University of Virginia Library

An Application of Semiotics to the Definition of Bibliography
by
Ross Atkinson

I

Bibliography, as Greg remarked in 1914, "suffers from its name."[1] The problem is not, however, that the word can no longer be precisely defined on the basis of its etymologically derived parts. Some of its current uses, as will be noted later, do in fact correspond quite closely to the combination of βιβλιον and γραøω, but even if this were not so, verbal meaning does after all vary with time and place, and it is neither unusual nor any cause for concern if our usage of the word does not coincide with that of Dr. Johnson or Lucian. The problem today in English lies rather in the multiplicity of meanings assigned to the name "bibliography," as well as in the relationship between them, for bibliography "is not 'a subject' but a related group of subjects that happen to be commonly referred to by the same term."[2]

It is through the application of a variety of descriptive adjectives to the term "bibliography" that its various "subjects" are usually differentiated. These adjectives, like so much other fundamental terminology in the field, have gained acceptance as a result of the persistent efforts of Professor Bowers to provide bibliography with a theoretical basis. Robert Heilbroner, in a recent article on Karl Marx, has maintained that the reason "why Marx asserts and reasserts his intellectual thrall" in the social sciences is "not because he is infallible," but rather "because he is unavoidable."[3] The same may be said of the "intellectual thrall" Anglo-American bibliography finds in the work of Fredson Bowers, and while we obviously cannot claim, as Heilbroner does for Marx,

that Professor Bowers created "a whole new mode of inquiry," we can nevertheless say that Bowers has essentially created the basis upon which the whole "mode of inquiry" known as bibliography views itself and defines its various activities. Any discussion of the general nature of bibliography must, therefore, be founded upon the fundamental pronouncements of Professor Bowers, according to whom there are four—or sometimes five—distinct but related branches of bibliography: enumerative, analytical, descriptive, textual, and—sometimes—historical.[4]

In the area of descriptive bibliography there was a call not long ago to replace the work of the "splitters" with that of the "groupers," in other words to cease expending so much energy on the description of minute details of individual documents and to begin to concentrate on the assemblage of broader bibliographic data of more demonstrable value to critics and collectors.[5] A similar tendency to replace splitting with grouping is already apparent in some current efforts to categorize and define the various branches of bibliography. Paul Dunkin, for example, has recently seen fit to define all bibliography as simply that field of study designed "to locate books."[6] Most grouping efforts, however, have attempted to split bibliography into two broad categories. Probably the clearest and best received of these has been Lloyd Hibberd's suggestion that the subject be divided into "physical bibliography" and "reference bibliography,"[7] about which distinction more will be said

later. Another of the most recent attempts to accomplish a general grouping may be found in the current Encyclopaedia Britannica article on "Bibliography" by Sir Frank Francis, who conservatively divides the entire field into "descriptive bibliography" and "critical bibliography."[8] Aside from Francis' surprising choice of the old and ambiguous term "critical bibliography"[9] to describe "the study of books as material objects," the most remarkable part of his article is his apparent combination of descriptive bibliography in the sense of Bowers with enumerative bibliography.[10]

The suggestion that descriptive bibliography replace its "splitters" with "groupers" was understood by its originator to be "a call back to the past" (Liebert, p. 10), and it may appear that recent grouping efforts in the area of general definition are similar attempts to return to a pre-Bowers state of innocence. This would be, however, an incorrect assumption, for Professor Bowers, unlike Greg, has in fact never denied that bibliography is divisable into two broad categories. On the contrary, he actually adopted this perspective in his own Encyclopaedia Britannica article:

The next question we must ask, then, is whether such a bifocal view of the word "bibliography" is justified, especially since the second meaning—the study of books as material objects—appears to be limited to

English usage. According to Rudolf Blum, who has made a thorough study of the history of the use of the term in French and German speaking countries, many influential continental bibliographers included at one time the physical history of printed matter as part of the subject of bibliography. One of the first of these was the Viennese scholar Michael Denis, whose widely used Einleitung in die Bücherkunde (Vienna: Trattner, 1777-78) was divided into two parts: "Bibliographie" and "Literargeschicht." While the second part has an enumerative function, listing the main literature of various disciplines, the first part, "Bibliographie," is devoted to such matters today referred to in English as historical bibliography.[11] Blum maintains (col. 1112), furthermore, that it was Denis who probably provided the basis for the distinction made by Jean François Née de la Rochelle between "literary bibliography" and "typographical bibliography":

As a result primarily of the categorical formulations of J. S. Ersch and C. G. Schütz, the term in German was already restricted to its enumerative meaning by the late eighteenth century. Historical bibliography was given the separate designation "Geschichte des Buchwesens," which it still has today (Blum, cols. 1149-1153). In France, however, the state's sudden acquisition of rare and valuable books confiscated from private libraries during the Revolution brought about a requirement for extensive research into historical bibliography.[13] Enumerative and historical bibliography were thus necessarily fused in French bibliography throughout much of the nineteenth century, and it was not until 1934, according to Malclès (p. 7), that bibliographie in France was officially restricted to its enumerative sense.

English usage apparently never abandoned the double view of the term. The 1830 edition of the Encyclopedia Americana still includes both meanings in its definition: bibliography "is, therefore, divided into two branches, the first of which has reference to the contents of books, and may be called for want of a better phrase, intellectual bibliography; the second treats of their external character, the history of

particular copies, etc., and may be termed material bibliography."[14] In his 1892 inaugural address, Copinger was still making use of the same terminology.[15]

Professor Bowers' division of bibliography into two general categories is, consequently, the latest stage in this process. Yet here must be injected a word of caution, for between Copinger and Bowers something happened which is usually referred to as the New Bibliography. It is not to be assumed that the uniquely Anglo-American New Bibliography simply carried forward the old meaning of material bibliography. On the contrary: we might even go so far as to say that the New Bibliography actually eliminated material bibliography as an independent subject and squeezed aspects of it into the other side of the field in a way not unlike that which occurred in France and Germany in the previous two centuries. Copinger's idea of material bibliography was "for example, Brunet's Manuel; Ebert, &c." (p. 31), names which today are primarily associated with enumerative, or at most proto-descriptive bibliography. Professor Stokes clouds the issue somewhat when he suggests that Hibberd's divisions are similar to the old divisions (p. 21), for material bibliography—as it was understood when the term was in current use—is quite a different thing from Hibberd's "physical bibliography" or Lancour's "Book as Physical Entity,"[16] since it knew practically nothing of the recovery of the mechanical processes of individual book production. Indeed, what was essentially new about the New Bibliography was that it did not so much augment nineteenth-century material bibliography as replace it with something different which today goes under the name of analytical bibliography.

There is, therefore, no such established term as analytical bibliography in French or German because there is to no great extent any such thing.[17] Any objections to the continued use of the designation "analytical

bibliography" on grounds that it is either antiquated or that it does not coincide with the meaning of the term "bibliography" in other Western European languages are consequently unfounded.[18]

It may be accepted, then, as an established condition, that the subject of bibliography in English is divisible into two broad categories or divisions—one primarily concerned with the listing of documents, the other with the study of documents as material objects—and that other branches of bibliography may be arrayed as subdivisions beneath these two divisions. What has not yet been sufficiently established, however, is a precise definition of these divisions and subdivisions in relation to each other. Professor Stokes, who has written an entire book on the categories of bibliography, provides a thoughtful and useful exposition of the characteristics of Bowers' five distinctive categories, yet although Stokes' purpose is "to bind up the parts and create a mutual awareness of interdependency" (p. 9), he does not believe in strict categorization. His book, therefore, is mainly descriptive: it contains no final conclusion and the reader is consequently left "to bind up the parts" for himself on the basis of the preceding descriptions.

Hibberd's laudable effort to divide bibliography into two generic categories acceptable to all bibliographers is weakened somewhat by his preference for subdividing these two categories into a set of phases rather than into distinct disciplines or activities. Also, it is incorrect to believe, as many bibliographers seem to, that Hibberd's "reference bibliography" is "the term suggested for what has been called 'enumerative' or 'systematic' bibliography."[19] The fact is that Hibberd (p. 128) includes all five of Bowers' branches, including enumerative bibliography, under the category of "physical bibliography." Enumerative bibliography is, in Hibberd's scheme, the first phase of both physical and reference bibliography (p. 130).[20]

It should, of course, be noted at once that Hibberd's aim—like that of Stokes—is not to posit hair-splitting definitions but rather apparently to throw light on the characteristics or processes of his two categories with the ultimate goal of showing that they are "not antagonistic but complementary" (p. 133). This is hardly surprising since both Hibberd and Stokes have relied heavily—as one must—on Professor Bowers' definitions which are themselves to a very great extent based on principles of reciprocal applicability. That is indeed the current trend which Professor Bowers has established in the definition of bibliography. His distinctions have always been immensely practical and are based less upon what a thing is than on what it does or rather how it can be used. Professor Stokes' rejection of abstract definition in favor of a description of applications (the "only valid definition of an orange is to eat one" [p. 10]) is indicative of this trend.

But bibliography not only does, it also is. Thanks especially to the work of Professor Bowers the field is fairly well established as a legitimate discipline, and we who have inherited this legacy need no longer feel constrained to provide pragmatic justifications for its every activity. It is now possible—and necessary—to view bibliography and its various branches as entities unto themselves and to define them no longer from the standpoint of their applicability to each other or to allied disciplines, but rather on the basis of their own characteristics, which is only to say on the basis of how as activities they relate to their objects of study. This is important: for if we intend to establish exactly how the various branches relate to each other, we must first define them on the basis of their separate relationships to their objects.

And what is, generally speaking, bibliography's object of study? In all cases it is the same: material media designed and used for the transmission of information. Since information is essentially immaterial,[21] it must be encrypted into a material sign system by some "emitter" (as an author) in order to be transmitted to some receiver (e.g. the reader) and subsequently decrypted. The objects of study of bibliography are material sign systems. It is therefore reasonable to make use of some of the elementary concepts of semiotics—the science of signs—in order to determine

the relationships of the various branches of bibliography to their objects and ultimately to each other.

II

Semiotics (sometimes called semiology) originated at about the same time as the New Bibliography, in the early years of the present century. Its places of origination were Switzerland and the United States. The reason for this dual citizenship of semiotics is that the field was created at about the same time by two scholars working independently in different disciplines: the Swiss linguist Ferdinand de Saussure (1857-1913), the founder of structural linguistics, and the American philosopher C. S. Peirce (1839-1914), usually credited with the foundation of American pragmatism.[22] Since most of the work of both men was published posthumously, it is not possible to say precisely when semiotics came into being. Neither Peirce nor Saussure had apparently any knowledge of the work of the other. Their separate systems, therefore—Saussure's linguistically based semiotics and Peirce's logically based semiotics—seldom overlap, so that combining the two into one great system would be a very complex task. Most work in semiotics is consequently accomplished through the application of either one system or the other.

It is to Peirce's logical semiotics that we should now direct our attention, or rather to one small aspect of it, namely the Second Trichotomy of Signs. While the Second Trichotomy may have attained some notoriety among philosophers because it is "Peirce's most important division of signs,"[23] the popularity it has achieved outside of philosophy is probably more a result of the fact that it is, at least on the surface, one of the easiest of Peirce's many sets of definitions to understand. Even so, it should be well noted that what follows in no way presumes to be anything like a complete exposition of Peirce's Second Trichotomy, for Peirce's philosophy was systematic,[24] and to discuss any single part without reference to the whole is open to justifiable charges of misrepresentation. Our subject is bibliography, however, not philosophy, and if we

end up ignoring the full philosophical implications of Peirce's categories—which we will—it is because the full implications are not essential to our subject. We are concerned here not with what Peirce ultimately meant, but rather with the application to our own subject of an isolated segment of what he said.

Peirce was of the opinion that logic is basically semiotics, the "quasi-necessary, formal, doctrine of signs" (II,134). A sign is simply "something which stands to somebody for something in some respect or capacity" (II,135). Words, for instance, are signs, most of which stand for specific objects or actions. All such semiotic relationships consequently consist of three parts: there is the sign (Peirce's word is "representamen"), there is the entity for which the sign stands (its "Object"), and there is one's mental image (the "interpretant") of the object triggered by one's encounter with the sign.

Semiotics—or logic—consists, according to Peirce, of three parts: speculative grammar, critical logic, and speculative rhetoric. Speculative grammar, "the first and most fundamental of Peirce's three main divisions" is primarily concerned with "the meaning of signs and symbols" (Feibleman, p. 89). A central aspect of speculative grammar is the idea that all signs are divisible into three trichotomies.[25] The second and most famous of these divides signs into three categories from the standpoint of their relationships to their Objects. Peirce named these three categories of signs "icons," "indexes," and "symbols."

An icon is a "sign which refers to the Object that it denotes merely by virtue of characteristics of its own . . ." (II,143). It is linked to its object by virtue of similarity. A portrait of someone, for example, is an icon.

An index is a "sign which refers to the Object that it denotes by virtue of being really affected by that Object" (II,143). It is linked to its object by virtue of causality. The symptoms of a disease are indexes to that disease.

A symbol is a "sign which refers to the Object that it denotes by virtue of a law . . ." (II,143). Peirce's concept of the symbol is often accepted as being very similar to Saussure's definition of the linguistic sign. Most linguistic signs are arbitrary, or rather the connection between the sign and its Object (the "signifier" and the "signified" to use the English equivalents of Saussure's own terms) is an arbitrary one. There is no reason, for example, why the sounds which make up the word "tree" should refer to a large plant with branches. They do so only

on the basis of convention, or as Peirce would have it, by virtue of a "law." A word, then, in Peirce's terms, is a symbol.

Before proceeding to apply these categories to the field of bibliography, we must first make one terminological adjustment. While Peirce uses the word "Object" to denote that entity for which the sign stands, we shall instead use the word referent. Thus the word "tree" is a sign and its referent is a large plant with branches. We will continue to use the word "Object" in a non-semiotic sense specifically to denote the object of study of a particular discipline. Thus the Object of all bibliography, as we have already noted, is a document or any material medium (i.e. sign system) designed and used for the transmission of information.

III

The decryption of written or printed signs (graphemes) on the basis of linguistic convention is a definition for the process of reading. The signs fall under the Peircean category of symbol. A thorough analysis of the meaning and significance of such signs in certain kinds of documents (especially those which convey imaginary information) is the primary activity of the professional reader or literary critic.

Sign decryption in the field of analytical bibliography, on the other hand, is quite different since, to begin with, it is not accomplished on the basis of linguistic convention.

Although it is unlikely that Professor Bowers had in mind the Peircean concept of "symbolic meaning," his explanation nevertheless coincides neatly with the semiotic distinctions outlined above. It is not that the inked shapes on the page have no meaning in analytical bibliography, but rather that they indeed have no symbolic, i.e. verbal meaning. They do, however, constitute a sign system—or at least part of one—and they do consequently transmit meaning for the analytical bibliographer (otherwise there would be little point in his study of them). Both the bibliographer and the reader, therefore, assign separate referents, i.e. separate meanings, to essentially the same set of signs.[27]

If the reader, then, accomplishes sign decryption on the basis of linguistic convention, how is essentially the same set of signs decrypted by the analytical bibliographer? "Analytical bibliography concentrates on the examination of books as tangible objects in order to recover the details of the physical process of their manufacture."[28] In other words, the purpose of analytical bibliography is to establish, through intensive examination and with the assistance of previously gathered information on techniques of book production (historical bibliography), how a particular document (or set of documents) came physically to be the way it is. That is the essential function of analytical bibliography.[29] The relationship between the sign and its referent is for the analytical bibliographer causal. Thus signs which from the standpoint of the reader are symbolic, must be understood by the analytical bibliographer to be indexical.

Let us now turn to enumerative bibliography. The enumerative bibliographer is traditionally concerned with the intellectual content of publications.[30] Yet again: our purpose here is to establish the precise relationship of enumerative bibliography to a sign system constituting a document. We must do this in order to obtain an exact definition, for to define enumerative bibliography as, for example, "primarily concerned with the ideas in books and their circulation among men"[31] is to fail to note those characteristics which render its activities unique and distinct from all others. Does the enumerative bibliographer, then, like the reader, indeed view a text as a set of symbolic signs, the decryption of which depends upon linguistic convention? Yes and no. Clearly the enumerative bibliographer must understand the verbal content of the text, yet while this is the essential function of reading, it is only a means to accomplish the essential function of enumerative bibliography, namely the production of "a list of books arranged according to some permanent principle."[32] It is therefore reasonable to maintain that ultimately the enumerative bibliographer does not view the document as a sign system but rather as a set of referents to which a sign system of his own

generation, i.e. his bibliography, is to refer. This is admittedly a difficult point, but it is valid. The reader is concerned about what the verbal signs in the text signify. The enumerative bibliographer is rather concerned only that they signify concepts which fall within the confines of the "permanent principle" upon which his selection of documents is based (although obviously in order to accomplish this he must be aware of what they verbally signify). The essential activity of the enumerative bibliographer, then, involves in all cases the generation of what we might call "meta-signs," that is to say signs which specifically signify other signs. And what, finally, is the relationship of enumerative bibliography as sign system to its Object, i.e. to the documents it lists? "Les répertoires [bibliographiques] imprimés depuis le XVe siècle, qui se comptent aujourd'hui par dizaines de milliers, sont des nomenclatures de textes, également imprimés."[33] Bibliographies are "printed" works listing "printed" works. Their purpose, and the purpose of enumerative bibliography in general, is duplication or representation. This is not to say that enumerative bibliography is not a creative discipline. It is very much so: its creativity is, however, restricted to the origination of meta-sign systems and their subsequent coordination (i.e. classification according to the "permanent principle"), and does not involve the relationship of the meta-signs qua signs to their referents. An enumerative bibliography reproduces its Object in microcosm; it is a reflection, a picture of its Object. As such, the relationship between sign and referent in enumerative bibliography is one of similarity and may consequently be designated iconic.[34]

This most essential feature of enumerative bibliography is clearly visible in all areas of that field. The purpose of subject bibliography, for example, is to provide a relatively concise representation of an individual subject through the abbreviated duplication of some or all of its constituent documentation. Part of this duplicating or representational process may consist of what are traditionally known as annotations. Hibberd's "text-as-text" versus "text-as-content," and "external description" versus "internal description" (p. 131) are valid and useful distinctions,

yet only inside the sphere of iconic sign systems. A description of the contents of a publication—which is the function of most annotations—remains at its most basic level an (often difficult) exercise in duplication. That is the essential relationship between enumerative bibliography and the document. Since every document is itself a sign system, and since every sign system involves "something" standing for "something else," it is only natural that efforts are made not only to duplicate certain signs contained in a document but also to represent certain parts of the meaning for which certain signs stand. "External description" tends to reproduce signs, "internal description" tends to reproduce referents. The point is in any case that enumerative bibliography does not create new content but rather duplicates (re-presents) content which already exists.[35]

If the relationship of the signs of an enumerative bibliography to their referents defines such signs as essentially iconic, the relationship of the signs of a descriptive bibliography to their referents must be the same, only more so. The function of descriptive bibliography—whether comprehensive or degressive—is to duplicate certain portions (Hibberd would say "external" portions) of documents with great precision. The call for the use of photographic processes to aid description is indicative.[36] Even if the ultimate purpose of the activity is understood to be the description of an ideal copy, moreover, the relationship of the description to its Object is none the less one of duplication, for its Object consists of certain features of a document as the printer intended them to appear. It is a central quality of the icon that "it affords no assurance that there is any such thing in nature [i.e. its referent]. But it is of the utmost value for enabling its interpreter to study what would be the character of such an object in case any such could exist" (Peirce, IV,359). Any elementary book on physics will normally contain a diagram of an atom, even though no one (yet) has ever seen one distinctly. The diagram is rather a representation of how an atom should appear on the basis of all currently available evidence. The description of an ideal copy is a similar effort to represent the characteristics of an entity not itself subject to autopsy.[37]

It is time, then, to reject the view that there is a generic difference between enumerative and descriptive bibliography. The difference is quantitative only. This is clear from many formulations made by Professor Bowers:

Both enumerative and descriptive bibliography, moreover, rely on analytical bibliography, although it would probably be most accurate to state that enumerative bibliography does or should rely on the results of analytical bibliography, while descriptive bibliography relies on its techinques. It is obviously impossible to produce a descriptive bibliography without an analytical study of the various conditions of the documents described. Yet analysis and description are nevertheless two entirely separate operations: in the former the document is treated as a product of certain mechanical procedures, in the latter the document (in its various conditions) is approached as a set of representable characteristics—a raw material—from which a product, the description, is to be created. This latter treatment of the document is, at this general level of categorization, identical to that of enumerative bibliography.

Another point used to differentiate enumerative from descriptive bibliography is that the latter should be confined to the description of primary works.[38] There appears, however, to be no reason to preclude the application of Professor Bowers' excellent system of description to secondary works as well, always assuming, of course, that the result justifies the effort. The designations "primary" and "secondary" are, in any case, relative. The Life of Samuel Johnson may be either primary or secondary, depending on whether it is being regarded from the standpoint of Johnson or Boswell.

One value of the original distinction between enumerative and descriptive bibliography was that it emphasized the degree of precision possible in bibliographic description. That was a point well worth making and a point well made, but the time eventually comes when categories introduced with the best of didactic or polemical intentions outlive their usefulness and begin to impede the progress of the discipline

they were originally designed to promote. When that point is reached, such categories must be discarded and replaced by others more reflective of actual conditions.

It would be comfortable to stop here—to sit back and complacently view the symmetry. Once we have embarked upon a particular analytical route, however, we are obliged by the requirement for consistency to continue upon that route until we arrive at its ultimate conclusion. And the ultimate conclusion is this: if the relationship of enumerative bibliography to its Object does indeed involve the reproduction of portions of that Object, and if that of descriptive bibliography is the same only with greater precision and in greater detail, then there must be a final level of such a representational relationship at which the goal is to reproduce the Object with maximum precision and in every detail. The name we give to the discipline practiced at that level is, of course, textual criticism. That is to say, then, that the activity of textual criticism, by virtue of its most elementary function—the reproduction with the greatest possible accuracy of sign systems which either once actually existed or which were intended to exist—can and must be described as the concentration of the essential representational activity of enumerative and descriptive bibliography onto a single, total document. The only difference, at this level of basic definition, between the citation of Marlowe's Doctor Faustus in the NCBEL, its description in Greg's A Bibliography of the English Printed Drama to the Restoration, and its reproduction in the textual portion of Greg's 1950 critical edition of the play, is the degree to which the document is duplicated.

This conclusion may be underscored somewhat by recalling that the relationship between textual criticism and reading is practically identical to that noted above between reading and enumerative bibliography. Reading provides a means to carry out the operations of textual criticism, but it does not constitute the essential characteristics which distinguish textual criticism from all other activities. Once again: while the reader uses sign systems to arrive at verbal meaning, the textual critic uses verbal meaning derived through the process of reading as an assistance to arrive at the proper representation of the sign system.

Textual criticism, like descriptive bibliography, draws heavily upon analytical bibliography as well. It is now a fairly well accepted premise (at least in Anglo-American scholarship) that any literary critic who relies on a text produced without the assistance of analytical bibliography will sooner or later end up criticizing soiled fish. The application of analytical bibliography to textual criticism is indeed so fundamental that a name has actually been coined for it: textual bibliography. Two

remarks must be made on this term, one concerning its nature, the other its effect. First, textual bibliography is not—or at least not yet—a discipline in itself: the term rather signifies the application of one discipline to another (what we will later designate an "efficient relationship"). As such it can have no place in any strict taxonomy of bibliographic disciplines. Second, introduction of the term has tended to re-enforce the notion that analytical bibliography and textual criticism are generically related.[39] Such a notion is typical of inaccuracies which can arise when activities are defined on the basis of their relationship with each other rather than first on the basis of their relationship with their Objects. Textual criticism is far more closely related as an activity to descriptive (or even enumerative) bibliography than to analytical bibliography, for like descriptive bibliography it views its Object essentially as a set of representable signs from which a product—a critical edition—is to be fashioned (always assuming, of course, that we equate textual criticism with editing). This is an entirely separate activity from approaching the document as a set of signs to be decrypted, which is the essential function of analytical bibliography.

By its very nature bibliography is a highly conservative field and reevaluation of established definitions and categories is perhaps not as welcome as it is elsewhere. It should be noted, however, that a definition of bibliography in the sense of the iconic constellation of "enumerative bibliography—descriptive bibliography—textual criticism" (hereafter: "EDT") as a reduplicative sign system is, from one point of view at least, a very conservative one, for it reflects the original meaning of the word quite closely. βιβλιογραøια means "the writing of books," not in the modern sense of "authoring" but in the literal, physical sense, i.e. duplication by means of written signs. A βιβλιογραøοζ is a scribe or copiest, so that to define certain forms of bibliography in terms of their function of reproducing all or part of selected texts is hardly to change the original meaning of the word at all.

We are left now only with the problem of historical bibliography. Professor Bowers' most precise definitions are found in his Encyclopaedia Britannica article. There he does not consider historical bibliography as properly belonging to the field, but states instead: "Ancillary to analytical bibliography . . . are the numerous fields concerned with the study of printing and its processes both as art and as craft" (p. 588). One must agree that historical bibliography is not, properly speaking,

bibliography at all. This is because it does not have as its Object material sign systems or documents. Its Object rather consists primarily of certain mechanical techniques and as such it must be considered not part of bibliography but a constitutent of such fields as the history of technology or, perhaps, information science. Its relationship to analytical bibliography, moreover, is frequently somewhat similar to the relationship of language to reading: historical bibliography often provides the causal basis for sign decryption in analytical bibliography just as language provides the conventional basis for sign decryption in reading.

Let us return now to the subject of analytical bibliography and summarize the semiotic basis for its isolation.

a. the sign: while many (but by no means all) of the signs considered by the analytical bibliographer have the same form (i.e. inked shapes on paper) as those considered by the reader, they invariably have a different referent or meaning. For the analytical bibliographer they do not equate to mental images but rather to activities of those responsible for the production of the document. This distinguishes analytical bibliography from reading.

b. the referent: the essential activity of analytical bibliography involves the acceptance of the Object as a closed set of signs which require decryption. It does not approach the document as a referent for which a new set of signs (what we have called "meta-signs") is to be generated. This distinguishes analytical bibliography from EDT.

c. the relationship of the sign to the referent: in analytical bibliography the sign is related to its referent on the basis of physical causality, not on the basis of similarity or convention. The relationship in Peircean terminology is consequently indexical rather than iconic or symbolic. This distinguishes analytical bibliography from both EDT and reading.

IV

We may now turn finally to the problem of the relationships between the various branches of bibliography. Most of what has been previously established can be summarized in the following diagram which, for purposes of completeness, includes the process of reading. Σ represents a closed material sign system or document (what we have been calling the Object), R stands for reading, and A for analytical bibliography.

Broken lines in the above diagram represent objective relationships, i.e. that of an activity to its Object. These are the relationships we have been using to define the branches of bibliography, but they may also be used to assist in the definition of the relationships between the branches. Thus the unbroken vertical lines (between EDT) represent generic relationships. The activities in such a relationship form a class, the basis for which is that their relationships to their Object do not differ in essential quality but only in degree. Members of this class are related through the possession of similar attributes, but not by any operational interdependence. Each activity operates independently of the others, and knowledge of any special, non-essential requirements of one such activity is not necessary for the successful operation of another. Since EDT does form a distinct class, we may now provide it with a generic designation: representational bibliography.

Horizontal arrows stand for efficient relationships. Such a relationship arises when one activity is necessarily affected by another essentially different activity. We must use arrows rather than simply lines because such an efficient relationship is directional and the arrows indicate the direction of influence from the causal entity to the affected one. Thus while the elements in a generic relationship are independent or simultaneous, an efficient relationship is one of dependence or succession.

One particularly problematic area in the distinctions made by Professor Bowers has been that his preference for describing textual criticism and descriptive bibliography mainly in terms of analytical bibliography has tended occasionally to draw attention away from the unique characteristics of the latter. This has, per force, weakened another of his central arguments, namely that analytical bibliography "can be pursued independently of any limited objective" (Ency. Brit., p. 588). The distinction between generic and efficient relationships should alleviate this problem, for it establishes analytical bibliography as a generically separate activity in its own right which, at the same time, can be related

on a different level than the generic to the separate activities of representational bibliography.

But on what basis may we posit such efficient relationships? Again, only on the basis of objective relationships. The relationship between reading and its Object is essentially interpretive (ι1). It functions to interpet the individual meanings of the signs in the system as verbal symbols. The relationship between analytical bibliography and the same Object is also essentially interpretive (ι2), for it, too, interprets the meanings of signs in the system, although it views them as mechanical indexes. This is significant: for to state only—as has been the common practice—that analytical bibliography serves textual criticism which in turn serves literary criticism is a rather limited, "two dimensional" view of the functions of these fields. Only by adding the "third dimension" of the objective relationship do we realize that analytical bibliography as an activity has much more in common with literary analysis (i.e. reading) than with either descriptive bibliography or textual criticism, since both reading and analytical bibliography are essentially interpretive activities. If we were to oversimplify the entire system, which might be justifiable, say, in the classroom for pedagogical purposes, we could state that analytical bibliography, as opposed to representational bibliography, is simply another form of reading: the analytical bibliographer merely "reads" in a different "language," namely that provided often by historical bibliography. The semiotic systems recognized by the literary critic and the analytical bibliographer are different, but the fundamental skills applied to the two systems are essentially the same.

The objective relationship, on the other hand, of representational bibliography is—as we have defined it—one of duplication (δ). Its essential function is to duplicate sign systems, yet prior to such duplication, decisions must be made concerning which signs are to be duplicated: which titles in which order, which "internal" and "external" data, which variants. Most of these decisions can only be made through interpretation of the signs available. Thus interpretation necessarily precedes and affects duplication, although it remains a generically separate operation.

It should, in conclusion, be emphasized that analytical bibliography, no less than reading, may as a distinct activity, produce results in any single instance which do not necessarily affect representational bibliography. Chain-line direction, for example, may be a sign which sometimes decrypts to half-sheet imposition, yet that information may not necessarily influence decisions made in the critical edition of many documents where such indexical signs are found. What is the value, then, of results of analytical bibliography which do not specifically affect representational

bibliography? Many bibliographers begin as literary critics, yet they quickly seem to forget that the practitioners of humanistic literary criticism (as opposed to Marxism or other forms of criticism which accept absolutes outside of the text) have no easy time justifying their existence. To say, then, that analytical bibliography is the "grammar" (or the "dictionary") of literary investigation, to maintain that it exists only to be applied to the production of accurate texts upon which literary criticism may securely ply its craft is not much of a justification. The simple fact is rather that any set of signs, once recognized as such, necessarily contains within itself a number of problems: what do the signs signify? how did they come to be the way they are? how do they relate to other signs? can they be replicated or imitated? The only difference between disciplines is that each recognizes different sets of signs or different varieties of problems inherent in them, and it is in each case the simple recognition of such problems which justifies the search for their solution.

Notes