University of Virginia Library

II

The recent essays on this subject continue the historical trend toward the criticism of the scientific analogy: they find fault, in one way or another, with the supposedly scientific pretensions of bibliography. Insofar as they touch on the nature of bibliographical evidence or the historical aspect of the field and fail to make distinctions between one kind of bibliography or one aspect of science and another, they repeat past history. In this sense the TLS is right in saying that they have "revived the old argument about the scientific nature of bibliography" (though apparently it was never dead). But in another sense they are pitched on a different level, for they offer extended discourses on the philosophic background, the methodology, and the logic of bibliographical demonstration. It does not matter if, for purposes of argument, they assume greater claims for the scientific rigor of bibliography than have normally been advanced; but they do little to alter one's feeling that the question of science in bibliography, initiated as a metaphor to help elucidate the nature of the subject, has developed into a verbal smog which threatens to hide it.

McKenzie's "Printers of the Mind"—the starting point for the current debate—is essentially a statement of the weaknesses of the inductive method.[50] Many of the conclusions reached through analytical bibliography, McKenzie shows, are unsound or less certain than

they were thought to be, because in each case a generalization was based on an insufficient body of inductive evidence. The question which obviously follows is whether any body of inductive evidence can ever be large enough to support more than a reasonable guess. Although McKenzie is ostensibly criticizing bibliography for not being sufficiently "scientific," his discussion demonstrates that bibliography is like "science" in proceeding by empirical observation and that the problem of induction is therefore basic to both. Philosophers have never proposed a satisfactory solution to the problem of induction. Indeed, in the form in which it is often posed, there can be no solution: for if induction is by definition not a form of deduction, and if valid conclusions can result only from a deductive argument, then induction must be ruled out as a legitimate process of logical demonstration.

McKenzie's way of dealing with this dilemma is a standard one: to advocate the insertion of qualifications in any inductive generalization and thus the conversion of such generalizations into hypotheses to be tested deductively. In his words, "A franker acceptance of deductive procedures would bring a healthy critical spirit into the subject by insisting on the rigorous testing of hypotheses, and the prime method of falsification—adducing contrary particulars—would impose a sound curb on premature generalizations" (p. 61). This line of reasoning—given its classic statement in Karl Popper's The Logic of Scientific Discovery (trans. 1959)—rejects inductive generalizations in favor of unfalsified hypotheses; but it does not confront squarely the logical objection to inductive evidence, since any finite body of evidence which fails to provide falsification for a hypothesis would be open to the same kind of objection. There would seem to be little difference between a generalization held provisionally to be true on the basis of examined evidence and a hypothesis for which no falsifying evidence has yet been located. In either case, further investigation may overturn present judgments. This sort of argument, in other words, appears to make little distinction between induction and deduction, except for the supposed greater caution of the latter. But if the goal of observation is to find some kind of regularity that will be useful in making further observations, excessive qualification may almost negate the process. As Max Black says, "In converting a purportedly inductive argument into a valid deductive one, the very point of the original argument— that is, to risk a prediction concerning the yet unknown—seems to be destroyed."[51] One could perhaps restate McKenzie's observation, without

recourse to induction or deduction, simply by saying that bibliographers should be more careful in framing general statements and more thorough in surveying the relevant evidence. Clearly this is sound advice, and the most impressive part of McKenzie's essay is his effective account of instances in which bibliographers have jumped to conclusions that must be modified in the light of further evidence. McKenzie's article is important and timely: his work on the Cambridge University Press records has put him in a position to understand the value of knowing in detail the various jobs in progress in a printing shop at any one time, and one of the weaknesses of much bibliographical analysis in the past has been that the production of a single book was looked at in isolation, without sufficient regard for the total activity of a shop. The great value of McKenzie's essay, in other words, seems to me to lie in its challenge to widely held generalizations rather than in its theoretical discussions about the logic of bibliographical investigation.

Nevertheless, the objections to induction which McKenzie summarizes ought to be faced by bibliographers—anyone whose work involves argument from empirical observation should give some thought to the logical validity of what he is doing. The inconclusiveness of inductive reasoning cannot be denied, but it seems shortsighted to limit "scientific" argument to the deductive. Philosophers of science recognize that there is no such thing as "the" scientific method, except perhaps in the broadest characteristics.[52] One can say that "science" or "scientific method" involves scrupulous fidelity to evidence obtained empirically and a systematic means of handling that evidence. But the details of the procedure will vary from one kind of situation to another or from one area of endeavor to another. Inductive investigations can be "scientific" in this sense, and to deny their legitimacy is greatly to restrict the range of research. In justifying induction one must finally turn to the pragmatic or common-sense argument of common experience. Everyone, from birth, learns to get along through an inductive process. From time to time one's generalizations are proved incorrect, when the expected does not occur, and one makes adjustments in the generalizations; but the whole concept of "rationality" or "rational behavior" depends on expectations of regularity based on past experience. Perhaps there is no ultimate regularity in the universe; the point, however, is that the projection into the future of a seeming regularity from the past appears to be the only way of proceeding in

the short run. If induction is denied, all human concepts would seem to be destroyed with it.[53] Furthermore, a deductive argument is conclusive only in terms of its premises, which may themselves be unrelated to the "real" world (that is to say, logical validity and truth are separate concepts). Therefore, to establish "truth"—that contact with the "real" or "objective" which is the aim of research—involves the testing of those premises by what amounts to an inductive procedure, even if it is expressed in terms of Popper's theory of falsification. In other words, one is driven to induction on pragmatic grounds, despite the unassailability of logical objections to it. I am making this amateurish summary of a familiar philosophical debate in order to suggest two points: first, bibliographers—though they should understand the implications of inductive reasoning—need not hesitate to proceed inductively, so long as they do so with care and responsibility; second, to collect and examine evidence with care and responsibility is by definition to be scientific, and discussions about whether or not bibliography resembles one particular scientific pursuit or another seem somewhat fruitless exercises (except perhaps to demonstrate the multiplicity of individual paths which scientific endeavor takes).

The more direct and positive approach to scientific method in bibliography is to accept induction openly and to set about examining what constitutes responsible handling of inductive evidence in this particular field, given the nature of the problems which bibliographers wish to solve. Fredson Bowers did exactly that fifteen years ago in his Lyell Lectures. After distinguishing analytical bibliography (concerned

with books as "tangible objects") from textual bibliography (in which analytical bibliography is applied to "internal form, or contents" of books) and suggesting in general the relations of bibliographical research to editing, Bowers examines the nature of bibliographical evidence and states that one of the "laws" of bibliographical procedure "requires us to reason inductively from specific, concrete evidence in the text" rather than deductively from "our general ideas about printing practice" (p. 36). Of course, if "our general ideas" were adequately buttressed with evidence, there would be no problem, but finding that evidence returns us to an inductive search—thus the inductive process is basic, whatever it is called, and Bowers is not interested in debating the terminology.[54] Instead, he proceeds to—what is the heart of the matter—the question of the interpretation of inductive evidence, and he sets up "three orders of certainty": the demonstrable, the probable, and the possible. Now to say that inductive evidence can ever lead to a "demonstrable" case (one in which physical evidence "leaves no loophole for opinion") entails certain assumptions —that all relevant evidence is known and has been examined[55] or that extreme coincidences do not in fact take place. In other words, one has to begin with some notion of the range of occurrences which it is reasonable to expect. Bowers calls this notion the "postulate of normality," which "depends on the working hypothesis that all we know at any given time must be the truth, and therefore the details of the printing process and their handling that have been recovered (when tolerably full) must represent 'normality' unless we have stubbornly inexplicable evidence to the contrary" (p. 72). The phrase "when tolerably full" underscores the central problem, since one must have surveyed a certain quantity of evidence in order to interpret a new piece of evidence, and yet without that new evidence itself the interpretation may be faulty. Nevertheless, some assumption of normality is unavoidable: At this point McKenzie would say that we do not have a large enough body of evidence to define satisfactorily any kind of "normality."[56] Still, his description of a deductive process based on a recognition of "the partial and theoretic nature of bibliographical knowledge" is not, in practical terms, very different from Bowers's picture of an inductive procedure in which explanations "based on imperfect evidence" are modified or corrected by new evidence. Obviously we never know enough; but if we are to proceed, we have to assume that we know enough to get on with. Bowers's discussion, by providing numerous examples of what he regards as demonstrable, probable, and possible interpretations of bibliographical evidence, shifts the focus from the theoretical to the practical. He is not principally concerned with arguing the philosophical question of inductive versus deductive reasoning; the assumption lying behind his analysis seems to be that, since the demonstration of a "truth" finally rests on empirical observation, one might as well accept induction and proceed to confront and examine the evidence that turns up. As a result, his book is a more direct investigation of the "scientific" nature of bibliography than any other discussion, for, instead of concentrating on how well bibliography conforms to certain abstract qualities of "science," he looks at concrete examples of what bibliography in fact consists of, in order to see what particular brand of "scientific method" emerges as most appropriate for dealing with bibliographical evidence.

In the same year in which McKenzie's essay was published, James Thorpe delivered a paper entitled "The Ideal of Textual Criticism" at a Clark Library Seminar.[57] One section of this paper collects quotations

from several bibliographers—especially Greg and Bowers—which seem to assert the scientific nature of "textual criticism," and it cites a few instances of quantitative and mechanized approaches to textual problems. Thorpe then proceeds to conclude, "I can see nothing in the present or future of textual criticism, however it is carried on, which will make it answerable to the term 'science' or 'scientific'" (p. 68). Of course, if "science" is taken to mean "the physical sciences," it is easy to agree with him; but, since no analysis of the term or of the nature of textual criticism accompanies the statement, it has the effect of being simply an assertion, placed in opposition to a series of other assertions. The scientific analogy is worth analyzing if some illumination of the nature of the subject emerges, but little is gained by asserting its inadequacy or inappropriateness, particularly since leading bibliographers over the years have repeatedly made the same point. Although the bibliographers cited by Thorpe did make the comments he quotes, I hope that my earlier historical survey has shown the general drift of the major statements of the last half-century to be in the direction of finding fault with the scientific analogy and recognizing the important place which critical judgment occupies. The difficulty here—as in so many similar discussions in the past—is one of definition. Does "science" mean the same thing throughout all of Thorpe's quotations and in his own remarks as well? More to the point, does "bibliography" mean the same thing, and the same thing as "textual criticism"? Although Thorpe elsewhere discusses at length the relation of bibliography and textual criticism,[58] he does not at this point raise the issue of shifting definitions. His subject is specifically "textual criticism," yet the word "bibliography" is what apears in a number of his quotations; and it should be clear by now that the two terms have not normally been regarded as synonymous. (Even when "text" or "textual criticsm" appears, it is by no means certain that the writer is asserting the "scientific" nature of every step of the editorial process.) The trend in recent decades, as shown above, is to think of the techniques of analytical bibliography (analysis of physical evidence) as perhaps having certain "scientific" qualities but to regard textual criticism (often defined to include editing) as being—true to its name—critical. Thorpe may be concerned about the increasing amount of attention which an editor is expected to give to bibliographical information, but that is a different matter from suggesting that the editorial process itself is claimed to be mechanical (which is what "scientific" often means in this context). If Thorpe's conclusion is that editing involves critical judgment (or literary criticism), most people would undoubtedly agree, including those from whom he quotes. One can concur, in other words, with what Thorpe appears to be saying at the end and yet not see how he is led to that statement by quoting comments on "scientific" bibliography and textual criticism from various periods—especially without analyzing the sense in which each writer was using the key terms.

A similar problem emerges in Peter Davison's incisive discussion[59] of McKenzie's position, for Davison is chiefly interested in textual criticism, while McKenzie is concerned with analytical bibliography. In what is surely one of the most penetrating analyses of the nature of bibliography yet written, Davison argues persuasively that McKenzie's view of scientific method is oversimplified and that his view of bibliography as amenable to the "hypothetico-deductive method" is unrealistic. Nevertheless, Davison's examples are editorial problems, and he shows the shortcomings of the deductive method in terms of editing. He points out, for example, that editors "have to provide answers even if evidence is insufficient or contradictory" (p. 13); and he goes on to explain that the deductive method

All this seems reasonable, but it does not meet McKenzie on his own grounds, since McKenzie is talking about analytical bibliography. What starts out as a criticism of McKenzie's position—the advocacy of the deductive method for a particular purpose—turns into a criticism of the appropriateness of the deductive method for a different purpose. I do not believe that McKenzie would disagree with Davison's position in regard to editing: that an editor must frequently make decisions on the basis of his own interpretation and judgment (informed by whatever data are available) rather than on the basis of conclusively established facts. But McKenzie would still say that the deductive method should be followed in bibliographical analysis—in establishing, that is, the facts and hypotheses which may turn out to be of use in the process of editing. It is possible to meet this argument—as I tried to show earlier—by examining the general problem of induction; but Davison, though he makes an effective case against deduction, does not really speak directly to McKenzie's point, since he shifts the area of application to textual criticism. Once again, a debate about the scientific aspects of bibliography is rendered less clear than it might be through the failure to draw distinctions among different kinds of bibliography (or between "bibliography" and "textual criticism").

Davison's important essay takes up a still larger issue. The existence of essays like McKenzie's and Thorpe's, he believes, may suggest that bibliography is at a "moment of crisis," that it is engaged in what Thomas S. Kuhn calls "paradigm rejection"[60] —the replacing of one paradigm by another when the former is judged to be inadequate to handle the problems with which it is faced. He cites examples of dissatisfaction with the usual concepts of "author" and "text" and with the stemmatic approach to textual criticism. He then argues—in the most intriguing part of his discussion—that, just as a creative writer reflects the changing world-views that result from new scientific theories, so an editor (who responds to "the needs, general and scholarly, of his own society") should perhaps "take note of these changes in the physical explanation of our world and the response of creative writers thereto" (p. 27). The rise of the "new bibliography" is placed in this context:

One illustration is the study of the history of textual transmission: Insofar as this argument says that a man inevitably reflects the characteristic interests and approaches of his intellectual milieu, it is making a generalization about all men and not about bibliographers in particular. But in applying this observation to bibliographers, it has the merit of stressing the humanistic aspects of the field, of saying that bibliographers are like "creative writers," historians, and others who meditate on human behavior, in their reaction to scientific theories about the physical universe. By proposing a pervasive influence of science on bibliography, Davison is paradoxically setting the two apart, for he associates that influence not with specific methodological changes but with an altered outlook that bibliographers share with other thinking human beings. The result is to provide a strong affirmation of the creative in bibliography and to reject the idea that bibliography is like "science" (although the rejection is on a deeper level than is usually implied by the comparison).

The affirmation is salutary, and it gains weight from the thoughtful analysis lying behind it. But one is surprised to find the conclusion couched in language which seems to reopen the troublesome issue of "scientific bibliography." Davison is urging bibliographers to a renewed faith in intuition and subjective judgment:

Here is the most sophisticated use yet made of the scientific analogy. But to claim that the recognition and acceptance of creativity in bibliography somehow involve modification of the bibliographical paradigm is to suggest that bibliography has been more rigidly mechanical than would appear to be the case, judging from the statements, and the work, of its practitioners. Even in the limited area of analytical bibliography—or in Newtonian physics—imagination plays its role (in recognizing significant evidence, in devising ways of arranging it, in making connections between related occurrences). Furthermore, if the bibliographical methods springing from McKerrow and Greg are regarded as the reflection, in the bibliographical area, of the nineteenth-century scientific revolution, it is hard to see the aptness of comparing a further development of those methods with Newtonian science. In any case, the emergence of ways of looking at the universe which go beyond Newtonian physics can indeed be said to require modification of a paradigm, since the Newtonian laws were thought to be universal; but the rules of procedure in analytical bibliography were never claimed to have the same kind of universal application throughout the whole realm of bibliographical pursuits. (Newtonian laws might be thought of as operating in all areas of one horizontal plane—which serves well enough to provide a perspective for everyday purposes—though we recognize the existence of other planes; in contrast, rules of analytical bibliography might be thought of as operating in the limited areas of several planes forming one vertical segment of the bibliographical whole, though we recognize that other segments—editing, for example—border on it.) Besides, to compare "Newtonian physics" and the "precise techniques" of analytical bibliography is seemingly to mix explanations with approaches, though both are called "methods." If the methods of the two areas are compared, it is true that both require care and accuracy, but so do all scholarly pursuits; if the discoveries of the two are compared, both share the ultimate inconclusiveness of all inductive generalizations, though one offers in support an incalculably greater body of evidence than the other. That bibliography and science can be compared in certain carefully defined respects and contrasted in others is not at issue. But Davison's closing comparison, like so many similar ones in the past, diverts attention from, rather than clarifies, his main point and therefore does less than justice to what he has to say.

Another essay which stresses the humanistic nature of bibliography appeared at about the same time as Davison's. Morse Peckham, in "Reflections on the Foundations of Modern Textual Editing," is chiefly concerned with examining the concepts of "text" and "author,"[61] but he begins by questioning the appropriateness of the words "mechanical" and "scientific" as descriptive of bibliography. To

think of book production as a mechanical process in which the "human factor" must be adduced to explain anomalies is, he says, an inversion of the truth, since book production is essentially a form of human behavior, and its study is therefore a branch of historiography: That bibliography is a historical study has been expressed before, by Greg and others, in less elaborate language, and Peckham is right in saying that it does not need to be repeated—except in the hope that one more repetition may convert those who apparently do not yet understand. But are there really any bibliographers who do not understand that they are dealing with human productions and human behavior? It seems unlikely, and yet the kind of terminology Peckham objects to undoubtedly persists. The problem, of course, is one of rhetoric. The old scientific analogy has become so entrenched as a cliché that it continually turns up in one form or another. While its history shows that it is productive of enough confusion that it might better be avoided, it obviously still serves a purpose for some writers. I do not believe that most bibliographers who use the term "human factor" would disagree with Peckham's statement that the term "actually explains nothing" and "only admits that the explanation has broken down." One impulse to use the scientific analogy comes from the need to explain just what approach the analytical bibliographer is taking toward human behavior; since he is dealing with an area which involves the use of mechanical instruments—pens, presses, type-formes, type matrices, paper moulds, and so on—his approach is to see how much can be explained by factoring out the "human" element and concentrating on those instruments. Obviously what he is trying to describe ultimately is a human action, but he wants to see how far he can go in that direction by examining the products of mechanical instruments. He can never go all the way, and sometimes he can hardly get started; when he resorts to a term like "human factor," he is admitting that his explanation can go no farther. Certainly the suggestion that his work is a "scientific" treatment of "mechanical" operations is an overstatement and a cliché, but the motivation for it is clear enough. The difficulty comes—as it has repeatedly—when people react to the rhetoric on a different level from the one intended, and Peckham's complaint about "human factor" is another instance.

Nevertheless, he usefully redirects our attention to the basically historical nature of bibliography and recognizes where that leaves the scientific question: "The scientific status of analytical bibliography is the scientific status of any historiographical construct" (p. 132). This, too, has of course been said before, but what Peckham adds that is new is his approach to the definition of historiography. Since statements about past events cannot be verified by empirical observation, he argues, their interior logic has no relation to any "truth" outside the historical account unless the historian, like the scientist, makes predictions about currently existing artifacts, which are thus subject to repeated direct observation. As he puts it in another essay, historical statements "cannot tell us how to locate the phenomenally perceptible, but only how to construct other statements that may, or may not, successfully instruct us how to locate something in the world before us."[62] In other words, the historian "predicts about where he is going to find documents and artifacts and what their attributes are going to be" (p. 133); and "like any scientist," Peckham says, the historian, after checking his prediction, may have to adjust the thinking that led to it. The question whether history is a science has been more widely debated than the question of bibliography's scientific status,[63] and what Peckham has done is to offer another explanation of the sense in which history does resemble science. His argument, though expressed in different terms, arrives at essentially the same point as McKenzie's: that analytical bibliography—or history—is scientific insofar as it continually

tests hypotheses against directly observable evidence, insofar as the "printers of the mind" move outside the mind.[64] In contrast to the natural sciences, however, with the large body of evidence which they have amassed, "analytical bibliography certainly is not a very highly developed science" (p. 134)[65] —a point which Greg made long ago. Peckham's analysis does not reach any new conclusions, but it goes farther than previous discussions in treating the relations of bibliography and science in the context of historiography.

The recognition that analytical bibliography is history should answer any questions about whether it is an independent pursuit or only the servant—"handmaid" is the favorite term[66] —of another pursuit. But the question has generated a considerable amount of heat, and recourse to the analogy of "pure" versus "applied" science has not helped to answer it.[67] When Copinger, and Greg after him, called bibliography the grammar of literary investigation,[68] they did not mean to imply that the grammar was of no interest in its own right. But the fact that analytical bibliography grew up in the hands of people who were concerned with literary texts led a number of literary scholars to believe that it existed only as an aid for establishing texts.[69]

As a result of this way of thinking, Bowers found himself in the position of insisting on the seemingly self-evident: that bibliography is "an independent discipline of scholarship and not merely an ancillary technique to literary investigation."[70] The possibility of understanding the word "bibliography" in various ways is again at the root of the matter, as it is when Thorpe returns to this issue. He quotes S. L. M. Barlow as saying, "It is none of the business of the bibliographer or the pure scientist what use is made of his findings";[71] and he objects by replying, "In my way of looking at textual criticism, its value derives only from serving the useful purpose of helping to present the text which the author intended" (p. 68). But "bibliographer" does not necessarily mean "textual critic"; it can mean "analytical bibliographer," whose field of interest—the printing practices of a given period or a given shop as revealed by physical evidence—is surely a legitimate subject of inquiry in its own right. If it had been as popular to call bibliography "history" as it has been to call it "science," these matters would probably have aroused less controversy; and Peckham's emphasis on the historical nature of analytical bibliography is therefore welcome.

Forty years ago Georg Schneider, in his book on reference bibliography, said, "It makes little difference whether bibliography is termed a science or an art, a technique or a skill, or even all of these together."[72] One is particularly ready at this point to apply the statement

to other areas of bibliography as well. The act of classifying a subject in terms of a larger framework ought to help clarify the nature of that subject, but the history of the association of bibliography with science shows that there are exceptions. Part of the trouble, it is evident, is that bibliography is not "a subject" but a related group of subjects that happen to be commonly referred to by the same term. There should be no problem in recognizing historical bibliography (the study of printing, publishing, and associated areas at particular times in the past) as history; nor is it hard to move from there to an understanding of analytical bibliography (the examination of the physical evidence in books as a clue to the processes of their production) as history. Descriptive bibliography again is history (the history of the forms in which a given group of books has appeared), drawing on both historical and analytical bibliography. It may have occasion, depending on the nature of the problems encountered and the level of detail contemplated, to utilize instruments or methods of measurement generally regarded as "scientific,"[73] but that fact does not make it a "science," except in the general sense that it is striving for accuracy; it may often take as its field of investigation the books written by a particular literary figure, but that fact does not make it a "literary" study. Textual criticism and scholarly editing,[74] however, though they draw on these three historical kinds of bibliography and though they aim at establishing the history of particular texts, deal with questions of meaning in texts which can frequently be resolved only by literary sensitivity, and they can reasonably be thought of as part of the field of literary criticism[75] —but they could also be defined as a form of history, and to debate the matter would be as fruitless as to debate whether they are a form of science. To regard bibliography as principally historical is not to settle anything, since the status of history is also in question;[76] but it places the debate about the scientific nature of bibliography in the context of a larger debate, about which much more has already been written, and it associates bibliography with other pursuits that concentrate on unique past events,[77] thus providing a more immediately acceptable analogue (if indeed it is not a tautology).

The impulse to use a scientific analogy is ultimately the natural human inclination to believe that what one is doing now is more rigorous and precise than what people were doing in the past. Speaking of textual criticism, Housman, in a well-known passage, says that "the most frivolous pretender has learnt to talk superciliously about 'the old unscientific days.'" But the truth is, he continues, "The old unscientific days are everlasting; they are here and now."[78] To talk about what one is doing can sometimes help one to proceed; but there are other times when it seems best to get on with the work and to define the work by doing it.[79]