University of Virginia Library

An Inquiry into the Social Status of Texts and Modes of Textual Criticism
by
Peter L. Shillingsburg

The main purpose of this paper is to ask questions about what an editor owes to the "social contract" when preparing a scholarly edition. A survey of specific issues in editorial debate preceding these questions can be found in Textual Criticism since Greg, A Chronicle, 1950-1985 by G. Thomas Tanselle.[1] In what follows I assume readers to some extent familiar with the history of that debate through the early 1980's as it applies to editing modern works. A historical background is important because various spokespersons for new social theories of textual criticism claim to be proponents of a "new historicism" replacing the principles of New Criticism underlying much textual criticism of the 1960s and 70s.[2] But many spokespersons for traditional historical criticism claim never to have lost track of history and think the new historicism is valid primarily because it returns to old principles, rejecting in the process some of the "aberrations" of recent critical theories, deconstruction particularly, in which history, including the author, seems to have disappeared.

In dealing with the history of textual criticism and the newer ideas, I mention theorists by name occasionally, but I am not trying to survey any particular person's ideas; rather I am attempting to sort out different possible ways of looking at and using evidence in textual criticism—particularly along lines that might be regarded as a social theory of texts. Only God does not need to learn from experience (though my wife points out Eve was made second), and many writers to whom I refer, especially those exploring ideas of social influence, have changed their minds or modified their theories. I daresay they will again. As Morse Peckham has recently observed, "we have after all learned a great deal about something that has turned out to be a more difficult and confusing problem than was originally thought."[3]

I

The roots of modern textual criticism lie in classical and biblical textual scholarship which generally speaking assumes an ur-text, most often no longer extant. The editorial task in such situations is to reconstruct this ur-text, which is usually conceived to have been single and completed. If variant readings occur in competing modern editions of classical or biblical works, they usually reflect the difficulty editors have in sifting the evidence of extant derivative source texts and their differing assessments of the importance of

specific problems. The differences seldom represent disagreements about the aim or purpose of editing, which is to recover or reconstruct the archetype or ur-text.

When editors turned to textual criticism of modern works for which authorized texts were extant and for which even authorial materials such as manuscripts survived, the idea of an archetype or ideal text was not easily discarded. The comparative richness and authenticity of materials merely made it seem more likely that success could be achieved in editorial work. The concept of definitive texts seemed plausible.

Generally speaking, the evolving editorial principles in the mid-twentieth century all assume that the literary work was the product of an author whose wishes concerning the text were to be paramount, and that the work should be edited in such a way as to produce the "established" text. The purpose of critical apparatuses was to show what the editor had done to produce the "established" text and to show what historical authorial forms of the text had been used as a basis for establishing the new text.[4]

The editorial discussions of the twenty years preceding 1980 did not seriously question these assumptions, concentrating instead on issues related to "final intentions" (exploring distinctions between artistic and commercial intentions and debating whether the terms "final intentions" and "last intentions" were synonymous),[5] and on issues related to the treatment of accidentals (exploring distinctions between purely formal elements and "semi-substantives"[6]).

Some potentially interesting questions were raised about fulfilling authorial expectations (as opposed to authorial intentions), but even these did not clearly challenge the underlying assumption that the work belonged to the author and that the task of the editor was to serve the author. Most of the arguments on the side of serving the reader (objections to "barbed wire" editions, encouragement of modernized editions)[7] were dismissed as corrupting influences that pandered to popular audiences. Only one "reader service" principle has found general acceptance in American scholarly editing—the clear reading text, uninterrupted by footnotes or note indicators in the text. It is probable that this principle reflects more than anything else the result of belief in the established definitive text, the recovery of Pure Virgin Text.[8]

A few relatively serious but largely unsuccessful attempts were made in that period to legitimize the role of publishing and production crews in the eyes of scholarly editors. That is, there were some attempts to show that secretaries, editors, compositors, and advisors often enough gave "good" advice and that sometimes their good effects had to be carried out in spite of the author's wishes or, in the case of absent or dead authors, without any indication of the author's wishes. One of the first such efforts, by James Thorpe, defined the work of art as merely "potential" until it was published.[9] The effect of this definition was to acknowledge the production process both as an aid to the author in finalizing the work and as a necessary fact of life enabling works of art to be known to readers. To support the first of these propositions, Thorpe argued from specific cases in which the author

received needed help for which he was grateful.[10] The second proposition received its best support from Donald Pizer in a series of objections to the editing of texts (particularly Stephen Crane's Red Badge of Courage by Henry Binder and the Pennsylvania Edition of Dreiser's Sister Carrie) in which the pursuit of purely authorial forms produced a text so different from the "cultural" artifact that had been known so long that, Pizer thought, serious confusion resulted about what the "real" text was and what it meant.[11] Philip Gaskell generated a considerable amount of press for Thorpe's basic premise in the editorial advice appended to his New Introduction to Bibliography and in his discussion of textual problems in Writer to Reader. Nevertheless, these editorial arguments remained author-centric, for they argue from the belief that the production process was an extension of authorial intention and merely fulfilled expectations for improvement of the work.

Discussion of the "critical" nature of editing did, however, dethrone the concept of definitive texts. Its most visible manifestation was in the emblem of approval designed for CEAA sponsored editions, which reads "An Approved Text." The point was made that the emblem did not say "The Approved Text." To emphasize the point, the Committee on Scholarly Editions, the Modern Language Association's replacement for the CEAA, redesigned the emblem to say "An Approved Edition," a label meant to indicate that the editorial work was definitive, though the text itself was not necessarily so. However, the arguments about critical editions and about what part of them was or was not definitive did not go beyond an acknowledgement that different editors attempting to produce a text that best represented the author's intentions were confronted from time to time with inconclusive evidence and might exercise their judgment differently in those instances.[12] Whatever else they might disagree about, G. Thomas Tanselle wrote, scholarly editors were in general agreement that the aim of editing "is to discover exactly what an author wrote and to determine what form of his work he wished his public to have."[13]

Most editors are well aware of the difficulties the idea of intention represents. Much of editorial debate, as well as the generation of new approaches to editing, is fueled by the problem of identifying intention and of using the concept responsibly in practical ways. Some editors are attracted by theories which attempt to avoid the problem of intention altogether.[14] It is not my purpose to rehearse any of these arguments. Nor do I mean to imply, in quoting Tanselle's remark about the general agreement among editors about the aim of editing (that it is to recover the author's intentions), that he was unaware of the underlying problem; indeed, he has written the most detailed discussion of it.[15]

II

In the 1980s the tenor and range of editorial discussion suddenly changed and broadened. The changes have taken three primary modes, but all three derive impetus from the newer critical theories and particularly from new attention to the act of reading as well as social theories of communication.

Morse Peckham had tried something similar in the early 70s with a behavioral model of communication, but it had little influence in the editing community.[16] The three new modes have certain things in common, but the differences lead to radically different editorial results. Two of the modes remain author-centric; the third is socio-centric. Ultimately, it is upon this third, the social view, that I wish to focus.

The first of these new modes dubbed itself The New Scholarship; its most frequent spokesman has been Hershel Parker.[17] The reading and editing principle for this approach is to see the work in the context of its creation in order to best understand what the author was trying to do and to see that effort in relationship to what the author did do—the text which he produced. Parker questions a basic assumption of the New Critics, that works of art are unified and make sense and that the critic's job is to discover and reveal that unity and sense. Exploring examples of works in which he believes the author did not fulfil his own intentions, or where a decently fulfilled intention was marred by the author's own later revisions, Parker dethrones the text as the verbal icon. He argues that textual criticism pursued with the diligence and methods of "The New Scholarship," can reveal flawed texts with "adventitious" meanings resulting, not just from corruptions imposed by editors, but from the author's failure to embody a coherent intention in the produced text. While this approach is wide-ranging in its search for evidence of all sorts that may have influenced the creative process and while it is equally wide-ranging in its awareness of influences on reading and interpreting the work, it remains an author-centered approach. As such it is primarily a refinement of the more traditional (CEAA) modes, the primary difference being that it eschews "final intentions" per se as a touchstone for determining the authenticity of the text. Often, in the examples Parker provides, initial intention is preferred to final intentions, since by the time of "final" intentions, Parker says, the author may have ceased the creative mode and adopted an editorial mode.[18] The approach does bear a similarity to the other two new editorial modes in that it emphasizes a concept of the author in a social continuum that includes economics, politics, the psychology of creativity, and book production. It represents, however, a refinement of an authorial concept of authority, not a dismissal of it.[19]

The second new editorial mode reacts strongly against that "single adequate established text" idea. The "multiple text" editors are not really new, but much new editorial debate emanates from their point of view. It is represented by Hans Gabler, editor of Joyce's Ulysses, and has roots in German editorial traditions as expressed early on by Hans Zeller.[20] I have adopted a good deal of this approach in my own editing of W. M. Thackeray. Its primary difference from previous editorial principles is its insistence on multiple or fluid texts. Authorial intention is recognized as less frequently monolithic than it is developing or changing—not only in a continuous progression but reaching stages of supposed completion or being transformed by conflicting or mutually exclusive intentions. In theory no particular preference is given

to the final-intention text, as was the stated goal of CEAA editions, or to the text best representing the author's creative involvement with the work, as advocated by Parker in Flawed Texts. This position insists that the editorial aim is to present the multiple authorial texts—representative of developing or changed intentions—in such a way that the work may be read radially, each version in relationship to its other manifestations. Each authorial version of the work adds to the whole critical perception of what the author was doing and therefore of what the work means. Further, it allows for the view that a new authorial intention does not deprive an original or intermediate intention of any "authority"—it is merely different; all authorial intentions may be equally authoritative.

It is much easier to think of the multiple-text concept in theory than it is to construct a multiple-text edition. "Everybody knows" texts are linear, not multiple, and "everyone" seems to want a text to be comfortable with.[21] But comfort and the practical problem of how to present the work of art are secondary to the significant issue here—that, except for unrevised texts, a single text does not adequately represent the work for scholarly study and that no single text can best represent the work for a critical reader—other single texts may be equally successful or useful.

There have been several efforts to make practical multiple-text editions. In Hans Gabler's edition of Ulysses [22] one can see the "synoptic text" on left-hand pages tracing the development of authorial intention through a sequence of variant texts preceding the final fair copy used to set the first edition. The synoptic text superimposes all the preceding texts on one page through an elaborate, but decipherable, coding system. On the facing page the reader sees a linear text (clear reading text) representing the final manuscript version. The reader is encouraged to read the right-hand linear page in tandem with the synoptic or radial left-hand page. Michael J. Warren's Complete King Lear reprints two basic variant texts in parallel on facing pages with further variant versions indicated in marginalia. The idea is that the editor's critical preference, if he has one, is submerged in the arrangement of competing documentary textual forms. That the linear text is a limitation which scholarly editions can, to some extent, overcome was clearly apparent from the many iterations of the idea at the 1987 meeting of the Society for Textual Scholarship in New York.[23]

As different as these various approaches to editing may seem, they have certain things in common. In each the work of art is seen as an authorial communication normally reaching its verbal form with minimal reference to book production, which is seen largely as an enabling process owing its allegiance to the author but frequently defaulting in that allegiance. Given this set of assumptions the critic is likely to see the interference of the publisher, editor and printer as legitimate only if it does not violate authorial intention or authorial practice. The production crew, from publisher to binder, is entrusted with translating the author's manuscript into print—or in more characteristic terms, entrusted with enabling the author's communication

to be seen by many in the form of a printed book. The production process is necessary but is always on notice that it may fail its handmaiden role.

III

The third new approach is sociological; it is both radically different from and blandly similar to traditional approaches. Its primary spokesmen are Jerome McGann and D. F. McKenzie, whose writings on the subject have generated a good deal of excitement and confusion. The idea that authors live, breathe and have artistic being in a social complex that includes more than the words surviving on pages of manuscripts and printed books is not new. Most editors consider themselves historical researchers—it goes with being a textual critic. It is the arguments of the social contract, not the new evidence, that make the theory worth considering. But the confusion of argument has been and may for some time continue to be a major block to its influence.

There are five and possibly six distinct arguments involved in the sociological approach. The first three are old and ultimately irrelevant if not contradictory; for years they have failed to influence the mainstream of scholarly editing. The fourth and fifth have grown out of critical theory and are sophisticated and well worth attention regardless of whether one adopts the conclusions of their present proponents. Most discussions of social texts, including the first three, have focused on the role of the production process in creating the work of art, but the concept is not limited to that subject, as the fourth and fifth arguments will demonstrate.

Earlier I mentioned attempts by Thorpe, Gaskell, and Pizer to legitimize the production process. It will help to begin with another look at the basic arguments they use, for the first three arguments are really theirs, though McGann occasionally uses them, too.

The first line of reason focuses on the idea that the production process is a necessary and desirable finishing or completion process in which the author turns to the publisher for help—and usually gets it. The argument is supported with examples of felicitous collaborations, all tending to show that the normal process of book production is a happy one.[24] I oversimplify somewhat, but the result of this argument is that, normally, the printed text is preferred as copy-text and, normally, the editor restores "pre-copy-text" forms only when demonstrable errors marred the process. Notice, however, that the justification in this argument is based on a critical assessment of the effect of changes. Regardless of whether the editor acknowledges it, such assessments depend on the aesthetic principles of the critic/editor. This is true even when the editor sets aside his own preferences and pursues what he takes to have been the author's. A basic premise of this argument is that the author's intention is only tentatively or provisionally recorded in the manuscript and that the collective experience and judgment of publishers and editors help the author fulfil his intentions—thus, authorial intention and editorial judgment combine for a better end product.

The counter argument, that editorial interference is detrimental introducing corruptions, is often laced with examples of the silly results of production errors or angry remarks by disgruntled authors.[25] It leads to Greg's rationale of copy-text, so that authorial forms will be more likely to survive without these "corruptions." Later printed forms are adopted only where they are demonstrably authorial or are more than plausible corrections of manuscript errors. Though many editors would want to deny it, the basis for this counter argument is also a critical assessment of the effect of production crews (proofreaders, editors, compositors, and/or any other persons who help or hinder in the production of the published work). Scholarly editors who recognize the critical bases for judgment tend to say that rules cannot be made and each textual situation should be judged on its own merits—after all, they say, what is critical editing if it is not critical?

A second argument for social texts, one used by both Donald Pizer and McGann, is very different because its logic denies the relevance of either authorial intention or critical judgment. The argument is that production and the resulting book is a part of history which has integrity of its own regardless of the author's or the readers' assessment of its success. Thus the fact that William Faulkner's Sartoris was published as it was becomes more important than might be the work that was to have been Flags in the Dust. In that particular case the enormous difference between the two versions, combined with the changed title, makes it easy to consider the two forms as completely different works for which it is silly to ask "which is the real X---." But when the difference is not so great and the name remains the same, as with Crane's Red Badge of Courage or Dreiser's Sister Carrie, the historical product, the first edition, has a validity or status that seems to some people to be threatened by editorial efforts to undo or redo the work of the original publication process. Hence the argument for the historical integrity of the first edition. In this argument many will recognize what has often been called documentary integrity. Historical integrity and documentary integrity need not, however, be the same thing. Tanselle, for example, describes the work of eclectic critical editors as historical, though they do not confine themselves to the readings of any single document. In addition, one might distinguish between the integrity of the document as it represents a historical event and its integrity as it is a physical unit connected to the author.[26] In the former the text of a document may be thought inviolable, while in the latter, the document itself is the important unit.

The third argument asserts that the published texts have become culturally validated by the decades or centuries of readers who have known the work in that form. Readers and critics have come to know and in many cases love the work in this socialized form. If an editor changes the text so that the work means something new or different, these people and their sense of the cultural heritage will be confused, if not absolutely violated. This is true even though the changes are "restorations" of previously unpublished authoritative forms, which perhaps should have been a part of the original work save for the interference of "unauthoritative" influences.[27]

Although, to my knowledge, literary editors do not use the second and third arguments (about the historicity and the cultural validation of texts) to support the idea that NO emendation should be undertaken, that is, strictly, the logical conclusion of both arguments. That a first edition is a historical document, that it was the basis for the work's first readers' acquaintance, and that it can still so serve new readers are unquestionable propositions. That an emended edition would misrepresent the integrity of the historical document is also true—though not necessarily important. It would be ludicrous, however, to expect all students of a text to read only the first edition or some other "authoritative" edition that had comparable legitimacy—such as "the author's revised edition." One could, with the argument concerning cultural validation, opt for editions that were straight reprints (unedited) with the idea that the linguistic text, more or less as it was first published, is what has been validated. But as soon as emendations are introduced, the center of this argument for "historical integrity" is weakened and, in fact, some other notion of integrity has replaced it.

Social contract theorists, or editorial theorists inclined to accept a social validation for texts, such as Thorpe, Pizer, McGann, and McKenzie, may have tended to attack the "Greg/Bowers" or authorial intention editorial theories in order to make room for their own. In a sense, editors are in competition for the attention of readers—each wants readers, scholars, and critics to think his edition is the "standard" edition. Sometimes, therefore, arguments are used against other propositions instead of for an alternative proposition. The idea of historical integrity and cultural validation may be primarily arguments against the authority of authorial intention rather than for a social theory of texts. It is unfortunate that these arguments are used in this way for it confuses the issue to argue that the production process is good because it helps (i.e., authors need help and textual critics can judge when that has happened) and that the production process freezes and or validates whatever came out (i.e., authors get what they get, sorry, judgment has nothing to do with it).

The fourth and fifth arguments for social texts are much more interesting and sophisticated and provide the proper ground for serious consideration of a social contract theory. The fourth argument focuses again on the production process. But rather than saying that editors and publishers help authors to implement their intentions in a physical book, or that there is an antiquarian integrity to the document containing the text, this approach sees the production process as a cultural phenomenon without which books do not exist. The influence of production on the book does not begin when the author hands a completed manuscript to the publisher; it begins when the author raises his pen for the first word of a work intended for publication, because of a consciousness of the way books gets published. Publishers, too, are only part of an even larger phenomenon that includes language and usage and everything that forms the sociological context within which authors are enabled to write and can hope to be understood. Any legal contract which the author signs with a publisher is, in this view, merely a confirmation of a

predetermined contract that exists, whether acknowledged or not, among authors, publishers, and readers. Publishers, therefore, are not primarily handmaidens to authorship exercising helpful servant roles, which they may fail to do well; they are, instead, part of the authoritative social complex that produces works of art. They are a fact of life that cannot be "edited out" of the text any more than the effects of an author's breakfast and subsequent indigestion can be edited out. There is no tangible support for this point of view. We are asked to acknowledge a higher order of historical determinism that operates regardless of individual intentions. We all know this in our bones, perhaps. We all see it working in "movements"—be they political, moral, social, or aesthetic. The author becomes the pawn of time—as we all are. If one does not feel this argument in his bones, I suppose it will not be very convincing.

There is a fundamental shift in the concept of authority involved in this argument. Rather than judging the effect of publication on the author's words, this argument simply vests authority for the text in a socio-economic environment which "contains" the author's initiating creative activity and the publisher's ongoing process of moving composition into production. The textual changes introduced by the production process are accepted by the editor, not because they are better, not because they are historical, not because they are sacred, but because they are social—representing a necessary bio-socioeconomic relationship between author, publisher, and audience (including the editor). The result of this argument is that the printed text is preferred as copy-text, emendation is discouraged but allowed to complete or correct the socializing process, and the critical apparatus is seen as an important record of social/cultural dynamics.[28]

Intrinsic to this argument, with its redefinition of textual authority, is a concept of transferring authority from the author to the publisher. Different theorists would argue in different ways about how this transfer takes place; it is different, perhaps, for every author and work and, therefore, looks different to each editor. The evidence of transfer is not new, though the conclusion that the author transfers authority over the text to the publisher when he "submits" the manuscript is, perhaps, new in editorial theory. It is, of course, a quite old idea in commercial practice. Editors who use this argument are, however, usually comfortable with it only as long as the production process changed insignificant details or "improved" the work. When actual damage resulted from the publication process, the theory ceases to be appealing and recourse is taken to another.

A very interesting discussion of the various attitudes toward the author's "submission" to the publisher and the relationships that they represent was presented by James L. W. West at the Society for Critical Exchange.[29] He made a potentially useful distinction between attempts by the production crew to enhance the author's "active" intentions and its attempts to enhance the "programmatic" intentions, terms borrowed from Michael Hancher.[30] West's position was that only limited transfers of authority may occur and that it is part of the scholarly editor's duty to determine which were legitimate

fulfilments of active intention and which were illegitimate attempts at programmatic changes. It is clear that the details of an individual case would influence any editor's assessment of the process. The argument in the hands of a "social contract" theorist would tend, however, to acknowledge the authority of the production process in all cases. McGann uses this argument to deal with Byron's works, in A Critique of Modern Textual Criticism, but he is reluctant to apply it to Shelley's works in which the author was judged to have been more clearly in control of the effect of punctuation and the publication process was judged to have marred that control rather than enhanced it.[31]

The fifth argument adds several new dimensions. Rather than focusing on the production process and its socializing influence, this argument focuses on the physical artifact that results. It does not argue that the "known" text is part of our cultural heritage to which we have become accustomed and which has influenced so much critical debate in the form it was produced. Rather it says that the physical object is a version of the work that itself generates meaning. It argues that the linguistic text generates only a part of the meaning of a book; its production, its price, its cover, its margins, its type font all carry meaning that can be documented.[32] In short, the physical book, of which the linguistic text is but a part, is important not because we have become accustomed to it, and not merely because it is a part of history, but because its form and historical entry into the culture determined the cultural acceptance it received. For example, the elegant first edition of Byron's Don Juan was received as a witty and spirited performance, but reviewers of the cheap pirated editions called it an immoral travesty. The same verbal text in differing physical formats meant different things.[33] When the fact that the pirated cheap "immoral" texts were the ones that most people read and that made Byron's "literary reputation and popularity," is added, we have a very complex textual situation. The implications of this phenomenon to the social contract editorial school is that the physical books, as products of the social context and contract, are physically the works of art.

There is a basic validity to this argument that I think few persons would deny. One's reaction to a work is conditioned by his knowledge of a variety of factors having little if anything to do with the linguistic text itself. Book designers know this well; cheap literature in small type and double columns is printed on newsprint or other inexpensive paper. Lurid covers indicate something about how the publisher hopes readers will react to the text and may also indicate what kind of reader is being addressed. Likewise, hard covers, ample margins, generous leading, and heavy paper often imply the social status of the publisher and of the readers. They may also imply the literary value or durability of the "work of art" thus produced. N. N. Feltes, in a recent discussion of W. M. Thackeray's Henry Esmond pursues this line of investigation. He argues that, by its physical format (a Victorian three decker), its price (36 shillings—out of the reach of most ordinary book buyers), its genre (historical fiction), and its publisher (George Smith, an executive

member of the Bookseller's Association) Esmond represents "the establishment" and is unlike Thackeray's earlier parts-issued Vanity Fair and Pendennis which, by their price, formats, and publisher represented the proletarian economic revolution.[34]

This is a very interesting approach to the meaning of Esmond, bringing into serious question the effect of reprinting the novel in modern dress (one volume, modern type font) and obscuring by that process these meaningful elements of the first edition. This is true even though Feltes' conclusion, that Thackeray's hand was limited by these "establishment" forces and that the meaning of Esmond is "determined" by the social contract over which Thackeray had no control, is not convincing.[35]

It is one thing to recognize the implications of the physical embodiment of a linguistic text, quite another to identify the text in its physical format as the work of art, and still another to say that the social and economic context within which the work was published determined what the author would or could do. There may be some truth in all of these propositions, but it would be very hard to make a rule out of them that would apply well to many texts. In some ways this argument sounds rather familiar; in 1952, for example, Gordon Ray made much the same case in "The Importance of Original Editions" but drew a different conclusion, since Ray saw the cultural history of books as external to authorial intention and as an added, rather than substitute, interest.[36]

The concept that physical books have integrity and meaning in themselves bears more thinking about. The meanings referred to can be not only in addition to the linguistic text, but separate from or unified with that text. They can be meanings of which the author was unconscious or meanings the author consciously exploits or manipulates. There are many well-known instances of authors consciously seeking to enhance their text by controlling the physical form of the works: the green cover of Leaves of Grass, the shaped lines of George Herbert's "Easter Wings," the private limited printing in 1926 of The Seven Pillars of Wisdom, Blake's lithographed combinations of text and illustration. Other cases are known of authors manipulating or taking advantage of reader reactions to physical formats: Thackeray's references to the jaundiced yellow covers and Trollope's remarks about the third volume of the novel. But for the most part authors do not control these matters and the "meanings" or implications of the physical text, if anything, reveal unconsciously things about the author or publisher—rather like dramatic irony where the physical book is the situational context of the linguistic text, though sometimes the implications may not be ironic.

Readers, by and large, of classic texts in new editions are taught to ignore the extra-linguistic contexts of the physical books in the same way that theatre goers are taught to ignore the stage relation to the audience through a willing suspension of disbelief. Thus we read Tennyson's In Memoriam in the cut version in the Norton Anthology and "understand" the poem as if its placement had nothing to do with how we read the poem or what the poem

means (willing suspension of disbelief, indeed). The social contract theory asks readers to pay attention to the physical setting of "authoritative" versions (not the Norton Anthology) of the physical work.

Before proceeding, let us summarize for convenience the five arguments:

• 1. Production crews improve works of art—the editor and critic recognize this improvement by judging the results.
• 2. Publication is a historical fact—emendation would violate the historical event represented by the published text.
• 3. Years of use have validated the familiar text, it has been appropriated by the cultural heritage—emendation would confuse people.
• 4. Publication is an integral and necessary part of the social act of producing works of art; production influences pervade composition from its inception and should be accepted as a social fact—part of the definition of "literary work of art."
• 5. The physically produced work is the work of art, its physical form reflects social contexts revealing the true character of the work.

The most important difference between the social contract approaches to textual problems and previous editorial principles lies in the radically altered definition of textual authority that they offer. Whereas earlier principles for critical editing accepted without question the authority of the author, differing primarily in determining what the authorial intention was, the social contract principle denies the author the pride of place and substitutes the social event that produced the book as the authority. In that complex, the author is merely one of several authorizing forces; authorial intention may still have relevance to editorial practice but it is no longer the central focus; other "larger" issues enter the equation.

At first, looking primarily at the first three arguments for "social texts," it would appear that the principle of social contract would eschew critical editing because those arguments identify the work of art with the commodity offered for sale, the socialized product. Perhaps some of its adherents would make that identification. In its most rigid form, this principle would require that printed texts should be chosen as copy-texts and that emendations should not be allowed. But if those two arguments are disposed of, as perhaps they should be, and the last two arguments assessed critically, we can see that a concern for authorial meaning is merely different from the concerns emphasized here, not a concern denied by them.

There is, perhaps, a sixth argument relevant to a social theory of texts, though in some ways it may be merely the extrapolation of the fourth. Its summary for the chart above would be: 6. The social determinism of texts.

It may be that this concept has nothing to do with a social contract theory of editing, but it does seem to underlie a good deal of the thinking behind the other five arguments I have been trying to sort out. I try to approach it merely from the point of view of a scholar wanting to know everything I can about the work, its author, and its historical and linguistic context in an attempt to understand it as a whole.

This idea concerns the importance of the entire bio-socio-economic context in determining the work of art. If we do not understand this idea thoroughly, it may appear to have unqualified power or authority over texts, leaving us to conclude simply that whatever text was produced by the publishing contract is the right text. That conclusion appears contrary to any editorial principle in which authority is defined as author-centric. This idea is alluded to by McGann when, in reviewing McKenzie's 1985 Panizzi Lectures, he speaks of "the people who belong to the text."[37] The usual approach is to see things the other way round: the text that belongs to the author or to the publisher or to the reader. That has been the conventional way of seeing "authority." That is the "old" way of shaping editorial policy—by reference to those persons who authorize or determine what the text should be. McGann and McKenzie seem to make the text and the social forces it represents larger than any person. They include the text itself as a part of the "contract" or the complex of social, political, and economic forces that determine life. Social determinism becomes literary determinism. The language speaks the author; the social complex makes the book. The book is not "a container or transmitter of meaning"; it is "a meaningful agent itself."[38] And what it has come to mean is not within the author's or any other person's control. N. N. Feltes, in the discussion of Esmond mentioned above, concludes from his study of the "establishment" elements in the book that Thackeray was forced into uncharacteristic, reactionary meanings—that authorship is largely determined by economic forces and that these forces reduce or replace the play of individual judgment.[39]

The idea of socially determined texts may best be understood by asking several related questions designed to clarify the way in which a definition of authority controls editorial decisions and to highlight the difference between author-centric and socio-centric approaches.

What happens when the contracted socializing process violates the confidence or trust which the author unwittingly placed in it? Is the social contract or the author's wishes of greater authority? Similarly, what is the status of the "contract" when the socializing process takes place without the author's consent, as in posthumous publications? Is there to be a presumption that the socializing process is proper, legitimate and good unless something can be shown to be wrong with it? If the answer to this last question is yes, it follows that publisher's punctuation would be preferred to author punctuation. One could ask why that must be so—what is the applicable rationale for copy-text and what distinction is made in that rationale between accidentals and substantives. None of the theorists proposing arguments on behalf of the social contract seem to have a well-developed or convincing rationale for dealing with accidentals.

What happens when the succession of reprints that perpetuates and validates the cultural heritage perpetuates demonstrable flaws?[40] Is a restoration of text which creates an unfamiliar version disallowed because it violates the cultural artifact?

What happens when the authorial intention is weak or flawed but is

substantively improved by editorial advice?[41] Is the social contract, or in this case collaboration, acceptable only when the results are judged acceptable? Similarly, what happens when the authorial intention was strong and unorthodox but is substantively weakened by the publication process? Does that prove that no contract exists? If the results can be judged unacceptable, then the criterion is aesthetic not contractual; there is no contract. And what if that impoverished text makes a cultural impact validated by large sales and broad influence?

What happens when a text considered as a cultural object with meaning and significance derived from its physical form becomes re-objectified in new editions? Is each re-objectification a new version of the work worth editorial attention? No one, I think, would deny that each is worth attention as a fact of cultural history, but does that make each publication a part of what the work of art is from a critical point of view? If the answer is yes, how does it affect editorial policy?

Listing out these questions and contemplating their possible answers suggests that the "social contract theory" is just as complex and variable as the author-centric approaches have been. It would therefore, I think, be short-sighted and narrow-minded to try to persuade editors that a social theory of texts with a definition of authority marginalizing the author is the only responsible way to edit.

IV

The common link of all these questions and of the theories available with which to answer them is the concept of authority. Authority is not something "out there" to be discovered and analyzed. It is a concept brought to the situation by the observer. It is whatever it is defined to be. The hesitation one has in answering these questions—and, therefore, of knowing whether authorial acts are determined or free—comes from the different definitions of authority implied by the way the questions state the supposed conditions.

The idea that authority is not intrinsic or discovered in the textual problems but is, instead, brought to the problem by the editor to help evaluate the problem is not one I expect most persons to recognize immediately.[42] It is worth, therefore, a separate discussion of authority before we can get on properly with the questions I have just posed.

The greatest obstacle to understanding differences in editorial principles and in particular to understanding the claims of the social contract is the word "authority." I think this is so not only because it is used to denote several different things, which users suppose to be clear from the context, but because it is an emotionally loaded word that grants approval; it is not cleanly descriptive.

I would like to sort out various definitions of the word derived from its use to describe textual phenomena, show that it is used to indicate concepts at various levels of thinking, show that sometimes it is used as a specific description (as we use the word "yellow" in the phrase that is a yellow schoolbus) and other times as a comparative word (as we use the words "clear" and

"clearer" with an eye-doctor to indicate which lens clarifies the eye chart better), and show how those things can be thought of separately from the approval or disapproval indicated by the words "authoritative" and "nonauthoritative."

One meaning of authority is "deriving from the author." This usually means that a word, phrase, or point of punctuation can be shown to have been written or dictated by the author. Melville's "coiled fish" has authority; the compositor's "soiled fish" does not. Such words are said to have authority because we know they came from the author. One could call this specific kind of authority "primary authorial authority," which may sound redundant but is not when we consider other meanings for authority. An author's manuscript, of course, has primary authorial authority. So do the autograph alterations in proofs or setting-copies for revised editions and instructions in letters or notes to secretaries, compositors, or publishers.

Another meaning for authority is "having a demonstrable, though not precisely known, relation to the author." This usually means that it is generally known that the author did revise or proofread the text which is said, therefore, to have authority. The specific words, phrases or punctuation that have "primary authorial authority" are not known. One could call this "secondary authorial authority." Instances of secondary authorial authority can be seen in a scribal copy or typescript made from a lost manuscript or a magazine publication made from a lost revised carbon typescript where the fact of authorial revision is not in dispute but the details of specific revisions cannot be recovered.

Another meaning for authority is "deriving from a document with 'primary' or 'secondary authorial authority.'" This usually means that the text referred to is the closest known text to one the author wrote or otherwise supervised. It does not mean that the author necessarily had anything to do with the typesetting of proofreading. This could be called "primary documentary authority" to distinguish it from later editions. Instances of "primary documentary authority" would be the so-called good quartos of Shakespeare's plays or any first published appearance of a book for which prepublication forms have disappeared and for which it is reasonable to suppose that setting copy was the author's manuscript or fair copies made from documents with "primary authorial authority."

Another meaning for authority is "having a precedent in a historical document." This usually means that the text, as preserved in physical documents that may be (and probably are) corrupt, has an unknown relation to the author and may or may not preserve the authorial forms as successfully as other documents with similar characteristics. An instance of this kind of authority would be the existence of say three syndicated magazine versions of a story no longer extant in manuscript. The differences between the versions probably originated in the composing rooms of the various publications. There is no other "authoritative" source for the story, hence these three documents are the authorities. This could be called "radial documentary authority."

Supposing that one of the three documentary authorities corresponds in the pattern of punctuation more closely to the patterns familiar in the author's surviving manuscript, that document may be said to have "more authority" than the other two, at least with regard to punctuation. Notice that this brings into play the word "authority" in a comparative rather than a descriptive sense. Here the word is used evaluatively and reflects critical judgment in analyzing punctuation patterns. It would be more accurate to say that this one document is "more likely to be authoritative in its punctuation than the other two" than to say that "it is more authoritative," and it would be better yet to have a word other than authority or authoritative to describe what is being said. Perhaps it would be wise to revert to the terms "more or less sophisticated" and "more or less corrupt" to indicate our sense of the relative amount of intentional and unintentional alteration introduced by compositors or other production personnel.

Another instance of "documentary authority" illustrates yet another way in which the word authority is used evaluatively. If one has a "difficult" reading in a text with "secondary authorial authority" or even with "primary documentary authority" which has been altered to an "easy" reading in a purely derivative "nonauthoritative" text, editors sometimes resist the temptation to adopt the easier reading by "sticking to authority." This use of the word places value on the tangible text over the critical conjecture of later compositors or modern editors. The possibility that the physical text may be wrong can never be proven, though there are some difficult readings that no one (so far) has tried to defend; but "Mne Serephim" in Blake's "The Book of Thel," "spitting" rather than "splitting" in Shakespeare's Henry VIII (2.4.181), and the auctioneer who "repeated his discomposure" in Thackeray's Vanity Fair are just a few of myriads of cases where the editor is tempted to supply conjecture rather than "stick to authority." Just as the documentary reading cannot be proven to be wrong, so the conjecture can never be established to be correct (if it could it would cease to be conjecture). The point here is that the word authority is used evaluatively to defend an action taken rather than descriptively to indicate the nature of a situation. It would be just as accurate to say of the editor who "stuck to authority" that he knuckled under to the tyranny of copy-text or that he is a very conservative editor or even that he is an unimaginative editor. All these terms are evaluative and not indicative of anything demonstrable. What can be said specifically and objectively of "Mne," "spitting," and "repeated his discomposure" is that they occur in texts with "primary" or "secondary documentary authority"— that and no more; anything else would be a matter of judgment.

All of the foregoing definitions reflect an author-centric view of textual criticism. Especially in the use of "authority" comparatively, one can see the controlling influence of an attitude of respect for authorial autonomy with regard to txets. There is nevertheless considerable room for disagreement within these views. Depending on the stress one puts on the authority of one document over another, depending on the sense of historical or documentary

integrity one espouses, one can within these definitions be a documentary editor or a critical editor.[43]

Hershel Parker invokes yet another definition of authority when he limits it to the product of the act of creativity during the time that creativity as he defines it is in fact going on. As the author works he imbues the work with his intention and perhaps finds within the work the work's intention, which then takes over and becomes the controlling force of the work. That controlling force becomes the "authority" of the work and its intention— and it therefore becomes the editor's aim to edit the work in accordance with that authority. To do so, he may have to reject a good deal of editorial tinkering that the author may have done to the text after he had lost the creative urge that produced the work.

A totally different level of meaning is invoked when one asks "by what authority does the editor emend the text?" Here the word "authority" has no direct reference to the source of a specific reading; whether a word began with the author or a compositor or whether it is in a text with primary or secondary documentary authority are irrelevant questions. What this question asks is, what is the critical theory about works of art that leads you to believe that the author is the ultimate source of authority? Why, for example, is it that the text you are trying to present need not have "better" readings, why is it sufficient that they be authorial?

Until a person is willing to consider the definition of authority at this level there is no point in wasting words about social theories of text or about the aesthetics of textual criticism. As long as the author-centric view is the only legitimate view, James Thorpe and Jerome McGann will remain outsiders whose editorial principles need not be taken seriously.

It is painful to consider the question, "by what authority does the editor emend the text?" because a serious consideration of the answer requires examination of fundamental assumptions about things that turn out to be problematical rather than solid and objective. As long as one remains settled inside the author-centric world of ideas, the concept of authority is very useful practically in the business of producing scholarly editions, documentary authority is a firm resting place, primary authorial authority is like being in clover. The problem with new editorial discussions is that they are successfully forcing thought to be expended on this higher level of authority. They are succeeding because it has become de rigueur to be self-conscious about our critical theories.

In order for it to be acceptable to consider McGann's or McKenzie's editorial theories, one must put in abeyance the definitions of authority I have surveyed above (put them on hold, so to speak), shift gears so as to consider the larger questions of authority which a "social theory of texts" demands, and anticipate the possibility that one's author-centric definitions of the materials of textual criticism will have counterparts reflecting a different higher level concept of authority.

It appears to me, therefore, that the major question, the question of

authority, may usually be argued and settled before the evaluation of specific evidence concerning the composition and production of a given text takes place. Editors may approach a text with preconceived ideas, about the authority of the author or the social contract or the roles of publishers, that will predispose them to interpret the evidence in certain ways. In general terms the evaluation of specific evidence may not be determined by its intrinsic meaning—that which "the facts cry out"—but rather by some previous determination concerning the nature of the work of art. For example, it could be argued that Dreiser's friend Arthur Henry gave him bad advice about Sister Carrie that was designed to avoid problems Henry himself had encountered but were not designed to help Dreiser fulfil his own artistic intentions with regard to the novel. One editor will say, "save Dreiser from his misguided friend," another will say, "Henry is a fact of life, his advice was taken, the book that Sister Carrie would have become without his intervention is not recoverable." It could further be argued that Mrs. Page, the publisher's wife, influenced the book even more significantly than Henry, since by her insistence Sister Carrie was suppressed, though technically it was published. She did to the book something that can be measured in the circulation and reception of the book. Our first editor would say, "rescue Dreiser's book from the publisher's influence"; the second might say, "the publisher is a fact of life without which Sister Carrie would never have become a suppressed book—along with all the meaning that condition adds to the book." My point is not to say which is right but to question the proposition that the facts of publication history "cry out" for a specific treatment. What an editor will see as appropriate treatment will depend on a prior commitment to a definition of textual authority.

Seen from a sociological perspective, the specific evidence of the social context will "cry out" for attention and recognition as legitimate operants or influences in the production of texts. Those facts will cry equally loudly for a place in the reader's interpretation and appreciation of the work as representation not only of the aesthetics of its time but of the economics and politics of its time. The word "legitimate" is not exactly right; the fact that production (the world of publishers, editors and printers) is an acknowledged, normal part of the creation of book texts makes the participation of production staffs an integral part of the creation of the book, of the work of art. It is argued that authors write with the expectation of receiving production help in completing their art; therefore, the help they get is a "legitimate" operant or influence.

Further, given this essentially marxist orientation and principles, an edition that systematically eliminates the influences of all contracting parties but the author (such as was goal in the Pennsylvania edition of Dreiser's Sister Carrie) will be seen as partial, distorted, and misrepresentative of the historical, socio-economic and political event that produced it.

Marxist critics take the social implications of texts more seriously than most critics. In so far as they recognize the implications of editorial principles aimed at recovering the author's final intentions by eliminating the "external

influences," they can be nothing less than disappointed and will more likely be enraged by what they will see as the desecration of texts. For the marxist critic, the editor's pursuit of the pure virgin text is a hypocritical, evil coverup, unless it be an ignorant naive game.

The general rules or orientations that tend to predetermine how the evidence will be used often are forced upon the evidence. One can have rules, said A. E. Housman about editors nearly 100 years ago, but they will lead you wrong. That is, when an a priori set of principles leads to a mechanical or rigid manipulation of the facts of a given case, the potential provided by the overall view for generating exciting new insights is hobbled by an unexciting tyranny.

A look at specific cases will illustrate the uses and potential abuses of the fact that a general orientation predetermines the use of evidence. But first let me reiterate the basic questions that stimulated this discussion and encapsulate the competing and apparently irreconcilable general principles involved.

The questions are, what is textual integrity (the unity and honesty and authenticity of the text of a work of literary art)? Is it possible for there to be more than one "legitimate" integrity for a single work? What is meant by textual authority? What should it be? Do editors influence interpretations (how do they do it, should they do it, can it be avoided)?

These questions arise because traditional consensus about the real nature of works of art has been challenged by a competing notion that appears antithetical, and may in fact be antithetical, to it. Persons in our profession can no longer assume that everyone sees the work from a common point of view, but many of us act as though we still should be able to agree. I do not mean we can no longer count on standard interpretations; I mean we no longer agree about the nature of the thing we call a literary work. I also do not mean we no longer agree on the canon or on which works are literary; I mean we no longer agree about the foundation of words and linguistic meaning. One of these notions, the more traditional one, is that the work of art is a personal communication from an author to an audience. The assumption is that when the author writes he has something he wishes to communicate; when he puts words on the page he is trying to create certain effects in readers; when an author has developed the artistic and technical skill to create works of art, he is in control and does things deliberately. These assumptions I would call a commonsense approach because generally speaking people think that way about their own speech acts and because that is what I am trying to do right now. This view accords with the editorial principle of pursuing authorial intention and with the critical hermeneutical principles discussed by E. D. Hirsch, Jr., in Validity in Interpretation. And this general notion of authorship views the production process of turning manuscripts into books as a service to the author helping him get his creative effort into a form for general dissemination.

The apparently antithetical competing notion is that the work of art is social rather than individual. Rather than the artist using language to create

a new work of art, the language speaks the artist. Rather than the production crew being assistants in the author's effort to communicate, the publishing world is a cultural agency which employs authors to aid them in providing society with works of art. The actualizing or domesticating of art for society is the production function. Both the author and the production personnel are to some extent cultural puppets, products of their time producing inevitable art. The way we know that what came out of the process was inevitable is that something else did not materialize instead. The function of criticism in this view is to interpret the texts as cultural artifacts in their historical settings.

If nothing else, this survey of competing views of editing has convinced me even more that editing is a critical enterprise that not only involves criticism but is in fact a form of literary criticism. Criticism is interpretation; editions, like other interpretations, can be supported by evidence and argument, but they cannot be proven or validated. They are not definitive. Consequently, the interpretation of the work, which an edition is, must have its literary critical basis clearly acknowledged by the editor for the edition user. I believe no theorist should say that his method is the only responsible one, though I think it is possible to discover that some methods are irresponsible.

One very interesting idea arising from D. F. McKenzie's remarks about the object that is the work of art is that the physical object that is the scholarly edition changes and enriches the work much more than it preserves or restores it. In critical editing, not only is the edited text itself one which never existed before, it is surrounded by alter-texts and related historical materials which have never before been attached to it physically. Scholarly editions invite a kind of reading no other textual form comes close to suggesting. As McGann has put it, scholarly editions invite the reader to read linearly, radially, and spacially all at the same time.[44] Scholarly editors like to think of themselves as historians and preservers when in fact they are the most progressive innovators of new texts and new contexts in the profession.

More questions become possible now: What is the textual editor's responsibility to the author and to the social contract? Is that one responsibility or two? If two, how can the editor balance them? I would like to think it is possible to identify and respond to both responsibilities. But an extended examination of a specific actual case is the only way to demonstrate how that can be done. To do so will lead to assessments of a related question: What is the importance of authorial intention and of the social contract to literary criticism?

I originally intended to include a lengthy discussion at this point of Thackeray's Henry Esmond as an editorial problem to be considered from both the authorial and the social points of view as a trial answer to these questions. I believe, however, this discussion is long enough without that attempt. My essay on Esmond is in my nearly completed book on Thackeray and his publishers. I believe most editors can develop their own answers just as well in reference to specific editorial problems with which they are

familiar—provided of course that as A. E. Housman stipulates they have brains not pudding in their heads.

One way to balance awareness of cultural context, including the ministrations of production crews, with a respect of authorial intentions was suggested by Fredson Bowers in his address to the Society for Textual Scholarship in 1987: "As a textual critic I am inclined to suggest that awareness of this phenomenon should encourage an editor to remove from the text as much as possible of this non-authorial accretion when it is actually of no material aid in assisting the latest original authorial intention" (my italics).[45] The suggestion is worth pondering though there may be some contortion involved in believing original authorial intentions to be also final intentions and though the process limits critically the force of any supposed "social contract."

There are serious, legitimate differences in point of view in these matters. We may need, for certain important and rich works of literary art, several scholarly editions: the edition representing the author's final intentions, the edition representing the historical event of first publication, the edition representing the thorough revision—each would possibly affect the student in a different way. None would of itself be THE work of art. At the very least, the facts of controversial cases point to the need for editors to be clear-eyed and honest about the particular principles they follow and to clearly identify those principles and the kind of edition they produce. It is not enough to call it a scholarly edition—even if it is approved by the MLA. Ultimately, it is an impossible quest to produce "the edition that conveys the author's most comprehensive intentions"; what we can hope to do is produce "the edition that conveys the author's intentions most comprehensively."[46]

Notes