Any encounter with Blake is a meeting across media; watercolors, pen and ink, engraving, pencil sketches, and the reverse writing of illuminated printing are found in small handmade books and in anthologies of British literature, on display in museums and on the walls of private collectors, and on the computer screen in digital archives, Youtube videos, and multimedia projects such as if:book’s Songs of Imagination and Digitisation. Blake’s characteristic integration of text and image, which makes possible such an abundant multi-media experience of his work, has also engendered a complex and varied editorial legacy. “The editorial history of Blake’s art” has included, in the assessment of Morris Eaves, “three distinguishable historical phases—radical normalization in the decades following Blake's death; consolidation and institutionalization in the twentieth century; and, most recently, a digital superconsolidation that is simultaneously progressive and conservative" (“Crafting Editorial Settlements” “Abstract”). These editorial compromises, or “settlements” in Eaves’s terminology, result from the limitations of media, both print and digital. Print editions of Blake that integrate text and image are often prohibitively expensive for publishers and readers alike, and while digital editions may in part resolve the schism between text and image, they too are “restricted by interlocking technical compromises imposed by the present, rather severe, limits of memory, bandwidth, software and hardware design, institutional requirements, and our own editorial imaginations” (Eaves “Crafting Editorial Settlements” par. 25).
In his foreword to Electronic Textual Editing (2006), G. Thomas Tanselle acknowledges the exciting possibilities of digital technology, such as increased accessibility, efficiency, and new modes of textual production and display, but also warns against an uncritical embrace of a digital editing revolution. He reminds us that editorial “procedures and routines will be different; concepts and issues will not" (6). He goes on to say,
These desirable changes [offered by the digital medium] do not alter the questions we must ask about texts or guarantee a greater amount of intelligent reading and textual study. We will be spared some drudgery and inconvenience, but we still must confront the same issues that editors have struggled with for twenty-five hundred years. (6)
Whether working in print or digital media, editors still need to consider questions about authority, intention, interpretation, textual variants, and presentation or display.
While fundamental issues might remain the same in both print and digital scholarly editing, procedural differences are extremely important. Major differences in editing procedure, for example, may actually prevent scholars from producing electronic texts. Digital editions, unlike their print counterparts, often require technical knowledge about encoding, display, and even programming. Peter Robinson asserts that many scholars lack the tools and support to begin a digital edition on their own, and that the encoding systems and programs currently available to digital editors are difficult for non-specialists to use (“Current Issues”). Preserving such projects can also be a challenge. Marilyn Deegan describes the “tension between the new possibilities offered by the electronic edition and the need to preserve the scholarly record” (365). While “the underlying scholarly practices are much the same” between the production of print and digital scholarly editions,
the technical issues […] need to be resolved. The resolution is difficult, because at the moment electronic editing is characterized more by innovation, experimentation, and new developments than by established practices—that is what makes it so exciting. Electronic editing is also caught up in the world of hardware, software, applications, and standards, which change with dizzying speed. An editor is caught between taking advantage of all these new developments and trying to ensure that the work survives for the long term. (Deegan 362-63)
These problems in the procedures of technical editing revolve around the limitations of current digital tools and the pace of their development. They can in fact prevent new digital editions from being created and may raise doubts about the stability and reliability of evolving electronic projects.
The fluidity of digital editions, however, can also offer unique opportunities for experimental scholarly editing. Eaves’s concept of “x-editing” offers a new model for scholarly editing in a digital medium that embraces instability. Emerging out of the editorial settlement with the limitations of digital media, x-editing represents a break with traditional print-based editorial practice and attempts to evolve alongside changing technologies. Eaves argues that
The pressures [exerted by the “harsh conditions of unstable but relatively inflexible digital environments”], I would maintain, produce—demand—forms of editing different in degree and kind from their print-based relatives. X-editing does not simply complete or improve earlier kinds; it breaks with them, not out of editorial desire but out of desire fused with necessity [….] The conditions of our work and medium force us, in some measure, not just to adjust and improve but also to lay new foundations that will have their own evolutionary cycle, as yet unrevealed. (“Crafting Editorial Settlements” par. 30)
X-editing produces “a torrent of freewheeling experimentation characterized by multilateral problem solving, trial and error, approximation, compromise, revision, and (always) unintended consequences" (Eaves “Multimedia Body Plans” 220-21). This mode of editing is a departure from the stability of printed texts, but it also offers an important flexibility that allows for the evolution of electronic texts within emerging theoretical frameworks.
Despite the advantageous flexibility of an editorial model that evolves alongside technologies, there are also pitfalls in a system that emphasizes constant evolution. “Fluidity in editing,” as Deegan points out,
can cause serious problems at the same time as it conveys many benefits [….] Fluidity can be a strength for an editor, who can adapt and change the edition as new information comes available, but a weakness for a user who may not know what changes have been made and for a librarian who needs to deliver and preserve the materials: what version of a text becomes the preservation version? (362)
Deegan is more concerned with the problem of preserving fluid, electronic editions, but she’s right to point out that the instability of x-editing can become a problem for readers who need stable, reliable scholarly texts. If changes in editorial practices are not made public, the usability, accessibility, and validity of such projects will be compromised.
Documenting editorial decisions has long been a practice in print editions, but transparency in electronic scholarly editions takes on a new significance —especially when it describes the failures of experimental editing. John Unsworth asserts the importance of documenting failure as a lesson for future scholars:
We are in an important evolutionary moment: an important transformation is taking place, and we are a part of it. Many things that we take to be trivial, or embarrassing, or simply wrong, will be of interest to our peers in the future. Our first responsibility, therefore, is to document what we do, to say why we do it, and to preserve the products of our labor, not only in their fungible, software-and-hardware-independent forms, but also in their immediate, contemporary manifestations. The greatest mistake we could make, at this point, would be to suppress, deny, or discard our errors and our failed experiments: We need to document them with obsessive care, detail, and rigor.
Shifting the focus away from the digital product and onto the process of electronic scholarly editing opens up failure, not only as a lesson, but also as a necessary condition of experimental editing.
This piece documents our experiences editing Blake’s manuscript, An Island in the Moon, a forthcoming electronic publication in the William Blake Archive. The evolution of our XML tag set for manuscripts and the development of a color code for the online transcript show x-editing at work; the editorial process has been and continues to be collaborative, transparent, and adaptable. From its inception in 1996, the William Blake Archive has sought to take advantage of the democratizing and transformative capabilities of digital media to set “a new standard of accessibility to a vast array of visual and textual materials that are central to an adequate grasp of the British art and literature of the eighteenth and nineteenth centuries” (“Archive at a Glance”). Free access and rigorous scholarly standards make the Blake Archive a valuable resource for Blake scholars and non-specialists alike. The Blake Archive, writes Thomas Benton in The Chronicle of Higher Education, “is a gorgeous, meticulously edited compilation of [Blake’s] many illuminated works […] [and] mark[s] an important point of departure from expensive clothbound volumes available in university libraries—and unique items in private collections—to high-resolution facsimiles freely available to anyone with Internet access.” A project that would have been prohibitively expensive in print, and impossible for only one person to undertake, has become one of the most collaborative, authoritative, and long-standing digital archives in the comparatively short history of the Internet.
The Blake Archive’s success is due in part to its hybrid nature, integrating Blake’s text with images, and offering unified access to disparate materials. Historically, the codex form has limited the scope of scholarly editions, and in particular, Blake’s integration of media, excess of variant texts, and broad range of artistic and commercial work are a challenge for any medium—but especially for print. Jerome McGann’s 1995 essay, “The Rationale of Hypertext,” argues that hypertext media can in some ways ameliorate the limitations of the codex form:
To date, for example, it has been impossible to produce a true critical edition of the works of Blake. Because Blake's texts operate simultaneously in two media, an adequate critical edition would have to marry a complete facsimile edition of all copies of Blake within the structure of a critical edition. One needs in such a case not a critical edition of Blake's work, but a critical archive. This archive, moreover, must be able to accommodate the collation of pictures and the parts of pictures with each other as well as with all kinds of purely textual materials. Hypermedia structures for the first time make this kind of archive possible. (McGann par. 37)
The Blake Archive has taken advantage of these hypermedia structures to present color corrected images of Blake’s works alongside diplomatic transcriptions, explanatory textual notes, introductions to individual copies and series of works, and image descriptions. The editors frame the Blake Archive as “a hybrid all-in-one edition, catalogue, database, and set of scholarly tools capable of taking full advantage of the opportunities offered by new information technology” (“Archive at a Glance”). The hybridity of the Blake Archive is twofold: along with reuniting Blake’s text and image, detailed facsimiles appear within a robust scholarly apparatus.
From their computers, scholars can access a diverse range of primary materials that are “highly disparate, widely dispersed, and more and more often severely restricted as a result of their value, rarity, and extreme fragility” (“Archive at a Glance”). The Blake Archive now “has permission to include thousands of Blake's images and texts without fees,” from contributors ranging from the Victoria and Albert Museum in London to Essick's own personal collection (“Archive at a Glance;” “Contributing Collections”). The Blake Archive’s expansive digitized collection of fragile primary documents partially fulfills the role of research libraries as imagined by Donald Reiman in his 1987 essay, “‘Versioning’: The Presentation of Multiple Texts.” In the introduction, he observes that it is becoming increasingly difficult for scholars to collect primary materials themselves, and therefore, there will be a
greater need for all libraries that support scholarship and criticism to have available accurate (i.e. photographic) facsimiles of manuscripts, first editions, and other textual authorities if they do not have the originals, and if they do own the originals, the fragility of these primary witnesses and the danger that heavy use may destroy them may render the availability of facsimiles all the more urgent. (168)
The urgency for access to primary materials stems from the difficulty of creating personal collections, and electronic facsimiles not only make these documents available, but also preserve them for the public record. The importance of access is also tied to Reiman’s major thesis about the fundamental importance of making available multiple versions of literary works. As he famously argues,
I suggest that it may be possible to make available to the public enough different primary textual documents and states of major texts (not all of which may need be critically edited) so that readers, teachers, and critics can compare for themselves two or more widely circulated basic versions of major texts. (169)
Since Blake himself published multiple versions of his own work, he is a particularly interesting case study for the “alternative” editing practice of versioning. The Blake Archive actively facilitates the reader’s own exploration and analysis of Blake’s work not only with the presentation of variant copies, but also with analytical tools, such as image searching and the “Compare” function.
In addition to increased access to Blake’s work and new scholarly tools, the digital archive offers opportunities for both selective and radial reading. A reader can view images of an Illuminated Book separately from the transcriptions. Hyperlinked text notes appear in a new window. Each work also includes a detailed introduction, “Copy Information” (detailed bibliographic information and provenance), and an illustration description (when applicable). Because of linked navigation, readers can choose to focus specifically on Blake’s text or his images, and can tailor the scope of the scholarly apparatus to best fit their needs. When used in conjunction with the Blake Archive's advanced searching and analytical tools, the high-resolution scans of Blake's works, particularly his Illuminated Books, enable a more interactive and far-reaching exploration of the visionary poet than a traditional monograph would allow.
Despite the advances in digital technology, however, Blake’s work has always been difficult to represent, and continues to challenge editors and readers alike. One problem for the editor of Blake is the integration of multiple media within his primary documents. Illuminated printing emphasizes Blake’s own hybridity as an author, painter, engraver, etcher, and printer, and necessitates his working within—and between—multiple media. While his new “method of Printing which combines the Painter and the Poet” (qtd. in Viscomi “Illuminated Printing,” “New Printing Technologies” par. 1) solved some of the problems posed by commercialized book making, it also compromised his success within each artistic discipline. As Eaves and Viscomi point out, eighteenth-century book making was characterized by a total separation between the processes that reproduced images from those that produced text. Blake’s illuminated printing united textual and visual elements within the process of production, a process that was “appealingly domestic and autographic, as well as reasonably fast, flexible, and inexpensive, at least by standard methods of reproductive engraving" (Eaves “Graphicality” 105). However, the hybridity of illuminated printing prevented the full reception of Blake’s work within his lifetime, and limited access in the following generations. Eaves links this problem of audience to Blake’s own interest in pursuing “an artistic technology that fits many categories and none" instead of more traditional, mainstream vehicles for his work, such as oil painting or epic poetry (106). This early alienation of audience and the institutional divide between Blake’s poetry and art gave rise to an editorial tradition of separation.
The editorial divide between text and image in Blake’s work has been substantially lessened over the last few decades of scholarship, but images continue to remain a problem within technologies of storage, manipulation, and reproduction. Pictures in both print and digitization remain resource-intensive, requiring vast amounts of time, money, and equipment to reproduce. Because of the increasingly sophisticated tools and technologies needed for successful image reproduction, explains Eaves, “images resist easy technological assimilation.”
Video and audio on the (relatively) amazing multimedia scene of the World Wide Web remain grotesquely primitive, complicated, and elusive beside the reassuringly streamlined and stable letters and lines of standard, searchable, intermeasurable ASCII text. Pictures remain formidable problems, and the relation of the graphical to the textual remains an unsolved foundational issue. (“Graphicality” 118)
These conditions of technological impairment give rise to various editorial settlements: the Blake Archive restores the union of text and image by offering both, yet text and image searching are entirely separate activities. Hyperlinked navigation allows for a more customizable reading of Blake, yet it also reproduces the divisions between facsimile images and transcriptions. Although one of the founding aims of the Blake Archive was the reunion of Blake’s text with his designs, “more than a decade of collective experience has revealed that the Archive is concerned at least as much with recapitulation and recycling as with restoration, and as much with disciplined fragmentation as with integration” (Eaves “Crafting Editorial Settlements” “Abstract”).
Image searching, for example, is an important scholarly tool of the critical archive, but it also represents another compromise with the limitations of digital technologies. Searching the Blake Archive is separated between text and image; in fact, image searching is actually a search for text. Illustration descriptions, written collaboratively by the editors, generate key words, which then constitute the controlled vocabulary of the search page. Readers must select terms from a controlled vocabulary, rather than enter their own keywords. Essentially, text is used to stabilize images. Discussing the scholarly treatment of images in the Blake Archive, John Walsh writes that "the controlled vocabulary is an excellent tool for finding what one wants to find. The controlled vocabulary […] serves a much-needed purpose, like an index or page numbers, for locating material of interest. Having then found interesting content, the scholar can reflect on the subtlety and richness, ambiguity and clarity of the work" ("Survey," "Individual authors" par. 4). Once they have located an image through searching, scholars can then manipulate the images of primary sources. At the time of this writing, the Blake Archive uses Inote to annotate images. When viewing an image, the reader can select the Inote tool, which displays the image descriptions associated with that image in a separate window, along with the original image marked by quadrants (which are used to organize and display the annotations). Inote also works within image searching to zoom directly to the region of the image that contains the search term. Inote, however, will be replaced with Virtual Lightbox, an open source software tool for viewing and annotating images. Originally developed in 2000 by Matthew Kirschenbaum and Amit Kumar at the Maryland Institute for Technology in the Humanities, Lightbox is being further developed by the Blake Archive’s Technical Editor, William Shaw (Kirschenbaum and Kumar). The Blake Archive’s version of Lightbox is still in rounds of revision and testing, but the core purpose will be to provide users a new interface with which to view, manipulate, analyze, collect, and compare images.
The Blake Archive’s technical standards and encoding practices are also evolving experiments in digital editing. In 2005, the Blake Archive converted to eXtensible Markup Language (XML), a non-proprietary standard that allows more flexibility in document encoding and compatibility across platforms and file formats. The switch from SGML and DynaWeb gave the Archive a “stronger foundation” from which to work (“Archive at a Glance”) and brought it up to date in relation to current digital scholarship. In this new phase of its continued evolution, the Blake Archive continues to adapt general encoding standards to fit its own needs. In general, XML allows users to define their own elements, but there are also discipline-specific standards, such as the guidelines developed by the Text Encoding Initiative (TEI), an international consortium that has set the standard for the encoding of electronic texts in the humanities. The Blake Archive integrates several aspects of the TEI's most recent release of guidelines (at the time of this writing, P5), in addition to developing and revising elements already in use.
XML elements, or tags, are used to encode textual data based on the document's structure and content, rather than its typographic appearance (as HTML would encourage). A canceled word in a manuscript, for example, is tagged as a “deletion,” instead of being encoded as a “strikethrough.” This key difference between semantic and visual markup results in a greater flexibility in the document's final format. Within XML documents, the appearance or rendering of the text is determined solely by the stylesheet, also called XSL (eXtensible Stylesheet Language). XSL stylesheets “can turn XML into many data formats, including ASCII text, HTML, and PDF” (“Technical Summary”). The Blake Archive's XML documents are transformed “into HTML for display on the Web” (“Technical Summary”). As display standards change over time or become obsolete, XML documents are still relevant and usable; they merely have to be processed through new XSL stylesheets to meet the demands of different data formats. While the syntax of the XML encoding might not be visible to readers of the Blake Archive, XML is a better archiving and editing tool because it focuses on the document’s structure and content, rather than its appearance.
The advantages of XML, however, certainly do not negate the difficulties of editing Blake. In fact, the very potential, freedom, and flexibility inherent in digital publication create new editing challenges. The hierarchical structure of XML, for example, becomes difficult to maintain when encoding Blake's revision-heavy, non-linear manuscripts. The XML tags which appear so easy to define in theory become inadequate when faced with complex revisions, partial letters, backwards writing, or indeterminate marks that could be either altered text or an incomplete image. As reading Blake—a richly-textured encounter with a visual artist, poet, printmaker, engraver, and self-editor—becomes encoding Blake, we (as the Editors and Project Assistants) are forced to consider key questions about our project's purpose, transcription policies, and intended audience, and are compelled to develop an XML tag set able to contain his elusive works.
One of the current works-in-progress, An Island in the Moon, will set the encoding and display standards for future manuscripts published in the Blake Archive. The manuscript, Blake’s longest prose work, exists in one unique draft without a title (An Island in the Moon is extrapolated from the first line). Island is an incomplete manuscript written in pen and ink, and contains dialogue, songs, and a page of sketches, signatures, and partially legible words and letters unrelated to the textual content. Probably composed in 1784-85, when Blake was around twenty-seven years old, Island was first published as a whole in 1907. The work satirizes contemporary fashions, literature, philosophy, scientific experimentation, and the superficiality of the salon lifestyle, and may include caricatures of Blake’s acquaintances in the Mathew Circle. The character Quid the Cynic seems to represent Blake himself, while Suction the Epicurean may be his brother, Robert. Its reference to the short-lived fad of balloon hats links the piece to early demonstrations of ballooning, while its location on the moon places it in a longer tradition of moon voyage or flight narratives (Smith 60). Its form (dialogue interspersed with song lyrics) links it to contemporary theatrical pieces, while its satirical approach is a familiar narrative mode within the eighteenth century. The manuscript also notably contains the earliest drafts of three poems from Songs of Innocence: “Holy Thursday,” “The Nurse’s Song,” and “Little Boy Lost.”
The Blake Archive’s work on Island began in 1999, when the manuscript pages were photographed and color-corrected. Transcription and encoding started in 2003, and manuscript images and transcriptions are currently available on the Blake Archive’s “Testing” site, a non-public site that contains both previously published works and those in production. Editors and assistants can access works in progress to proofread transcriptions, check links, and test displays. Island is currently undergoing final revisions to the encoding, and awaiting pre-publication feedback and proofreading. Because it will be the first manuscript published by the Blake Archive, and therefore will utilize a revised tag set and a new color code designed specifically for manuscripts, this process of feedback and revision will be especially important.
Island's tag set (see “Manuscript Tag Set”), which defines XML tags and their attributes, has been in development for several years. It was initially based on the tag set used for the Illuminated Books, but ultimately, we needed a new vocabulary to represent the intricacies of the manuscript. The XML elements “addition” <add> and “deletion” <del>, for example, are seemingly straightforward changes to mark; words inserted with a caret, for example, have clearly been added to the text, and words crossed out have been obviously canceled. The original tag set defined a “deletion” as having several attributes. The most common type is “overstrike,” which describes a cancellation made by a line or lines drawn through the text—horizontally, vertically, or a simple looping scrawl (“Filling out an XML BAD File”). Another type of “deletion” is “obscured,” which signifies text that has been heavily canceled through washes, charcoal, or crayon (“Filling out an XML BAD File”). As we worked on Island, we discovered that “obscured” was too general and vague for our purposes. A heavily canceled word in ink, for example, might be described as an “overstrike,” (because the medium is ink), but also “obscured,” (since it was heavily canceled). Some of the letters and words in the manuscript were clearly “washed,” yet they were not “obscured”—they were, in fact, clearly legible. To clarify the types of “deletions,” we revised the tag set to include more precise descriptions of the medium used, and aimed for tag definitions which were less dependent on subjective analysis of the deletion, such as a “simple” scrawl or “heavy” cancellation. The original and revised definitions for “deletion” are below.
Original Tag Set (circa 2007)
CANCELLATIONS: <del><del>Material between tags cancelled.</del>
A <del>> tag (a.k.a. “deletions”) is used to mark deletions that are recoverable/legible. The del tag may have the following attributes:
overstrike (a line or lines drawn through the text, horizontal or vertical, or a simple looping scrawl)
erasure (text that has been obscured by attempt to remove writing mark, such as by rubbing)
overwrite (text that has been replaced by text written directly over it)
obscured (for washes, charcoal, crayon, etc. used to cancel text by obscuring it heavily)
Revised Tag Set (2009)
DELETIONS: <del><del>Material between tags is cancelled.</del>
A <del> tag is used to indicate text that has been deliberately cancelled, but remains legible. <del> may have the following attributes and values:
overstrike: Text that has been deleted by a line or lines drawn through the text. Pen and ink is the default.
overwrite: Text that has been deleted by new text written directly over it. (Always grouped with an <add>, as part of a <subst> unit.)
erasure: Text that has been deleted by an apparent attempt to remove the text by rubbing.
wash: Text that has been deleted by a wash.
Our decisions regarding the markup of these revisions will not only affect the transcript display and encoding of Island (and future manuscripts), but will also necessitate a return to previously published works in the Blake Archive that will benefit from the updated tag set. As we create XML elements and redefine their attributes, we are building a new editorial vocabulary with which to describe Blake's texts. Ultimately, our encoding choices are editorial acts of interpretation, and how we decide to encode text emphasizes certain manuscript features while leaving others unmarked.
The robust structures and varying detail of TEI-conformant encoding for manuscripts have nearly limitless possibilities. Since time and financial constraints make it impossible to describe every possible feature of the text, the limits of the project must be determined.
Early in the process, the editor must determine how fine-grained the transcription is to be, because markup permits the specification in minute detail of the paleographic features of the document [….] Virtually any linguistic or prosodic feature of interest can be represented; consequently, editors must make fundamental choices as to what features they will mark. (Fenton and Duggan 244)
There are tags to describe the physical object, such as its current location, material, dimensions, watermarks, ink color, handwriting, and damage. XML elements can map the text's content through rhyme schemes, grammatical structure, technical language, or foreign expressions. Projects concerned with linguistics can encode parts of speech, while others may choose to include the GPS locations of place names mentioned in the text or link to other relevant material available on the web. Transcribers can include alternate readings for an unclear word, and translators can use linking structures to provide multiple translations for a single line, stanza, or entire poem.
Adhering to our diplomatic transcription policies while serving our audience, both scholarly and non-specialist, motivates our encoding. The “utterly fundamental” first principle of the Archive is its emphasis on the physical “object” (the Illuminated Book page, the manuscript page, the letter) over the “textual unit [...] the poem or other work abstracted from its physical medium” (“Editorial Principles”). The effort to transcribe text at a visual rather than a contextual level roughly translates into a policy of “transcribe what you see,” which “implicitly raises some rather complex ideas in the form of questions about what constitutes a text, a work, a copy, an image, a picture, a representation” (Essick qtd. in Kraus 34). While these questions of what constitutes a work are certainly not new in the realm of textual criticism, they do take on renewed significance in a digital environment that can easily present multiple versions of a literary work. In electronic versions, the XML documents are both an important piece of critical apparatus and yet another version of the original work. In addition to manuscript images, textual notes, work introductions, and transcriptions, XML documents also contain important editorial notations and explanatory information. In a digital edition, the tag set, and the XML documents it helps to create, become a formalization of editorial intention. We are constantly redefining our object of study—Blake’s page—as we revise the tags we will use to encode it. The tags we choose reflect the Archive’s concern with Blake’s revisions and the physicality of the page. Changing the tag set means updating our XML documents as well—creating new electronic “versions” of Blake’s page with every XML or tag set update. The fluid evolution of the Archive’s critical apparatus demonstrates the importance of experimental modes of editing, such as versioning, in the production of scholarly electronic editions.
At the Blake Archive, our editorial interest in An Island in the Moon lies in its physical properties; namely, the revisions in Blake's hand—the manuscript’s additions, gaps, substitutions, and deletions. We want to depict the relationships of revisions within his substitutions, describe the sketched images of the last page, delineate partial letters from scribble, and clearly represent these complex features in the manuscript. We strive to reunite Blake’s visual work with his textual, while at the same time treating the page as a visual, physical object. The focus of our encoding necessarily overlooks other features of the text, such as the possibility that Island’s characters satirize real people in Blake's social circle, the various narrative modes at work (such as poetry, song, and satire), or explicit references to the popular culture of Blake's day, such as balloon hats and George Cumberland’s new methods of printing (Phillips 10-12). Although some of this material may be included in the brief introductions to a work or textual notes, our encoding is of necessity “incomplete.” In keeping within our view of Blake’s page as a physical object, we want to represent revisions to the manuscripts in both our encoding and transcript displays. While the process of developing a tag set should keep the audience in mind, it must also be responsive to the primary work itself; encoding thus functions as a mediation between Blake, our editorial intentions, the anticipated scholarly audience, and the limitations of the technology.
One of our biggest obstacles in creating a detailed—yet concise—tag set for Blake’s manuscripts has been the problem of substitutions; that is, when one word (or word ending) replaces another. Such acts of revision, although two separate changes (a deletion, and the addition of new material), actually constitute a single intervention in the text. This is what we wanted to show. But, at the time, there was nothing in the TEI standards that described this situation. We had to invent our own tag. We developed a “replacement” tag that we defined as follows: “A <rep> tag is used to mark text that apparently replaces earlier text that has been deleted by the writer. In practice, a <rep> will always be preceded by either a <del> (most often) or a <gap> - that is, text that the writer had deliberately canceled in order to replace it with different text” (Eaves “Update for MS. Tags”). The simplest kinds of replacements involved word endings, or single words and short phrases that were clear replacements in the text.
For example, the word below originally read “endeavourd” and was changed through overwriting to read “endeavouring.”
The markup for this word would have read:<l n="bb74.1.ms.01.17" justify="left">the three
Philosophers at this time were each endeavour
<rep type="overwrite" place="over">ing</rep>
This line in the transcription also includes an explanatory textual note: “Blake may have first written ‘endeavourd’ or ‘endeavoured’ before altering the word to ‘endeavouring.’”
Another example shows the replacements of noun markers, “any” for “a.” In keeping with our emphasis on the typographical page, we transcribe the inserted word on its own line. In this case, the “replacement” appears first in the encoding, and a text note explains the relationship between these two lines.<l n="bb74.1.ms.06.10" justify="left" indent="3">
<rep place="supralinear"><hi rend="subscript">any</hi>
<note>This line was inserted above, and appears to replace,
the deleted "a" in line 11.</note>
</l><l n="bb74.1.ms.06.11" justify="left">that
natural fool would make a clever fellow if he was properly
<note>Phillips reads the deletion as "a-" (page 39).</note>
While using <rep> worked well for “simple,” straightforward replacements, such as changed word endings (from “d” to “ing”) or alternate phrases, it became a problem in the display for more complex revisions. We were struggling with definitions, and had trouble deciding what in fact constituted a “replacement.” We realized that while some revisions were deliberate replacements, not all text that follows a “deletion” could be defined as a “replacement.” The simplest kinds of replacements involved word endings, or single words and short phrases that were clear replacements in the text.
While the <rep> tag does serve its function in these examples, it causes a few problems. Since the <add> and the <del> occur on separate lines, it is not immediately apparent that they constitute a single act of revision. This gets at a larger problem of definition; although we are trying to represent a single act of revision comprised of two XML elements (an <add> and a <del>), the <rep> tag emphasizes only one part of the manuscript change. By definition, the <rep> tag emphasizes the “addition,” and is an unbalanced representation of the total revision. While <rep> did serve our purposes, it ultimately did not present the entire substitution as a single unit of revision.
In 2007, the TEI updated their encoding standards. The P5 update contains the new element “substitution,” which links an “addition” element and “deletion” element together. The definition of this tag, from the TEI guidelines, is that “<subst> (substitution) groups one or more deletions with one or more additions when the combination is to be regarded as a single intervention in the text” (TEI Consortium). Integrating this tag into our manuscript tag set clarified how to encode revisions that were comprised of combinations of smaller changes. This linking of elements clearly represents the revision as a single unit, while maintaining the separate actions of adding and deleting. For example, the changes to the word “endeavour” are now encoded like this:<l n="bb74.1.ms.01.17" justify="left">the three
Philosophers at this time were each endeavour
<add type="overwrite" place="inline">ing</add>
The <subst> tag is far more flexible than our original <rep> tag, since <add> and <del> remain separate, yet we can also show their relationship to one another.
As Blake Archive project assistant Christopher Jackson points out, developing the tag set “led us not only to see something new in Blake’s process of revision, but also to formulate new editorial ideas in response” (“island article”). This kind of responsive editing is made possible by the digital environment, which allows us to rethink and revisit previous readings of Blake, including our own. Editing Blake thus becomes a process of discovery. Jackson notes that
The concept of replacement or substitution text is not mysterious. It seems obvious once you have it pointed out to you. But we had to discover it on our own–and it was the process of engaging with the vocabulary of our [manuscript] tags […] that allowed us to notice the feature of Blake's text. (“island article”)
This process of “discovery”–of both complex revisions in Blake’s manuscript and the XML tags we needed–is a result of reading Blake through the lens of encoding. Our working vocabulary of XML elements forces us to closely inspect and reconsider authorial interventions in the text. The shifting symbiosis between our tag set and Blake’s revisions is the impermanent condition of experimental editing, one in which encoding may reveal new aspects of a work, and these characteristics in turn require new XML elements to describe them.
The formal publication of Island will also mark the implementation of a color-coded transcript. By using font colors and highlighting instead of traditional editorial markings, our goal is to represent and clarify Blake's original manuscript with minimal editorial presence. The color code works to balance the Archive’s own editorial choice to highlight Blake’s process of revision, while also enabling the reader to approach the text through a systematized transcription. Colors have been carefully chosen by project assistants and editors to coordinate harmoniously while simultaneously representing a range of textual edits; for example, red font represents complex deletions, black highlighting represents illegible or obliterated text.
The color code also appears on every transcription page for easy access (as a pop-up window). We have also restricted the colors used to avoid overwhelming the reader visually–by limiting the number of colors to a specific, clearly defined set, the color code will eventually become intuitive for frequent users. Along with the color code, the Archive also uses one of the more traditional editorial symbols, the strikethrough, to indicate simple deletions. This selective combination of traditional, familiar typographic symbols with the more experimental color code allows the Blake Archive to diversify and strengthen its means of transcribing Blake's often challenging and convoluted revisions.
The use of a color code rather than traditional textual symbols to track manuscript edits is not particularly new to digital scholarship. Projects like The Walt Whitman Archive have employed similar transcription displays to capture the complexities of manuscripts. But a color code will be a first for the Blake Archive, which has predominantly focused on publishing Blake's illuminated books, drawings, prints, and commercial works up to this point. These previously published works exist as fair or finished copies, and therefore do not contain complex revisions. Island, on the other hand, was never finished or copied, and thus is riddled with misspellings, cryptic letterings, and multiple edits and re-edits. The manuscript forces us to consider how these unusual features can be transcribed diplomatically, with as little conjecture as possible, and displayed with minimal reconstruction of authorial intention.
The color code offers some key advantages over the more traditional method of using typographic symbols to represent editorial changes. While color codes and typographic symbols work in very similar ways to visualize textual alterations in a manuscript, a color code allows editors more creativity in how changes display in a transcription, while also enabling a deeper look into the often messy details of the Island manuscript. For example, transcriptions using traditional typographic symbols do not have a particularly effective way to denote that a word or phrase is replacing, or substituting, another word or phrase. These symbols tend to show only that one word has been deleted and is followed by an added word. The display that depends upon these more traditional transcription methods does not effectively capture the idea that the deleted and added words form a unit. In the example below, there are clear units of substitution, marked by yellow highlighting (from object 3 in Island).
Traditional editorial symbols drawn from the tradition of the printed monograph transcription are restricted by the fact that they must be standardized typographic or stylistic markings. While a color coded display might overcome some of the limitations of typographic symbols, it is still an editorial settlement. Any transcription will be necessarily different from the manuscript, making it even more important that editorial choices are transparent.
Edits in the Island manuscript can be difficult to interpret and hard to track. Because of damage to the manuscript, illegible handwriting, or multiple edits of one particular word or phrase, Blake's changes can be challenging for his editors, and particularly for the Blake Archive, which has set in place an editorial policy based on diplomatic transcription and adherence to only what is visually discernible from the source material (the page). When the editors / assistants want to make a particular claim or put forth a variant reading for a complicated edit, they have adopted the textual note. Text notes are linked from the transcriptions of illuminated books, individual etchings (such as Laocoön) and now manuscripts. They can provide detailed information not included in the transcription or the tag set, thereby allowing the Blake Archive to maintain its editorial policies while simultaneously providing a platform for clarification, explanation, or translation. The inclusion of text notes provides readers with yet another way to interact with a work, while also enabling a project like the Blake Archive to become more transparent through detailing the choices assistants and editors make when transcribing a work.
Although including text notes with a transcription is standard practice for many editors working in print, the Blake Archive is able to go a step further by adding text note images to highlight particularly tangled cases within the Island manuscript. These text note images appear alongside the written content of the notes themselves, providing the reader with the specific manuscript content in question. The example below gives an illustration of this–the text note and the text note image appear together to clarify the complicated nature of the overwrite here (from object 9 in Island):
This not only gives the reader a direct reference to troublesome or heavily edited spots in the manuscript, but also allows the Blake Archive to unite the textual (in the form of the text note) with the visual (the manuscript image itself). "Depending on the needs of the project,” writes John Lavagnino, “it may be desirable to represent […] unusual [textual] features with images instead of relying solely on transcription and tagging" (336). By supplementing the text of transcription and encoding, images in the editorial notes show another kind of relationship between text and image–one that is marked by extension, instead of separation. The text note image continues the Blake Archive's philosophy of joining the textual and visual sides of Blake's work, and brings those two previously separated realms together in the sphere of critical editing.
Referencing and representing the visual component of Blake's work within the text notes is perhaps most important because it allows the editors to display complicated images alongside transcriptions that often lose (and indeed, should lose) the occasional incoherence and difficulty of the original text. The transcription should ideally work to clarify the manuscript, rendering difficult revisions in a clear and straightforward manner. However, the text note image can highlight particularly problematic or noteworthy cases that may interest the reader. The text note image thus becomes another way to privilege the page while simultaneously heightening the transparency, readability, and usefulness of the transcription. Below is an example of one revision that the assistants and editors felt needed further explication (from object 8, line 29 of Island). The transcription renders this particular spot in the manuscript thus:
The transcription is clearly legible, and the color code signals that there have been a series of revisions (something illegible is followed by the deleted “man,” which seems to be substituted by “Gent”). But the sequence of deletions and additions is not as simple as it appears here. The manuscript presents us with a tangle of revisions that looks like this:
From the text note image, we can see that “Gent” replaces the entire deletion, rather than just “man,” as the transcription might lead some readers to mistakenly think. However, its positioning in the manuscript, over an illegible deletion and before the deleted “man,” cannot be replicated exactly in the transcription. Because the transcription must operate within specified parameters set in place by the editors, representing this case presents some serious issues. As Jackson explains in an email discussion, the Blake Archive has conventionally transcribed the earliest of Blake’s revisions in a given sequence first, in a movement from left to right, rather than attempting to duplicate the appearance of the manuscript exactly (“Morris, Ali, Rachel”). Thus, the text note image here provides a necessary clarification for the reader, who can examine it alongside the transcription to get a clearer idea of Blake's chain of revision. While the transcription removes a level of complexity from Blake’s manuscripts, and therefore might hinder a reader’s comprehensive understanding of the text, the text note images restore the problematic layers of Blake’s edits and disclose our editing decisions.
The final page of the Island manuscript, object 18 in the Blake Archive's pagination, has been one of the most difficult to transcribe and display digitally, precisely because it integrates, even confuses, textuality and graphicality.
Object 18 of Island
This final page, which resembles that of a sketch book, contains images of horses, livestock, and human faces in profile; signatures, probably deleted but still legible; single letters repeated in rows; backwards writing; and partial letters. The text that does appear on the page, unlike the rest of the manuscript, is very striking visually–it is not written in lines, but fills in the spaces around the images. Some of this text is heavily deleted, some is washed over in lighter shades of ink, and some even intersects with the images of horses and human faces. The images and text that appear together on this page appear to have no connection to one another beyond the shared space of the manuscript page.
In his facsimile edition of Blake's Island, Michael Phillips does not include this final page in his transcription of the work. To meet our goal of joining the visual and the textual within Blake, however, the Blake Archive must consider this page in both of its aspects. In addition to describing the images and making them available through image searches, we also transcribe the text–to the best of our abilities. The digital representation of this page becomes problematic since the transcription might visually misrepresent the manuscript's complex features, and our capabilities to encode and display certain components of the text, such as partial lettering, are limited. These challenges redefine the problem of graphicality and textuality within the critical apparatus of a digital work.
The partial letters challenge the Blake Archive's transcription policies because they exist on the border between image and text–neither fully one or the other, these lines hint at legibility, but also refuse it. Two sets of partial letters visually resemble lowercase “n's” or “m's” in cursive and another line may be the backstroke to a capital “T” or “F.”
Electronic publication limits how we can display unique semi-characters that cannot be fully represented by modern typefaces. While we could infer that these marks were most likely the beginning backstrokes of an “n” or “m,” we could not confidently transcribe them as such. There seemed to be no clear way to transcribe the ambiguity of these markings: not representing them as text in the manuscript misrepresents the fact that they are more textual than they are graphical, transcribing them as either the character “n” or “m” misrepresents our own confidence in that reading, and transcribing these lines as two characters simultaneously just doesn't seem feasible, and certainly does not preserve our goal for a transcript, which is to clarify and make legible the manuscript page.
In addition to the problem of displaying partial characters, we were also not sure how to encode them in XML. There are several options available within TEI, but none seemed to be a perfect solution. Our decision for now is to encode a partial letter with the <gap> element, which is used to mark text that is illegible, usually because of damage or cancellation. Attributes within the element specify the reasons for the illegibility. To encode partial letters, we added the attribute “partial” to explain why these letters are illegible.
While this is a good working solution for our encoding, we are not sure it will be an effective display option. The <gap> element is currently displayed as black highlighting, and we must consider the visual noise this might create in the transcript for object 18.
While black highlighting is quite similar to cancellations in other areas of the manuscript – deletions that are made with heavy strikethroughs or puddles of ink–it might misrepresent the partial letters on this page, which are not covered in heavy ink, and are in fact very present as gray-washed lines on the page. As Lavagnino suggests, there is less room in a digital display for visual ambiguity: "There is less room in a digital edition for evading interpretive questions by printing something with an ambiguous appearance. To make an edition work as intended, it is generally necessary to interpret features and not merely reproduce their appearance" (338). To the extent that the transcription of object 18 “misrepresents” the appearance of the original, the color code represents our editorial interpretations, not the literal appearance of the page.
The problems of encoding and displaying certain textual elements on this page raise important questions about graphicality and textuality not only in Blake's work, but in the scholarly edition itself. Reading partial letters, we rely on their graphical resemblance to complete characters. Describing the images on this page, we focus on the most recognizable and complete images–animal bodies and heads, human faces in profile. Several indeterminate lines on the page, neither complete image nor partial text, remain outside of both the transcription and illustration description for this page. In our production of the Archive edition, the scholarly version of Blake's work, we alter the visual aspects of Blake's work in the transcript, and incompletely describe the images in our prose descriptions and search vocabulary. While the Archive does help remedy the long-standing disciplinary division between the visual and the textual in Blake, our work leaves out still other aspects of Blake’s work as we make difficult decisions about what we can adequately encode, describe, and display in an online version.
To the extent that the publication of Island will mark a new stage of evolution for the Blake Archive, it is also a transitional project. A product of x-editing, itself a provisional, speculative process, the online edition of Island may someday become obsolete as electronic editing continues to evolve. Increased interest in folksonomy (social tagging) and the semantic web, for example, may significantly alter the scope, detail, and extent of markup. In an article discussing the reliability of electronic texts, Phill Berry, Paul Eggert, Chris Tiffin, and Graham Barwell write: “No editor can foresee all the uses to which an electronic scholarly edition can be put or all the interpretive markup that will be required. The more the attempt to provide interpretive markup is pursued through increasingly heavy tagging, the more the reliability of the text is put to risk" (271). In electronic textual editing, the tasks of the editor extend beyond meeting obligations to the audience, primary materials, and one’s own editorial goals, but must also include producing a reliable–but flexible –electronic text.
Increasingly, visions of future electronic editions put more of the editorial process in the hands of the reader. In a 2004 essay, Peter Robinson advocates “fluid, co-operative and distributed editions” which will be “the work of many, the property of all” (“Where We Are”). In these fluid editions, readers perform editorial acts to primary documents. A reader, for example, might “want to attach commentary, annotations, or translations” to a digital manuscript, import additional manuscripts not included in the original electronic edition, “make his or her own edited text, perhaps by taking over an existing edition and substituting his or her own readings at various points,” or even continue the markup by encoding unmarked elements (such as people names or places) to allow for linking and data analysis (Robinson “Where We Are”). These fluid, cooperative editions will “present materials which can be dynamically reshaped and interrogated, which not only accumulate all the data and all the tools used by the editors but offer these to the readers, so that they might explore and remake, so that product and process intertwine to offer new ways of reading” (Robinson “Where We Are”). This model of scholarly edition which opens itself to readerly interrogation and intervention parallels open source projects, such as Wikipedia, which succeed precisely because they create “an architecture of participation” (qtd. in Shirky “Gin, Television, and Social Surplus”).
In “Reflections of Scholarly Editing” (1996), however, Tanselle maintains the necessity of historical expertise and the vital role specialists play in interpreting textual artifacts of the past. In response to “theorists who say that readers should decide for themselves how (or whether) they wish to alter documentary texts,” Tanselle suggests that readers might want the historical knowledge of experts, as they themselves might not be “equally qualified to engage in historical reconstruction, which involves knowledge as well as imagination” (“Reflections”). Countering the objection that critical editions are too prescriptive, Tanselle asserts that readers can choose for themselves “how much or how little they wish to rely on the historical activities of the readers who preceded them” (“Reflections”). These activities, which include editing and publishing scholarship, are the products of systematic efforts to interpret the past. In Tanselle’s view, readers might not be altering documentary texts themselves, but they do retain the power of choosing how much of the critical apparatus and scholarly research to incorporate into their reading.
Regardless of whether readers actually encode and edit electronic editions, the future of scholarly digital projects still rests with its readers. As Robinson writes, “the best guarantee that an electronic edition should remain usable is that it should be used” (“Where We Are”). In an effort to increase transparency and maintain usability, we at the Blake Archive continuously reexamine and revise our tools, goals, and procedures. At the moment, projects under discussion include a major site redesign, a new imaging application (Lightbox) that will allow users to annotate and manipulate Blake’s images on their own computers, and an unofficial blog (The Cynic Sang) that attempts to shed light on working for the Archive while exploring wider topics in digital humanities.
Just as Blake’s work is “intercanonical” and “straddles two strongly defined conventional canons whose borders are institutionally guarded,” the work of the Blake Archive attempts to break new ground in digital humanities while staying true to many of the criteria for producing, and evaluating, a scholarly edition in print (Eaves “Graphicality” 105). Uncertainty about the long-term future of the Blake Archive, (indeed, about any digital project), motivates our goals and decisions today, though it leaves us “with no answer to the haunting question of where and how a project like this one will live out its useful life" (Eaves “Multimedia Body Plans” 218). Our struggle to best represent the complexities of Blake’s work situates us firmly within the realm of x-editing, but it also links us to Blake himself, whose own experiments with print technologies led occasionally to failure, but also, more often than not, to invention.