INTRODUCTION  ❖  Editing a Fluid Text · Navigating the Typee Manuscript · Writing Typee · Scenes of Revision

Navigating the Typee Manuscript

The online edition is an electronic archive consisting of digital reproductions of the Typee manuscript, a diplomatic transcription of the document, a reading text available with or without mappings of the documents revision sites, and links to the revision sequences and revision narratives for each site. The following describes each feature and how to use it.

The Typee Manuscript

The manuscript is a three-chapter fragment of the working draft manuscript of Melville’s first book, Typee (published in 1846). The so-called “First Draught” actually represents an intermediate stage of composition in which copying from earlier drafts, fresh writing, proofreading revisions, and other phases of writing are combined.

The physical document consists of a single stitched booklet (measuring 7.5 x 12.5 inches) of twenty pages of durable white wove paper, six additional loose leaves torn from similar booklets, and one smaller insertion slip (measuring 8 x 4.33 inches). The manuscript text is inscribed in one kind of ink, now brownish in color, entirely in Melville’s hand; however, some light pencilings of words and marginal markings, probably in Gansevoort Melville’s hand, appear sporadically throughout.

An inscribed paper folder, customized with reinforced edges, was crafted perhaps by Melville or a sister to hold the booklet and loose leaves. The front of the folder bears the following inscription: “First Draught of ‘Typee’—after which much was added & altered. Written in the Spring of 1845—Began in New York in the winter of that year and finished in Lansingburgh in the early part of the summer.” Also inscribed on the front of the folder is a large numeral “3,” indicating that the entire manuscript may have been divided and stored in several similar, appropriately numbered, folders. On the front inside of the folder are several experimental spellings, in Melville’s hand, of the name that finally became “Marheyo.”

The manuscript text, designated as chapters 10, 11, and 12, corresponds to chapters 12, 13, and 14 of the print edition of Typee. This discrepancy in numbering indicates that two chapters of text were added to earlier chapters in the manuscript some time after the “first draught” was completed. The text of the document is a virtually continuous version of the three chapters, with three exceptions: the manuscript chapter 10 begins in mid-sentence with text from the third paragraph of print chapter 12 (that is, the opening of that chapter is missing), the text of one unlocated leaf (designated as Leaf 12) is missing, and the texts of several unlocated insertion slips are also missing.

The booklet and five of the six extant loose leaves were discovered in 1983, in Gansevoort, New York, and since 1984 have been located in the Gansevoort-Lansing Collection of the Rare Book and Manuscript Division of the New York Public Library. One exception to this provenance is Leaf 13, which was acquired for the NYPL in the 1940s by librarian Victor H. Paltsits and transcribed in an essay published in 1943.1 This leaf, also transcribed in the NN Typee, pp. 363–74, is encapsulated in plastic and resides now alongside the other loose leaves in the New York Public Library. How or why Leaf 13 was separated from the others is not known.

The digital reproduction of the manuscript and folder were produced by the New York Public Library and appear here with its permission.


The Typee manuscript transcription is a simulation or “diplomatic” rendering of the manuscript text.2 It presents Melville’s handwritten text—including his baseline inscription, cancellations, and insertions—typographically as it appears on the document page. Melville’s inserted texts have been positioned in proximal relation to their placement on the document, and Melville’s carets (when used), specialized insertion devices, paragraph marks, and marginal doodles have also been reproduced. The pen strokes and cross-hatching of Melville’s cancellations have been reproduced as precisely as possible to indicate the exact beginning and end of each set of canceled words. Every attempt has been made to decipher obscured and canceled text; undeciphered words are indicated by a question mark in parentheses: (?).

Beginning in 1985, I prepared the transcription by first inspecting a microfilm of the manuscript provided by the New York Public Library and xerographic enlargements drawn from that microfilm. I then compared my initial transcription draft to the actual manuscript several times from 1993 to 1999 and corrected the draft accordingly. In preparing the revision sequences and narratives from 1995 to 2003, I deciphered certain previously undeciphered words, corrected one or two misreadings, and altered the transcription accordingly. A final proofreading of the transcription against the manuscript was made during the preparation of this electronic edition.

Melville composed on both sides of his paper. Each leaf of the manuscript (including the missing Leaf 12) has been given a consecutive number, and the two pages of each leaf have also been numbered. Both leaf and page numbers appear on the upper right-hand corner of each transcription page. The NN edition page and line numbers for the corresponding print text of each manuscript page appear below the “Leaf” and “Page” designations. Each manuscript line is numbered in the left-hand margin. Each numbered line of the transcription corresponds to a single inscribed baseline in manuscript. This transcription lineation is sustained in other features of the edition, including the reading text base version and the relevant corresponding sections of the British first edition text.

The following symbols are used to convey specific physical aspects of the document and Melville’s inscription practices:

Marginal notes describe more idiosyncratic physical features of individual manuscript pages (such as tears and pinholes) as well as the series of penciled notations (attributed in Melville Unfolding to Gansevoort Melville).

In general, I have attempted to preserve Melville’s imperfect spellings. All handwriting, especially in working drafts, involves idiosyncratic flourishes, abbreviations, unfinished words, and the inadvertent dropping of letters; and such shorthand does not constitute actual misspelling. In these instances, I have given the full word intended, not the abbreviated scribble actually written. Conjectured or debatable readings of undeciphered words are followed by a question mark in parentheses, “(?),” and any word that has not been deciphered in this edition appears with its legible letters printed and each undeciphered letter indicated by a “?.”

No effort has been made in the transcription to go beyond a visual representation of the document itself and indicate, through arcane genetic transcription symbols and the like, the process of revision. The chief value of the transcription is that it enhances the document’s legibility; it enables readers to read Melville’s scrawl and to discern his canceled words.

Reading Text

The central feature of the edition is the reading text of the manuscript, which consists of an editorially constructed “base version” and a set of links, one for each of the manuscript’s one thousand or so revision sites. Essentially, the base version text is an edited version of the manuscript’s “final reading.” That is, it represents the resultant text when all of Melville’s instructions to add and delete text are followed. This final reading version varies significantly from the text of the first print edition of Typee; indeed, about half of the revision sites mapped on the reading text indicate changes made to Melville’s text after he completed his first draft of the manuscript.

To explain the reading text and provide a rationale for its use as this edition’s base version, we need to recall Melville’s revision method and recognize the limitations of standard scholarly editing in rendering that process.

Base Version

In manuscript, Melville performed several phases of writing: he inscribed initial words on the baselines of his pages either by copying from previous (now lost) texts or through immediate invention; he canceled words as he wrote, revised later as he proofread, and revised several times again in later stages of composition. And this series of manuscript writing events can be witnessed in over five hundred revision sites. Moreover, as noted above, when we compare Melville’s manuscript text to the text he actually published in his first British edition, we discover another five hundred or so revisions made either by himself when he created his own fair copy (derived from the working draft and now unlocated), by his editors when they prepared the British edition, or by his brother Gansevoort when he corrected page proofs. Obviously, the diplomatic transcription of the manuscript, by itself, cannot possibly represent Melville’s revision processes. It merely depicts the revision markings and wordings in manuscript; it does not show what Melville did to create the mess, nor what he did after he tried to clean it up, nor how he continued to revise while making his fair copy and preparing his text for publication. What is needed in addition to the transcription, then, is some suitable “base version” of Typee that can be used to display, like a map, all of Melville’s discernible acts of revision, both in and subsequent to the manuscript.

The idea of a base version is drawn from the traditional notion of “copy text.”3 A primary concern for anyone attempting to edit a literary work (whether a fragment like the manuscript or the entire work called Typee) is selecting from the available versions of the text the one most useful for copying. Typically, in modern scholarly editing, the editor selects as “copy text” the version that most closely represents the writer’s “final” intention before going to press: either a fair-copy manuscript, corrected page proofs, or a first-edition text. (Editors aiming to represent the writer’s words without editorial intervention might want to use the writer’s fair copy, if it can be found. Since the Typee fair-copy manuscript has not been recovered, the editors of the NN Typee settled on the closest version of the fair copy they could find, the first British print text, for their copy text.) The editor then reproduces this copy text, emending any passages judged to be in need of emendation, either by substituting for them variant texts (drawn from other versions) that are deemed to be more reliable than the copy-text version, or by inventing suitable wording (see the discussion of “nations” and “matrons” in “Editing a Fluid Text: Rationale and Approach”). This form of eclectic critical editing can effectively represent an editor’s conception of final intention, but as I note more fully in The Fluid Text, it is a highly problematic approach to displaying the revision text of a literary work spread out over multiple versions (in both manuscript and print forms) of that work, which is itself the accumulated effect of numerous moments of intentionality. If our goal is to make a full range of revision available to readers—throughout its manuscript and print versions—we need a base version upon which we can map out all of the literary work’s revision sites. For various reasons, discussed below, no single historical print version of Typee exists that is suited to this end; therefore, a base version derived from the manuscript needs to be editorially constructed.

The problem for the fluid-text editor is to establish a base version that is large enough in its textual field to encompass the total range of revision so that all revision sites can be displayed. In the case of the Typee manuscript, we have three texts that might possibly serve as a base version: Melville’s initial baseline inscription, the “final” version that includes his cancellations and insertions, and the first British edition. The first option is not advisable because Melville inscribed his text in alternating waves of copying, fresh composition, and subsequent multiple stages of revision. An initial baseline version is simply not coherent without incorporating certain revisions, and since some of these revisions (even baseline cancellations) cannot be determined as having happened at the time of initial inscription, we cannot establish a historically stable initial inscription version with any certainty. The third option (the first British edition) is a fully readable and coherent text, but it is not viable becuase its text necessarily incorporates the collaborative revisions of editors as well as Melville. Of course, collaboration is a major cause of revision in any fluid text and is not, in itself, a problem. But if we seek to differentiate Melville’s revisions from his editors’, we can best demonstrate that collaboration by avoiding a collaborative text as a base version. Better to use the purely authorial version that precedes the collaboration as a foundation upon which to map out authorial and editorial revisions.

With these restrictions in mind, I have chosen to construct a base version on the second option: the “final” manuscript version. To establish this text, I did nothing more than follow the editorial instructions implicit in Melville’s cancellations and insertions. That is, I canceled what he canceled and inserted into his baseline text what he inserted. In performing this mechanical process of addition and deletion, I created what may be called a kind of fair copy of the working draft manuscript. Of course, my “fair copy” should not be confused with the actual fair copy Melville prepared himself. In preparing his own fair copy to send out to publishers, Melville most certainly revised what he was also copying. (Indeed, some of those revisions appear in the variation of text between the manuscript and first British edition versions.) In preparing my mechanical fair copy for the base version of this fluid-text edition, I have, needless to say, refrained from any revising of my own, including any of the small logical corrections to the resultant text that Melville (or any writer) would have made. That is, in a handful of cases Melville inadvertently dropped necessary words or added unnecessary ones while he revised so that the final reading includes several obvious inadvertencies such as “a a college of vestals.” Since these inadvertencies are few and inconsequential, I have not edited them away. The resultant base version text, then, is quite readable but not perfectly “smooth.” Happily, these minor incoherencies do not seriously impede the flow of reading, and I preserve them as a vestige of the roughness of Melville’s original document and as flaws that remind us of the “constructedness” of the base version itself.

Revision Sites

Despite its one or two rough spots, the base version text provides a “smooth enough” reading experience. But though it can be read conventionally for its surface content, the base version principally functions as a map for designating Melville’s revision sites. As such, the mapping feature of the base version may pose small challenges to our normal reading expectations. Each numbered line corresponds precisely to the lines in the transcription. Since some lines are affected by cancellation, they will appear shorter than others, or simply as a blank “canceled line.” Other lines include lengthy insertions and are therefore so long in accumulated content that they cannot fit on a single line. Thus, while most consecutive lines in the base version appear in uniform lengths, those lines with substantial revision will vary remarkably in length.

The base version can be viewed as a clear text (without “mapping”) or with its color-coded revision sites mapped in three ways; each mapping is available under Revision Sites in the frame selector. In these views, each revision site is also designated by its own revision code that pops up when the reader moves the cursor over any site. (The numbering of the codes restarts at “one” for each of the three manuscript chapters.)

To illustrate the mappings, let’s look at four sample lines from chapter 10 in the Reading Text.

Located under Texts in the frame selector is the “base version,” which offers a clear reading of the final wording of Melville’s manuscript:

22 ¶ Here were situated the Taboo Groves of the valley — the scene
23 of many a sensual feast, of many a horid rite.
24 Beneath the deep shadows of the consecrated Breadfruit trees
25 there reigned a solemn twilight, a cathedral like gloom.

Users may read this text without having to bother with any codings, and this version may be used in comparison to the first print edition (also located under Texts in the selector) with its altered text indicated in pink. This color-coded print text, however, is not “clickable.” For the links to revision sequences and narratives, readers must use the mapped reading texts located under Revision Sites.

By clicking on “In manuscript” under Revision Sites, users may view the same four sample lines, but here with the manuscript revision sites highlighted in yellow.

22 ¶ Here were situated the Taboo Groves of the valley — the scene
23 of many a sensual feast, of many a horid rite.
24 Beneath the deep shadows of the consecrated Breadfruit trees
25 there reigned a solemn twilight, a cathedral like gloom.

The yellow sites indicate places in the text Melville revised in manuscript only. When you move the cursor over a site, that site’s revision code will pop up; the revision code for the site in line 22 is RS10ms40, meaning revision site 40 in manuscript chapter 10. When you click the site, a separate, resizable window containing the site’s revision sequence and narrative will appear.

By clicking on “Later stages” under Revision Sites, users will find the text color-coded in pink to indicate revisions to Melville’s text that were made subsequent to the completion of his working draft manuscript.

22 ¶ Here were situated the Taboo Groves of the valley — the scene
23 of many a sensual feast, of many a horid rite.
24 Beneath the deep shadows of the consecrated Breadfruit trees
25 there reigned a solemn twilight, a cathedral like gloom.

These changes, evident by a comparison of the base version to the first print edition, were made either in the preparation of his fair copy (now lost) or in galleys and page proofs (also), by Melville himself, his brother, or his publisher, editor or printer. These sites are also linked to corresponding revision sequences and narratives.

Finally, by clicking on “All combined” under Revision Sites, the user will find all revision sites displayed, both the manuscript and print-text sites in their respective, yellow and pink, highlighting.

22 ¶ Here were situated the Taboo Groves of the valley — the scene
23 of many a sensual feast, of many a horid rite.
26 Beneath the deep shadows of the consecrated Breadfruit trees
27 there reigned a solemn twilight, a cathedral like gloom.

In addition, sites revised in both manuscript and later stages, such as “deep shadows” in the illustration above, appear in orange highlighting to indicate the overlapping of the separate revision processes.

The revision site codes are assigned in a routine and linear manner, following the text from left to right on a line and top to bottom on a page, one page after another. Thus, the reader can be assured that RS10ms41 follows RS10ms40 and RS10e33 follows RS10e32 on the reading text; however, the numbers have no relation to the actual temporal sequencing of the revisions themselves. That is, Melville may have performed the revision at site RS10ms41 before revising at RS10ms40. Similarly, a revision on one page may have been performed at the same time or in coordination with a revision a page or two later. Again, the numbering of codes is purely arbitrary and designates the physical location of the sites on the document, not their sequence in time. In order to clarify the temporal sequencing of revisions at a particular site and to tell the story of those textual events, we need revision sequences and narratives.

Revision Sequence

In traditional scholarly editing, variant texts are represented in a textual apparatus, either as footnotes or in an appendix, often in a severely concise, highly encoded format to convey the necessary information in as little space as possible. In the genetic text editing of the one extant leaf of the Typee manuscript available to the editors of the 1968 NN edition of Typee, an attempt is made within a transcription format to give bracketed and coded instructions to readers on how and when Melville revised certain bits of text. The resultant genetic transcription is not meant for casual reading, and the technique has proven to be a barrier to critics and scholars as well as general readers. Even if a reader can decode either the footnoted or textually integrated transcriptions of revisions, he or she has little hope of tying one revision within the transcription to another, or determining the sequence of revisions that produced each new wording, or discerning the larger revision narrative that links the variants together. In fluid-text editing, the editor is obliged to work out the sequences for the reader, and to tell their stories.4

The purpose of a “revision sequence” is to display in precise and numbered steps the actual wordings Melville considered as he revised, step by step, at a given revision site or set of sites. Even though the full texts of these wordings do not, of course, appear physically spelled out in the manuscript, they are not conjectural but represent what had to have gone through Melville’s mind as he revised. When the user of this electronic edition clicks on a highlighted revision site, a separate window opens containing that site’s revision sequence with its series of steps. The first step in each sequence represents Melville’s earliest documentary inscription as he inscribed it; subsequent steps (usually several, but as many as twelve or eighteen in certain cases) repeat the previous step but add or delete words depending upon Melville’s insertion or cancellation at the site. If, for instance, a manuscript revision site indicates that Melville stopped writing to cancel a word or set of words or a false start on a word, the wording of that step ends where Melville stopped and the words he canceled are indicated in bold (step 1). The subsequent step repeats any text of the previous step left uncanceled, indicating the canceled wording with a bracketed ellipsis ([…]), and continues on in boldface with the newly inscribed wording (step 2). If they exist in the selected revision site, any revisions found in the British edition print text will appear as the final step of a sequence (step 3), with a revision narrative code (RN) referring the reader specifically to that particular British edition revision.

As with the mapped base version, the revision sequence requires a new kind of reading. Rather than reading linearly, one must read in both regression and progression as one moves from one step to the next, for each new step contains the same phrase or sentence of the previous step but also the textual modification that moves the revision forward in that new step. While moving through the revision sequence, the reader can watch the full text of the passage grow or shrink as Melville makes each successive revision indicated in the revision site. In addition, each numbered step in the revision sequence is keyed to the same numbered step in the revision narrative so that readers can easily find an explanation of each step.

The shaping of the revision sequences is not always a simple or routine matter. To be sure, many revision sites involve the alteration of a single word, requiring only a two-step revision sequence, and there is little likely debate over the shaping of that sequence. And even in the many cases involving a lengthy and complex set of steps, the sequencing is often indisputable because the logic of grammar determines the only possible next step. That is, because each revision step must be linguistically viable, matters of syntax and sentence structure frequently dictate a necessary progression. But in certain ambiguous grammatical situations, a site will lend itself to variant hypothetical sequencings, and in such cases the editor is obliged to voice as many hypotheses as possible. In revision sites with variant sequencings, each set of possible sequences is designated with a letter (A, B, C, D) prefixing each numbered step within the variant sequencing (A1, A2, A3, then B1, B2, B3, etc.). For example, in RS10ms4, we find the phrase “previous to his taking his departure” with “taking” and the second “his” canceled in two separate strokes. Two revision sequences are possible: (A) Melville stopped after writing “taking his,” canceled both words, then continued with “departure”; or (B) Melville composed the entire phrase and returned later to cancel “taking his.” The two revision sequences would appear as follows:

While A2 and B2 are essentially the same text, they are derived through a different possible revision sequence, and it is impossible to tell whether Melville decided to remove “taking his” as he wrote, or later while proofreading.

In many cases, a revision site cannot be explained independently of other sites, so that readers will find combinations of consecutive revision sites arrayed, sequenced, and narrated as a unit. Often with these compound revision sites, the sequence of revisions actually follows the linear progression of the arbitrarily designated numbering of the revision sites themselves; thus, in the compound revision site RS10ms79-80, the revision at ms80 actually followed in time as well as in space the revision at ms79. Accordingly, in the revision sequence it can be assumed that step 2 treats ms79 and step 3 treats ms80. And in such cases where the temporal sequencing happens to follow the spatial arrangement, no indication is made to designate which step accounts for which revision site. However, in larger compound revision sites, the temporal sequencing may not follow the spatial arrangement, so that, as in the case of RS12ms67-72, the revisions do not follow the sequence of the arbitrary numbering of sites (67, 68, 69, etc.), but actually occurred in the following pattern: 69, 70, 68, 67, 72, and 71. In such cases, each step ends with the appropriate revision site number in brackets.

Revision Narrative

Revision sequences are essentially an array of textual data arranged in a logical, temporal order. By themselves, however, they do not explain the strategies of revision embedded in each step. The purpose of a “revision narrative” is to tell the story of each revision site. Each narrative is a concise but readable tale that explains the textual events or revision steps occurring at a given revision site. Traditional scholarly editors may feel that the use of the word “narrative” suggests a departure from the rigorous discipline of objective analysis. But in fluid-text analysis, each revision site is taken as the scene of a textual event, and to explain the dynamics or “work” that constitutes each event requires us to consider writers as people motivated for one reason or another to perform actions that have one meaning rather than another. Moreover, a revision site can be the product of multiple hands and shifting intentions. As language is changed, strategies are revealed; inevitably, the sense of an audience or actual audiences come into play. What’s more, in analyzing a textual event, the editor has no choice but to imagine a likely scenario, and tell a story. Analysis—or ratiocination as Poe calls it—is an act of imagination. Scenes, play, strategies, writer, audience, actions, motives, language: these are the elements of textual scholarship, and of narrative. It is inevitable and not simply a critical conceit that analysis and narration are conjoined, especially if we are to analyze the events that create textual objects (variants and versions), not just the objects themselves.

Like any critical construction of a historical event, an edition (whether traditional or in the fluid-text mode) is an imagined thing; it represents what the historian or editor imagines is the case. Rather than dismiss the thousand or so revision narratives as merely “imagined,” I argue that they represent what is “imaginable” and “plausible.” They relate what probably had to have happened. And their credibility stems from the reliable methods with which they were produced. To begin with, each revision narrative is firmly rooted in an analysis of the material condition of each site. Direct, repeated, and magnified inspection of the document itself—the layering of one pen stroke over another, the squeezing of words in unexpected places, the thinness or thickness of an inking, the appearance of penciled markings, the disparity of chapter numbers—these and other peculiarities help establish sequencing. As noted above, the logic of grammar, syntax, and sentence structure along with the writer’s (any writer’s) desire to reduce repetitions in one place or increase them elsewhere, the varied effects of variant word choices, all become evidence of possible rhetorical strategies in the text. Of course, assembling data at such a microscopic level has the odd effect of rendering one insensible, from time to time, of the larger flow of sentences, paragraphs, and chapters that make up the writer’s larger work. Therefore, the editor also needs to get away from the document and simply read the book itself, and to teach it. This provides a needed perspective (not to mention an annual income), and this larger perspective allowed me to consider broader strategies and even textual events that had to have taken place beyond the realm of the manuscript fragment, such as the growth of Typee chapter by chapter and Melville’s use of sources (as discussed in Melville Unfolding).

In writing the revision narratives, I developed a concise manner of exposition to give a unity and consistency throughout. I began as early as 1993, sharing samples with participants in my NEH Summer Seminar on Melville and asking them to create sequences and revisions of their own. In the process, I learned more efficient ways to explain complex events, and this forced me to revise narratives composed earlier on. At the same time, I found that in some cases the writing of the narrative made me see a more likely sequencing of revisions and thus forced me to revise the revision sequence itself, which in turn required me to reinspect the manuscript directly. And on one such reinspection, I discovered the nearly invisible set of pencil marks tucked away in the margins of the manuscript, which in turn required me to establish new revision sites and further revise other, related revision sequences and narratives. In this process of writing and research, not only does analysis become narrative but narrative induces further analysis.

In each revision narrative, I lay down facts, arguments, and theories concerning the given revision site. Each step in the revision sequence receives a straightforward explanation that has its own corresponding number placed in brackets. Thus, readers of a particular step in a revision sequence can quickly find my explanation of it by finding the corresponding step number in the revision narrative. To achieve a unified and unobtrusive voice in my narrations, I developed a consistent habit of using the present tense to describe Melville’s revisions as he is composing and the past tense to describe his subsequent proofreading revisions. At each step, I describe the revision process in terms of the actual words Melville writes as he writes them (or in the case of false starts, intends to write them), and indicate the possible phases of writing Melville employed and the stages of composition in which each revision may have occurred. In each case, I also provide a plausible rationale for the revision in terms of grammar, aesthetics, social or cultural pressures, or rhetorical strategy. Where appropriate (either regarding the penciled markings or the fair copy and British edition revisions), I relate not only how a revision occurred but who performed it—Melville, his brother Gansevoort, his publisher John Murray, or copy editor Henry Milton—and why. Also where applicable, I provide cross-references to other revision sites that are related to or coordinated with the revision site under narration. In compound revision sites in which text revised in manuscript was also revised in fair copy or for the British edition, I always place the revision narrative for the British edition in its proper sequencing, after the manuscript narrative.

Next: Writing Typee
Previous: Editing a Fluid Text


1. Victor H. Paltsis, “Herman Melville’s Background and New Light on the Publication of Typee,” in Bookmen’s Holiday: Notes and Studies Written and Gathered in Tribute to Harry Miller Lydenberg (New York: New York Public Library, 1943): 266–68.

2. In constructing my transcription, I have profited greatly from Mary-Jo Kline, A Guide to Documentary Editing (Baltimore: Johns Hopkins University Press, 1987).

3. See Bryant, Fluid Text, chap. 7, in particular “A Rationale of Base Version,” pp. 151–54.

4. See Bryant, Fluid Text, chap. 7, in particular “A Map of Revision” and “The Revision Narrative,” pp. 155–61.