This is an online version of an article published in Literary and Linguistic Computing 12 (1997), 185-95.

The Electronic Beowulf and Digital Restoration

Andrew Prescott

A modern edition, made by a person really conversant with the language he illustrates, will in all probability be much more like the original than the MS. copy, which, even in the earliest times, was made by an ignorant or indolent transcriber.

J.M. Kemble, Beowulf (1835 ed.), p. xxiv.             

I.

On 26 October 1815, Richard Price, an English bookseller and antiquary, wrote from Munich to the Danish scholar, Grímur Jónsson Thorkelin. He began as follows: 'Your letter of 26th July did not reach me until my arrival here in the beginning of October and its content surprised me not a little as I had already in May last intrusted the promised plate to a friend who was about to visit Copenhagen. But if my astonishment was the great I leave you to conceive what it this day has been upon receiving back my packet accompanied by a letter informing me that my friend has for these last four months lain dangerously ill at Gottingen: This has thrown me into a dilemma from which I hardly know how to extricate myself and my regret is so much the more increased as the poem is now published. I have however determined upon writing you for instructions and shall detain the plate till I hear from you' (Edinburgh University Library, La. III. 379, no. 677*).

The book to which Price refers was Thorkelin's edition of Beowulf, the first complete edition of the poem, which had been published earlier that year in Copenhagen. Thorkelin had already suffered many vicissitudes in producing this edition. He had first seen the only medieval manuscript of the poem at the British Museum in the 1780s (Kiernan 1986). He had made substantial progress on his edition when his library was wrecked and many of his papers destroyed during the British siege of Copenhagen in 1807. He was ready to give up work on the poem but, according to Conybeare, 'the encouragement...of some powerful friends, induced the literary veteran to recommence the task of preparing the work for the press' (Conybeare 1826, p. 32). Price was presumably one of those friends who urged Thorkelin to complete his task. The lost plate was doubtless an illustration of the Beowulf manuscript. It would have made a handsome addition to the volume, and its failure to reach Copenhagen must have been a bitter blow to Thorkelin. It is not known what Price eventually did with the plate, which is a pity, since it would have provided valuable evidence of the state of the manuscript at this time. Thorkelin does not seem to have remembered Price kindly. Writing four years later to Francis Douce, formerly Keeper of Manuscripts at the British Museum, he recalled how 'A certain Mr Price came here, you know him perhaps, and that is enough, for I never learned to know him, that gentleman preyed the groves of our Danish muses, and carried away spoils, equal to the Golden Fleece. He boasted to have done the same in England ' (Bodleian Library, MS. Douce d. 23, ff. 123-4).

The story of the lost plate for the Thorkelin edition of Beowulf might be seen as prefiguring some of our experiences with the Electronic Beowulf. In October 1993, the editor of the electronic facsimile, Professor Kevin S. Kiernan of the University of Kentucky, was returning to the United States after a successful trial of the digital camera at the British Library. He was carrying on a SyQuest cartridge some images of the manuscript he was intending to use in a presentation announcing the project at a conference in Washington D.C. organised by the Association of Research Libraries. Professor Kiernan was concerned that the data might be affected by the x-ray equipment at the airport, but he was assured it was safe. On returning to Kentucky, he was disconcerted to find that all the images had been erased from the cartridge. Frantic attempts were made to transfer copies of the images to the United States in time for the presentation. At that time Internet availability in the British Library was limited and, in any case, trying to transfer large image files across the small capacity lines then available was a very time-consuming process. My PC had to be dismantled to turn it into an FTP site and technicians had to get up in the middle of the night to supervise the file transfer. Eventually all but one of the images were successfully transferred from London to Kentucky and the presentation went ahead successfully (Kiernan 1994).

This story perhaps illustrates some of the advantages of digital technology in that, unlike Price and Thorkelin, we were able to dispatch lost images quickly over large distances. However, the most striking feature of these two stories is the way in which, two hundred years after Thorkelin, scholars feel that existing images do not sufficiently convey the manuscript context of this enigmatic poem, and consider it worth struggling with the latest technologies to try and give a better understanding of the problems posed by the Beowulf manuscript. A distinctive feature of Beowulf studies throughout this time has been the way in which scholars in each generation have used new technical aids to explore the manuscript. In 1882, Beowulf was one of the first medieval manuscripts to be made available in photographic facsimile (Zupitza 1882). Just before the Second World War, A. H. Smith used ultra-violet photography to try and decipher badly faded folios in the manuscript (Smith 1938). In the 1980s, Kevin Kiernan experimented with the use of medical imaging equipment to try and read erased portions of the manuscript and also used fibre-optic light to read letters obscured by conservation work (Kiernan 1984, 1991). The pioneering use of technical approaches of this kind is very unusual in medieval studies. One cannot imagine any other medieval text in which the use of photography or ultra-violet light are accepted as key developments in the critical history of the text, as they are, for example, in the Beowulf Handbook recently produced by Robert Bjork and John Niles (1996, pp. 35-37).

The Electronic Beowulf therefore belongs to a long tradition of technically-aided exploration of the Beowulf manuscript. This technical analysis of the manuscript is not simply due to scholars trying to squeeze more and more information from an already over-studied manuscript. It is due to particular problems posed by the manuscript itself which reflect its very chequered history. In order to understand why these various technical approaches have assumed such an important role in Beowulf studies, it is necessary to look more closely at the history of the manuscript.

II.

The history of the Beowulf manuscript was described in detail for the first time by Kevin Kiernan in his groundbreaking book Beowulf and the Beowulf Manuscript (1981), a reprint of which is being bundled with the Electronic Beowulf CD ROM. The following account of the manuscript is based largely on Kiernan's work. The only known medieval copy of Beowulf is found in Cotton MS. Vitellius A.xv, a volume from the library of Sir Robert Cotton (1571-1631). Cotton's library was given to the nation by his grandson in 1700, transferred to the British Museum on its foundation in 1753 and thence to the British Library on its creation in 1973. Cotton's library was relatively small -- less than a thousand volumes at the time of his death -- but contained many of the most important manuscripts of medieval English history and literature. Among his treasures were the Lindisfarne Gospels, two of the original letters patent of 1215 by which Magna Carta was promulgated, the only known manuscripts of Sir Gawain and Pearl, and several versions of the Anglo-Saxon Chronicle. At the time of Cotton's death, his library was in the process of being reorganised into a pressmark system based on the names of the roman emperors whose busts stood on the top of each press of manuscripts. Thus the pressmark for the Beowulf manuscript, Vitellius A.xv, indicates that it was the fifteenth volume on the first shelf of the press marked by the bust of the Emperor Vitellius.

Vitellius A.xv in fact contains at least two separate and unrelated manuscripts bound together by Cotton. The first contains an ownership inscription recording that it belonged in the fourteenth century to Southwick Priory in Hampshire and is known as the Southwick Codex. The second is known as the Nowell Codex, from an ownership signature of the antiquary Laurence Nowell, dated 1563. The manuscript entered Cotton's ownership before 1612. The Nowell Codex contains, apart from Beowulf, four other unrelated Old English works: the St Christopher fragment, The Wonders of the East, Alexander's Letter to Aristotle, and the Judith fragment. Cotton used a leaf from a mid-fourteenth century illuminated manuscript (now reconstructed as Royal MS. 13.D.I*) as a flyleaf for the volume, something he did for a number of his most important manuscripts (Tite and Carley 1992).

The history of the Beowulf manuscript in Cotton's hands already creates a number of problems. Cotton's action in binding together the Southwick and Nowell volumes was undoubtedly by modern standards cavalier. If he was willing to create an artificial volume by joining two manuscripts together, perhaps he went further. It has been suggested, for example, that the Judith fragment was originally separate and was added to the volume in early modern times. If the Judith fragment was separate, how can we be sure that the Beowulf section was not also originally a separate manuscript? There are some hints that perhaps this may have once been the case. Trying to establish the original physical structure of a manuscript such as this requires attention to small physical details such as patterns of ruling, the arrangement of hair and flesh sides of the vellum, and even the appearance of burn marks or worm holes. In analysing fine details like these, technical aids are very useful -- high magnification, for example, is one obvious requirement.

Cotton's librarian, Richard James, added a list of contents to the beginning of the manuscript. Intriguingly, he seems to have been baffled by Beowulf and left a blank for that article. Consequently, Beowulf is not mentioned in the description of Vitellius A.xv in Thomas Smith's 1696 catalogue of the Cotton library. Vitellius A.xv was used by Franciscus Junius who transcribed the Judith fragment sometime between 1628 and 1650, but he apparently took no interest in Beowulf. The credit for first drawing attention to Beowulf belongs to Humfrey Wanley, who in 1705 described the poem in his catalogue of Old English manuscripts, transcribing a few lines from it, and drew attention to it in his correspondence. Wanley's catalogue was apparently responsible for leading Thorkelin to the manuscript, thereby inaugurating the modern editorial tradition.

While Thorkelin's edition undoubtedly finally put Beowulf on the literary map, it was far from being a scholarly success. In the letter already cited, Richard Price described his attempts to promote Thorkelin's edition in England: 'In conformity with my promise I applied on my arrival in England to those booksellers whom I knew offering the sale of the intire impression of the poem and producing your letter for a description of its contents. I am however sorry to add without success. The principle and never-failing objection was the Latin translation which they declared would not be read or understood by those upon whom they would reckon for the most certain sale namely the English Antiquary.' More seriously, N.F.S. Grundtvig listed in a review of the edition numerous errors by Thorkelin and in a Danish translation of the poem published in 1820 gave forty-five pages of corrections to Thorkelin's edition. In 1826, John Conybeare in his Illustrations of Anglo-Saxon Poetry also drew attention to many errors by Thorkelin, but charitably reminded the reader of the difficult circumstances under which the edition had been produced (1826, pp. 30-32). More harshly, John Mitchell Kemble who produced a new edition of the text in 1833 roundly denounced Thorkelin: 'not five lines of Thorkelin's edition can be found in succession, in which some gross fault either in the transcript or the translation, does not betray the editor's utter ignorance of the Anglo-Saxon language' (1835, p. xxix).

Kemble's edition was dedicated to Jacob Grimm and, applying the insights of the new philological learning, Kemble dealt almost as severely with the medieval scribes as he did with Thorkelin: 'The numerous blunders both in sense and versification, the occurence of archaic forms found in no other Anglo-Saxon work, and the cursory illusions to events which to the Anglo-Saxons must soon have become unintelligible, are convincing proofs that our present text is only a copy, and a careless copy too' (1835, p. xxi). Kemble went on to declare that 'All persons who have had much experience of Anglo-Saxon MSS. know how hopelessly incorrect they are in every way...which can perhaps only be accounted for by the supposition that professional copyists brought to their task (in itself confusing enough) both lack of knowledge and lack of care' (1835, pp. xxiii-xxiv).

It would be an over-simplification to suggest that many modern scholars still share Kemble's contempt for the original scribes. Kemble was firing the opening shot in a still unresolved battle over how far editors are entitled to emend texts. However, the modern scholarly consensus still perhaps reflects Kemble's disdain for the Anglo-Saxon scribe in assuming that the manuscript we have is a more or less accidental survival with no special authority. The Beowulf manuscript dates from the first quarter of the eleventh century. The orthodox view is that the poem itself is fundamentally far older, dating back perhaps to the eighth century. The exact mechanism by which it was transmitted over two hundred years is not agreed -- presumably a mixture of oral transmisson and, in the later stages, written exemplars. If this is the case, one might assume that the existing manuscript would be relatively straightforward in character, but it is not.

The manuscript is the work of two scribes. They carefully checked their work. As Kevin Kiernan put it, 'The extensive proofreading done by both scribes, and the nature of their corrections, are proof that the Beowulf MS. was subjected to intelligent scrutiny. The Beowulf MS. contains scores of erasures and scores of written corrections' (1981, p. 191). Kiernan sees these corrections as suggesting that the poem might be contemporary with the manuscript, but even if this is not accepted, it is clearly essential in establishing the nature of the text to investigate these corrections further. This is another area where technical aids can be of great importance. Many of these erasures can be read with the aid of ultra-violet light and, as Kiernan shows, they raise doubts about many accepted readings.

In other places, more radical problems occur with the manuscript. Thorkelin drew attention to various gaps in the text, which he attributed to the great age of the manuscript and later damage. However, the origin of these problems may lay further back. On the folio numbered 192v on the manuscript (196v according to the current British Library foliation), something very strange has happened. It looks as if the ink in some words in the middle of lines 2886-2891 of the poem has run as a result of water damage. But closer inspection suggests that water was probably not the culprit, as there is no evidence of water damage elsewhere on the folio. Another explanation might be that the letters have been touched up by a more recent hand, but again the way in which the ink has spread outwards makes this unlikely. Above these words is an erasure with an interlineation. Kiernan points out that, under ultra-violet light, the erasure can be read, and consists of the same words as the interlineation (1981, p. 218; cf. Kiernan 1994). This suggests that perhaps the scribe had had problems at this point with the surface of the vellum, and some time after he first wrote the folio, restored parts of the text which had become damaged.

On the folio numbered 179 on the manuscript (182 according to the current British Library foliation), even more extensive textual loss occurs, but not in a consistent pattern. Whole words and phrases have disappeared, either as the result of erasures or later damage. Some of these lost sections have a curious grey discoloration. Tilman Westphalen first suggested that this page was a palimpsest, a proposal which Kiernan endorsed, pointing to other evidence apparently supporting it. This view has not been generally accepted, but it is impossible to deny that something very extraordinary has happened to the manuscript at this point. The ultra-violet photographs in the black and white facsimiles fail to convey adequately the dramatic appearance of this folio. This is partly because the ultra-violet light fluoresces with the grey patches, obscuring rather than revealing the lost text beneath. The frontispiece of Kiernan's 1981 book is a superb colour photograph of this folio (unfortunately very badly reproduced in the reprint), which conveys the appearance of this folio more effectively than the black and white facsimiles. The erased words look tantalisingly legible in the colour photograph, but, although enough can be made out to throw doubt on some of the traditional readings on this page, a consistent reading of the lost text is not possible.

It was problems such as these which prompted an early experiment in photography of the manuscript. On 9 April 1864, the Anglo-Saxon scholar Oswald Cockayne was authorised by the Trustees of the Museum to make a complete transcript of Vitellius A.xv. He evidently ran into difficulty with some of the damaged pages, and on 23 July 1864, the Trustees agreed 'to have the effect of Photography tried on a leaf or two of the injured Cottonian MS. Vitell. A.XV, with a view to the recovery, if possible, of some portion of the now illegible text' (British Library, Department of Manuscripts Archives). A collation by Cockayne of Thorpe's edition of Beowulf with the original manuscript apparently made at this time survives in the Houghton Library at Harvard (Houghton Library, 28286.27.7), and a note by Cockayne on the pastedown of the back cover confirms that 'Fol. 179a. is the one to try by Fotography'. (I owe this reference to Kevin Kiernan). The photographs made for Cockayne unfortunately do not survive. Cockayne's hope that photography would help make out these difficult passages was not as quixotic as it might seem. At that time, manuscripts were read in the gloomy natural light of the Round Reading Room, and it is likely that the strong lights required for photography would have made parts of the text legible which could not be seen in the Reading Room.

In 1981 Kevin Kiernan urged the use of electronic photography and digital image processing to study the 'palimpsest folio' further -- a visionary viewpoint at a time when the personal computer was only just appearing and most humanities research was pencil driven (Kiernan 1981, p. 233). Inspired by the work of John Benton at the Jet Propulsion Laboratory, Kiernan attempted in 1984 to videotape images of the manuscript under special lighting conditions to provide input for a Kontron Mipron-D microcodable array-processor at the University of Kentucky, in the hope of being able to enhance these images digitally and make out the lost words. The conversion of a British videotape to American standards meant that the results were less satisfactory than hoped, but they nevertheless showed that digital image processing could greatly improve the contrast of the image and thus potentially help in reading erasures. Three years later, Kiernan set up a complete image processing station at the British Library to make direct images of the manuscript (producing the then enormous quantity of ten megabytes of data). This improved the legibility of some sections of the page and raised some doubts about accepted readings, but by no means resolved the problems raised by this folio. It was however a significant pointer to the future (Kiernan 1991).

The medieval text itself thus also raises many major issues, which again are susceptible to technically-aided analysis. However, this by no means exhausts the problems raised by the Beowulf manuscript. Between the time that Wanley borrowed the manuscript and Thorkelin first saw it, it suffered another great trauma. The gift of the Cotton library to the nation in 1700 had embarassed the government, which was unwilling to meet the costs of providing adequate accommodation or other administrative expenses. It was left in Sir Robert Cotton's decaying house at Westminster. In 1717, John Elphinstone, the Keeper of the Library, petitioned the Treasury, pointing out that, at the time of the gift, 'King William of glorious memory did promise that a sufficient allowance should be paid to those imployed to attend the said library; which was not done in his said Maiesties Reign'. Not only had Elphinstone not been paid for attending the Library himself, but he had been forced to pay for cleaning and so on out of his own pocket. At the time of George I's coronation, 'Cotton House and Gardens were thought a convenient place, for dressing the dinner, and preparing other necessaries', so that Elphinstone, 'fearing any accident that might happen by the many fires made there, was at the charge of setting up a bed and attending day and night in the said Library...at extraordinary trouble and charge.' (British Library, Additional MS. 61615, f. 81).

When Cotton House finally fell into a completely ruinous state, the manuscripts were moved to Essex House near the Strand. Essex House was, however, thought to be a fire risk, so the manuscripts were transferred again to Ashburnham House in Little Dean's Yard at Westminster. On 23 October 1731, Dr Bentley, the former Keeper of the Royal Library, staying at Ashburnham House with his son, the then Keeper, was woken up by the smell of smoke. While the librarians busied themselves trying to rescue the manuscripts, the fire spread to the presses, so that it was necessary to break open the cases and throw volumes out of the window.

The morning after the fire, Little Dean's Yard must have been a sad sight, littered with fragments of burnt manuscripts, which the boys of Westminster School picked up and kept as souvenirs. Vitellius A.xv was comparatively lucky. It was badly singed around the edges, and left smoke stained and brittle. Water from the fire engines had stained many of the folios, causing the ink to run in some places. The heat had caused some edges of the manuscript to stick together, and it seems from the introduction to his edition that when Thorkelin looked at it fifty years later he had to force some of the leaves apart (Bjork 1996, pp. 312-3). The conservation technology of the eighteenth century was not equal to the task of stabilising the condition of a heat-damaged vellum manuscript such as this one. It is possible that some rudimentary conservation work was undertaken on the manuscript just after the fire, to dry it out and prevent the growth of mould, but basically it was left unconserved, and was transferred in this state to the British Museum in 1757. Following the move to the Museum, the manuscript was made freely available to readers. It was consulted by Thomas Astle, John Topham and others in the 1770s, who were perhaps responsible for helping to draw the manuscript to Thorkelin's attention (Prescott 1995).

The Royal Library in Copenhagen contains two transcripts of Beowulf used by Thorkelin. The first, known as Thorkelin A, was made by an unknown copyist who did not know Old English but made a spirited attempt to represent the appearance of the original manuscript. The second, Thorkelin B, is in Thorkelin's own hand and is apparently a working transcript used by him in preparing his edition. These transcripts are precious documents since they record hundreds of letters which afterwards crumbled away from the fire-damaged manuscript (Kiernan 1986). The textual loss may have been made worse by a misguided attempt to bind up the manuscript undertaken by the Museum binder Elliot under the direction of Joseph Planta at the time he was preparing his catalogue of the Cotton collection between 1793 and 1796. Since Elliot's binding apparently did not protect the fragile edges of the manuscript, it did little to stop the gradual crumbling away of the manuscript (Prescott 1997). It was used by Sharon Turner sometime before 1805, by Conybeare in 1817, by Frederic Madden in 1824, by Grundtvig in 1829, by Benjamin Thorpe in 1830, by Kemble sometime before 1833, and doubtless by many others besides (Bjork and Niles 1996, pp. 35-6). Every time one of these readers used the manuscript, more pieces of text fell off the edges and ended their life on the floor of the British Museum.

In 1837, Madden became Keeper of Manuscripts at the British Museum, and, prompted by the discovery of a great mass of unrestored fragments of Cotton manuscripts lying forgotten in a garret room, determined to have all the manuscripts fully restored (Prescott 1997). He used a binder named Henry Gough to inlay each leaf of the damaged vellum manuscripts (Kiernan 1981, Prescott 1997). Gough's technique as applied to another Cotton manuscript was vividly described by Oswald Cockayne in 1864: 'The binder first soaked the ruins [of the manuscript] in water, to make them limp; he then flattened them, and for this purpose was obliged often to cut through the edges, and stretch them by pins, widening all the flaws; stout pieces of cardboard were then prepared as a frame to carry the leaves, which were fixed into these paper frames by ligaments of goldbeaters skin' (Cockayne 1864, pp. lxxv-lxxvi).

The precise technique used by Gough in preparing the paper frames which protect the edge of the manuscript has been elucidated by Kevin Kiernan (1981, pp. 68-70). The material used by Gough was more akin to cartridge paper than cardboard. He traced around the edge of each damaged leaf on a sheet of this heavy paper. He then cut a hole of a few centimetres inside the traced outline to leave a retaining edge for the vellum. The leaf was then glued to the paper and secured with tissue and skin where necessary. This work was undertaken under the close supervision of Madden, who probably devised the techniques used by Gough. On 7 August 1845, Madden noted in his diary that 'Gough is getting on nicely with the restoration of the injured vellum manuscripts and brought me up today the Bede of the 8th century, Tib. A. XIV, and the Beowulf and other Saxon treatises, Vitell A. XV, both inlaid and perfectly repaired and preserved' (Prescott 1997). Madden then sorted the inlaid sheets of Vitellius A.xv into an appropriate order and sent them off to be bound by the Museum binder, Tuckett. Madden's work in sorting the leaves can be traced from various pencil numerations made by him at the bottom right hand corner and (partly cropped away by Tuckett) the extreme top right hand corner of each leaf (the so-called fifth and fourth foliations) (Prescott 1997).

Gough's work did indeed, in Madden's words, 'perfectly repair' Vitellius A. xv. It has not suffered further textual loss since 1845. Gough's restoration is, however, now beginning to show signs of age. In certain places the vellum is beginning to come adrift from the paper frame, and Tuckett's binding, the back of which has been broken by years of exhibition display, will require replacement before too long. Since Gough's work was not (to use the modern conservation jargon) 'reversible', any future restoration may well destroy some of the evidence carefully preserved by Gough -- another reason for making a full record of the condition of the manuscript at the present time. Nevertheless, Gough's work was a marvel for its time. The skill and utility of the restoration have been well summarised by Kiernan: 'The new binding is a remarkable piece of craftsmanship, as well as of preservation...In addition to protecting the edges of the vellum from further deterioration, the heavy paper frames also obviate the need to handle the vellum at all while reading the MS.' (1981, pp. 68-9).

The preservation of the manuscript, however, came at a price. In order to have a retaining edge for the paper frame, Gough had to conceal letters around the edge of the verso of each leaf. Zupitza lamented the loss of these letters in 1882, and attempted to make them out by holding them up to the light (Zupitza 1882, p. vi). But, as Kiernan has emphasised, at least there is something left to decipher -- if Gough had not covered up these letters, they would have joined the others on the floor of the Reading Room. In 1984, this conservation strategy was triumphantly vindicated by Kiernan, who pointed out that by holding a powerful fibre-optic light (a cold light source which cannot damage the manuscript) behind the concealed letters, many of the letters which Zupitza could not make out could be read (Kiernan 1984). He was able to transcribe many of them, but felt frustrated that it was impossible to obtain a photographic record of them. In order to make out the concealed letters, the fibre-optic light often had to be held at a very oblique angle. Kiernan guessed that by the time a photograph had been taken and the film processed, it would be impossible to tell if it was an accurate record of what the fibre-optic light had revealed. Subsequent tests showed that this was indeed the case. The problem of obtaining images of these concealed letters has been a central concern of the Electronic Beowulf project.

At every stage of its history, the Beowulf manuscript presents problems to which technical aids apparently offer help, if not solutions. The Electronic Beowulf represents the most concerted attempt to date to try and bring technical assistance to bear on the investigation of these issues.

III.

The Electronic Beowulf project began in 1993. During a conference at the University of Wisconsin, Madison, Professor Paul E. Szarmach, formerly of the State University of New York, Binghamton, and now Director of the Medieval Institute, Western Michigan University, suggested to me that the time was right for such a project. Our first action was to contact Kevin Kiernan, who reacted to the idea with enthusiasm -- for some time, he had kept an e-mail folder called e-Beowulf, and was delighted at the possibility that this dream might be realised more quickly than he had imagined. As is evident from the preceding account of the history of the Beowulf manuscript, Kiernan's work had already anticipated many of the themes which are drawn together by the electronic facsimile. His 1981 book had put forward the controversial view that not only was Beowulf contemporary with the Beowulf manuscript but that 'the extant MS. actually was the poet's (or poets') working copy' (Kiernan, 1981, p. 270). Kiernan's arguments have generally not been accepted, but it has meant that scholars have had to pay closer attention to the manuscript, as the only piece of physical evidence for the origins of the poem. Moreover, Kiernan's subsequent work has emphasised how the history of the reception of this text is a very complex one. He produced the first textual analysis of the Thorkelin transcripts and, as has been seen, made the first modern transcription of the letters in the manuscript concealed by Gough's conservation work (Kiernan 1984, Kiernan 1986).

Prescott and Szarmach had in their initial discussions naively imagined the Electronic Beowulf as a series of straightforward digital images of the manuscript which would have been relatively straightforward to complete, taking perhaps no more than the three weeks or so required for the recent CD-ROM of the Exeter Book. Kiernan's more profound knowledge of the history of the Beowulf manuscript lead him to propose a much more challenging project, but one which makes much more intelligent use of the technology. He has sought to create an image edition which draws together and juxtaposes all the primary evidence for the transmission of Beowulf, and exposes the different layers of evidence on which the received text depends. The result could perhaps be described as a diplomatic edition done with pictures rather than words, but even this does not convey the radical nature of the edition. Conventional editions are interpretations of a manuscript text by an editor, designed to convert manuscript sources into print media. Thus, for Beowulf, the standard editions include many corrections and emendments, some of an extremely subjective character, which editors feel are necessary to produce a readable text. In order to help establish the origins of the text, Kiernan had in 1981 called for the production of a 'new, truly conservative edition', purged of such corrections (Kiernan, 1981, p. 278). He will achieve this in the Electronic Beowulf in a way which would perhaps have surprised him at the time he wrote his book.

The work in preparing this image edition has been more complex than that normally associated with a conventional facsimile or text edition. Not only have there been the inevitable difficulties associated with the use of any new technology, such as managing the transfer across large distances of huge quantities of data or moving 24-bit colour images across different technical platforms, but it has been necessary to devise innovative technical solutions to achieve the desired end result. At all stages, Kiernan has taken the lead in determining the technical approach of the project -- for example, he identified the camera used in the project and proposed the solution to the difficult problem of providing a suitable front end. As editor, he has had not only to identify and sort the hundreds of different images used in the project, but also create the HTML which links them together. Since this is a new type of edition, it has required a new type of editor. Moreover, although the final result reflects one man's vision of the manuscript, the creation of it has required a new type of collaboration between scholar, curator, conservator, photographer and technical expert.

The project was financed as part of the British Library's Initiatives for Access programme. This provided finance for image capture equipment at the British Library and also paid for staff time on the project, in particular providing the services of a professional project manager, John Bennett of Strategic Information Management, to supervise the development of the technical infrastructure at the British end. In America, the University of Kentucky provided a massive HP Unix Workstation for editorial work, gave access to its Convex mass storage system for archiving data, and provided technical assistance. A summer stipend from the National Endowment for the Humanities helped finance Professor Kiernan's travel costs, and The Andrew W. Mellon Foundation also gave a grant to Professor Kiernan to allow him to spend a year working on the edition -- as with all such projects, the most precious resource has been time.

In order to facilitate image processing work and to allow the use of special lighting effects, it was felt that any imaging process which required the scanning of photographic media would be unsuitable for this project. On conservation grounds, it was not acceptable that the manuscript should come in direct contact with the scanning device. Any digital camera used in the project would need to offer good quality colour combined with high resolution and ease of use. At Professor Kiernan's suggestion, the ProgRes 3012 camera, manufactured by a German medical imaging company, Kontron, was used. The camera perfectly met our criteria by offering images at 3096 x 2320 pixels (an upgrade offering 4490 x 3480 pixels is now available) with full 24-bit colour. It had an on-screen focussing facility, which few systems at that time offered. A separate feed to a black and white monitor was also available, which proved invaluable in setting up shots with fibre-optic light.

The technical features of this camera are described in detail by Peter Robinson in his book on The Digitization of Primary Textual Sources (1993, pp. 45-7), so suffice it to say here that the camera has proved the most reliable and versatile of the pieces of equipment used in the project. The main difficulty has been lighting. David Cooper has already described the problems involved with trying to use cold light with the Kontron camera. The approach in the British Library has been to use photo-flood and latterly halogen lights, on the grounds that the lux exposure was broadly comparable to that used for video filming and acceptable for a one-off project unlikely to be repeated in the near future. However, in order to ensure that the heat of the lights does not cause any harm to the manuscript, all photography has taken place under the supervision of both a conservator and curator. The temperature during shooting has been carefully monitored, and both manuscript and equipment frequently rested. As a result of these necessary restrictions, the digitization process has been relatively slow and expensive. We have had to focus very much on assembling the materials for the Beowulf project, and have not been able to draw together as large a collection of digital images as has been possible for the Oxford project.

The Kontron camera was used at the British Library on a PC platform, initially Windows 3.1, then Windows 3.11. This was a purely pragmatic decision. Support for Unix and Mac in the British Library was at that time very limited, and it seemed that it would be difficult to get the project off the ground if one of these platforms was used. This has created a number of problems. The DOS/Windows architecture was ill-suited to handling the large 21 MB image files produced by the Kontron camera, and the large 486 and latterly Pentium machines used to take and manipulate the images proved very unstable and unreliable. Another problem was colour. The PCs produced beautiful images, but did not display the full 24 bits. It seemed to miss the point to say that, to see the image at its best, one needed to take a plane to Kentucky and view it on Kevin Kiernan's Unix workstation. Although this was frustrating, it was not at first a serious problem. However, as the software available to take images on the PC improved and 32-bit emulators for the PC became available, the colour of the image became distorted when it was transferred to other platforms. This has been a considerable obstacle for the project in its latter stages. By contrast, the Unix and Mac machines used in Kentucky have been much more reliable in their handling of the image files. If I was to make one change in the way the project has been organised, it would have been to try operating the camera in London on Unix or Mac machines.

The question of the storage of the large quantities of data produced by the camera when working at full capacity has also been a major issue. The scan time of the camera working on a Pentium machine under Windows 3.1 is 15 seconds (this is considerably reduced when Windows NT or Windows 95 is used). This means that the speed of scanning can potentially rival microfilm, but something has to be done with the data -- over two gigabytes can be produced in a normal working day. The Library does not possess its own mass storage system for this quantity of data. Initially, it was felt that the best solution was off-line backup onto DAT. This offers large amounts of storage very cheaply -- in 1993, two gigabytes for less than twenty pounds. However, the backup procedure was extremely time-consuming, and the tapes needed regular checking to ensure that the data could still be read. Moreover, the files had to be transferred to Kentucky. The proprietary DOS software used in the British Library for backup could not be run on the Unix machines in America. Although a tar utility eventually made the transfer easier, the cheapness of the DAT proved to be a false economy. The length of time spent in maintaining and recovering data from the DAT wiped out savings achieved through the cheapness of the tape.

It was observed to me in a seminar in 1994 that these problems were not so much ones of storage as of bandwidth, and indeed they eased as the capacity of networks improved. When the high speed SuperJanet link became available within Britain, images could be transferred to to the large scale storage facility at the University of London Computing Centre. This has proved a more effective method of archiving shots as they are taken, but the time taken in transferring images is still much greater than that required to shoot them. >From ULCC, it has proved possible to transfer images by FTP to Kentucky, and this is currently the normal means of transfer. Network links are still, however, not as reliable or as speedy as one might like, and it is possible that the use of a CD writer or external drive might provide a better solution of these issues.

This gives some idea of the technical infrastructure. But what does the Electronic Beowulf consist of? The first level comprises colour images under normal lighting conditions of the whole of Vitellius A.xv, providing the first facsimile of the complete volume and the first colour facsimile of the Nowell Codex. The resolution of the images varies depending on the exact distance of the camera from the surface of the manuscript, but generally they are in excess of 300 dpi. The colour images are already a considerable advance on the existing black and white facsimiles. Words and letters running into repairs on the edge of the folio can be read in the colour facsimile which cannot be made out in black and white. Kevin Kiernan has given a striking example of the value of colour images in the three lines at the top of folio 180 (183 according to the British Library foliation), which look in black and white as if they are simply indistinct as a result of the general damage to the manuscript. In colour, however, it seems as if these lines have been deliberately erased by the scribe (a point only previously noticed by Thorkelin) (Kiernan 1994; cf. Kiernan 1981, pp. 245-50).

However, if the aim had merely been to produce a colour facsimile, it would have been easier to do it by conventional means. The colour digital images are merely the starting point. We hope that they will provide the raw material for digital image processing of damaged or obscure text, but have not had the chance to make much progress with this ourselves yet. The images can be magnified to examine physical details of the manuscript or details of scribal technique. Different parts of the manuscript can be compared side by side. Users can rearrange the manuscript into different collations to test rival theories about its structure.

The manuscript alone, however, does not tell the story of the Beowulf text. In a digital environment, it is possible to place side by side documents which are widely dispersed. This sounds fine in theory, and an obvious application of this technology is the reintegration of membra disiecta. One of our primary aims was to provide images of the Thorkelin transcripts, which had not been seen side by side with the Beowulf manuscript since they were sent by special railway delivery to the British Museum for Zupitza to use them in 1880 (British Library, Department of Manuscripts archives). The practical issues involved in realising this aspiration were, however, considerable, since it required the transportation of the camera, PC, monitors, lights and other equipments to Copenhagen. This was a considerable logistical exercise, similar to transporting a film crew. The excess baggage bill alone was enormous. The Royal Library at Copenhagen were very accommodating hosts and staff there responded enthusiastically to the opportunity to see this technology at work at first hand, even when the two 2KW photo-flood lights blew out the electrical circuits in the area where scanning was taking place.

We had initially assumed that, for modern transcripts, scans at a lower resolution than that used for the Beowulf manuscript would be acceptable, but as scanning progressed it became evident that full resolution shots were also required of the Thorkelin transcripts. Where these transcripts record letters that have since disappeared, minor details such as erasures, corrections and alterations can be just as important for determining particular readings as in medieval manuscripts. Working under the supervision of Professor Kiernan, David French, the Library conservator who has provided much of the technical support for the project, completed the scanning not only of both sets of transcripts but also of important ancillary material in the Royal Library and the Danish national archives.

As has been seen, Thorkelin's edition prompted a much closer examination of the Beowulf manuscript. This was frequently recorded in the form of corrections to Thorkelin's printed edition. The first such collation was made by John Conybeare in 1817. He makes many corrections to Thorkelin's text, which again are of great importance where letters have disappeared from the original manuscript. He also noted where text had been lost from the manuscript since Thorkelin had used it, indicating this by underscoring the relevant section of the edition with small dots. His collation thus is also an important witness to the deterioration of the manuscript at the end of the eighteenth and beginning of the nineteenth centuries. Conybeare's collation was owned by Professor Whitney Bolton of Rutgers University, who described it in an article for English Studies in 1974 (Bolton 1974). In order to facilitate the scanning of Conybeare's collation for the Electronic Beowulf archive, Professor Bolton very generously presented the volume to the British Library, where it is now Additional Manuscript 71716. This was an exceptionally generous gesture and an unexpected and very concrete bonus for the Library from the project. Conybeare's collation was also scanned at full resolution for the burgeoning electronic archive.

A more accurate collation of Beowulf was also made in 1824 by the future Keeper of Manuscripts at the British Museum and restorer of the manuscript, Frederic Madden. This is prefaced by an eerily realistic drawing of the first page of the poem, the only extant representation of the manuscript before its restoration in 1845. Madden's collation was lot 293 in the sale of his library at Sotheby's on 7 August 1873 and is now in the Houghton Library at Harvard University (Houghton Library, 28286.24.3; Kiernan, 1997). Scanning of this collation required another off-site shoot, the costs of which were met by the University of Kansas and Harvard University, both of which were anxious to see demonstrations of the camera. The organisation of this shoot was simpler than with Copenhagen -- the components of the PC used for image capture were transferred to a small tower which could be managed by one person, while lights, monitors and other standard equipment were hired in the United States. However, rough handling of the PC during the flight meant that it had to be virtually rebuilt in the United States with the assistance of technicians from the University of Kansas. Despite this crisis, however, the shoot was completed successfully, although it seems at the time of writing that the use of the latest Adobe Photoshop plug-in for the camera on a Windows 95 platform may have created some problems with the colour on these images when viewed on Unix or Mac machines.

By providing straight shots of Vitellius A.xv, the Thorkelin transcripts and the Conybeare and Madden collations, the Electronic Beowulf will provide the user with all the extant manuscript evidence for the history of Beowulf. A user will not only have at his or her fingertips all the key evidence for the history of this text but will also be able to juxtapose them in ways which would be impossible even if all the volumes were assembled in one room. This already at one level represents a kind of restoration of the damaged manuscript, in that it enables the user to reconstruct text lost as a result of the fire damage and potentially get some idea of the structure of the manuscript before the fire. By facilitating comparison between the Thorkelin transcripts and the original manuscript, the user can see at a glance how far the original condition of the manuscript has changed. For example, the first folio of the poem in the original manuscript is badly rubbed and worn in the bottom right hand corner. This looks at first sight as if it is the result of over-use of the manuscript in recent years, but comparison with the Thorkelin transcripts shows that its condition was equally bad in the 1780s.

However, the digital restoration of the Electronic Beowulf goes further than this, and it is really in this kind of work that the digital technology permits new approaches to the manuscript which would not be possible in any other medium. It has been known for many years that ultra-violet light allows faded and damaged parts of the text to be more easily read. The 1959 version of Zupitza's photographic facsimile of the manuscript incorporated a number of ultra-violet photographs. Unfortunately, it was not apparently practicable for the facsimile to offer both ultra-violet photographs and pictures made under normal light. The choice of which pages were reproduced under ultra-violet light was apparently made by the Museum's photographer, and not the editors of the volume. In an electronic facsimile, one need not be restricted to one or the other -- both types of images can be readily made available and the choice left to the user.

Moreover, the digital camera allows ultra-violet images of the manuscript to be made more easily. Conventional ultra-violet photography is very time-consuming and expensive, because of the long exposure times required, characteristically fifteen to twenty minutes for each shot. This is potentially hazardous for both the photographer and the manuscript. It was found that very good ultra-violet shots could be made with the Kontron camera in the normal fifteen-second scan time. The essential prerequisites were powerful ultra-violet lights and the careful recalibration of the camera under ultra-violet light. Because the Kontron only takes colour shots, the result can often be at first only a murky blue patch, but adjustment of contrast and brightness and transfer of the image to grey-scale will cause a very good ultra-violet image to appear. This opens up the prospect of large-scale ultra-violet photography of the burnt Cotton manuscripts. Many of the most badly damaged manuscripts, which under normal light look like burnt toast, can in fact be easily read under a light source giving a combination of ultra-violet and blue light, but the cost of conventional ultra-violet light has meant that it is impracticable to undertake substantial ultra-violet photography of the collection. With a digital camera, it becomes a more practicable possibility.

The Kontron camera similarly provides good images under infra-red light, but this requires the removal of an infra-red filter within the camera. Although infra-red light is of little value in making out damaged sections of the Beowulf manuscript, this technique may prove to be of value with other categories of material.

The heart of the project however has been our 'backlit' shots - images of the letters concealed by Gough's paper frames. As has been noted, although these letters can be made out with the aid of a fibre-optic cable held behind the paper frame, it is difficult to make them out even with the naked eye and impossible to make a picture of them with a conventional camera. The advantage of the digital camera is that it is possible to see the digital image in a matter of seconds and establish whether or not the elusive letter has been recorded. It took some time to work out the best method of obtaining these shots. An experiment was made with a flat A4-size fibre-optic pad to see if this would provide a kind of x-ray image of the hidden letters, but the light source was not sufficiently powerful to provide a clear image. Attempts were also made to illuminate whole sections of concealed text, but this also proved unsatisfactory. In the end, it was found that the best method was to clamp the fibre-optic cable into position just behind the folio, illuminating two or three letters at a time. Checking the readings and setting up the shots was extremely demanding, and could not have been accomplished if it had not taken place under the immediate supervision of Professor Kiernan. Many hundreds of such concealed letters have now been shot, covering not only the Beowulf poem but also the whole of the Nowell Codex. Already, images of fibre-optic letters have been used by Professor Kiernan to help correct entries for the Dictionary of Old English.

It would not have been possible to record these hidden letters without a digital camera, but having them in digital form of course then makes it possible to do other things with them. At a simple level, the hidden letters can be pasted back into their appropriate place in a digital image of the whole folio, thus recapturing something of the appearance of the manuscript prior to Gough's restoration. In this way, the process of creating the digital facsimile becomes an extension of the process begun by Gough of conserving and restoring the fire-damaged manuscript, but one performed without direct intervention on the manuscript itself.

This does not exhaust all the levels of the Electronic Beowulf. Although at the heart of the project is the idea that editions should now be done with images not text, and the perception that text is (for the medieval period at least) image, it is recognised that some users may prefer to work with a transcription of the poem. This will therefore be provided, with full SGML tagging, as well as a translation. Links will be provided from the text to show supporting evidence for particular readings. There will also be a glossary.

Knitting together the many different layers of the Electronic Beowulf proved to be no easy problem. The idea was to give the user hyperlinks between each different level of the images. The user would be able to click on the appropriate part of an image of a folio from the original manuscript and be able to see hidden letters and ultra-violet readings in their appropriate place. With a further click, he or she would be able to call up images of the relevant sections of the Thorkelin transcripts or Conybeare or Madden collations. We were anxious that the package would be available for Unix, Mac and PC platforms. A Unix programme allowing users to use the images in this way was developed under the supervision of Professor Kiernan at Kentucky in 1993. A version of this programme was also developed for Mac users, called MacBeowulf, which worked very well, but the creation of a PC version proved problematic, because it was difficult to create a programme which would perform reliably on all the different flavours of PC available. Moreover, it was not clear how this software would be supported. Worst of all, it seemed likely that the editorial work would have to be done three times over, once for each version of the CD.

While the issues associated with this development work were being investigated, the network browsers had made great strides forward. Netscape version 2.0 for the first time offered the ability to display frames containing different images side by side, precisely how we wanted to display the Beowulf materials. With the development of the Java programming language, it also became possible to develop tools which would help the user of the images, such as a tool for zooming in on different parts of the manuscript. The browsers in fact offered all the functionality that was required for the Electronic Beowulf. The only problem was that the networks did not have the capacity to handle very smoothly the large image files, even when in a compressed JPEG format. In fact, even though the image files will be distributed as JPEGs, it will be difficult to fit them all on two CDs. However, the network browsers not only read networked files but also files held on a local disc. So the final CD-ROM will use Netscape 3.00 as a front end to read HTML files and images held on a local CD -- the kind of hybrid approach which is likely to be increasingly common until network capacities are substantially increased. This offers a number of advantages. The materials on the CD-ROM will essentially be independent of the software -- the prototype package works under both Internet Explorer and Netscape. It will also be very simple to make the package available on the Internet when network speeds improve. The CD-ROM is currently scheduled for publication in the summer of 1997.

IV.

Most of the arguments in favour of the digitization of manuscript materials have been framed in terms of increasing access. Digitization is seen by some commentators as providing a kind of superior microfilm which will be available to anyone who has a modem. The practical difficulties raised by these kind of assumptions are beginning to emerge as we come to terms with the technology. The premise has, of course, always been that digital images will be free or at least very cheap. This is a false impression created by the fact that governments and other bodies are at present willing to fund digital projects to keep their researchers in the technical vanguard. As programmes for digitization start to have to recover their costs, the free ride will come to an end, and it is by no means clear at present that libraries will be able to provide digital images as cheaply as microfilm or indeed photography. Although the speed of capture of digital images can bear comparison with microfilming, the post-processing and storage costs can be much greater. At the moment, the costs of network access are borne by academic institutions, but this may not continue to be the case in the future. Already, it sounds as if the costs of access to the broad-band successors to the Internet may be prohibitively expensive for some institutions.

Against this background, it is perhaps worth starting to consider what digital technology can do that other media cannot. The Electronic Beowulf provides a powerful demonstration of the way in which a digital imaging project can produce research information of fundamental value which could not be assembled in any other way. It offers more than access, since the user will have more information about the text available to him than if he was sitting in the British Library with the manuscript itself in front of him. Kevin Kiernan has called this approach 'digital preservation', and at one level this is an accurate description, since the digital images will preserve information about the manuscript (such as the hidden letters) which it may be difficult to preserve in future years. However, as has been seen, this approach goes beyond simple preservation of information about the manuscript towards the active recreation of previous states of the manuscript. This is a direct extension of the activities of Gough, Madden and others in restoring the damaged manuscript, and may perhaps be described as 'digital restoration' (Kiernan 1994).

It may seem that the approach described here is only really applicable to Beowulf, but Vitellius A.xv was only one of more than a hundred manuscripts badly damaged in the Cotton fire, all of which were conserved using similar methods to those in Vitellius A.xv. These include a number of manuscripts of great importance for early English literature and history which have not been very carefully studied. A programme to apply the techniques used in the Electronic Beowulf to all these manuscripts would occupy many lifetimes, but would perhaps finally realise the vision of Sir Frederic Madden of the perfect restoration of this great library. One obvious candidate for such treatment would be Sir Robert Cotton's pride and joy, the Cotton Genesis, Otho B.vi, a Byzantine manuscript dating from the sixth century and one of the earliest illustrated Christian manuscripts, which was reduced to a pile of cinders in the fire. The illustrations of this manuscript are now shrunken, warped and cracked, and attempts to reconstruct their original appearance have had to rely on poor quality photographs or artists' impressions. Digitally aided 'restoration' of some of these illustrations is a simple proposition, and an obvious area for further experimentation.

Already, Kevin Kiernan has applied ultra-violet imaging to Otho B.x, an important manuscript of Ćlfric which was very badly burnt in the Cotton fire and was inaccurately pieced together by one of Madden's assistants, N.E.S.A. Hamilton. Similar experiments have also been made by him with Otho A.vi, the only known manuscript of the prosimetrical Alfredian translation of Boethius (Kiernan 1997). He has shown how the use of imaging can help reconstruct an Old English fragment at the University of Kansas. In this case, the fragment was used as a pastedown. When the fragment was removed from the cover, part of the text adhered to the cover. Using digital images, Kiernan was able to reintegrate the offset and the fragment. Another dramatic illustration has been the use of the 'Remove Dust and Scratches' setting in Adobe Photoshop to improve the legibility of a badly burnt fragment in a Cotton manuscript which had had gauze stuck on it in the 1950s.

Of course, much of this work is experimental, and is not always successful. Cotton owned two of the original letters patent of 1215 by which Magna Carta was promulgated. One of these, now Cotton Charter xiii.31a, was the only such charter with King John's Great Seal still attached. However, it was damaged in the fire of 1731 and the great seal reduced to a shapeless blob. The charter was further damaged by heavy-handed conservation work in 1834, and is now largely illegible (Prescott, 1997). We attempted to apply some of the ultra-violet and backlighting techniques used on Beowulf to this manuscript, but it appears that chemicals were applied to the surface of the document, perhaps during the 1834 conservation work, and, despite processing of the images, little could be recovered. However, more sophisticated image processing techniques may yet perhaps allow greater success in recovering this document.

In developing such techniques further, humanities computing will start to make novel and complex demands on computing technology. This indicates the need for closer collaboration with computer scientists and other disciplines. In recognition of this, an inter-disciplinary group was formed at the University of Kentucky called GRENDL (Group for Research into Electronically Networked Digital Libraries). A major symposium was organised at Lexington in November 1995 to discuss the ways in which humanities research can present technically challenging problems for computer science. The theme of the conference was 'Reconnecting Science and the Humanities through Digital Libraries', and this is perhaps no naive hope. The possibilities offered by digital restoration may yet bear larger and more unexpected fruit than have hitherto been imagined.

References

Bjork, R.E., 'Grímur Jónsson Thorkelin's Preface to the First Edition of Beowulf, 1815', Scandanavian Studies 68 (1996), 291-320 .

Bjork, R.E., and Niles, J.D., eds., A Beowulf Handbook (Exeter: University of Exeter Press, 1996).

Bolton, W.F., 'The Conybeare Copy of Thorkelin', English Studies 55 (1974), 97-107.

Cockayne, T. Oswald, Leechdoms, Wortcunning and Starcraft of Early England, 1 Rolls Series 35 (1864) .

Conybeare, J.J., Illustrations of Anglo-Saxon Poetry (London: Harding and Lepard, 1826).

Kemble, J.M., The Anglo-Saxon Poems of Beowulf, The Travellers Song and the Battle of Finnesburgh (London: William Pickering, 1835) .

Kiernan, K.S., 'Beowulf' and the 'Beowulf' Manuscript (New Brunswick, New Jersey: Rutgers University Press, 1981).

Kiernan, K.S., 'The State of the Beowulf Manuscript 1882-1983', Anglo-Saxon England 13 (1984), 23-42.

Kiernan, K.S., The Thorkelin Transcripts of 'Beowulf', Anglistica 35 (Copenhagen: Rosenkilde and Bagger, 1986) .

Kiernan, K.S., 'Digital Image Processing and the Beowulf manuscript', Literary and Linguistic Computing 6.1 (1991), 20-27 .

Kiernan, K.S., 'Digital Preservation, Restoration, and the Dissemination of Medieval Manuscripts' in Okerson, A., and Mogge, D., eds., Scholarly Publishing and the Electronic Networks: Gateways, Keepers, and Roles in the Information Omniverse (Washington D.C.: Association of Research Libraries, 1994), 37-43 .

Kiernan, K.S., 'The Conybeare-Madden Collation of Thorkelin's Beowulf' in P. Pulsiano and E. Treharne, eds., Anglo-Saxon Manuscripts and their Heritage (Ashgate, 1997), 117-136.

Kiernan, K.S., 'Alfred the Great's Burnt Boethius' in G. Bornstein and T. Tinkle, eds., The Iconic Page in Manuscript, Print, and Digital Culture (Ann Arbor: University of Michigan Press, 1998), 7-32.

Prescott, A. 'The Early Modern History of the Beowulf Manuscript', paper to the International Society of Anglo-Saxonists, Stanford, 1995 (unpublished) .

Prescott, A. '"Their Present Miserable State of Cremation": the Restoration of the Cotton Manuscripts' in C.J. Wright, ed., Sir Robert Cotton as Collector (London: British Library Publications, 1997).

Robinson, P., The Digitization of Primary Textual Sources, Office for Humanities Communication Publications 4 (1993).

Smith, A.H., 'The Photography of Manuscripts', London Medieval Studies 1 (1938), 179-207.

Tite, C.G.C., and Carley, J., 'Sir Robert Cotton as Collector of MSS and the Question of Dismemberment: BL MSS Royal 13.D.I and Cotton Otho D.viii', The Library, 6th series 14 (1992), 94-99 .

Zupitza, J., Beowulf, Early English Text Society 77 (1882); reprinted with new photographs and an introduction by N. Davis, Early English Text Society 245 (1959).