering Publisher (1924) and Sidney Kramer's A History of Stone & Kimball and Herbert S. Stone & Company (1940) bring together at one place reproductions of all the devices used by these firms; each is numbered and can be referred to by number in the description of any title page (or spine) bearing it. A Coleridge bibliographer, when he comes to the 1844 Pickering Poems, can thus refer in his title-page transcription to "[dolphin and anchor device within oval scrollwork frame (Keynes ix)]." As more studies of major publishers become available, recording not only devices but all patterns used by these publishers, the task of specifying such patterns in bibliographies will become increasingly more simple and more exact. In the meantime, most descriptions of patterns occurring in letterpress sheets will have to be accompanied by specific illustrations.

Greater standardization in the specification of patterns in descriptive bibliographies is desirable in order to make verbal descriptions more precise. After all, any rendering in words of visual characteristics depends for its meaning on conventions, and the more detailed the conventions the more exact the verbal reference can be. Standardization should not, however, be carried to the point where it restricts rather than facilitates; and the infinite variety of possible patterns in books raises this problem in an acute form. In the case of certain patterns — cloth grains and marbled designs — a comprehensive standard and generic scheme of classification can be devised without becoming so involved or cumbersome as to defeat the purpose; and other patterns which by their nature must be illustrated specifically rather than generically can still be covered comprehensively for certain periods without becoming unmanageable. But in some cases the attempt to bring together all patterns falling within a given category (such as all stamped patterns on nineteenth-century publishers' casings) might not be worth the effort, if the result was so unwieldy that reference to it was more time-consuming and less meaningful than reference to a limited set of illustrations provided for the specific purposes of one bibliography.

Whenever a comprehensive standard set of photographs is possible, it is to be preferred to separate photographs in individual bibliographies, for two reasons: since terms become precise to the extent that everyone uses them to refer to the same things, widespread reference to a single set of photographs serves to encourage this precision;

in addition, reference to a standard more comprehensive than is possible on the basis of a single author-bibliography places the individual pattern in a larger context and provides a more meaningful form of description. Whether or not a comprehensive standard is possible — or presently available — in a given area, the general principle is the same: a bibliographical description must remain a verbal construction, but its descriptive phrases can have exact meanings only if they are supported by reference to pictorial representations. This double approach lays the groundwork for a generally understood terminology and therefore, however clumsily, bridges the gap between the verbal and the visual.

Notes