The Legacy of Wiglaf: Saving a Wounded <i>Beowulf</i>

This is an electronic version of an essay published in Beowulf Basic Readings (Garland 1995) and The Beowulf Reader (2000), edited by Peter S. Baker; pp.195–218.

The Legacy of Wiglaf: Saving a Wounded Beowulf

Kevin Kiernan

[This essay was delivered as a lecture at the University of Kentucky on 9 December 1983; it was printed in The Kentucky Review 6 (1986): 27–44. For the present reprint the author has added annotations both supplementing and updating the text.]

To try to dignify my fascination with the Beowulf manuscript, I will liken it to Wiglaf’s attempt, at the end of the poem, to help Beowulf fight the fire-drake. As a Beowulf scholar, I fight like a loyal thane to save the poem from fire-damage and other forms of draconic emendation. In other words, I want to revive an Old English Beowulf, the one still surviving in the manuscript. I am depressed by the cosmetics of the mortuary, the neat and tidy but still rather stiff view of Beowulf I think we get in modern editions of the poem. What makes me a little nervous about my analogy is that all of Wiglaf’s efforts were in vain. Beowulf died, and Wiglaf’s whereabouts are unknown. Nonetheless, a modern-day analogous Wiglaf limps among you.

The single surviving medieval copy of Beowulf is in a hefty composite manuscript known as British Library manuscript Cotton Vitellius A. xv. It is called this because the book was owned by a seventeenth-century antiquary named Sir Robert Cotton who kept track of his manuscripts by their shelf position in bookcases surmounted by the busts of Roman emperors. Thus, Cotton Vitellius A. xv was the fifteenth book on the first shelf of the Vitellius case. If we look inside this big book we find that Cotton bound together two distinct and quite unrelated manuscripts. The first 90 folios are in a twelfth-century handwriting, and we call this part of Cotton’s book the Southwick Codex, based on the notice of ownership—actually a chilling curse on anyone who stole the book—on the second folio. The last 116 folios are copied by two early eleventh-century scribes, and we call this part of Cotton’s book the Nowell Codex, because a previous owner, Laurence Nowell, left his name in it in 1563.¹

The Nowell Codex is the part of Cotton Vitellius A. xv we are interested in: it contains a fragment of the life of St. Christopher; a couple of treatises (one illustrated) describing the kind of monsters who live in the East; the apocryphal Biblical story of how Judith lopped off the head of Holofernes; and of course the true story of how Beowulf, among other things, lopped off the head of Grendel.²

We don’t really know exactly when people began to study the poem in modern times. Perhaps it was Nowell in the sixteenth century who underlined some of the proper names in the manuscript, if indeed he ever attempted to read it.³ But the seventeenth-century table of contents in Cotton’s book leaves a blank for Beowulf; probably reflecting his librarian’s utter bewilderment.⁴ The first intelligible reference to the poem was in 1705, when Humfrey Wanley mistakenly described it as a story about Beowulf the Dane who fought with Swedish princes. At any rate the Beowulf manuscript survived intact, if virtually untouched, until 1731, when a disastrous fire decimated Cotton’s Library and left the Beowulf manuscript badly scorched along its outer edges.⁵

Wanley’s inaccurate description, making our hero a Dane instead of a Geat, can be indirectly credited for preserving a large part of the poem for us.⁶ In 1786, Grímur Jónsson Thorkelin, an Icelander who worked in Denmark as an archivist, and who eventually became the Danish National Archivist, went to England to find Danish heroes in British archives. He learned about Beowulf in Wanley’s description, and in 1787 he hired a professional scribe to copy the manuscript for him, and later made a second copy himself. The great value of these two transcripts is that they alone preserve nearly 2000 letters which subsequently crumbled from the scorched edges of the manuscript.⁷ Thorkelin used his two transcripts to produce the first edition of the poem in 1815, and all modern editors use them, too, to fill in the gaps in the manuscript.⁸

Ten years before Thorkelin’s edition appeared, and exactly one hundred years after Wanley’s brief description of the poem, Sharon Turner, in his history of the Anglo-Saxons, took issue with Wanley, saying that his “account of the contents of the manuscript is incorrect. It is a composition more curious and important.”⁹ Turner should have stopped while he was ahead. He goes on to say, “It is a narration of the attempt of Beowulf to wreck the fæhthe or deadly feud on Hrothgar, for a homicide which he had committed.” Turner was the first scholar to attempt to translate parts of the poem into English. He came up with the following excerpt from the celebrations after the building of Heorot, Hrothgar’s great hall:

There every day
He heard joy
Loud in the hall.
There was the harp’s
Clear sound—
The song of the poet said,
He who knew
The beginning of mankind
From afar to narrate.

“He took wilfully
By the nearest side
The sleeping warrior.
He slew the unheeding one
With a club on the bones of his hair.”

Turner remarks that “The transition to this song is rather violent, and its subject is abruptly introduced, and unfortunately the injury done to the top and corners of the MS by fire interrupts in many places the connections of the sense” (402). Fortunately the fire-damage to the manuscript is nowhere near as serious as Turner indicates. His main problem, in addition to a very rudimentary understanding of Old English poetry, was that the leaf containing Grendel’s attack on Heorot was misplaced in the manuscript when Turner used it.¹⁰ His translation, though, helps illustrate the state of Beowulf studies as late as 1805; and Thorkelin’s Latin translation confirms that the discipline was not much further along as late as 1815.

Although Beowulf now plays a primary role in the history of English language and literature, it has played this role for a shorter time than (say) Wordsworth’s Lyrical Ballads, about as long as the later Romantics. The study of Beowulf is, in other words, a relatively young discipline despite its formidably hide-bound aspect. English literary history was aleady well-established when Beowulf arrived on the scene, and there was no doubt at all where it belonged in the grand scheme.”¹¹The situation was somewhat like the posthumous publication of Hopkins’s poetry in 1918. By then, scholars had characterized the nature of Victorian verse and it was perfectly clear that Hopkins had no place in the conitinuum. He became for a while a 20th-century poet.¹² I believe this is basically the way Beowulf became the earliest English poem.

The situation with Beowulf was, of course, far more complicated. It was an international phenomenon, not only the earliest English epic, but the earliest Germanic epic, an ancient record of a Germanic language, and a new window to the pan-Germanic heroic age, through which everyone eagerly peered. They saw Danes and Swedes, Geats and: Goths, Angles and Jutes, Franks and Frisians, Finns, Norwegians, Vandals, and more, and heard from them all in an early Germanic dialect then known as Anglo-Saxon.¹³ No wonder there was little interest in the unique manuscript, dating perilously near the conquest of Anglo-Saxon England. From the start the manuscript was dismissed as a late, corrupt copy, and scholars set to work trying to reconstruct the ruined original, or what they imagined it to be.¹⁴

The manuscript was so fully ignored, in fact, that until 1916 scholars with unrelated interests could still refer to the date of the Beowulf manuscript as around the year moo, but to the preceding prose texts, in the same handwriting as the first part of Beowulf; as mid-eleventh and even twelfth century.¹⁵ Needless to say, when the mistake was discovered the eleventh and twelfth century dates were quietly moved back to around the year 1000, with a quiet effect on literary history. Scholars had previously thought that two of the prose texts, The Wonders of the East and Alexander’s Letter to Aristotle, were among the last books written in Late Old English.¹⁶ They thought the new fascination with the East must have been imported from the Continent around the time of the Norman Conquest. Now the Anglo-Saxons had to be interested in this kind of romantic lore much earlier, around the year 1000.¹⁷

In fact, there is good linguistic evidence to date Alexander’s Letter, like Beowulf; sometime after 1016—that is, after the Danish conquest and during the reign of Cnut the Great. We can apply this evidence to our dating of Beowulf. The letter exhibits clear, explicable cases of linguistic change, amelioration of the word here, “Danish army,” and pejoration of the word fyrd, “English army.” These words had definite connotations for the Anglo-Saxons. The Bosworth-Toller notes that here “is the word which in the Chronicle is always used of the Danish force in England, while the English troops are always the fyrd, hence the word [here] is used for devastation and robbery.” The same dictionary refines this statement in volume 2, by adding that “in the annals of the eleventh century here is used in speaking of the English.”¹⁸ Obviously it lost the connotations of devastation and robbery. The reason for the semantic amelioration is that, after Cnut’s accession in 1016, the Danish here in word and deed was the English army. How does this criterion affect our dating of Alexander’s Letter? Alexander consistently refers to his special Greek forces as a here, the equivalent of the Anglo-Saxon “select fyrd,” and to his combined forces as a fyrd, the same as the Anglo-Saxon “great fyrd.”¹⁹ He notably refers to his enemy’s armies as a fyrd, the term the Anglo-Saxons reserved for themselves in both senses of the word. In this text, the linguistic pejoration of fyrd (the enemy) and amelioration of here (the good guys) can only be explained by assuming that the translation was made sometime after 1016, after the Danish conquest. It might well be added that the collocation of a Greca (or “Greek”) here, in the sense of a select imperial guard, received its first historical warrant when Danish Vikings served as a here for the Greek emperors in the Varangian guard, which only came into existence in the closing years of the tenth century, just in time for our translation of Alexander’s Letter.²⁰

The usage of here in Beowulf falls in line with the usage in the Letter. Hrothgar’s Danes are specifically singled out as a þeod tilu, “a good nation,” because of their readiness for war both at home and in the here (1246–50). In nearly a score of here-compounds in Beowulf not one carries a pejorative connotation. By contrast, Beowulf’s cowardly thanes, the ones who run away at the end, are called fyrdgesteallum, “companions in the fyrd” (2873). The Anglo-Saxon fyrd earned a similar reputation in their late conflicts with the Danes. I mention this case here to allay any nagging doubts you may have that linguistic evidence precludes an early eleventh-century date for Beowulf. I have yet to hear of any linguistic evidence showing that Beowulf predates its manuscript. The pseudo-evidence always brought forth to bolster an early date is dubious, at best, perhaps because scholars never felt the need to make a strong case for something they deemed self-evident.²¹ In any event, the preconceived notions of where Beowulf belonged in literary history had a profound effect on the text we read in the editions available today. At first, because of a romantic desire to put Beowulf into the pagan past, scholars had to explain all of the implicitly Christian elements, from Genesis to Doomsday, in the poem. That seemed easy. Christian scribes copied the poem over the centuries, and as they did so they merely interpolated Christian parts in precisely the same style as their pagan source.²² This theory exploded when scholars began removing the supposed interpolations, leaving behind a poem in little bitty pieces. The only way to put it back together again without seriously disrupting literary history was to move it into the eighth century, as near to the pagan era and as far away from the manuscript as possible.

The move had many apparent advantages. We still had the earliest English epic, the earliest Germanic epic, an ancient record of a Germanic language, and at least a decent view of the pan-Germanic heroic age. So what if the poet was a Christian? At least he only quoted the Old Testament. Some readers, disturbed by the way Beowulf grew younger as the years wore on, seized on the absence of references to the New Testament as an indication of the Anglo-Saxons’ recent conversion, as if they were first converted to Judaism before being persuaded to switch to Christianity. The real motive, I think, in this line of argument was to root Beowulf in the eighth century, where it could not get any younger.

A more gripping argument with the same motive closes off the ninth and tenth centuries, when the viking invasions traumatized the island.²³ No matter where they sailed from, the vikings were called Danes by the Anglo-Saxons, and it is hard to imagine a poet during these times creating peace-loving, home-body Danes more interested in sleek architecture than sleek warships. It was not a time to be admiring the stout Sea-Geats, either, those unabashed vikings who lost their king Higelac in a raid on the Rhineland. As long as viking raids continued in England, no Anglo-Saxon scop in his right mind would chant the opening lines of Beowulf before a live, beer-drinking audience: “Yes, we have heard about the glorious deeds of the Spear-Danes of the old days—how those noble ones performed deeds of glory; [how] Scyld Scefing deprived so many people of their beer-halls, terrified men .. . until everyone had to obey him and hand over their money. That was a good king!” (1–11).

Without the ninth and tenth centuries, editors were left with what everyone seemed to want, an early poem and a late manuscript. Editors could still make hundreds of changes in the text on the assumption that the late scribes, through laziness, ignorance, and indifference, contributed to the hundreds of blunders the poem accumulated as it lumbered through its supposed transmission.²⁴

What the editors ignored was that the argument ruling out the ninth and tenth centuries for the composition of the poem also wiped out these centuries for the transmission of the text. If no Anglo-Saxon poet would create the poem, no scop recite it, and no audience listen to it, why would ninth and tenth century scribes copy it? The usual scribes of the time were the same monks whose rich monasteries were prime targets of the vikings. If Beowulf is an eighth-century poem, its transmission in Anglo-Saxon times must have been abrupt, from a time when Danes were not synonymous with viking marauders to a time when they ruled England and thus put a stop to the marauding. The regnal list of Danish Scyldings at the start of Beowulf would have been a compliment to Cnut, the latest member of the line, but it is hard to see it as anything other than an insult to any other king of England before him. In the ninth century, King Alfred, who for political reasons reasons seems first to have appropriated Scyld for his line of Anglo-Saxon Scyldings, would not have appreciated the enemy’s version in the prologue of Beowulf.²⁵ And at the end of the tenth century, I am sure that Ethelred Unraed would have been ready to include our scribes in the St. Brice’s Day massacre if he found out about it.²⁶

Using political history to help date the poem early, but not to explain its preservation in a late manuscript, modern editors of Beowulf have always assumed that the ninth and tenth centuries participated in a long and complicated transmission of the text, which included corrosive copying through all the main dialect areas.²⁷ While they agreed that the extant manuscript preserves the poem in more or less standard Late West Saxon, they thought they found the linguistic residue of an ancient, multi-dialectal transmission in one defunct (and in fact imaginary) instrumental reading, an early West Saxon linguistic form here, a Kentish form there, a Mercianism hither, a Northumbrianism yon, with a dash of Saxon patois for good measure.²⁸

But as scholars have increasingly come to recognize, the mixture of linguistic forms in Beowulf is not extraordinary. Most of the archaic forms are part of a poetic word-hoard used also in undoubtedly late poems like Brunanburh and Maldon. The mixture of cross-dialectal forms shows up in late prose, as well as verse, and so must be a reflection of copying conventions in some late scriptoria.²⁹ Keep in mind that Late West Saxon was a literary dialect used throughout England around the year 1000. When it was used in Mercian scriptoria, Mercianisms naturally crept into it; when it was used by a southerner in a Mercian scriptorium, both southern and Mercian features were likely to occur.

Think about our own literary dialect, used throughout the world now regardless of spoken dialects. It contains archaisms like “knight” (Old English cniht) and “should” (Old English scolde), Scandinavianisms like “they” (Old English hi) and “skirt” (Late Old English scyrte, which comes down to us as “shirt”), and a rich mixture of cross-dialectal forms, including those from other ancient and modern languages. A writer with a Mississippi accent today communicates in writing, at least, with readers in the Bronx. Yet even today some spellings differ from center to center. In Binghamton the word “center” in Center for Medieval and Renaissance Studies ends in “-er,” while across the border in Toronto the same word in Centre for Medieval Studies ends in “-re.” Across the same campus, at the Pontifical Institute of Medieval Studies, the word at the Centre spelled “medieval” is at the Institute spelled “mediæval,” another archaism. In short, if we wanted to, we could use the same linguistic criteria used to date Beowulf early to create long and complicated transmissions for texts written the day before yesterday in North America. Throw in an Eastern Kentucky scribe with a sense of humor and he might add an Old English relic like “hit,” standard English “it,” for us to ponder.

My first interest in the Beowulf manuscript had nothing to do with the date of the poem. I accepted the conventional dating. I only wanted to read the poem in its Old English version, freed from all of the modern emendations. In my view, modern editors had done to Beowulf precisely what they had accused the eighth, ninth, and tenth-century scribes of doing. They created a new poem from an ancient source. But whereas the editors could only imagine their pristine source emanating from the misty moors of prehistoric times, I had a hard copy of my ancient source in a photographic facsimile, and the ancient source itself at the British Library on a misty, but well-travelled street in modern London.³⁰

For a long time, the facsimile alone served my purposes. I was mainly interested in verifying manuscript readings where the editions I used had changed them. I will try to give you some idea of the range of these editorial changes without driving you outside or into the arms of Morpheus. I want to show you, in particular, how seemingly innocuous emendations, based on alliteration and meter, have far-reaching consequences.

Theories about Old English prosody, the art of versification, invariably join forces with theories about the date of the poem to justify emendations.³¹ The Old English verse line is divided in two main parts, called an on-verse and an off-verse. Each half-line has two heavily stressed syllables and a variable number of less heavily stressed or unstressed ones. Sophisticated metrical studies have revealed that there are only 5 main patterns, or types, of metrical stress in the half-lines, though these types are by no means as regular as the iambs, trochees, dactyls, and anapests of later English poetry.³² The two half-lines of Old English poetry are linked by alliteration, which can occur on the first or second stress (or both) of the on-verse, but only on the first stress of the off-verse, that is, on the third stress of the full line. So for a line of Old English verse you cannot have “Peter Piper picked a peck,” but only “Peter Piper picked a bushel,” “Ethelred Piper picked a bushel,” or “Peter Cnutsson picked a bushel.”

The problem is that not all of the lines in Beowulf, which was used to establish the rules of Old English prosody, follow these rules. Sometimes there is no alliteration; sometimes there are three stresses in a half-line; sometimes there is only a half-line and sometimes there are three instead of two; rather often an atypical metrical type surfaces. These aberrations naturally offend the sensibilities of metrists and alliterationists, and with all of the positive evidence they have amassed they have generally been able to persuade editors to get rid of the few bits of negative evidence.

According to the rules, any vowel can alliterate with any other vowel, but a particular consonant (including h) is strictly bound to alliterate only with itself. The fact that words beginning with h often alliterate with words beginning with vowels in Beowulf is a sure sign, according to the rules, of scribal corruption.³³ But the relentless application of this rule may be hiding some linguistic evidence from us. We would say that the alliterative phrase “honest Abe” alliterates vocalically with the phrase “heir apparent,” but not with the phrase “hairy head.” “Hairy head” alliterates for me with “humble Harry,” but for some English speakers “’umble ’arry” alliterates with “honest Abe.” In late Old English times the quality of h was protean, too. That’s how Hroðgar and Hroðulf ultimately became Roger and Ralph.

Emending this kind of evidence out of existence can have a major impact on our interpretation of the poem.³⁴ According to our modern editions, there is a character in the poem named Unferð, whose name, the editors tell us, means “mar-peace” (un means “not” and ferð, actually frið, means “peace”). It seems like a good name for the troublemaker who quarrels with Beowulf on his first night in Heorot. One wonders, though, why his parents named him Unferð. Elaborate interpretations have evolved making the character an allegorical representation of Discord or Dissension.³⁵ Yet the name that four times appears in the editions as Unferð appears four times in the manuscript as Hunferð, a fairly common name in the Anglo-Saxon period. The first time it appears in Beowulf the name begins a new section of the poem, and here the scribe went to special trouble drawing a large, unusually ornate, capital H for it.³⁶ Are we to suppose that an ignorant, lazy scribe made such a self-confident and industrious change, and kept his eye peeled for the three additional uncapitalized cases as he copied? As we have seen, the name Hunferð may have been pronounced “’unferth” in Old English times, but neither the poet nor his audience would be likely to interpret the pronunciation, just as we would not be inclined to interpret Cockney “’arry” to mean “light as air, delicate, or graceful.”

Consider another far-reaching emendation involving alliteration. A line lacking it occurs in the manuscript at the point where Beowulf is greeted at Heorot. In the on-verse, the first half-line, Hrodgar tells his messenger to say that Beowulf and his men are welcome by Deniga leodum, “the people of the Danes.” In the off-verse, the second half-line, Word inne abead, “he brought the message in.” Editors have supplied the missing alliteration by creating a modern “Old English” off-verse to alliterate with the real Old English on-verse, and a modern “Old English” on-verse to alliterate with the authentic Old English off-verse. The most influential modern interpolation now reads, þa to dura eode widcud hæleð, “then the famous warrior went to the door—and brought the message in.”³⁷ While the manuscript forgiveably ignores the movements of the messenger, the editor puffs the messenger’s walk to the door into epic proportions. In view of the high stakes, we ought to be able to tolerate a few lines in Beowulf that lack alliteration. Here the allure of alliteration allots a line of modern Old English to the poem. Some Beowulf scholars, who believe that the number of verse-lines in the poem was of special significance to the poet, have even been including such modern interpolations in their calculations.³⁸

The metrists, like their cohorts the alliterationists, believe that an ur-text of Beowulf once existed whose meter unalterably followed their rules. As one metrist says in the opening sentence of his book, “Metrical studies of ancient poetry have at least two immediate aims, the establishment of the text and the recovery of the pleasure inherent in verse.”³⁹ Metrists, to put it in a more skeptical way, aim to emend the manuscript. Their emendations, moreover, are circuitously linked to the belief in an early date for the poem, since the manuscript readings they change undermine their rules. Thus they change by deletion the off-verse in line 9, þara ymbsittendra, “of the neighboring peoples,” to ymbsittendra, “of neighboring peoples,” because þara (“the”) would not be used in this way, they claim, in early poetry.⁴⁰

A final example of how metrical theory and dating theory converge and collaborate can be seen in the on-verse of line 6. The manuscript reads egsode eorl, “terrified the man,” but singular eorl is routinely emended to plural eorlas (“men”) or to the proper name Eorle (“the ancient tribe of Erulians”) because the metrical rules demand an unstressed syllable after it. However, if we do not change the manuscript, the word eorl may be seen as a roughly datable anachronism. Old English eorl took on the meaning of Danish Jarl in Anglo-Saxon England after the Danish invasions, and survives today in the modern title of Earl.⁴¹ Another plausible case occurs later in the poem in the phrase eorl Ongenþio, in reference to a king of the Swedes. Parts of what is now Sweden were ruled by Cnut’s Danish jarls at the time of the Beowulf manuscript. In the context at the beginning of the poem, Scyld Scefing, among troops of his enemies, deprived many tribes of their meadhall benches, but his final victory was that he egsode eorl, terrified the ruling chief, or petty king.

The desired meter for the phrase can be achieved, moreover, without resorting to emendation by pronouncing eorl in two syllables—eor-el. We all know people who say both “athlete” and “ath-e-lete,” or “twirl” and “twir-el” to cite a current “r-l” example, and distinguished poets such as Chaucer, Shakespeare, and Milton have been known to use side-by-side such forms as “Canterbry” and “Canterbury,” “alarm” and “alarum,” “sprite” and “spirit,” solely on the dictates of meter.⁴² The Beowulf poet uses, for example, Dena beside Denia, along with a host of other metrical variants.

In the light of this kind of evidence, I am not persuaded by the metrists’ contention that certain contractions in Beowulf prove that the poem is early. True enough, in the old days the whiskey, Scotch, was Scottish and the people, Welsh, were Wealhish. But there is reason to believe that speakers of Late Old English would have naturally pronounced the contractions in Beowulf in two syllables, rather than one, despite the conventional spellings, particularly if the meter encouraged them to do so. There are only a few such contracted forms in Beowulf and one explanation can serve for all.

In Old English the ending for all verbs in the infinitive form was -an, but at an early stage the stem of the verb do (our “do”) had contracted with this ending to produce the form don instead of doan. But native speakers would have recognized by analogy with all the other infinitives that this was a contraction. Native speakers of modern English recognize, with much less linguistic reinforcement, that “don’t” is a contraction of “do not.” We know, moreover, that in late Old English times the old pronunciation survived or revived in some dialects, for uncontracted spellings like doan re-emerge in some late texts, if not in Beowulf.⁴³

Remember, though, that the Beowulf manuscript is written in the standard late West Saxon literary dialect, and that its spellings do not necessarily reflect the pronunciations of non-West Saxon poets.⁴⁴ Note that today in formal prose we always write “do not” even where we would naturally say “don’t.” If Beowulf is a late poem, the poet may well have decided to use a conservative spelling instead of a provincial one, since he knew that the word would be pronounced correctly in any case. Our standard literary dialect gives us an old, conservative spelling for the number 2, “t-w-o,” but everyone I know now pronounces it [tu] “too,” not [two] “two.”

My quarrel (or quar-rel) with the metrists is not with their aim to recover the pleasures inherent in verse. They have surely hit the target, if not the bull’s-eye, in their analyses. I think they are right that the few contractions in Beowulf usually need to be decontracted to sound like verse. I think they are quite wrong to assume that contractions thereby prove that Beowulf is an early poem preserved in a late, corrupt manuscript. My quarrel with them, with most textual critics of the poem, and with all modern editors of it, is their common aim to establish the text by making emendations to fit their theories.⁴⁵

I was content to believe that Beowulf is an early poem preserved in a late, reliable manuscript, until I studied the manuscript at first hand at the British Library. What I found there was a hoard of evidence that had never been mentioned, much less taken into account, by the metrists, the textual critics, and the editors. Before giving you an inventory of its extraordinary features, I need to tell you what the Beowulf manuscript looks like today. The fire of 1731 destroyed Cotton’s binding and left the outside edges of the manuscript crumbled and charred. In short, the fire had left behind what was essentially a big stack of separate vellum leaves, rather than a book. In the middle of the nineteenth century, in 1845 to be exact, the officials at the British Museum decided to rebind Beowulf and the other manuscripts of Cotton Vitellius A. xv, to make the stack of leaves a book again.

The officials realized that they could not simply slap a new cover around the stack and call it a book. Skilled bookbinders had to connect the leaves somehow, to recreate gatherings that could then be stitched together in a conventional binding. Morever, they had to come up with a way to protect the charred, crumbling edges of the vellum. The method they decided on was to mount each vellum leaf in a paper frame and to bind the paper frames. For each leaf they made a tracing on heavy paper and cut the center out, leaving a retaining space of 1–2mm around the edge. They put paste in this retaining space and then carefully pressed the vellum leaf into place. To secure the leaf from the front, they then pasted transparent strips around the edges. When the work was done they bound the paper frames in a brown-calf facsimile of the original covers.⁴⁶

The new binding was a triumph of book preservation, for it stopped the crumbling of the vellum in its tracks. However, I think it is fair to say that the binders were more interested in the outside appearance of the book than with the inside preservation of the text and the physical features of its original manuscript. The retaining space of the paper frames covers hundreds of letters of the text. Moreover, the decision to mount each leaf separately meant that any vestige of the original vellum gatherings had to be sacrificed.

In 1882, Julius Zupitza attempted to record all of the covered letters for his facsimile edition by holding the manuscript up to the light. He did a great job for someone working before the days of the light bulb. In the summers of 1982 and 1983 I used fiber-optic light to check his work and found few mistakes, but over 300 letters and letter-fragments he had been unable to see.⁴⁷ It is symptomatic of the general neglect of the Beowulf manuscript that no one bothered to verify Zupitza’s claims for a hundred years. And no one had ever tried to reconstruct the original gatherings of the Beowulf manuscript from any scraps of physical evidence that might remain of them.⁴⁸

Any medieval manuscript, quite apart from the texts it preserves, tells a unique story. There are no two alike. Before Beowulf could slay his monsters, someone had to slay a lot of sheep. It was a costly proposition to copy a poem of the length of Beowulf; and someone had to believe that it was worth a small flock of sheep, and a large flock of precious time and labor. Once the sheep were slaughtered and skinned, the hides had to be washed, limed, unhaired, scraped, dried, washed again, stretched on frames, scraped again to remove blemishes, smoothed and polished with pumice, softened with chalk, and cut to size.⁴⁹ When the prepared sheets of vellum finally went to the scriptorium, more work had to be done before any copying could begin. Someone, presumably the scribe, had to rule the sheets of vellum for lines of text and margins, and arrange the sheets in small booklets called gatherings. The usual gathering at this time was made up of four folded sheets, providing eight leaves or sixteen pages to write on.⁵⁰

The way booklets were put together in manuscripts sometimes tells us something worth knowing about the intended use of the texts. For example, in manuscripts of homilies, scribes often copied separate texts in single gatherings, small, irregular, self-contained units that could be removed from the manuscript for use in preaching.⁵¹ Though the original gatherings of the Beowulf manuscript were permanently obscured by the fire and the new binding, anyone with my blind faith in this manuscript wanted to hear its hidden secrets. Part of its story was buried in those original gatherings.

My approach was to collect any extraordinary manuscript evidence that had been overlooked, ignored, or forgotten. For instance, I had to determine for each vellum leaf which side was the hair (or wool) side and which side was the flesh side of the animal skin. If I was going to reconstruct the original gatherings, I could not deduce an original sheet that had hair on one side of its fold and flesh on the other.⁵²

I also had to make sure that the scribal rulings lined up properly. The scribes ruled the vellum for lines of text and margins a gathering at a time. First, they punched tiny holes through the stacked sheets of a gathering along both sides for lines of text, and along the top and bottom for the margins. Then they drew the lines with a ruler and an awl, using the holes as guide-marks. Although these guide-marks were destroyed in the fire, the writing-grids they helped create of course still survive. The awl left rulings in the form of furrows, or indentations, which show up in reverse on the other side of the sheet. I could not come up with a sheet that had furrows on one side of the fold and reversed markings on the other. Nor could I come up with a sheet that had 20 rulings on one side of the fold and 22 on the other, since the same sheet would have had the same number of guide-marks for the rulings.

By this simple process, I was able to establish the most probable construction of the original gatherings and definitely eliminate some alternatives that once had seemed more likely. I discovered that the two scribes of Beowulf had constructed their gatherings in completely different ways. The first scribe had made 4-sheet gatherings, ruled (with one exception) for 20 lines of text, and had consistently arranged his sheets with hair sides facing hair sides and flesh sides facing flesh sides, to obscure the contrast when the book was opened at any point. This arrangement was typical for early eleventh-century manuscripts. But the second scribe had made 5-sheet gatherings, ruled for 21 lines of text, and had invariably arranged his sheets with hair sides facing flesh sides, as if to highlight the contrast between hair and flesh wherever the book was opened in his part. It is a striking change in format.

Knowing this kind of information can have important consequences. I believe, for example, that I have been able to identify another manuscript from the same scriptorium on the basis of striking paleographical and codicological similarities. The manuscript is the famous Blickling Homilies codex in the Scheide collection at Princeton. A paleographical connection was noted long ago, but no one ever noticed that the Beowulf manuscript and the Blickling Homilies manuscript share the same odd features in the sheet arrangement of the gatherings and that the size of the writing grids are virtually identical.⁵³ What makes this discovery so exciting to me is that it explains in the best possible way why the description of Grendel’s mere is so like the description of Hell in Blickling Homily 16.⁵⁴ The Beowulf poet had access to this manuscript of homilies, which is dated internally in the year 971. Beowulf must have been composed after that.

You can understand, then, my interest in the original gatherings of the Beowulf manuscript. Through my analysis of the sheet arrangement, I learned that the make-up of the first gathering in Beowulf was extraordinary. In 1705, when the original gatherings were still intact, Wanley had told us that Beowulf began a separate manuscript, but later scholars preferred to think that the poem had been copied continuously with the prose texts that precede it, and that copying of the poem began in the middle of the last prose gathering. Since they had no evidence to5 contradict Wanley, I think these scholars wanted to believe that Beowulf was copied mechanically and that its manuscript was in no way special to the scribes. The hair and flesh arrangement of the leaves supported Wanley’s statement, while the rulings from the Beowulf leaves did not line up properly with rulings on the relevant prose leaves.⁵⁵

There is a good deal of corroborating evidence that the first page of Beowulf originally served as an outside cover. I will mention only one aspect of this evidence. The page shows unmistakeable signs of unusual wear and tear that cannot be attributed to exposure in modern times. Most of the damage is in the lower right corner, where some of the text has worn off and is no longer legible. It looks like the result of excessive handling, as if the book had been repeatedly held by the corner. The damage presumably occurred in Old English times, since Wanley, in his transcript of the first page, copied one of the partly illegible words as it now appears and then stopped transcribing when he reached the other illegible words. The Thorkelin transcripts unequivocally show that the damage was as advanced in 1787 as it is now.

There was inarguable evidence, in addition, that the last page of Beowulf had also served as an outside cover, making the manuscript what appeared to be a special, self-contained unit. The most obvious evidence was that the scribe had to use a plethora of abbreviations in order to squeeze in the last lines of the poem on this page; that he later had to freshen up the ink where readings had worn off; and that a medieval bookworm feasted on the last pages of Beowulf before the Judith fragment became part of the codex.⁵⁶

To make a long lecture somewhat shorter, I found some remarkable things going on in this newly separate, special manuscript. Those ignorant, lazy scribes had both carefully proofread their work and had made nearly 200 corrections of their mistakes. I can’t believe they overlooked up to 350 additional mistakes, about one every ten lines, as the modern editions maintain. The second scribe had even proofread the first scribe’s work, and in addition to making some corrections had made a few minor emendations.⁵⁷ There was no comparable evidence of proofreading, by either of the scribes, in the prose texts. This convincingly proved that the scribes were neither ignorant nor lazy, that they well understood what they were copying and that they worked uncommonly hard to provide a reliable text.

But there were more remarkable things going on. The second scribe, who took over copying in the middle of one of the first scribe’s gatherings, ostentatiously ignored the rulings on four consecutive pages and between the first and last rulings adroitly inserted more lines of text than the rulings provided for. To appreciate his feat, try doing it yourselves in a lined notebook. In the immediately preceding gathering, the first scribe, who always ruled his gatherings for 20 lines of text, suddenly ruled one for 22 lines, before resuming his normal format. The second scribe, after squeezing in those extra lines in the first scribe’s gathering, used a totally different format for his own gatherings, with more sheets, more lines of text per page, and wider margins.

The first leaf of his first gathering was a full palimpsest—that is, the original text on it had been completely eradicated, and a new, shorter text, written in a slightly different script with a few strange spellings, had replaced it. Parts of it were later erased, and a full restoration was never achieved. On the reverse side of the second leaf three lines had been deliberately erased, with no attempt to replace or restore them.⁵⁸ In the midst of all of these remarkable features, the first scribe, who numbered the sections of the poem after copying them, introduced an error in the number sequence. The second scribe sometimes forgot to leave space for numbers in his part, but otherwise continued numbering his sections based on the first scribe’s erroneous sequence.⁵⁹

It all seemed like the locus Anglo-Saxonicus of the right hand not knowing what the left hand was doing, or the mental breakdown of a schizophrenic manuscript. All of the traumas were clustered at the breaking point, where one scribe abruptly (in the middle of a verse) stopped copying and the second took over.⁶⁰ Moreover, they were all clustered in the section of the text known as “Beowulf’s Homecoming,” a loose transition that fuses the story of Beowulf’s youthful exploits in Denmark with the story of his confrontation as an old man with the dragon in Geatland.⁶¹

For me, all of the evidence led to an unorthodox, but seemingly inescapable conclusion. Two separate stories and two separate manuscripts had been linked together in the same manuscript that has come down to us from the early eleventh century. The Beowulf manuscript was not a late copy of an early poem, but a revision-in-progress of a contemporary one. It was not planned in advance, to judge by the sudden breakdown in the format.

Both scribes copied parts of the new transitional text. The first scribe stopped where he did to go back and supply his share of the revision in the preceding gathering. The length of the new text did not permit him to use his normal-sized gathering ruled for 20 lines to the page. He was thus obliged to rule it for 22 lines to the page. If part of his new transition had deleted a former section of the poem it would explain how he messed up the number sequence. He recopied the old numbers, not remembering that one of the old numbers was now gone. Since the second scribe often forgot to leave space for numbers, it follows that he had no numbers to miscopy from his exemplar. This deduction explains why he innocently resumed the number sequence where the first scribe had erroneously left off.⁶² The second scribe, moreover, was obliged to squeeze in extra lines of text, in disregard of the rulings, because he had already copied the last two gatherings, containing the dragon episode. It was not easy to link up two completely different stories in two different manuscripts.⁶³

The palimpsest suggests that the second scribe many years later decided that the transition was not as it should be. After erasing all of the original text on the first folio (front and back) of the dragon episode, and three related lines on the reverse side of the next folio, he provided a new start for this episode. The current state of the text on this folio indicates that it was still in a draft stage when the poem’s Old English history came to an abrupt halt.⁶⁴ It is well to remember that at this time Anglo-Saxon history was about to come to an abrupt halt, too.⁶⁵ The poem remains unfinished on this folio to this day, making the manuscript, at least in part, an early eleventh-century record of an early eleventh-century poem.⁶⁶

Notes

1. For a full analysis of the composite codex with illustrative plates (one in color) see Kiernan, “The History and Construction of the Composite Codex,” Beowulf and the Beowulf Manuscript (New Brunswick, 1981), 65–169, hereafter referred to as BBMS. For the Nowell Codex alone, see also Kemp Malone, The Nowell Codex, British Museum Cotton Vitellius A. XV. Second Manuscript, Early English Manuscripts in Facsimile 12 (Copenhagen, 1963).

2.The Judith-fragment is itself headless, and scholars do not agree how much of the text is actually lost at the beginning, a head only or a torso too. Moreover, its concluding lines are only preserved by an early modern hand, suggesting that the rest of its original manuscript was still intact at the time, but for some reason was dismembered. I argue that the fragment was added to the Nowell Codex when its monastic library was dissolved in the Reformation in BBMS, esp. 59–60, 150–159. Kenneth Sisam first suggested that the Nowell Codex was conceived as a book of monsters in Studies in the History of Old English Literature (Oxford, 1953), 66; cf. Michael Lapidge, “Beowulf, Aldhelm, the Liber Monstrorum and Wessex,” Studi Medievali 23 (1982): 151–192.

3. George Hickes made the earliest known reference to Beowulf when he replied in a letter to Humfrey Wanley on 20 August 1700, “I can find nothing yet of Beowulph” (BBMS, 133 note 44). The letter was recently printed in A Chorus of Grammars: The Correspondence of George Hickes and His Collaborators on the ‘Thesaurus linguarum septentrionalium’, ed. Richard L. Harris, Publications of the Dictionary of Old English ₄ (Toronto, 1992), 337.

4. See BBMS, plate 2, 75. This and two other early modern flyleaves in Cotton’s binding unfortunately threw off the numbering of the manuscript leaves when in 1884 they were included in a count now regarded as “official,” even though one of the three prefixed leaves is no longer in the codex. For students of Beowulf, the best recourse is to use the older foliation still visible on the manuscript leaves, as nearly all modern editors have done.

5.Sir Frederic Madden, Assistant Keeper of Manuscripts in the British Museum, 1828–37, and Keeper, 1837–66, began restoring the burnt Cotton manuscripts in 1839. His ledger book, Cottonian MSS, Repairing and Binding Account, is now BL MS Add. 62577.

6. Readers of translations may not know about Beowulf the Dane, who often shows up in them as Beow. In The Norton Anthology of English Literature, 5th ed. (New York, 1986), E.T. Donaldson asserts that “most scholars now agree” that the MS readings at lines 18 and 53 should be changed to “Beow” (note 2). An informal poll of Anglo-Saxonists on ANSAXNET, the computer network for AngloSaxonists, seemed to suggest the contrary, despite a spirited defence by Howell Chickering of his use of Beow in his dual-language edition. The discussion can be found in ANSAXDAT, http://www.mun.ca/Ansaxdat/.

7. For an account of his research trip see Kiernan, “Thorkelin’s Trip to Great Britain and Ireland, 1786–1791,” The Library, 6th series, 5 (1983): 1–21; and Part One of The Thorkelin Transcripts of Beowulf (Copenhagen, 1986).

8.De Danorum Rebus Gestis Secul. III & IV. Poëma Danicum Dialecto Anglosaxonica (Copenhagen, 1815). Editors have usually relied on Julius Zupitza’s transliteration, which incorporates lost readings from the Thorkelin transcripts. Klaeber, for example, says in a note preceding the text that readings quoted from A and B are “from Zupitza’s notes.” “The Text,” Beowulf and the Fight at Finnsburg, 3rd ed. (Boston, 1950), facing p. 1.

9.Sharon Turner, The History of the Manners, Landed Property, Government, Laws, Poetry, Literature, Religion, and Language, ofthe Anglo-Saxons (London, 1805), 4:402.

10. See BBMS, 137. In a recent study also citing Turner’s blunder, Allen J. Frantzen refers to the misplaced leaf as “fol. 137” (Desire for Origins: New Language, Old English, and Teaching the Tradition [New Brunswick, 1990], 194–195). In the manuscript foliation the leaf was numbered 131 when it was out of place. Now that the leaf is back in place between folios 146 and 147, we can keep track of its former misplacement without disrupting the useful manuscript foliation by referring to it as fol. 147A(131).

11. After Thorkelin suggested in the title of his editio princeps that the poem derived from the third or fourth century, scholars slowly moved the date forward to a fairly firm consensus in the eighth century. For an overview see Colin Chase, “Opinions on the Date of Beowulf 1815–1980,” The Dating of ‘Beowulf’ ed. Colin Chase (Toronto, 1981), 3–8.

12. Hopkins died in 1889, but still seemed to belong to “The Twentieth Century” (between Thomas Hardy, who died in 1928, and Bernard Shaw, who died in 1950) as late as the 4th edition of The Norton Anthology of English Literature (New York, 1979), 2:1785–97. He is, however, referred to as a Victorian poet in the 5th ed.

13. J.S. Cardale reminds his readers at the start of the 19th century that Hickes in the early 18th century distinguished three Saxon dialects in England: Britanno-Saxon from the 5th-century Anglo-Saxon invasions until the coming of the Danes in 793; Dano-Saxon, until the Norman Conquest in 1066; and Normanno-Dano-Saxon after that. Cardale presumed that King Alfred spoke “pure Anglo-Saxon” (King Alfred’s Anglo-Saxon Version of Boethius De Consolatione Philosophiæ: with an English translation, and Notes [London, 1829], no page numbers). Today most Anglo-Saxonists have forgotten all about the Dano-Saxon dialects that must have been spoken in the vast Danelaw areas of Anglo-Saxon England.

14. J.M. Kemble set the prevailing tone in the English editions when he said, “All persons who have had much experience of Anglo-Saxon MSS. know how hopelessly incorrect they in general are. . . . we are yet met at every turn with faults of grammar, with omissions or redundancies of letters and words, 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.” He adds that “A modern edition, made by a person really conversant with the language which 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” ( The Anglo-Saxon Poems of Beowulf; the Travellers Song and the Battle of Finnes-burh [London, 1833], xxiii–xxiv).

15. Stanley Rypins, ed., Three Old English Prose Texts in MS. Cotton Vitellius A xv, EETS 161 (1924), xi–xiv. As Rypins says, Sisam first pointed out in print the identity of the first hand of Beowulf with that of the prose texts. Madden had in fact sorted them out in his journals nearly a century earlier, as I mention in “Madden, Thorkelin, and MS Vitellius/Vespasian A. xv,” The Library 8 (1986): 130.

16. As Richard Wülker says, “Dieses Stück [i.e. Wonders of the East) wie das vorige [Alexander’s Letter] entstand wohl kaum früher als um die Mitte des elften Jahrhunderts” (Grundriss zur Geschichte der angelsächsischen Litteratur [Leipzig, 1885], 505).

17. The loose dating of the Beowulf manuscript “around the year 1000” or “circa 1000” has effectively prevented scholars from investigating the implications of Ker’s paleographical dating of the script any time in the 50-year span from the last quarter of the 10th century through the first quarter of the 11th. Thus a recent book on Cnut, who ruled England during Ker’s dating limits, avoids the issue by presuming that “the surviving manuscript of Beowulf with its story of pagan Scandinavia,” comes from the reign of Æthelred, who was deposed by Cnut’s father in 1014 (M.K. Lawson, Cnut: The Danes in England in the Early Eleventh Century [Longman, 1993], 130–131).

18. J. Bosworth and T. Northcote Toller, An Anglo-Saxon Dictionary (Oxford, 1898), 532, and Toller, Supplement (Oxford, 1921), 537. For a review of opinions on these words, see Phillip Pulsiano and Joseph McGowan, “Fyrd, here, and the Dating of Beowulf,” Studia Linguistica Posnaniensia 23 (1990): 3–13.

19. As this use of fyrd shows, Pulsiano and McGowan misunderstand my position when they say that by my theory “here would have come to mean by the reign of Cnut the forces of the allies; fyrd the forces of the enemy” (11). Both words can be used with neither positive nor negative connotations in Beowulf. My point is that fyrd had pejorated to the extent that it was no longer reserved for English forces, and that here is used honorifically.

20. See Christine Fell, “The Icelandic Saga of Edward the Confessor: The Version of the Anglo-Saxon Emigration to Byzantium,” Anglo-Saxon England 3 (1974): 179–196; and Sigfús Blöndal, The Varangians of Byzantium: An Aspect of Byzantine Military History, translated, revised and rewritten by Benedikt S. Benedikz (Cambridge, 1978). The most famous former Varangian guard associated with England at this time was Harold Hardråde, who died at Stamford Bridge trying to conquer England in 1066.

21. Ashley Amos provides a comprehensive analysis of these tests in Linguistic Means of Determining the Dates of Old English Literary Tests (Cambridge, Mass., 1980). See also Angus Cameron et al., “A Reconsideration of the Language of Beowulf” in The Dating of Beowulf, 33–75, and Kiernan, “The Linguistic Tests for an Early Date,” BBMS, 23–37.

22. Karl Müllenhoff, “Die innere Geschichete des Beovulfs,” ZfdA 14 (1869): 193–244. Walter A. Berendsohn; Zur Vorgeschichte des ‘Beowulf’ (Copenhagen, 1935). Klaeber notes with exasperation that “even the exact number of lines credited to each one of the six contributors was announced by Müllenhoff,” adding that “Ettmüller in his edition (1875) pared the poem in its pre-Christian form down to 2896 lines,” while “Möller condensed the text into 344 four-line stanzas” (cii, n. 1).

23. Dorothy Whitelock, The Audience of ‘Beowulf’ (Oxford, 1951), 24–25.

24. For an historical overview of editorial changes, see Birte Kelley, “The Formative Stages of Beowulf Textual Scholarship: Part I” Anglo-Saxon England II (1983): 247–74, and “Part II,” Anglo-Saxon England 12 (1984): 239–75. Cf. Tilman Westphalen, Beowulf 3150–3155: Textkritik and Editionsgeschichte (Munich, 1967).

25. See A.C. Murray, “Beowulf; the Danish Invasions, and Royal Genealogy,” in The Dating of Beowulf, 101–111.

26. See the Chronicle entry for 1002. Although Ethelred’s order could not have had much effect in the Danelaw, there were massacres, as Ethelred himself makes clear in a charter renewal in 1004 for the monastery of St. Frideswide, Oxford. As one of his scribes records, “to all dwelling in this country it will be well known that, since a decree was sent out by me with the counsel of my leading men and magnates, to the effect that all the Danes who had sprung up in this island, sprouting like cockle amongst the wheat, were to be destroyed by a most just extermination, and this decree was to be put into effect even as far as death, those Danes who dwelt in the afore-mentioned town, striving to escape death, entered this sanctuary of Christ, having broken by force the doors and bolts, and resolved to make a refuge and defence for themselves therein against the people of the town and suburb; but when all the people in pursuit strove, forced by necessity, to drive them out, and could not, they set fire to the planks and burnt, as it seems, this church with its ornaments and its books” (English Historical Documents, I, c. 500–1042, ed. Dorothy Whitelock [London, 1955], nos. 127, 545).

27. Sam Newton attempts to resuscitate this theory in The Origins of ‘Beowulf’ and the Pre-Viking Kingdom of East Anglia (Cambridge, 1993), 10–11.

28. Thus Klaeber characterizes the spelling ēo, for instance, as variously reflecting Anglian, Kentish, Saxon patois, Northumbrian coloring, and even “partly” West Saxon dialects (lxxx). There is a full discussion of wundini (1382) in BBMS, 30–37. Wrenn called this form “the only certain evidence for dating Beowulf before circa 750 on purely linguistic grounds” (Beowulf with the Finnesburg Fragment [London, 1953], 21, note 2). The manuscript reads wundmi, as Thorkelin A records.

29. While she agrees with me that “the mixture of spellings does not necessarily mean a long history of textual transmission” (411, note II), Janet Bately for dating purposes singles out the Early West Saxon shibboleth ie as phonological rather than orthographical evidence (411–415). She admits, however, that ie spellings occur even in 12th-century manuscripts (412). See “Linguistic Evidence as a Guide to the Authorship of Old English Verse: A Reappraisal, with Special Reference to Beowulf,” Learning and Literature in Anglo-Saxon England, ed. Michael Lapidge and Helmut Gneuss (Cambridge, 1985), 409–431.

30. Both modern facsimiles use the same photographs. See Julius Zupitza, ed., Beowulf Reproduced in Facsimile, 2nd edition by Norman Davis, EETS 245 (London, 1959); and Malone, The Nowell Codex.

31. The most recent proponent is R.D. Fulk, who defends the 19th-century theories of reconstruction in his exhaustive History of Old English Meter (Philadelphia, 1993).

32. In the Preface to The Metrical Grammar of ‘Beowulf’ (Cambridge, 1992), ix–x, Calvin Kendall confides, “I began this study many years ago in the fond hope of reducing to a neat system (neater than the systems of Sievers, Pope or Bliss, whose basic patterns kept dissolving, as it seemed to me then, into a welter of anarchic subpatterns) the seemingly endless varieties of rhythmic possibilities in Beowulf . . . That goal,” he continues, “now seems to me a will-o’-the wisp.”

33. This is the only motivation for emending handlean to andlean ‘retribution’ (1541, 2094) or hondslyht to ondslyht ‘counterblow’ (2929, 2972). For an excellent discussion of the phonological value of h in Beowulf see M.F. Vaughan, “A Reconsideration of ‘Unferð’,” Neuphilologische Mitteilungen 77 (1976): 32–48.

34. Historical linguists have traditionally preferred the diachronic reconstruction of a prehistoric stage of Old English to a synchronic analysis of unique or unusual manuscript forms, which may in fact give us important linguistic information if it is not hidden by emendation. For example, at Beowulf 2295, þone þe him on sweofote sare geteode, Klaeber notes without apparent motivation that “sare is an adverb, not the object of the verb, the fem. gender of the noun sar being more than doubtful” (210–11). He does not mention that at line 2468b the MS. reads sio sar (fem. nom. sg.), which he quietly emends to to sar. We may have here a late confusion of forms. Cf. the strong scod (1887b) and weak sceþede (1514b), morphological variants of the verb sceþðan ‘harm’, or the phonological variants hraðe (which usually alliterates with h, but sometimes with r) and raðe (724). Because of editorial emendation based on 19th-century interest in reconstruction, it will be necessary for linguists to return to Old English manuscripts to discover much of this kind of evidence.

35. Morton Bloomfield, “Beowulf and Christian Allegory: An Interpretation of Unferth,” Traditio 7 (1949–50): 410–415.

36. Fol. 141r15 (line 499). The fit numbers in the manuscript are also routinely ignored. For an excellent study, see Eamonn Ó Carragáin, “Structure and Thematic Development in Beowulf” Proceedings of the Royal Irish Academy 66 (1967): 1–51.

37. Lines 389–39o. Although Birte Kelley, “The Formative Stages,” does not mention it, Kemble began this radical line of emendation in his 1835 edition when he commented that “Probably two lines have been omitted, of which the second may have been Wulfgár maþelode” (28).

38. See, for example, T. Hart, “Ellen: Some Tectonic Relationships in Beowulf and Their Formal Resemblance in Anglo-Saxon Art,” Papers on Language and Literature 6 (1970): 263–90; David Howlett, “Form and Genre in Beowulf,” Studia Neophilologica 66 (1974): 309–325. Even if something is missing by haplography in lines without alliteration, it might just as well be three lines as one or four. The advantage of following the manuscript version is that we can be sure an Anglo-Saxon audience knew Beowulf in that form.

39. John C. Pope, The Rhythm of ‘Beowulf,’2nd ed. (New Haven, 1966), 3.

40. Klaeber was also bothered by the rare anacrusis in the off-verse of a Type D meter, but he omits þara, he says, because it is “syntactically objectionable” (280). I would scan 9b without emendation as x x / / x x, Type C. As E.V.K. Dobbie observed, “the metrical arguments for the omission of this word are not as convincing as they once were, and the argument from syntax (the infrequency of se, seo, þæt as a definite article in the early poetry) is refuted by ymbesittendra ænig ðara, 1. 2734, where ðara is required by the meter.” Beowulf and Judith, Anglo-Saxon Poetic Records 4 (New York, 1953), 114.

41. See OED, sense 2, “In late OE: a Danish under-king (see JARL); hence (under Cnut and his successors) the viceroy or governor of one of the great divisions of England, Wessex, Northumbria, Mercia, etc.”

42. After noting that the line as it stands in the manuscript might be metrically acceptable, Fulk objects to my suggestion of an intrusive vowel in eorl because of a lack of “orthographical evidence for such anaptyxis” in this word (History, 205, n. 70). See the variant spellings erel, errel, erell errille of Earl in OED, which also cites the Old Norse runic spelling erilaR. Intrusive vowels are in fact most likely to occur in the environment of liquids r and l, even when spellings fail to reflect them.

43. A. Campbell points out that “analogical forms with the contraction eliminated quite often occur.” Among his examples are Northern forms of the verb “to do,” dōa for Southern don (Old English Grammar [Oxford, 1959), §239, p. 104). Both the Dictionary of Old English, ed. A. Cameron et al., Fascicle D (Toronto, 1986), 509, and the Microfiche Concordance to Old English, ed. Antonette diPaolo Healey and Richard Venezky (Toronto, 1980), 211–213, list many uncontracted forms, including doan and the inflected infinitive to doanne.

44. C.L. Wrenn, “The Value of Spelling as Evidence,” Transactions of the Philological Society, London (1933): 14–39, and, more recently, Angus Cameron et al., “A Reconsideration of the Language of Beowulf,” Dating of Beowulf 33–75, and Joseph Tuso, “Beowulf’s Dialectal Vocabulary and the Kiernan Theory,” The South Central Review 2 (1985): 1–9.

45. Kendall says he has “resisted the temptation to emend the text to suit my theories” (xv), but by “the text” he does not mean the manuscript, but Klaeber’s edition, which is already heavily emended for metrical reasons.

46. Scholars sometimes talk about the manuscript without having seen it. For example, J.D.A. Ogilvy and Donald C. Baker wrongly assert that “Today each leaf of Vitellius A. 15 is encased in transparent plastic to protect it from further harm” in their chapter, “The Manuscript,” in Reading ‘Beowulf’: An Introduction to the Poem, Its Background, and Its Style (Norman, Oklahoma, 1983), 4. Some of the mistakes in Leonard Boyle’s “The Nowell Codex and the Poem of Beowulf” (Dating of Beowulf 23–32) can be explained because he formulated it first in Toronto in April 1980, before he had ever seen the manuscript.

47. “The State of the Beowulf Manuscript, 1882–1983,” Anglo-Saxon England 13 (1984): 23–42.

48. Neil Ker determined the hair/flesh patterns in the single quire of Judith, but for some reason did not try to ascertain them in Beowulf where they can help establish the original construction of the dismembered gatherings (Catalogue of Manuscripts Containing Anglo-Saxon [Oxford, 1957], 282). Cf. Malone, Nowell Codex, 16.

49. There is a good introduction to these methods in Christopher de Hamel, Medieval Craftsmen: Scribes and Illuminators (Toronto, 1992). See also the section on “Materials and techniques” in Michelle Brown, Anglo-Saxon Manuscripts (Toronto, 1991), 46–53.

50. The characteristic methods mentioned below for arranging, folding, and ruling gatherings are based on Ker’s findings, Catalogue, xxii–xxv; see also L.W. Jones, “Pricking Manuscripts: The Instruments and Their Significance,” Speculum 21 (1946): 389–403, and the excellent guide by Alexander Rumble, “Using Anglo-Saxon Manuscripts,” forthcoming in Basic Readings in Anglo-Saxon Manuscripts, ed. Mary P. Richards (New York: Garland, 1994).

51. P.R. Robinson, “Self-Contained Units in Composite Manuscripts of the Anglo-Saxon Period,” Anglo-Saxon England 7 (1978): 231–38.

52. After describing them correctly as five- and three-sheet gatherings, Max Förster mistakenly describes the first two gatherings of the Nowell Codex as four-sheet quires. The hair/flesh patterns of the second proposed quire unequivocally show that only one set of folios could have been conjugate (“Die Beowulf Handschrift,” Berichte über die Verhandlungen der Sächsischen Akademie der Wissenschaften zu Leipzig, Philologisch-historische Klass 71 (1919): 7–8 and 21–23). Cf. Ker, Catalogue, 282, and Boyle, Dating of Beowulf, 23–24, for the same mistake.

53. Scheide MS 71. For a full facsimile see Rudolph Willard, ed., The Blickling Homilies (The John H Scheide Library, Titusville, Pennsylvania), Early English Manuscripts in Facsimile 10 (Copenhagen, 1960); the collation is more accurately presented in Donald Scragg, “The Homilies of the Blickling Manuscript,” Learning and Literature, 299–316.

54. See Rowland Collins, “Blickling Homily XVI and the Dating of Beowulf,” Bamberger Beiträge zur Englischen Sprachwissenshaft 15 (1983): 61–69.

55. Wanley not only notes that Cotton Vitellius A. xv is a composite codex (ex diversis simul compactis), but also describes the prose texts alone as belonging to the Nowell Codex. For him Beowulf began a new book. I think the scribe, wishing to make Beowulf a self-contained unit for ease of access, pulled a sheet from the previous gathering when he began copying the poem. See my article, “A Long Footnote for J. Gerritsen’s ‘Supplementary’ Description of BL Cotton MS Vitellius A. xv,” English Studies 72 (1991): 493–496 and nn. 12–16.

56. In “The Place of Judith in the Beowulf-Manuscript,” Review of English Studies 41 (1990): 470, n. 21, P.J. Lucas asserts that “Kiernan’s claim (Beowulf p. 151) that the depth of the ruled frame in Quire 14 [i.e. Judith] is half a centimetre less than that of the previous quires is not observable.” Lucas was presumably observing Malone’s facsimile, rather than the manuscript, where one can both observe and accurately measure the rulings. Boyle, whom Lucas cites as an authority on this manuscript (465), radically mismeasures them at the start of Beowulf when he says “the spaces between rulings 18 and 19 (giving line 19) and between 19 and 20 (giving line 20) are 9 mm. and 10 mm. respectively for each frame in folios [123–130], whereas the space ruled for folios [132–139] is the regular 7 mm. or so that one usually finds throughout this part of the Nowell Codex and for the rest of the lines in each frame in folios [123–130]” (23). If Boyle’s measurements were correct, the usual text frames would measure only c. 133 mm. from top to bottom (7 mm. x 19 spaces between ruling 1 and 20 = 133), rather than c. 175 mm. for the written space, as Ker says (Catalogue 282).

57. See BBMS, 272–277. An example is the second scribe’s emendation of the first scribe’s customary spelling of -scaðan to -sceaðan at fol. 140v14.

58. Both Thorkelin transcripts single out these lines, A by leaving three lines blank in his transcript (67, lines 1–12), and Thorkelin himself by writing in the margin that they were deliberately deleted in the manuscript (104a). It is surprising that no modern editor, including the editors of the facsimiles, even mentions Thorkelin’s comment.

59. For a full discussion see BBMS, 264–270.

60. To suggest that the first scribe may have abruptly stopped copying Beowulf because “he had taken ill, and died” (32), Boyle says he may have stopped, not at the end of line 3 of fol. 172v, but after writing most- of moste in the next line. Boyle’s argument is paleographically untenable: the st ligature is distinctive of the second scribe, and is never used by the first. If he gave up copying because he died, we must place his demise at the end of line 3, not at the beginning of line 4.

61. See Levin Schücking, Beowulfs Rückkehr (Halle, 1905).

62. The evidence is somewhat obscured by the alterations someone made to fit numbers XXV–XXVIIII, which led editors to think (e.g. Klaeber, number [XXVIII–XXX] at line 2039) that two fits belonged after fol. 174v19, where the second scribe simply omitted [XXX] for lack of space. See BBMS, 264–270.

63. It is not unprecedented, however, as Guthlac A and B and Genesis A and B both attest.

64. See “The Palimpsest and the New Text of Folio 179,” BBMS, 219–243.

65. In 1066 two Scandinavian warlords who thought they had claims to Cnut’s Anglo-Danish descent invaded England from different directions when Edward the Confessor died. Harold Hardråde of Norway arrived first and died in battle at Stamford Bridge. William of Normandy landed at Hastings a few weeks later and defeated the exhausted English forces.

66. David Dumville seeks to restrict the dating of the script to the opening years of the eleventh century in “Beowulf Come Lately: Some Notes on the Palaeography of the Nowell Codex,” Archiv 225 (1988): 49–63. Ker specifically cautions against this temptation. “The failure of Anglo-Saxon minuscule at the end of the tenth century,” he says, “led to a period of some fifty years, approximately 990–1040, during which there was great variety in the writing of books and charters in England, with some good writing and, especially in the vernacular, much rather poorer imitative writing with no character of its own. In this period great differences are to be seen between the hands of scribes writing at the same time and in the same place, between, for example, the first and the second hand of Beowulf.... These examples illustrate the impossibility of dating script of this period at all closely, and in particular hands which are either dully imitative like our scribe’s or which have gone some way towards Caroline minuscule, like the first hand of Beowulf ....” (The Will of Æthelgifu [Oxford, 1968], 45–46). If one follows Ker, there is no paleographical evidence for excluding the writing of the Beowulf manuscript during or even sometime after the reign of Cnut the Great, 1016–1035.