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

Editing a Fluid Text

Marnoo now sought to learn my version of the story as to how I came to be an inmate of the Typee valley.

—Chapter 18, Typee

There’s another rendering now; but still one text.

—“The Doubloon,” Moby-Dick

In nearly all cases, literary works exist in multiple versions. They are what I call fluid texts,1 and each version represents a revision process triggered, in some moment of intentionality, by the individual writer, or through a collaboration of writers and editors, or because of pressures by certain reader groups. But while all literary works are fluid texts, readers rarely get a chance to witness their fluidity, owing to the private nature of writing, the way books are printed and marketed to the public, and what readers expect to read. The primary cause of textual fluidity is revision, and writers and their editors often consider revisions to be discardable anticipations of a finished readable text; publishers place great stock in being able to offer to the public a “definitive,” that is fixed and established, text; and readers are inured to reading one version of a work, and one version only.

But it is also true that when readers happen upon different versions of a text (whether in the Bible or Shakespeare, in Typee or Moby-Dick, in Ulysses or even in an alternate ending to a film stored on DVD), they perk up and lean into the variants wondering why and how they came about, and what they mean for the writer, themselves as readers, and their culture. Just as there is a powerful desire to experience the pleasures of a fixed and invariant text, there is also a submerged but no less compelling desire to experience the pleasures of a fluid text. This online edition of Melville’s Typee manuscript is designed to exercise this other pleasure.

How can we provide readers access to a fluid text and, more importantly, let them witness the modes of revision that generate textual fluidity? The former problem is a matter of textual editing; the latter, a matter of literary and historical analysis. Both problems are addressed in this online, fluid-text edition (part of the University of Virginia Press’s Rotunda series) and in its print companion, Melville Unfolding (forthcoming from the University of Michigan Press). The online edition offers a full archive of materials that reproduce, transcribe, and analyze the document itself, a heavily revised working draft of three central chapters in Melville’s first published book. Melville Unfolding offers a fuller analysis and a critical approach to Typee based on manuscript findings, and it includes a selected edition of the manuscript as an appendix. The online site and book can be used independently, but they draw upon and reflect upon each other. The book and its selected edition demonstrate how one might use the archive critically; the archive contains a wealth of material for further analysis, but it, like any edition, is itself a critical endeavor.

Rationale and Approach

The now-standard approach to the editing of literary works, as developed through the modern theories of Gregg, Bowers and Tanselle, has been to focus on a single moment in the creative process, usually that moment in which the individual writer surrenders the text to editors, and to build an edition based on that moment, emending it in sensible ways to reflect more accurately the writer’s intentions, at that moment.2 But to edit a literary work in such a way as to showcase revision, versions, and the multiple and shifting moments of intentionality throughout the creative process presumes the validity of all moments of revision and requires an editorial approach that can accommodate and, most importantly, render readable the flow of shifting intentions. This new focus on revision requires editors to develop strategies of displaying texts to reflect a work’s textual fluidity; at the same time, readers need new strategies in how to read revision and versions.

The goals of editing a fluid text are to display a full range of textual moments and, more importantly, to make the invisible distances between versions visible and readable. Typically, modern editing has focused on the creation of the eclectic critical edition, which features a clear reading text derived from a conflation of authorial variants and (where deemed necessary) editorial invention. But this approach requires the editor to privilege one set of variants, relegating the others to a textual apparatus. Readers seeking to follow a path of revision must labor to construct stages of revision from an apparatus that may itself be heavily encoded. Fluid-text editing offers a reading text that gives readers more immediate access to the full range of revision texts associated with a literary work.

Of course, the fundamental problem in editing revision is that the text of any given revision site is invisible. When writers revise, they engage in a kind of coded linguistic process: they cross out words, insert words, or do both at the same time; they draw lines, arrows, and circles, and add carets or insertion devices to mark a path to an added text. They do not write out at each moment of revision a full rendering of the exact wording they have in mind. Thus, while these wordings exist and have interpretable meaning in the context of the revision process, they do not appear visibly upon the manuscript page as a complete or coherent text. They are invisible wordings, so to speak, registered in mind only, or if registered on the page, then legible only up to the moment of deletion. One illustration of this phenomenon, discussed in chapter 2 of Melville Unfolding, occurs on an insertion slip that Melville made to augment his description of a Typee village edifice. Referring to the decayed sacrificial offerings found there, Melville writes: “the putrefying relics of some blo{ody} recent sacrifice.” Here, the false start “blo” records Melville’s intention to write “bloody,” but the word is canceled before it is completed. The implication here is that Melville momentarily intended to recast the Typeean sacrifices as human (or at least animal) rather than just vegetal but that he changed his mind in mid-revision. It would be inaccurate to say that Melville actually wrote the words “bloody recent sacrifice,” for in fact he did not write that text; however, it is certain that he thought that text. Thus, the text “bloody recent sacrifice” is invisible to us even though it is real, and interpretable. The goal of fluid-text editing is to make such invisible texts visible. In the Typee manuscript there are over a thousand such revision sites, each representing a set of sequentialized, invisible texts made visible in this edition for the first time.

Because the witnessing of invisible revision texts requires close inspection of manuscript and print variants and because giving access to the revision process requires speculative editorial construction, the task of editing a fluid text like the Typee manuscript immediately raises the problems of interpretation and editorial practice. Of course, all textual editing involves interpretive critical judgments. When, for instance, the editors of the 1968 Northwestern-Newberry (NN) edition of Typee changed the phrase “Lacedemonian nations” to “Lacedemonian matrons” (215), they did so even though “matrons” never appears as a manuscript or print variant. They based their change on several judgments: that neither “nations” in the plural nor “nation” in the singular makes sense; that “matrons” would make sense in context; and that in fact, given Melville’s handwriting, “matrons” could have been misread in manuscript as “nations,” such that the erroneous word entered as the printed text and was never corrected by the author. Unfortunately, because the manuscript section of Typee in which the phrase “Lacedemonian matrons” might have appeared has not been recovered, the editorial decision to print “matrons” instead of “nations” cannot be tested. Even so, the argument that “nations” is really intended to be “matrons” is plausible, even compelling, and in revising the printed word “nations” to “matrons” in their edition, the NN editors have, through several interpretive acts, made visible what they take to be the once-invisible (but intended) word “matrons.”

Readers may argue that the editor has no “right” to make changes, but the fact of the matter is that editors not only have such a right, they are in a sense obliged to enact it. Texts are representations of intended wordings, and writers will perform various attempts to make their written or printed text correspond effectively with their desired wording. At the same time, publishers, editors, even printers make changes, thus complicating the process. Texts are therefore quite mutable; they exist in different versions and forms; they are revisable, and are inevitably revised either accidentally or intentionally. They are fluid texts, existing in variant versions, and their full reality exists in the total flow from version to version. But since we can read only one version at a time, we expect single reading texts even when we are aware of the existence of multiple versions. Thus, if only to sort out the versions and choose one for reading, a text must be edited, and decisions about wording are unavoidable. In the end, the function and role of an editor as a changer of texts grows out of the very nature of textuality. In the case of “matrons,” the NN editors judged “nations” to be an unintended rendering of “matrons” and altered their version of Typee to conform at that textual point to their understanding of Melville’s intended word. A more cautious editor might refrain from making the change to “matrons,” but he or she would nevertheless earmark the word “nations” to raise the question of its status. The issue, again, is not whether editors have the right to edit, but what sort of changes they may make, for what reasons, and with what signals to the reader. Along these lines, a more pertinent complaint concerning the NN edition’s change to “matrons” is that, given the NN edition’s policy of featuring a clear reading text that lacks any overt, on-the-page signal of editorial change, readers are not adequately informed of the editorial decision. They will read “matrons” on page 215 but may discover its status as an editorial invention only if they consult the textual apparatus on page 332 or 338.

Like the standard eclectic approach followed by the NN edition, fluid-text editing is also a critical and interpretive endeavor. However, it adds entirely new categories of critical intervention and interpretation. One practice in fluid-text editing is to make a text’s authorial and editorial variants an integral part of the reading experience. To make the invisible texts of revision visible requires acts of verbal construction that textual editors are not usually expected to perform. In the “bloody sacrifice” example already introduced, we may take two variants of the passage as endpoints of a process of revision: Melville first wrote “some blo”; he ended up printing “some recent sacrifice.” An editor must consider making several judgments: Did Melville write “blo” thinking to say “some bloody recent sacrifice” or simply “some bloody sacrifice”? Is “recent” a substitute for “bloody,” or was a compound adjective intended? More deeply, in considering “bloody” at all, is Melville attempting to copy that text from an earlier version (in which case he had already inscribed “bloody” and would now be rejecting it mid-word while copying it)? Or is he thinking of “bloody” on the fly as a new option, which he then rejects in mid-execution? In either case the rejected word requires even deeper conjecture: why and in what manner is the idea of bloody (hence human or animal) sacrifice impinging upon his image of the less offensive decayed vegetal offering, and what does this encroachment say about Melville’s frame of mind concerning civilization and savagery as he is writing? An editor will surely have an answer for each question. But more importantly, the debates over “blo” are matters for readers to cogitate as well. Readers cannot tell if these options have meaning critically unless they know they are interpretive options to begin with. But they cannot know they exist, critically or otherwise, unless each option is made visible to readers by editors, who, to make them visible, must already engage themselves in the critical debate over the textual option itself. Thus, editors seeking to make invisible texts of revision visible must understand the dynamics of revision, which can be known only as critical constructions of past textual events. Fluid-text editing (more than many forms of textual editing) cannot happen without critical analysis and interpretation.

Because editing that focuses on process as well as objects requires some at-times adventurous theorizing upon the invisible, the editor is obliged to shed an authoritarian mode of address and become a teacher.3 A fluid-text edition must be a heuristic device that invites readers into the analysis of variants, versions, and revision. That is, if the otherwise invisible text of revision is to be made visible to readers in a critical edition, the editor must be prepared to guide readers through the reasoning behind each editorial judgment and to help them with the new kinds of reading that an edition focusing on textual fluidity requires. Readers need to recognize how an editor locates revision sites in a document like the Typee manuscript, derives revision sequences, and concocts a revision narrative, not simply so that they may understand these features but so that they may engage the array of variants and versions themselves and concoct sites, sequences, and narratives of their own. Editors must invite readers to become editors, or to think editorially, as they read, and thus become attuned to the pleasures of the fluid text.

If the goal of any fluid-text edition is to enhance the pleasure of reading multiple versions by giving readers fuller access to the processes of revision, the editor must also devise new ways of displaying text. But statically displaying the texts of versions side by side in a book, or even on a screen, by no means adequately renders the energy that was expended to create the variations between the side-by-side texts, that is, the private and cultural energies that contribute to the process of revision itself. Moreover, not all versions exist as coherent, physically separable units; several versions of a work may, for instance, appear interspersed and in fragmentary form throughout a single document (which itself may be a fragment, as is the case of the Typee manuscript) so that widely dispersed and seemingly unconnected revision sites, visible here and there on a document, might be related to each other, whereas adjacent and seemingly connected revisions might not be related at all. In the case of the Typee manuscript, Melville performed several phases of writing at different times, which in Melville Unfolding I relate to three distinct versions. These three versions—transcription, transformation, and translation—represent three modes of revision related to memory, narrative, and rhetoric. Since Melville did not always perform these revision modes in sequential fashion, each version in a sense overlaps the others, and in uneven ways. Therefore, a sequence of revisions at one revision site might reflect one revision mode, the next site might reflect two modes, and a third consecutive site might reflect a combination of two other modes. This actual imbrication of revision texts spread out like a patchwork quilt over the document makes a standard side-by-side display of these three versions impossible. Even so, the versions and sites are discernible, but not without editorial direction.

Accessing the versions requires us to engage different edited forms of the manuscript in different ways. Readers can inspect the edition’s diplomatic transcription, which makes Melville’s near-impossible handwriting and deleted words legible, but that feature alone does not itemize and group the many localized revision sites, make visible the invisible links between otherwise seemingly unconnected revisions throughout the transcription, or reveal the broader context that relates a set of revisions to one version or another. In short, it cannot display the layerings of revision texts and versions represented on the document. The fluid-text editor’s first major critical step, then, is to devise a base reading text upon which revision sites may be mapped.

But what is affixed to this map requires other specialized editorial interventions and new ways of reading. Because a passage in manuscript may be revised repeatedly on the document itself and then changed again when it is transferred into print, a fluid-text editor must establish for each revision site on that document a reasonable sequencing of these revisions: a set of steps, each step textually coherent, each representing the actual optional wording the writer had to have considered. Such a revision sequence has no meaning for readers—its steps cannot be read—without an accompanying revision narrative to explain the motivation, direction, and rhetorical strategy that moves us from step to step. Together, the revision sequences and narratives—like a map and its legend—provide a detailed representation of the workings of the revision process. Because the editor must devise these steps and tell stories, no editorial rendering of revision can be anything other than a critical and interpretive engagement with the materials at hand.

These interventions are Herculean as well as speculative, for each sequence has its own logic, probabilities, and necessities, and each narrative will have a constructable arc (or two, or more). Multiplying those tasks by a thousand, as in the case of the Typee manuscript (itself only a fragment) is exhausting just to contemplate. So one might argue that a cautious editor should just settle for the standard transcription only and let readers determine sites, devise sequences, and invent narratives by themselves. But the diplomatic transcription only casts into sharper focus the problem of reading the invisible text of revision in the manuscript. It is a trusty editorial construct, to be sure, but its legibility is deceptive, for even though it clarifies the words located at a thousand revision sites it is inadequate to the task of rendering each site’s various recombinant wordings. That is, if an editor is obliged to make texts visible, his or her transcription only partially enacts that obligation, for the words of a transcription, no matter how sharply delineated, still conceal rather than reveal the invisible texts of their genesis and evolution. An editor’s more fully articulated revision sequences and narrative, then, are not simply a convenient service performed for the reader; they are a plausible editorial, historical, and interpretive explanation, one that guides the reader through the editor’s encounter with the otherwise invisible reaches behind the deceptive clarity of the transcription of a very messy but very important document. And in modeling for the reader that encounter in these revision narratives, the editor implicitly invites readers to make narratives of their own. In doing so, they will discover more text—and more about the nature of texts—than they had previously imagined.

Reading the Edition: Book and Archive

Currently, the fluid-text edition of Melville’s Typee manuscript exists in two formats: the selected book version attached to Melville Unfolding, my critical analysis of Typee, and the full online version you are presently viewing. The selected book edition features those revision sites (about 10 percent of the whole number) used to demonstrate various arguments in my reading of Typee. The book also examines the sexual, political, and rhetorical implications of Melville’s broader revision strategies. The online version offers the complete archive of revision sites, sequences, and narratives, but refrains from broader critical and cultural interpretation. Melville Unfolding, then, is one example of how the online archive may be drawn upon for analysis. At the same time, it contains fuller narratives of how Typee grew as a book, what the manuscript tells us about Melville’s early use of sources, and how others may have collaborated with Melville in producing his text. No single revision narrative in the online archive can tell these stories, and readers may therefore want to consult Melville Unfolding for these fuller analyses. At the same time, it is hoped that readers of Melville Unfolding will want to inspect more than the partial selection of revision sites found in that book, and that they will search the online edition for evidence of their own in thinking about Typee, or Melville, or American and Polynesian culture, or colonialism and postcolonialism, or any area of interest in which textual fluidity may be of use. In short, the two editorial forms are designed to be used synergistically, each feeding into the other.

Because Typee grew beyond its manuscript and first print versions and appeared in still other radically different versions, including the expurgated 1846 edition and the 1892 revision (both of which Melville supervised) as well as various modern illustrated editions, a film, a comic book, and the scholarly NN critical edition, I hope that this online edition featuring the Typee manuscript will grow to include these numerous other instances of this remarkable fluid text, as part of the projected Melville Electronic Library. I would also hope that scholars and critics will plunge into the archive and write books of their own, to keep Melville Unfolding company on the shelf.

Next: Navigating the Typee Manuscript


1. John Bryant, The Fluid Text: A Theory of Revision and Editing for Book and Screen (Ann Arbor: University of Michigan Press, 2002).

2. See Bryant, Fluid Text, chaps. 1–3; in particular, see my discussion of the Northwestern-Newberry (NN) edition of Typee, pp. 38–40. The NN edition is Herman Melville, Typee: A Peep at Polynesian Life, ed. Harrison Hayford, Hershel Parker, and G. Thomas Tanselle (Evanston and Chicago: Northwestern University Press and Newberry Library, 1968).

3. See Bryant, Fluid Text, chap. 6, in particular my discussion of heuristic versus rhetorical editions, pp. 115–22.