"The ELECTRONIC Beowulf"

by Kevin S. Kiernan

The "Electronic Beowulf" project is part of the British Library's "Initiatives for Access" program to make its collections more available to the public through new technologies. Beowulf, one of the first sustained imaginative works to emerge from a modern language in the early Middle Ages, holds a special place in the history of Western culture. School children keep it moving from the Middle Ages to the present by reading exciting versions in modern translation that highlight Beowulf's victorious wrestling matches with male and female monsters in his youth and his courageous death in a head-to-head dragon fight as an old man. The text is compelling to grown-ups, too, for its capacity to sustain the detailed intellectual interests of linguists, literary theorists, textual critics, editors, historians, translators, poets, metrists, folklorists, archaeologists, paleographers and codicologists. More has been written about Beowulf, in fact, than any other literary text in the English language. The electronic archive will assemble an encyclopedic resource, founded on the only extant manuscript, of Beowulf materials for all these diverse interests.

The unique Beowulf manuscript preserves special problems in addition to the ones caused by the monsters in the poem. In particular, the manuscript was severely damaged along its outer edges in a fire in 1731, and about 2,000 letters subsequently crumbled away along these edges. Toward the end of the eighteenth century, after the fire but before the brittle edges of the manuscript had fallen away, two complete transcripts of Beowulf were made, one by G.J. Thorkelin, the first modern editor and translator of the poem, and one by his hired scribe. In the middle of the nineteenth century, many hundreds of additional letters and parts of letters that had survived were covered along the burnt edges by paste, tape, and paper frames used to protect each folio leaf from further damage in a restoration of the binding. Today the transcripts, now in the Royal Library of Denmark, are essential for restoring the lost letters and for helping to confirm the identity of many of the covered ones, which are still visible to varying degrees in the manuscript when illuminated from behind the paper mounts.

This past summer the Royal Library allowed me and David French, a conservator in the Department of Manuscripts of the British Library, to digitize the two eighteenth-century Thorkelin transcripts in Copenhagen. We have also begun to add important nineteenth-century collations of the manuscript with the first edition (1815) of the poem. In the earliest collations nineteenth-century scholars, in particular J.J. Conybeare and Sir Frederic Madden (who later became Keeper of Manuscripts at the British Museum) were able to examine the manuscript before the burnt leaves were mounted in the paper frames. To facilitate its inclusion in the electronic archive, Whitney F. Bolton recently donated Conybeare's first collation (1817) to the British Library. Meanwhile the Houghton Library at Harvard University allowed me to begin digitizing Madden's second collation (1824) and other pertinent material. We had previously digitized the entire manuscript itself, including hundreds of unrecorded letters and parts of letters along the covered edges, which were for the first time made visible by fiber-optic backlighting. All of these images are now part of the Electronic Beowulf, which so far consumes many gigabytes of disk space at both the University of London and the University of Kentucky.

The technological complexities of this huge, distributed, electronic archive accumulating around the manuscript have fortunately begun to engage the theoretical and practical research interests of computer scientists, library and information scientists, mathematicians, and statisticians at the University of Kentucky. Because of its age, its broad cultural appeal, and the various damage it has sustained by erasure, fire, and restoration schemes, the Beowulf manuscript presents a number of interesting problems whose solutions will apply to many other fields. To make the electronic facsimile available to others in a useful and usable form, a team at Kentucky that calls itself GRENDL (Group for Research in Electronically Networked Digital Libraries) is developing software that hypertextually links all related images in the archive. A reader studying a given folio of Beowulf will now be able to superimpose enhanced, back-lit, or ultraviolet images over covered or erased readings; bring in for comparison the corresponding pages from a transcript, a collation, an edition, a translation, or other popular portrayals; access an Old English glossary or a related archaeological find; or initiate a bibliographical search and retrieval of a related article. The program is obviously applicable to any text and the British Library plans to use it for other important documents in its incomparable collections.

The basic archive is made up of huge 21MB image files produced by a digital camera that scans at 2000x3000 pixels in 24-bit color. With the proper equipment, including an ethernet connection, a powerful workstation with 128MB of RAM and relatively unlimited storage facilities, and a 24-bit color monitor, I as editor can retrieve the gigantic images from England or Kentucky, display them with excellent fidelity, image-process them to improve their clarity, and link them with one another and with other related databases. The challenge is to make the archive truly accessible to those many users who will not have such specialized equipment. We must devise fast, efficient ways to move high-resolution image files across the network and then display them on the customary 8-bit monitors of PCs and Macs without appreciable loss to the quality of the image. Our Unix-based program, already well underway, will next be compiled for PCs and Macs, as well, using scaled images that relatively normal personal computers can handle. The British Library is running a beta-version of the archive in the Students' Room of the Department of Manuscripts.

In the next stage of the project we will continue to develop the software at Kentucky and to recover by image processing as much of the covered readings as possible, and then link these enhanced images to their appropriate places in the manuscript. The software already enables me to hypertext or interactively link these images with one another. One of its editorial, compositional, features is that it automatically writes a permanent data file with the coordinates of each image as I place it where it belongs on the relevant folio. Thereafter any subsequent user can choose to view the folio as it appears in the manuscript or with boxed areas showing where enhanced images are available. With the click of a mouse in any of these boxes the user can instantly restore an enhanced image to its proper place on the manuscript page.

There are currently ftp sites at the British Library (othello.bl.uk) and the University of Kentucky (beowulf.engl.uky.edu), where anyone can retrieve over the Internet a selection of scaled images from the Beowulf manuscript. There is also a World Wide Web presentation of the project at URL http://www.uky.edu/~kiernan/welcome.html. Thanks to the initiative of my academic co-director, Paul E. Szarmach, Director of the Medieval Institute at Western Michigan University, the Electronic Beowulf is now a project of the new Richard Rawlinson Center for Anglo-Saxon Studies and Research. We expect that the project will eventually make available on multiple platforms and on the Internet a comprehensive interactive electronic archive, an on-line exhibition, and a diversified series of CD-ROM-type packages of Beowulf and ancillary materials for scholars, teachers, and students at all levels.

Acknowledgments: The project in this phase has received substantial support at the University of Kentucky from the College of Arts and Sciences, Research and Graduate Studies, and the Center for Computational Sciences. Particular thanks are due to Brent Seales, Dave Hart, Ashley Coutinho, Ken Kubota, Todd Rutland, and Praveen Devulapalli.