Maþeliende

Matheliende is published quarterly by the students and faculty of Anglo-Saxon Studies at The University of Georgia.

By Michel Aaij

To most readers, Beowulf is a text like any other, neatly translated into Modern English and available in any anthology of early English literature. But the translation is not the only barrier between the modern reader and the older text. For Beowulf survives in only one manuscript, and that a badly damaged one. The handwriting of the two scribes, the orthographic conventions, and the damage done to the manuscript make it very difficult to decipher and understand the text. But modern transcriptions, translations, and interpretations are also influenced by hypotheses about the origins of the poem and the author. Many scholars have assumed that the poem originates in the eighth century and was orally transmitted, through a variety of Old English dialects, to the tenth or eleventh century, when it was written down by scribes who had no relation to the original text and copied mechanically.(1)1 Beowulf, then, would be a late (Christian) copy of an earlier Germanic epic, a view that still persists in most anthologies. This essay discusses the value of the manuscript, and how assumptions on the dating of the manuscript have led editors to discredit the quality of the scribes' work, and thus discredit the authority of the manuscript. I will try to show that opinions on the dating of the text have had an impact on the manuscript's authority, but that arguments for an early dating are based on emendations that presuppose an early dating.

Beowulf was saved for posterity by, among others, Sir Robert Cotton (1571-1631), an avid collector of manuscripts. At present, it is in the collection of the British Museum, and part of a composite codex cataloged as BL MS Cotton Vitellius A.xv. This codex is made up of two different codices, the Southwick Codex (containing four items) and the Nowell Codex (five items, including Beowulf). The part containing Beowulf is dated roughly between 975 and 1025 (see Kiernan "Legacy" 30, Manuscript 2nd ed. xv-xix[2]2). In 1731, Ashburnham House, which housed the Cottonian library, caught fire, and the Beowulf codex was seriously damaged. The edges of the codex were burned, and the binding destroyed. No restoration work was done on the manuscript until the middle of the 19th century. During this period, over a century, the manuscript continued to deteriorate the edges of the leaves, charred and brittled, simply crumbled, and some 2,000 readings were lost.(3) Eventually, the individual leaves were mounted in heavy paper frames, and fastened to the frames with transparent paper strips. The manuscript is still in this state. Reading and transcribing the manuscript is thus problematized by water damage, fire damage, and the restoration work that obscured the text along the edges. But fortunately, in the 18th century, before further deterioration, a transcription was made, by a Dane named Grimus Jonsson Thorkelin. He can also be credited with the modern discovery of the manuscript. Thorkelin was the first to study the manuscript, and publish it.

When Thorkelin arrived in England by royal appointment in 1786, his mission was to gather as many documents as he could, of all kinds, pertaining to the history of Denmark and the Danes. It was probably a somewhat misleading catalogue-description that awakened his interest in the Beowulf manuscript.(4) Thorkelin obviously realized the value of the text, because he procured the services of an employee of the British Museum to make a copy of it. This man, whom Kevin Kiernan identifies as James Matthews (Thorkelin 21-24), produced, on Museum issue paper, the transcript that is known as Thorkelin A (listed as Ny kgl. Saml. 513 in the Kongelige Bibliothek in Copenhagen). When Thorkelin A was made, Thorkelin himself was busy with other documents(5) and traveling throughout England and Scotland. However, the insular hand of the Beowulf manuscript posed obvious difficulties for Mr. Matthews, who had knowledge of neither Old English, nor this particular handwriting.

Matthews (very consistently) misread' Old English runic characters as Roman letters for instance, ð for d and þ for p. The ligature æ he often copied as e, the typically difficult insular t he copied as c.another problem is familiar to anyone who has ever looked at an Old English manuscript: long series of letters like n, I, and u ( minims') look like nothing but a row of short vertical lines ( downstrokes'), because the ligatures connecting the individual downstrokes are extremely thin and often faded. Matthews solved this problem by reproducing, as accurately as possible, each downstroke, thereby sometimes producing words that are really not words at all, without asking himself what the text really read and offering his conjectural version. In some cases, Matthews simply left blanks where he could not decipher the text, as a copyist in the strict sense of the word: he copied the signs, without thinking about what the text or even the individual words meant. The apparent contradiction is that since he did not know Old English or the insular hand, he provided the most reliable transcription by faithfully copying what he saw, without any effort to edit his product. In fact, his errors are for the most part so regular and predictable that the original text can be easily retraced (see Dobbie, xx-xxii). The blanks, and the mistakes that must have been obvious to anybody who knows Old English, left Thorkelin dissatisfied with the product and he took it upon himself to make another copy, now called Thorkelin B (Ny kgl. Saml. 512).

Thorkelin was an editor, rather than simply a copyist. The collation by Kevin Kiernan of both transcriptions and the original(6) shows that Thorkelin was not content with merely copying, but as he copied, he corrected what he supposed to be scribal errors and often enough came with interpretations of his own. Kiernan's collation also shows that Thorkelin must have had A and the original side by side, and used both in producing what in the end was a half-edited version rather than a transcription. Matthews's copying must have deteriorated what was badly damaged, for Thorkelin B has blanks where Thorkelin A apparently did not have a problem. When Thorkelin prepared his editio princeps, he did not have the original manuscript at hand anymore, and although he claimed that his transcription (B) was the first complete and correct one, it is evident that he had to rely heavily on the A copy to produce his 1815 edition and translation.(7) In 1815, Thorkelin published his edition of the poem, and the modern history of Beowulf started.

One of the first editions was Benjamin Thorpe's, from 1830, which was based on a collation of the manuscript and Thorkelin's edition (not the two transcriptions). Thorpe's edition, together with Kemble's 1833 transcription, were the two most authorative editions in the 19th century. At the end of the century, in 1894, Wyatt published his edition, which was revised in 1914, by Chambers. Klaeber's edition (originally published in 1922) is probably the most widely used. It has a long introduction and bibliography, and many useful comments and critical apparatus. He used a number of different editions, notably Thorpe's and Kemble's, and both Thorkelin transcriptions, and has extensive notes on text and emendations. Another popular modern edition is Wrenn's, from 1953, revised by C.F. Bolton in 1972 and in 1988. All these editions use feature a large number of alternative readings and emendations.

Any edition of the poem will necessarily have editorial emendations, due to the state of the manuscript and the problems in deciphering the handwriting. Most editions must rely on Thorkelin A and B for the aforementioned lost readings. Thorkelin A has always been taken as less trustworthy than B, because at first glance the transcription seems far from the original. Most editions perceive B as more authoritative than A, or the manuscript. The most important example of B's authority is a supposed archaic (pre-750) instrumental wundini ( wound') in line 1382,(8) where the manuscript reading looks like wundmi. This, however, is not possible in Old English, and Thorkelin realized that when he made his transcription. He proposed wundini, and this has been taken at face value by subsequent editors.(9) The problem with this reading, however, is that it puts the cart before the horse: it presupposes that Beowulf originates from the eighth century and inserts a conjectural reading to prove that. This particular reading is unparalleled in the manuscript, and is in fact the only linguistic proof' of its eighth-century origin (Kiernan Manuscript 23-37). But the row of four downstrokes can be more reasonably interpreted as wundun, a leveled 10th/11th century form of wunden (a regular Late West-Saxon dative/instrumental), paralleled at least thirty-two times in Beowulf.(10) (This emendation is unproblematic, and repeated on the same folio where inihtigan is invariably and correctly emended to mihtigan.) It is no exaggeration to say that an assumption on the dating of the poem influences the reading of manuscript, which in turn creates a linguistic argument for an early dating of the poem, which discredits the authority of the manuscript again by dating it as a late copy of an early original.

The various emendations in the different editions usually go further than simply proposing different words or letters, which might be scribal errors. Sometimes, they go as far as inserting lines or half-lines where it is assumed the scribes have left material out. Editor's readings are strongly influenced by assumptions about what they think the poem should look like. Since Beowulf is an alliterative poem, some emendations try to repair what is considered faulty. For instance, in line 461 the unusual non-alliterative manuscript reading gara cyn ( spear people') is in most editions replaced by the conjectural but alliterative reading Wedera cyn (Klaeber 18), an emendation that is absolutely unnecessary as far as the text or its meaning is concerned.(11) A similar emendation is made by Klaeber in lines 389b - 390a. He inserts two half-lines, urged by the lack of alliteration. Benjamin Thorpe supplies a footnote to these lines (in his edition beginning with 783, because he numbers half-lines), and says "the text is evidently defective; two lines at least are wanting" (27, see also Kiernan "Legacy" 36-37). Alexander inserts two half-lines similar to Klaeber's; his only comment is that they are "supplied" (220). Elsewhere Klaeber inserts a half-line (403b) to get an even number of half-lines where an uneven number might be a welcome change to the regular meter.(12) Apparently, poetic license is not given to the scribes, neither in alliteration nor in metrics.

Beowulf was copied by two scribes, with very distinctive handwritings, the second one taking over the first scribe's work in line 1939.(13) Now, if these scribes were indeed purely mechanically copying a text, one might feel the need to improve on their work. The problem with this attitude is that it fails to recognize some very basic facts about the manuscript. The manuscript was proofread by both scribes, and was considered important enough to be a separate text, not bound as the next to last part of a codex. Apparently the text had a lot of authority in the 11th century, and the poem must have enjoyed great popularity.

It has always been assumed that the Beowulf-text, the fourth text in the Nowell-codex, was simply a part of a whole, preceded by three prose-texts, which were copied by the first scribe, and succeeded by the poem Judith, in the hand of the second scribe. Because the binding is destroyed, it is very hard to reconstruct the original form of the manuscript, and indeed a thorough investigation was not done before 1981, when Kiernan's Beowulf and the Beowulf Manuscript was published. He argues, from physical evidence, that Beowulf existed as a separate text before it was bound in the codex. The last folio of the Beowulf manuscript is heavily damaged by wear and tear to the point of almost being illegible damage that can only be fully explained by this folio being the last of a quire, indeed the last of a codex (Kiernan Manuscript 126-27,134-47). It is hard to see how it could have come to the state it is in were it originally followed by Judith.

As to the scribes' work, what editors have failed to appreciate fully is the huge amount of corrections made by both. The first scribe proofread and corrected his part, and the second scribe (who took over in line 1939) proofread and corrected the whole text, including the first scribe's work, after he finished copying. Klaeber mentions the corrections, but sees no significance whatsoever in them (Klaeber c). No such attention was given to any of the four other works in the codex, which have what might be called a normal amount of scribal errors, and this only strengthens the contention that the Beowulf manuscript had a special significance.

Beowulf apparently was so important to the second scribe that he, some twenty years after the original copying, worked on the manuscript again. But contrary to what earlier editors have said (notably Zupitza, who in his 1882 edition of the poem was content with studying negatives of the MS rather than the MS itself), he did not retrace the original words on a faded text, while introducing an incredible number of mistakes.(14) Klaeber says in a note to line 2207, the beginning of folio 179: "Folio 179 . . . is the worst part of the entire MS. It has been freshened up by a later hand, but not always correctly" (82). However, this later hand' in fact belongs to the second scribe. Rather than a restoration, it is a revision of the original text folio 179 is a palimpsest: the original text was (poorly) erased and a new one written. The present defective state of this folio is probably due to the new text having been written on a page still damp from the process of erasing, and readings that appear to be errors are caused by the old text shining through (Kiernan Manuscript 219-43). The new text is notably shorter than the old one, and three more lines have been willfully erased from the top of folio 180. All this points to a revision rather than to a simple restoration of a faded text.

Of course, this leads to the question why just one folio of the manuscript had to be revised and a tentative and exciting answer is that during the copying process (rather than centuries before) two originally separate stories were put together. Folios 179 and 180 contain the transition between Beowulf's fights with Grendel and Grendel's dam, and his fight with the dragon: in lines 2207-8, at the top of 179 recto, the Geats' realm passes into Beowulf's hands. Physical evidence from the manuscript proves that the copyist was more than a simple monk who restored a text and did a very poor job. He seems to have played an editor's role if not a writer's.

As can be seen in the various editions of Beowulf, editors have commonly taken the manuscript as less authorative than their own interpretations of the text. Klaeber, Thorpe, Wrenn, Wyatt, all have long lists of emendations that are more often than not based on what they think should be in the manuscript. One of the reasons why editors have treated the manuscript with some disdain, and the scribes as simple and average copyists, is their idea of the origin of the poem. Naturally, if a poem is already three or four centuries old before it is written down, and orally transmitted through a variety of dialects, it is reasonable to assume that it has been changed and corrupted in the course of time (see also Frantzen, ch. 6 "Writing the Unreadable Beowulf," esp. 179-90; Rochester). But the linguistic evidence for this checkered history' (Klaeber lxxxviii) is not very strong, as Klaeber himself admits when discussing wundini (cix-x). In fact, an eleventh-century provenance of the actual poem is a possibility that no editor has yet dared to bet on.

While critics no longer search for a Germanic Ur-Beowulf, the editions of the poem that we study today have to a considerable extent been influenced by the assumption that there was an Ur-version of our poem, and that the Beowulf-manuscript is at the end of a long line of transmissions. But physical evidence from the manuscript itself validates its authority in relation to text and content. The manuscript was originally a single, individual text, and the scribes' minute attention, proofreading, and revising point to a contemporary interest (if not a creative interest) that counters the modern readiness to improve upon it. Perhaps it is time for a new edition.(15) Modern technology (IR and UV photography, better lighting), but more importantly, a different attitude may enable editors to better appreciate the manuscript as we have it, and thus create a new Beowulf. Modern criticism has done away with many of the assumptions of earlier critics, but the next step is a necessary one: to produce a new edition of the poem, untainted by assumptions that color the Beowulf in our text books.

1. In fact, Beowulf is written in regular Late West-Saxon. A few Mercianisms' point to Mercian origin of the poet (or maybe the scribes), but the conservative Anglo-Saxon poetic diction' hardly proves anything about date or original' dialect.
2. All quotations from Kiernan's Beowulf and the Beowulf Manuscript are from the first edition, except where otherwise indicated.
3. Reading' is used in the sense of individual sign.' Letter' is not really a proper expression, since Old English scribes often used one-letter signs to stand for short, often-used words. For instance, on the very first page of the MS the sign þ', with an additional cross-stroke stands for þæþ.'
4. The first known description of Beowulf is by Wanley, who identifies the protagonist Beowulf as a Danish hero at war with Swedish kings, perhaps due to a possible scribe's Beow/Beowulf confusion in lines 18 and 53 of the poem. Wanley transcribed lines 1-19 and 53-73. Whether there are two Beowulfs, or one Beow and a Beowulf, is not important here: the Beow/Beowulf of lines 18 and 53 is a Dane, and not the hero of the poem (Kiernan Thorkelin 35, Wrenn 11, Wyatt ix).
5. The Reading Room Register shows that Thorkelin himself did not study any Old English material in 1787 at the British Museum (Kiernan Thorkelin 18-19).
6. Kiernan was not the first to do this, but he was the first to make a collation with the explicit objective of trying to evaluate the value and accuracy of the Thorkelin-transcriptions. Thorpe did a collation for his 1830 edition. Zupitza made a collation in 1882 for his transliteration, but he used photostats, not the actual manuscript.
7. Thorkelin may have wanted to impress his employer King Christian VII, because the catalog-numbers of A and B were originally 512 and 513, in chronological order. He himself reversed that order, and put the new numbers on the manuscripts, but the old numbers can still be seen underneath the paste-downs with the new numbers (Kiernan Thorkelin 14).
8. This is the reading in Thorkelin B. Thorkelin A has wundmi (Kiernan Thorkelin 67).
9. E.g. Wrenn 16, 153; Wyatt 69. Kemble has wunden from MS reading wundum. See note 10 below. Kemble's preface exemplifies scepticism towards scribes' authority: "Anglo-Saxon MSS . . . [are] hopelessly incorrect . . . , we are yet met at every turn with faults of grammar . . . professional copyists brought to their tasks . . . both lack of knowledge, and lack of care" (xxiii-iv).
10. Kemp Malone, "When Did Middle English Begin?," Curme Volume of Linguistic Studies. Baltimore, 1930, 110-117, quoted in Kiernan's Manuscript 36. Thorpe does not emend to wundini; he has wunden from his MS reading wundum. But wundum could never be correct, as the photograph of folio 160v in Kiernan's "Manuscript" 35 clearly shows. The d is followed by four downstrokes, not five. Since Thorpe collated Thorkelin with the MS, how could he have read wundum? Klaeber has wundnum, from MS reading wundini or wundmi, no less than three downstrokes more than the MS. Alexander and Jack include two more downstrokes, emending wundnu or wundini or wundmi to wundnum (Alexander 222, Jack 111-12).
11. Kemble, the most conservative of editors, retains gara. Thorpe prints gara, but proposes Wara in a note, because alliteration is wanting. But he acknowledges that he doesn't have a clue about who or what the Waracyn are: a Wara or Wedera tribe is unknown. Wrenn follows Thorpe. Dobbie, Alexander, Wyatt, Jack, all have Wedera.
12. That modern editors still take cues from Klaeber is evident in Alexander's 1995 edition. For 403b, Klaeber inserts "heaþorinc eode," and Alexander establishes difference from and conformity to Klaeber's emendation by supplying "eode heaþo-rinc." Dobbie, and Jack, simply indicate omission' for both 389b-390a and 403b.
13. Copied from what?' is a very interesting question, one that will probably never be answered. As far as I know, Kiernan is the only one to suggest an answer: "The transmission of Beowulf may have only been from the poet's wax tablets to the extant MS" (Manuscript 171).
14. "Nearly one blunder every two and a half lines" according to emendations by editors (Kiernan Manuscript 227). See also Wrenn's inconclusiveness on the subject (12), who sees the problems connected to the freshening-up' theory, but does not venture on an alternative hypothesis. He does say that this part of the MS may have been handled separately, even been bound separately, "because of its special popularity" (12), but this supposed popularity is never explained. In any case, without a discussion of the manuscript and which folios were the beginning of which quires, such a claim cannot be made if the bindings have disappeared.
15. Most modern translations, including Donaldson's 1966 prose translation, reprinted in the Norton anthology, are based on Klaeber.