University of Virginia Library

The Editing of Historical Documents
by
G. Thomas Tanselle

If the third quarter of the twentieth century can be considered—as it often is—an age of editing, one of the principal reasons is the existence and influence of two American organizations: the National Historical Publications Commission (NHPC), renamed in late 1974 the National Historical Publications and Records Commission (NHPRC); and the Center for Editions of American Authors (CEAA), succeeded in 1976 by the Center for Scholarly Editions (CSE). The NHPC (NHPRC)[1] has since 1950 given encouragement and assistance to a large number of multi-volume editions (more than four dozen) of the papers of American statesmen, especially those of the late eighteenth and nineteenth centuries; the CEAA, from 1963 through 1976, gave its official approval to volumes in fourteen editions, predominantly of the works of nineteenth-century American literary figures.[2] As a result, massive scholarly editions have been produced in an unprecedented quantity during these years; hundreds of scholars have been connected with these projects, and widespread discussion and awareness of the problems and aims of editing have been engendered. The presence of these editions has dramatically altered the scholarly landscape in American history and literature within a generation.[3]

When there is so much editorial activity directed toward material from a single country and, for the most part, a single century, one would expect a great deal of communication among the editors involved; indeed, the creation of coordinating organizations like the NHPC and the CEAA suggests a recognition of the need for such communication. However, the fact that two organizations have seemed necessary indicates that the communication has not very readily crossed the boundary lines between academic disciplines. Regrettably, but undeniably, editors of "literary" material and editors of "historical" material[4] have gone their separate ways; members of each group have discussed common problems among themselves but have remained remarkably uninformed about what was taking place in the other group. One does not have to examine many volumes to recognize a central difference between the historical and the literary editions: the historical editions in general give more attention to explanatory annotation than to the detailed recording of textual data, whereas the literary editions reverse this emphasis. It is a fact that most of the historical editions do not meet the standards for reporting textual information established by the CEAA and would therefore not qualify for the award of the CEAA emblem. Whether those particular standards are justifiable is a separate question; what is disturbing is that such different standards should prevail in the two fields. If one could argue that the material edited by historians is different in kind from that edited by literary scholars, there might be some reason to expect different approaches. Indeed, the NHPRC editors do have more occasion to deal with manuscript letters and journals than with texts which were published by their authors, and for CEAA editors the opposite situation prevails. No doubt these relationships are largely responsible for the lesser concern of historians with questions of copy-text

and textual variants and for the greater concern of literary scholars with these matters. Nevertheless, literary editors frequently must edit letters and journals, and historical editors must handle statesmen's published, as well as unpublished, works. The editing of literary and of historical material should have many more points of similarity than of difference; and a greater understanding of mutual problems, between the two groups of editors, is bound to have a salutary effect on the editing produced by both groups.

There have recently been some encouraging signs to suggest that the dangers of editorial parochialism are perhaps becoming more widely recognized. Most notable is the broadening of the scope of the Modern Language Association's committee on editions: no longer limited to editions of American authors, it now provides simply a "Center for Scholarly Editions"—editions of any kind of material from any time and place—and it has shown itself to be concerned with promoting greater contact between editors in different fields. A similar development is the careful editorial attention which has lately been given to certain philosophers: Jo Ann Boydston's edition of John Dewey (1967- ), Fredson Bowers's of William James (1975- ), and Peter H. Nidditch's of John Locke (1975- )—the first two are CEAA editions—manifest an approach to textual matters which had previously been limited almost exclusively to more clearly bellettristic or "literary" writing.[5] In 1972 Edwin Wolf, 2nd, published a timely and well-considered appeal for historians to begin applying to historical works the techniques of analytical bibliography which have long been associated with literary studies, particularly with the editing of English Renaissance drama.[6] He calls attention to the historian's lack of sophistication in dealing with printed texts by pointing out that two of the most respected editors of historical manuscripts, Julian P. Boyd and Leonard W. Labaree, "never questioned the validity of the text of only a single copy of any printed work" (p. 29). After citing some examples of variants in American printed works of the eighteenth century, he again laments the "tradition of a wall separating bibliography as applied to literary works from bibliography as

applied to historical or political works" (p. 37). Nicolas Barker has also found occasion recently to comment on this point: in one of his editorials for the Book Collector he rightly says, "Historians, even more than literary scholars, have been apt to neglect the physical form in which the evidence on which they subsist has been preserved."[7]

In many other respects, the situation in which historical editors find themselves is similar to that of literary editors. In each field there was increased recognition, in the years following World War II, of the need for new editions of basic writings. In each field there was one man whose work provided the impetus and model for further work: the first volume of Julian P. Boyd's edition of Jefferson in 1950 set the pattern for many later historical editions, and the publication of that volume was the occasion for President Truman's reactivating the NHPC (which had originally been established in 1934);[8] the first volume of Fredson Bowers's edition of Hawthorne in 1962 was influential among literary editors in showing how the editorial techniques developed for Renaissance plays were applicable to nineteenth-century literature, and soon after its publication the CEAA was formally constituted (1963).[9] In each field there is thus an agency which serves as coordinator and clearinghouse, though with some differences: the NHPRC[10] is a government

agency (part of the General Services Administration and housed in the National Archives building), which undertakes to do some research (such as locating relevant manuscripts in archives) for editors; the CEAA[11] was, and the CSE is, a committee of the Modern Language Association of America, which draws some funds from the National Endowment for the Humanities[12] and which calls attention to excellence in editing by awarding an emblem to volumes that qualify (after being requested to inspect printer's copy for those volumes by their editors). In each field there has been some controversy surrounding the new editions, though for characteristically different reasons: criticism of the literary editions has been concerned principally with textual matters, whereas the main questions raised about the historical editions have had to do with the quantity of annotation, the justification for letterpress rather than microform publication, and the choice of material to be edited in the first place.[13] And in each field the editors have found that a great many of their colleagues neither understand nor respect editorial work;[14] in both fields an attempt has been made to improve graduate training in editing and to bring about a greater interest in and acceptance of editorial projects for dissertations—though the historical field, with the various NHPRC conferences, institutes, and fellowships in editing, has been more active in this regard that the literary.[15]

Despite some differences, editors in the two fields are in similar enough positions and face similar enough problems that one would expect them not only to be conversant with each other's work but to approach each other's concerns in an understanding and constructive spirit. In fact, however, there is, in the extensive editorial literature in the two fields,[16] practically no discussion which takes up the NHPRC and CEAA editions together or which examines the textual policies of the NHPRC editions in the way those of the CEAA editions have often been examined. The most publicized article of this sort is unfortunately one which confuses the issues more than it clarifies them. Peter Shaw, writing for a general audience in the American Scholar and interested in exploring textual matters,[17] was in a position to inaugurate a period of productive interdisciplinary discussion; but the regrettable tone of some of his remarks, as well as the fact that they are sometimes uninformed and incoherent, results in an essay which cannot command respect or offer a fruitful basis for further discussion. Shaw believes that the historical editors "unquestionably have had far greater success than their literary counterparts" (p. 739) and finds the literary editors' "tragic flaw" to be "their respect for language" (p. 740). But when he then praises the historical editors' "respect for historical fact," since for them "both the text and its variants qualify as historical facts" (p. 743), one

becomes lost. His point lacks any real substance because it is based on the superficial view that a modern literary editor produces an "eclectic text" and a historical editor a "faithful transcription of a single text" (p. 739)—without examining, for instance, what kinds of texts and textual histories may lead to a literary editor's decision to be "eclectic" or what kinds of textual facts are not recoverable from many historical editors' "faithful" transcriptions. It is naïve to suggest that "the historical editor requires a literary appreciation of nuance, while the literary editor needs the historian's respect for fact" (p. 740); but one can nevertheless agree with Shaw that "each set of editors might usefully have advised the other"—though not because they have "opposite kinds of problems."

What is needed is mutual discussion of common problems, and in this spirit I should like to raise a few questions about the textual policies of some of the historical editions, in the light of what has been learned about editing by the literary editors. In order fairly to assess Shaw's assertion that the historical editors have been more successful, one must examine carefully the editorial rationale and procedures followed by those editors. A survey of the differing practices of a number of editions of letters and journals—both historical and literary—will lead, I think, to a consideration of some underlying issues—issues basic not merely to the editing of the papers of American statesmen but to documentary[18] editing in general.

I

Three statements of editorial policy for historical editions appeared within the space of five years in the early 1950s; all three have been influential, and an understanding of modern American documentary editing must begin with them. The first, and the most influential, was Julian P. Boyd's account of his "Editorial Method" (pp. xxv-xxxviii) in the first volume of The Papers of Thomas Jefferson, published by Princeton University Press in 1950.[19] Boyd states that his general aim is "rigidly to

adhere to scrupulous exactness in the presentation of the texts as Jefferson wrote them" (p. xxviii), but he recognizes that "complete exactitude is impossible in transmuting handwriting into print"; he has therefore worked out a "standard methodology which, though sometimes consciously inconsistent, is nevertheless precise" (p. xxix). From this, one assumes that the only changes to be introduced are those necessitated by the typography. As soon as he starts to explain the methodology, however, one begins to wonder how it supports his aim of adhering to the text with "scrupulous exactness." He says that he is going to follow a "middle course" between "facsimile reproduction" and "complete modernization," except in the case of business papers and of certain important documents (like the Declaration of Independence), which are to be "presented literally." There are thus two categories of material, accorded different treatment: letters and ordinary documents, presented with some degree of "conventionalization";[20] and business papers and important documents, presented as literally as print allows. Only the treatment of the second category would seem to fulfill the goal of presenting with "scrupulous exactness" the texts "as Jefferson wrote them" or of providing "as accurate a text as possible" which preserves "as many of Jefferson's distinctive mannerisms of writing as can be done" (p. xxix).

In the first, and larger, category, spelling, grammar, and capitalization remain unchanged, except that each sentence is made to begin with a capital letter (in contrast to Jefferson's practice). As for punctuation, however, "for the sake of clarity this literal policy will be less rigorously applied" (p. xxx): periods are supplied, when lacking at the ends of sentences, and unnecessary dashes, such as those which follow periods, are deleted.[21] Although this alteration of punctuation is minimal, one may well ask what is gained by eliminating these dashes; they could not cause a modern reader to misinterpret the sense, and, if they are a characteristic of Jefferson's style, to delete them is at best to modernize and at worst to risk losing a nuance of meaning. More troublesome is the treatment of abbreviations and contractions. They are "normally" expanded,

with the exception of those designating money or units of measure and weight, those standing for proper names, and a miscellaneous group containing such forms as "wou'd," "do." (for "ditto"), and "&c." (though "&" alone is altered to "and"). The rationale for this arbitrary list of abbreviated forms to be retained is not clear, especially since Boyd recognizes that some of them will require editorial expansion in brackets. If there is a value in preserving these contractions, why should others be expanded silently? Boyd gives an example to show Jefferson's extensive use of abbreviations in hurried jottings: "wd hve retird immedly hd h. nt bn infmd" is expanded into "would have retired immediately had he not been informed" (p. xxxi). The expanded text, Boyd argues, "represents the kind of clear and readable form that Jefferson himself would have used for a document intended for formal presentation in print. It makes for clarity and readability and yet sacrifices nothing of Jefferson's words or meaning." But the document was not in fact intended for formal presentation, and to smooth its text out silently is to conceal the essential nature of the preserved document. And if the nature of a document is misrepresented, even if the literal "meaning" is preserved, can one say absolutely that the meaning has in no way been sacrificed? It is true that a long passage full of such abbreviations would slow the reader down, but the reader's convenience is surely not the primary consideration here. The argument presented for expanding contractions like "wd" and "hd" could just as well be applied to "Wmsbgh," yet contractions of geographical names are allowed to stand. Perhaps this distinction is one of the conscious inconsistencies Boyd alludes to, but the reason for it remains unclear. It is disturbing because it would seem to reflect a wavering between two editorial approaches—an indecisiveness whether to transcribe or to normalize.

Three basic decisions about the nature of the edition are implicit in what has been said up to this point. One is that the text is to be critical, in the sense that it incorporates certain kinds of changes dictated by the editor's judgment. A second is that the original text will not be fully recoverable from the data provided; some editorial changes, in other words, will not be recorded. And the third is that the edited text will not be "clear text"—that is, it will incorporate bracketed editorial insertions. These decisions also evidently underlie the treatment of substantive matters, which Boyd turns to next. Conjectured readings are placed in roman type in square brackets and editorial comments (such as "In the margin") appear in italics in square brackets. Such intrusions suggest precision, and it is therefore unfortunate that a bracketed reading in roman type followed by a question mark can mean two different things: either a conjecture at a point where the manuscript is mutilated

and part of the text is missing or else an attempt to read a faded passage or one that is "too illegible to be deciphered with certainty" (p. xxxii).[22] Obvious errors in the original texts are corrected, again indicating that the edited text is a critical one. In writings by Jefferson, the original readings in these instances are provided in notes; in writintgs by others (such as letters to Jefferson), the original readings are not reported—"though," Boyd adds, "if an error has psychological significance it will be allowed to stand, with a note when required." Once it is recognized, however, that errors can have psychological significance, it becomes hard to justify a policy that conceals any of them. And this treatment of errors—emending the text and recording the original readings in notes—is a further reflection of editorial indecisiveness, for it represents a third approach in contrast to the treatment of conjectured readings and of some contractions. In the case of errors, the text is emended but is kept free of editorial symbols; conjectured readings are also placed in the text but are marked there as such; and certain contractions remain unemended but are explained by an editorial insertion in the text. Finally, if two or more copies or drafts of a document exist, variant or canceled readings are reported in notes only when they are "significant." (The variants in fact may not always be known, for it is stated a few pages later that "The editorial policy does not call for full collation of every document extant in more than one version" [p. xxxvi].)[23] Nothing is said about the possibility that a variant reading could call attention to an error in the copy-text, which might then be emended with that variant reading. Of course, if the editorial policy regards each edited text as an edition of a single copy of a document, emendations from other copies would not be allowed. But emendation to correct "obvious errors" is permitted here, and such a category is naturally a subjective one. Can a policy be logically defended which allows the correction of errors that a given editor discovers without recourse to another copy of the text but does not permit the correction of errors that he locates only through examination of another copy? Any procedure that might be called "eclectic" is automatically rejected by some editors. But if a text is not to be presented literally, then the editor's judgment is involved in determining at each point what ought to be in the text;[24] and it is hard to draw a line between being critical (using one's judgment) and being eclectic (considering readings which come from outside a given copy of a text, whether from the editor's head or from another copy of the text). Perhaps such a line could, with careful definition, be worked out; but Boyd's discussion does not acknowledge the existence of this problem, though it implicitly raises the issue.

All these points, one must remember, relate to the treatment of letters and "ordinary documents." The other category of texts, "documents of major importance," are handled very differently. They are presented literally, exactly as found in the document supplying the copy-text—though with bracketed editorial insertions when required for clarification. Variant readings, as before, appear in notes; but all of them, not just the "significant" ones, are recorded. Canceled passages, however, are now given in the text, in italics within angle brackets, placed before the revised wording. Aside from the fact that it is unclear why canceled matter should be reported within the text for major documents and in notes for ordinary documents, the approach employed for the major documents is far simpler and more satisfactory than that for the ordinary documents. With the major documents, no complicated rules are necessary, and yet the reader knows exactly what he is using (with one exception to be noted below); with the ordinary documents, in spite of the complex guidelines, he cannot always know the reading of the original or what evidence is available in other copies or drafts. It may be true that fewer people will be interested in textual details about the ordinary documents; but, if those documents are less important, why should considerable editorial effort be expended to make them more conveniently readable, especially when that effort serves to conceal some evidence that could conceivably be of use? The juxtaposition of the two kinds of texts is in itself somewhat awkward; and the straightforward handling of the major documents makes the compromises involved in the treatment of the ordinary documents appear all the more unsatisfactory by contrast.

There is, however, one serious weakness in the presentation of the major documents: the system used for recording canceled passages. The simple insertion of canceled matter in angle brackets cannot possibly inform the reader in many cases of the true textual situation, especially when no provision is made for labeling which words or syllables are entered above the line. For instance, in the edited text of Jefferson's first draft of the Virginia constitution of 1776, the following appears:

"unless suspended in their operation for his <con> assent" (p. 338, lines 4-5). One would naturally assume that Jefferson had started to write "consent," changed his mind after writing the first syllable, then marked it out and wrote "assent." But a check of the manuscript (reproduced facing p. 414) shows that Jefferson actually wrote "consent" and at some time after that crossed out the first syllable and inserted "as" above it.[25] The printed transcription not only misrepresents the manuscript but fails to show that the revision may have occurred at a time later than that of the original inscription. A few lines later occurs the phrase "endeavoring to prevent the population of our country <by> & for that purpose obstructing the laws" (338.16-17); Jefferson's revision becomes clear only when one knows that "& for that purpose" was inserted above the line at the time when "by" was deleted. Beginning in the next line the edited text contains a phrase that is bound to leave readers even more puzzled: "raising the conditions of new appropriati<ng>ons <new> of lands" (338.18-19). One can of course read the final text here; but if one wishes to know how it read earlier, one cannot simply add the bracketed letters, because no indication has been given of what words or letters were added at the time when the bracketed material was canceled. The manuscript shows that Jefferson first wrote "conditions of appropriating lands." After this "of" the word "new" is careted in; "on" is written over the "ng" and followed by "s"; and after that another caret points to "new of" with the "new" marked out. Thus Jefferson first revised his wording to "conditions of appropriating new lands"; then he further altered it to "conditions of new appropriations of lands." These examples are enough to show that the system is inadequate; reporting cancellations in this way serves little purpose because it does not provide enough information to allow one to reconstruct the stages of revision.[26]

What I have been saying about the textual policy of the Jefferson edition is not meant to cast doubt on the accomplishment of this edition in other respects. It is surely a great achievement in its assemblage and arrangement of material, its exemplary historical annotation, and its generally efficient physical presentation (with each document followed by concise descriptive, explanatory, and—in some cases—textual notes). And it deserves to be praised for the role it has played in causing serious scholarly attention to be turned to the full-scale editing of important statesmen's papers—it has eloquently demonstrated why the scholarly

editor must place "the exacting claims of history" above "the amenities and a respect for the privacy and feelings of individuals" (p. xxviii). What is to be regretted is that an edition in such a strategic position of influence is so unsophisticated in its handling of the actual text. There is no single right way to edit a text, but the editorial policy of the Jefferson edition does not suggest that the alternatives have been clearly thought through. As a result, there is indecision as to whether the text is to be literal or critical, whether it is to be modernized or unmodernized, and whether it is to incorporate apparatus or have the apparatus appended. The reason given for retaining "&c." is that "it was widely used in eighteenth-century printing" (p. xxxi), but Jefferson's "&" is expanded in ordinary documents to "and," presumably because it would not have appeared in an eighteenth-century printed version. Yet, as Boyd recognizes, an editor cannot undertake to capitalize various nouns for Jefferson, even though Jefferson's "extreme economy" in the use of capitals was a matter in which he "differed from his contemporaries" (p. xxx). Is the question of how a given letter or private note would have appeared in print in the eighteenth century even a relevant one, when such documents were not intended for print? The way Jefferson wrote them, however unconventional it may have been, is what the reader is interested in. This view prevails part of the time, since the editor has thought it worthwhile to transcribe the major documents literally. But at other times there seems to be a feeling that formal matters are really not important and that a partially "conventionalized" rendering is all the reader needs. The statement of editorial method, in short, reflects no coherent textual rationale.

Two years later Clarence E. Carter published Historical Editing (1952), a 51-page pamphlet which in some ways is the counterpart, for the historical field, of the CEAA's Statement of Editorial Principles and Procedures (1967, 1972). Although it was not meant to be an official statement of the NHPC (as the CEAA's pamphlet was a committee position paper), it was published as Bulletin No. 7 of the National Archives and was written by a man with extensive editorial experience in connection with a government project, The Territorial Papers of the United States (1934- ). Unlike the CEAA's pamphlet, which emphasizes printed texts and devotes most of its space to discussion of textual matters, Carter's booklet deals with manuscript texts and spends only ten pages on textual questions. Carter refers favorably to Boyd's work early in his discussion (pp. 10-11), but it is clear that Carter's position is more conservative than Boyd's and that he places a higher value on the formal aspects of a text.

Carter begins his account of "Textual Criticism" (pp. 20-25) with

the problem of establishing the authenticity of a document, and then he turns to "the operation designed to clear up such corruptions as may have entered it" (p. 23). This statement suggests that the kind of edited text which Carter envisions is a critical one, not an exact transcription. The matter soon becomes less clear, however. Although he admits that originals may contain errors, he discusses emendations only in regard to copies. He implies that originals are not to be emended, because even in the copy retained by the writer "no editorial emendations are permissible": "it is an official record, and the only resort is to call attention to the presence of specific errors" (p. 24). A copy made by someone else, in contrast, may be emended—but whether silently or not is uncertain. "Conjectural emendations," he says, "are recommended only when it is clear that the errors are due to the inadvertence of the scribe." But, he goes on, "such emendations should be plainly identified as such in footnotes or by editorial brackets in the text" (p. 23). Yet on the next page he says that "slips of the pen" by the copyist can be corrected by "unidentified emendations." Apparently the second category is meant to consist of obvious errors, such as "the transposition of letters in words, or the repetition of words or lines," and the first of less obvious errors. But such a distinction is not definite enough to provide a workable basis for deciding which emendations are to be silent. There is a curious mixture here of strictness and leniency: nothing, not even errors, can be altered in a text from a document in the author's hand; but scribal copies can be emended, sometimes silently. This mixture also reflects an indecision similar to Boyd's about the nature of the editor's task—whether it is to produce an exact transcription of a surviving document or a critical text not identical with the text in any single extant document. The issue emerges squarely in Carter's paragraph on "the occasional needs to reconstruct a document when two or more textual versions are encountered, each of which possesses attributes which stamp it as authentic" (p. 24). The word "reconstruct" suggests the production of an emended text; but his "harmonizing of the various versions" amounts to "the choice of the one which seems to be the most complete one of chronological priority," with readings from the other versions placed in brackets or in footnotes.[27]

Carter says nothing further about emendation but instead turns to "Transcription" (pp. 25-30), where the emphasis is clearly on what he calls "exact copy." His comments are based on a thorough understanding of the value of retaining the original punctuation and spelling; he cites some useful examples illustrating the importance of punctuation

in official documents (p. 26) and notes that the "interest in bad spelling lies partially in that it indicates the current pronunciation" (p. 28).[28] He believes that superscript letters, ligatures, abbreviations, date-lines, addresses, signatures, and the like should all be reproduced exactly.[29] Canceled matter, he says, can be inserted into the text, appropriately marked, or reported in notes—but not simply ignored. To eliminate these passages, as he rightly points out, "omits an element that often indicates what was actually passing through the mind of the writer which he concluded not to set down, and of course it also represents carelessness in many instances—a not unimportant facet of a writer's character" (p. 29). Carter's discussion of "Transcription," taken by itself, sets forth an intelligent and well-considered approach, which is admirably put in practice in his own work on The Territorial Papers (commented on further below).

Although he stresses objectivity here and throughout, he is aware that subjective judgment enters into transcription. When a mark of punctuation is not clearly identifiable, for instance, "it becomes the editor's responsibility to determine from the sense of the passage what was probably intended, and to proceed accordingly" (p. 26). This view is more realistic than the one expressed at the end of the preceding section, where he says that "the editor must eschew any and all forms of interpretation; he cannot deal with his documents in a subjective manner" (p. 25). What he is primarily getting at in this earlier statement is that the editor should not interpret the facts presented in his text, leaving that task for "the historian who uses the edited documents as a basis of historical composition." He is adamant on this point: "It cannot be too strongly emphasized that the editor's sole responsibility, after having established the purity of the documents, is to reproduce them with meticulous accuracy." Despite his insistence, the issue is not so easily settled, for it can be argued that the editor, having thought deeply about the text, is in the best position to suggest interpretations of it in his

annotation. In any case, this question does not affect texual policy. But Carter does not perhaps sufficiently recognize the extent to which judgment inevitably enters the editorial process, especially when emendation is allowed. His discussion, like many others in the historical field, neglects printed texts and (perhaps partly for that reason) fails to confront adequately the issues involved in an editor's decision to produce a critical text; the issues are present even when the only choice for copy-text is a holograph letter, but they may call themselves more forcibly to the editor's attention when he has more occasion to deal with multiple versions of a text. Nevertheless, Carter's comments are generally sensible, as far as they go, and he at least takes notice of—if he does not fully pursue—the problems of choosing a copy-text when one is faced with several copies, none of which is in the author's hand, or with multiple possibly authoritative texts. Certainly his views on punctuation and spelling and on the necessity for recording variants deserve to be heeded more than they have been.

A third influential statement on historical editing was published two years after Carter's, in the Harvard Guide to American History (ed. Oscar Handlin et al., 1954)—which contained a short section on "The Editing and Printing of Manuscripts" (pp. 95-104), prepared primarily by Samuel Eliot Morison. Because of the wide circulation which the Guide has achieved, a great many people have been exposed to this discussion, and it has often been referred to in historical literature as a standard account of editing. When the Guide was revised in 1974 (ed. Frank Freidel et al.), the editors apparently saw no need to alter this section, for it was retained in practically identical form ("Editing and Printing," pp. 27-36).[30] Yet it is a superficial treatment of editing which, like Boyd's and Carter's, oversimplifies or fails to touch basic questions which any editor must consider.

The discussion attempts "to set forth general principles of editing American documents" and begins with the usual point that "printing is unable to reproduce a longhand manuscript exactly." But from there on, difficulties arise. Three methods of preparing texts are announced—called the Literal, the Expanded, and the Modernized—and a preliminary section offers directions that apply to all three. Some of these directions are overly precise and unnecessary—such as specifying that a salutation should be printed in small capitals or that the date line, regardless

of where it appears in the original, should consistently be "printed either in italics under the heading, or at the end" (I.2). What such directions do reveal is that some silent alterations of the original are to be allowed—even in the Literal Method, since these directions apply to all the methods. Three other preliminary directions indicate further—and more objectionable—silent alterations. When a manuscript is torn or illegible, editorial comments are to be inserted in italics within brackets and conjectured readings in roman type within brackets, as Boyd recommended; but, unlike Boyd, the Guide claims that "if only one to four letters [of a long word] are missing, brackets are unnecessary and pedantic" (I.3)—on the grounds that the editor can be sure in those cases of what had originally been written. Yet obviously one cannot really be certain what spelling was used; not to indicate in some way what the editor has done misrepresents the surviving evidence by offering as a fact what is actually an inference.[31] Another direction calls for inserting "[sic]" after "a very strange spelling or mistake of the original writer" (I.5), implying that mistakes are not to be emended. yet the same direction states, "One may correct, without notice, obvious slips of the writer's pen such as 'an an hour ago.'" As in Carter's discussion, nothing explicit is said about what distinguishes errors to be silently corrected from those to be retained. The two categories in fact represent very different approaches to editing, and their juxtaposition here requires further explanation. Still another direction, dealing with manuscript alterations, asserts that "canceled passages are omitted unless they contain something of particular interest, when they may be inserted in a footnote" (I.7). No discussion of what value canceled passages may have is given, nor of what might cause some to be of particular interest; if the point had been taken up and analyzed, the difficulty of regarding any cancellations in a letter or journal as insignificant would have become apparent.[32]

The subsection on the Literal Method begins with the statement, "Follow the manuscript absolutely in spelling, capitalization, and punctuation"—unaccompanied by an explanation of how this directive is consistent with such earlier rules, applicable to all methods, as the one permitting silent corrections of slips of the pen. And it is immediately followed by a troublesome exception: "in very illiterate manuscripts,

where little or no punctuation is used, a minimum necessary to understand the text may be supplied; and in documents where the writer begins practically every word with a capital, the editor may use his discretion" (II.1). Although the editor is told that he should state "the practice followed" in a preliminary note, there is no requirement for him to record his alterations. Obviously the point of a literal method is to reproduce the text of a document exactly as it stands; if a manuscript is "illiterate," the reader of a literal text of it will expect to see the characteristics that make it illiterate. There is no logic in setting up a category called "Literal Method" and then saying that an editor can, in extreme cases, make changes for the convenience of the reader and still produce a literal text. Even if there were really much difficulty in reading a text in which most words are capitalized, the ease of readibility is not a criterion for a literal text. A few changes, of purely typographic significance, can be defended in a literal text, such as the elimination of the long "s"—a literal text, after all, is to be distinguished from a type-facsimile. Manuscript abbreviations, however, constitute a difficult category: one would expect an abbreviation to be reproduced, not expanded, in a literal text, and yet some abbreviations would require specially cast types to be printed. The rule given here is to print abbreviations and contractions "exactly as written within the limitations of available type" (II.4) and otherwise to expand them without brackets (II.5). This procedure is defensible as a practical compromise; but unfortunately the impression is given that an editor need not explain exactly what he has altered in this respect.

For the so-called Expanded Method, taken up next, the Guide recommends Boyd's practice, though it prefers more expansion of abbreviations and more standardization of designations for money, weights, and measures. In fact, most of the discussion is concerned with the treatment of abbreviations, the general policy being to "spell out all abbreviations except those still used today . . . and those of months, proper names, and titles" (III.2). No rationale is given for the aims of the Expanded Method, but since the goal is not to produce a modernized text (that is the subject of the third method) it is not clear why the present-day currency of an abbreviation is relevant. Nor is it clear just what changes are to be made silently. All sentences are to begin with a capital and end with a period, "no matter what the writer does" (III.1); these changes and most expansions of abbreviations are apparently to be made without comment, but supplied letters which follow the last one in a superscript abbreviation are, inexplicably, to be enclosed in brackets ("m°" becomes "mo[nth]"). Except for the treatment of the opening and closing of sentences, the original capitalization and punctuation are to be retained

(III.1), and the spelling as well, even if inconsistent (III.5); the point in standardizing the money, weight, and measure designations, therefore, becomes less clear by contrast.[33] Indeed, the point of the Expanded Method as a whole is puzzling. It is not, as one might at first suppose of an emended but unmodernized text, to correct errors, nor is it to produce consistency, except in a few minor respects; it is simply, as the name indicates, to expand some of the abbreviations. But this expansion does not really constitute a separate "method"; it is more accurately regarded as a form of annotation. One could just as well have a literal text with the explanations of the abbreviations in brackets or notes; indeed, such a procedure would be preferable to the uncertainties suggested here. If the Expanded Method were truly a different method of editing, it would have to involve a basically different approach to the text—a critical approach, for instance, in which the text is emended to correct errors and resolve cruxes. Despite the confusions of the section on the Expanded Method, it ends with a salutary caution: Ironically this closing statement, which contradicts the opening point of the section ("always capitalize the first word and put a period at the end of the sentence no matter what the writer does"), is the most sensible one in the whole discussion.[34]

The subsection on the Modernized Method requires little comment. Modernization is said to be for "the average reader who is put off by obsolete spelling and erratic punctuation." The extent to which the average reader is "put off" by such features of a text is probably not so great as many editors seem to think. In any case, the modernization

recommended here is a confused concept. The first direction is the expected one: "Modernize the spelling, capitalization, and punctuation, but pay scrupulous respect to the language" (IV.1)—although one might not expect the additional statement, "Paragraphs and sentences that are too long may be broken up." What is confusing, however, is that the same instruction also contains this sentence: "Where the original writer has obviously omitted a word like not, or, for instance, has written east when you know he means west, the editor may add or correct a word; but he should place it within square brackets." The correction of errors is an entirely separate matter from modernization, and the two should not be linked together here as aspects of the same "method." One can modernize a text without correcting errors, and one can emend without modernizing. An introduction to editorial method which does not make this distinction will only encourage illogical thinking.

The confusions which underlie the Guide's whole discussion are epitomized in the concluding remarks on "Choice of Method" (VI). The choice is said to depend "partly on the kind of document in question, but mainly on practical considerations, especially on the purpose of the publication." The nature of the document does determine whether expansion of abbreviations or modernization is required, once it has been decided that the edition is aimed at an audience which would require such alterations; but that decision comes first, since for some purposes only the literal approach will suffice, regardless of the complexities of the document. To say that documents of the sixteenth and seventeenth centuries "full of contractions" should be printed literally "in a publication destined for scholarly readers only" is both to underestimate the capacities of a wider audience and to ignore completely the possibility of accompanying a literal text with textual annotation. But why anyone, scholar or not, needs an unmodernized text does not seem to be fully grasped: an expanded text is said to be better for the student than a modernized one "because the wording, spelling, and punctuation of the original give it a certain flavor"—a statement suggesting only a trivial interest in these matters (and again including "wording" as one of the concerns of modernization). The assertion that "for a new edition of some classic such as the Virginia 'Lament for Mr. Nathaniel Bacon,' or the poetry of Edward Taylor, the Modernized Method is best" shows a complete failure to understand the serious reasons for being interested in spelling and punctuation and implies that those features are of less concern in "literary" than in "historical" documents. (An earlier similar comment claims that the "texts of recent editions of Shakespeare, Dryden, and the King James Bible have been established

by this [modernized] method"—as if modernizing could "establish" a text, instead of being a way of altering a text, once established.)[35] The motto offered at the end of the section is in the spirit of the rest of the discussion: "Accuracy without Pedantry. / Consistency first, last, and always." The accuracy required for establishing a text may be regarded as pedantry by some, without affecting its desirability, and what excessive accuracy might be is not defined. If consistency of editorial treatment is the prime virtue, then surely a logical consistency of editorial rationale is a necessity; the Guide in this respect sets a poor example.[36]

These three statements of editorial method were not the only ones available to historical editors of the 1950s and 1960s. Thirty years earlier, for instance, the Anglo-American Historical Committee produced a two-part "Report"[37] —the first dealing with medieval and the second with modern documents—which was in many ways an intelligent and carefully considered statement. Unfortunately it recommended modernizing punctuation for all documents;[38] but, unlike some later treatments, it recognized the importance of recording cancellations and revisions and of providing a detailed account of the practice of the manuscript text in any respect in which the editor alters it.[39] Boyd, Carter,

and the Harvard Guide, however, are more important for anyone examining the NHPRC editions. Boyd's edition led the way for the later editions and was taken as a model, and the other two discussions followed in quick succession at a time when some of the later editions were being organized. The first and third especially have had a considerable influence on a large number of American editions, which either refer to them explicitly or are modeled on other editions that follow their recommendations. If that were not the case, they would hardly deserve the attention given them here; but their deficiencies have apparently not been regarded as obvious. The discussion in the Guide is the least satisfactory, as Carter's is the best, of the group; all three have serious shortcomings, but the one with the most merit ironically has been cited the least often. A recognition of the indecisiveness of these discussions— particularly the two most influential ones—in regard to editorial theory and procedure suggests what a weak foundation they provide for the massive superstructure later erected.

II

A brief survey of some of the historical editions which followed, beginning in 1959 with the Franklin, Calhoun, and Clay editions, will illustrate how similar their characteristic position is to that of one or more of the three statements of the early 1950s.[40] Leonard W. Labaree, in The Papers of Benjamin Franklin (Yale University Press, 1959- ),[41]

sets out to follow "a middle course between exact reproduction . . . and complete modernization" (p. xl)[42] and cites the Harvard Guide for "a discussion of principles which the editors have in general followed." The aim is "to preserve as faithfully as possible the form and spirit in which the authors composed their documents, and at the same time to reproduce their words in a manner intelligible to the present-day reader." Insofar as the second aim involves alteration of the original, it would seem to be incompatible with the first. Labaree distinguishes his treatment of printed copy-texts from that of manuscript copy-texts. The former, he says, are "considered as having been edited once from an original manuscript" and therefore are presented as originally printed, except for the silent alteration of certain typographic conventions (italic proper names are made roman and words in full capitals are made lower case) and the silent correction of "obvious" errors (otherwise, "no attempt will be made to reconstruct the original version"). In manuscript copy-texts, however, contractions are expanded, periods are placed at the ends of sentences, and punctuation is altered in various other ways: "A dash used in place of a period, comma, or semicolon will be replaced by the appropriate mark of punctuation . . . Commas scattered meaninglessly through a sentence will be silently omitted" (p. xlii). These procedures leave the editor in the ironic position of treating printed texts—which are at least one step removed from the author's manuscript and may contain compositors' alterations—with greater respect than authorial manuscript texts, in which there is direct evidence of the author's practice. Furthermore, there is no recognition of the fact that printed texts may vary from copy to copy or that manuscript texts may be of a kind that were never intended for publication. The idea that a printed copy-text has already "been edited once" and thus requires less alteration implies that the scholarly editor's function, like that of the printing- or publishing-house editor, is to put a text—regardless of its nature—in "publishable" shape. But, as Labaree knows, a scholar is interested in the "form and spirit" of Franklin himself; and most of the silent changes described here can only take one farther away from him. Part of the texture of contemporary detail is sacrificed for the sake of a supposedly more readable text, though many of the deleted features would not have caused a reader any real difficulties in the first place. One must wonder why, if a partially modernized text of Franklin had to be produced, it could not have been accompanied by a record of editorial alterations.[43]

The same year, in The Papers of John C. Calhoun (University of South Carolina Press, 1959- ), Robert L. Meriwether took a different position from Labaree, arguing that printed texts could be treated more freely than manuscript texts because Calhoun was not responsible for printed reports of speeches and the like; yet the freedom employed— involving the silent revision of capitalization and punctuation and the breaking up of paragraphs—seems excessive, especially in view of the fact that Calhoun probably revised the reporter's accounts in some cases (p. xxxv). In manuscript texts, the editor does not allow Calhoun to employ two marks of punctuation together (one is chosen), and dashes at the ends of sentences are silently changed to periods. The most confusing device in this edition is the use of roman type in square brackets to represent both editorial restorations and authorial cancellations. W. Edwin Hemphill, taking over with the second volume (1963), makes explicit reference to the Expanded Method of the Harvard Guide (p. xxvii). By contrast, The Papers of Henry Clay (University of Kentucky Press, 1959— ), edited by James F. Hopkins,[44] says little about editorial method and nothing about punctuation, except that the lowering of superscript letters sometimes affects the punctuation. Presumably punctuation is otherwise unaltered, and the "original spelling and capitalization have been retained" (p. ix), so that this edition may come closer to offering a literal treatment than the others of 1959—although "typographical errors" in printed texts are silently corrected. The problem of variant texts, frequently slighted in historical editions, is at least commented on here: "When several contemporary copies, but not the original letter of delivery, have been discovered, that which most closely approximates the form identified with the sender has been used. When there are several versions of a manuscript in the inscriber's hand, that which most closely represents his final intent has been accepted." This statement shows no awareness of the intricacies of textual criticism. The first sentence does not recognize the possibility of constructing an "eclectic"

text; it assumes that the task is to edit a single document, not the text which is found embodied in several documents. Yet when errors in a printed copy are silently corrected, the editor is concerning himself with an idealized text rather than with the reproduction of a specific embodiment of the text; the principle that is recognized in handling a printed text is not extended to situations involving scribal copies, though both may obviously contain departures from the author's manuscript. And the second sentence does not suggest the difficulties of determining "final intent" or the importance of variant readings among the holograph drafts.

In 1961 two more large editions began publication. One was The Adams Papers (Belknap Press of Harvard University Press)—which, like the Jefferson and Franklin editions, had been designated a priority project by the NHPC.[45] Lyman H. Butterfield, describing his editorial method in the first volume of The Diary and Autobiography of John Adams, praises those other two editions, and it is clear that his procedures closely resemble those of the Jefferson edition (with which he had earlier been associated).[46] He aims at a "middle ground between pedantic fidelity and readability" (p. lvi) and adds that scholars who are "concerned with the ultimate niceties of a critical passage" can "resort" to the microfilm edition of the Adams papers.[47] It is true that the availability of the papers on microfilm makes it earsier for a scholar to check readings in the manuscripts, but that fact has no bearing on the editor's responsibility for producing a sound text in a letterpress edition. The reason for undertaking a letterpress edition of material available on

microfilm is not simply to offer a more readable (that is, partly modernized) text; it is to furnish readers with a text which has benefited from the editor's critical thinking about what the writer meant to have in that text.[48] Of course, a scholar under any circumstances may wish to consult the original manuscripts (just as he might wish to check on any other documentation); but to justify silent alterations in a printed text on the grounds that a scholar can always look at the manuscripts is to conceive of editing as little more than styling for present-day readability. In addition, the discussion suggests that only a few scholars will be interested in such matters as punctuation and even takes a disparaging tone toward anyone concerned with them. Rather than counting on the reader's agreement that it is "pedantic" to be interested in the "ultimate niceties" of a text, it would be more positive and productive to assume that readers will want to understand the text as fully as they can and will not wish to slight any aspect of it in the process.

As with many other historical editions, the determination here not to emend from a variant text is in odd contrast to the leniency with which the selected text is handled. Relevant texts are collated and "significant" differences are recorded; however, Butterfield says, "Whatever version is found in the manuscripts being edited has perforce been considered the 'basic' text in the present volume" (p. lix). Two years later, in the opening volume of Adams Family Correspondence, a supplementary editorial discussion marks a notable departure from this practice: the comparison of copies, it is said, can call attention to clarifications of grammar, corrections of spelling, and the like, and such changes are adopted silently (p. xlv). The fact that their immediate source is another document makes this an "eclectic" procedure, and the statement is a welcome recognition of the possibility of editing a text rather than a document. The Adams edition, unlike many of the literary editions of published works, does not fully carry this approach through; but it has gone farther than most of the historical editions in enunciating the principle on which the establishment of critical texts rests.[49]

The other edition beginning in 1961, The Papers of Alexander Hamilton (Columbia University Press), edited by Harold C. Syrett,[50] places even more stress on modernization: not only are punctuation and capitalization altered "where it seemed necessary to make clear the sense

of the writer" but a "special effort has been made to eliminate the dash, which was such a popular eighteenth-century device" (p. xvii). The reader is at more of a loss than usual to know what the editor has done, because "unintentional slips" are handled in one of four ways (they are allowed to stand, explained in a note, corrected with bracketed insertions, or corrected silently), but there is no discussion of the circumstances for choosing one method over another. Deletions are reported only when "the significance of a manuscript seems to warrant it," as is also the case in The Papers of James Madison (University of Chicago Press, 1962- ), edited by William T. Hutchinson and William M. E. Rachal (and, later, Robert A. Rutland).[51] Because Madison made some revisions in his papers long after they were written, the editors rightly feel that these later alterations should be distinguished from the earlier ones: "Changes which the editors believe that Madison made in later life, when looking back over his papers, are given in footnotes" (p. xxxvii). But since the determination of which revisions fall into this class rests on editorial judgment and since cancellations are not reported ordinarily, there is the possibility that in some instances Madison's later revisions have been incorporated into the text, with no record of canceled readings to call attention to the potential problem. Donald Jackson's edition, the same year, of the Letters of the Lewis and Clark Expedition, with Related Documents, 1783-1854 (University of Illinois Press, 1962), is again a partly modernized edition: "When in doubt as to how to proceed in a trivial matter I silently follow modern practice." He employs identical policies in two later editions, The Journals of Zebulon Montgomery Pike (University of Oklahoma Press, 1966- ) and The Expeditions of John Charles Frémont (edited with Mary Lee Spence; University of Illinois Press, 1970- ). Like many of his fellow editors, he insists on normalizing the end-punctuation of sentences and eliminating superfluous dashes.[52] He is also characteristic in neglecting the possibility of authoritative variants in printed texts; as he says in the Frémont, "Material taken from printed texts is so indicated . . . but no attempt is made to record other printed versions." His departures from his copy-texts are in general said to be "based on common sense and the current practice of scholars." Whether that current practice is in turn based on a coherent and defensible editorial rationale is not inquired into; practices which are current tend to become self-perpetuating by inspiring uncritical acceptance.[53]

Some of the other editions of the late 1960s follow the same path. The goal of The Papers of Henry Laurens (University of South Carolina Press, 1968- ), edited by Philip M. Hamer (and, later, George C. Rogers, Jr.),[54] is to follow "with some deviations" the Expanded Method of the Harvard Guide. Although the object is "not only an accurate but a readable text," the word "accurate" here cannot refer to punctuation, and modernization seems to take first place: "The flavor of the eighteenth century . . . has been maintained where clarity would not be sacrificed" (p. xxxi). The editorial function is conceived of as the accurate conveyance of "meaning" rather than of a text: "Superfluous commas may be omitted or reduced in number, and commas will be added when they will assist the reader, but no punctuation will be changed unless it is clear to the editors that no change of meaning will result." What is clear to one informed person, of course, may not be so to another, and it is debatable whether the "readability" gained is worth the price of not knowing what is in the original; reporting the evidence would not settle the question whether modernization is desirable, but it would make the situation more tolerable. The Correspondence of James K. Polk (Vanderbilt University Press, 1969- ), edited by Herbert Weaver, also modernizes for "clarity," including grammar in what can be altered. "These changes have generally been made silently," Weaver says, "rather than risk cluttering the pages with editorial props that divert attention from the meaning or spirit of the writers" (p. xii)[55] —though the alterations themselves have already done that to some extent.

Not all the editions of the late 1960s, however, conform to the prevailing pattern. One is pleasantly surprised to find that Arthur S. Link's edition of The Papers of Woodrow Wilson (Princeton University Press, 1966- )[56] makes very few—and clearly defined—silent emendations

(such as lowering superscript letters and replacing dashes with periods at the ends of sentences); otherwise, each document is "reproduced exactly as it appears in the original" (p. xvi), with any change marked by brackets (and deleted matter reported in angle brackets). It is true that the changes are made "for the sake of clarity," as in the other editions, but here the reader knows where they occur. Similarly, LeRoy P. Graf and Ralph W. Haskins, in The Papers of Andrew Johnson (University of Tennessee Press, 1967- ),[57] make no changes of spelling or punctuation without using brackets (and apparently the only alteration of punctuation is the insertion of bracketed periods), although they add in the second volume (1970) that slips of the pen are eliminated. A third edition of these years, John Y. Simon's edition of The Papers of Ulysses S. Grant (Southern Illinois University Press, 1967- ),[58] is particularly commendable. It can state flatly that "None of Grant's spelling, grammar, or punctuation has been altered" (p. xxxi), and it reports deletions in canceled type.

Most of the historical editions which followed in the 1970s unfortunately did not imitate these three editions but continued in the familiar pattern of partial modernization and selective recording of evidence. Robert A. Rutland's edition of The Papers of George Mason (University of North Carolina Press, 1970) states that it is following Boyd's Jefferson; while it retains inconsistent spellings, it silently regularizes the punctuation of sentence-endings, reduces Mason's capitalized pronouns to lower case, and inserts periods "in place of many a semicolon or colon that the writer obviously intended to function as a break rather than a pause" (p. xxii). Haskell M. Monroe, Jr., and James T. McIntosh, in The Papers of Jefferson Davis (Louisiana State University Press, 1971- ) also silently emend punctuation according to modern standards, sometimes "correcting" a colon to a comma or a period; but, oddly, they do not insert what they regard as needed punctuation where no punctuation is present in the manuscript, representing the lack instead by an extended space. The Papers of Joseph Henry (Smithsonian Institution Press, 1972- ), edited by Nathan Reingold, takes the Adams edition as its model and incorporates canceled matter in angle brackets if of "historical, psychological, or stylistic significance" (it is hard to

see how any canceled matter could be eliminated on these grounds). Although punctuation and spelling are said to be "usually faithfully preserved," "ubiquitous dashes are converted to modern commas and periods, and a few commas and periods are inserted silently where absolutely necessary for clear understanding" (p. xxxv).

Louis R. Harlan, in the second volume (1972) of The Booker T. Washington Papers (University of Illinois Press, 1972- ),[59] describes his policy of silently correcting "typed and printed errors" and regularizing some punctuation, "except in semi-literate letters, which are reproduced exactly as written in order to avoid an inordinate amount of editorial intrusion into the document." A more valid reason for printing them as written is that the documents are more revealing unemended—an argument which could be applied to a much wider range of material. The first volume of this edition, containing Washington's published autobiographical writings, illustrates the way in which editors who primarily work with single manuscript texts sometimes fail to report adequately on multiple printed texts. Harlan's brief textual comment on Up from Slavery, for instance, merely says that the first book edition is used as copy-text in preference to the serialization in the Outlook because the magazine "did not include all that later appeared in the book version" and because "Negro" is spelled with a capital, as Washington wanted it, in the book but not in the magazine. Nothing is said to characterize the material added to the book or to explain the relation of the book text in other respects to that of the magazine, and no lising of variants is provided. The two texts do differ occasionally in punctuation and spelling ("coloured" in the book vs. "colored" in the magazine, for example), but the question of which text better reflects Washington's practice in these respects is never addressed.

In E. James Ferguson's The Papers of Robert Morris (University of Pittsburgh Press, 1973- ), slips of the pen and "casual or incorrect punctuation" (p. xxxiv) are corrected: "Dashes and commas randomly distributed in the manuscripts are silently removed." Herbert A. Johnson's The Papers of John Marshall (University of North Carolina Press, 1974- ) also silently emends some punctuation but interestingly confuses the author's intention with standards of correctness for a published work: sentences are supplied with opening capitals and closing periods "as necessary to preserve the original intention of the writer" (p. xxxvi). Apparently printed texts are reproduced with greater fidelity than manuscript texts, if that is what is meant by saying that dashes at the

ends of sentences are "silently omitted from documents other than those that reproduce a previous imprint." In other respects punctuation is not emended silently, for Johnson recognizes "the uncertainties involved in correcting any given writer's use of the comma." He very sensibly continues, "Should considerations of clarity dictate some explanatory insertion, the editors have added punctuation in square brackets, thereby permitting the reader to reach his own decision concerning the propriety of the editorial decision."[60] The Papers of Daniel Webster (University Press of New England, 1974- ), edited by Charles M. Wiltse,[61] is similarly cautious about silent changes and makes none except to replace the dashes "intended" as periods; it is careful to retain misspellings and abbreviations or to alter them only in brackets. Merrill Jensen's two recent editions, however, go to the opposite extreme: both The Documentary History of the First Federal Elections, 1788-1790 (with Robert A. Becker; University of Wisconsin Press, 1976- ) and The Documentary History of the Ratification of the Constitution (State Historical Society of Wisconsin, 1976- )[62] remove capitals and italics "except when they are evidently used by the author for emphasis," add punctuation "if needed to clarify meaning," and modernize spelling except for personal names (p. xvi); although official documents and a few others are given in a literal text, other printed texts are emended to eliminate certain eighteenth-century practices, "except when capital letters and italics were evidently used for emphasis by the author or the printer."

Enough has been said to show the characteristic textual practices of the NHPRC editions and other editions modeled on them. But I do not wish to imply that "historical" editions are the only ones which have indulged in partial modernization and selective reporting of emendations and have in general taken a superficial view of textual matters. A number of editions of the letters of literary figures—not particularly influenced by the modern practice of historians—are equally unsatisfactory. The influence, in fact, may go the other way, because The Yale Edition of Horace Walpole's Correspondence (Yale University Press, 1937- ), edited by Wilmarth S. Lewis,[63] was the first of the modern

large-scale editions of a single figure and has been cited as an influential force in some of the historical accounts of the NHPRC editions.[64] Lewis did set a good example in his thorough explanatory annotation and in his careful headnote to each letter giving details about the manuscript. His treatment of the text, however, raises some questions. Although he indicates, with brackets, emendations of words, he makes numerous silent emendations of punctuation and spelling. The policy is to retain Walpole's punctuation (but not that of his correspondents) and his spelling of proper names, but "to normalize other spellings and capitalization." One of the justifications offered is "a considerable gain in readability and appearance." The "considerable" is debatable, but readability is at any rate the standard argument for modernization—although the question remains why thorough modernization is not therefore undertaken to make the text even more readable. Another justification is more troublesome: "What is amusing and 'flavoursome' in small doses becomes wearisome in large, and it imparts an air of quaintness to a text which was not apparent to the correspondents themselves" (p. xxxvi). Surely no serious reader will regard any characteristics of a particular time in history as merely quaint; all characteristics are part of the evidence for historical understanding, and it is an insult to the reader to suggest that he can better perceive the intended tone of a letter if certain features of it have been altered for him.

Similar problems arise in many other literary editions. Theodore Besterman's edition of Voltaire's Correspondence (Institut et Musée Voltaire, 1953-65)[65] is famous because of its enormous size; the completion of an edition of 21,000 letters is indeed an accomplishment, to say nothing of bringing it out a second time in a revised "definitive edition" (Correspondence and Related Documents, 1968-76). Although Voltaire's alterations are recorded in notes, the treatment of the main text is disappointing: the first edition reports that apostrophes are inserted and "a minimum of capital letters and punctuation, where lacking" (p. xiii), and the revised edition follows the same policy (pp. xvii-xviii; Besterman says, "without attempting to modernize, I have introduced a measure of regularity"). The edition offers an example of the kind of inconsistency which partial modernization almost invariably leads to: "When Voltaire used an accent it has been reproduced even if

it now looks wrong, but when he omitted one it has been supplied."[66] Gordon S. Haight, in The George Eliot Letters (Yale University Press, 1954-55), says that his "principal concern has been the reader's convenience" (p. xxxv); though he retains spelling, he treats punctuation "a little more freely, adding or deleting an occasional mark to save rereading." In the same year Allan Wade, in The Letters of W. B. Yeats (Hart-Davis, 1954), argues for "correcting" both spelling and punctuation on the grounds that Yeats was poor at both. To retain Yeats's spelling would "in the long run appear merely tediously pedantic" (p. 16); Yeats's "faults" in punctuation, he says, "I have silently corrected, and I have not hesitated to introduce commas into sentences which, without them, are either ambiguous or almost meaningless" (p. 17)—obviously running the risk of giving those sentences meanings which Yeats did not intend. E. S. de Beer does not attempt to normalize punctuation in The Diary of John Evelyn (Clarendon Press, 1955) but does supply "without note a certain amount of punctuation" aimed "solely at intelligibility," arguing that for "strict linguistic study" one must consult the manuscripts (p. 68). In The Swinburne Letters (Yale University Press, 1959-62), Cecil Y. Lang says, "I have always tried to make readability my first concern" (p. xlix), and he follows the practice of reproducing printed texts "faithfully" but making some alterations in manuscript texts.

The same approach continues to appear in literary editions of the 1960s and 1970s. Harry T. Moore, in The Collected Letters of D. H. Lawrence (Heinemann, 1962), comments on some of Lawrence's seeming deficiencies of punctuation and states, "rather than belabour the reader by calling attention to these peccadilloes I have quietly done what was needed" (p. xxi). Rupert Hart-Davis silently emends spelling, capitalization, punctuation, and paragraphing in The Letters of Oscar Wilde (Hart-Davis, 1962). Wilde's habitual dashes, he says, "make the letters difficult to read, and I have re-punctuated normally as the sense seems to demand" (p. xi). Wilde also liked to capitalize words beginning with "t" and "h," "presumably because he enjoyed making those particular capitals more than their lower-case equivalents." Hart-Davis believes that "to perpetuate this whim would only irritate the reader," and he has "followed the standard usage wherever the capital clearly has no significance." But he has just told us what significance those two capitals have. Why should a writer not be allowed to indulge his

"whims" in a letter? It is a perfect place for him to do so, because the text will not have to go through the hands of a publisher or a printer before reaching the intended audience. The Letters of Henry Wadsworth Longfellow (Belknap Press of Harvard University Press, 1966- ), edited by Andrew Hilen, is like some of the NHPRC editions in silently correcting "mere slips" but not altering errors or variations in proper names. "Occasionally," Hilen says, "I have silently provided punctuation, or deleted it, in order to clarify meaning" (p. 13). Leon Edel, in the Henry James Letters (Belknap Press of Harvard University Press, 1974- ), makes silent corrections "where they were obviously called for" (p. xxxv), but in the letters of the young James he retains "relevant misspellings" because "they are a part of the flavor of the letters." Unfortunately he does not extend this argument to the later letters.

I do not wish to prolong this litany unnecessarily. I have merely tried to cite a sufficient number of examples to show that there is a considerable body of editors whose approach to the editing of letters and journals is in the spirit of the policies set forth in Boyd's Jefferson and the Harvard Guide. And it is by no means only the historians who fall into this group. While it is true that most of the NHPRC editions—with only a few exceptions—are of this type, there are certainly a great many literary editors whose practice coincides with that of the NHPRC editors.[67] Most of the editions mentioned are praiseworthy in many respects: most of them reflect thorough research and exemplary annotation. But their treatment of the actual texts is relatively casual and unsophisticated by comparison. It is clear, from this survey, that one widely followed approach to editing documents assumes that some modernization is essential and that a silently modernized or corrected text can serve most purposes of historical study. The assumption is made, however, without adequate consideration of the role which such features as spelling and punctuation play in private documents and the extent to which they constitute part of the total body of evidence that the historian needs to have at his disposal. What I have said about these editions can perhaps begin to indicate why their textual policies are bound to seem unsatisfactory to anyone who has given careful thought to textual matters and the nature of written communication.

III

At the time when Boyd's Jefferson was about to come out and the NHPC to be revitalized, there were some editions other than the Walpole

which might have been turned to as models, and it is unfortunate that they did not have more influence at that strategic moment. The Walpole edition, because of the enormous size of the undertaking, may have seemed a closer parallel to the large editions which were projected to accommodate the masses of papers accumulated by statesmen; but certain smaller editions could have offered a sounder textual policy. Paget Toynbee and Leonard Whibley's three-volume edition of Correspondence of Thomas Gray (Clarendon Press, 1935), for instance, states, "The text is printed as Gray or his correspondents wrote it, with the spelling, punctuation, use of capitals, and abbreviations of the originals" (p. xxiii); and Ralph L. Rusk's six-volume edition of The Letters of Ralph Waldo Emerson (Columbia University Press, 1939) requires little space for an explanation of editorial policy, for Rusk says simply, "I have tried to print a literal text, with no interpolated corrections or apologies" (p. v). Gordon N. Ray's four-volume edition of The Letters and Private Papers of William Makepeace Thackeray (Harvard University Press, 1945-46) is a model edition. Ray presents "a literal text" and is not bothered, as so many editors seem to be, by sentences which end with dashes rather than periods. In an admirable statement, he sums up why it is important to preserve in print the spelling and punctuation of the manuscripts: Although the details which lead to this conclusion might have to be altered somewhat in the case of other writers, it is difficult to see how the conclusion itself could be improved upon as a guiding statement for all editors of letters.

Another notable edition, which began to appear just after the first volume of the Jefferson but early enough that it could have been influential in the formative days of the new NHPC, is Elting E. Morison's eight-volume edition of The Letters of Theodore Roosevelt (Harvard University Press, 1951-54). The letters are "printed as written without further indication of Roosevelt's frequent and startling departures from the norm of accepted usage in spelling." Morison, like Ray, has given careful thought to the rationale for such a policy, and he makes an intelligent statement of the case:

Also during these years historical editors in particular should have been aware of the excellent example being set by Clarence E. Carter in his major project, The Territorial Papers of the United States (Government Printing Office, 1934- ); it was in 1956, in the introduction to the twenty-second volume, that he made an important statement of his aim of "literal reproduction."[68] Even more persuasively than in his Historical Editing, he pleads the case for an unmodernized text: Modernization, he rightly concludes, "tends to obscure rather than to clarify." Some literary editors, too, were commenting in the 1950s on the importance of exact transcription of letters and journals. R. W. Chapman, reproducing the manuscripts "as closely as typography admits" in his three-volume edition of The Letters of Samuel Johnson (Clarendon Press, 1952), points out the value of errors: Kathleen Coburn, at the beginning of The Notebooks of Samuel Taylor Coleridge (Pantheon Books [later Princeton University Press], 1957- ),[70] agrees, stating that "Slips of the pen are respected, in conformity with the argument of Dr. Chapman in editing Johnson, that such things have their own interest and significance" (p. xxx), and she adds that Coleridge himself remarked on this point.[71] Howard Horsford, editing Melville's Journal of a Visit to Europe and the Levant (Princeton University Press, 1955), suggests the importance of precision in his careful descriptions of cancellations and his thorough discussion of the difficulties of Melville's handwriting. Hyder Edward Rollins, in The Letters of John Keats (Harvard University Press, 1958), notes that "Keats penned his sentences rapidly and spontaneously, not carefully and artfully" (p. 17), and therefore his "queer punctuation" and "occasional grammatical slips" are indicative and should not be rectified. And Thomas H. Johnson's edition of The Letters of Emily Dickinson (Belknap Press of Harvard University Press, 1958) presents all holograph letters "in their verbatim form" (p. xxv), which involves many dashes.[72] With editions of this kind available to point the way, the NHPC editors of the late 1950s were unwise to turn in a different direction.

In 1960 four editions appeared which, in their somewhat differing ways, represent the approaches followed by the best of the literary editions of the 1960s and 1970s. All are characterized by scrupulous reporting of details of the manuscripts, but what distinguishes a number of them from most earlier careful editions of manuscripts is a system—not unlike that long in use for printed copy-texts—whereby certain categories of emendation can be allowed in the text, with the original readings preserved in notes or lists. Henry Nash Smith and William M. Gibson's edition of Mark Twain-Howells Letters (Belknap Press of Harvard University Press, 1960) involves no normalizing of punctuation or spelling, and it records significant cancellations. James Franklin Beard, in The Letters and Journals of James Fenimore Cooper (Belknap Press of Harvard University Press, 1960-68), does alter some punctuation

for clarity and amend some spellings, but these editorial alterations are recorded in footnotes (except for a few specific categories),[73] while "legible cancellations" are incorporated into the text within angle brackets. The text of Merrell R. Davis and William H. Gilman's edition of The Letters of Herman Melville (Yale University Press, 1960) also incorporates a few emendations of punctuation for clarity, but they are all listed in the meticulous textual notes at the end. These notes additionally include such details as foreshortened (hastily written) words: one can learn from them that what appears in the edited text as "thing," for example, resembles "thng" in the manuscript (merely misspelled words, of course, are not altered). Cancellations are all transcribed, either in the text (in angle brackets, along with braces for insertions) or in the textual notes. The Journals and Miscellaneous Notebooks of Ralph Waldo Emerson (Belknap Press of Harvard University Press, 1960- ), edited by William H. Gilman et al., goes farther in the use of symbols to record as much textual information as possible within the text. It aims to come "as close to a literatim transcription" as is feasible in print (p. xxxviii) and does indicate the stages of Emerson's revision with great precision; some categories of editorial alteration, here too, are not labeled in the text but are reported in textual notes at the end. The volumes of Emerson Journals which appeared after the CEAA emblem was instituted have received the emblem, and later CEAA editions of journals further illustrate the modern practice of the full recording of manuscript characteristics. Washington Irving's Journals and Notebooks (University of Wisconsin Press, 1969- ), as edited by Henry A. Pochmann et al.,[74] is uncompromisingly literal (it respects Irving's lowercase sentence openings, for example) and contains one of the most thorough discussions in print (pp. xix-xxvi) of the problems involved in exact transcription (amply demonstrating that the process is not mechanical). Claude M. Simpson's edition of The American Notebooks of Nathaniel Hawthorne (Ohio State University Press, 1972), as is usual with CEAA volumes, makes some emendations in the text but records them, as well as authorial alterations of the manuscript, in lists at the end. And Mark Twain's Notebooks & Journals (University of California Press, 1975- ), as edited by Frederick Anderson et al., offers an excellent discussion of editorial procedures (pp. 575-84) and is a model of how to combine the emendation of certain obvious errors (always listed at the end, accompanied by "doubtful readings") with the preservation of "the texture of autograph documents" (which contain "irregularities, inconsistencies, errors, and cancellations").[75]

These are not the only praiseworthy editions of letter and journals in the 1960s and 1970s,[76] and a few others deserve mention not simply for their high standards of literal transcription but for the cogency of their statements justifying that approach. Shelley and His Circle (Harvard University Press, 1961- ), edited from the holdings of the Carl H. Pforzheimer Library by Kenneth Neill Cameron (later by Donald H. Reiman),[77] surpasses all these other editions in its efforts to reproduce in type the features of manuscripts—printing careted material, for example, above the line and in smaller type. The aim is "the traditional one" of producing "a foundational text . . . from which other editors may depart as they wish," and the rationale is stated with great effectiveness:

Herbert M. Schueller and Robert L. Peters, in their edition of The Letters of John Addington Symonds (Wayne State University Press, 1967-69), give some additional reasons for offering a literal text: In The Journals and Letters of Fanny Burney (Clarendon Press, 1972- ), Joyce Hemlow[78] allows errors to stand "as the normal hazards of hasty or spontaneous writing" and believes that "the twentieth-century reader probably needs few such props" as modernization (p. lviii). Leslie A. Marchand, in his editorial introduction to "In my hot youth": Byron's Letters and Journals (Murray, 1973- ), adds further to the strength of the case: Marchand, like most of the other advocates of this point of view, adds that the unmodernized text is not difficult to read; but the reasons for not modernizing, it is clear, are of sufficient weight that the question of whether the resulting text is somewhat difficult to read is of secondary importance.[79]

The statements quoted here, which make a number of different points and refer to a variety of periods and kinds of material, add up to an impressive argument and are no doubt sufficient in themselves as a criticism of the partially modernized and silently emended editions described earlier. Merely to juxtapose comments on editorial policy from the two kinds of edition is to show up the weaknesses of attempting to justify modernization and silent alterations in scholarly editions of historical documents. But it will perhaps be useful to try to sort out more clearly the issues involved, especially since there has been so little discussion of the matter, at least in connection with the editions of statesmen's papers. Although a voluminous literature has grown up around the NHPRC editions, it contains very little commentary on textual procedures, and what there is seldom touches on fundamental questions. The NHPRC editions have probably been more extensively reviewed than the CEAA editions; but in both fields it is difficult to find reviewers who can adequately analyze textual policies, and the reviews of NHPRC volumes in particular have almost consistently slighted—or ignored completely, except for a perfunctory word of praise—the textual aspects of the editions.[80] The historical significance of the contents of these editions

and the quality of the explanatory annotation—on which the reviews concentrate—are important matters, but the way in which the text has been established and presented is surely of first importance in evaluating an edition.

Considerable criticism has been directed at the NHPRC editions but for essentially irrelevant or trivial reasons. One objection, raised by Leonard W. Levy, for example, in his reviews of the Madison edition, is that the explanatory annotation is carried to excessive lengths.[81] Another criticism questions the choice of material to be edited. J. H. Plumb, among others, believes that too much attention is paid to unimportant documents,[82] and Jesse Lemisch argues that the pattern of figures chosen to be edited reflects a bias "in the direction of white male political leaders."[83] Whatever justice there may be to these opinions, they have nothing to do with the quality of the editions themselves. If the annotation is accurate and helpful, it will be of use, and there is little point in wishing there were less of it; and any document or figure is of some historical interest. Individual tastes regarding what material is worth spending time on, and judgments about priorities, will naturally vary; one may deplore another's choice of subject, but it is unrealistic to criticize accomplished work for having usurped time better spent on something else. Still another frequent complaint is that letterpress editions

are too expensive and time-consuming to produce and that microfilm publication of the documents would be cheaper, faster, and more appropriate.[84] Certainly the well-established microfilm publication programs of the NHPRC, the National Archives, the Library of Congress, and various state historical societies are to be praised;[85] but making photographic reproductions of document collections widely available is by no means a substitute for editing those documents,[86] as Julian Boyd and Lester Cappon, among others, have effectively pointed out.[87] The skilled editor, employing his critical intelligence and fund of historical detail, establishes a text which marks an advance in knowledge over the mere existence of the document itself. Microfilm editions of unedited documents do not obviate true editions; but editing takes time, and one is back at the earlier question of individual priorities for spending time.

These controversies are really peripheral to the main business of editing. Since individual priorities do differ, anyone may decide not to become an editor; but for those who elect to undertake editorial projects, surely the first priority is the text itself, its treatment and presentation. And when one considers the divergences of textual policy which

distinguish most NHPRC editions from the CEAA and CSE editions, the first question to ask is whether there is an essential difference between the materials of historical interest and those of literary interest that would necessitate differing treatments. Nathan Reingold, in a letter to the American Scholar (45 [1976], 319) commenting on Peter Shaw's article, suggests such an explanation, pointing out that the CEAA editors work with printed texts, whereas the historical editors for the most part deal with thousands of "scrappy and informal" bits of manuscript. It is true that the bulk of the CEAA and CSE editions are of works which have previously appeared in print,[88] but those editions do include numerous volumes of manuscript letters and journals, and of course in the literary field in general many editions of such material exist. It may also be true that letters predominate in editions of statesmen's papers, but the comprehensive editions do include speeches, reports, and other works of a public nature normally intended for distribution in printed form. Is a letter written by a literary figure in some way fundamentally different from a letter written by a statesman? Both are historical documents: literary history is still history, and all letters offer historical evidence. And either letter may be regarded as "literature": a statesman may produce masterly letters, and a literary figure may write pedestrian ones. Is a novel or a poem fundamentally different from a work which a statesman prepares for publication? At their extremes, imaginative literature and factual reporting seem to be different kinds of communication, but in between there is a large area in which they overlap. No clear line can be drawn between writing which is "literature" and writing which is not. Certainly the editor of an individual's whole corpus of papers is likely to encounter writings which can be regarded either way: some of Franklin's and Jefferson's best-known writings have often been classified and analyzed as literary works, whereas Hawthorne's Life of Franklin Pierce and Whitman's journalism are not always considered literature. There sometimes seems to be an assumption that close attention to textual nuances—and thus the need for recording textual details—is more vital to the study of literary works and other writings by literary figures.[89] Apparently that is part of Frederick B. Tolles's point when he criticizes the "zeal" of the editor of George Mercer's papers for her "reverent handling" of the texts: "it seems important to remind ourselves," he says, "that they are not sacred codices of Holy Writ or variant quartos of Hamlet."[90] He also means, of course, that Mercer's papers are not as important as Hamlet. But neither the importance nor the literary quality of a piece of writing determines the amount of attention that must be paid to nuances of expression; if one seriously wishes to understand a text, whatever it is, no aspect of it can be slighted.[91] There is no fundamental distinction, then, from a textual point of view, between the materials edited by the historian and those edited by the literary scholar. Letters, journals, published works, and manuscripts of unpublished works fall into both fields; all of them are historical documents, and any of them can be "literary."[92]

A distinction does need to be made, but not between literary and historical materials. Rather, the important distinction is between two kinds of writings which both historians and literary scholars have to deal with: works intended for publication and private papers.[93] Works intended for publication are generally expected to conform to certain conventions not applicable to private documents. For example, a finished work is expected to incorporate the author's latest decisions about what word he wishes to stand at each spot; in a private notebook jotting, however, or even in a letter to a friend, he can suggest alternative words and is under no obligation to come to a decision among them.[94] Similarly,

he can spell and punctuate as he pleases in a private document, but he will have difficulty getting a work published if it does not conform, at least to some extent, to current standards. Whether or not a writer really wishes to have his manuscript altered by a publisher's editor or a printer to bring it into such conformity is a complex question of intention, and editorial debate on this issue is likely to continue. Some editors feel that a surviving completed manuscript of a published work is the proper choice for copy-text because it reflects the author's characteristics more accurately, while others feel that the published text should be the copy-text because the author expected his manuscript to be subjected to the normal routines of publishing. No doubt the answer will vary in different situations, but this is not the place to explore the question.[95] The point here is to contrast that situation with the very different one which exists for private documents. In the case of notebooks, diaries, letters, and the like, whatever state they are in constitutes their finished form, and the question of whether the writer "intended" something else is irrelevant. One still sometimes hears the argument that an editor must make alterations in such documents because the writer would have expected to make changes in them for publication. If the writer had in fact prepared them for publication, they would then no longer be private documents but works intended for the public; they would have passed through the usual steps leading to publication, as any other work would, and the author probably would have made alterations in them, since the original documents would be parallel with the rough or semifinal drafts of other kinds of works. But when the writer did not prepare his own letters or diaries for publication, they remain private papers. The scholarly editor who later wishes to make them public is not in the same position as the writer or the writer's contemporary publisher. Not only is it impossible for him to know what the writer or his publisher would have done to them; but if he presents them as anything more polished or finished than they were left by the writer, he is falsifying their nature. A journal, as a piece of writing for one's own use, is in its final form whenever one stops adding to it; a letter, as a communication to a private audience of one or two, is in its final form whenever it is posted. The writer is under no constraint to conform to any particular convention in these writings, except to the extent that he hopes a letter will be comprehensible to its recipient. Any idiosyncrasies in them—however contrary to the standards for published works—are an essential part of their character.

These considerations lead to the conclusion that a scholarly edition of letters or journals should not contain a text which has editorially been corrected, made consistent, or otherwise smoothed out. Errors and inconsistencies are part of the total texture of the document and are part of the evidence which the document preserves relating to the writer's habits, temperament, and mood. Modernization, too, is obviously out of place. While it is not the same thing as the correction of errors or inconsistencies, the line between the two is often difficult to establish. Even in many published works the spelling, punctuation, and capitalization are inconsistent, and to assume that the writers or publishers intended them to be consistent or cared whether they were consistent or not is to read into the situation a point of view held by many people today but one that has apparently not always been held. Correcting errors is somewhat different, since by definition an error is not intended; but it is frequently difficult to avoid a modern bias in deciding what constitutes an error. Editors of published works are increasingly recognizing that to regularize or to make certain supposed corrections is to modernize.[96]

In the case of private documents, then, where errors and inconsistencies are an integral part of the text, the argument against modernization is doubly strong. Indeed, the position that the text of a scholarly edition of any material can ever be modernized is indefensible. Many editors of literary works have long understood this fact,[97] and it is difficult to explain why such a large number of editors of private documents have, during the same period, neglected it. They are not always cognizant of a distinction between correcting and modernizing; but to subject

such documents to either is to violate their integrity. Ultimately the position of these editors rests on a failure to grasp the significance of punctuation, capitalization, and spelling as functional elements of written expression. They think, as a result, that they can make alterations "for clarity" and "for the reader's convenience" without affecting the content of the document in any important way. In most instances, they greatly exaggerate the difficulty of reading the original text, and it is hard to see how the reader's "convenience" is really served by changing a dash to a period, an ampersand to "and," or an upper-case letter to lower case.[98] What, in the end, do they accomplish, other than depriving the reader of the experience of reading the original text? Is the text "clearer" as a result of their labors? Frequently it is less clear, because documentary editors rarely modernize more than a few features, leaving the text with a confused and unhistorical mixture of elements that clash with each other.[99] What is intended as a help becomes a barrier between the reader and the text he is interested in reading. Anyone who has examined a number of the partially modernized editions of letters can only react with incredulity at the things which editors seem to think readers need to have done for them. The modernizing editor is both condescending and officious: he assumes that the reader is not serious enough to persevere in reading a work if the punctuation, capitalization, and spelling do not conform to present-day practice, and his belief in the necessity of making changes blinds him to the triviality and senselessness of many of his alterations.[100] Modernization, or partial modernization, is clearly incompatible with the goals of the scholarly editing of private documents—a fact which points to the most tragic weakness of many of the NHPRC editions.

Once it is settled that letters and journals are not to be presented in a corrected or modernized text, there still remains the question of whether editorial symbols are to be employed within the text or whether the text is to be free of such symbols ("clear text"). Even though no corrections are made,[101] there will be occasions when the editor needs to introduce a comment, such as "word illegible," "edge of paper torn, eliminating several words," or "written in the margin." Whether these explanations are entered in brackets in the text or printed as appended notes is to some extent merely a mechanical matter. But there is a theoretical aspect to the question. It is often argued that novels, essays, poems, and other works intended for publication should be edited in clear text, because such works are finished products, and the intrusion of editorial apparatus into the text (recording emendations or variants, for example) would be alien to the spirit of the work. For this reason the CEAA editions of this kind of material are in clear text, with the textual data relegated to lists at the ends of the volumes.[102] Private documents are different, however, in that they are often characterized by not being smooth—by containing, that is, false starts, deletions, insertions, and so on. The problem of how to handle deletions gets to the heart of the matter. Simply to leave them out, as is often done (or done on a selective basis), is indefensible, since they are essential characteristics of private documents.[103] One solution would be to leave them out of the text and report them in notes. But to do so would make the text appear smoother than it is; no evidence would be lost, but the reader would have to reconstruct the text of the document, which is after all of primary interest. If, on the other hand, the deletion is kept in the text but clearly marked as a deletion (with angle brackets or some other device),

the nature of the original is more accurately rendered in print. In reading the original, one would see a phrase with a line through it, for instance, and then read the phrase which replaced it; by keeping the deleted matter in the text, the editor allows the reader to have the same experience. But when canceled matter is recorded, it is essential at the same time to indicate whether the replacement (if any) occurs on the same line or is inserted above the line, so that the reader can tell whether the revision was made in the process of writing the words or perhaps at a later time.[104] A number of the NHPRC editions devote some attention to cancellations, but their frequent failure to specify interlinear insertions makes it impossible for the reader to use properly the texts of the cancellations which they do provide. Some situations can become very complex and may require an editorial description of what has happened as well as editorial symbols. This description might well be placed in a note rather than in the text; but since the text will contain editorial symbols in any case, one could decide to include editorial comments—at least the brief ones, like "illegible"—within the text.[105] The crucial point is that if a private document is presented in clear text it loses part of its texture.[106]

The argument thus far has assumed that for any given text the evidence available to the editor is a single document in the hand of the author. In those cases the editor's goal is to reproduce in print as many of the characteristics of the document as he can. The goal is not, in other words, to produce a critical text, except to the extent that judgment is involved in determining precisely what is in the manuscript. And judgment is inevitably involved: the editor of Shelley and His Circle points out that if a word clearly intended to be "even" looks more like "ever" it is still transcribed as "even." Distinguishing between

an actual misspelling or slip of the pen and merely indistinct or hasty handwriting requires careful judgment. The editor, even in presenting an "exact" transcription of the text of a document, must keep the writer's habits and intention in mind, if he is to be successful in discovering what that text actually says at difficult spots. If, for instance, a manuscript clearly reads "seperate," there is no doubt that the author wrote the word with a middle "e"; whether or not the author intended to misspell is irrelevant, so long as one agrees that an author's errors in private documents are of interest and should be preserved. But if the word only looks like "seperate" because the author has been careless in forming an "a" in the second syllable, the editor who prints "seperate" is neither transcribing accurately nor respecting the author's intention. In a case like "even"/"ever," the intention as determined by the context plays a greater role: deciphering handwriting and understanding the content are inseparable.[107] It is frequently necessary, therefore, even in connection with a so-called "literal" transcription, for an editor to append notes recording editorial decisions, if the reader is to be fully apprised of the state of the manuscript. But these decisions, it should be clearly understood, result from the effort to determine what the text of the document actually says, not what the editor believes it ought to say.

The situation is different, however, when the textual evidence is not limited to a single holograph document; there may be several drafts, versions, or copies, and they may be in the hand of a copyist or in printed form. In such cases the editor has a fundamental decision to make about the nature of his edited text: is it still to be a transcription of the text of a single document (with evidence from related documents given in notes), or is it to be a critical text which attempts through emendation (based on a study of all the documents) to represent the writer's intentions more fully than any single surviving document can? This decision will rest on the nature of the surviving documents—on their relative authority and completeness. When there are various versions or drafts of a letter in the author's hand, the editor would normally choose the one actually posted, if it survives, or the retained copy or latest surviving draft if it does not, as the document to be edited; variant readings and canceled matter in the other documents might then be added in notes, but—in line with the reasoning suggested above—they would not be emended into the text itself. If, on the other hand, the extant version or versions of a text are not in the author's hand—as when a letter

survives only in several scribal copies or in print—the editor is faced with the problem of distinguishing those features which reflect the author's intention from those which result from the habits and errors of another person (the copyist, the compositor, the printer's or publisher's reader, and so on). Since the interest is in the characteristics of the author's expression, not in those of a copyist or compositor, this problem is worth solving. For if an editor presents the text of a nonholograph document in an exact transcription, as he would that of a holograph document, he is respecting equally its authorial and its nonauthorial features; but if he attempts, so far as his evidence allows, to remove some of the nonauthorial features, he comes that much closer to offering what was present in the author's manuscript.

Editors of works which were intended to be made public commonly have this problem to deal with. When confronted with a printed text or texts, or with a printed text which differs from the author's manuscript, or with a scribal copy or copies, these editors frequently take it as their responsibility to evaluate the evidence (on the basis of their specialized knowledge of the author, his time, and the textual history of the work) and then to choose and emend a copy-text so as to obtain a maximum number of authorial readings and characteristics and a minimum number of nonauthorial ones.[108] The CEAA editions of works intended for publication have taken this approach, on the ground that more is to be gained by encouraging a qualified editor to apply what judgment and sensitivity he has to the problem of determining the author's intended text than by requiring him to reproduce the text present in a particular surviving document. Some mistakes are bound to result, but in general a text produced in this way is likely to come closer to what the author intended than a single documentary text could possibly do. (An accompanying record of emendations and variant readings is naturally important, so that the reader can reconstruct the copy-text and reconsider the evidence for emending it.) Editors of letters and journals will perhaps less frequently encounter similar situations, but when they do they should remember that preparing a critical text of nonholograph materials is not inconsistent with a policy of presenting a literal text of holograph manuscripts. Rather, it is an intelligent way of recognizing that a consistency of purpose may require different approaches for handling different situations. The aim of an edition of a person's letters and journals is to make available an accurate text

of what he wrote; that goal cannot be achieved as fully for nonholograph documents as for holograph ones, but it is the editor's responsibility to come as close as he can in either case.[109]

When Peter Shaw claims that the NHPRC editions show more respect for historical fact than do the CEAA editions, he fails to recognize that an edition with a critical or "eclectic" text does not necessarily conceal historical facts and that an edition of a single documentary text does not necessarily reveal all relevant facts. Whether they do so or not depends on their policies for recording textual data.[110] CEAA editions are required to include textual apparatuses which contain records of all editorial emendations as well as several other categories of textual information;[111] most of the NHPRC editions, on the other hand, incorporate several kinds of silent emendations.[112] Readers of the former are able to reconstruct the original copy-texts and are in possession of much of the textual evidence which the editor had at his disposal; readers of the latter cannot reconstruct to the same degree the details of the original documents and are not provided with carefully defined categories of textual evidence on a systematic basis. The CEAA editors fulfill an essential editorial obligation: they inform their readers explicitly

of what textual information can and what cannot be found in their pages.[113] The truth is, therefore, that the CEAA editions are actually more respectful of documentary fact, and at the same time they recognize more fully that fidelity to a writer's intention demands, under certain circumstances, an eclectic approach to the documents. Comparing a CEAA edition of a novel with an NHPRC edition of letters creates a false opposition; but when CEAA and NHPRC editions of similar materials—two volumes of letters[114] or two volumes of works intended for publication—are compared, the CEAA volumes characteristically exhibit a more profound understanding of the problems involved in textual study and a greater responsibility in treating textual details. The NHPRC editors have undeniably been successful in the nontextual aspects of their work, and the CEAA editors could learn from them in regard to explanatory annotation. But in textual matters the CEAA editors are far in the lead.

This state of affairs is a depressing reminder of how little communication sometimes exists between fields with overlapping interests. In 1949, the year before the first volume of the Jefferson edition appeared, Fredson Bowers commented on the importance of textual study for all fields of endeavor:

In the twentieth century scholars of English literature—especially of Elizabethan drama—have taken over from the Biblical scholars and classicists as leaders in the development of textual theory and practice; and in the last generation the editing of nineteenth-century American literature has been a focal point in this continuing tradition. But the principles that have been emerging are not limited in their applicability to the field of literature. Students in all fields have occasion to work with written or printed documents, and they all need to have the habit of mind which inquires into the "purity and accuracy" of any document they consult. The NHPRC volumes have been singled out here because they constitute a prominent block of modern editions and can serve as an instructive example: the difference between the way American statesmen and American literary figures have recently been edited is a striking illustration of how two closely related fields can approach the basic scholarly task of establishing dependable texts in two very different ways, one of which seems superficial and naïve in comparison to the other. But history and literature are not the only fields that would mutually profit from a more encompassing discussion of textual problems; many editorial projects are now under way in philosophy and the sciences, and the fundamental questions which editors must ask are the same in those fields also. Editing is of course more than a matter of technique; a text can be satisfactorily edited only by a person with a thorough understanding of the content and historical and biographical setting of that text. Nevertheless, there is a common ground for discussion among editors in all fields. The time for closer communication of this kind is overdue; not only editors but all who study the written heritage of the past will benefit from it.

Notes