University of Virginia Library

Science, Method, and the Textual Critic
by
Peter Davison

I

The theatre, it is frequently remarked, is always in a state of crisis. Controversy about the place of analytical bibliography in English studies, not to mention editorial conjectures and refutations, suggests that it has more in common with the drama than a concern for its texts. Even when all is well, bibliography, as David Gallup has said, "can hardly be recommended as an occupation for completely sane people."[1], and for reasons beyond those he gives.

Quite recently Peter Brook argued in an interview with the late James Mossman that the theatre is in crisis, not for any of the old and relatively superficial reasons — because money, or authors, or audiences were in short supply — but because it is no longer rooted in shared, common experience. One does not have to describe how, in many academic communities, there has been a weakening of 'shared understanding' of the nature and purposes of the academy. It is a time of disquiet when basic assumptions are being challenged and it would be surprising, and even alarming, if fundamental questions were not now being asked about the nature and practice of textual criticism. The most notable, and the most sustained, of these critiques is D. F. McKenzie's "Printers of the Mind: some notes on bibliographical theories and printing-house practices" which appeared in Studies in Bibliography in 1969, but McKenzie is not alone. The editors of the Ohio Browning have questioned the meanings of 'author' and 'text'; James Thorpe, the author of a distinguished essay on "The Aesthetics of Textual Criticism," has attacked what he regards as the scientific pretentions of bibliographers;[2] and of much longer standing, though

almost unnoticed by bibliographers concerned with later English literature, George Thomson and George Kane have questioned the very foundation of the use of stemmatics in editing complex manuscript traditions.

It could be that contemporary uncertainty is akin to those "Revolts, republics, revolutions" described by Tennyson in The Princess as "No graver than a schoolboys' barring-out;" but it is salutary to recall that The Princess was published only a year before The Communist Manifesto.

The situation described by McKenzie and implied by the editors of the Ohio Browning is (if, risking Mr. Thorpe's displeasure, I might make a comparison with science) not unlike that defined earlier in the decade by Thomas S. Kuhn and called by him 'paradigm rejection.'[3] Received or normal science, he argues,

Thus, one set of traditions — one paradigm — is replaced by another: Copernican for Ptolomaic astronomy, Newtonian for Aristotelian dynamics. This scientific paradigm serves not for replication as does a grammatical paradigm (so that amo, amas, amat may be replicated as laudo, laudas, laudat) but as "an object for further articulation and specification" (23). Thus the scientific paradigm is open-ended (10) and encourages the solution of puzzles within the terms of the paradigm. If these puzzles prove insoluble within the paradigm they become counterinstances which, when sufficiently significant, lead to the rejection of the paradigm. Are we faced in analytical bibliography with sufficiently serious counterinstances for us to reject our current paradigm?

Kuhn's description of the moment of crisis for the scientist is instructive:

If for 'normal science' one reads 'current bibliographic practice' it is possible to see why it might be true that we have encountered in recent months not just a moment of individual or passing uncertainty but a deeper and more serious crisis. In another respect there seems to be an analogy between McKenzie's response to what he regards as the inadequacy of current bibliographic practice and Kuhn's description of the effect of crisis on some scientists. Dr. McKenzie, it will be remembered, began by arguing that bibliographers must proceed by the hypothetico-deductive method. He pointed out, quite rightly, that our understanding of what went on in the Elizabethan printing house is based on limited evidence and hence assertions of normality must be tenuous. Then, basing what he said on his study of printing at Cambridge University Press between 1696 and 1712 and on the work of two eighteenth-century printers, Charles Ackers and William Bowyer, he considered such problems as workmen's output, edition sizes, and the relationship of composition to presswork. He then attacked current theories of compositors' measures, cast-off copy, skeleton formes, proof correction, and press figures. He argued that bibliographers ought to show a greater concern than they had for historical perspective and concluded by suggesting what seemed to some readers to be what was tantamount to a turning away from analytical bibliography: This is not dissimilar to the effect of crisis on scientists as Kuhn describes it: McKenzie's plea that bibliographers should be more modest in their claims and his preference for historical studies and enumerative bibliography over a proliferation of ill-founded theories is acceptable enough, but his article seems to me to reveal an unease that goes much deeper. It is not surprising that Robert Donaldson, reviewing what McKenzie had written, should say:

McKenzie's strictures are salutary but not, I think, quite so devastating as to lead me to desert bibliography and the practice of editing because I cannot tolerate this crisis (to adapt Kuhn's words). I should like to comment on what McKenzie and Thorpe have to say and to discuss some of the points raised by the editors of the Ohio Browning. I shall try to suggest what seems to me the proper relationship of bibliography to science and what part the hypothetico-deductive method can play in textual studies. I shall endeavour to outline some of the considerations which a modern editor ought to bear in mind in his use of bibliography in the light of what may be a shift in the way in which we apprehend our world which could affect not only bibliography but studies of many kinds and, indeed, creative literature itself.

II

The likening of bibliography to science is understandable viewed in the context of English studies as a whole. The extent to which one can make objective literary critical pronouncements is limited and a technique that attempts to demonstrate something, however modest, which can be checked and agreed against tangible evidence, must seem not unlike a scientific demonstration to those unaccustomed to science and perhaps a trifle envious of it. This, of course, does not make bibliography scientific and a responsible editor (and a responsible reader) will bear in mind Fredson Bowers's words in Textual and Literary Criticism:

Nevertheless, as Bowers's quotation suggests, there is advanced by bibliographers a claim to operate a scientific method and there is no doubt that some bibliographers have deceived themselves into imagining that what they do is scientific. Editors can easily imagine that bibliography is a scientific tool. The desire for objectivity, the attempt to put back the moment of conjecture, the awesome achievements of science, have all contributed to this assumption. It would be surprising if bibliographers could entirely escape the influence of what is, in P. B. Medawar's words, "an immensely prosperous and successful enterprise."[6] And to those seeking objectivity, the scientific method — a scientific method — must have seemed the way ahead.

It is, therefore, not difficult for James Thorpe to select seemingly definitive statements by the most distinguished textual critics — Housman, Greg, Bowers — which bluntly state that textual criticism (or bibliography) is a science.[7] La 'Pataphysique est la science. . . .

But Thorpe's selection of quotations do less than justice to their authors. He himself indicates, in a passage included in a note (a device he justifies by explaining somewhat ingenuously that the passage originally appeared in one of Greg's footnotes: but when Greg made his comment it was a genuine aside, whereas it is central to Thorpe's argument) that by 1950 Greg had modified his claim made in 1932 for "the science of the transmission of literary documents" to the point that

he would rather rely "upon the impression of a critic in whose judgement I feel confidence than in an accumulation of mechanical tests."[8] The quotations from Bowers are nearly all modified by the use of single quotes, and significantly in the two most recent passages Thorpe quotes (1959 and 1964) Bowers correctly qualifies scientific by the word 'more.' A fairer and more useful picture might have been presented had Thorpe quoted Bowers's expression of preference for the taste of a Kittredge or an Alexander.

The most striking discrepancy between utterance and quotation is to be found in the use made of Housman. Thorpe writes:

Before going on to disappoint Thorpe's hopes, I would make one or two points about the selection of this quotation. First, this is not all Housman said. Immediately after stating what the name denoted — that is, in the very next sentence — Housman wrote: "But I must also say something about what it does and does not connote, what attributes it does and does not imply; because here also there are false impressions abroad." Housman went on to state, quite categorically: "Textual criticism is not a branch of mathematics, nor indeed an exact science at all."

Rather than quote selectively in order to support such an argument, it might have been more profitable to have considered why scholars of no mean distinction should have presented this apparent paradox. Might not Housman, a textual critic and creative writer of considerable

insight, be more sensitive to the relationship of textual criticism to science and art than Mr Thorpe's strictures suggest?

Earlier I quoted P. B. Medawar. I should now like to continue that quotation:

Now it will be apparent (as Medawar himself indicates) that philosophy, history, and literature are not natural phenomena. The part played by nature and art (or artifice) in the creative arts has been argued since Pindar, but the materials upon which bibliographers work are clearly to be distinguished from natural phenomena. In this most important sense, bibliography cannot be scientific. But, for all that, it is in certain important respects, of science. I disagree most with Thorpe when he says he "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'" (pp. 18-19).

It is not difficult to find in the writings of as distinguished a scientist as P. B. Medawar a description of the scientist which, in its generalized form, applies with equal validity to the textual critic:

In his essay "Two Conceptions of Science" in the same volume, Medawar discusses what seem to some people alternative, or even competing, accounts of that process which he argues as being "two successive and complementary episodes of thought that occur in every advance of scientific understanding" (p. 133). (Here he refers to K. R. Popper's critical analysis, "On the Sources of Knowledge and of Ignorance," in Conjectures and Refutations (1963), for the conception that truth is manifest.) All this strikes me as relevant to bibliography and the work of bibliographers. It is also (although I shall come to this later) an interesting passing comment on McKenzie's insistence on the hypothetico-deductive method, which seems to demand regimentation and book-keeping and to ignore imagination, creativity, and poetry.

But Medawar also presents the contrary case. In his introductory essay to the whole collection of essays, The Art of the Soluble, he quotes from Arthur Koestler's The Act of Creation, which he had reviewed:

"Good scientists," Medawar goes on, "study the most important problems they think they can solve." Whether this is so or not, it is very different from the situation that faces the textual critic. Later in this collection of essays Medawar quotes from Introduction à l'Étude de la Médécin Expérimentale by Claude Bernard:[11] "if one proposes a hypothesis which experience cannot verify, one abandons the experimental method." Hypotheses must be of a kind that can be tested, says Medawar, following Bernard, and "one should go out of one's way to find means of refuting them" (p. 171). This accords exactly with the position taken up by McKenzie and it is not surprising, and perfectly logical, that McKenzie should argue, as Medawar does for scientists, that bibliographers "should normally proceed" by the hypothetico-deductive method which, whilst welcoming conjecture, yet faced with "but one piece of negative evidence, one contradictory occurrence" will conclusively falsify that hypothesis.[12] What McKenzie regards as normal he does not state, though the tone and rigour of his approach is such that it is difficult to conceive of abnormalities acceptable to him.

McKenzie is of course right to protest at the use of unexamined evidence. But, and it is a very significant 'but,' if science is concerned with the soluble, and the hypothetico-deductive method is its Method (even if, as Medawar correctly puts it, "rightness" that lies beyond the possibility of future criticism cannot be achieved by any scientific theory'[13]) is this applicable to textual criticism? Is textual criticism only concerned with the soluble in the sense that Bernard and Medawar imply? I think not. Kuhn states that "with scientific observation . . . the scientist can have no recourse above or beyond what he sees with his eyes and his instruments" (p. 113); can the textual critic be so limited? Must he not, by the very nature of his subject, venture beyond what is before his eyes? Can the relationship of the texts of Hamlet be verified by experience? Can the lost manuscripts that underlie those we have of the Oresteia or of the Epistles of St. Paul be restored and tested in a way that is scientifically acceptable? The answer must be 'No.' But the non-scientific nature of the problem is even more striking in those cases where we may have most, or even all, the texts. Can we scientifically demonstrate an edition of The Prelude — an ideal text of the kind, say, proposed by De Sélincourt?[14] Will hypothetico-deductive reasoning and tests for falsification produce a text for Murder in the Cathedral, complicated as it is by conflicting authorial versions?[15] Will we not in these cases have to go beyond what we can see with eyes and instruments? Contradictory occurrences are so prolific in these texts that the temptation to turn to enumerative bibliography is very strong.[16]

But is not the temptation to run away based on a misunderstanding of the nature of bibliographic studies? Scientific research may be based solely on the art of the soluble but it should be obvious that bibliography poses problems that are not soluble in the Bernard-Medawar sense. It is not inferior because of that: but it is different. McKenzie seems to be implying that although bibliography is not a science, a specific scientific method — the hypothetico-deductive method — is essential to bibliographic studies and, where it cannot be applied, these studies should not be pursued. (There is here, I think, a discrepancy between his opening in which, among other things, he seems to accept Bald's conclusion that bibliography cannot claim the same validity as is appropriate to the exact sciences (pp. 2-3) and the tenor and conclusion of most of what follows.) Furthermore, McKenzie, in his advocacy of the hypothetico-deductive method seems under the impression that there is a scientific method, and that this is it and, indeed that science itself is all of one kind. Medawar has argued, far more authoritatively than can I, how erroneous it is to imagine that there is any such thing as a Scientific Method.

Pointing out the manifold errors of our colleagues, salutary though it is for them, does not, of itself, demonstrate the efficacy of the hypothetico-deductive method. McKenzie's strictures point to no more than the invalidity of specific arguments and they have less to do with general principles than he, or those such as the reviewer of his paper in the T.L.S., 22nd May 1969, imagine. Perhaps I might take an editorial

problem, where all the evidence is available, and apply the hypothetico-deductive method. I have chosen an instance in which the editor is demonstrably wrong, in which the problem is delightfully simple, in which 'the Method' (McKenzie will be our Stanislavsky) proves sterile, and in which, in the end, depends for an answer on the editor's taste and judgement.

The problem concerns the notes to be played by the timpani in two places in Bizet's Symphony in C. I refer to the Edition Eulenburg, edited by Hans-Hubert Schönzeler; it is undated but the Editorial Comment is signed from London, June 1961. The text is in English and German and I refer only to the former.

The editor makes what to most contemporary textual critics will seem a curious statement about his editorial procedure:

Are we to assume that the autograph was only looked at when there was a discrepancy between Choudens and Philharmonia? What standing have these editions? Will this procedure result in the Eulenburg edition following the autograph in all details — surely not?

Later in the Editorial Comment the editor refers to the special problem posed by the timpani part. Choudens and Philharmonia vary widely from the autograph and the editor has "as in all other respects," considered the autograph as final.

There appears, however, to be "a mistake on Bizet's part in the first movement." The autograph unmistakably shows that the timpani are to play D two bars before No 5 and in the fourth and fifth bars after No 6. The editor concludes: "As it has always been the policy of Edition Eulenburg to adhere to the letter of the autograph, these doubtful timpani notes have also been retained in the present score." Quite apart from what, in the retention of what the editor himself describes as a mistake, amounts to an abrogation of the editorial function, is this, in fact, what happens?

Two bars before figure five they are certainly given D to play in the Edition Eulenburg. As the cellos, basses, and horns are all playing concert D, it is possible that, if the timpani is to play, it should be D, in which case there is no mistake — though retuning would be necessary. But it could be we have here our old friend, "attraction to a word or letter previously copied."[16a] However, at the fourth and fifth bars after figure six, where in the autograph the "notes unmistakably are D," we find that the timpanist is given C to play (as in the second and third bars). Here also the cellos, basses and horns play concert D for all four bars and it could be that a discord is intended — and is required before figure five also.

Clearly the editor has not done what he said he would do (even if that is what he should have done). Obviously D should not appear in one of these places and C at another — but which note ought the timpani to be playing, D or C? Will the hypothetico-deductive method help? I doubt it. If one hypothesises C it can be argued that the autograph clearly has D, and that is certainly "one contradictory occurrence;" if one hypothesises D, one can argue that the part is clearly marked in C and G and there is no indication "muta in D."[17] If one hypothesised that the notes should not be played at all, apart from the fact that notes are written down, one would have to explain the appearance of notes in the second and third bars after figure six. (From five onwards there is a reduction from ff to f which might account for the cessation of the timpani at the sixth bar). Both notes played simultaneously is possible but extremely unlikely — and that, too, would require retuning. A variety of hypotheses can be advanced but, even though we have a holograph, none can be 'proved true.' Now this is not to suggest that the hypothetico-deductive method is without its uses here. It can help us to discriminate between hypotheses, but it cannot itself provide an answer.

It may be that the deliberately heavy weather I have made of this little crux will illustrate the limitations of the application to humanistic problems of a technique appropriate to science or logic. Demonstrating Schönzeler's error (as McKenzie has done for many of our colleagues) will not itself solve the crux. Suggesting that the hypothetico-deductive

method is negative rather than positive certainly does not mean that bibliographers and editors can throw logic to the winds. McKenzie's strictures are necessary: but they must be seen in proportion.

In practice editors, or those acting in that capacity, have to provide answers even if evidence is insufficient or contradictory. Thus, to remain in the sphere of music, Ansermet, Karajan, and Munch include the trumpet fanfares which Debussy omitted from the second edition of La Mer because he was hurt by the suggestion that they were reminiscent of those in Puccini's Manon Lescaut. Some conductors, including Toscanini and Bernstein, follow the second edition. Which approach is truer to the music? If one adopts the composer's second thoughts as being, after all, his final (even if mistaken) decision, how will one be placed in presenting Bruckner's Second Symphony at the point the composer substituted a clarinet for a horn solo, though there is no doubt the horn is much more effective — especially if it is correct that the substitution was only made because at the rehearsal for the first performance the horn player could not manage his solo? It is not responsible, if one cannot 'prove' a solution in such cases, to turn to enumerative bibliography. That will hardly serve the best interests of the music — which is surely its performance.

It is sometimes argued that if an editor provides all the evidence, as the editors of the Ohio Browning set out to do, the reader can be left to make decisions for himself. It does not matter greatly, it can be argued, which text is given 'above the line' provided that, below it, a full register of variants is supplied. Quite apart from the effect that differences of typography between main text and notes have on the reader, the invalidity of this argument is at once apparent if performance is involved. One cannot perform simultaneously several variant readings of music or a play without doing serious damage to the author's creation. (A note omitted in one version and replaced by a rest but included in another version would present a particularly intriguing problem in such simultaneous performance!) The duty of an editor can surely not be less than that of a conductor or a producer and he ought to be in a much better position to make a judgement.

If the hypothetico-deductive approach cannot solve this little crux in Bizet, how will it resolve the problems posed by Piers Plowman, or Hamlet, or even Murder in the Cathedral, where we have all the evidence? The short answer is that it cannot be more than a useful tool which may help us avoid the avoidable. Thus, in practice, one often has to choose between various courses, none wholly satisfactory, and

the hypothetico-deductive method is a convenient means of testing the choices open to an editor, helping him to decide to which choice he should give preference.

I am not implying a weakness in the hypothetico-deductive method but simply pointing to its relative inadequacy in dealing with the insoluble — that is in a field for which it is not designed. Thomas Kuhn, however, does argue that the hypothetico-deductive method is limited in use in scientific research itself. He concludes that "If any and every failure to fit were ground for theory rejection, all theories ought to be rejected at all times." It is the very "incompleteness and imperfection of the existing data-theory fit that, at any time, define many of the puzzles that characterize normal science." If only

It is, of course, verification theories that appeal most to bibliographers (and least to Dr. McKenzie) and especially that aspect of verification that asks, in Kuhn's words, "not whether a theory has been verified but rather about its probability in the light of the evidence that actually exists" (144). Later he argues:

The best procedure, as I see it, in the practical dealing with texts, is a combination of probabilistic theory and falsification tests, endeavouring, where the hard evidence permits, to ensuring that this is subjected to a process of rigorous logic and being fully conscious of the moment when one is departing from the demonstrable and falsifiable, and into the realm of the subjective and unverifiable.

One cannot, as a bibliographer, avoid sharing a certain fellow feeling with Bishop Butler when he says in the Introduction to The Analogy of Religion (1736), "but to us, probability is the very guide of life." If this seems inappropriate — perhaps even immoral — in so scientific an age, it might be worth noting that this is a position akin to that adopted by the great physicist Werner Heisenberg, the author of "The Principle of Uncertainty:"

That describes, for me, exactly the position in which a bibliographer and an editor is likely to find himself. It does not mean logic can be disregarded; it does not licence woolliness; but it does recognize the realities of bibliography and editing in a way which, with all respect to Dr. McKenzie's learning, I think he fails to do.

III

I should now like to consider the larger matter of which McKenzie's and Thorpe's arguments may be symptomatic: the possibility that the current paradigm in bibliographic studies is no longer acceptable. They are not alone in expressing dissatisfaction with conventional bibliographic and editorial practices. The editors of The Complete Works of Robert Browning [19] reject the conventional meanings of 'text' and 'author' and argue that punctuational changes ought not to be differentiated from substantival variants, or, to put it in their own language:

I have always taught that substantive variations may include punctuation. Thus, in The Dutch Courtesan, the 1605 corrected Quarto reads:

(The uncorrected quarto reads "to worship;") Martin Wine, in the Regents edition, 1965, takes this to mean: The Fountainwell edition, 1968, which rather unsatisfactorily partially modernizes (so 'to' means 'too') reads: Whatever the correct reading (and as that is not my concern here, reference should be made to the arguments in the respective editions), punctuation certainly affects meaning and the effect is substantival. The reader of Marlowe will not need reminding of the substantival effect on Edward II of an unpointed line: there was nothing accidental about Young Mortimer's omission.

But even if one accepts such punctuational variants as substantival, this is not an open invitation to include all accidental variants as substantives. Even if the argument of a continuum is acceptable (and the word is overused by the Ohio editors, being applied to authorial, editorial, and semantic functions), if given effect it means readers will be faced with a welter of variants, from which, even when presented with the typographic elegance of this edition, it is extremely difficult to distinguish the significant from the insignificant. But, more seriously, are not the Ohio editors abrogating their editorial duty even if these poems do not, like music or a play, have to be performed? Can editorial responsibility be argued away by a continuum of semantic niceties (or, perhaps, by an anxiety to avoid assuming god-like pretentions)?

The Ohio editors are right, I think, to be uneasy about the conventional meanings of 'text' and 'author', and they are very properly much alive to the significance of the least comma in Browning's work. Nevertheless, their analysis of the problem posed is confused and their solution is unsatisfactory. One senses, perhaps unfairly, that, having decided to abandon the variorum edition originally advertised, they settled on a policy of reprinting the final edition published in Browning's lifetime, supporting it with a register of all variants known to them, and that they then looked round for a policy to justify this procedure. The choice of such a text is unhappily reminiscent of a practice castigated by Housman in the preface to his Juvenal, 1905:

We seem, indeed, to be back with the idea of a codex optimus and one feels that in the Ohio Browning inadequate attention has been

paid to the readings of all the significant editions at the points at which they diverge.[21] On top of that there is a failure to differentiate between the authorial and editorial functions. In: and: (to which one could add friends such as Joseph Milsand (xii)), there is surely far more confusion than discrimination. What the logic of their argument for a continuum demanded was a genetic text like that produced by Harrison Hayford and Merton M. Sealts Jr. of Billy Budd in 1962. There is, of course, no reason why editors should have to produce a variorum, a critical, or a genetic edition if none of these will serve their purpose, but what the Ohio editors have produced is not, I think, the happiest solution to their problem.

Nevertheless, criticism about the way the Ohio editors interpret 'text' and 'editor' should not lead us to assume that the conventional meanings are without problems. The Ohio editors have touched on a matter of some concern.

E.A.J. Honigmann has recently argued against the idea that there is but a single version of a Shakespearean text:

This is surely correct and not only for Shakespeare, as Honigmann shows. The sort of problem this poses for an editor may be represented by extracts from the 1869 and 1875 editions of Culture and Anarchy and J. Dover Wilson's treatment of one of these passages in his edition of 1932. These changes were made by Arnold himself, in order, Dover Wilson suggests, to suppress or tone down personal allusions or because the allusions were no longer topical. Whilst the two readings are not the result of the kind of instability Honigmann describes, the problem facing the editor is similar. As Dover Wilson prints the text it appears at one point as: In the notes to this passage[23] Dover Wilson explains: Dover Wilson therefore gives us a version which Arnold never authorised in this form. Whether it is acceptable is another matter; it is certainly convenient but obviously it can only work if the words changed are complementary — one cannot have the notes C and D simultaneously in the Bizet example, it will be recalled. Honigmann's solution, which is no more in accord with the hypothetico-deductive approach than is Dover Wilson's, is quite different:

IV

The problem is much more difficult in the case of such works as Piers Plowman, The Prelude, and Murder in the Cathedral. These all challenge very strongly the concept of a single but lost text which the editor must seek to restore. This has implications of considerable importance for all textual critics and especially for those who can take their work no further back than an archetype which may be several removes from the author's original, which, of course, may not have ever existed in a single finalised form. Indeed, even if there is an absence of contamination, I am inclined to doubt how securely one can go back beyond the exclusive common ancestor (or hyparchetype). It seems to me increasingly doubtful to what extent one can relate one exclusive common ancestor to another in order to find their common ancestor (or archetype). The existence of more than one finalised text of so many works must make one wonder whether the various exclusive common ancestors do not descend from stages of these and that one is setting about doing what Dover Wilson has done with Culture and Anarchy — but at several removes and unaware that it is this kind of problem that one is tackling.

Even in what might at first sight seem to be the most certain cases of single-text sources — private letters written before the age of the typewriter and carbon paper — we can be deceived. Many examples spring to mind, but one will suffice. Thomas More's letter to Henry VIII of 5th March 1534 also exists, as George F. Warner notes, "in More's hand with a few verbal differences, in the Public Record Office" as well as in the British Museum, the version he reproduces.[24] And, of course, letters might be copied after the author's death. Drake's letter to John Foxe included in a recent collection[25] is dated about 1615 by the editors and is most certainly not in its author's hand. To what extent do, say, the letters of St Paul go back to versions made, even by their author (or authors) for communities other than those to which they were specifically addressed?

In no way is the paradigm accepted by textual critics more seriously in question than that of the modern theory of stemmatics. The nature of this acceptance is to be seen in the approval given to Paul Maas's

Textual Criticism.[26] Fredson Bowers begins his Bibliography and Textual Criticism by remarking: "The general procedures of textual criticism as it deals with manuscript study have been formulated for some years." There is then appended in a footnote: "The most convenient summary of the principles for dealing with manuscript texts is Paul Maas, Textual Criticism." Bowers goes on to suggest that the editor of a classical or medieval text "can attack the problem from a position of strength" and he concludes his paragraph by saying: These traditional methods do not, however, receive universal approval. H. J. Chaytor (who does not mention Maas) summarizing the work of Bedier, Dom Quentin, et al., in 1945 said: Earlier he had argued that a genealogy required that the approximate dates of the manuscripts must be established and went on: And L. D. Reynolds and N. G. Wilson also point to the limitations of the stemmatic method:

The crucial weakness of the stemmatic method is, of course, its inability to deal with texts suffering from contamination. Maas certainly mentions contamination, stating quite clearly (paragraph 6) that his method is based on the assumption "that the copies made since the primary split in the tradition each reproduce one exemplar only, i.e. that no scribe has combined several exemplars (contaminatio)." And in paragraph 10 he states that if contamination has occurred, eliminatio "is greatly hindered, if not made impossible." And in the very last sentence (before the retrospect of 1956 which does not mention contamination) he states: "No specific has yet been discovered against contamination."

There seems little doubt that contamination was extensive in classical, Biblical, and medieval texts. Kane found that so much convergent variation had taken place in the A manuscripts of Piers Plowman that random groupings were extremely numerous and that it was difficult, if not impossible to show the whole genetic relationship of these manuscripts. Among the authorities he quotes, one is of particular interest in connexion with what I have to say about textual studies, science, and the rejection of paradigms.[29] According to E. Vinaver, writing in 1939:

As a gesture to Mr. Thorpe, perhaps I should not omit the last sentence Kane quotes from Vinaver: "Nothing has done more to raise textual criticism to the position of a science than the realisation of the inadequacy of the old methods of editing."[30]

In recent years some of those of us in the Departments of Greek, Theology, and English in the University of Birmingham who are interested in textual studies[31] have organised a joint seminar for our undergraduates to enable them to gain some insight into kinds of textual problems that they would not normally encounter in their main subjects. It is a modest enough scheme but it includes some account of problems presented by Aeschylus, the New Testament, the early Christian fathers, Piers Plowman, Elizabethan texts, The Prelude, and one or two modern authors such as Joyce and Eliot. What has proved most striking to those of us conducting the seminars has been the unanimity with which, quite independently, we have come to the conclusion that, because of contamination, the stemmatic approach is only of limited use — and then, as Kane suggests, often in a negative fashion:

In this area of textual studies it does seem that the old paradigm, represented by Maas's Textual Criticism, is no longer acceptable and one must evolve a new approach. For Thomson this has centred on the classification of errors.[33] This is closely related to "the identification of typical scribal substitutions" which, in Kane's view, "is, in default of recension, the main resource of the editor of these manuscripts." Kane's aim is "to establish presumption of originality among available readings, or, less often, to reconstruct the original reading from the variants."[34]

Kane's method may be illustrated from this example:

It will be apparent that such an approach is hardly likely to stand up to a Popperian falsification test; nor may it be verified, except subjectively. Its appeal can only be on the grounds of probability: that one hypothesis is, in the light of the knowledge of the text, language, literature, and ideas of the period, and of the scribes' involvement in what they were doing, less unacceptable than another hypothesis. The hypothetico-deductive method can play a part in discriminating between hypotheses, but ultimately, as Bowers puts it, one has to prefer the taste of a Kittredge, an Alexander, or, in this case, a Kane, to a quasi-systematic, quasi-objective approach.

The one element that Kane seems to take too little into account (in the light of what Honigmann has since written on the stability of texts) is the possibility of there being more than one authorial reading. This seems at least a possibility in the case of the example just quoted. Kane states that in Piers Plowman A he knows of no instance when it can be shown conclusively that because of authorial revision several variants are original (p. 147 fn 2). Of course, it cannot be shown conclusively: but then neither can Kane's editorial procedure prove it conclusively.

V

This brief account of Kane's approach, and the discussion that led to it, may have indicated not only how the old paradigm has given way to a new one in English textual studies but that in a limited way the experience of the old paradigm can be put to use, if in a very different manner from that originally intended (I refer to Kane's use of groups of manuscripts, not to create a tree but to limit support for a reading to the exclusive common ancestor of the group). But it is in that area of textual studies more generally called bibliography that McKenzie launched his attack. Is the current paradigm generally unacceptable here?[35] Have the puzzles become anomalies? The short answer,

so far as I am concerned, is that I do not know. I suspect that the answer is 'no,' or rather 'not quite yet.' Nevertheless I feel concerned enough at the possibility of a shift in our paradigm to prepare the students I teach not only to understand current bibliographic practice but to try to prepare them to respond to a new paradigm.

Of the need for there to be a paradigm, even if its validity is uncertain, I am convinced. A paradigm does not need to be wholly successful:

The most striking example I know of the need being felt for a paradigm when it seems inadequate is in Herman Bondi's response to Heinrich Olbers's paradox that the sky at night, though apparently dark, ought to be light. My concern is not with this paradox but in Bondi's response.[37] Bondi showed that all Olbers's conclusions followed from his assumptions and he then argued that in consequence one of the assumptions must be wrong. Which he chose and why need not trouble us. What is of interest in this context is what Bondi accepted: that the known laws of terrestrial physics (i.e. the paradigm) must apply. To Bondi that assumption could not be usefully rejected since the system of terrestrial physics was all that was known and to abandon it required giving up the whole inquiry. A passage in Bowers's Bibliography and Textual Criticism is relevant here:

It is characteristic of scholars that before taking decisions they require to have "all the evidence." It is this that especially distinguishes the scholar in the pursuit of his studies from the man who must base his decisions upon calculated risk. The bibliographer's and the editor's dilemma is that temperamentally he will think as a scholar but in his actions he will, in Werner Heisenberg's words, "always have to act on insufficient evidence." It is simply not possible to wait until one has all the information before one comes to a conclusion in bibliography or editing. One might wait for ever and attempt nothing. Even when all the evidence is available an 'answer' may still not be scientifically provable. It is exactly as Heisenberg puts it in speaking of life itself: "Even the most important decisions in life must always contain the inevitable element of irrationality. The decision itself is necessary. . . ." The bibliographer and the editor have to realise that their kind of scholarship lies not solely in the collection and arrangement of information (vitally important though that obviously is), nor even in the scientific demonstration and proof of what they have decided (desirable though that is) but in the resolution of problems that are humanistic, not scientific or enumerative. One must, as it were, find the narrow way between Popeian eclecticism and Popperian hypothetico-deduction.[38]

At its highest pitch the work of the bibliographer and the editor is creative. It is not creative in the same way that the original author was creative (and I am not arguing for the continuum of creativity favoured by the Ohio editors of Browning, therefore). Whereas the author, given a certain romantic disposition, can argue that his audience is solely himself, or, hopefully, that it is for all time, the editor's very justification is that he acts in response to the needs, general and scholarly, of his own society, bringing his author's work before the editor's society so that it may be enabled to respond to it. At its simplest this may be no more than a matter of typography (and rarely the editor's responsibility); at a rather higher level it is the provision of matter explicatory and critical; but at its most significant, it is the bringing to bear upon the text of all the scholarship (bibliographical, historical, literary and linguistic) and all the intellectual and aesthetic

insight characteristic of the editor and his age, without being restricted to that age or inhibited by any aspect of it.

It was the new awareness of science and man which developed in the nineteenth century (and which can be seen in the great creative writers as well as the scientists of the time) which came to be applied to textual studies in English literature from the time McKerrow and Greg met at Cambridge in the 1890s. In the work of editors of the twentieth century one can see, covert and unrealised, or explicit and declared, the scientific and socio-realistic concerns of 'the new bibliography' brought to bear on the textual problems posed by the texts they edited. The response to the spirit motivating the understanding of man in society which influences the creative writing of, say, a Zola or a Shaw, or even a Lawrence or a Joyce, influences also that aspect of textual studies which seeks to discover what happened to texts in the societies which produced and transmitted them. This is particularly obvious in the way Kane takes account of the personal involvement of scribes in what they copied and the way this affected the texts that have come down to us.[39] It is this that has led to so many investigations into, say, the Stationers' Company, or Henslowe, and it is precisely this that underlies the second question Bowers poses for editors.[40]

VI

To what in our agonised society should the textual critic now be responding? It occurs to me that McKenzie, in his concern to import into bibliographic studies an approach appropriate to science and Thorpe, in his desire to dissociate bibliography from science, have in mind a kind of science which is only partially relevant to our world as it is now understood. One can see the same sort of thing in one of the most interesting critical studies of the last decade, George Steiner's The Death of Tragedy.[41] In part of his argument he suggests that: "When the new world picture of reason usurped the place of the old tradition in the course of the seventeenth century, the English theatre entered its long decline." Later on he argues: "The myths which have prevailed since Descartes and Newton are myths of reason, no truer perhaps than those which preceded them, but less responsive to the claims of art."

My concern here is not whether the conclusions which Steiner

draws are sound (though I believe these do not necessarily follow from his analysis, brilliant though that analysis is), but of the change since Einstein, Heisenberg, and Gödel pronounced The Special Theory of Relativity, The Principle of Uncertainty, and the Theory of Incompleteness, respectively. No longer is the world within the atom to be understood in terms of the laws of cause and effect, nor that universe beyond our world as ruled by time as marked at Greenwich. If within the limits of atom and universe our world works well enough according to Newtonian physics it is only with an awareness that this is only a partial explanation of physical being.

Whether directly or not, creative writers have responded, since Strindberg wrote The Dream Play a year or two before Einstein published his theory, to this new understanding of the nature of our world. It is surely not mere coincidence that those in the forefront of physics and mathematics, and those in the avantgarde in the arts, are concerned in their very different ways and to very different purposes, with uncertainty, incompleteness and irrationality? In this at least, and most excitingly, science and art are related. If we do sense that we are at a point of crisis in bibliographic studies — that our paradigm is inadequate to cope with the anomalies with which it is faced — then perhaps it might repay us to take note of these changes in the physical explanation of our world and the response of creative writers thereto.

What we could find is that the more precise techniques developed by 'the school of Bowers and Hinman' (if I may use such an expression) are to us not unlike what Newtonian physics is to scientists, but that outside the usefulness of these methods (which are, after all, rather extensive) we ought not to be afraid of irrationality and infinite

coincidence. Or, to put it more conventionally, imagination and taste. Thus might we still be able to respond to the question Erasmus asked of Martin van Dorp: "What are we to say when we see that the exemplars of this edition do not agree?"[43] and we may, in Ben Jonson's words, in an aptly named work, "doe a welcome worke yet to helpe posterity to judge rightly of the old.",[44]

Notes