The ELECTRONIC BEOWULF Project

Electronic Beowulf

Kevin Kiernan and Paul E. Szarmach, Academic Directors
Kevin Kiernan, Editor of the Electronic Facsimile


The Electronic Beowulf project began as part of the British Library's "Initiatives for Access" program to make its collections more available to the public through new technologies. It won the 1994-95 Library Association / Mecklermedia Award for Innovation Through Information Technology for the technique of using fiber-optic light and an electronic camera to reveal hundreds of covered letters and parts of letters along the damaged edges of the manuscript. The long-term goal of the project is to assemble an open-ended electronic archive of materials founded on, but by no means limited to, the only surviving manuscript of Beowulf and ancillary texts. In other words an electronic archive should grow as any other archive.


Overview

An Overview
(November 1995)

The Electronic Beowulf project has so far assembled a huge library of digital images of the Beowulf manuscript and related manuscripts and inaccessible printed texts. The archive includes, in addition to color images of each page, hundreds of fiber-optic readings of hidden letters and ultraviolet readings of erased text from the early 11th-century manuscript; full electronic facsimiles of the two indispensable 18th-century transcripts of the manuscript by G.J. Thorkelin and his hired scribe; the first edition of the poem by Thorkelin (1815); the first collation of this edition with the manuscript, showing readings that were visible to J.J. Conybeare in 1817, before the 19th-century binding frames covered the burnt edges of the manuscript; and selections from other documents needed to restore the damaged text. The model for such an archive is a special collection in a rare book library rather than a single book. My current role as editor of the electronic archive is to establish an accessible resource of these images for scholars and editors.

An image archive is fundamentally different from a text archive, comparable in some respects to the distinction between Kemp Malone's facsimile editions of the Nowell Codex and the Thorkelin Transcripts and Frederick Klaeber's edition of Beowulf. Imagine, however, a situation in which Malone had all of the photographs of the Nowell Codex and the Thorkelin transcripts, but no tried and true way to publish them. That's more or less the situation we face with the Beowulf images. We have hundreds of high-resolution, full-color images, each one of which is 21 megab ytes, each one far too big for any ordinary computer. The only way to solve this primary problem of size is to scale the images to sizes compatible with standard equipment. Fortunately there are image formats that provide considerable compression withou t apparent loss of resolution so that the compressed images far exceed the appearance of any published black-and-white facsimiles.

GRENDL

GRENDL

Another big difference between an image and a text archive is that computers cannot yet search and otherwise manipulate images, even ones of manuscripts that contain texts, the way they so efficiently do with plain texts. Neither TEI (the Text Encoding Initiative) nor SGML (Standard Generalized Markup Language) can tag images, although it would be highly desirable to have an efficient way to markup images. Computer scientists are only beginning to work in this area, however. What scholars in the human ities urgently need is a well-focused IEI, or image-encoding initiative. Researchers in the Computer Science department at the University of Kentucky have begun to develop a "paleographical tool," which Brent Seales demonstrated in May at the Medieval Co ngress in Kalamazoo. The tool is actually part of a larger research effort to address the problem of searching images and multimedia in general. One result of an international symposium on "Reconnecting Science and Humanities in Digi tal Libraries" held at Kentucky from 19-21 October 1995, was the foundation of GRENDL, the Group for Research in Electronically Networked Digital Libraries, by the University of Kentucky, the British Library, and CIMATS (the Centre for Information Man agement and Advanced Technology for Scholarship) at London Guildhall University. Its aim is "to provide a new means of communication between the sciences and the humanities and to develop projects which will explore the computationally challenging proble ms presented by the humanities." Its first project will be concerned with the encoding of images.

Lacking for now the technological means to search images, however, I am assembling the Electronic Beowulf archive around individual folios of the manuscript. It can be argued, in any case, that the images themselves provide the most compelling "markup la nguage" of all for transcriptions derived from them. Each folio is accompanied by a transcription with cross-references to line numbers and linked to all other images that directly relate to it, and I welcome any suggestions for adding search capabilitie s to the transcriptions and other text parts of the archive.

My initial approach toward making the electronic facsimile available to others in a useful and usable form was to hire programmers to develop Graphical User Interfaces (or GUIs) for Unix, Mac, and PC platforms that would hypertextually link all related im ages in the archive. In short, I found myself in the software business, developing low-end software to open the images we had collected. The first version was made to run on a Unix platform running X-Windows with a higher level "widget set" of routines (in effect, a repository of mini-programs that perform specific tasks) called OSF/Motif. This GUI works well, but it cannot be run on Unix stations not using OSF/Motif without reprogramming it, and it cannot be run at all on Macs or PCs, the platforms us ed by most people in the humanities. To make the same type of GUI available on Macs and PCs it was necessary to develop entirely new and different programs, mimicking the Unix program but using the operating systems of the Mac and PC platforms.

The construction of a Mac version proceeded smoothly, because of the uniformity of Mac machines. The features of the Unix program were simply reprogrammed following Mac procedures and utilizing the Mac menu system. "MacBeowulf" is currently limited to t he image format known as "pict," however, requiring pervasive conversion from the "tiff" image files used by the Unix program and then relinking of the hundreds of images. The PC version, programmed in C, ran into lots of problems, because of the lack of uniformity among PC clones and because of the change in operating system in Windows95. The beta version we developed was limited to the image format know as "bmp," which once again necessitates massive conversion and, again, relinking of hundreds of ima ges. Because of the many ways the various clones distribute their color maps, moreover, we could never predict how well an image would display on any particular PC machine, even if it was capable of displaying our images. Instead of getting thousands of colors or even 256 colors required for adequate displays of high-resolution color images, the PCs would sometimes deliver only 16 or 8 colors, making the full-color, high-resolution manuscript pages look like they were colored in by children with either bad taste or a bad selection of crayons.

148v1-3

The separate development of three separate programs for the three major platforms is not a practical solution for other reasons as well. Commercial programmers would in all likelihood be able to work out all of the different kinks of the multiple platfor ms and produce stable programs to run on CD-ROM. But even then each GUI would quickly become obsolete, and the Electronic Beowulf project would not be able to provide ongoing support or updates for any of the versions. One solution we are pursuing is to develop a fourth version using the program "Tcl/Tk," which promises to provide cross-platform compatibility. We now have beta versions that work on Unix and to a lesser extent on PCs, but the development of Tcl/Tk itself has been slow and it is not read y to display large, high-resolution images like ours on any platform. The biggest drawback is that this approach places the emphasis on the interface rather than the archive, which ought to be the center of focus and ought to be readily accessible as an archive.

The best solution, in my opinion, is to switch directions entirely and take advantage of the recent remarkable advances in network browsers like Netscape and Mosaic. There are many apparent virtues to adopting a browser as an interface to the archive. O ne big advantage is that it separates the software, which the companies marketing browsers will continue to improve, from the electronic archive of images and the documents written to link them. Users could also study individual images in the archive wi th ordinary image viewers instead of browsers. Another advantage is that a single image archive in one format, such as jpeg, can be accessed by any platform -- Unix, Mac, or PC -- just as they all access the same sites on the World Wide Web. Browsers ar e not limited to use on the Internet and can in fact open local files on disk or CD-ROM. Users of the archive, therefore, would not be bound to an ethernet cable and access to the World Wide Web. Most people do not have, and in view of the limitations o f bandwidth may never have, unlimited access to the World Wide Web. A local archive such as a CD-ROM would certainly be a more convenient way to use the collection for ease of reference as well as for speed in opening the images. At the same time the Br itish Library would always be able to link the archive to the Internet, which remains the most flexible medium for adding new features and for establishing interactive contacts among people using the archive.


It is easy enough to illustrate the virtues and vices of using a network browser. A user must have a computer with Netscape 1.1 installed, an image driver and monitor capable of displaying at least 256 colors, and an Old English font, such as Times-Old E nglish, chosen as the preference font under Netscape Options.


Beowulf, Collation for fols. 129-130,


Folio 129 recto

Fol. 129r, verse lines 1-21b:

HWÆT WE GARDE
na ingear dagum. þeod cyninga
þrym ge frunon huða æþelingas elle(n
fremedon. Oft scyld scefing sceaþen(a)
þreatum monegum mægþum meodo setla 5
of teah egsode eorl syððan ærest wear(ð
fea sceaft funden he þæs frofre geba(d
weox under wolcnum weorð myndum þah.
that him æghwylc þara ymb sittendra
ofer hron rade hyran scolde gomban 10
gyldan thatwæs god cyning. ðæm eafera wæs
æfter cenned geong ingeardum þone god
sende folce tofrofre fyren ðearfe on
geat hie ærdrugon aldor (le)ase. lange
hwile him þæs liffrea wuldres wealdend 15
worold are forgeaf. beowulf wæs breme
blæd wide sprang scyldes eafera scede
landum in. Swa sceal ge(ong g)uma gode
ge wyrcean fromum feohgiftum on fæder

Vi 9


Notes

Thorkelin A and B, Conybeare, Madden:


Folio 129 verso

Fol. 129v, verse lines 21b-46b:

...r)me thathine onylde eft ge wunigen wi(l
gesiþas þonne wig cume. leode ge læsten
lof dædum sceal inmægþa gehwære mange
þeon. him ða scyld gewat toge scæp hwile
fela hror feran onfrean wære hi hyne 5
þa ætbæron tobrimes faroðe swæse gesiþas
Swa he selfa bæd þenden wordum weold
wine scyldinga leof land fruma lange
ahte þær æt hyðe stod hringed stefna isig
ut fus æþelinges fær. aledon þaleofne 10
þeoden beaga bryttan onbearm scipes
mærne be mæste þær wæs madma fela
offeor wegum frætwa gelæded. Ne hyrde
ic cymlicor ceolge gyrwan hilde wæpnum
heaðo wædum billum byrnum him on bear 15
me læg madma mænigo þahim mid scol
don onflodes æht feor ge witan. Nalæs
hihine læssan lacum teodan þeod gestreo
num þonne (þ)a d(y)don þe hine ætfrumsceafte
forð onsendon ænne ofer yðe umbor we ...


Further Information

Although the current archive was assembled for the use of scholars and textual editors, later additions to the electronic archive should incorporate digital material of anything its users deem appropriate, including for example manuscript illuminations and modern illustrations; material culture from the Anglo-Saxon period and contemporary representations; links with the Toronto Dictionary of Old English project and with the comprehensive Anglo-Saxon bibliographies of the Old English Newsletter; medieval glossaries and modern translations; electronic articles and on-line discussion groups; and in fact any related multimedia resources.

For background to the project see the 1993 announcement in my Digital Preservation, Restoration, and Dissemination of Medieval Manuscripts, or the recent cover story in "The Electronic Beowulf," Computers in Libraries (February 1995), 14-15. Some scaled image files have been mounted on two sites, one at the University of Kentucky and another in The British Library. These images are for personal use only, and may not be reproduced in any form without the permission of the British Library. For additional information you may contact kiernan@beowulf.engl.uky.edu.



Copyright © 1995, The British Library Board