University of Virginia Library

A System of Color Identification for Bibliographical Description
by
G. Thomas Tanselle

The introduction of publishers' cloth in the 1820's was an unfortunate day for bibliographers, since the description of bindings has turned out to be perhaps the most troublesome aspect of the description of nineteenth- and twentieth-century books. Other parts of the description, which generally apply to earlier books as well, have become by now reasonably standardized. The bindings on earlier books, since they are products distinct from the process of publication, are not ordinarily the concern of the descriptive bibliographer; and specialists in the history of bookbinding have developed a vocabulary for dealing with them.[1] Even publishers' "bindings" (or casings), of course, are not bibliographical objects, strictly speaking, since they are not part of the letterpress. However, aside from the fact that binding variants can sometimes help to determine the priority of an issue,[2] the cloth is part of the dress in which an author's words are presented to the public, and its appearance therefore deserves to be recorded by the historian of such matters, the descriptive bibliographer.

To frame in words an adequate description of a cloth binding requires essentially some kind of notation of the texture, or "grain," of the cloth and some indication of its color. The first of these problems, though by no means solved, has received a great deal more attention than the second and is much nearer solution. Michael Sadleir, in his

pioneering Trollope: A Bibliography (1928), used such terms as "silk-grained" or "morocco-grained" to describe binding cloth, a gain in precision over "grained cloth" or simply "cloth," if one could visualize what the words implied; and two years later, in The Evolution of Publishers' Binding Styles, 1770-1900, he furnished photographs of four common cloth grains (facing p. 46). Then John Carter, the other pioneer historian of edition binding, made the next step forward in Binding Variants in English Publishing, 1820-1900 (1932), with his preliminary section on "Terminology of Grains and Designs" (pp. xvi-xviii). Here he supplied a plate illustrating twelve grains and provided a table of equivalences between descriptive terms like Sadleir's and the letter designations in use by Winterbottom's, the principal supplier of book cloth.

This table suggests the inevitable dichotomy in the verbal presentation of visual data: one may either use a precise, technical term, which often has little immediate meaning for the uninitiated reader, or else a more readily visualized term, which often is less exact and which breaks down when fine discriminations are needed. Carter declared his preference for the Sadleir terms — that is, "diaper" instead of "H" cloth, "sand-grain" instead of "C", and so on — but not all bibliographers have agreed with him. The two most important recent sources for photographic identifications of cloth grains represent these two approaches. Sadleir, at the end of the first volume of his great catalogue, XIX Century Fiction (1951), includes four plates showing twenty-four grains and giving them descriptive names; Jacob Blanck, at the front of each volume of the Bibliography of American Literature (1955- ), illustrates twenty-eight grains, assigning them the letter symbols used in the trade.[3] Either of these sets of photographs provides a basis for standardization of nomenclature, if followed scrupulously by bibliographers. Perhaps a chart should be issued making these standards more readily accessible, and perhaps bibliographers should, for precision, use both terms — such as "bold-ribbed (T) cloth."[4]

When one turns to the other basic ingredient of the description of cloth, the indication of color, one is surprised to find that practically no attention has been given to the matter. In the Bibliography of

American Literature, for example, the careful specification of cloth grains is in sharp contrast to this comment about color: "No attempt has been made to give other than a brief statement regarding color, and commonly accepted designations are used; variations in tone are recorded" (I, xxxiii).[5] One does not expect bibliographers who strive for precision in most respects to emphasize the casualness of their approach to color, as in these remarks:

One is not surprised, however, given the subjective nature of color descriptions produced in this casual way, to find that any two bibliographers in the past, treating the same book, have been likely to come up with two different designations of the cloth color. Thus T. J. Wise, in his Browning bibliography (1897), describes the wrappers of Pippa Passes (1841) as "yellow" (p. 7), though he explains that they are sometimes "pale cream" or "light brown"; the Broughton-Northup-Pearsall bibliography (1953), on the other hand, calls these wrappers "light apple-green" (p. 4). Similarly, J. W. Robertson (1934) sees the covers of Poe's Al Aaraaf, Tamerlane, and Minor Poems (1829) as "purple" (p. 38), while for Heartman-Canny (1943) they are "grey-blue" (p. 23). Sometimes the difference is a matter of emphasis, as when Duval (1939) designates the cloth of Aldous Huxley's On the Margin (1923) "blue-green" (p. 28) and Muir-van Thal (1927) finds it "greenish-blue" (p. 20); or when McDonald (1925) considers D. H. Lawrence's The Widowing of Mrs. Holroyd (1914) to be "red" (p. 32) and The Prussian Officer (1914) to be "dark blue" (p. 35), while Roberts (1963) finds them, conversely, "dark red" (p. 24) and "blue" (p. 25). Even the relative proportions are not constant, for Hogan (1936) labels Edwin Arlington Robinson's The Man Against the Sky (1916) and Merlin (1917) equally as "maroon" (pp. 11-12), while

Beebe and Bulkley (1931) say that the first is "dark maroon" and the second plain "maroon" (pp. 14-15). For Currie (1932), Booth Tarkington's In the Arena (1905) is "dark olive" (p. 53), but for Russo and Sullivan (1949) it is "sage-green" (p. 14); for Williams and Starrett (1948), Stephen Crane's The Red Badge of Courage (1895) is "tan" (p. 18), but for H. F. West (in the Dartmouth catalogue, 1948) it is "yellow" (p. 5); for Parker (1948), Joyce's Dubliners (1914) is "plum" (p. 22), but for Slocum and Cahoon (1953) it is "dark red" (p. 12); both Stewart (1959) and Livingston (1927) agree that Kipling's The Seven Seas (1896) is "red" (pp. 136, 160), but for Martindell (1923) it is "maroon" (p. 53); for Sadleir (volume 12 of the Constable edition, 1923), Melville's Typee (1846) is "fawn" (p. 341) and Mardi (1849) is "dark green" (p. 348), but for Minnigerode (1922) the first is "yellow" (p. 102) and the second "dark brown" (p. 135).

Such a list could be extended indefinitely, but the process would be pointless, since confusion of this kind is to be expected when color names are chosen on the basis of personal preference, without recourse to any set of standards. There have been only a few signs in recent years that bibliographers are beginning to be concerned about the problem. John Carter, in his ABC for Book-Collectors (1952), understated the case when he said, "There has never been much precision or uniformity in describing the colours of cloth"; but he went on to make a specific suggestion: "until we all agree to use the official Dictionary of Colour Standards, this imprecision will no doubt persist" (p. 55). The following year Patrick Cahill, in The English First Editions of Hilaire Belloc, adopted standard 381c of the British Standards Institution, Colours for Ready Mixed Paints (1948; 3rd impression, 1951), and thus described his bindings with such terms as "dovegrey" or "deep buff" or "pale-crimson." Then in 1956 Raymond Toole Stott took up Carter's recommendation in his bibliography of Somerset Maugham and used one of the British Colour Council's publications — the Dictionary of Colours for Interior Decoration (1949). Though it may seem strange to read of "hay" or "biscuit" endpapers, the experiment was, as Stott recognized, "at least a step on the way to the systematized description of colours of binding cloth" (p. 8). And it was undoubtedly more efficient and precise than the method used by Frederick T. Bason in his earlier (1931) Maugham bibliography: the binding of Of Human Bondage (1915), labeled "petrol blue" by Stott

(p. 35), had been described by Bason as "dark sea-green cloth (a more distinct green than The Casuarina Tree)" (p. 33).[7]

These worthy efforts, however, did not convert other bibliographers from their subjective ways. Anthony Rota, reviewing Keynes' bibliography of Siegfried Sassoon (1962), remarked that "Sir Geoffrey's treatment of colours again points the need for the adoption of a common standard for defining binding shades"; and he quoted Keynes' description of The Old Hunstman (1917) as "drab or grey-blue" (with its ambiguous or, since the two colors do not seem synonymous) and of The Daffodil Murderer (1913) as "orange" (when to Rota's eyes it is yellow).[8] That this kind of confusion has not prodded bibliographers long before now to attack the problem of color designation is remarkable. On the other hand, this neglect can perhaps partly be explained by the fact that significant nineteenth-century binding variants generally do not depend on color differences alone; books were frequently issued in several colors simultaneously, and later bindings often involved a different cloth texture as well as color.[9] But in the twentieth century color variants may be more meaningful, since the simultaneous issue of multiple colors has not been customary. In any event, there should be a precise method for describing the color of a given binding whether or not the priority of an issue depends on its specification. No bibliographer would estimate the dimensions of a leaf without using a ruler; in the same way no bibliographer should make his own subjective estimate of a color without turning to a color chart, which ought to be an equally essential part of his equipment. The point is self-evident; there should be general agreement, in the words of the reviewer of Cahill's bibliography, that "it would be a great relief to all concerned if some standard scale could be adopted."[10]

The question then becomes the determination of the particular system best suited to the requirements of bibliographical description. And this decision is not to be lightly made; for any kind of standard, to serve its purpose, must be capable of wide acceptance and future

applicability. At the outset, it should be possible to agree that any system selected for bibliography must meet certain minimum conditions: (1) it must contain color chips or samples which can be compared easily with book covers; (2) it must include a sufficient number of different colors to be compatible with the degree of precision required in making the kinds of distinctions between bindings that are likely to be significant under varying circumstances; (3) it must assign to each color a name (not simply a number or symbol), avoiding names so fanciful as to carry no immediate connotation; (4) it must be compact in physical form and easily portable, so that the bibliographer can conveniently carry the standards with him to the libraries in which he is working; (5) it must be inexpensive enough that it can become a standard tool in every bibliographer's possession (for it is too much to expect, even for an accepted standard, that each collection in which one works will have a copy at hand); (6) it must provide strong assurance of continued availability in the future. The number of color systems which have devised since the time of Isaac Newton is vast,[11] and it is necessary to know something about the currently available ones in order to make an intelligent choice.

I

The fundamental scientific method for measuring or specifying a particular color was established in 1931 by the Commission Internationale de l'Eclairage (CIE).[12] In this system, the proportions of red, green, and blue light required to match a given color are measured with a colorimeter, and the chromaticity coordinates of the color are thus established. The Commission also defined the characteristics of the standard observer and the properties of three standard illuminants. The usual notation of a color consists of two of the chromaticity coordinates plus the luminance value as established by spectrophotometer; these tristimulus values represent dominant wave length, purity, and reflectance. For example, the color of a tomato might be expressed as follows: x = .622; y = .350; Y = 10.2%.[13] However basic this

system is for scientific measurement, its disadvantages for bibliographical description are obvious. To demand that bibliographers acquire the necessary knowledge of optics would decimate their already small ranks; and the prospect of setting up shop, with delicate and expensive instruments, in each rare book room would eliminate the few that remained — if, indeed, the librarians had not already resigned. Not only is this method impractical for bibliographical purposes; it is undesirable. For the degree of accuracy attainable with a spectrophotometer is not required in a bibliographical description (and is thus wasteful of effort); in addition, it could often be actually misleading, since it would continually reveal variants, most of which (depending on the tolerances established by the manufacturer of the cloth or the dye) would have no significance to the bibliographer. To put the matter another way, if the bibliographer makes finer distinctions than those required by the adopted tolerances of the manufacturer, he may find himself recording as variants bindings which came from the same bolt of material or consecutive ones.[14] Furthermore, the notation in tristimulus terms is not one which could be readily comprehended by the readers of a descriptive bibliography.

If the CIE system is not feasible for bibliographical work, the next question to ask is whether or not another more appropriate system exists which is at the same time scientifically accurate and respectable. The alternative to spectrophotometric measurement is visual comparision with material standards, such as a set of color chips (sometimes known as the "ratio method"). Depending on the selection and production of the colors represented in any given set of material standards,

this approach to color specification can be very precise and is scientifically valid. Although individual perception of a color is subjective, the act of matching a color with a carefully produced standard (which can then be referred to by other persons) reduces to a minimum the effects of subjectivity. And if the color chips have been chosen to represent particular CIE specifications that will result in a systematic sampling of color space, the whole process rests on a sound scientific foundation.[15] Aside from the danger of fading, an unavoidable problem in any material standard, the principal limitation of such systems is the necessity of interpolation. Since the eye can distinguish about ten million colors and since there are theoretically an infinite number of colors in three-dimensional color space, any color atlas or set of standards must represent a selection based on some principle of organization (hence the generic term "color-order systems"); therefore, depending on the number of colors selected and the accuracy of specification required, it is sometimes necessary to estimate the relation of a given color to two of the colors in the standard. This sort of interpolation, though it can be performed skillfully with practice, remains of course a subjective matter. A material color standard for bibliography should contain enough colors that, within the limits of accuracy desired, interpolation would seldom be required; on the other hand, it should not include so many colors that the process of matching becomes time-consuming and indecisive.

Color-order systems fall into three groups: (a) color-mixture (or additive) systems contain copies of colors established by mixing colored lights in particular proportions with a tristimulus colorimeter; (b) colorant-mixture (or subtractive) systems contain colors produced by mixing colorants (pigments, dyes) in various proportions; (c) color-appearance systems contain colors arrived at by means of psychological

perceptions of color differences and spacing. Some systems, such as atlases printed by a screen-plate process, are intermediate, partaking of the characteristics of both the color- and colorant-mixture types.[16] The methods by which the colors are produced, however, will be of less concern to the bibliographer than the physical features and range of the resulting standards; and the three most promising systems are surveyed here with the needs of the bibliographer in mind:

Of the other publications of the British Colour Council, two should be mentioned. In 1938 the Council issued (in collaboration with the Royal Horticultural Society) the first volume of a Horticultural Colour Chart, also called the Wilson Colour Chart after Robert F. Wilson, the active and prominent general manager of the Council at that time. It consists of a portfolio containing 100 loose plates of printed color samples (not affixed chips). Each leaf lists foreign synonyms of the color name and the equivalents in four other systems (BCC Dictionary, Ridgway, Oberthür-Dauthenay, Ostwald); it also indicates a special notation for the color, in which the last two digits stand for one of 64 hues and the preceding digit represents lighter tints (600's and below) and darker shades (700's and above) — thus "Rose Bengal" is 25 and "Phlox Pink" is 625 — with prefixed zeros for steps of graying. In 1941 a second volume of 100 more plates was published. These two volumes, with their somewhat awkward notation and nomenclature and their rather inconvenient physical form, would not be successful for bibliographical description. The other publication is the one used by Stott in his Maugham bibliography, the Dictionary of Colours for Interior Decoration (1949), with 378 colors (labeled CC1-CC378) displayed in silk samples in three volumes. The Council's regular Dictionary of Colour Standards contains cross references to this

chart, but the names do not always match — BCC47, "Victrix Blue," for example, is the same as CC287, "Ming Blue." This situation does not seem conducive to standardization of terminology, particularly when the terms are not self-explanatory. Stott, after using this Dictionary, reported that "many book cloths defy even the 378 examples shown on the chart" (p. 8); and a reviewer of his bibliography questioned the BCC nomenclature by asking whether the gain in accuracy was "worth achieving at the cost of superficial confusion in the minds of those who are more familiar with The Library than with Vogue."[22] Another disadvantage of all three publications was expressed by Cahill when asked why he adopted a paint standard instead of the British Colour Council's Dictionary: "its bulk and price make it an impossible tool for the Belloc collectors, enthusiasts and booksellers for whom my book is mainly intended."[23]

The bibliographer should be aware of the features of certain other systems, though none of them would serve as a practical choice for a bibliographical standard: (1) the Standard Color Card of America (9th ed., 1941), with 216 silk swatches, used mainly by the textile industry, employs rather bizarre nomenclature and is an unsystematic sampling of color space; (2) the DIN-Farbenkarte (1953), with samples representing equal psychological steps, is the official German standard, not very widely known in England and America; (3) the great Villalobos Colour Atlas (1947), probably the most extensive guide available, shows 7279 samples, each with a hole in the center to facilitate matching; (4) the Dictionary of Color (1930; 2nd ed., 1950) by Aloys J. Maerz and M. Rea Paul is the standard work on color nomenclature, with 7056 colors on 56 plates, and it is no criticism of the work's great authority to say that the small size of the color squares (usually 144 to the page, with no holes for comparison) and the historical purpose of the work (with some colors assigned no name at all) make it inappropriate as a standard for bibliography; (5) the Plochere Color System (1948) by Gladys and Gustave Plochere, with 1248 colors on 3" x 5" cards (or smaller mounted rectangles), is basically a guide for interior decorators; (6) Federal Standard No. 595 (1956), with 358 color chips, is not a systematic sampling and is mainly intended for the specification of paint colors in use by the United States government; (7) Robert Ridgway's Color Standards and Color Nomenclature (1912), with 1115 colors and names, was long a standard for naturalists but is now out of print; (8) the Nu-Hue Color Coordinator (1949, 1952), prepared by Carl E. Foss for the Martin-Senour Company, is

perhaps the most ambitious and elaborately produced of the colorant-mixture systems developed by paint manufacturers; (9) the 1962 Reinhold Color Atlas (in England the Methuen Handbook of Colour), by A. Kornerup and J.H.Wanscher, displays 1266 colors in extremely (perhaps overly) compact form. These systems are of course only a few out of a possible list of seventy-five or more; but the bibliographer cannot seriously consider the great majority of color charts, devised specifically for stamp collectors, horticulturists, interior decorators, paint dealers, textile dyers, or ornithologists.

Clearly the various attempts at color standardization have not in the past led to any general system which encompasses or coordinates a large number of them, and the bibliographer is faced with a multiplicity of systems, none of which precisely suits his needs. The choice of one with the fewest disadvantages becomes a matter of deciding which of the desired features are most important — whether it is better to have a standard with a large number of colors, or a satisfactory nomenclature, or a low price, or something else. There is no doubt that the problem of nomenclature is extremely important for bibliographical description, because the reader of a bibliography should not be required to consult a color chart except when a question arises. He should not be confronted with "8pa" or "13432," unaccompanied by a commonly understood color expression; yet the common expression must be firmly attached to a precise area in color space so that it will hold the same meaning for each user. Fortunately, such a system is now available.

II

In 1931 the Inter-Society Color Council was formed as an organization of national societies whose work involved color; it was not only to be a clearinghouse for color problems and research but more specifically was to assist in revising the color names used in the U. S. Pharmacopoeia.[24] E. N. Gathercoal, the ISCC's first chairman, was a member of the Pharmacopoeial Revision Committee, and he had arranged a symposium on color names at the 1930 Pharmacopoeial Convention in Washington. His goal was a color nomenclature "sufficiently standardized as to be acceptable to science, sufficiently broad to be appreciated

and usable by science, art, and industry, and sufficiently commonplace to be understood, at least in a general way, by the whole public." By 1933 I. H. Godlove had sketched out the foundation for such a system; by 1936 Kenneth L. Kelly of the National Bureau of Standards was working on the task of assigning boundaries to the suggested color designations; and in 1939 Deane B. Judd and Kelly published Research Paper 1239, "Method of Designating Colors," in which common color names were defined in terms of specific areas of the Munsell color solid.[25] This ISCC-NBS method, as it was called, was an enormous step forward, but by 1949 a committee of the ISCC had revised the color boundaries in response to certain criticisms which some users had made. Finally, in 1955, Kelly and Judd published the revised version of their 1939 work, which now also contained a dictionary relating the ISCC-NBS names to those in other systems. This remarkable book, The ISCC-NBS Method of Designating Colors and a Dictionary of Color Names, is NBS Circular 553, available from the Superintendent of Documents, U. S. Government Printing Office, at $2, and should certainly be a part of every bibliographer's equipment.

In the history of color nomenclature, the work is epoch-making in at least two ways. First, instead of assigning names to particular color chips, it proceeds in the other direction, setting up a system of easily understood names and then mapping out the entire color solid for the first time into segments which define the precise limits of each name. The system of names — which takes into account the three attributes of color — is a simple one based on ten hue names, three neutrals, their modifying forms, and ten other modifiers, as follows:

Hues

• B blue
• Br brown
• G green
• O orange
• Ol olive
• P purple
• Pk pink
• R red
• V violet
• Y yellow
• b bluish
• br brownish
• g greenish
• p purplish
• pk pinkish
• r reddish
• y yellowish

Neutrals

• Black black
• Gy gray
• White white
• blackish blackish
• gy. grayish

Value Modifiers (Lightness)

• d. dark
• l. light
• med. medium
• v. very

Chroma Modifiers (Saturation)

• gy. grayish
• m. moderate
• s. strong
• v. vivid

Value and Chroma Combinations

• brill. brilliant [light, strong]
• deep deep [dark, strong]
• p. pale [light, grayish]

The abbreviations provide a concise notation (as "v.d.pR" for "very dark purplish red") but of course are not essential. The point is that there are no "post office reds," nor "puces," nor even "garnets" here, but only combinations which — however complicated, as "dark grayish olive green" — at least give a suggestion of the particular color to any reader. The color solid is then divided into 267 named segments, and the boundaries of each are plotted on 31 charts representing ranges of Munsell hue. Thus, for example, both Munsell renotation 4R 3/6 and 3R 2.5/5 fall within the area designated as "dark red." Since colorimetric measurements have been made of the colors of Munsell renotation,[26] even a color identified in CIE terms could be converted to an ISCC-NBS name.

The second important feature of the work — the second part, added in 1955 — is the dictionary, which makes it equally easy to convert to these names from a number of systems other than Munsell. This dictionary is a compilation of the names used in fourteen previous charts or atlases: Maerz-Paul, Plochere, Ridgway, Federal Specification TT-C-595, Wilson's Horticultural Colour Chart, the Color Card of America, Taylor-Knoche-Granville (supplement to the Color Harmony Manual), the American Association of Textile Chemists and Colorists and the Society of Dyers and Colourists' standard, Commercial Standards CS147-47 (Urea Plastics) and CS156-49 (Polystyrene Plastics), the National Research Council's Rock-Color Chart, the Department of Agriculture's Soil Color Chart, H. A. Dade's Color Terminology in Biology, and W. H. Beck's Postage-Stamp Color Names. First comes a list (pp. 35-82) of the 267 ISCC-NBS names with the synonyms from these systems listed under each name. Then follows the dictionary proper (pp. 83-158), which lists alphabetically all the names from these fourteen systems (about 7500 names), giving for each the source and the ISCC-NBS designation (and serial number). Previous dictionaries drawing names together from various sources have not subordinated them to a new terminology nor attempted to provide names to cover the entire color solid systematically; the ISCC-NBS method is thus a kind of master-system, furnishing a common ground to coordinate earlier systems. If one person, using Maerz-Paul, describes a color as "Rhodonite Pink" and another, using Plochere, refers to "Orchid Mauve," both can consult the ISCC-NBS dictionary and discover that they are dealing with the same color, "dark purplish pink" (no. 251).

The advantages of this system for bibliographers (or for anyone else) are immediately apparent. It provides a consistent, standard, easily understood nomenclature for color, regardless of the particular set of color chips employed for matching. If one bibliographer prefers for some reason to use the Plochere system but hesitates to describe the color of a certain binding cloth as "Best Effort" (or as 1224 YYg 3-h), he can look that name up in the ISCC-NBS dictionary and find that it corresponds to ISCC-NBS 121, "pale yellow green." Then if another bibliographer, who wishes to check this color himself, has only a copy of Ridgway at hand, he may make the comparison, find that Ridgway's "Glaucous" is the one that matches, and check the dictionary to see that "Glaucous" is also within the area defined by "pale yellow green." The same would be true of a third bibliographer or collector who had identified the binding as "Rhone" (Maerz-Paul 18 B 3). In this way uniformity results in the final bibliographical description, despite the multiplicity of systems which may have been used by individual bibliographers. Even if a bibliographer buys for himself a copy of the Munsell Book of Color, for example, he may not always have it with him; if he finds himself in a library which has only a copy of Maerz-Paul, he may proceed with his description of the binding and later convert the Maerz-Paul term by using the ISCC-NBS dictionary. For accuracy and for general scholarly indication of sources, the ISCC-NBS name should be accompanied by a reference in parentheses to the actual color sample used — as "dark red (Maerz-Paul 6 L 11)," or "dark red (Plochere 353 R 3-a)," or "dark red (TCCA 65020)," or "dark red (Munsell 2.5R 3/7)." The Nickerson Color Fan is particularly convenient (aside from its price, size, and arrangement) because it designates on each sample both the Munsell notation and the ISCC-NBS name (thus obviating any reference to the dictionary itself after the process of matching). In any case, two points are essential: that bibliographers agree to compare binding colors with some collection of color samples and that they convert the identifications into ISCC-NBS names.[27]

The only real limitation of the ISCC-NBS system as published in 1955 was that it contained no actual color chips to illustrate the names.

However, by 1958 Kenneth L. Kelly had worked out the Munsell renotations for the center of gravity of each of the 267 ISCC-NBS color-name blocks,[28] and handmade charts illustrating these centroid colors were produced as an aid to ISCC-NBS Subcommittee on Problem 23, the Historical Expression of Color Usage, as explained in its Interim Report of 20 November 1960. The NBS then began preparing such charts for general distribution as a supplement to Circular 553, and they became available in February 1965 from the Standard Reference Materials Office at the National Bureau of Standards, as Standard Sample No. 2106, for $3 per set.[29] The ISCC-NBS Centroid Color Charts complete the system, and for a total of $5 the bibliographer can now equip himself with both the dictionary and the charts, which together comprise the most efficient method of color naming yet devised and the one most likely to become a general standard accepted by all fields. The charts alone are all that the bibliographer need carry with him, and they are convenient in size. There are twenty leaves, 10½" x 8", of which the first two contain a table giving the Munsell renotation for each centroid sample and the other eighteen are hue charts illustrating 251 of the 267 ISCC-NBS names in glossy chips 1" square. The chips are arranged on each sheet as they would appear on a Munsell hue plane (that is, with the grays at the left, the highly saturated colors at the right, the lighter colors at the top, and the darker ones at the bottom), against a neutral background of about their own value level; beneath each chip is the identifying number of that color-name block and the abbreviation of its color name. Using these charts the bibliographer will not have to make any conversions from one color name to another; he can simply find the chip which most nearly matches his binding and record it as "deep bluish green (Centroid 161)" or "vivid orange (Centroid 48)" or "dark grayish yellowish brown (Centroid 81)."

The first question which bibliographers are likely to raise is whether a system with 267 colors (and 251 chips) can be sufficiently accurate, particularly in view of Stott's comment, in his Maugham

bibliography, that the 378 samples in the BCC's Dictionary of Colours for Interior Decoration were not sufficient. The question may be answered in two ways. First, one must consider the principle of selection of the 267 colors. Though not a great number, they represent blocks which, taken together, comprise the entire color solid; the chips, in turn, represent the center of 251 of these blocks. In contrast to other systems, then, each chip illustrates a color characteristic of a well-defined area. When one decides that a particular binding color does not match precisely any of the centroid chips but comes closest to matching Centroid 16, one can be sure that the block of the solid labeled "dark red" contains the particular color, and "dark red" is thus an accurate term, even though the match with the color chip was not exact. To this degree, the system is capable of complete accuracy, whereas in other systems the fact that one chip is closer than another to the color in question has little significance for naming, since the boundaries of the color-name have not been defined.

Second, since the ISCC-NBS dictionary encompasses fourteen other systems (plus the Munsell name charts), it is always possible, when finer discriminations are required, to utilize a system with more chips and yet remain within the framework of the ISCC-NBS method. Indeed, Kelly has explained six levels of accuracy in color description attainable within this method.[30] In the first the color solid is divided into only 13 sections, corresponding to the ten hue names and the three neutrals; for some purposes it is enough to distinguish "pink" from "red," or "yellow" from "orange." The second level works with 29 name-blocks, consisting of all the hue terms — such as "olive brown," "greenish yellow," "yellowish brown," or "olive green." The 267 names produced by adding the modifiers descriptive of value and chroma, as in the ISCC-NBS dictionary and centroid charts, constitute the third level of refinement, the one on which bibliographers may normally find themselves. But many distinguishable colors naturally fall within each of these name-blocks, and it may be that certain binding variants are not distinguishable in terms of the centroid colors alone. One may then move to the fourth level, which involves consulting an appropriate color-order system — Munsell if possible, but, if not, perhaps Plochere, with 1248 colors, or Maerz-Paul, with 7056.[31] If an exact

match is not found, the next step (the fifth level) is to make a visual interpolation from the Munsell Book of Color. Because the Munsell system is based on psychological spacing (a color-appearance system) and because the Munsell notation is conveniently manipulated to reflect changes in any of the three attributes, a person with some experience can accurately indicate about 100,000 colors by such interpolation. The sixth and highest level of accuracy is of course the CIE method of spectrophotometric measurement, capable of dividing the color solid into about five million parts. Bibliographers will need to leave the third level only rarely, but the accuracy required at certain times[32] may necessitate moving to the fourth or fifth levels. In any event, the ISCC-NBS system adequately takes into account the fact that varying degrees of accuracy are desirable under varying circumstances.

Another problem arises from the fact that the centroid color chips are glossy and book cloth is not. Since the ideal system for bibliography is undoubtedly one with cloth samples, there is no answer to this objection, except to say that the system offers so many other advantages that it is still the best one to choose. Besides, the importance of the surface texture of the sample in any given case is a function of the accuracy required. In some instances, then, the bibliographer may wish to turn to the textile Color Card — so long as he realizes that its sampling of the color solid is not systematic — or to the British Colour Council's Dictionary. The difference in appearance between a glossy chip and a cloth swatch of the same color is also to some degree a function of the viewing conditions. It is normally recommended, in most systems of color identification, that the matching be done in natural light, preferably northern light and certainly not direct sunlight; the light should strike the surfaces to be matched at an angle of 45°, and the surfaces should be viewed from 90°. Some sets of chips include masks which can be used to block out the colors on the chart surrounding the chip being matched. Although the ISCC-NBS charts do not contain masks, it is a good idea to prepare a few of them by making a hole 1" square in a stiff piece of gray paper or cardboard.

Still another of the bibliographer's questions will concern notation — what form the color information is to take in a bibliographical description. The ISCC-NBS abbreviations — as "d.gy.G" — should

probably not be used, for they give essentially common terms an esoteric appearance; "dark grayish green" does not take up too much space. The centroid number can conveniently be inserted in parentheses following the name, just as the letter identifying a cloth grain can be placed after the verbal description of it. Thus what formerly might have been referred to as "dark green ribbed cloth" may now appear as follows: The length of the color expression, including its parenthesis, may be clearer and less awkward if it follows, rather than precedes, the designation of grain. A further convention may perhaps be agreed upon: the presence of a centroid number suggests only that the color falls within that color-name block and does not necessarily imply an exact match with the centroid color; however, if another set of chips is referred to as the standard for the identification, the match may be inferred as exact, unless the abbreviation "cf." precedes the notation. To illustrate: Obviously the last of these represents an extension of the fourth level of accuracy; it stops short of the fifth level because an interpolation has not been suggested. When an interpolation is made, the notation should be enclosed in brackets: In this system the color of Waldo Frank's Time Exposures (Boni & Liveright, 1926) is "brilliant yellow (Centroid 83)," of Vachel Lindsay's Adventures While Preaching the Gospel of Beauty (Kennerley, 1914) "very deep red (Centroid 14)," of Sherwood Anderson's Horses and Men (Huebsch, 1923) "strong reddish orange (Centroid 35)"; and a complete description — of Eunice Tietjens' Jake (Boni & Liveright, 1921) — would go something like this: In this instance the slight difference in color between the cloth and the endpapers could not be indicated by reference to Centroid 12, so approximations to two colors in the Nickerson Color Fan were used.

However, it should be emphasized here that, despite an occasional example of this kind, the bibliographer generally need feel no urge to go beyond the centroid charts. Given the nature of book cloth and of material standards, the majority of matches are going to be only approximate; and in any case the purposes of most bibliographical descriptions would not be further served by a more precise match. For these reasons the centroid colors, which are simply the representative colors of particular color-blocks, constitute an ideal frame of reference. If a bibliographer goes outside this system, one must assume either (1) that another system was the only one at hand when the comparison was made (with the result later converted to the corresponding centroid name) or (2) that greater accuracy was required — and obtained — by specification in terms of another system. It would be a mistake to overuse the "cf." designation, for, if the match is only approximate anyway, nothing is gained in precision over the simple reference to the centroid color.

Because of the advantages of the Munsell notation, it may be that some bibliographers, if they have had to turn to another set of chips for an exact match, will want to indicate the equivalent Munsell notation or renotation as a convenience to their readers. The Munsell equivalents of the most commonly used systems have been tabulated, and it may prove helpful to bring together the principal references to these conversion tables:

• Color Harmony Manual Walter C. Granville, "Munsell Renotations of Color Harmony Manual Chips (Third Edition) from Spectrophotometric Measurements," available from Container Corporation of America, Color Standards Department, 38 South Dearborn Street, Chicago 60603.[33]
• DIN-Farbenkarte W. Budde, H. E. Kundt, and Günter Wyszecki, "Überführung der Farbmasszahlen nach dem Farbsystem DIN 6164 in Munsell-Masszahlen und umgekehrt," Farbe, IV (1955), 83-88.
• Horticultural Colour Chart Dorothy Nickerson, "Horticultural Colour Chart Names with Munsell Key," JOSA, XLVII (1957), 619-621.

---

226




• Maerz-Paul Dorothy Nickerson, "Interrelation of Color Specifications," Paper Trade Journal, CXXV (1947), TS219-237.
• Plochere W. E. Knowles Middleton, "The Plochere Color System: A Descriptive Analysis," Canadian Journal of Research, XXVII (1949), F1-21.
• Ridgway D. H. Hamly, "Ridgway Color Standards with a Munsell Notation Key," JOSA, XXXIX (1949) 592-599.
• Standard Color Card Genevieve Reimann, Deane B. Judd, and Harry J. Keegan, "Spectrophotometric and Colorimetric Determination of the Colors of the TCCA Standard Color Cards," JOSA, XXXVI (1946), 128-159; or Journal of Research of the National Bureau of Standards, XXXVI (1946), 209-247.

Whether the original match was exact or approximate, an equals sign may be used to indicate the Munsell equivalent; but if one wishes to give only the centroid number, the symbol for "approximately equals" (≈) should be used: If Munsell renotation is employed, it should be so specified: The Munsell renotations of the centroid colors are provided in the table accompanying the centroid charts; if a binding happens by chance to match exactly a centroid chip, the coincidence may be indicated by including the renotation figure: These equivalences are of course simply additional information which may be furnished for the convenience of the reader in making his own further comparisons; they are by no means required. But knowledge of the existence of these conversion tables may be useful to the reader of a bibliography which does not provide the equivalences.

Another question — and one of the most troublesome — is the problem of fading, both of the color samples and of the bindings. As far as the samples are concerned, one should not expose them to light except when they are being used; and after extended use one should perhaps compare them with a new copy to see whether they have yet

faded. A consideration of faded bindings is tied up with the whole matter of the degree of accuracy required, for in some cases it may be necessary to determine whether a given binding is variant or merely faded. But in most instances the determination of a secondary binding will not turn solely on color discrimination, and, since a bibliographical description records the characteristics of an "ideal copy,"[34] the general rule is to include in the binding description a notation of the brightest copy examined. Then, if one wishes to provide the colormatching data for the other copies, the paragraph listing "Copies Examined" is the proper place for the information. In the case of certain scarce books, all copies may appear faded, even on the covers (as well as the spine, where fading is to be expected). If there is no small patch or strip of brighter color (as on the inside edge of a cover) to serve as a basis for the color identification, one may have to describe the color as it appears and append a note explaining one's hunch that the color is probably faded. A descriptive catalogue of the books in a particular collection, however, is obligated to describe a binding in whatever faded state it is found in that collection; but an energetic cataloguer will go further (by examining other copies or consulting a published bibliography) and indicate the extent of the fading, either through an exact match or through interpolation. Dorothy Nickerson has worked out a formula for an Index of Fading, whereby the amount of fading can be indicated in a single figure. Though the formula is not a complicated one, the single-figure index is more meaningful for expressing tolerances in the textile industry than for describing the fading of bindings, since the single figure (consolidating the differences in hue, value, and chroma) does not enable one to visualize the changes in the three attributes.

A final consideration has to do with the fact that colors in books are not limited to bindings. There are colored sheets, inks, dust jackets, and endpapers; and the ISCC-NBS names are appropriate for describing them all. In fact, Deane Judd has specifically commented on the applicability of these names for the paper industry and has shown some of the equivalents between the ISCC-NBS names and those in the Grading Committee of the Groundwood Paper Manufacturers' Association's

"Blue Book," Standard Color Nomenclature System and Manual (1936).[35] The methods developed for measuring the whiteness of paper, however useful for specification and standardization in the paper industry, go beyond the degree of accuracy required in the bibliographical description of white papers. Colors of printing inks may also be specified in ISCC-NBS terms; in 1935 the International Printing Ink Corporation issued A Series of Monographs on Color, which included a description of the advantages of the Munsell system. But the many atlases and sets of color chips circulated by both paper and ink companies are not convenient as general standards. Not a great deal of attention has been given to the description of dust jackets,[36] but the main problem is simply a question of the completeness of the description rather than of method; and the ISCC-NBS names should make such descriptions more accurate. Another kind of paper, however, causes greater trouble — the marbled papers sometimes used as endpapers or binding papers. The difficulty in describing them is analogous to that in specifying cloth grains and is not essentially a color problem. It would be very helpful, therefore, to have a chart illustrating such marbling patterns as "gold vein" or "nonpareil" — the same sort of chart (but on a more elaborate scale) that Bernard C. Middleton furnishes as the frontispiece to his A History of English Craft Bookbinding Technique (1963), where twelve common kinds of decorated papers are displayed.[37]

In 1953 Fredson Bowers remarked, "As a matter of fact, when the technicians really get to work on the problems of machine-printing, I rather suspect that the general reader and the bibliographer who has catered to him are due to suffer a shock."[38] Whether or not the method of color description outlined here will offer a shock to those bibliographers who fondly remember the good old days when it was possible to speak of "puce" or "Eureka" or "Victoria Lake," depending on one's mood, the fact remains that a move in this direction is inevitable. The ISCC-NBS system can be as simple or as complex as is required under different circumstances, and its use is no more difficult, and only slightly more time-consuming, than the measurement of leaves

with a ruler. The color names should shock no one, and the whole method seems ideally suited to descriptive bibliography — with the exception of the glossy chips. Perhaps a handbook can some day be produced which will contain illustrations and names of binding grains and decorative papers along with the centroid colors in cloth samples. In the meantime it is not asking too much that bibliographers compare each binding cloth with the centroid chips or some other collection of samples and express the color name in ISCC-NBS terms. Maerz and Paul, at the beginning of their Dictionary of Color, observe that, "while standardization has been arrived at in practically all other fields, in the use of color names for identifying color sensations a condition prevails that is usually characterized as chaotic." Bibliographers can ill afford to perpetuate chaos in any of their endeavors, if their general concern with order and accuracy is to be meaningful.[39]

Notes