Editing Tools
There is as yet no modern edition of the distinct version of Alfred’s Boethius preserved in the earliest extant manuscript, British Library MS Cotton Otho A. vi. Following its Latin source, this mid-tenth-century manuscript, which was severely damaged by fire in 1731, presents the text in a five-book structure with alternating prose and verse. The other surviving version, the twelfth-century Oxford, Bodleian Library, Bodley 180 manuscript, is arranged in forty-two chapters, all in prose. Before the fire Francis Junius collated Cotton Otho A. vi with his own transcript of Bodley 180. In view of the great difference between the verse sections of Otho A. vi and their corresponding prose renditions in MS Bodley 180, which made a simple collation impossible, Junius providentially chose to copy all of the verse parts in their entirety. Today Junius’s transcript, MS Junius 12, is the only source for those verse passages in Cotton Otho A. vi that the fire later destroyed.
In the standard modern edition of King Alfred’s Old English Version of Boethius “De Consolatione Philosophiae” (Oxford 1899), W. J. Sedgefield used as much of the prose of Cotton Otho A. vi as he was able to see through the fire damage, but he replaced the “meters” (as they are called) with the prose versions of Bodley 180 and also followed its forty-two chapter structure, instead of the five-book structure of the earlier manuscript. Other modern editions present the meters all by themselves, extracted from their original context. Founding the text on digitally enhanced ultraviolet images of the burnt Cotton manuscript, supplemented where necessary by Junius’s collations and transcripts of the meters, the Electronic Boethius will thus constitute the first edition of the earliest manuscript.
The immediate goal of an image-based electronic edition of Alfred’s Boethius is to restore as much as possible a clear view of the earliest surviving manuscript, and the only one from Anglo-Saxon times. A restored view of the prose-and-verse version in this manuscript is intended to provide support for the new print edition, based on this unique version, that Malcolm Godden and Susan Irvine are preparing in The Alfredian Boethius Project: Anglo-Saxon adaptations of the De Consolatione Philosophiae. Edition Production Technology (EPT) is a digital workbench that can preserve and display manuscript features while still making the text accessible to scholars and students alike. An image-based electronic edition of the largely inaccessible Cotton manuscript helps clarify the seemingly upside-down manuscript tradition of Alfred’s Boethius. The following description of the editing tools in the workbench is from Kiernan, “Remodeling Alfred’s Boethius with the tol ond andweorc of Edition Production Technology (EPT),” forthcoming in Making Sense: Constructing Meaning in Early English, eds. Antonette diPaolo Healey and Kevin Kiernan (Toronto Old English Series, University of Toronto Press, 2006).
The general idea of an apparatus for an image-based electronic edition is that an editor embeds information in unrestricted virtual space, rather than appending it in the limited areas of marginal notes, footnotes or endnotes, as in print editions. One advantage of an electronic apparatus over a traditional print one is that it can remain totally out of sight, yet have all its information instantly available for automatic searching and display. With a print version of the same information, the reader must hunt for the information and then decode its sigla and abbreviations before getting meaningful access to it. The apparatus of an electronic edition is built through systematic markup of the manuscript resources. The Boethius EPT workbench provides a menu of tools, including, ImagText (for linking images and text), Glossary, DucType (ductus + type, for paleographical analyses), Quires, MIP (for minor image-processing), SaMod (for collating) and Search. The icon for Team indicates the capability of all members of the Electronic Boethius research team to synchronize their work, commit individual changes, or update with the server in a controlled version system that protects the team’s work from accidental (or deliberate) losses.
[EPT toolbar of editing environments]
To understand how the EPT manages all of this complex interchange of information, a brief introduction to the purpose of editorial markup is in order. The editorial apparatus, as well as the computer’s dexterity, is initiated and empowered through encoding, also known as tagging and markup. The EPT markup is specifically designed for humanities scholars to provide descriptive encoding in plain text, which is not only intelligible to an editor of Old English texts, but also is independent of any particular software program. This freedom protects the encoding, as well as the text, from extinction, even if the software that runs it today meets a boethian shift in fortune. These diverse results are achieved through encoding, also known as tagging and markup. XML, eXtensible Markup Language, and its subset for humanities projects called TEI (Text Encoding Initiative), is a plain-text form of encoding that remains immune to changes in software.
The way the system works is that an editor first assembles a comprehensive set of elements and attributes that are required by the editorial apparatus. The elements and attributes are listed in a DTD (or document type definition), which defines by means of a complete list of elements the XML document structure of the edition. The DTDs developed for the Electronic Boethius project are extensions of TEI P4 (see TEI Guidelines for Electronic Text Encoding and Interchange) and include the tagsets for prose, verse, text-critical apparatus, and transcription of primary sources, plus several new elements and new attributes needed for the special conditions posed by image-based electronic editing. For the facsimile edition alone the apparatus must include, among other markup, tags for distinguishing quires, bifolia, folios, foliolines, words, prose sections, verse sections, damaged text, illegible text, offsets, and shinethrough; scribal hands, letterforms, abbreviations, punctuation, spacing, added text, deleted text, corrected text, omitted text, glosses, other languages; and the various kinds of scribal formatting from word-division to the larger issues of scribal “markup” for rubrication for book or chapter divisions and for prose and verse headings; as well as encoding to identify technological restorations from ultraviolet and editorial restorations from Junius 12 (replacing the temporary Old English Corpus ones from Sedgefield and Krapp) to provide a context for the fragmentary readings that survive in the manuscript.
The editing is done through a series of operations that progressively encode the text with searchable and otherwise accessible and displayable information. With a properly encoded text with carefully designed transformation scripts, the user may display the edition in a way suited to a scholar interested in seeing the details of damages and reconstructions, scribal features, editorial emendations and textual notes, or to a student only wanting to read an uncluttered text with ready access to an online dictionary and explanatory notes. The XML-based tools that make sense of manuscript features through systematic encoding integrate images and text, organize and display folios and quires, assemble pragmatic manuscript glossaries that reconcile the unexpected word divisions of script with real word boundaries, facilitate detailed paleographical analyses, mine statistical tendencies in preserved parts of the manuscript to enable plausible reconstructions of lost material, facilitate collations and restorations, and provide the means for multiple display options and complex searches of everything gathered.
To start a project in our system an editor simply creates an XML project file, a plain-text transcription or other base text, which must only include in its preliminary encoding the name of the collection, the name of the manuscript, and the name of the work. As there is no base text for the prose-and-verse version of Alfred’s Boethius, and it is impossible to make a meaningful transcript from the severely burnt and fragmented Cotton manuscript, the Electronic Boethius uses as a base text a compilation of Sedgefield’s and Krapp’s editions. With the permission of the editor of the Dictionary of Old English project, the editor compiled an electronic base text from Toronto’s Old English Corpus (OEC) files of Sedgefield (BoProem B9.3.3, BoHead B9.3.1, Bo B9.3.2) and Krapp (Met A6). OEC files divide their texts into meaningful units, by chapter, page, and line number in the case of the Boethius prose, and by meter number and line number in the case of the verse. This rough and ready base text thus re-enacts the process Alfred went through as he revised his all-prose draft to complete his prose-and-verse translation. At the same time, by putting the verse back were they belong and by removing the prose versions, this compilation of a base text ends up reversing what Sedgefield did in his edition. The base text thus supplies all of the parts of the text that were destroyed by the fire but saved by Junius, and subsequently used for restorations by Sedgefield and Krapp. Once the plain-text file and a set of images are assembled in folders in a project workspace, the EPT is ready for action.
ImagText
After opening the EPT from a desktop icon, the editor chooses a project (Cotton, Bodley, or Junius) from the Navigator, which displays all available files in the left frame. When a project opens, the default editing environment, called ImagText, appears in the main frame. ImagText, as its name suggests, is designed to integrate image and text. As such, it is the conceptual model for most of the other tools in the image-based electronic edition. All editing environments are controlled by the underlying xTagger software, which performs all the integrative tagging between the text and image. ImagText opens with the base text in the lower panel and, after preliminary encoding, a corresponding image in the upper panel. The editor is free to rearrange these default settings and to have more than one image showing at the same time.
[Full view of ImagText - click for larger view]
xMarkup
The narrow frame on the left is reserved for three utilities in addition to the Navigator: 1) an expandable menu called xMarkup, which organizes the encoding and opens with a click of the mouse the various templates that guide the editor through the encoding; 2) a Bookmarks facility to help the editor quickly find key places as the work progresses; and 3) a Thumbnail image for easily navigating the large image in the top frame of ImagText.
The xMarkup menu houses a wide range of templates organized under eight headings (Start Transcript, Glossary, Condition, Paleography, Codicology, Collation, Restoration and Edition) for encoding the relationship between parts of the text and the images on which it is based. An editor may customize this organization and the templates themselves, based on the particular needs of a project, without resorting to new programming.
[xMarkup menu and Folio templates]
Because the manuscript is so severely damaged, the encoding of the Cotton manuscript is extremely difficult and tedious, but the templates organize the work and steer the editor away from human errors of omission and commission induced by fatigue and boredom. The Cotton manuscript is so damaged by fire that it is almost always necessary to refer to ultraviolet images just to identify the boundaries of folios and foliolines. The Electronic Boethius therefore uses ultraviolet images as the foundation for all encoding. The first stage of the markup associates the image of each folio in the manuscript with the base text by identifying the boundaries of folios, the boundaries of folio lines, and the boundaries of words.
Editors do not want or need to think about, much less look at, the underlying markup that makes it possible for a computer to create a searchable text or glossary; to display an edition with or without scribal abbreviations; to show or hide brackets or other means of notation for emendations or conjectural restorations; to compile variant readings; or to record any of the other essential features of an editorial apparatus. We have accordingly built the editor’s working environment to focus on the editing and to keep the encoding in the background. The editor simply chooses an aspect of the apparatus for encoding, such as damaged areas, and then fills in the template, systematically encoding the edition while minimizing the possibility of errors of omission and commission. The editor may view, of course, the underlying encoding at any stage, but it is not necessary. In this way an editor can go about the real business of creating a scholarly edition without worrying about the tagging, just as one may use a word processor without knowing how its encoding works.
The systematic markup or encoding of folios and foliolines pays many dividends as the electronic edition grows. As the editor fills in and tags information in the templates, the EPT automatically inserts the markup in the transcript. The editor can reassuringly choose to show or hide this underlying encoding at any time. Once this preliminary markup is completed, anyone can search throughout the manuscript by folio and folioline and view the corresponding images in bright light or ultraviolet, because the transcript of each folio is now permanently linked to its images.
OverLay
For the tenth-century manuscript of Alfred’s Boethius, it is often impossible to read the text without the aid of ultraviolet fluorescent effects, which rescue much of what seems to be totally destroyed in the manuscript. At the same time the editor and eventual users of the edition will want to see the way the manuscript actually looks without special lighting techniques.
[Comparing images with OverLay]
The OverLay tool provides an easy and effective means to compare in minute detail images produced under different lighting conditions or at different times. The user simply chooses two corresponding points on both images and EPT places one image on top of the other one.
[Examining UV effects in OverLay]
After the images are superimposed, the editor can choose any area (or the entire folio) to examine and encode. A slide bar in the bottom left shows the underlying section chosen for examination. Employing xTagger, the editor can then encode the selected regions of both images simultaneously. Because it reveals what survives of the folio, OverLay is especially useful for recording areas of damage and showing where the manuscript is gone and requires restoration from Junius 12. The tool is indispensable for determining the folio-line boundaries and, when the damage is severe, even for determining the most probable boundaries for the folios themselves.
The OverLay tool is highly adaptable for a range of common manuscript problems. Because it does not change the original images, but only creates a new, layered, version of both, OverLay is an ideal tool for testing hypotheses for conjectural restorations. With a tool included for minor image-processing (MIP) or with professional image- processing software, the editor can remove obscuring shine-through or post-medieval additions. The OverLay facility can usefully supply the missing rubricated initials, for example, or more significantly test the validity of conjectural emendations of apparent errors and conjectural restorations of lost text.
The ability to compare images of the manuscript in bright light with ultraviolet enhancements furnishes many new insights into the text preserved in the 10th-century Cotton manuscript. For example, because of its ruined state, scholars have not understood the clear and careful formatting of the prose-and-verse version in the Cotton manuscript into five books, instead of the chapter divisions of the later Bodley manuscript. The Cotton scribe has taken pains to mark the book divisions, but Junius, because he emphasizes the chapter divisions of the Bodley manuscript, tends to obscure the formatting for books in his collation of the Cotton manuscript. The ultraviolet images of these sections in the OverLay configuration illustrates how this tool gives us a better understanding of the structure of the Cotton version.
[UV showing book formatting in Cotton]
Because the change from Book I to Book II in the Cotton manuscript does not coincide with the book division of the Latin source, Janet Bately has suggested that “its incorrect position is the work of an incompetent scribe, seeking to make good a perceived deficiency” (“Book Divisions and Chapter Headings in the Translations of the Alfredian Period,” Early Medieval English Texts and Interpretations: Studies Presented to Donald G. Scragg, eds. Elaine Treharne and Susan Rosser (Tempe, AZ 2000), p. 157). The OverLay with ultraviolet shows, however, that the scribe competently marked (or encoded) the new book division in four different ways. First, he marked the end of the book with special punctuation (“.,.”). Next, he drew a large capital H to prominently announce, Her endað sio forme boc boetius 7 onginneð sio æftere, “Here ends the first Boethius book and the second begins.” Then he left the next three and a half lines empty to emphasize the end of the book.
Starting Book II on a new page on the verso, the scribe similarly left a large space at the beginning of the first three and a half lines for a rubricator to supply a large initial. The scribe’s careful formatting of the new book division should leave no doubt that the book division is not incorrect, that Alfred intended to begin his revision of the second book of his Latin source at this point. Although this is not the place to show how Alfred revised Book I to lead to this new division between books, it might be noted in passing that Sedgefield’s austere formatting and punctuation submerges the book division (even while restoring Bodley’s omission of it) and enforces a sense of continuity with Bodley chapter XVI (38.6). Through its ability to disclose nearly invisible aspects of the manuscript, the OverLay tool not only gives us a better appreciation of the scribe, but also a better understanding of the earliest manuscript of Alfred’s Boethius.
Glossary
The Glossary tool provides a comprehensive set of grammatical templates for preparing an exhaustive glossary. The tool allows us to describe and simultaneously tag all parts of speech in the course of creating a glossary for both manuscript and edited text. Drawing on the previous markup for folios, foliolines, and words, the Glossary tool first generates an alphabetized wordlist with precise manuscript and edition locations for each occurrence. Glossary templates for each part of speech open with fields already filled with the word to be glossed, and its location.
[Glossaries for script and print]
The editor selects the tab for the appropriate part of speech and safely undertakes the challenging task of creating a complete glossary of the 50,000 words in Alfred’s Boethius. The program keeps track of work-in-progress by flagging with a box in red any unfinished entries or questions for the editor, and flagging in green completed work. As the alternation of prose and verse in the Cotton manuscript is encoded, too, the Glossary automatically keeps track of words that are only used in prose or only used in verse. When the glossary is complete, the interchange of the tools means that the reader can click any word in the transcript, even when it is divided by scribal spacing, and it will open in the glossary, or click any word in the glossary and it will open in its correct location in the edition.
Some of these features may be briefly illustrated with the word boc. The Dictionary of Old English lists many different meanings for boc, two of the primary ones being “book” in general (A.1), and “major division of a larger work” (B). Alfred’s Boethius uses both meanings, and frequently uses boc and bec (its form in genitive, dative and accusative singular, and nominative and accusative plural) in reference to one of the five books in both manuscripts.
[Definitions of boc]
The entries for the nominative form, boc, include two examples in the prose proem (ppr) from a folio destroyed in the fire. Lost folios are represented by 00, until they are assigned a folio number after their virtual restoration; thus, the next entry, boca, genitive plural, occurs in line 52 of the first meter (00m1.52), from another lost folio. They both use the first meaning of a book in general. The remaining three examples refer to the book divisions between Books I and II, III and IV, and IV and V. In the missing division between Books II and III, Alfred has used the hapax compound word, froferboc, “book of consolation, comfort-book.” In the manuscript frofer ends line 7 and boc begins like 8, but the Glossary automatically lists it correctly under froferboc. Because the editor has identified <w>frofer boc</w> through underlying markup as a single word, the Glossary tool is able to achieve this level of native Anglo-Saxon literacy, without the modern support of a hyphen, even after adding manuscript spacing, in this case, a line boundary. The preliminary markup of words allows the editor to insert scribal spacing and other manuscript features without adversely affecting the glossary.
[fol. 38v7-8 <w>frofer boc</w>]
By encoding modern word boundaries, the Glossary tool also helps modern readers understand a manuscript and its scribal word divisions with the skill of a native Anglo- Saxon. On the surface is the text of a document in an image-based electronic edition, but beneath the surface there is a text of the work, grounded in the manuscript evidence.
As this example also illustrates, the EPT filters out markup, which is amassing into a vast sea of angle brackets, far exceeding the size of the Old English text as encoding of it proceeds. But that sea of markup, all mercifully concealed from the reader, makes it possible to search and find ever widening kinds of information, from words with scribal spacing to definitions, from faded or lost readings to Junian restorations of them, from ruined folio boundaries to illegible book divisions.
DucType
The process of encoding or tagging or marking up the edition for subsequent access, most obviously in searches by users of the electronic edition, but also by other tools in the EPT, is well illustrated by the special DucType template for paleographical encoding. DucType is a portmanteau for ductus and typeface, emphasizing its integration of image and text. It helps the editor gather detailed information about the scribal handwriting. To record paleographical features, the editor opens the DucType tool from its icon on the menu bar, and a letter template appears in the ImagText editing environment. Because folios and foliolines are already encoded, the editor can simply choose a letter from the text and Ductype supplies the letter, its location, and a model of the selected letter in the template.
[DucType paleographical encoding]
When the editor drags the mouse over the corresponding letter in the image, the scribe’s sampled letter appears in the template, too, and the coordinates are automatically recorded in the markup. The editor then chooses the upper, body, and lower properties of the letter from a drop-down menu to complete the encoding of the letterform. The amount of paleographical detail is up to the editor, but the list of features is easily customized and expanded.
A simple interface for adding, deleting, or editing paleographical features in the template allows the editor to modify the DucType template without the need to hire a programmer to change the code. If, for instance, the editor comes to a crossed þ for the first time, it is easy to add the abbreviation stroke to the upper features of þ. Macrons and accents above letters are similarly treated as upper features of the relevant letterforms, making it possible to search for all the instances of a, as well as for all the instances of a with an accent, a with a macron standing for an abbreviation, or any combination of these searches. Ligatures are treated under the first character, so that high-s with w or p are both listed under s; the many ligatures with the e-tongue (e.g., ea, ec, ef, eg, etc.) are listed under e.
StaTend
Used in conjunction with a tool for gathering basic statistical tendencies (StaTend), DucType helps refine our understanding of the frequency distribution of scribal letterforms in the manuscript. To learn about the distribution of letters, spaces, and lines for any folio, the editor types in a folio number and presses the Calculate button. The display shows the line count, the average number of letters per line, and the total number of characters on the folio, along with a view of the folio showing the precise figures. To avoid skewed results, the editor may choose to exclude content governed by any particular encoding.
[StaTend excluding selected markup]
In his description of the hand of Cotton Otho A. vi in his Catalogue of Manuscripts containing Anglo-Saxon, N.R. Ker states that the use of eth is “regular,” while the use of thorn is “rare” (Oxford: Clarendon Press, 1990), p. 218. The StaTend tool reveals that þ is actually used 22 times on fol. 38v, compared to 18 times for ð. Excluding instances of “damaged text” restored by Junius 12, StaTend shows that in surviving cases þ is still used 22 times compared to ð 14 times.
The StaTend tool is also very useful as a guide for virtual reconstruction of folios destroyed in the Cotton fire. Using data from the surviving folios and the Junius transcript and collations, the editor can make informed judgments on the distribution of letters per line, lines per folio, and even probable folios per quire, in those places where the manuscript was destroyed, but the text was preserved by Junius. To learn about the distribution of letters and words and lines in folios near or surrounding a lost folio, the user enters a folio number and studies the results in the display. Another button reveals the same kind of raw information for the entire manuscript.
Folio Reconstruction
Armed with this information, the editor can conjecturally restore the general dimensions of the lost leaf. For example, we can tell from Junius’s transcript and collation that a folio is missing between folios 127 and 128. We can also determine the probable distribution of the text on the lost folio by observing that the surrounding folios contained 27 lines per folio averaging about 34 characters per line. With the help of a rudimentary interface for folio reconstruction, the editor can distribute the unformatted Junius restoration into 27 “manuscript” lines averaging 34 characters per line for each side of the virtual folio. The interface lets the editor use word boundaries or divide the lines more precisely at the number of letters and spaces per line.
[Folio Reconstruction]
The editor then encodes the recto and verso of virtual folio 128A (named for its position between folios 127 and 128) for foliolines. Fol. 128A is of course only a rough estimation, but it is based on information of the surrounding folios, and so the lost folio presumably contained the same number of lines per page. The editor now has something to work with to fit the apparent circumstances.
[Virtual fol. 128Av]
For example, a prose section ends and a verse meter begins on the verso of virtual folio 128A around line 14. We know that in this manuscript the scribe always leaves an opening space of 3-4 lines for a rubricated initial at the start of both prose and verse sections. The editor should therefore distribute the lines in the same way, before tagging the virtual folio for folio lines. Through this process of conjectural reconstruction based on statistical tendencies the editor can learn more about the lost parts, as well as the surviving parts of the damaged manuscript as well as the lost parts.
Before the fire, Cotton Otho A. vi was a codex made up of bound gatherings of folios. Although the fire destroyed all the gatherings, the individual folios still contain residual information about the original gatherings. This evidence includes notes mentioning the former misplacement of leaves, the hair and flesh arrangement of the surviving leaves, and the number of rulings, as distinct from the number of written lines, per folio. Unless contrary signs appear, it is reasonable to presume that the original sheets were most likely gathered in true quires, groups of four, which provide eight folios, or sixteen pages, per gathering. As the charts from StaTend suggest, the last sixteen pages, including the virtual fol. 128A recto and verso, of the Cotton manuscript may have been a four-sheet quire.
[Statistical tendencies of the last 8 folios]
The number of lines per folio, as well as the hair and flesh arrangement of the surviving leaves, previously encoded by the Folio template) supports the hypothesis that the last eight folios of the Cotton manuscript comprised a natural quire of eight folios. The Quires tool can begin to make sense of the statistical tendencies.
Quires
Quires is designed to facilitate the encoding of sheet collations of the gatherings of a codex. The template for constructing the quire diagrams makes it easy to change sheets to half-sheets or leaves, if the hair/flesh pattern shows that an expected sheet does not have conjugate leaves. Once again, because the folios are already tagged, the Quires tool can draw on previous encoding to automate the encoding of entire gatherings. The editor simply enters the number of the gathering and the number of leaves in it. Through the interface the editor can revise the hair and flesh patterns of the leaves and arrange them into sheets or half-sheets as the evidence leads. While 128A may have been part of an original sheet, it is more likely that it was a folded-in leaf, which would account for separation from the codex during or after the fire. For manuscripts like Bodley 180 that are intact and may even contain supporting sheet or quire signatures, the tool makes quick work of showing the makeup of the medieval book. But it can also conjecturally reconstruct lost gatherings in the many manuscripts, like the prose and verse version of Alfred’s Boethius, whose gatherings were destroyed by time.
[Constructing quires]
By constructing lost gatherings, Quires and the diagrams it made lets us visualize the sheet collation and the ruined quires.
[Interactive sheet collations]
After the diagrams are constructed, they become interactive with the images of the folios that comprise the gathering. If one chooses the middle point of a sheet, for instance, Quires will display the conjugate leaves; if one chooses a diagonal line, H (hair) or F (flesh), it will show the recto or verso of one leaf from, in this case, sheet 128-124. The H/F distribution is easily changed as the evidence requires. If the reconstruction is correct, these two folios once comprised a single folded sheet (or two folded-in leaves, if not conjugate) in the original gathering. The user may examine these folios in detail by enlarging them through the Quires interface.
[Folios and quire collations]
To complete the theoretical reconstruction of the gathering, the editor can provide a place-holding image of the lost folio, virtual folio 128A.
As we have seen, many tools contribute to the process of reconstruction. The xMarkup of folios and foliolines provides StaTend with the statistical tendencies for all folios in the manuscript. The Folio Reconstruction feature in StaTend sorts the Junius restorations following these statistical guides and, when the editor is satisfied with the layout, xMarkup encodes the folio and foliolines of the reconstructed folio. Quires, too, draws on the same encoding to markup groups of folios into gatherings. The diagrams, which the Quires tool generates, are all dynamically linked to the images of folios these diagrams represent. Even the DucType tool for paleographical analysis contributes to the process of reconstruction. By saving the exemplars or models of the individual letters, the editor accrues through DucType a comprehensive library of letterforms, each described and encoded by its upper, body, and lower features. These images in turn provide the means for creating instructive facsimiles of lost words, phrases, lines, folios, and even quires of the fire-ravaged manuscript.
RamSome
The RamSome tool, so named because generating a facsimile on the fly requires some RAM and the result looks like a medieval ransom note, transforms the plain-text form of the virtual folio into a visualized representation, using the scribe’s own letters.
[RamSome letters]
No one with any knowledge of paleography would mistake this facsimile for a real folio. For one thing, the ascenders and descenders do not invade each other’s upper and lower minim spaces as they do in real manuscripts, which renders the virtual facsimile considerably longer than it should be. Unlike the scribe’s handwriting, each letter is separate and distinct, with no ligatures, no variety in letterforms, no use of abbreviations or other distinctive signs a scribe manifests in the course of creating a unique copy. It is, in short, quite like the plain-text version it has transformed.
Notwithstanding its evident shortcomings, the RamSome facsimile remains an effective pedagogical tool. The facsimile is not as static as it appears. Each letter is in effect a drop-down menu, from which the user, with a control-sequence, can select any other exemplar of that letter saved in the letter-library. If, for example, the library includes the scribe’s habitual sc-ligature with high-s, the editor may replace the non-ligatured low-sc instance in line 2 and elsewhere. Likewise, most of the e’s and æ’s on the page should be ligatured with the e-tongue to the following letters (e.g., in order of occurrence, æc, ea, es, en, ew, er). In view of the scribe’s practice, the crossed-thorn abbreviation should probably appear for the two occurrences of þæt in the last line of the facsimile, and there are other possibilities for abbreviations with macrons over vowels on the page. Because all abbreviations are encoded, an established system of markup based on statistical tendencies in the surviving parts of the manuscript could be employed to produce a more realistic representation of the scribe’s use of abbreviations, as well as of variant letters and ligatures. The descenders and ascenders could even share the same space between lines with an adaptation of the OverLay tool, not to create an illusion or a forgery, but to learn the personal ductus of the scribe and the general aspect of the page.
SaMod
SaMod (Old English “together” ) is a collation tool we are developing for comparing
different versions of the same text and automatically encoding variants in a base
document. At its simplest level, then, it can compare a prose section of the text from
Cotton with the corresponding passage from Bodley, and then tag the Cotton text with
any differences found in Bodley. In other words, it is designed to do what Junius did in
his collation, only more accurately and thoroughly, and without messing up the
appearance of the original text, because the markup is out of sight. The process is more
complicated than this description implies, however, because “Cotton” is at the outset a
base text that combines Sedgefield and Krapp for the lost parts of the Cotton manuscript.
To establish an accurate transcript, the editor identifies by means of ultraviolet images in
ImagText what actually survives in the manuscript, and encodes as “damaged” anything
else that is restored by the Sedgefield and Krapp texts, which are in turn based on Junius
12. Thus, in collation with the restored damaged sections, Junius 12 should agree with
Sedgefield and Krapp, except for their emendations. SaMod refines the
[SaMod Verse vs Prose Meter - click for larger view]
However, we are also developing SaMod to deal with the far more complicated problem that Junius encountered when comparing the Bodley prose meters with the Cotton verse meters. In this role, SaMod is designed to help a scholar analyze the substantial differences between texts with the same essential source, but radically different representations of it, one in prose and one in verse. In this case, instead of locating differences, SaMod finds and highlights identical words and phrases. For example, in the collation of the verse meter on Homer (Cotton Meter 30) with the prose meter, SaMod would highlight in the first line only the name Omerus and the phrase in crecum, “among the Greeks.” SaMod lets the editor set rules (to ignore minor spelling differences, for example, or to establish synonyms as acceptable matches). The idea is to provide an analytical setting for understanding the precise relationship between the two Old English versions, which according to the traditional view is that the verse is a translation of the prose. By making it convenient to examine the versions in close proximity, SaMod is likely to lead to new insights about the two translations and perhaps even reopen seemingly settled issues, such as the common descent of the two translations, and maybe even the order in which they were made.
Search
The last tool is the biggest beneficiary of all the others. The Search facility has the capacity to find and deliver all of the information systematically encoded by all the various templates of the other tools. Through its organizing power, the Search tool can sort through the myriad of angle brackets. The most difficult aspect of the Search is designing an effective “instrument panel” for the end-user. In order to deliver pertinent information, the user must be able to define and narrow a search. One of the constraints we have imposed on ourselves is the assumption that most humanities users will have no knowledge of technical query languages. As a result, we must design an interface that provides maximum access with minimum help from the inquirer.
The interface we have designed, which is modeled on the ones for edition and transcript searches of the Electronic Beowulf, provide access to both unstructured and structured (i.e. encoded) searches. An unstructured search can reveal a great deal of information that would otherwise be difficult to find. The statistics on the occurrence of þ compared to ð in StaTend, for instance, which is essentially the result of a substring search, is an example of an unstructured search. Anytime an unstructured search narrows its range to a group of folios, to a particular scribe’s work, to a series of lines, to a metrical section, or to a particular chapter or book, however, the search has availed itself of the encoding of a structured search. Indeed, everything on the user panel is an implicit use of markup.
[Search panel with defined search]
In this example in Figure 22, a boolean search of the entire Boethius is intended to discover all cases of boc and bec and their compounds, and to avoid all false hits of (-)bæc.
[Boolean search result]
But the results are structured by the folio ranges and by the requirement that the search goes by folio lines, rather than by some other feature, such as verse lines. Unstructured searches have the advantage of being unique and unrestricted, and as such can lead to new knowledge and innovative interpretations of it. But, as powerful a research tool as they are, unstructured searches cannot find any of the kind of information the templates encode.
Without encoding, a Search facility cannot find a quire, a sheet, a folio, or a folio line, nor recognize a noun, an adjective, a pronoun, a verb, an adverb, a conjunction or even a preposition. The distinction between an editorial emendation and a conjectural restoration eludes it, and it sees no difference between damaged and undamaged readings, with or without ultraviolet fluorescence. Textual notes and explanatory notes are just plain text, like everything else. Inflections, abbreviations, and past participles do not exist. The Search could not tell a book from a chapter in any light, not to mention a prose meter from a verse meter. With tags, a Search facility can make sense out of chaos. A reader can open a transcript search, for instance, and quickly locate all damaged letters that are made more legible by ultraviolet:
[Search of damaged letters enhanced by UV]
In a search of the edition a textual scholar can find all cases of conjectural restorations between lines 456 and 765, examine them one by one or as a group, and click any result to go the full text linked to an image of the folio.
[Search of emendations and restorations]
And while the edition may expand all scribal abbreviations, as is customary in editions of Old English texts to make them easier to read today, the interested scholar may search the transcript for any or all scribal abbreviations, and for anything else that the editor has taken the pains to encode.
[Search of abbreviations]
As we have seen, the EPT allows us to reconstruct many of the features of the prose and verse version preserved in Cotton Otho A. vi unavailable to scholars in any of the editions and indeed largely unviewable in the manuscript itself without the aid of special lighting and the enabling tools described above. The new evidence allows us to re-evaluate some of the things that do not seem to make sense in the manuscript and editorial traditions of the Alfredian Boethius. This formerly inaccessible and obscure evidence includes words and parts of words that seemed gone, even in bright light under the persistent and searching eyes of scholars like Sedgefield; the special attention the scribe gave to book divisions and the consistent formatting of the alternation of prose and verse that characterize this manuscript; the scribal word divisions imbued with meaning; the vestiges of the ruined gatherings and the recovery of lost folios and lost quires.
With the earliest manuscript of Alfred’s Boethius in the best shape it has been since the day before the fire of 1731, scholars and editors will have the opportunity to reassess the theory that the all-prose text of the Bodley manuscript precedes the prose-and-verse text of the Cotton manuscript. There is no reason to doubt Alfred’s statement that he first made prose drafts of the meters before redoing them as verse. But there are sufficient grounds to question whether Bodley 180 goes back to these drafts or instead represents a post-Alfredian creation of an all-prose version, made by scholars in Anglo-Saxon times who simply wanted a better text for philosophical inquiry than the Old English verses supplied. This new theory would make new sense of the manuscript tradition, in which Bodley claims that it contains verse, and provide the impetus for a new editorial tradition. For a more complete context for this new theory, see Kiernan, “Remodeling Alfred’s Boethius with the tol ond andweorc of Edition Production Technology (EPT),” forthcoming in Making Sense: Constructing Meaning in Early English, eds. Antonette diPaolo Healey and Kevin Kiernan (Toronto Old English Series, University of Toronto Press, 2006).
Back to Headings |
© 2005 Electronic Facsimiles & Texts | Last modified by KSK  |