Perhaps the most important criterion to be met by any document published in a scholarly edition is that it is, indeed, the intellectual product of the person or organization purported to be its author. Editors must rule out forgeries and misattributions while simultaneously making every effort to identify documents not previously recognized as the work of their subjects.

The papers of well-known individuals, like the works of great artists, are most likely to become the targets of forgers. The commercial market for letters and other manuscripts penned by heads of state, military and naval leaders, famed scientists and inventors, and literary lights is centuries old in the Western world, and forgery is almost as ancient as that trade. Some editors must be more conscious of the danger of forgeries than others. The editors of the Papers of Abraham Lincoln, for instance, maintain an extensive file of known Lincoln forgeries.

Joe Nickell’s Pen, Ink, and Evidence: A Study of Writing and Writing Materials for the Penman, Collector, and Document Detective and its bibliography provide a useful introduction to the history of documentary forgers. Experienced dealers or auction house experts familiar with an edition’s subject can also offer hard-earned knowledge of such fakes.

Almost as common is the problem of ruling out published or unpublished works that, while historically genuine, have been erroneously credited to an edition’s subject. Perhaps the best-known recent discovery in this field came when editors of the two-volume Political Correspondence and Public Papers of Aaron Burr (1983) revealed that the cipher letter from Burr to James Wilkinson that led to Burr’s denunciation as a traitor was not, in fact, written by Burr but by his associate Jonathan Dayton. Less well publicized but more historically significant was the Madison Papers editors’ decision to omit from volume 17 of their series the 1799 Virginia Resolutions and Address of the General Assembly. On the basis of both contextual evidence and a statistical analysis of word frequency from a model in Frederick Mosteller and David Wallace, Inference and Disputed Authorship: The Federalist, they became convinced that Madison was not the author of these statements.

Traditionally, scholars have shown more concern for the attribution of works to famous authors than to renowned politicians or generals, and attributions of anonymous and pseudonymous literature to novelists and poets are more likely to be subjected to exhaustive investigation. A notable exception is Benjamin Franklin, whose newspaper contributions were the subject of J. A. Leo Lemay’s monumental Canon of Benjamin Franklin, 1722–1776: New Attributions and Reconsiderations. For any published author, computer analysis of usage may help corroborate or disprove attributions.

In most cases, the editor will not be the first scholar to examine the subject in question, and questions of attribution will be addressed throughout the editorial project, from the time of collection until the point when informational footnotes are added to the editorial text. The editors of the Papers of John Adams were able to credit Adams with drafting a March 1773 message of the Massachusetts House of Representatives only after they had reviewed all of Adams’s correspondence concerning the edition (1:312–13).

The decision to omit from or include in an edition unattributed verbal documents must be weighed carefully. It involves a combination of factors including writing style and word usage and circumstantial and contextual evidence—all of which presuppose that the editor is familiar with a large cross-section of the documents. It means that verbal documents have to be scrutinized as carefully as graphic works have been when experts authenticate drawings and paintings.

It is not only published works that may raise questions of attribution. Scholars who deal with the papers of any figure who commanded a large civilian or military staff must establish guidelines that define when a document may be considered that figure’s work and when it should be credited to a junior associate who drafted the document, which may have been dispatched without a moment’s review by the person whose signature it bears. The editors of the papers of Dwight D. Eisenhower and George C. Marshall admit that their task here was made easier by the clear chain of command both generals demanded. They were sticklers not only for precisely established office procedures but also for putting those procedures on record. This meticulousness enabled scholars to understand patterns of documentation and to recognize which set of initials or stamps identify documents by Marshall or Eisenhower.

Even those editors, however, faced a decision requiring Solomon-like judgment when they confronted a radio message to Douglas MacArthur of 7 February 1942. As a member of Marshall’s staff, Eisenhower drafted the dispatch, and a typescript copy of that draft was submitted to Marshall, who then covered margins and interlinear spaces with his additions and changes. So extensive were Marshall’s revisions that a complete retyping was necessary before the message could be coded and dispatched. Here the two sets of editors saw two documents with two different authors. The Eisenhower edition (1:101–3) printed the typescript portions of the draft, noting Marshall’s revisions in footnotes. The text in the Marshall edition (3:100–102) was based on the typescript that incorporated those changes, the version that reflected Marshall’s wishes.

Authentication and attribution are only the beginning for the editor who must choose source texts. Only a few candidates for source texts will be excluded here, and the process of evaluation goes on.

II. Types of Documentary Materials

The methods used to create the documents at hand are likely to dictate criteria for evaluating source texts. These rules for selection, like all editorial procedures, should be stated clearly in an edition’s introduction. For modern and Anglo-American documents, the criteria generally involve little more than common sense, and they relate to how a document came into being.

A. Inscribed Sources: Handwritten or Typewritten Materials

1. The manuscript or a reliable photocopy or scanned image is to be preferred over any later scribal copies or transcriptions as the source text.

2. If the original has been lost or destroyed, contemporary copies are preferred over later ones unless evidence demonstrates that later copyists both had access to the original and were more accurate than the earlier scribes.

3. In general, the most nearly final version of a document is the preferred source text. Editors can take comfort in the thought that variants can and should be noted in the editorial apparatus. By choosing one version over another, editors do not deny readers access to significant differences among versions of a document. The source text is simply the one that serves as the best working basis for the edition and most closely meets the needs of the edition’s audience.

Other criteria for establishing priorities among such materials are listed below.

1. Letters

The first order of preference is given to a version of the original letter (preferably signed to denote authorial approval) known to have been received by its addressee. The best evidence for identifying a recipient’s copy of a letter may be an attached address leaf (for letters before the mid-nineteenth century) or a mailing envelope (for later correspondence), the recipient’s endorsement, or the location of the letter in the addressee’s papers. Even when no address leaf or envelope survives to prove that a copy of a letter passed through the mails, fold patterns and other physical evidence often mark a particular document as the recipient’s copy. The editor familiar with a subject’s letter-writing habits will generally be able to identify “finished” versions with little trouble.

A special problem may arise with the correspondence of statesmen of earlier centuries who were assigned to foreign posts in times of national emergency. At such periods, diplomats frequently sent duplicate or even triplicate or quadruplicate signed copies of their dispatches to friends or government offices at home. When two or more such multiple addressees’ copies survive, the editor has to decide which version to use as a source text. The decision may be based on the completeness of the variant copies or the care that the letter’s author or secretary used in copying the dispatch. A triplicate letter in which words are carelessly omitted or in which the handwriting reflects haste would be a less desirable text than the quadruplicate of the same letter in which the text is complete and the manuscript shows that it was copied more accurately than the triplicate. Nontextual considerations, however, may be decisive here—in particular, evidence showing which copy of the letter was received first or received at all. A carelessly copied duplicate version of a letter from abroad that reached the State Department in June and was the basis for a foreign-policy decision would be a better source than an elegantly inscribed first copy that arrived several weeks or months later.

A second order of preference is given to copies of the letter in the hand of the author or someone under his or her direction that were designed to be retained, not transmitted. Such versions fall into four categories: drafts, letterbook copies, copies that present fairly exact facsimiles of the recipient’s version (such as letterpress or carbon copies), and independently prepared loose file copies.

Preliminary draft versions on separately inscribed sheets are easily identified, but some of the other categories are not mutually exclusive. In the eighteenth and early nineteenth centuries, bound volumes of blank pages were used to record outgoing letters. Many authors used such letterbooks as a convenient place in which to draft correspondence, recopying the final version on a fresh sheet for mailing. Others copied texts of a letter’s final version into the letterbook. Thus, some letterbooks represent volumes of drafts while others are assemblages of file copies of polished versions of outgoing correspondence. More difficult still are public figures who did not leave well enough alone. George Washington could not resist tinkering with youthful letterbooks once he achieved middle age, correcting spelling and awkward usage. A clerk then set to work recopying the emended results, creating a textual nightmare for present-day editors (Abbot, “An Uncommon Wareness of Self”).

Documents that represent contemporary facsimile versions include letterpress copies, carbon copies, and modern photocopies and computer files. The efforts of American letter writers before the era of carbon paper to create files of outgoing correspondence without recopying the letters are a tribute to their ingenuity. In the late eighteenth century, Thomas Jefferson and others enthusiastically experimented with every invention that could ease their burden. Stylographic pens and pencils, which produced duplicate images of holographic materials, were employed. Jefferson and Charles Willson Peale purchased polygraphs, devices that made two pens move simultaneously to produce two copies of each letter, one for transmittal and one for filing.

Last, but not least, there was the letterpress. In the earliest of these devices, a thin, nearly transparent sheet of paper was moistened and pressed against the inscribed surface of handwritten material. Ideally, enough ink was transferred from the original to the back of the blank sheet so that the handwritten words showed through and could be read from the front. In practice, most letterpress copies were fragile, smudged horrors until the mid-nineteenth century, when more sophisticated techniques produced more satisfactory copies.

The invention of the typewriter and the use of carbon paper added still another form of simultaneously created facsimile copy. Although carbon copies of typed materials were part of American life and usage for more than a century, editors have seldom written of the problems of rendering intelligible versions from uncorrected carbon copies. They’ve been even less eager to comment on copies created by photocopying or printed from computer files. (A notable exception was Fredson Bowers’s “Multiple Authority: New Problems and Concepts of Copy-Text.”) Until more such literature appears, editors can only rely on their own common sense and the experience of other editions.

A third order of preference is given to transcriptions and printed copies. Editors commonly designate as “transcriptions” copies made substantially later than a document’s composition and executed by someone acting without the authority or assistance of the author. As source texts, such typed or handwritten transcriptions and printed versions of letters rank far below an existing contemporary copy. To complicate the matter, the families of many well-known figures like Thomas Jefferson began copying an author’s letters in his or her lifetime and continued doing so in the years after the celebrated author had died. It may be impossible to determine which copies are contemporary and which posthumous. The editors of Jefferson’s papers are fortunate, since the original documents, in Jefferson’s hand, usually survive for use as source texts, while the copies and transcriptions need only be cited in notes. Other editors aren’t so lucky.

Editors must often choose among several transcribed copies of the same document to determine which is the best source text, the one closest to a vanished original. For modern materials, nontextual evidence may identify the best transcription: patterns of handwriting that indicate age, results of chemical tests on paper, or a comparison of typefaces in printed sources. When nothing useful can be learned from such historical evidence, editors use more advanced methods of classical textual filiation, whose commonsense rules apply to modern as well as ancient materials. These rules require that the editor (a) attempt to learn something about the copyist for each scribal version and (b) identify patterns of common and unique errors in the copies to determine the sequence in which the variant transcripts came into existence. Transcripts that contain the same errors are likely to be part of the same family, one descended from the other in order of recopying.

Like handwritten transcriptions, previously printed versions of a letter can reflect the interpretation and even the style of a later copyist, not the author’s intention. The problem addressed here is not that of establishing the preferred text of a work written for publication and printed during an author’s lifetime, but of letters or other private materials for which the original has disappeared and only a version printed in some earlier edition survives. When more than one such printed text survives, the editor must decide which has the greatest claim to authenticity and accuracy. Some of the techniques for choosing among handwritten or typewritten transcriptions apply. Obviously, the editor has an advantage in establishing chronological patterns among printed texts, which usually bear a date of publication. But even this can be misleading when earlier editors prepared their printed versions of a source text by referring independently to the same archetype. The eye of the editor who published a text in 1830 need not have been more reliable than the one who went to work in 1880. When evidence indicates that two printed versions or two transcripts of an original were both based on the same source text, the editor relies on such internal evidence as faithfulness to known patterns of authorial spelling, punctuation, and capitalization, as well as the accuracy of the copyists’ readings of proper nouns. Many printed versions dating from the nineteenth and early twentieth centuries appeared without statements of editorial policy, and modern readers must resort to their own critical judgment in selecting a source text.

As Herman Saatkamp pointed out in “Private Rights vs. Public Needs,” editors may also have to deal with letters that were intended to be private but were published nonetheless. He gives a hypothetical example of an unknown author whose letter became a rallying point for political revolution and was published and republished, with significant variants. The documentary editor’s decision whether to publish the earliest or the last version is a choice between authorial intention and the letter’s context and significance. While the author’s intention would be represented best in the earliest published version, the political movement’s use of the letter would be more evident in the last published version. A volume of the individual’s writings might well include the earliest published version of the letter, while an edition of the records of the political movement would demand use of the last published form. Here, the “context is not the letters of an individual, but the letters of the revolution” (92).

2. Papers

In general, the rules that apply to an individual’s personal or professional correspondence also apply to selecting source texts from variant versions of his or her papers, records of professional activities or public life that cannot be defined as any form of letter or diary. This category includes the legislative reports of a lawmaker, the technical notes of a scientist, the general orders of a military commander, or the lecture notes of a reformer or educator.

Special considerations sometimes may make the best source text something other than the most nearly final version that survives. These are very practical issues relating to the availability of reliable editions of these final versions and the absence of such accessible editions for preliminary versions that may display significant variants. The problem arises frequently in the papers of women and men who held public life. Official government publications often provide adequate, accurate versions of legislative committee reports and formally adopted statutes. Given their limited resources in terms of money or size of publication, modern documentary editors may hesitate to republish an already available printed version of the same materials. If the search for the edition uncovers manuscript or typewritten draft versions of such reports or pieces of legislation, these little-known documents might better serve readers as source texts. Their notes can direct scholars to the location of the printed final versions, which can be used to create parallel texts of the evolutionary stages of the same document.

3. Diaries and Journals

Personal diaries, like some letterbooks or pocket notebooks, have been subject to subsequent revision and improvement by their authors or later editors. When the resulting earlier printed versions have achieved wide circulation and popularity, a new edition of the source, restoring its value as documentary evidence, can be one of the greatest challenges in documentary or textual editing. C. Vann Woodward recounted his efforts to deal with this problem in “Mary Chesnutt in Search of Her Genre,” and Joel Myerson and Daniel Shealy discuss their more conservative methods of dealing with another nineteenth-century woman’s journals in “Editing Louisa May Alcott’s Journals.”

Usually, however, only one version of a journal or diary survives. While its pages may be difficult to transcribe and its references may be baffling to annotate, there are seldom rivals for the title of source text. If more than one does survive, the one closest to the original inscription has the most documentary value.

4. E-mail and Other Personal Electronic Records

“Inscribing” one’s correspondence or daily reflections on a keyboard connected to a computer will not necessarily make life easier for your future editors. Electronic forms of correspondence and diary keeping can create variant versions as easily as those created by pen or pencil or typewriter. Even though those versions may be more legible than scrawled handwriting, they don’t save an editor from decision making. Not only may an author’s files contain drafts of a message, that message may have been sent to multiple parties. Authors who maintain personal journals based on bits and bytes rather than ink and paper are still free to rethink and revise these entries.

These convenient electronic writing systems create their own challenges as well. E-mail makes a distinction between “cc” (carbon copy) and “bcc” (blind carbon copy) transmissions. Editors who catalog e-mail files must decide early on whether to retain the original e-mail address and domain and server names tracing the route of transmission. Beyond individual e-mail exchanges, there are the treacherous reaches of the “Listserv.” As this Guide focuses on references to existing editorial practice rather than speculative advice, we will do little but recite the special problems of this new form of text. Readers will be reassured to know that the issues have already been considered in a recent presidential address to the ADE, Beth Luey’s “Editing Bill Gates’s E-Mail.” Luey employs humor and insight to analyze the problems of editing a large corpus of the papers of a figure who spans the twentieth and twenty-first centuries.

B. Printed Works

The literary works of any writer—prose and poetry composed for publication and issued in print more or less as intended by the author—present special problems for the documentary editor. Of these, the first is the choice of which version of the published work will be the basis for the text in the new edition. The editor who aspires to a CSE emblem will naturally refer to the guidelines of that organization before choosing a version for transcription. The choice there may be the most appropriate copy-text, not a source text, and the criteria for its identification lie outside the boundaries of this volume.

A copy-text may be the foundation of an emended critical text whose aim is the representation of the author’s final intentions, intentions that may or may not appear in all respects in any single surviving copy of the work in question. The critical edition will then be a new document itself, recording the best judgment of the editor, not the words or punctuation of any single version published in the author’s lifetime or available to any earlier reader. The documentary edition of such a work, however, is normally a noncritical one based on a single printed version that was actually read by the audience for which it was originally intended.

The differences between the goals of a textual and a documentary editor may extend to the choice of the version that will be the basis for the editor’s work. Traditionally, textual editors of literary works prefer the printed version that reflects the author’s most fully realized intentions. They assume that most of their readers are concerned with the development of an author’s literary craftsmanship, which is often reflected in her or his correction of earlier printed errors. Modern textual scholarship has questioned these assumptions, but a focus on authorial intention is still an overriding consideration for many. Documentary editors must also weigh factors such as the historical impact of specific editions or printings of the same book or essay. Many considerations can make something other than the final version in a set of printed variants far more important in documentary, evidentiary terms than the last, most polished edition.

As an example, textual and documentary editors might make very different decisions were they to choose between two versions of a syndicated column published in a 1940 newspaper and then revised by its author for inclusion in a book-length anthology of such pieces twenty years later. An editor concerned with the journalist’s literary intentions might conflate the elements of the two into a critical text, but a documentary editor would weigh the public influence of each version. If the 1940 column was an important factor in that year’s presidential race, while the 1960 revision languished in books that quickly went to remainder tables, the documentary editor would choose the first. The pamphlet’s significance for an audience of historians of politics or journalism is its impact in 1940. The author’s later improvements could be recorded as variants in notes, but it would be nonsensical to use the 1960 revision as a source text, forcing readers to reconstruct the 1940 column through laborious reference to footnotes or tables of emendations.

For printed material, as for any source text, both the source and the projected audience will influence editorial decisions. Whether literary or historical values are the primary reason for republishing a printed document, the choice of source text or copy-text demands a thorough knowledge of the document and its creator and the needs of those likely to consult the new edition. An editor must assume the role of historical bibliographer to master the story of a work’s composition and publication before deciding on the appropriate basis for a new edition. Knowledge of the technology of publication methods must be part of this skill.

Newspapers present a special challenge, for pieces published in that medium are peculiarly affected by changes in the mechanics of dissemination. Until the mid-nineteenth century, the newspaper that first printed an author’s essay or articles was usually the one to which the writer had submitted a manuscript, and its columns were likely to carry a version that had received some authorial review. Later newspaper printings in this era customarily drew on the first, and the variations that appeared were more likely to be typographical errors that could be noted in the editorial apparatus. Tracing authorial contributions to the press in this age can be challenging, but textual problems are comparatively straightforward. (See, for instance, John M. Robson, “Practice, Not Theory: Editing J. S. Mill’s Newspaper Writings.”)

This situation changed with the development of newspaper syndication and the telegraphic transmission of text in the late nineteenth century. If an author’s article or story was published in syndicated form, the editor’s research must expand to the methods used by the news syndicate in question. Here the methods of filiation may again be helpful, for the original can often be reconstructed by analyzing the variants in second- and third-generation copies.

Simply put, editors preparing noncritical documentary texts of printed sources cannot be uncritical in their methods. They may not ignore tools of textual scholarship or other disciplines that can helpfully be adopted. Many printed items can be viewed as documents: official government publications, pamphlets, essays, and books. Within each category, the editor may need to become familiar with all the tools of bibliography and textual criticism, even though the editorial product itself will not be a critical one. To edit printed sources as documents, one must investigate their printing history with all the care employed by an editor hoping for a CSE emblem. Various—and variant—printings of the document may need to be collated as rigorously as for any MLA series, and the results of that historical research and the mechanical collation should be analyzed scrupulously.

Neglecting the tools of sophisticated bibliography for documentary editions can have embarrassing results. The editors of the Laurens Papers admit openly, if not happily, to such a blunder in the fifth volume of their series, where they presented a scrupulously printed facsimile of Laurens’s 1767 pamphlet A Representation of Facts. Variant readings from all surviving copies of the pamphlet are recorded in textual notes, but the introduction to the document states that the editors were unable to discover the precise order in which the last two versions of the pamphlet were issued. Had the editors used the tools of modern bibliography to compare the variants for such clues as broken type, they might have determined the sequence in which the versions of the pamphlet were run.

In documentary editing, the compositor’s type font should be regarded as an element of inscription as important as a scribe’s copying practices, the configuration of the keyboard of an 1882 typewriter, or Woodrow Wilson’s use of Graham shorthand. Editors of printed documents cannot assume that their task is easier because their sources already exist in typeset form. Instead, they must make themselves experts in this method of documentary inscription. Fredson Bowers’s Principles of Bibliographical Description will disabuse such editors of the notion that editing printed sources is simple, and the members of the CSE are ready to offer advice and encouragement to novices in their fields of specialization.

For a sharp reminder of the ways in which a disregard for bibliography and printing history can invalidate an edition of printed documents, see Jack Warren’s “The Counter-Revolutionary Career of Peter Porcupine.” Assessing modern reprintings of pamphlet literature that pretend to scholarly quality, Warren remarks:

A pamphlet edition ought to present complete and authoritative versions of clearly defined source texts, sufficient to make it unnecessary for scholars pursuing most lines of inquiry to refer to the originals; the edition as a whole should be based on well-defined selection criteria relevant to the contemporary importance of the writings involved; the notes should provide as much information as possible about the publishing history of the texts and should point out any significant variations between contemporary editions. Such basic standards may seem obvious to most documentary editors, but they are often neglected by those editing pamphlets. (93)

C. Orally Generated Texts

Until the development of practical sound recording at the end of the nineteenth century, speeches, conversations, sermons, and interviews could survive only as inscribed records, handwritten, typed, or printed texts. In the last century, phonography—sound recording and reproduction by machines—has provided facsimiles of spoken archetypes to complicate and enrich the life of scholarly editors. First, we’ll examine the choice of source texts for the prephonographic era—a wide and varied menu, indeed.

1. Lectures, Speeches, and Sermons

It is not uncommon for such oral presentations to survive only in the handwritten or typed version from which their author read aloud, with no records made by witnesses to the event. Such a “reading text” or “pre-text” is the only surviving text, and the editor is spared further work in identifying the source text.

However, popular clergymen and other speakers were often not the only persons likely to leave records of these events. The editors of the Frederick Douglass Papers found such an embarrassment of documentary records that they devised useful categories into which they grouped the records left by witnesses to his speeches: summary, narrative, extract, and stenographic text. The first three are usually brief paraphrases of the words actually spoken, and the distinctions among them refer to the completeness of the paraphrased record left behind. When one of these forms is the only record of an oral text, it is transcribed and reproduced in a documentary edition as literally as possible. Editorial notes can explain the nature of the source, and readers can judge for themselves how accurately the reporter has mentioned, narrated, or summarized the speech or conversation involved.

Through the early nineteenth century, various mentions, narrations, and summaries of the same speech or conversation often survived in textually irreconcilable versions. In the absence of systematic shorthand, no two reporters left accounts of the same spoken words that could be viewed as variants of the original. Verbatim extracts and stenographic reports of speeches are more common in documents inscribed after the mid-nineteenth century, when systematized methods of shorthand became increasingly common.

The editor may be unable to identify one source text here, choosing instead to give readers access to all conflicting reports. Should one record of the speech be a ten-page narration, while the others are paragraph-long summaries of the same words, editors usually transcribe and print the longest record as the editorial text, reporting transcriptions of the shorter versions in notes or as separate documents.

Lectures and political speeches are only one variety of orally transmitted texts to have received attention from modern documentary editors. Sermons have also received considerable attention, and many American divines have left modern editors both manuscripts used as reading texts in the pulpit and manuscript or printed versions modified later for the reading public. For a valuable discussion of the differences between such sermons as oral events and as literary documents, see Wilson Kimnach’s “Realities of the Sermon: Some Considerations for Editors.”

2. Conversations and Interviews

Other difficulties in editing oral communication were faced by a group of editors of conversations with well-known writers that were published in the Victorian era. In any series of published interviews, the personality of the interrogator may be as important to the reading public as the words of the person being interviewed. When more than one version of an interview survives, the expected audience of the modern edition may dictate the choice of source text. An earlier, uncorrected version would serve those interested in a biography of the interviewee; while the published, final versions better serve the study of that same figure’s public reception or reputation (Patrick Scott and William Thesing, “Conversations with Victorian Writers: Some Editorial Questions”).

3. Oral History Memoirs and Other Recorded Interviews

An even more complicated form of orally communicated record arose from the oral history movement initiated nearly sixty years ago by Alan Nevins at Columbia University. In earlier years, oral history projects tended to interview well-known figures, but today interview subjects represent all walks of life. Indeed, it is now recognized that the records of such interviews may provide scholars with unique evidence from men and women who would otherwise have no voice in the public forum. Oral histories have special value now precisely because they can record the memories of people unlikely to keep extensive diaries, maintain large personal archives, publish memoirs, or attract the notice of contemporary media.

Whether the subject is well known or comparatively obscure, oral histories are firsthand spoken narratives recorded with high-quality sound or video equipment in a carefully structured interview setting. Such interviews are quite distinct from recorded speeches or reminiscences in that interviewers do their best to elicit detailed responses to open-ended questions. The very setting of the oral history interview introduces an intellectual challenge. As Ronald Grele pointed out, unlike written diaries and letters, “oral history interviews are constructed, for better or worse, by the active intervention of the historian” (“Movement without Aim: Methodological and Theoretical Problems in Oral History”).

The process of transcribing the interview can never capture all its nuances, but methods for representing the false starts, breaks in thought, and other nonverbal signals are established in numerous oral history resources. Less standardized are the methods oral history interviewers or projects use to control the “authorship” of those transcriptions. Most oral history programs ask interviewees to review initial transcriptions of their interviews, encouraging them to correct errors and allowing them to delete passages they regret or to place restrictions on the use of other sections. Some programs provide interviewees with a chance to review even these “corrected” versions as well. Quite simply, this means that the choice of a source text for an oral history interview may be determined by factors beyond the editor’s control. If the oral history project’s guidelines make the original audio recordings or uncorrected transcriptions unavailable, there will be no choice but to use transcriptions reviewed and emended by the interview.

The Margaret Sanger edition found that even a complete and apparently straightforward source text for an interview had its pitfalls. A videotape of Sanger’s television interview by newsman Mike Wallace had survived in excellent condition, and the network had created a transcription of the Sanger-Wallace conversation at the time. The surviving transcription was so incomplete and inaccurate that Sanger’s editors had to view the videotape again and again, correcting the original transcript to produce a truly accurate text for their own edition. The supposedly authoritative contemporary transcription became merely a reference to be mentioned in the source note; the videorecording itself was the source text.

4. Audio-Recorded Speeches

Some of the methods adopted for oral history interviews can apply here. The luckiest editors of publicly recorded speeches are those with the option of using sound recordings as their source texts for transcriptions and even of making these recordings available to the public. Here, too, issues of a special kind will appear.

Recordings of arguments before the U.S. Supreme Court since 1955 have proven so popular that they are now the basis for competing editions, some free and some fee based. The first was Peter Irons’s 1997 May It Please the Court: The First Amendment: [Live Recordings and] Transcripts of the Oral Arguments Made before the Supreme Court in Sixteen Key First Amendment Cases. This set consisted of four cassettes and a 262-page book containing explanatory notes and transcriptions of the materials on the tapes. It was followed two years later by Jerry Goldman’s edition, The Supreme Court’s Greatest Hits, a CD-ROM containing recordings of arguments before the Court and editorial annotation based on Goldman’s classroom use of the materials. The Greatest Hits appeared in a second edition three years later, this time expanded to two CD-ROMs. In addition, Goldman’s online “Oyez Project” makes much of the material free over the Internet in MP files.

A more conventionally scholarly example of such publication is the Presidential Recordings Project at the University of Virginia. This project transcribes, edits, and publishes selections from recordings made secretly in the White House, 1940–1973. The audio files include meetings and telephone conversations of presidents from Roosevelt through Nixon. This historical record is more than 5,000 hours long, and neither the complete audio files nor their transcriptions will be mounted on the Web site nor published in the sets of volumes in the Thematic Series or the Reference Series (both available from W. W. Norton). To provide access to the larger archive, the project maintains its Presidential Recordings Program on the Web with a host of information on the recordings—including their provenance and articles on the generation of tapes in different administrations—along with downloadable files of the complete unedited tapes as provided by the National Archives. When available, there are flash modules for the recordings, linked to transcriptions.

III. Horrible Exceptions to These Simple Rules

A. Reconstructing Source Texts

Occasionally, editors will have to apply scholarly skill and knowledge simply to reunite the physical components of a source text. André De Tienne recounts the heroic measures demanded of the editors of Charles Peirce’s manuscripts in “The Peirce Papers: How to Pick Up Manuscripts That Fell to the Floor.” In the decades after his death, Peirce’s papers were disorganized and reorganized by a series of well-meaning students and admirers. The problem was exacerbated by Peirce’s disinclination to supply page numbers or dates in any consistent fashion. Today his editors must use a “hanging page display” of photocopies that offers “all the advantages of a recycled laundry line” to bring order to the chaos. To aid them, the project maintains separate databases for manuscript organization, publication history, and a calendar.

While few editors will face a task like that of the Peirce scholars, many will encounter the phenomenon of a paper-based document that was simply torn in two. For the detective work needed to resolve such a dilemma, see John Lupton’s “Putting the Pieces Together,” the tale of a Lincoln legal document ripped in half, with each segment finding its way to a different repository. The reconstruction of Lincoln’s professional papers for a comprehensive edition enabled editors to reunite the pieces of the whole.

B. Source Texts Demanding Translation

Some documents demand more than traditional textual methods of transcription and emendation to make them intelligible to their intended modern audience. These are the texts that must be translated from a foreign language or from authorial shorthand or cryptography of some sort.

1. Foreign-Language Materials

Editorial policies on the translation of foreign-language materials in an edition are discussed below, in chapter 6. Final decisions in this area may depend both on the number of such texts and on the intended audience. Whatever the choice, it will have an impact on the choice of source texts.

In past decades, American documentary editors have followed a variety of paths. A few editions publish all documents in their original languages, but this is rare. Some leave foreign-language materials untranslated if they are in a language the recipient was able to read. Others translate all non-English materials. Of these, some publish both the foreign-language original and its translation, while others publish the translations alone. Some publish contemporary translations prepared for the recipient of a letter or document, whether or not these translations are entirely accurate. Others commission new English translations. We offer some examples to make clear their rationales.

A single volume for Woodrow Wilson’s presidential years may include materials in French, German, Spanish, and Japanese as well as English. The foreign-language materials that qualified for inclusion in the Wilson edition formed but a small percentage of the whole, and the editors assumed that diplomatic historians interested in specific documents were already masters of the languages in which these letters and state papers were inscribed.

Other editions have tailored—and sometimes altered—their policies on foreign-language sources more flexibly. Whenever possible, the Washington Papers edition publishes the translations actually used by Washington, those prepared by his secretaries and aides and filed with the foreign-language originals. Originally, the Thomas Jefferson edition did not translate foreign-language materials if they were written in languages Jefferson himself could read easily. More recently, however, the editors of both Jefferson editions provide readers with translations of all foreign-language materials.

The Lafayette editors followed still another course. Lafayette’s papers for the Revolutionary era include so many documents in French that only bilingual readers could have read the series had all documents appeared in their original tongues. The editors chose to translate all materials into English, the language of the audience to whom the letters were primarily addressed. These translations were clearly labeled as such, and they appeared in the annotated, chronological series of editorial texts in the Lafayette volumes. An appendix to each volume printed transcriptions of the French originals.

The editors of Albert Einstein’s papers based their decisions on sources whose audience would be substantially different from the readers of the Lafayette edition. There was never a question of presenting English translations of Einstein’s writings in German as editorial texts. Unlike Lafayette, who learned English as a young man and who maintained a substantial correspondence with English-speaking friends throughout his long life, Einstein thought and wrote in German—only rarely did he compose in English or French. Many argued for parallel texts of non-English documents and their facing-page translations in the printed volumes, but Einstein’s life and intellectual contributions made him a figure of world interest. Americans might have been well served by English translations in the annotated volumes, but the books would then have been of little use for the cosmopolitan audience they deserved. It was finally determined to publish all documents in German, English, and French in their original languages; translations appeared for the small number of sources inscribed in other tongues. A simultaneous microform of unannotated English translations for German and French texts appeared with the first printed volume, but interest in these first translations proved so great that a paperback edition was issued as well, and this format continues throughout the series.

John Stachel, the Einstein edition’s first director, explained his rationale: “Placing the translations in a separate and clearly subsidiary position has an additional advantage: it reminds readers that German was Einstein’s language, which he habitually used when he wanted to express himself precisely; hence, a struggle with the German text, using the translation as a pony if need be, is highly preferable to assuming that any translation has adequately conveyed the full import and all the nuances of Einstein’s deceptively simple German” (Einstein from “B” to “Z,” 64).

2. Shorthand

Almost everyone adopts some system of abbreviations or symbols to save time and effort when taking notes or drafting a letter or other document. This may be no more sophisticated than using “&” instead of “and” and scribbling Arabic numerals instead of writing out “one,” “two,” “three.” A few figures who have become the subjects of editorial analysis were trained in formal shorthand systems, and many more have employed secretaries with these skills. All of these forms will eventually have to be translated if an edition is to be intelligible.

Among American political leaders, Woodrow Wilson ranks first as a shorthand writer. Unfortunately, Wilson learned the rather obscure Graham system instead of the far more common Pitman shorthand technique, and he used the handy method throughout his life. His editors at Princeton counted themselves blessed when they found an expert who could still translate the archaic symbols. Here the only possible source texts were the sheets Wilson had covered in Graham symbols. Until they were translated by the Graham expert, they were meaningless to anyone else on the staff.

3. Codes and Ciphers

Coded and enciphered communications present textual problems analogous to those of standardized shorthand. In both cases the editor should make every effort to translate symbols into verbal equivalents that the modern reader can understand. Systematic codes and ciphers are customarily used to ensure confidentiality in a writer’s communications with a second party. Such cryptic passages usually appear in communications of considerable historical significance, such as diplomatic dispatches or private correspondence between political leaders. These documents are in the form of communications between two parties. Not only should the translated clear text of codes enable the reader to see just which sections were entered in code and cipher, but the text or the accompanying notes should also record what the editor has been able to determine about the recipient’s success in mastering the ciphered passages. Indicating which words, phrases, or sentences were significant enough to deserve encoding allows the reader to see exactly which information in the letter was judged confidential and which facts the writer felt free to leave open to prying eyes. Noting both the author’s skill in encoding his or her own words and the correspondent’s accuracy in using the key to the code is critical in showing the effectiveness of the transmission of the ciphered information.

Diplomatic records after the mid-nineteenth century usually contain exhaustive files of every version of an encoded message, and their editors are seldom left in doubt about what message could be read by whom, but editors of the writings of statesmen of the American Revolution frequently confront greater challenges. Many of these leaders established personal codes or ciphers for use in private correspondence during their public service in wartime, and the same men often received diplomatic appointments that required them to use official government ciphers for their correspondence.

This means that possible source texts may survive in several forms. In a draft letter, symbols for passages the author entered in cipher may be written above their verbal equivalents. The author’s file copy may carry interlined cipher equivalents, or it may be copied directly from the final recipient’s version, with only numerical symbols or hieroglyphics for the coded passages. Among the recipient’s papers, the editor may find a virgin copy, with the encrypted text deciphered on a separate sheet. Some luckless editors can find no deciphered version of a text. If they can find the code or key to the system employed, they go to work as their own decoders, working from the most authentic version of the ciphered original as the source text. The source text for an encoded document from this era must be chosen with special care to ensure that it represents the most significant version of the message. As the editor may not be able to make a final decision here until different copies of the coded document have been translated into clear text, there is no disgrace in choosing several candidates for decoding and transcription.

The editors of modern political and military records seldom need to decode enciphered materials. Unlike eighteenth-century diplomats, who were responsible for encoding and decoding their own official correspondence, twentieth-century statesmen and generals seldom saw the ciphered version of confidential dispatches. Instead, aides or technicians unraveled the mysteries of increasingly sophisticated systems of encryption.

The messages’ addressees customarily saw only deciphered texts or summaries of them prepared by these aides. Editors of documents in this category try to identify the version read and acted on by the recipient for use as a source text.

C. Transmission of Documents in the Ages of Electricity and Electronics

Along with visible agents of inscription such as pens, pencils, crayons, and typewriters, the currents that made possible electrical telegraphy and the computer systems that have given us word processing and other electronic records are responsible for their own challenges and complications in choosing source texts.

1. Telegrams

At first glance, editors with the text of received telegraphic messages deal with nothing more than a ready-made translation of Morse code signals. But the words of any telegram existed at one time or another in at least three different versions: the text the sender submitted at the telegraph office, the Morse code signals transmitted by the telegrapher while looking at the sender’s text, and the translation of those signals as inscribed on the telegraph form finally delivered to the addressee. While it’s rare for a record of the second coded form to survive, the tripartite nature of telegraphic transmission must be remembered when dealing with cabled communications.

The editor generally has at hand either the sender’s manuscript or the recipient’s translation of Morse code. While the first represents an author’s final intentions, the second is evidence of what was actually communicated to the recipient. The puzzle is compounded in coded telegraph transmissions, where the possibilities for corruption reach extraordinary levels. The files of World War II military leaders’ headquarters are filled with requests for retransmission of garbled coded materials. The George Marshall and Dwight Eisenhower editions generally contain the paraphrased texts of such messages as they reached the recipient’s office, because these, at least, are evidence of what each commander learned from the cables.

In general, most editors will choose among the possible source texts from the standpoint of their own editions. When draft or final versions of outgoing cables survive in the hand of the edition’s subject, they will be chosen as source texts. For incoming cables, the editor will choose as a source text the decoded copy of the telegram that reached the subject’s hands, the information that influenced that person.

2. Electronic Records

This volume is confined to the experience and comments of American scholarly editors, who have not yet addressed the special problems created by electronic documentary records. While we have suggested some ways that e-mail and other products of a personal computer or laptop may further complicate an editor’s life (see pp. 95–96, above), we hesitate to go further. Instead, we will repeat the commonsense words of G. Thomas Tanselle in “Textual Criticism and Literary Sociology”: “Computerization is simply the latest chapter in the long story of facilitating the reproduction and alteration of texts; what remains is the inseparability of recorded language from the technology that produced it and makes it accessible” (88).

IV. Conclusion

Choosing source texts is only the first step in a long series of critical judgments an editor makes on the road to publishing an edition. Once that text has been identified, an editor must cautiously move from the “original,” the document that holds evidentiary value, toward some accessible version that will serve the needs of a wide audience. Next comes transcription, the conversion of those source texts into physical forms that can serve the purposes of a modern editorial project.

Identifying Source Texts

I. Authentication and Attribution