The Metrical Grammar of Beowulf (Cambridge Studies in Anglo-Saxon England)

CAMBRIDGE STUDIES IN ANGLO-SAXON ENGLAND

THE METRICAL GRAMMAR OF

BEOWULF

CAMBRIDGE STUDIES IN ANGLO-SAXON ENGLAND EDITORS SIMON KEYNES MICHAEL LAPIDGE Editors' preface Cambridge Studies in Anglo-Saxon England is a series of scholarly texts and monographs intended to advance our knowledge of all aspects of the field of Anglo-Saxon studies. The scope of the series, like that of Anglo-Saxon England, its periodical counterpart, embraces original scholarship in various disciplines: literary, historical, archaeological, philological, art-historical, palaeographical, architectural, liturgical and numismatic. It is the intention of the editors to encourage the publication of original scholarship which advances our understanding of the field through interdisciplinary approaches.

Volumes published: 1

Anglo-Saxon Crucifixion Iconography and the Art of the Monastic Revival by BARBARA C. RAW

2 3

The Cult of the Virgin Mary in Anglo-Saxon England by MARY CLAYTON Religion and Literature in Western England, 600-800,

by P A T R I C K SIMS-

WILLIAMS

4

Visible Song: Transitional Literacy in Old English Verse by K A T H E R I N E O'BRIEN O'KEEFFE

5

The Metrical Grammar of Beowulf by C A L V I N B . K E N D A L L

THE METRICAL GRAMMAR OF BEOWULF CALVIN B. KENDALL Professor of English University of Minnesota

The right of the University of Cambridge to print and sell all manner of books was granted by Henry VIII in 1534. The University has printed and published continuously since 1584.

CAMBRIDGE UNIVERSITY PRESS CAMBRIDGE NEW YORK

PORT CHESTER

MELBOURNE

SYDNEY

Published by the Press Syndicate of the University of Cambridge The Pitt Building, Trumpington Street, Cambridge CB2 1RP 40 West 20th Street, New York, NY10011, USA 10 Stamford Road, Oakleigh, Melbourne 3166, Australia © Cambridge University Press 1991 First published 1991 Printed in Great Britain at the University Press, Cambridge British Library cataloguing in publication data

Kendall, Calvin B. The metrical grammar of'Beowulf. (Cambridge Studies in Anglo-Saxon England). 1. Poetry in Old English. Beowulf. I. Title 829.3 Library of Congress cataloguing in publication data

Kendall, Calvin B. The metrical grammar of Beowulf / Calvin B. Kendall, p. cm. — (Cambridge studies in Anglo-Saxon England; 5) Includes bibliographical references and index. ISBN 0 521 39325 6 1. Beowulf-Versification. 2. English language - Old English, c. 450-1100 - Rhythm. 3. English language - Old English, c. 450-1100 - Grammar. I. Title. II. Series. PR1588.K4 1991 829'.3-dc20 90-2646 CIP ISBN 0521 39325 6 hardback

For Kay and Ned

Cambridge Books Online © Cambridge University Press, 2009

Contents

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Preface List of abbreviations List of changes from Klaeber's text The Beowulf-pott and his metrical grammar The alliterative and metrical principles of Beowulf. Kuhn's 'laws' and the transformational rule The three kinds of half-lines: extra-metrical alliteration and type A 3 Displacement Stressed proclitic adjectives: X-positions and the insertion rule Problems with the identification of clause-non-initial half-lines: the proclitic onset Half-lines with internal clause divisions: the transformational rule (revised) The alliterative requirement of unstressed prefixes and the copulative conjunctions The alliterative requirement of prepositions and the proclitic adverbs and instrumental The alliterative requirement of proclitic adjectives and pronouns: the alliterative rule of proclitics Displaced and detached proclitics The three classes of compounds: the alliterative requirement of class I compounds The alliterative behaviour of class II compounds The alliterative behaviour of class III compounds and simplexes vii

ix xiii xv 1 13 27 43 60 74 89 100 116 131 146 159 175 191

Contents 15 16

Metrical typology and the metrical grammar Conclusions

200 206

Appendix: A typology of the metre of Beowulf in relation to the metrical grammar Glossary of technical terms Bibliography Index of alliteration, scansion and metrical clause structure Index of verses specially discussed

219 221 225 231 309

via

Preface

The purpose of this book is to describe the principal features of the metrical grammar of the Beowulf-pott. It is not a new theory of metre or of rhythm. I have my views on these matters, but, since all theories of metre and rhythm must necessarily be grounded in the fullest possible understanding of the principles of the metrical grammar, I have tried to keep them in the background in the hope of making this description as useful as possible to theorists of whatever persuasion. Moreover, I want to emphasize that this is not a description of the common metrical grammar exhibited by all the Anglo-Saxon poets, although many principles of the Beowulf-poet's grammar can surely be generalized. The presentation of the metrical grammar is unavoidably detailed and argumentative. I have felt obliged to offer the evidence on which I base my conclusions, and this often consists of the listing of half-lines and the systematic review of real or apparent exceptions to the generalizations or 'rules' which I have proposed. This procedure, though necessary, may seem tedious. Therefore I have tried to make each chapter as self-contained as possible, so that the reader who is interested in a particular feature of the metrical grammar can move directly to the discussion which he or she finds relevant. When the argument depends upon a principle developed in an earlier chapter, I give a cross reference. For the further ease of the reader, a Glossary of Technical Terms provides a brief definition of a number of terms which I employ throughout the book. 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 IX

Preface

Beowulf — so different from the regular accentual rhythms of so much of later English poetry as well as from the precise quantitative measures of classical verse. That goal now seems to me a will-o'-the-wisp. The application of the Parry-Lord theory of oral-formulaic composition to Anglo-Saxon poetry in the 1950s and 60s opened up new ways of studying the traditional verse patterns. As I got further into my own study, the really fascinating questions for me came to focus on the relation between tradition and the individual. I wanted to understand better how a poem of such formal complexity could be so 'formulaic'. But I found that the interesting question here was not whether Beowulf was an oral poem (a question which probably can never be answered satisfactorily), but rather what the assumption that the Beowulf-poet was steeped in a cultural tradition of oral poetry could tell us about the way the poem works. This led finally to the idea of a metrical grammar which I present in this book. In the course of studying the metrical grammar of Beowulf I have acquired infinite respect both for the irreducible complexities of Old English versification and for the precision with which it was described by Sievers, Pope and Bliss. Time and again I have found that a conclusion which I had laboriously reached was anticipated by their work. If I have managed to see anything new, it is because I have been able to stand on their shoulders. The Bessinger—Smith concordance of The Anglo-Saxon Poetic Records appeared as the book was getting under way — just at the right time for my purposes. Though I never cite it in the notes, it is the basis for numerous assertions that something is, or is not, a feature of the Anglo-Saxon poetic tradition. Three full-length studies of Old English metrics and poetic style appeared after the first draft of this book was complete. I refer to David Hoover's A New Theory of Old English Meter, Geoffrey Russom's Old English Meter and Linguistic Theory, and Daniel Donoghue's Style in Old English Poetry. Though none of them focuses on the metrical grammar of Beowulf per se, they are all relevant to my purposes, and I have incorporated their insights where I could. Donoghue works from premises which are similar to my own; his is a book which I wish had been available when I began — it would have changed the way I analyzed lines with auxiliaries. His use of the word 'style' in his title, however, is a useful reminder that the phenomenon he is studying, the placement of the auxiliary in the verse clause, though it is clearly an aspect of the metrical grammar of the various poets, is one where the metrical grammar provided options. A poet's preference for one

Preface

word order over another (where three orders were available) is a mark of his style. I attend rather to the ways in which the metrical grammar dictated the placement of half-lines in the verse clause and imposed alliterative requirements, in part in order to be able to distinguish the elements of form which derived from the poetic tradition and were carried by the metrical grammar from stylistic choices made by the poet. At different stages of the writing of this book, I have benefited from the criticisms of Professor Dennis Cronan, Mrs Patricia M. Eldred, Mrs Cornelia Haley, Professor J. Lawrence Mitchell, Dr Lawrence Moe, Professor Eric G. Stanley and Professor George T. Wright. I want particularly to acknowledge the invaluable help of Mr Roy White, who was my Research Assistant while the manuscript was undergoing its final revision. His stamp is on nearly every page. My thanks go to them all along with the obligatory acknowledgement that they are not responsible for shortcomings and errors which remain. The Graduate School of the University of Minnesota provided generous support. Several paragraphs in ch. 1 are taken from my article in Anglo-Saxon England', portions of chs."2, 3 and 4 appeared in somewhat different form in Speculum. I am grateful to the editors of those journals for permission to use these materials here.

Cambridge Books Online © Cambridge University Press, 2009

Abbreviations

ASE ASPR BGDSL Bliss, Metre Bosworth-Toller

Anglo-Saxon England The Anglo-Saxon Poetic Records Beitrdge zur Geschkhte der deutschen Sprache und Literatur A. J. Bliss, The Metre of Beowulf J. Bos worth and T. N. Toller, An Anglo-Saxon Dictionary Clark Hall-Meritt J. R. Clark Hall and H. D. Meritt, A Concise Anglo-Saxon Dictionary E. V. K. Dobbie, ecL, Beowulf and Judith Dobbie The Early English Text Society EETS ES English Studies F. Holthausen, ed., Beowulfnebst dem FJ inns butgHolthausen Bruchstuck Journal of English and Germanic Philology JEGP F. Klaeber, ed., Beowulf and The Fight at Finns burg Klaeber H. Kuhn, 'Zur Wortstellung und -betonung im Kuhn Altgermanischen' Modern Philology MP Monumenta Germaniae Historica MGH Epist. Epistolae Neuphilologische Mitteilungen NM The Oxford English Dictionary OED Philological Quarterly PQ J. C. Pope, The Rhythm of Beowulf Pope, Rhythm Proceedings of the British Academy PBA Publications of the Modern Language Association of America PMLA Review of English Studies RES Xlll

List of abbreviations E. von Schaubert, ed., Heyne-Schiickings Beowulf K. Brunner, Altenglische Grammatik nach der angelsdchsischen Grammatik von Eduard Sievers neubearbeitet TPS Transactions of the Philological Society Wrenn—Bolton C. L. Wrenn, ed., Beowulf with The Finnes burg Fragment, rev. W. F. Bolton Wyatt-Chambers A. J. Wyatt, ed., Beowulf with The Finns burg Fragment, rev. R. W. Chambers von Schaubert Sievers—Brunner

Changes from Klaeber's Text

The basis of this study of the metrical grammar of Beowulf"is the text of the poem as printed by Klaeber (3rd ed., 1950). I have resisted the temptation to emend the text to suit my theories, but I have accepted a few, variously recognized, changes, which are here listed. . . . wiggetawum {with a short a) . . . guSgetawum (with a short a) . . . hador (with a short a) se \>e wi5 Brecan wunne on sidne sae (repunctuated) ymb sundflite (compound noun) 507b on breostum laeg, 5 5 2-5 3a Beadohraegl broden golde gegyrwed (repunctuated) . . . geneh[h]ost . . . 794b 811b he fag wiS God (MS) 875a . . . Sigemunde (MS) egl unheoru (MS) 987a 1067l>-68a maenan scolde: Finnes eaferum (repunctuated; MS) anwiggearwe (triple compound) 1247b . . . anraed (with a long initial vowel) 1529a J?aet gebearh feore 1548b-49b wiS ord ond wiS ecge, ingang forstod (repunctuated) ond halig God 1553b-55 geweold wigsigor, witig Drihten, rodera Raedend, hit on ryht gesced (repunctuated) 1575a . . . anraed (with a long initial vowel) 368a 395b 414a 506b-07a

XV

List of changes from Klaeber 1717-18a

eafejmm stepte ofer ealle men, for5 gefremede {repunctuated) ceol up ge)?rang, 1912b-13b lyftgeswenced on lande stod {repunctuated) . . . Hea3abear{d}na 2032b Hea5abear[d]na . . . 2037b 2152b eaforheafodsegn {triple compound) 2525a oferfleon . . . {without a circumflex over the eo) 2636a . . . guSgetawa {with a short a in the penultimate syllable) . . . hyt . . . {with a long y) 2649b anhydig {with a long initial vowel) 2667a Lig ySum for, 2672b-73 born bord wiS rond; byrne ne meahte {relineated\ repunctuated) . . . wundum . . . {MS) 2687a hwate Scildingas {MS) 3005b . . . strude; 3O73b^75a naes he goldhwaete gearwor haefde agendes est {repunctuated; MS)

XVI

1 The Beowulf-poet and his metrical grammar

Beowulf is a poem of infinite variety. If we cannot know precisely what it may have meant to its original, probably monastic, audience, it certainly speaks to modern audiences in many modes and moods. It has been called a heroic epic, a wondertale, an elegy for a bygone age, a mirror for princes, a celebration of pagan Germanic values, an allegory of Christian virtues, an exploration of the moral ambiguities of life and so on. The style and structure of the poem are as rich and complex as the matter which they embody. So intertwined and mutually reinforcing are content, style and structure that J. R. R. Tolkien compared the shape of its narrative to the form of the alliterative long line, l and John Leyerle pointed out the resemblance of its structure to the intricate interlace designs of the Anglo-Saxon art of the seventh and eighth centuries. 2 The Beowulf-poet was fascinated by the ways and means of oral composition. He incorporated several narratives of storytelling into his own narrative. Hrothgar builds the great hall Heorot, and his scop sings a song of creation, the substance of which the poet summarizes (90b-98). This recitation is part of the joyful noise of celebration which angers Grendel and incites his attack on the hall. When Beowulf arrives at Heorot, bringing an offer of help against Grendel, the Danes celebrate again, and the scop from time to time sings with his clear voice (496b—97a). Then, after Beowulf's victory over Grendel, the scop recites the story of Sigemund:3 Hwilum cyninges |?egn, gidda gemyndig,

guma gilphlaeden, 1 3

2 'Monsters and Critics', pp. 36-7. 'Interlace Structure', p. 1. For important commentary on 874b—77 and its significance for understanding the organizing principle of Beowulf\ see Creed, 'Singer as Architect', pp. 133—6.

1

The Metrical Grammar of Beowulf se 3e ealfela ealdgesegena 870 worn gemunde, word o(?er fand soSe gebunden; secg eft ongan siS Beowulfes snyttrum styrian, ond on sped wrecan spel gerade, wordum wrixlan; welhwylc gecwaeS, 875 J?aet he fram Sigemunde[s] secgan hyrde ellendaedum . . . 4 Facility, memory, tradition, allusion, improvisation, alliteration, form and variation - these are apparently what the Beowulf-pott thought mattered in the art of Hrothgar's scop, and it is not difficult to conclude that they are what he valued in his own art. Unfortunately, nothing is known about the Beowulf-pott himself beyond the fact that he was an Englishman 5 who lived sometime after the death of King Hygelac in the first half of the sixth century, 6 which he reports (esp. 2354b-79a), and before the penning of the unique manuscript of the poem at the very beginning of the eleventh century. All we have is his voice, speaking the poem. Still, each of us living with the poem over a period of time and pondering all that has been said about it is bound to flesh out the voice somehow and give this shadowy figure 'a local habitation and a name'. My impressions of the poet and his milieu help form the assumptions that underlie this study of his metrical grammar. It seems only fair to share them with the reader. Here then is an imaginary portrait of the artist. The Beowulf-pott came from an aristocratic family and grew up at a royal court. He learned at an early age the rituals of courtly behaviour. He assimilated the ideals of the warrior class to which his family belonged. At 4

5

6

'From time to time the king's thegn, a man filled with heroic language, who remembered tales, a very great many old sagas, fashioned new phrases truly linked; he then began skilfully to narrate Beowulf's adventure, and fluently to recite, varying his phrases, a fitting tale; he told everything that he had heard said about Sigemund's courageous deeds . . . " References to Beowulf are to Klaeber's text, unless otherwise noted. Other Old English poems are cited according to the texts in ASPR. Translations are my own, unless otherwise noted. Even this 'fact' involves assumptions, namely that the poem is the work of a single author, that the language of the poem is a guide to the poet's native land, and that the poet was a man and not a woman. The identification of Hygelac with the Chlochtlakhus mentioned by Gregory of Tours (Historia Francorum III.3) seems certain.

The Beowulf-^w^ and his metrical grammar

mealtimes and great feasts and during the long winter evenings he listened to the boasts of the young warriors, to the tales of the wise men who could remember what had been done in days of old in the lands from which his tribe had come and to the formal speeches of the king and queen whose power flowed as much from their control over their words as from the king's prowess in war and the queen's generosity. The retainers who could find the right words to tell what men had done in the past and what they should or would do in the future were highly honoured. The king's scop sat at the feet of the king. The words that everyone admired most were recited or sung in the alliterating rhythmical cadences of the heroic lays and praise poems 7 which were the favourite entertainment of the court. On formal occasions the audience was mostly males of fighting age and their elders, along with a few of the higher ranking women in the court. But there were many opportunities for everyone who lived in the vicinity of the court, men, women and children, to hear the same fare. In the remoter villages and farms the entertainment was similar, but the formal standards of the storyellers were not so exacting and their subjects tended to be homelier. Everybody had a chance to participate: there were women's songs and cowherds' songs. The means of expression in the court and in the countryside were always the same — the alliterative metre which their ancestors had brought with them from the continent. Most people could recite, more or less from memory, and in uncertain approximation of the traditional alliterative measures, a short poem in praise of their family's ancestors or the like. A few acquired a reputation for the polish of their performance or for their ability to extemporize a poem in honour of the king or in commemoration of a battle. The very best of the singers from the warrior class were attached to the person of the king. As a young boy, the Beowulf-poet listened to hundreds of recitations. Like the singers of tales whom Milman Parry and Albert Lord observed in Yugoslavia,8 he absorbed the recurrent formulae of the scops and, with practice, learned to fashion similar expressions of his own. He found his voice and made it heard. His genius manifested itself in early adolescence; he was not a late bloomer, like Csedmon. His rank and his talent brought him to the king's attention. But for all his love of the martial spirit and 7

8

On the importance of praise poems, or eulogies, see Opland, Anglo-Saxon Oral Poetry, esp. pp. 172-4. See Lord, Singer of Tales, esp. pp. 3-29.

The Metrical Grammar of Beowulf exploits of his family and people, he had a deeply reflective and questioning nature. He entered a monastery, choosing to become God's thegn rather than the king's. In the monastery, the Beowulf-pott learned to read and write Latin and English. He became acquainted with Vergil's Aeneid and Prudentius's Psychomachia and gained an appreciation of the possibilities of extended epic narrative. He knew by heart the 150 psalms of the Old Testament, but he could not erase the memory of the heroic lays of his countrymen. The brothers of the monastery were delighted to have a gifted singer in their midst. 9 With the abbot's permission he sometimes recited the exploits of their favourite heroes in the refectory at mealtime (the abbot was later to be severely criticized for this indulgence). What they loved was the way he could turn a story familiar to them from childhood into a meditation on the dilemmas of human conduct and the universal dependence of pagans and Christians on God. The abbot thought so highly of his talent and of the exemplary character of his poems that he encouraged him to write out some of them on the monastery's precious parchment. One of these was copied again sometime around AD 1000. This manuscript of Beowulf10 is the only copy of the only one of his poems which has escaped the destruction of the centuries. The truth about the Beowulf-pott may have been very different from the portrait I have painted. I had better say plainly that I do not really think it is likely that he was a monk under Bishop Higbald of Lindisfarne at the end of the eighth century. n Recent work has re-opened the possibility that the poem might after all be a product of the ninth or even tenth century. 12 But I should insist on the probability of certain features. That the Beowulf-pott was a supremely competent oral poet who had grown up in a community where oral composition was intensively practised seems to me certain. Virtually every half-line of the poem is 'formulaic' in the sense that its 9

10 II

12

On the interest of Anglo-Saxon monks and priests in heroic saga, see the comments of Mayr-Harting, The Coming of Christianity, p. 223. London, British Library, Cotton Vitellius A. xv, 129r— 198v. Higbald was the bishop of Lindisfarne who received the famous letter from Alcuin (MGH Epist. 4 (Epistolae Karolini Aevi II), 124) containing the admonition, which he doubtless passed on to the abbot: 'Let the Word of God be heard at the meals of the brethren. There it is proper to hear a reader, not a harper, the sermons of the Fathers, not the songs of the pagans. What has Ingeld to do with Christ? The house is narrow; it cannot hold both of them' (trans. D. W . Robertson, Jr). See the variety of opinions offered by the contributors to Chase, Dating of Beowulf.

The Beowulf-/w# and his metrical grammar cadence (the portion of the verse beginning with the first metrical lift) exhibits a common metrical—syntactical pattern with alliteration marking certain syllables according to precisely describable rules. A literate poet who had not previously acquired fluency in oral composition could probably imitate in a loose way the alliterative style of traditional Germanic verse, but the inner logic which governs virtually every syllable o( Beowulf in accordance with regularities which become apparent only after an exhaustive study of the poem would be beyond his or her reach. And even if it were possible, what would be the point of such minute imitation? That the poet composed Beowulf pen in hand seems to me nearly as certain. 13 The argument for written composition does not stem from the fact that the poem is long (though some scholars question whether there was a tradition of extended oral narrative in Anglo-Saxon England 14) and of high quality. Parry and Lord's study of the living tradition of extended oral compositions in Yugoslavia conclusively demonstrates that oral literature can be both. But Beowulf nz.% nothing in common with the narratives they describe. For one thing, even though Beowulf is a poem about a hero who engages monsters and a dragon in mortal combat, it is extraordinarily lacking in action. The poet spends most of his time circling about a few moments of intense activity. He meditates, and his characters meditate, on the meaning of the events which occur or which have occurred or which are likely to occur. I do not want to claim that an oral poet could not conceivably create such a poem in an extemporaneous performance before an audience, but the likelihood of it seems remote. Nor is the likelihood much increased by supposing that the artificial circumstances in which a poem by an illiterate oral poet in Anglo-Saxon England would have been committed to parchment — namely, by dictation to a scribe — would have altered the conditions of the performance. A dictated oral poem might differ considerably in degree from a typical performance; it might be longer, or more carefully thought out, or less spontaneous, but there is no reason to believe that it would differ appreciably in kind. The assumption that the Beowulf-poet was a monk who knew how to read and write, and who, being an accomplished oral poet, could therefore dictate a poem to himself at a leisurely pace over a period of days or weeks, now rapidly 13

14

For general arguments against the assumption that formulae in Old English poetry necessarily imply oral composition, see Benson, 'Literary Character', pp. 334—41. See esp., Campbell, 'Old English Epic Style', p. 13.

The Metrical Grammar of Beowulf composing an extended 'run', now pausing to reflect on what he had written, or proceeding meditatively a line at a time, seems to me the one which best explains the kind of poem we have. For another thing, even though the language of Beowulf is formulaic to the core, the style and structure of the poem seem thoroughly literary, much more so in my opinion than another major medieval poem which is also demonstrably formulaic and which has been described as the 'prototype of a transitional text', 15 the Chanson de Roland. There is in Beowulf more enjambment than might be expected in an oral poem. The poet frequently separates a subject from its verb with a parenthetical clause which may extend for two or three half-lines (e.g., 2237b—39 interrupted by 2238t>-39a; 2258-60a interrupted by 2258b-59). Elliptical constructions require the reader to hold a controlling auxiliary in mind for several lines at a time (e.g., lines 1855-61). John Niles has convincingly argued that many of the major and minor structural patterns of Beowulf ate the product of'ring composition'. He acknowledges that the techniques of ring composition 'would be useful to an oral poet or performer', but he points out that 'oral texts taken from the field have not been shown to exhibit complex ring structures comparable to those evident in the Iliad and Beowulf. 'Perhaps these structures exist', he goes on, 'but the necessary field work and analysis have not been done.' 16 The art of Beowulf is more grammatical than rhetorical. Constantly shifting patterns of alliteration and metre enthral the ear of the audience and the eye of the reader. At first the patterns seem random, almost formless, until one realizes that they are predictably related to, and therefore a function of, the syntax of the verse clauses. But of course this is only a partial truth. Syntax is just as much a function of alliteration and metre as the other way around. The art of the poem must be sought in the delicate interplay between the differing constraints of alliteration, metre and syntax. The Beowulf-pott gives a bravura performance in the art of alliteration in the opening lines of the poem: Hwaet, we Gar-Dena in geardagum, J?eodcyninga J?rym gefrunon, 15 16

Curschmann, 'Oral Poetry', p. 47. Beowulf, pp. 152—62; quotations from p. 160.

The Beowulf-/w# and his metrical grammar hu 5a ae)?elingas ellen fremedon! Oft Scyld Scefing scea(?ena (?reatum, 5 monegum maeg(?um meodosetla ofteah, egsode eorl[as], sySSan serest wearS feasceaft funden; he |?aes frofre gebad, weox under wolcnum weorSmyndum (?ah, o3 \>&t him aeghwylc ymbsittendra 10 ofer hronrade hyran scolde, gomban gyldan; J?aet wses god cyning! 17

The cross alliteration of the first line (G-D : G-D) is followed by alliteration linking the final lifts of lines 2 and 3 (F/F). Then a run of double alliteration marks the first half of six of the next eight lines. No two of the verses with double alliteration are just alike. The double alliteration of Oft Scyld Scefing (type C1) is prompted by the use of a patronymic after a proper name. Monegum mcegpum (type Al), with resolved stress in the first lift, has mandatory double alliteration in consequence of the insertion of the stressed proclitic adjective monegum before a stressed element. The primary alliteration in egsode eorlas (type A3) is carried by the initial diphthong of the noun eorlas, supplemented by the extra-metrical alliteration of the verb egsode. The F alliteration in feasceaft funden (type A2a) which builds to the consonant cluster FR in the second half of the line harks back to the final lifts of lines 2 and 3 (consonant cluster FR). Alliteration on funden is made obligatory by the fully stressed compound feasceaft, the second element of which picks up the thread of the SC alliteration that began with line 4 and concludes with the last lift of line 10. Weox under wolcnum (type A3) is another verse with primary alliteration on the stressed syllable of a noun {wolcnum) and extra-metrical alliteration on a finite verb (weox). Single alliteration marks the first halves of lines 9 and 10 (although an interesting pattern of incidental vocalic and H alliteration may be noticed). In gomban gyldan (type Al), the poet returns to double alliteration and to the initial alliteration of the poem, which leads him to the exclamatory conclusion pcet 17

'Listen! We have heard of the glory of the Spear-Danes, of the people-kings, in days of yore, how the nobles performed deeds of courage. Often Scyld Scefing deprived of their mead-seats troops of enemies, many tribes, he terrified the earls, after he had been found destitute in the beginning; he experienced consolation for that, he grew beneath the skies, he prospered in honours, until each of the neighbouring tribes over the whale-road had to obey him, to give him tribute; that was a good king!'

7

The Metrical Grammar of Beowulf wees godcyningl. Since we know from the rules of the metrical grammar that the double alliteration which links gyldan to gomban is optional, the effect must be one which the poet intended. One of the reasons for undertaking an investigation of the poet's metrical grammar is to enable us to distinguish alliteration which is dictated by the rules of the grammar from alliteration which is chosen according to the rules of the grammar. Only then can we compare the formal elements of the poem with its content in a meaningful way. The interpretation of data of this kind tends to be subjective. All Old English poetry is alliterative, and patterns of alliteration linking successive lines can be discovered in any passage of comparable length from any Old English poem. Most sensitive readers of Shakespeare's sonnets agree that their innumerable patterns of alliteration, assonance, consonance and rhyme are an essential part of the aesthetic experience and meaning of the poems. Of course, analysis of any paragraph in the morning's paper will turn up similar patterns to which no one will attach the slightest importance. In the end it comes down to the belief which most of us share that all the formal properties of a poem that moves us are somehow significant and find their echo at the level of the narrative. These alliterations of the opening lines of Beowulf are the poet's way of showing his mastery of the form which will carry the story of Beowulf through the 3182 lines of the poem. The astonishing fact is that, without abandoning his alliterative form, he modulates in the final lines of the poem into what later generations would think of as rhyme. He tells how the Geats raise a monument for their dead leader — a great funeral mound on a headland. And then twelve noblemen's sons ride about the mound, lamenting their chief: Pa ymbe hlaew riodan hildedeore, 3170 se|?elinga beam, ealra twelfe, woldon care cwi&an ond kyning m&nan, wordgyd wrecan, ond ymb wer sprecan; eahtodan eorlscipe ond his ellenweorc duguSum demdon, — swa hit gedefe biS, 3175 baet mon his winedryhten wordum herge, ferhSum freoge, )?onne he forS scile oflichamtf» laeded weor&zw. Swa begnornodon Geata leode hlafordes hryre, heorSgeneatas; 3180 cwsedon f>aet he waere wyruldcyninga 8

The Beowulf-^/ and his metrical grammar manna mildust leodum ViQost

ond montSw&rust, ond lofgeorn&tf.18

Only readers thoroughly familiar with the style of the poem will recognize how extraordinary this is. Isolated cases of terminal identity like 3171 and 3177 can be found elsewhere in the poem, 19 and occasionally internal rhyme within a half-line, 20 but nothing comparable to this accumulation of rhyme within the space of fourteen lines, including the full rhyme wrecanlsprecan and the quadruple repetition of the suffixes -ustl-ost in the last two lines which brings the poem to a full stop. From the perspective of the Beowulf-poet what we see as rhyme would probably have been considered the happy consequence of a final intensification or heightening of the appositive style, as Fred C. Robinson has so aptly called it, 2 1 that characterizes the poem. And still the alliterative form carries on in its endless variation. To notice only the three final lines: in 3180 there is transverse alliteration — C-W : W-C — which is made possible by incidental alliteration on the finite verb cwcedon. And then the double alliterations in the first halves of the last two lines, alliterations which were not mandated by the rules of the poet's metrical grammar, but were optionally imposed, 22 draw the alliterative patterning to its stately close. Given the Beowulj-poet's attention to the form of his beginning and ending, it will not be superfluous to point out that the last lift of the last line is the first syllable of -geornost with a return to the first alliteration (G) of the poem. The poet's voice was modulated by his metrical grammar, the frame18

19 20 21

22

'Then the battle-brave men, the sons of the nobles, twelve in all, rode around the mound. They wished to bewail their sorrow and lament their king, to deliver a eulogy and speak of the man. They esteemed his nobility and praised highly his courageous deeds. It is fitting that one praise his friendly lord in this way with words, love him in one's heart, when he must be led forth from the body. So the people of the Geats, the hearth-companions, lamented the fall of their lord; they said that he was, of the kings of the world, the mildest of men and the gentlest, the kindest to his people and the most eager for praise.' Even in consecutive lines, e.g., 7 6 9 - 7 0 . E . g . , 279a, 656a, 1422a, I 6 l l a a n d 1864a; 1008b. See his Beowulf and the Appositive Style. It may not be superfluous to point out that the widespread use of rhyme, especially internal leonine rhyme, in late medieval Latin poetry owed much to the encouragement by such scholars as Bede of the use of the rhetorical figures of homoeoteleuton (like-sounding endings in parallel constructions) and homoeoptoton (same case endings in parallel constructions). However, manna mildust (3181a) is a traditional formula: cf. Exodus 550a, and the O l d H i g h G e r m a n 'Wessobrunn Prayer', 8a.

The Metrical Grammar of Beowulf work of the poetic language he learned in becoming an oral poet. The metrical grammar of the Beowulf-poet is the complete set of syntactical, alliterative and metrical rules and conventions that go into forming the lines that make up the poem as we know it.23 It might be supposed that the description of any metrical grammar would simply require the addition of the rules of alliteration and metre to the grammar of the poet's native language, and that a rigid metre would occasionally introduce distortions which would account for differences between everyday language and the language of his or her poetry. But this would be a misleading way to look at poetry which has its roots in an oral tradition. The language of poetry which Anglo-Saxon oral poets internalized during their years of apprenticeship in the poetic traditions of their culture contained grammatical, as well as metrical and alliterative, features which were specific to it. The 'metre' of an Old English poem is only one function of a set of regularities that make it something we call verse rather than prose. Separately these regularities may be described as 'rules'; taken as a group, the rules comprise a metrical grammar. No two scops' metrical grammars would have been exactly alike; in addition to individual differences, there must have been regional and dialectal variations, although the poetic tradition ensured remarkable uniformity over a wide area and a considerable period of time, and only at the end of the Old English period, with let us say The Battle of Maldon, are significant changes manifest. Further investigation is needed to determine which rules of the metrical grammar of Beowulf can be generalized. 24 Daniel Donoghue has called attention to a number of specific ways in which the metrical grammar of The Metres of Boethius differs from that of Beowulf. He tentatively attributes this difference to the survival in them of the verse style of the early Germanic lays, which Campbell has described. 25 Many of the rules of the Anglo-Saxon scop's metrical grammar corresponded nearly or exactly to the grammatical rules of his native language. A 23

I have developed the notion of a 'metrical g r a m m a r ' in part from the remarks of Lord on the g r a m m a r of poetry, in Singer of Tales, p p . 35—6. See further, Bliss, 'Auxiliary a n d Verbal', p . 157, K e l l o g g , 'South G e r m a n i c Oral T r a d i t i o n ' , esp. p . 6 7 a n d Kendall, 'Prefix un- , p p . 3 9 - 4 1 . It is i m p o r t a n t to p o i n t out t h a t I intend 'metrical g r a m m a r ' t o refer to t h e rules of alliteration as well as those of syntax and m e t r e .

24

O r t o n , 'Verbal A p p o s i t i o n ' , p . 1 5 8 , calls for investigation of this k i n d . See below,

25

Donoghue, 'Word Order', pp. 190-2; Campbell, 'Old English Epic Style', pp. 13-26.

ch. 16. Donoghue modifies his conclusions somewhat in Style, pp. 100-1.

10

The Beowulf-/we/ and his metrical grammar

few differed markedly. Some, and in particular those governing alliteration and stress, were unique. From these last we abstract our notions of metre as it is commonly understood. The scop internalized the rules of metre along with the other rules of his metrical grammar in the course of learning his craft. The chances are that he was never conscious of metre as a separate entity, especially if he had not received schooling in the metres of Latin. If we could return to the seventh century and Abbess Hild's monastery at Whitby for the purpose of quizzing Caedmon, we would probably find him as tongue-tied in the face of questions about his metre as he would certainly be about the way in which he constructed his sentences. Parry and Lord, working with illiterate singers of tales in Yugoslavia, discovered that they had no conception of so basic a notion as a line of poetry. 26 For this reason, it is unlikely that the typical unlettered scop ever consciously bent the rules of his metre for poetic effect. His poetic utterance was governed by the metrical grammar he had acquired. To speak poetically was to speak metrically, because metre was an integral part of that grammar. Of course he might rnis-speak in rapid oral composition and thereby produce a metrically anomalous line. This is quite different from playing with the expectations set up by metre, as a modern poet might do. Even the Beowulf-poet, who was almost certainly literate, depended on the resources of his metrical grammar rather than on self-consciously imposed deviations from it for special effects. Consider, for example, a line in which he put the alliterative stress on two demonstratives: On \>abm daege

J?ysses lifes

(197)

Although this looks like a place where he chose to give rhetorical emphasis to normally unaccented words, it turns out that the the 'licence' was provided by a rule of his metrical grammar. 27 Line 197 is a formula in Beowulf (see 790 and 806) which belongs to the poetic tradition. 28 Similar Violations' fall into fairly predictable categories, and all but a few can be accounted for within the rules of the metrical grammar. The view, which has had some currency, that Old English metre was so 26 Lord, Singer of Tales, p. 25. 27 28

For the insertion rule which governs this usage, see ch. 5. For on pczm dage w i t h alliterative stress on pcem, see Christ 1096b and 137 l b ; for pysses lifes w i t h alliterative stress on pysses, see Genesis 1120b, 1 6 0 0 b , 2 4 5 2 b , Guthlac 7 4 b , Phoenix 1 5 1 b , The Gifts of Men 19b and Solomon and Saturn 2 4 2 b . Cf. t h e formulaic system . . . peos I pas woruld with alliterative stress on peos I pas, Genesis 1126b, Christ 1583b, Guthlac 125b, The Phoenix 501b, The Wanderer 58b and Deor 31b.

11

The Metrical Grammar of Beowulf accommodating that it routinely permitted similar half-lines to be variously scanned to suit the exigencies of alliteration or the rhythmical whims of the poet is simply untenable. 29 By and large the scansion and alliteration of any given half-line of almost any Old English poem can be determined from its own grammatical properties without reference to the context in which it appears. I do not mean to suggest that scops could not vary their lines to suit their contextual purposes. The Beowulf-poet, at least, was enormously skilled at attaining variety in the shape and sound of his lines. But nearly all the flexibility he needed was available in the formal principles of his metrical grammar, which is the subject of the following chapters of this book. 29

The view can be illustrated passim in Daunt, 'Old English Verse', and Baum, 'Meter of Beowulf.

12

The alliterative and metrical principles of Beowulf: Kuhn's 'laws' and the transformational rule

A description of the metrical grammar of Beowulf may begin by taking notice of the most striking formal feature of Old English poetry — alliteration. In this poetry alliteration is not, as it would later become in English, an incidental adornment to be applied or not at the whim or will of the poet. Nor is it, like rhyme, a repeated pattern of sound affecting whatever part of speech happens to occupy a fixed position, usually the last, in a line. Alliteration marks the stressed syllables of syntactically prominent words in predictable ways. The poet who wrote, or sang, secga Ipreate, was not free to alliterate on S or P. Alliteration on S was required: secga Create.

Swa hit siSSan gelamp

(Elene 271)

Compare: Pa gesamnodon

secga (?reate

{Andreas 1636)

On the other hand, Caedmon's metrical grammar required that hergan rather than scylun alliterate when he sang: Nu scylun hergan hefaenricaes uard (Ccedmoris Hymn, Northumbrian Version 1) So did the metrical grammar of the poet who wrote, with the word order reversed: Nu we herian sceolan

her for life {The Seasons for Fasting 39)

The rules which assign alliteration are a reflex of the metrical grammar, or grammars, of the Old English poets. The alliterative long lines of Beowulfr, like those of all other poems which survive from the Old English period, are made up of two half-lines linked 13

The Metrical Grammar of Beowulf together by the alliteration of their chief syllables. The first half-line will be referred to as the a-verse, the second as the b-verse. The basic alliterative rule of the metrical grammar may be simply stated: at least one and no more than two syllables in the a-verse must alliterate with one and no more than one syllable in the b-verse. Thus the alliterative possibilities are two (where alliteration is represented by a lower-case a): aa : a and a : a. It remains to be specified which syllables will be marked by alliteration. The rules which assign metrical stress are likewise a reflex of the metrical grammar. The notion of 'metre' in Old English poetry is, it must be acknowledged at once, problematic. No single metrical template exists which can describe the alliterative long line. The nearest approach to one is some version of the two-stress theory championed in the nineteenth century by Max Rieger and Eduard Sievers according to which each half-line displays two major metrical stresses (or 'lifts') with two (or sometimes three) weaker metrical positions distributed in variable, but precisely describable, relation to them. But even this theory cannot be objectively verified, because there are whole classes of verses with only one clearly ascertainable stress, where the assignment of the second stress has to be imposed arbitrarily in order to satisfy the theory. The least controversial, though necessarily incomplete, description of the metrical pattern of the long line might therefore be: N : N, where N stands for the first metrical stress in each half-line. The first metrical stress falls on the stressed syllable of the leftmost 'stressed element' or its equivalent in the half-line. l However, all b-verses and many a-verses have at least one more syllable after this N with some potential for stress. 2 If N is also used to represent this second 'stressed' syllable or lift, there are two possible metrical patterns in the long line: N N : N N and N : NN. (Many half-lines have three prominent syllables and a complete metrical description must take 1

2

The first metrical stress cannot fall on a syllable in an X-position (see ch. 5 for definition and discussion), with one exception. In practical terms this means that the initial N will always be followed by at least two syllables, with the exception of one metrical contour in the a-verse (type A3) which may be followed by only one syllable. It has never been satisfactorily determined whether or not all b-verses have at least two stresses. Bliss, Metre, esp. §§ 9-11 and 67-75, is a strong advocate of the position that many b-verses have only one stress. The problem centres on the question of whether secondary or tertiary stress is found on certain syllables. It is not my intention to take a stand on this issue here. The essential point is that a second lift syllable can be specified. In the Index of Alliteration, Scansion and Metrical Clause Structure I assume stress.

14

Alliterative and metrical principles

this fact into account. However in almost every case one of these syllables is unambiguously subordinate with respect to the other two. In the present discussion N is to be understood as referring to the most prominent or to the most prominent followed by the second most prominent syllable in the half-line.) In the Beowulf-poet's metrical grammar, the fundamental alliterative principle which relates the patterns of alliteration to the patterns of metre may be stated as follows: In any half-line prior to the insertion of the half-line into the verse clause the stressed syllable of the leftmost 'stressed element' (or of the leftmost sentence particle marked for metrical stress by the transformational rule) always bears the first metrical stress, which is marked by alliteration? Natural

stressed elements include initially stressed compounds with two fully meaningful elements; nominals (nouns and attributive adjectives, including participles used attributively); initially stressed compound verbs and adverbs, such as dndswarian and unforhte\ most quasi-compound phrases4; and a few other classes and specially marked words (e.g., forms beginning with ceg-, the pronoun self, probably the pronoun gehwceper* etc.). Natural stressed elements always take metrical stress. Other parts of speech and some phrases may become the equivalent of stressed elements by displacement, emphatic stress or insertion into alliterating position in the b-verse, but they must be distinguished from natural stressed elements. 6 3

4 5

6

I take this wording with some modification from Kendall, 'Prefix un- , pp. 43—4, and n. 17. Sentence particles which receive metrical stress by means of the transformational rule do not become stressed elements; for discussion, see below, pp. 24—6 and ch. 7. See below, chs. 12 and 13. In the corpus of Old English poetry the pronoun gehwceper occurs fourteen times, always with metrical stress. In the a-verse it always alliterates: three times in the first lift and once in a hypermetric verse where it is the third alliterating element (The Rune Poem 28a). In on gehwcepere hand (The Battle of Maldon 112a), it leads a noun, which however is in the final X-position (see ch. 5). In the b-verse it alliterates five times and falls in the final non-alliterating position five times. On the other hand, except for the verse in The Battle of Maldon, when gehwceper is in the first lift, the following word, if any, is a pronoun or adverb, and the verse is clause-initial (see ch. 3). In short, it is difficult to be sure whether gehwaelper had become a stressed element. In Beowulf, the half-line ond gehwceder o&rum (217 la) is difficult to explain except on the assumption that gehwteder is a stressed element (scanning it by the transformational rule (see below) would result in a type A 3 with alliterating stress on odrum). I mark it so, throughout. See Sievers, Altgermanische Metrik, pp. 22—46; Heusler, Deutsche Versgeschichte, pp. 105-13; and Slay, 'Some Aspects', p. 1. For internal displacement, see below, ch. 4; for emphatic stress, see below, ch. 5.

15

The Metrical Grammar of Beowulf The syllables marked by the alliterative principle in the two half-lines provide the alliterative and metrical link between them. When the alliterative principle marks a syllable which corresponds to the first alliterating position in the two possible alliterative patterns in the a-verse, the alliterative and metrical systems intersect to generate three patterns of metrical alliteration: (1) (2) (3)

AA : AN AN : AN A : AN 7

When the alliterative principle marks a syllable which corresponds to the second alliterating position in the a-verse, two more patterns of metrical alliteration are generated: (4) (5)

aAN : AN aA : AN

In these two patterns the first alliterating position remains outside the metrical contour of the a-verse.8 If we look at a half-line in isolation, the alliteration may not be apparent (it is never apparent in a b-verse; it is apparent in an a-verse only when the a-verse displays 'double alliteration'). Nevertheless, even in isolation, the alliterating syllable can be certainly identified in almost all cases. In 1933 Hans Kuhn formulated two rules concerning word order and stress in the oldest poetry of the Germanic languages which have a bearing on these alliterative patterns. 9 In order to do so, he found it convenient to classify words (and some prefixes) into three categories on the basis of stress. He distinguished words which are independently stressed within the structure of the clause from unstressed elements which attach themselves to a single word or part of a clause (i.e., 'proclitics'). Then he subdivided independently stressed words into 'stressed elements', i.e., 7

8

9

That is, when the first 'a' on either side of the colon in the two alliterative patterns aa : a and a : a corresponds to the first N on either side of the colon in the two metrical patterns NN : NN and N : NN and the second 'a' in thefirstalliterative pattern corresponds to the second N in thefirstmetrical pattern, the coincidence of a and N being represented by A. See the discussion of extra-metrical alliteration in ch. 3 and of external displacement in ch. 4 for an explanation for these two patterns. Kuhn, pp. 1—109. Important discussions or applications of Kuhn's discoveries will be found in Bliss, Metre; Campbell, 'Verse Influences', pp. 93—8; Cosmos, 'Kuhn's Law', pp. 306-28; and Slay, 'Some Aspects', pp. 1-14.

16

Alliterative and metrical principles words of heavy stress within the clause, and 'sentence particles', i.e., words of a presumed lesser degree of stress. 10 His three categories therefore are: (1) stressed elements; (2) sentence particles; and (3) proclitics. (1) Stressed elements are the words described above. Under certain circumstances, as will become clear in the following chapters, sentence particles and proclitics become stressed elements. (2) Sentence particles include in Kuhn's words 'substantive pronouns, many adverbs and finite verbs, conjunctions, to some extent also adjectival pronouns, occasionally infinitives and predicate nouns, and possibly also vocatives'. n Kuhn's treatment of verbs, especially his separation of lexical (non-auxiliary) verbs from the class of nouns and adjectives, has been criticized on linguistic grounds. 12 There is no doubt that lexical verbs behave differently from auxiliary verbs and that they are more likely to be heavily stressed. Nevertheless, Kuhn's decision to treat most verbs as sentence particles rather than as stressed elements proves to have been the right one for the purpose of analysing the metrical grammar. I use his classification with only slight changes. Adverbial conjunctions (sentence particles) must be distinguished from copulative conjunctions. I include infinitives among the sentence particles. A few quasi-compound phrases are sentence particles rather than stressed elements. (3) Proclitics are the definite articles, adjectival pronouns, some adverbs, prepositions, copulative conjunctions and the unstressed prefixes. 13 Proclitics therefore include both separate word classes and what we now regard as bound morphemes - the unstressed prefixes - which attach to stressed elements and sentence particles. In the manuscripts themselves a prefix is no more likely to be attached to the word to which we think it belongs than is any other proclitic. 14 Kuhn called his first rule the law of Germanic sentence particles: 'Sentence particles are grouped in the first dip of the clause. The first dip precedes 10

Kuhn, pp. 4-5.

11

'In den altgerm. mundarten waren satzpartikeln substant. pronomina, viele adverbien und finite verben, bindeworter, zum teil auch adject, pronomina, gelegentlich infinite verbformen und pradicatsnomina, vielleicht auch vocative' (Kuhn, p. 5). It would be more precise to define categories (1) and (2) as, respectively, 'stressed elements minus their weakly stressed prefixes, if any' and 'sentence particles minus their weakly stressed prefixes, if any'. See Cosmos, 'Kuhn's Law', p. 308. See Kuhn, p. 43. Kuhn's term for 'proclitics' is sentence-part particles {satzteilpartikeln). Russom, Old English Meter, pp. 8-9, offers reasons for treating the unstressed prefixes in Old English as function words.

12 13 14

17

The Metrical Grammar of Beowulf either the first or the second stressed word in the clause.' 15 A dip is a string of syllables which does not contain a stressed element. The first dip of a clause usually coincides with a weak metrical position in a half-line. (A 'weak metrical position' is usually referred to as a 'dip'. I use the more cumbersome phrase to avoid confusion with the notion of a clausal dip.) Exceptionally, however, as we shall see, the dip includes elements which are metrically stressed. Sentence particles in the dip are presumed to be weakly stressed relative to the first or second stressed element. When a sentence particle is moved out of the first dip of the clause, or displaced, it takes metrical stress or exhibits alliteration or both. l 6 The normal 17 word order which Kuhn's first law describes can be illustrated by two independent clauses, each of which is just one half-line in length (the accented syllable of the first or the first and second stressed element of the clause is numbered; sentence particles are italicized): Eode pa to setle.

Pser waes symbla cyst

Hream weard in Heorote;

heo under heolfre genam

(1232) (1302)

The initial clausal dip consists of all syllables preceding the first stressed element {eode pa to) or between the first and second stressed elements (weard in). In both half-lines the sentence particle(s) are grouped in the first dip of the clause which occurs immediately before either the first or the second stressed element. Frequently however, one of the sentence particles is displaced. Here is a clause consisting of two half-lines: sipdan goldsele

Grendel warode

(1253)

The finite verb warode has been displaced from the first dip of the clause. As a result of displacement it takes metrical stress and can be counted among the stressed elements. 15

16

17

'Die satzpartikeln stehen in der ersten senkung des satzes, in der proklise entweder zu seinem ersten oder zweiten betonten worte' (Kuhn, p. 8). Kuhn assumes that all displaced sentence particles become stressed elements, because he takes alliteration to be evidence of metrical stress. See, e.g., his pp. 13-14. I prefer to leave open the question of whether additional stress always accompanies displacement. For further discussion see below, ch. 4. 'Normal' is a loaded term; I use it here and elsewhere to refer to a presumed simplest or least marked linguistic state. The 'normal' state is not necessarily the statistically most frequent one, nor is there anything necessarily 'abnormal' about its opposite.

18

Alliterative and metrical principles

Kuhn called his second rule the law of Germanic clause openings: 'A clause upbeat must contain sentence particles.' 18 An upbeat is an initial dip which comes before the first stressed element. A clause need not have an upbeat (as 1302a cited above does not). But if it does, the upbeat will contain at least one sentence particle (and therefore, by Kuhn's first law, any and all other undisplaced sentence particles as well). If the clause lacks an upbeat, sentence particles may or may not be found in the first dip. These possibilities can be illustrated by two examples. First, a five half-line clause with an upbeat: Se was HroJ?gare hselef>a leofost on gesiSes had be saem tweonum, rice randwiga . . . 19

(1296-8a)

The upbeat, se wees, contains sentence particles, which conforms to the second law. (And since it contains all the sentence particles, it also satisfies the first law.) Second, a three half-line clause without an upbeat: ides aglaecwif

Grendles modor, yrm|?e gemunde . . . 20

(1258b-59)

Here the first dip of the clause follows the first stressed element. It is occupied by the unstressed second syllable of Grendles. The only sentence particle in the clause, the verb gemunde, has been displaced to a stressed position. Prosodists have given less attention to Kuhn's second law than they have to the first. The second law has crucial implications for the study of the metrical grammar. From the last two examples, it should be clear that the second law predicts that a half-line like on gesides had, which has an initial weak metrical position consisting exclusively of the proclitics on and ge-, will not be found at the head of a clause. In conjunction with the first law it predicts that a half-line like se wees Hropgare, which has an initial weak metrical position with sentence particles in it, must be clause-initial. It puts no restrictions on a half-line like Grendles modor. And in fact, Grendles modor, which here stands at the head of its clause, appears elsewhere in final position (1282a). I take up the implications of these facts in ch. 3. 18 19

' I m satzauftakt miissen satzpartikeln stehen' ( K u h n , p . 4 3 ) . ' H e was to H r o t h g a r the clearest of warriors in the rank of a retainer between the seas, a powerful shield-warrior

20

'Grendel's m o t h e r , a lady m o n s t e r - w o m a n , r e m e m b e r e d her misery . . . "

19

The Metrical Grammar of Beowulf There is something odd about Kuhn's laws. Kuhn based his study primarily on the oldest monuments of Germanic poetry (Anglo-Saxon, Old Saxon and Old Norse) and on the metrical principles set forth by Sievers and Heusler. His laws seem not to apply to old Germanic prose, but they are not rules of metre either. Instead they describe certain features of the language of poetry which belong to what I am calling the metrical grammar; for present purposes, to the metrical grammar of Beowulf. Whether these were archaic features of primitive Germanic which the traditional language of poetry preserved after they had more or less disappeared from the language of everyday speech or whether, on the contrary, they result from a tightening up in poetry of tendencies inherent in everyday language is unclear. 21 In order to see what Kuhn's laws reveal about the metrical grammar of Beowulf, it is important to keep the distinction between 'metre' and 'metrical grammar' in mind. The examples I have given have been unambiguous. Let us consider several that are less straightforward. Take this three half-line clause: Pa gyt hie him asetton heah ofer heafod . . . 22

segen gyldenne (47-8a)

Apart from metre, the clause as analysed here conforms to both laws. It has an upbeat (47a) which contains sentence particles (second law). All the sentence particles are grouped in the first dip of the clause (first law). Metre however requires that the second, alliterating syllable of asetton occupy a metrical lift. That would seem to add asetton to the stressed elements, and we could re-analyse the clause this way: Pa gyt hie him asetton heah ofer heafod . . .

segen gyldenne

It would still conform to both laws. Now take a similar four half-line clause: Ymbeode pa ides Helminga dugu(?e ond geogoj>e dad aeghwylcne . . . 23 21

22 23

(620-1)

Lehmann and Tabusa, Alliterations, p . 3 , describe t h e role of the traditional formulae in preserving archaic linguistic features w i t h admirable precision a n d caution. 'In addition they set a golden standard high over his head 'Then the lady of the Helmings went around to each group of the old and the young warriors . . . '

20

Alliterative and metrical principles Again the clause conforms to both laws, but again alliteration shows that the chief syllable of the verb occupies a lift. Kymbeode is one of the stressed elements of the clause, then pa must be stressed as well since it is now displaced from the first dip which is occupied by the prefix ymb-\ Ymbeode pa This re-analysis produces a regular metrical contour for 620a (type B) 24 and conforms to Kuhn's first law. However the clause now has an upbeat (ymb-) that lacks a sentence particle, in apparent violation of the second law. Finite verbs frequently appear before stressed elements in the same half-line. If the half-line opens a clause the verb will be in the proper position for a sentence particle, as in Gemundefia5a are,

(?e he him aer forgeaf

(2606)

But more often than not the verb shares in the alliteration, which is usually taken as evidence that it has been assimilated to the stressed elements of the clause. Consider the half-line clause: stone da aefter stane,

stearcheort onfand

(2288)

Alliteration suggests that stone might be a stressed element: stone da aefter stane This generates what appears to be a regular metrical contour (type A). The only remaining sentence particle (3d) is properly in the first dip of the clause, which no longer has an upbeat. There is no violation of either of Kuhn's laws. But consider: gemyne maerjx),

maegenellen cyS

(659)

Here, if we allow ourselves to be guided by alliteration and take gemyne to be a stressed element, re-analysis produces: gemyne maerj?o This too is a regular metrical contour (type C 1 , with resolution of the first stress). Since there are no remaining sentence particles, Kuhn's first law holds. However the clause now has an upbeat (ge-), which does not contain 24

By 'regular metrical contour' I mean a pattern of metrical stress conforming to any of the simplest realizations of the basic Sievers types. These are: A /(x)x/x A3 (. . , ) x x / x 1 3 D " //\x B (. . . ) x / ( x ) x / D4 //(x)x\ C (. . . ) x / / x E /\(x)x/

21

The Metrical Grammar of Beowulf a sentence particle. Kuhn's second law is not satisfied. Finally, take the four half-line clause: Gewat pa ofer waegholm

winde gefysed

fugle gelicost . . , 25

flota famiheals

(217-18)

Re-analysis on the basis of alliteration gives: Gewat pa ofer waegholm Several problems result. The only way to make the half-line conform to a regular metrical contour is to claim that there is an extra, unmeasured syllable before the first lift (anacrusis). Furthermore, either the particle pa is displaced (but there is no metrical precedent for giving it stress 26), or it occupies the second dip of the clause in violation of Kuhn's first law. The first dip, which is an upbeat, does not contain a sentence particle, in violation of the second law. Sometimes a sentence particle must receive metrical stress in order for the half-line to be scanned at all. Two examples appear above (47a and 620a). This is often the case when a verb stands at the head of a b-verse half-line. Take, for example, the two half-line clause: Gyrede him Beowulf eorlgewsedum . . .

(l44lb-42a)

The only possible metrical contour for line 144 lb is type A with alliteration on G — a scansion confirmed by line 144 la (gryrelicne gist). If gyrede is counted among the stressed elements, the half-line must be re-analysed: Gyrede him Beowulf The re-analysis does not result in a violation of either of Kuhn's laws. And in fact it happens that mandatory alliteration on a sentence particle in the b-verse never violates either law in Beowulf. How are these facts to be interpreted? If a verb in the first dip of a verse clause is stressed for the sake of the metre, should the distribution of the remaining sentence particles and proclitics still conform to both laws? With respect to the first law, Bliss has argued that it must. He claims that 25

26

' T h e n t h e foamy-necked ship, propelled by t h e w i n d , departed over t h e waving sea, most like a bird . . . " As Bliss, Metre, § 2 0 , points o u t .

22

Alliterative and metrical principles

the alliteration on the verb in Gewat pa ofer wcegholm (217a) is 'nonfunctional' (that is, not stressed for the sake of the metre) because otherwise there would be a violation of the first law. 27 So far, I would agree (though, as we shall see, I regard the alliteration on gewat as 'functional' in a different sense). In accordance with the same law, he argues that both sentence particles must be stressed in Ymbeode pa (620a). 28 But we have seen that Ymbeode pa, analysed in this fashion, offends against Kuhn's second law which he does not mention. A more radical solution must be sought. It derives from the observation that a-verses which contain verbs that are not preceded by any stressed elements behave differently from b-verses of the same sort. The alliterating verb in the a-verse typically shares in alliteration with a following stressed element. The alliterating verb in the b-verse alliterates in preference to a following stressed element. If alliteration of the verb is taken as evidence that it is a stressed element, as is usually done, then numerous a-verses violate one or both of Kuhn's laws, whereas b-verses always conform to them. But these differences in behaviour, including the apparent anomalies in the operation of Kuhn's laws, can all be explained if we assume that an alliterating verb which is not in the second lift is a stressed element only when it appears in the b-verse; in the a-verse it remains a sentence particle. As long as the verb is not a stressed element and remains in the upbeat of the verse clause it is subject to both of Kuhn's laws as they apply to sentence particles even when it takes a metrical stress. 29 Metrical stress must be assigned to sentence particles in the upbeat according to an independent transformational rule. Ymbeode pa has the metrical contour x / x /, as Bliss says. 30 But the sentence particle pa has not been displaced from its normal position? nor is ymbeode a stressed element. Instead, the 27 28

Ibid. Ibid., § 19. A n d r e w , Old English Alliterative

Measure, p p . 2 3 - 4 , proposes accounting for

the stress on pa in these cases by t a k i n g \>a as enclitic w i t h a stressed verb, b u t as stressed after an unstressed syllable. 29

K u h n , p . 10, seems to have suspected this a l t h o u g h he was misled, like everyone else, by the a s s u m p t i o n t h a t only a stressed element can alliterate: ' W e n n worter, die meist schwach- oder u n b e t o n t u n d d a n n satzpartikeln sind, vollbetont g e b r a u c h t werden, sind sie vom satzpartikelgesetz unabhangig. Jedoch uberwiegt bet ihnen wenigstens in der dichtung auch in diesemfalle die stellung, in die sie als satzpartikeln gehoren (my italics).

30

Bliss, Metre, § 19.

23

The Metrical Grammar of Beowulf

entire half-line constitutes the upbeat of the verse clause. In short, Kuhn's laws describe the metrical grammar rather than the metre of Beowulf.

The transformational rule of the metrical grammar which assigns metrical stress in such cases can be stated as follows: Sentence particles in a clause-initial half-line which lacks stressed elements acquire metrical stress from right to left in accordance with the stress and phrase rules of the metrical grammar until the first valid metrical contour emerges?1 The fundamental alliterative

principle will then operate to assign alliteration to the first lift. In the a-verse half-line Sona pat onfunde (750a), the stress on onfunde suffices to generate a type A 3 metrical contour, and the transformational rule goes no further. The rule never generates a type A in the a-verse. Comparison with line 750a shows that the alliteration on the initial adverb in Sona pcet gesawon (1591a) cannot be taken as evidence that the adverb is a stressed element. This too is a type A 3 . In the case of ymbeode pa (620a), the transformational rule first assigns metrical stress to the syllable pa. Since the contour x x x / is invalid, the rule moves to the next particle to the left and assigns the phrasal stress contour x / x / with metrical stresses on the syllables -eo- and pa. This contour is metrical (type B)v and -eo- alliterates by the fundamental alliterative principle: 32 Ymbeode pa

In Gewat pa ofer wcegholm (217a) gewat does not take metrical stress. The sentence particles gewat and pa are in the dip, and the transformational rule is not required because the stressed element wcegholm suffices to establish a valid metrical contour (type A 3): Gewat pa ofer waegholm The tendency of the finite verb to alliterate when it is in initial position has been a source of some confusion. Alliteration should not be taken as evidence for metrical stress when the verb is in the upbeat of a clause in an a-verse, unless the verb has acquired metrical stress through the trans31% 32

This rule will be slightly modified in ch. 7. A marks an alliterating lift syllable; N marks a non-alliterating lift syllable; (A) or (N) marks a short alliterating or non-alliterating lift and (Ax) or (Nx) marks a resolved alliterating or non-alliterating lift. 24

Alliterative and metrical principles

formational rule. Otherwise the simplicity and function of the basic metrical types in the metrical grammar are obscured. 33 The distinction between the presence or absence of metrical stress on a stressed element or a sentence particle is a formal one, deriving from the Beowulf-pott's metrical grammar. It should not be taken to imply that stressed elements and sentence particles must have exhibited significantly different levels of stress in comparable syntactical patterns in the spoken language at the time Beowulfwas composed. Furthermore, just as I assume that a relatively weak syllable can be metrically stressed under certain conditions, so I assume that a linguistically stressed syllable can appear outside of a regular metrical contour. The contour that I give here for line 217a does not affect the probability that it would have been recited with some stress on gewat ( x \ x x x / x ) , or even (possibly) with heavy stress on gewat and pa (x / \ x x / x). The principles of the metrical grammar underlie, but do not limit, the expressive possibilities of the poem, which are realized in performance. Nor am I suggesting that the alliteration on gewat can be ignored. It is part of the alliterative pattern of the verse which is subject to strict rules. 34 Though alliteration and metre are interdependent, they are nevertheless distinct systems, which are not entirely congruent with each other. What has been said about verbs can be generalized to other sentence particles as well. The transformational rule applies to any a-verse half-line which lacks stressed elements. There are 151 such verses in Beowulf55 In 33

For a s u m m a r y of the systematic relationship between metrical type and clause structure that emerges w h e n alliterating finite verbs are scanned as I suggest here, see below, ch. 15.

34

Gewat displays extra-metrical alliteration, which I discuss in ch. 3 .

35

O n e hundred and sixteen are type A 3 (28a, 47a, 118a, 175a, 272a, 290a, 301a, 307a, 344a, 361a, 372a, 391a, 393a, 395a, 402a, 415a, 429a, 433a, 435a, 442a, 463a, 503a, 520a, 535a, 544a, 559a, 560a, 574a, 595a, 632a, 691a, 694a, 706a, 713a, 731a, 750a, 751a, 798a, 809a, 825a, 890a, 905a, 937a, 967a, 1082a, 1092a, 1095a, 1130a, 1142a, 1175a, 1221a, 1223a, 1270a, 1283a, 1336a, 1347a, 1353a, 1392a, 1465a, 1497a, 1501a, 1508a, 1550a, 1560a, 1578a, 1591a, 1661a, 1671a, 1700a, 1826a, 1846a, 1855a, 1859a, 1933a, 1975a, 2026a, 2032a, 2036a, 2039a, 2063a, 2115a, 2124a, 2158a, 2195a, 2200a, 2204a, 2304a, 2337a, 2381a, 2403a, 2450a, 2609a, 2630a, 2634a, 2665a, 2702a, 2797a, 2821a, 2864a, 2934a, 2976a, 2982a, 3002a, 3009a, 3038a, 3079a, 3081a, 3087a, 3096a, 3101a, 3104a, 3107a, 3110a, 3137a, 3141a and 3180a); nineteen are type B (34a, 237a, 333a, 377a, 525a, 620a, 652a, 778a, 880a, 1356a, 1408a, 1583a, 1608a, 1870a, 2345a, 2481a, 2516a, 2848a and 3156a); sixteen are types C 1 or C 2 (96a, 115a, 144a, 480a, 536a, 627a, 630a, 740a,

25

The Metrical Grammar of Beowulf conjunction with the fundamental alliterative principle, the rule places alliterating stress on a final verb or verbal 36 109 times. It stresses a final adverb {cetsomne, gelome, tScer tnne, to lange, swa micle, swa side, swa swide,

sydSdn, togtedre, tomiddes and pearle) seventeen times; and a final indefinite adjective (cenigne, genoge, mara, mare and monige) five times. The remaining twenty half-lines are ones in which the transformational rule places the alliterating stress on a word which is not final — in all but one case, the next-to-last word. These are primarily verb + adverb or verb + pronoun phrases, such as Hwcet syndon ge (237a) or Oferhogode da (2345a), but they also include: ponne he swulces hwcet (880a), hwceper him cenig wees (1356a), ond o^er swyle (1583a), pat hit eal gemealt (1608a) and peak de oder his (2481a). The only half-line in which the transformational rule places the alliterating stress further to the left than the last or next-to-last word is: Donne wene ic to \>e

wyrsan ge(?ingea

(525)

Here the rule first assigns metrical stress to the phrase to pe. Since this does not generate a regular metrical contour, the rule then puts metrical stress on the phrase wene ic. The resulting pattern corresponds to a type B and wene must alliterate.

36

1363a, 1944a, 2177a, 2253a, 2933a, 2985a, 3159a and 3178a). In ch. 7 the transformational rule will be extended to apply to half-lines with internal clause divisions. For a definition of'verbal' and discussion, see below, ch. 4.

26

3 The three kinds of half-lines: extra-metrical alliteration and type A 3

The basic unit of composition in Beowulf is the formula or formulaic system which is normally one-half line in length. Formulae and formulaic systems are easier to talk about than to define satisfactorily. All language is formulaic in the sense that it is made up of words inserted into repeatable patterns of syntax, and everyday speech is heavily larded with exactly repeated commonplace phrases. A poetic formula adds the dimension of poetic form — metre and alliteration in Old English poetry — to the repeated phrase. I would define an Old English poetic formula as a half-line which is repeated verbatim from first stressed syllable to line's end, with only such minor variations in unstressed words or prefixes or in spelling or inflection as do not affect its essential meaning or metrical contour or placement in the whole line or verse clause. x A formulaic system may be broadly defined to include both a half-line of a given metrical contour and syntactical pattern which shares at least one word or element of a compound in common with another half-line of the same metrical contour and syntactical pattern, and a half-line which is repeated essentially verbatim but with such variations in unstressed words or prefixes or in spelling or inflection as 1

Riedinger, 'Old English Formula', p. 306, broadens the usual definition of a formula to include which she calls a 'set': 'A "set" may be defined as a group of verses usually sharing the same function and system in which one word, usually stressed, is constant, and at least one stressed word may be varied, usually synonymously, to suit the alliterative and/or narrative context. A system may contain several different sets, each of which is a different formula, but all the verses in a set constitute the same formula - whether or not they repeat each other verbatim.' This very useful concept enables her to reconsider (pp. 313—14) the notion of'thrift', which has been a stumbling block to critics attempting to adapt Parry's theories to Old English alliterative poetry: 'An Old English set is potentially full of phrases that could "replace one another" - save in one important respect: each has a different alliterator, which is a different kind of thrift.' 27

The Metrical Grammar of Beowulf result in a change in its meaning or metrical contour or placement in the whole line or verse clause. 2 My working assumption is that virtually all of the half-lines in Beowulf ate either formulaic themselves or modelled on the metre and syntax of traditional formulae. 3 Thanks to the pioneering efforts of Lord and Francis P. Magoun, Jr, who first applied the Parry—Lord theory of oral-formulaic composition to Anglo-Saxon poetry, 4 and to the work of their followers, we have a theory which enables us to understand how and why metrical, syntactical and alliterative patterns — in particular, the metrical—syntactical patterns described by Kuhn's laws — which are common to a widely diffused body of ancient Germanic poetry could persist in this poetry without reinforcement from the patterns of everyday speech and writing. They persist because they are carried in the traditional half-line formulae and formulaic systems which individual poets absorbed from the oral culture in their years of apprenticeship. Not every half-line in Beowulf is demonstrably formulaic, but the formulae and formulaic systems which the Beowulf-pott inherited are the basis for the metrical grammar which enabled him to compose well-formed lines in the traditional manner. Now Kuhn's laws are a function of the verse clause, and the verse clause, although it sometimes coincides with the half-line, may range in length from a part of a half-line to five or more half-lines. How then can the half-line carry the patterns? The answer is simple. Since the laws apply, positively or negatively, to the first dip and the first stressed element of the verse clause, the half-line, which is either formulaic or modelled on a 2

3

4

Riedinger, ibid., p. 305, defines a 'system' as: 'a group of verses usually sharing the same meter and syntax in which one word, usually stressed, is constant and the other stressed word or words may be varied to suit the alliterative and/or narrative context.' Her definition is, as she acknowledges, 'indebted to and derived from' that of Fry in his 'Old English Formulas,' p. 203. There are problems both with the definition of a 'formula' and with a 'formulaic system' which I do not mean to minimize by this brief discussion. Are we, for example, to regard the repetition of a single word which happens to constitute an entire half-line as evidence that we are dealing with a poetic formula? On the subject of formulae and formulaic systems, see, in addition to Riedinger and Fry, Watts, Lyre and Harp, passim; Niles, Beowulf, pp. 121-37. A 'traditional formula' is a formula which is found in the work of more than one poet, and which is, presumably (one cannot exclude the possibility of one literate poet copying from another), part of the oral heritage of poetry in the culture. See Lord, Singer of Tales, p. 130; Riedinger, 'Old English Formula', p. 295, n. 7. Lord, Singer of Tales, pp. 198-202 (based on analysis done in 1949); Magoun, 'Oral-formulaic Character', pp. 446-67.

28

Three kinds of half-lines

formula, is marked for admission to or exclusion from initial position in the clause. The typical verse clause consists of a string of one or more half-lines. First comes the initial half-line. If the clause is made up of more than one half-line, the subsequent half-lines are non-initial. For example, Beowulf begins with an independent clause which is followed by a dependent clause (a virgule marks the clause boundary): Hwaet, we Gar-Dena in geardagum, t?eodcyninga (?rym gefrunon, / hu 5a aej?elingas ellen fremedon!5

(1-3)

The independent clause (1—2) is made up of an initial half-line (la) and three non-initial half-lines (lb, 2a and 2b). The dependent clause (3) is made up of an initial (3a) and a non-initial half-line (3b). Some half-lines can be placed anywhere in the verse clause; others are restricted. There are half-lines which can be placed only in initial position. These may be called clause-initial (I) half-lines.6 There are half-lines which can never be placed in initial position. These may be called clause-non-initial (II) half-lines. Finally, those half-lines which are not limited in their placement may be called clause-unrestricted (III). These properties are what Kuhn's laws describe. To restate the description of the verse clause in the terms just given, the typical clause consists of either an initial (I) or an unrestricted (III) half-line followed by zero or more non-initial (II) and/or unrestricted (III) half-lines in any order. The first, independent, clause of Beowulf is made up of an initial (I) half-line followed by a non-initial (II) half-line and two unrestricted (III) half-lines. The second, dependent, clause is made up of an initial (I) followed by an unrestricted (III) half-line. Clauses that begin with an unrestricted (III) half-line are less common than ones that begin with an initial (I) half-line, but they do occur. An example is: Maere J?eoden, unbliSe saet . . . 7

ae(?eling sergod, 5

6

7

(129b—30)

'Listen! We have heard of the glory of the Spear-Danes, of the kings of the people, in days of yore: how the nobles performed deeds of courage!' Certain clause-initial half-lines are subject to displacement which moves them out of initial position in the clause. See below, ch. 4. 'The famous chief, the noble well-tested over the years, sat unhappy . . . '

29

The Metrical Grammar of Beowulf Here all three half-lines in the clause are unrestricted (III). Clause boundaries nearly always coincide with half-line boundaries. I take up the limited number of half-lines with internal clause boundaries in ch. 7; they need not concern us here. The classification of half-lines as clause-initial (I), clause-non-initial (II), or clause-unrestricted (III) is a logical consequence of Kuhn's theoretical description of the early Germanic clause. To recapitulate: Kuhn made two crucial observations, which he summed up in his 'law of Germanic particles', or first law, and his 'law of Germanic clause beginnings', or second law. The first law states that all undisplaced sentence particles are grouped together in the first unstressed position^, or dip, in the clause. This dip is located either before or immediately after the first stressed element. It follows from the first law that any half-line which has a sentence particle in its first weak metrical position (wl), which in turn precedes or immediately follows its first lift, must be initial (I). The second law states that there must be sentence particles in the upbeat of a clause. By upbeat, Kuhn meant a dip preceding the first stressed element of a clause. A clause need not have an upbeat; it may begin with a stressed element. But if it does have an upbeat, then the upbeat must contain at least one sentence particle. Notice that the restriction does not extend to the first dip of the clause when the first dip follows the first stressed element. Then no sentence particle need appear. Just as a half-line with a sentence particle in the wl either before or immediately after the first lift must be clause-initial (I), it likewise follows from the second law that a half-line with a w l which comes before the first lift and which does not contain a sentence particle cannot be clause-initial. That is, a half-line with a weak onset consisting solely of proclitics must be clause-non-initial (II). In the first clause of Beowulf, there is one verse with a weak onset which consists solely of the proclitic preposition in: in geardagum (lb). If this verse were placed at the beginning of the clause, in would be in the upbeat, and Kuhn's second law requires that the upbeat contain a sentence particle. Since in is the only word before the first stressed element and it is not a sentence particle, we can only conclude that this placement would be a violation of the second law. Therefore in geardagum and all other verses like it must be clause-non-initial (II). Provided that the second law is valid, half-lines with weak onsets which lack sentence particles can be identified as clause-non-initial by inspection. Finally, any half-line which lacks both a weak onset and a w l with a 30

Three kinds of half-lines sentence particle is unaffected by Kuhn's laws. It may be placed anywhere in the verse clause. In the first two clauses of Beowulf there are three unrestricted (III) half-lines: peodcyninga (2a), prym gefrunon (2b) and ellen fremedon (3b). None of the three has a weak onset. Line 2a lacks a w l either before or immediately after the first lift (peod-). In line 2b the first lift (prym) is followed by a w l which is occupied by the proclitic ge- and in line 3b the first lift (the first syllable of ellen) is followed by a wl which is occupied by the second, unstressed syllable of ellen. Neither the proclitic ge- nor the second syllable of ellen is a sentence particle. Therefore all three verses are unrestricted (III). Let me illustrate the essential features of the three types of half-lines by analysing a more extensive passage of the poem. The passage (lines 1888-1913) is made up of twenty-two clauses varying from less than one half-line (1894a) to five half-lines (1896-8a) in length. It comes at the conclusion of Beowulf's expedition to Hrothgar's hall to purge it of monsters. Beowulf and his men hurry down to the shore to return to their home in Geatland (sentence particles which appear in the initial dip of a clause are italicized): Cwom pa to flode felamodigra, haegstealdra heap; / hringnet baeron, 1890 locene leoSosyrcan. / Landweard onfand eftsiS eorla, / swa he aer dyde; of hliSes nosan / no he mid hearme gaestas grette, / ac him togeanes rad, / cwced I pcet wilcuman Wedera leodum 1895 seaman scirhame to scipe foron. saegeap naca / Pa wees on sande hladen herewaedum hringedstefna, mearum ond maSmum; / maest hlifade ofer HroSgares hordgestreonum. 1900 / He |?aem batwearde bunden golde swurd gesealde, / pcet he sySJ?an waes on meodubence maj?me \>y weor(?ra, yrfelafe. / Gewat him on naca drefan deop waeter, / Dena land ofgeaf. 1905 / Da wees be maeste merehraegla sum, segl sale faest; / sundwudu )?unede; / no peer wegflotan wind ofer ySum siSes getwaefde; / saegenga for, I fleat famigheals forS ofer ySe, 31

The Metrical Grammar of Beowulf 1910 bundenstefna ofer brimstreamas, / pat hie Geata clifu ongitan meahton, cu)>e naessas; / ceol up gej?rang lyftgeswenced, / on lande stod. 8 THE CLAUSE-INITIAL HALF-LINE ( i ) Since every half-line has, at a minimum, either two weak metrical positions or two lifts, it follows that a verse clause cannot be constructed in which the first dip before or after the first stressed element appears in the second or any succeeding half-line. Therefore a half-line with one or more sentence particles in its first weak metrical position is clause-initial (I). The first dip of the clause always coincides with (or overlaps) the first weak metrical position of its first half-line. As long as these particles are not and cannot be displaced, Kuhn's first law predicts that the half-line will not be preceded by another half-line in the same clause. Thus, the half-line/^z wees on sande (1896a) must be clause-initial (I), because otherwise the particles pa and wees would not be in the first dip of the clause. Obviously the same is true of every member of the formulaic system, pa wees + preposition + noun, e.g., Da wees be mceste (1905a). Twelve clause-initial (I) half-lines appear in this passage. 9 Each has at least one undisplaced sentence particle before the first stressed element. 8

9

'There came to the flood the band of brave-hearted ones, of young men. They wore mail-coats, locked limb-shirts. The guard of the coast saw the coming of the earls, just as he had done before. He did not greet the guests with taunts from the cliff's top, but rode to meet them, said that the return of the warriors in bright armor in their ship would be welcome to the people of the Weather-Geats. There on the sand the broad sea-boat was loaded with armor, the ring-prowed ship with horses and rich things. The mast stood high over Hrothgar's hoard-gifts. He gave the boat-guard a sword wound with gold, so that thereafter on the mead-bench he was held the worthier for the treasure, the heirloom. The boat moved out to furrow the deep water, left the land of the Danes. Then on the mast a sea-cloth, a sail, was made fast by a rope. The boat's beams creaked: wind did not keep the sea-floater from its way over the waves. The sea-goer moved, foamy-necked floated forth over the swell, the ship with bound prow over the sea-currents until they might see the cliffs of the Geats, the well-known headlands. The ship pressed ahead, borne by the wind, stood still at the land' (trans. Donaldson). 1888a, 1891b, 1892a, 1893b, 1894a, 1896a, 1900a, 1901b, 1903b, 1905a, 1907a and 1911a. These are actually what I call clause-initial, nondisplaceable (la) half-lines; clause-initial, displaceable (Ib) half-lines and the distinction between the two are discussed in ch. 4. 32

Three kinds of half-lines

Each therefore belongs to one of the metrical types — A 3 , B or C — which has a weak onset. I use the term 'weak onset' to designate a weak metrical position which precedes the first lift of a half-line. It contrasts with a 'strong onset', that is, an initial lift which is not preceded by a weak metrical position. Most clause-initial half-lines have weak onsets; the few with strong onsets, like Hream weard in Heorote (1302a), are almost always type A. A sentence particle in the first dip of a clause-initial half-line will sometimes alliterate. Often such alliteration is incidental, by which I mean that it is superfluous with respect to the formal alliterative pattern of the line. It falls on the less significant particles. The Beowulf-poet may have used incidental alliteration deliberately. Sensitive to alliteration as he obviously was, he must have been aware of the pile-up of initial H's in this line: ne hie huru heofena Helm

herian ne cu(?on

(182)

But in other cases incidental alliteration may have gone unnoticed. It can be found in either half-line: Heebbe ic eac geahsod, word wceron wynsume. maga rice.

J?aet se aeglaeca Eode Wealh)?eow for5

Me (?in modsefa

(433) (612) (1853)

However, if the alliterating sentence particle is a verb other than wesan or weorpan10 or an adverb other than one of weak semantic force, 11 the alliteration has certain formal properties. This can be described as extra-metrical alliteration.u In extra-metrical alliteration, the alliterating syllable does not take metrical stress. It lies outside (to the left of) the regular metrical contour of the verse. But, and this is the crucial point, extra-metrical alliteration obeys the basic alliterative rule which limits the 10 11

12

Perhaps also the verb agan (see 1727b). Adverbs of weak semantic force in Beowulf include: eac, efne,ful, huru, hwanon, hweedre (except in the phrase hweetfre swa peah), neefre, nces, ne, nealles, no (but see 2585b), nopy eer, nu (but see 1174b), swa, swylce (including eall swylce and efne swylce), to (only when it is proclitic on another adverb or adjective), pa, par (only in the phrases peer wees, ne wees . . . peer, peer is, nispeer, peer inne and peer on innan), ponan (but see 123b, 69 lb), ponne and pus. In Kendall, 'Displacement', p. 15, and passim, I used the term 'ornamental alliteration' for extra-metrical alliteration. This was an unfortunate choice of a term, since it implied what I did not intend, that extra-metrical alliteration is somehow casual or unimportant.

33

The Metrical Grammar of Beowulf number of significant alliterating syllables in the a-verse to two and in the b-verse to one. Extra-metrical alliteration is part of the alliterative pattern of the line. Since double alliteration is prohibited in the b-verse, extra-metrical alliteration is also excluded. It can only occur in the a-verse, and then only when the metrical contour of the a-verse displays single alliteration, for example (extra-metrical alliteration is marked with a lower-case a): Gewat pa ofer waegholm Ofersw<swz da sioleSa bigong

winde gefysed sunu EcgSeowes

(217) (2367)

But, where double alliteration is already present, extra-metrical alliteration on the verb is not possible: geafme sine ond symbel,

sibbe gemunde

(2431)

Among the twelve clause-initial half-lines in the passage are four half-lines which prosodists agree are type A 3 (1888a, 1892a, 1896a and 1905a). Type A 3 is, of course, restricted to the a-verse. But aside from this, it is usually assumed that the metrical types were completely interchangeable — that any metrical type could be substituted for any other in any place. 13 But this is not true of type A 3 in Beowulf. A significant and largely unnoticed feature of this type is that it is exclusively clause-initial (I). l 4 That this should be so is by no means self-evident. One can easily construct a hypothetical verse with the contour ( . . . ) x x / x which is not clause-initial: 15 #

ofer scildweall

(cf. 3118a)

But such verses do not occur in Beowulf. There are roughly 310 commonly recognized type A 3 in Beowulf.16 Since all type A 3 have at least one sentence particle in the weak onset, they must 13

14

15 16

Fry, 'Old English Formulas', p. 202, for instance, states: 'Absolutely no evidence exists to prove that Old English poets used a half-line of a specific metrical type at a certain place or for certain effects; any verse which would fit one of thefivetypes was legitimate.' Pope, Rhythm, p. 82, comes close to seeing this: 'type A3 is employed very frequently as a light introduction to the weighty verses that follow'. I use an asterisk to indicate a hypothetical verse introduced for the sake of the argument. I count 310; others will arrive at slightly different totals, depending on what they regard as genuine as opposed to accidental double alliteration, and depending on how many partially or totally reconstructed lines they admit (Bliss counts 363 (his types al, a2 and el); Pope counts 309). The 310 include six 'short' A3: 262a, 459a, 779a, 1514a, 1728a

34

Three kinds of half-lines be clause-initial according to Kuhn's first law. Only nine require special comment with respect to the claim that all type A 3 are clause-initial. Correlative constructions must be construed independently. This applies, for example, to line 943a in the sequence of clauses: Hwaet, J?aet secgan maeg / efne swa hwylc maegt>a / swa Sone magan cende aefter gumcynnum, gyf heo gyt lyfaS, J?aet hyre Ealdmetod este waere bearngebyrdo.17 (942h^46a) Similarly the conjunction buton once appears without a verb: Ne nom he in J?aem wicum, Weder-Geata leod, maSmaehta ma, (?eh he |?aer monige geseah, / buton J?one hafelan ond J>a hilt somod since fage . . , 18

(I6l2-15a)

In one case the relative pam (= pampe) acts both as the dative object of the preceding verb gescod and as the subject of the following relative clause. It must be construed as subject: Bill aer gescod — ecg waes iren — ealdhlafordes / pam Sara maSma mundbora waes longe hwile . . . 19

(2777b-80a)

Only the formula Mfter pcem wordum (1492a and 2669a) appears to contradict the claim that every A 3 has a sentence particle in the weak onset. There are six other occurrences of this formula in Anglo-Saxon poetry: Exodus 299a and 565a; Andreas 88a, 76la, 1026a and 1219a. Every one stands at the head of a clause. Although as an A 3 the formula is properly clause-initial, the lack of a sentence particle in the initial dip would seem to

17

18

19

and 2048a. As a consequence of the theory of displacement, which I propose in ch. 4, I would increase the total number of A 3 to 486. 'Indeed, she may say, whichever one of women as gave birth to that son among the race of men, if she yet lives, that the Ruler of Old was kind to her in regard to childbearing'; so also, 1092a, 1223a, 1249a, 1283a, 3057a and 3164a. 'He did not take from the dwelling, the warrior of the Weather-Geats, more precious treasures, although he saw many there, except the head and also the hilt, shining with ornaments . . . ' 'The sword of the old lord — its edge was iron — had before this destroyed the one who had been the protector of the treasure for a long time. . . '

35

The Metrical Grammar of Beowulf violate Kuhn's second law. But Kuhn points out that the demonstrative occupies a special position in the metrical grammar in that it is not clearly differentiated from the relative and often pushes into the head of the clause where it refers back to the preceding sentence. 20 Hence pcem (or pyssum 3X in Andreas21) should be specially marked in these verses. Non-initial (II) type A 3 can be found in other Old English poems. Most of the ones I have observed contain a demonstrative, as, for example, in the Exeter Book: eard weardade

Rondburgum weold, oftast symle

in pare ceastre

Commedia,

heold hordgestreon.22

(Juliana 19b-22a)

This use of the A 3 in non-initial position in other Anglo-Saxon poetry may be a consequence of the loss of the relative function of the demonstrative. 23 In any event, the Beowulf-poet's strict avoidance of the noninitial type A 3 is a distinctive feature of his metrical grammar. THE CLAUSE-NON-INITIAL

HALF-LINE

(il)

A half-line with a weak onset consisting entirely of proclitics may be placed in any position in the verse clause except the first. That is to say, we will expect to find at least one half-line in front of it in the same clause. Kuhn's second law predicts that the initial upbeat of a clause, if there is one, will contain a sentence particle. Therefore, any half-line that has a weak onset which lacks a sentence particle can not be clause-initial, for example: on J?isse meoduhealle

(638a)

20

Kuhn, pp. 4 4 - 5 .

21

Editors differ on the location of the medial caesura in Andreas 88: Krapp (ASPR 2) reads: y^fter Jjyssum wordum

com wuldres tacen

But Brooks (Andreas) prefers: /Efter J?yssum wordum com

wuldres tacen

See D o n o g h u e , Style, p . 66. W h e t h e r or not Andreas 8 8 a is a type A 3 does not affect t h e a r g u m e n t here. 22

' H e [Heliseus] ruled t h e citadels, g u a r d e d t h e country continually in the city of N i c o m e d i a , possessed t h e treasure-hoard.'

23

See further t h e discussion of t h e formula after pcem wordum in c h . 6 .

36

Three kinds of half-lines Since there are no sentence particles in the weak onset (on pisse), line 638a should not begin a clause, and in fact it is in the middle of a clause which consists of three half-lines: Ic gefremman sceal eorlic ellen, / o(?Se endedaeg on \>isse meoduhealle minne gebidan!24

(636b—38)

Whatever part of speech it is, the alliterating word in the b-verse must be a stressed element. If the alliterating word is preceded by one or more proclitics, then, by Kuhn's second law, the half-line ought to be clause-non-initial (II), for example: peah de he his broSor beam

abredwade

(2619)

The verb abredwade forms the second half-line in this two half-line clause. The proclitic prefix a- marks it as clause-non-initial (II). None of the eight half-lines of the type Ymbeode pa25 is found in the b-verse. Since the verb is not a stressed element, each of these verses is in its entirety the initial dip of a clause. The transformational rule supplies metrical stress. Is there any reason why these clause-initial (I) half-lines are confined to the a-verse? In the b-verse the alliterating verb would have to be considered a stressed element. The proclitic would then mark the verse as clause-non-initial (II). Evidently the economy of the oral-formulaic technique prevented a formulaic system from being employed simultaneously as clause-initial in the a-verse and clause-non-initial in the b-verse. In our illustrative passage (1888-1913), seven half-lines have weak onsets without sentence particles. They are (grouped according to their syntactical patterns): (a) (b) (c)

(d)

24 25

of hliSes nosan ongitan meahton ofer HroSgares on meodubence ofer brimstreamas toscipeforon on lande stod

(1892b) (1911b) (1899a) (1902a) (1910b) (1895b) (1913b)

'I shall perform a heroic deed, or experience m y final day in this mead hall!' 3 4 a , 6 2 0 a , 6 5 2 a , 1408a, 1870a, 2 3 4 5 a , 2 5 1 6 a a n d 3156a. See Bliss, Metre, § 19;

Willard and demons, 'Bliss's Light Verses', pp. 242-3. 37

The Metrical Grammar of Beowulf The first six are clearly non-initial (II); the seventh begins a clause in Klaeber's text. Let us extend the sample. Altogether in Beowulf there are thirty-six half-lines with the syntactical pattern of (a): preposition (+ demonstrative/ possessive) + dependent genitive + object. Two are in the a-verse and the rest in the b-verse. 26 None begins a clause. There are thirty half-lines with the pattern of (b): infinitive with unstressed prefix + auxiliary. All are in the b-verse. 27 Again none begins a clause. There are 183 half-lines with the pattern of (c): preposition (+ demonstrative/possessive) + object. 28 Of these, only the formula JEfter pcem wordum (1492a, 2669a) is found at the head of a clause. 29 There are thirty-four half-lines with the pattern of (d): preposition (+ demonstrative) + object + finite verb. Ten are in the a-verse; twenty-four 26

27

28

29

214a, 668a; 21b, 27b, 28b, 158b, 311b, 326b, 357b, 471b, 580b, 633b, 662b, 845b, 1114b, 1147b, 1208b, 1479b, 1672b, 1826b, 1861b, 1892b, 1950b, 1954b,2059b, 2147b, 2193b, 2320b, 2405b, 2455b, 2549b, 2675b, 2755b, 2803b, 2915b and 3096b. 308b, 355b, 511b, 571b, 648b, 738b, 780b, 910b, 9 6 l b , 990b, 1078b, 1140b, 1277b,1350b,1462b,1496b,1504b,1535b,1561b,1628b,1875b,1877b,1911b, 1919b, 1998b, 2090b, 2186b, 2588b, 2770b and 2954b. 10a, 37a, *73a, 85a, 117a, 125a, 174a, 178a, 180a, 187a, 200a, 239a, 243a, 257a, 285a, 297a, 304a, 317a, 324a, 382a, 383a, 385a, 396a, 410a, 425a, 434a, *443a, 461a, 462a, 481a, 492a, 519a, 537a, 566a, 584a, 601a, 638a, 647a, 695a, 738a, 824a, 838a, 840a, 851a, 859a, 885a, 899a, 921a, 944a, 964a, 971a, 976a, *981a, 1003a, 1012a, 1026a, 1029a, * 1052a, 1062a, 1067a, 1070a, 1094a, 1149a, 1213a, 1244a, 1258a, 1262a, 1315a, 1317a, 1320a, 1403a, 1419a, 1421a, 1433a, 1492a, 1561a, 1606a, 1635a, 1638a, 1639a, 1695a, 1704a, 1731a, 1851a, 1902a, 1922a, 1928a, 1938a, 1943a, 1952a, 1979a, 2010a, 2030a, 2049a, 2066a, *2083a, 2176a, 2197a, 2203a, 2232a, 2261a, 2298a, 2311a, 2366a, 2374a, 2416a, 2445a, 2452a, 2463a, 2465a, 2502a, 2515a, 2520a, 2531a, 2540a, 2581a, 2605a, 2627a, 2635a, 2639a, 2644a, 2652a, 2669a, 2750a, 2753a, 2786a, 2798a, *2856a, 2905a, 2907a, 2935a, 3004a *3010a, 3097a and 3177a; lb, 26b, 75b, 126b, 174b, 175b, 298b, 360b, 380b, 393b, 395b, 460b, 475b, 484b, 540b, 617b, 629b, 710b, 820b, 854b, 906b, 1073b, 1082b, 1284b, 1330b, 1429b, 1437b, 1578b, 1659b, 1771b, 1910b, 1963b, 2039b, 2079b, 2128b, 2204b, 2437b, 2665b, 2805b, 2815b, 2917b, 2925b, 2949b, 2993b, 3031b, 3083b, 3123b and 3136b. A half-line which resumes a clause after an interruption is usually, though not always, felt to be non-initial. See Kuhn, p. 9. The seven half-lines marked with an asterisk in this list are preceded by a parenthetical clause or phrase. See above, and also ch. 6.

38

Three kinds of half-lines

in the b-verse. Thirty are unquestionably non-initial. 30 The other four begin clauses in Klaeber's text (1913b and 552b, 1110a and 1684a). In 1110a the demonstrative has a relative function and must be marked as a sentence particle: 31 Here-Scyldinga betst beadorinca waes on bael gearu. JEt pcem ade waes e)?gesyne swatfah syrce, swyn ealgylden, eofer irenheard, ae(?eling manig wundum awyrded . . , 32

(1108b-13a)

Possibly the scribe was having problems at this point. The preceding half-line, wees on bcelgearu, violates Kuhn's first law. The scribe's exemplar may have been corrupt or difficult to read. Lines 552b and 1913b can be repunctuated. Here are the passages in which they occur: (1) Waes merefixa mod onhrered; }>aer me wiS laSum licsyrce min heard hondlocen helpe gefremede, beadohraegl broden, / on breostum laeg golde gegyrwed.33

(549-53a)

Since beadohrcegl is the subject of the verb Iceg, as well as a variation of licsyrce which is the subject of gefremede, the comma after broden can be omitted without any change of meaning and the clause division put back to the end of line 551. 3 4 30

60a, 212a, 1280a, 1737a, 1842a, 2208a, 2264a and 2380a; 225b, 332b, 507b, 531b, 544b, 587b, 766b, 1001b, 1144b, 1154b, 1158b, 1342b, 1592b, 1640b, 1895b, 2034b, 2048b, 2054b, 2203b, 2348b, 2429b and 2960b.

31 32

SeeKuhn, pp. 4 4 - 5 . ' T h e best of t h e battle-warriors of t h e War-Scyldings was ready on t h e pyre. A t t h e funeral fire was easily visible t h e bloodstained shirt of m a i l , t h e golden figure of a swine on t h e h e l m e t , t h e iron-hard boar figure, m a n y a noble done in by w o u n d s

33

' T h e anger of t h e sea-fishes was aroused; there m y mail-shirt, hard, h a n d - l i n k e d , aided m e against t h e hostile ones, t h e woven b a t t l e - g a r m e n t , it lay on m y breast adorned w i t h gold.'

34

O n t h e r e p u n c t u a t i o n of this passage, see below, ch. 6. Wrenn—Bolton o m i t t h e c o m m a after broden\ W y a t t - C h a m b e r s o m i t t h e c o m m a and p u t a semi-colon after gefremede. Sievers t h o u g h t it likely xhatficet me had fallen o u t of the text before 5 5 2 b ; see his ' Z u m Beowulf \ 1 3 8 . H i s e m e n d a t i o n would p e r m i t t h e clause division t o remain as is.

39

The Metrical Grammar of Beowulf (2) lyftgeswenced,

ceol up gef>rang / on lande stod.

(1912b—13)

These, the last three half-lines of the illustrative passage, offer an interesting example of the way literary interpretation colours the definitions of words when they are entered in glossaries. 35 The compound lyft-geswenced, which is found only in this passage in Beowulf, is now usually glossed 'driven by the wind'. 36 The gloss fits the punctuation adopted by recent editors, 37 and accordingly the passage must be translated: 'The ship rushed forward, driven by the wind; it rested on shore.' Now the literal meaning of the participle geswenced is something like 'harmed', 'vexed', 'afflicted'. The interpretation 'driven' is not only strained but it is responsible for another distortion. Editors and translators alike seem to have entered into a silent conspiracy to ignore the literal meaning of the adverb up. But if the comma after lyftgeswenced is shifted back to follow geprang, the lines can be translated so as to keep the literal meaning of the participle and give full force to up: The ship drove up (i.e., ashore}; wind-battered, it rested on land.' This represents a return to the translation of lyftgeswenced ('weather-beaten') and the punctuation approved by Bosworth—Toller (s.v., lyftgeswenced). The result is a more vigorous reading, and line 1913b turns out to be clause-non-initial (II) after all. Line 1684a is the only one of the four which clearly both begins a clause and stands in violation of Kuhn's second law. The passage in which it appears runs like this: hit on aeht gehwearf 1680 aefter deofla hryre Denigea frean, wundorsmi(?a geweorc; ond J?a |?as worold ofgeaf gromheort guma, Godes andsaca, morSres scyldig, ond his modor eac; / on geweald gehwearf woroldcyninga 1685 Saem selestan be ssem tweonum Sara \>e on Scedenigge sceattas daelde.38 35 36

Robinson calls a t t e n t i o n to this p h e n o m e n o n in 'Lexicography', p p . 9 9 - 1 1 0 . Klaeber; Clark Hall—Meritt; 'driven by the air', Wrenn—Bolton; ' w i n d - u r g e d , driven by the w i n d ' , W y a t t - C h a m b e r s ; 'luftgeschwenkt, v o m W i n d e getrieben', von Schaubert.

37

W r e n n - B o l t o n , W y a t t - C h a m b e r s , D o b b i e , Klaeber, von Schaubert. As in 5 5 2 b , Sievers, ' Z u m Beowulf,

38

1 4 1 , proposed i n s e r t i n g / ^ he in 1 9 1 3 b .

'It [the golden h i l t ] came into the possession of the lord of the Danes, after the fall of the devils, the work of the w o n d e r - s m i t h s ; and at that t i m e the a n g r y - m i n d e d warrior,

40

Three kinds of half-lines The subject of gehwearf in 1684a is the same as that of gehwearf in 1679b — the hilt of the sword which Beowulf found in the monsters' hall in the haunted mere and used to kill Grendel's mother. Kuhn argued that a preposition could stand alone in the upbeat of a clause lacking its subject (as here, where the subject of gehwearf must be understood from 1679b). 39 However the examples he cited can be otherwise accounted for, and this may not be a valid generalization. In any case the transition from the intervening clause (168 lb-83) to 1684a seems extraordinarily abrupt. The passage is awkward, as both Cosijn and Klaeber observed. 40 The apparent violation of Kuhn's law can be offered as an additional reason for suspecting that the text is disturbed at this point. 41 In sum, the illustrative passage we began with led to the investigation of 283 half-lines which promised from their syntax to be non-initial (II). Three had to be reclassified as initial. Of the remaining 280, 279 proved to be non-initial in fact — a striking confirmation of Kuhn's second law. THE CLAUSE-UNRESTRICTED HALF-LINE

(ill)

A half-line which lacks the distinguishing characteristic of an initial (I) or a non-initial (II) half-line — namely, a sentence particle in its first weak metrical position or a weak onset without a sentence particle — may be placed anywhere in the verse clause. Thirty-two of the fifty-two half-lines in the illustrative passage are

39 41

God's adversary, guilty of murder, had given up this world, and his mother had too; it came into the possession of the best of the kings of the world between the seas, of those who dealt out treasures in the realm of the Danes.' 40 Kuhn, p. 44, n 1. Cosijn, Aanteekeningen, p. 25; Klaeber, p. cvii. It is tempting to argue that the subject of gehwearf'in 1684a is not the sword, but Grendel (and his mother). The clause division could then be put back to the middle of line 1682, eliminating both the violation of Kuhn's law and the awkward transition that Klaeber objects to: 'and at that time the angry-minded warrior had given up this world; God's adversary, guilty of murder, and his mother too, came into the power of the best of the kings of the world between the seas, of those who dealt out treasures in Scedenig . But does it make sense to say that Grendel and his mother came into Hrothgar's power, when Beowulf killed them? Alternatively, if Beowulf is meant by 'the best of the kings of the world', there is the problem that Beowulf is never called a king until he becomes one in the last part of the poem and Scedenig is perhaps less likely as an appellation for Geat-land than it is for Denmark.

41

The Metrical Grammar of Beowulf unrestricted (III). Seven begin their clauses 42; twenty-five do not. 43 It is merely coincidental that the seven unrestricted half-lines in clause-initial position in the passage happen all to be b-verses. Unrestricted a-verse half-lines appear freely at the head of clauses elsewhere in the poem. A special variety of the unrestricted (III) half-line is the b-verse with an alliterating verb (or comparable form) and neither initial proclitics nor sentence particles in any weak metrical position. This can be illustrated in faeger foldan bearm; gist of geardum . . . 44

Da waes winter scacen, fundode wrecca, (1136b-38a)

Line 1137b begins a clause. However, a similar half-line (1128b) with the same metrical contour (type A) appears in the middle of a four half-line clause: Hengest 3a gyt waelfagne winter wunode mid Finne eal unhlitme . . . 45

(1127b-29a)

Displacement is not involved because the verb wunode, like fundode in 1137b, was already a stressed element at the time the half-line was formed. Half-lines like fundode wrecca and wunode mid Finne, in which a finite verb leads a nominal, are restricted to the b-verse; if they were to appear in the a-verse, alliteration would shift to the nominals. 42 43

44

1889b, 1888b, 1899b, 1910a,

1890b, 1898b, 1904b, 1906b, 1908b and 1912b. 1889a, 1890a, 1891a, 1893a, 1894b, 1895a, 1896b, 1897a, 1897b, 1898a, 1900b, 1901a, 1902b, 1903a, 1904a, 1905b, 1906a, 1907b, 1908a, 1909b, 1912a and 1913a.

' T h e n w i n t e r was g o n e , earth's bosom was fair; t h e exile was anxious to g o , the guest from the dwelling

45

'Still H e n g e s t lived w i t h Finn the slaughter-stained w i n t e r all involuntarily . . . '

42

4 Displacement

The mechanism of displacement plays an important part in the metrical grammar of the Beowulf-poet. When a sentence particle is displaced from its normal unstressed position in the first dip of the clause, it will be marked by alliteration or it will take metrical stress or both, l Displacement is spoken of as though it referred to a procedure whereby the poet would remove a word from its normal syntactical position and insert it somewhere else in the verse clause. But this assumption will not adequately account for all the observed phenomena. Furthermore, it is rooted in a prior assumption — that Old English poetry was essentially, not just secondarily or superficially, the product of a literate culture. We can readily imagine a poet, pen in hand, mentally shifting a word from its expected position in one verse and fitting it into some other verse, possibly several lines removed, while making the necessary adjustments in both verses in order to preserve their proper rhythms. But to suppose that an oral poet, employing the traditional half-line formulae, could or would engage in such mental gymnastics in the midst of an extemporaneous performance seems intuitively unlikely. Even if, as I assume, the Beowulf-pott and others like him in the monastic culture of Anglo-Saxon England wrote down their poems by a process of self-dictation, their steady reliance on the traditional formulae and formulaic systems implies that they continued to practise the techniques of oral composition. A more adequate and plausible explanation of the mechanism of displacement is needed. I suggest that there are, in fact, two types of displacement, internal and external, which can be distinguished by the kind of movement that is involved. 1

See above, ch. 2. A proclitic which is displaced from its normal position behaves similarly. I take up displaced proclitics in ch. 11.

43

The Metrical Grammar of Beowulf INTERNAL DISPLACEMENT

The movement of a sentence particle within the limits of a half-line may be called 'internal displacement'. A sentence particle which takes metrical stress as a result of internal displacement behaves in all respects like a stressed element. Internal displacement may be thought of as belonging to the first stage of composition — the formation of the useful half-line according to the principles of the metrical grammar. For example, there is an internally displaced verb in the clause-initial half-line herestrael hearda;

he on holme wees

(1435)

Since the verb wees is a sentence particle, its normal position would be with he in the dip before the first stressed element. The half-line *he wees on holme would make a perfectly regular type A 3 . Compare, for example: He is for eorSan

(Elene 591a)

He wses on elne

(Guthlac 328a)

Half-lines like these, with the verb undisplaced, are of course confined to the a-verse, because their metrical contour is type A 3 . But all that is required to turn them into half-lines which could go into the b-verse is to move the verb from the dip to a place after the first stressed element. Thus the metrical—syntactical type A 3 system, pronoun + monosyllabic auxiliary/verb to be + prepositional phrase, has a reflex in the type B system, pronoun + prepositional phrase + (displaced) auxiliary/verb to be. There are six half-lines, including he on holme wees, belonging to this reflex system in Beowulf.2 All of them are b-verses, which suggests that the poet used displacement in this instance for the purpose of generating b-verse equivalents for type A 3 systems. Once a new system of half-lines has been created by internal displacement, the placement of these half-lines in the verse clause will be governed by their onsets. The half-line he on holme wees remains clauseinitial (I), because it still has a sentence particle {he) in a w l before the first lift. But consider line 2048b in the following passage: Meaht Su, min wine, mece gecnawan, /J?one \>in fseder to gefeohte beer 2

753b, 816b, 1170b, 1435b, 2535b and 2612b.

44

Displacement

under heregriman dyreiren... 3

hindeman siSe, (2047-50a)

The verb beer in 2048b has been internally displaced and is a stressed element. The half-line, which would have been clause-initial if the verb had not been displaced, now has a w l occupied entirely by proclitics {to ge-). It should be clause-non-initial (II), and it is in fact the second half-line in its clause. Given the dramatic effects of displacement, it is vital to be able to recognize the characteristics and limits of the first dip of the clause. These differ according to which of the two possible positions the dip is in, that is, before the first stressed element or after it. The number of syllables which can make up the dip in the first allowable position (before the first stressed element of the clause) is theoretically unlimited, although five or six syllables would seem to be the normal maximum, given the syntactical patterns of the language. Ordinarily, this dip coincides with the first weak metrical position of metrical types A 3 , B or C. In exceptional circumstances it may extend over an entire a-verse (as in Ymbeode pa (620a)), in which case the transformational rule assigns metrical stress. The boundary of the a-verse appears to be the absolute limit, however. All verses like 620a are immediately followed by a stressed nominal in the b-verse. We shall see shortly that the alliterative nature of the whole line prevents this dip from occupying an entire b-verse. The second allowable position for the initial dip (immediately after the first stressed element of the clause) is largely confined to the one metrical type — type A — which has its initial weak metrical position between the first and second lifts. 4 The metrical contour of type A (/ (x) x / x) restricts this weak metrical position to a normal maximum, only infrequently exceeded, of two syllables.5 In Beowulf, sentence particles rarely occupy the second 3

4

5

'Can you, my friend, recognize the sword, which your father, protected by his war-helmet, carried to battle for the last time, the precious iron sword . . . ?' Five verses with sentence particles in the second allowable position can be analysed as expanded type D (473a, 612a, 1323b, 1724b and 1997b). Alternative scansions, involving in three cases emendations, have been proposed for all five. See Pope, Rhythm, pp. 309, 365 and 366; Bliss, Metre, §§ 44, 52 and 78 (Bliss, Index to the Scansion, scans each of the five as type A). Line 343b (Beowulf is min nama) is difficult. Half-lines with internal clause divisions (e.g., 183b, 186b and 603b) present special problems. These will be taken up separately in ch. 7. Handbooks of Old English versification give the misleading impression that the first weak metrical position in type A may extend for as many as five syllables. See, for

45

The Metrical Grammar of Beowulf allowable position. Any sentence particle which extends beyond the second syllable after the first lift can be considered displaced. Take, for example, the two half-line clause: helm ne gemunde,

byrnan side . . .

(1290b-91a)

The verb gemunde is a sentence particle which extends beyond the limit of the first weak metrical position of 1290b (= the proclitics ne and ge-) and is thus displaced. With the exception of finite forms oiwesan and weorpan no verbs are found in this dip in Beowulf. It might be argued that the initial dip in the second allowable position could extend as far as the boundary of the half-line (or beyond). Lines 1290b-9 la would then have to be analysed: helm ne gemunde,

byrnan side . . . The transformational rule could be modified to assign metrical stress to gemunde, but if it were, the modified rule would seem to permit certain patterns which are syntactically possible, but which do not in fact occur, such as: *helm gewat pa

Since verses of this kind are not found, this modification of the transformational rule is too accommodating. It seems better to regard the group of a-verse lines like 620a as an exceptional phenomenon, the explanation for which is to be found in the inherent ambiguity of sentence particles in the a-verse before the first stressed element. Normally, the initial dip of the clause coincides with the first weak metrical position of the half-line. 6 Since the second allowable position for sentence particles must come before the second stressed element of the clause, a compound with two stressed elements which is placed at the head of a clause pre-empts the second position. A compound with the form / x / (e.g., mil-gemearces) encloses the dip, thus preventing any sentence particle from occupying it. A compound with clashing stress, that is, with the form / / (e.g.,

6

example, Bright's Old English Grammar, ed. Cassidy and Ringler, p. 279- This results in part from including verses which are properly type A 3 . Exceptionally, the initial dip of the clause follows an internal clause division within the half-line. See below, ch. 7.

46

Displacement

mearc-stapari) excludes the dip altogether. 7 Therefore a verb which follows a compound is displaced from its position as a sentence particle and must be stressed, for example: Benc-)?elu beredon\

hit geondbraeded wearS

(1239)

A caveat must be entered at this point. Proper names which are etymologically compounds do not behave like other compounds with two stressed elements. 8 No ordinary compound in Beowulf is followed by an unstressed copula. But we find verses like — blaed wide sprang9

Beowulf wees breme beodgeneatas;

Beowulf is min nama

Wiglaf wees haten,

Weoxstanes sunu

(18) (343) (2602)

All are clause-initial, and the copulas are undisplaced. EXTERNAL DISPLACEMENT

The second kind of displacement occurs when a sentence particle is moved out of the initial dip of the clause and yet retains its position at the beginning of its own half-line. This may be called 'external displacement'. I will begin with an illustration and then describe what has happened. The verb scoc is in the normal position for a sentence particle at the head of a clause in the half-line scoc ofer scildweall

(3118a)

As it happens, however, the verse is the third half-line in a clause that consists of three half-lines: (weaxan wonna leg) 7

8

9

Nu sceal gled fretan wigena strengel,

Kuhn, p . 9, seems to have considered a compound with clashing stress to be the equivalent of a single stressed word, although his statement is not entirely clear. However, the conclusion that it must be the equivalent of two stressed elements in the metrical grammar of Beowulfseems inescapable. See Campbell, 'Verse Influences', p. 95; Bliss, 'Auxiliary and Verbal', p . 165. For one thing, the stress of the second element is reduced. See Campbell, Old English Grammar, § 88. The word Beowulf here refers not to the hero of the poem, but to Hrothgar's grandfather, whose name may originally have been a monosyllabic form like Beow or Beaw.

47

The Metrical Grammar of Beowulf l?one 3e oft gebad isernscure, / ponne straela storm strengum gebaeded scoc ofer scildweall, / sceft nytte heold . . . 10

(3114b-18)

The verb scoc is displaced from the first dip of the clause and does not appear in the half-line which contains the first dip (ponne). This is external displacement. The relationship between internal displacement and scansion is unambiguous. The displaced particle behaves like other stressed elements and inevitably receives metrical stress. But with external displacement the case is different. In external displacement it is not the particle which is displaced, but the half-line. So, in the example above, scoc has not been moved from beside ponne in line 3117a. Instead, the half-line scoc ofer scildweall\ which could in another context have stood at the head of its own clause, has been held back for later insertion, and the half-line ponne strcela storm begins the clause. Alliteration on the sentence particle scoc enables the half-line scoc ofer scildweall to be displaced, 11 but I would argue that the metrical status of the half-line is unaffected by its displacement. It should be scanned just as it would have been if it had appeared at the head of the clause. External displacement, that is to say, belongs to the second stage in the compositional process — the poet's creative arrangement of well-formed half-lines, still according to the principles of the metrical grammar, into the verse clauses which make up the poem. THE CLAUSE-INITIAL H A L F - L I N E , DISPLACEABLE

(IB)

A number of consequences derive from the principle that external displacement is always by half-line. They lead to a paradox: only a clause-initial half-line may be displaced. In the discussion that follows, 10

11

'Now theflamemust eat the prince of warriors (the darkfiremust increase), he who often endured the shower of iron, when the storm of arrows strongly impelled hastened over the shield-wall, the shaft did its duty Since internally displaced sentence particles inevitably move into the second lift position (which is metrically stressed), they may plausibly be said to acquire some additional degree of stress as a result of their displacement. In the earlier version of this chapter which appeared in my article, 'Displacement', p. 12, I argued that scoc must receive 'positional stress' as a result of external displacement. Though it seems intuitively likely that some additional degree of stress might accompany external displacement, the only certainty is that the sentence particle is marked by alliteration.

48

Displacement

'displacement' means external displacement, and all displaced half-lines will prove to be clause-initial (I). (a) Displacement is limited to half-lines with a single internally undisplaced

sentence particle. Since it is the movement of the entire half-line which displaces the sentence particle, the same movement would theoretically displace any other sentence particles in the half-line. A displaced verse with two or more sentence particles would be overburdened - it would have too many alliterating syllables. 12 Therefore no half-line which has two or more sentence particles in the first weak metrical position can be displaced. Hence there are two classes of clause-initial half-lines: nondisplaceable (la) 13 and displaceable (Ib). The clause-initial half-lines analysed in ch. 3 were all nondisplaceable (la). In his study of 'light' and 'normal' verses, A. J. Bliss divides those non-auxiliary finite verbs of Beowulf which are not preceded by a stressed element in the same verse by half-lines 'into nine groups according to the position of the verb in the verse clause'. l4 In four of his groups ( 1 , 2 , 3 and 7) the verb is the only sentence particle which can be displaced. The remaining groups (4, 5, 6, 8 and 9) are made up of half-lines in which the verb is accompanied by at least one other undisplaced sentence particle. The latter groups are invariably clause-initial, nondisplaceable (la), as is to be expected. Only in the former (1, 2, 3 and 7) does the possibility of displacement exist. In Bliss's group 1, 'the finite verb has been displaced from its normal position among the particles at the beginning of the clause, and must therefore be treated as a stressed element'. 15 We know that the verb has been displaced because there is a stressed element in the preceding half-line which is part of the same verse clause. But it would be a mistake to suppose that the poet moved the verb from its normal position and inserted it into a later part of the clause. What he did was to move the half-line with the verb in it. That is to say, group 1 really consists of half-lines which have been displaced from their position at the beginning of the clause. For example, song in searwum is clause-initial (I) because there is a sentence particle (song) in the first weak metrical position before the stressed element 12 13

14

See below, rule (b). In ch. 15,1 put nondisplaceable half-lines with a sentence particle in the second allowable position in a separate category, labelled type Ic. 15 Bliss, Metre, § 13. Ibid., § 14.

49

The Metrical Grammar of Beowulf searwum. It would be an appropriate verse with which to begin a clause. But instead the poet wrote: GuSbyrne scan heard hondlocen, / hringiren scir song in searwum . . . 16

(321b-23a)

He displaced the half-line from its normal position. Displacement of the half-line is made possible by the alliterating verb, which is the only particle in the weak onset. 17 Bliss's groups 3, where the Verb is the only particle before the first stressed element', and 7, where the Verb forms a whole clause in itself',18 are similar to group 1 except that these half-lines begin their clauses. They have not been displaced. Group 2 is made up of a series of a-verses, again with the verb in initial position, where it is in apposition to a verb in the preceding half-line. The first verb is internally displaced. Consider, for example: weox under wolcnum

he pees frofre gebad, weorSmyndum (?ah . . . 19

(7b-8)

Strictly speaking, the second verb weox should begin a new clause, but Bliss argues that 'from the point of view of stress, the two verbs are treated as equivalent so that the stress of the second depends on the stress of the first'.20 The frequency of alliteration on the verb (31 out of 33) leads him to this conclusion. I think he is mistaken, and that the half-lines in group 2 are better regarded as beginning new clauses. 21 But let us accept his 16

' T h e battle-corslet shone, hard, h a n d - l i n k e d ,

t h e b r i g h t iron rings sang in t h e

armour . . . " 17

Bliss includes several b-verses in g r o u p 1 ( 1 0 5 b , 1 1 2 8 b , 1 2 3 1 b (misprinted as 1128a, 1231a), 2 3 4 4 b , 2 6 6 3 b , 2 8 9 9 b and 3 0 5 5 b ) . These have not been externally displaced. See below.

18

Bliss, Metre, § 13.

19

' H e experienced consolation for t h a t , flourished u n d e r t h e skies, prospered in honours

20

Ibid., § 14.

21

L e h m a n n , review of The Metre of Beowulf\ 1 3 9 , disputes Bliss's conclusion. H e argues that t h e alliteration reflects the fact t h a t 'in Indo-European clause structure, the verb was stressed w h e n initial in its clause.' See also L e h m a n n and Tabusa, Alliterations,

p p . 5—7.

This m i g h t be an explanation for t h e alliteration on initial verbs in the b-verse, where t h e resulting stress patterns never violate K u h n ' s laws (see below), a l t h o u g h h o w far Indo-European clause structure m a y have influenced t h e syntax of O l d English verse

50

Displacement

conclusion for the sake of the argument. Suppose that these 'displaced' half-lines in group 2 together with those in group 1 were removed from their present positions and put at the head of new verse clauses. Without exception every half-line in these two groups would then have to be put in group 3 or 7. All the a-verses of group 1 and the a-verses of groups 2, 3 and 7 with double alliteration consist of clause-initial, displaceable (Ib) halflines. (b) Only a sentence particle which already alliterates can be displaced. Alliteration

is the key to external displacement. If the single sentence particle does not alliterate, the clause-initial half-line remains nondisplaceable (la). Among the sentence particles, finite verbs, not surprisingly, are especially likely to alliterate. 22 Bliss calculates that alliteration marks 81 per cent of independent (that is, non-auxiliary) finite verbs which are not preceded by a stressed element in the same half-line. 23 The number of finite verbs in groups 2, 3 and 7 which do not alliterate is small. Bliss counts twenty-three in a- and b-verses combined. In the absence of alliteration these verses could never be displaced (= la). All the remaining a-verse members of groups 2, 3 and 7 (eighty-four in all) are displaceable (= Ib). As half-lines they are indistinguishable from the displaced verses in group 1. In the passage which I used for illustrative purposes in ch. 3 (lines 18881913) there is one verse with an alliterating finite verb before the first stressed element — fleat famigheals (1909a). This is therefore a clause-initial, displaceable (Ib) half-line. The verb fleat is in apposition to a displaced verb in the preceding verse; Bliss therefore assigns 1909a to group 2 (i.e., he considers//&*/ displaced). There are three similar half-lines in Beowulf in which a monosyllabic alliterating verb is followed by compound subject with the stress contour / x \ (709a, 1713a and 2183a): (1) ac he wseccende ^^bolgenmod

wrajmm on andan beadwa gej?inges.24

(708-9)

syntax is anybody's guess. Alliteration in t h e a-verse m u s t in any case be separately explained. 22

See Cosmos, ' K u h n ' s Law', esp. p p . 308—12.

23

Bliss, Metre, § 29- By way of comparison, t h e incidence of alliteration on nominals in t h e same position approaches 100 per cent. In Klaeber's text there are only three exceptions (316a, 2 6 1 5 a and 3056a). Nearly everyone agrees that the text is disturbed at these points.

24

'But he [Beowulf) lying awake in anger at the hostile one awaited, enraged, the issue of

the battle.' 51

The Metrical Grammar of Beowulf beodgeneatas, (2) breat bolgenmod eaxlgesteallan, oj? J?aet he ana hwearf, msere |?eoden mondreamum from . . , 25 (3)

naes him hreoh sefa, ac he mancynnes maeste craefte J?e him God sealde, ginfsestan gife, heold hildedeor.26

(1713-15)

(2180b-83a)

Bliss assigns 709a and 2183a to group 1 and 1713a to group 3. The present analysis makes possible a simpler and more comprehensive statement: all four half-lines are clause-initial, displaceable (Ib). Two have been displaced (709a and 2183a); two have not (1713a and 1909a). (c) Displacement has no effect on metrical contour. The metrical contour of a

half-line is independent of its position in the verse clause. A displaced half-line has the same contour, the same scansion, that it would have had, had it not been displaced. The alliteration of the sentence particle, which was extra-metrical before the displacement of the half-line, remains so afterwards. In this respect external displacement differs from internal displacement, where, as we have seen, the chief syllable of the displaced particle acquires metrical stress. External displacement is always marked by alliteration. Nevertheless, the alliteration is not part of the metrical contour. Take, for example, the undisplaced, clause-initial (Ib) half-line forbarn brodenmcel (1616a), which describes what happens to the sword which Beowulf used to kill Grendel's mother: sweord aer gemealt, I forbarn brodenmael; wses (?aet blod to (?aes hat, aettren ellorgaest, se J?aer inne swealt.27 (1615b—17) The verb forbarn is not a stressed element because if it were the clause would have an upbeat (for-) without a sentence particle in violation of Kuhn's second law. The fundamental alliterative principle assigns primary alliteration to the first syllable of the compound brodenmcel. Alliteration on the 25

26

27

'Enraged, he {Heremod] cut down his table-companions, his shoulder-companions, until he moved alone, the renowned chieftain, away from the joys of m a n k i n d . . . " ' H i s [Beowulf's] heart was n o t savage, b u t , brave in battle, he possessed t h e greatest strength of m a n k i n d , t h e great gift, which G o d gave h i m . ' ' T h e sword had melted, the wave-patterned sword had completely burned away; the blood was so h o t , the alien spirit was so poisonous, he w h o died there w i t h i n . '

52

Displacement

verb is quasi-mandatory because forbarn is not an auxiliary, but it is extra-metrical. Compare such clause-initial (la) half-lines as: wolde blondenfeax Itetad hildebord

beddes neosan her onbidan

(1791) (397)

All three half-lines have the same metrical contour: x x / x / (type B). Forbarn brodenmcel is a formula. Alliteration on the verb, which is the only sentence particle, permits the half-line to be externally displaced: forbarn brogdenmael, hatost heaj?oswata.28

Pa [>aet hildebil swa J?set blod gesprang, (I666b-68a)

The half-line could not have been displaced without the alliteration on the verb, but the metrical contour of the formula remains unchanged (type B). The alliteration which unlocks the clause-initial half-line, making displacement possible, is extra-metrical. Practically speaking, if a verse can be scanned without placing metrical stress on an initial particle, alliteration on the particle can be assumed to be extra-metrical, (d) The sentence particle in an a-verse displacement never occupies a metrical lift.

Extra-metrical double alliteration marks all a-verse displacements. Or, to put it the other way around, only those a-verses may be displaced in which the sentence particle alliterates with the syllable in the first lift. The first lift coincides with the second alliterating syllable. A displaced half-line scans like its undisplaced equivalent, and neither of them differs in scansion from a similar half-line with a sentence particle which does not alliterate. A sentence particle with extra-metrical alliteration may resemble the first alliterating lift of type A, D 1 or D 4 , though not of type E. Thus, the displaced (Ib) half-line, song in searwum (323a), with the metrical contour x x / x (type A 3 ), mimics type A (/ x / x). Indeed, if song acquired additional stress as a result of displacement, the Beowul]-poet may scarcely have distinguished the two rhythms. The distinction is crucial, however, when the sentence particle is accompanied by one or more proclitics. In order to accommodate these, prosodists have created a number of supplementary types with 'anacrusis' or internal 'expansion' which go far toward obscuring 28

'Then the battle-sword, wave-patterned, completely burned away, as soon as the blood sprang forth, hottest of battle-sweats.' 53

The Metrical Grammar of Beowulf the metrical order of Old English poetry. Take the displaced (Ib) half-line acigde ofcordre (312 la). It is the third in a clause made up of six half-lines: Huru se snotra acigde of corSre syfone (to)somne,

sunu Wihstanes cyniges )?egnas J?a selestan . . . 29

(3120—2)

Prosodists describe this as type A with anacrusis (x 11 / x x / x). 30 But it is simply type A 3 with the metrical contour x x x x / x and extra-metrical double alliteration. If alliterating sentence particles in initial position in a-verses were stressed elements, we might expect to find some type E's among them (e.g., trisyllabic verb + monosyllabic noun). There are none. 31 Furthermore, b-verses with sentence particles that have been assimilated to the stressed elements of the clause never display 'anacrusis' or 'expansion'. 32 (e) A displaceable (Ib) half-line conforms to both of Kuhn's laws prior to displacement. The displaceable (Ib) half-line contains just one sentence particle. As long as the half-line is undisplaced, the particle is in the upbeat of the verse clause. Alliteration on the particle permits the half-line to be displaced and frees the particle from the requirement that it be grouped with any other sentence particles in the upbeat. (f) A b-verse can not be displaced. External displacement is a function of the rule that permits double alliteration in the a-verse. Since the same alliterative rule prohibits double alliteration in the b-verse, it also prohibits displacement. Some b-verses look as though they had been displaced or could be displaced. These include all the b-verses in Bliss's groups 1, 3 and 7 which alliterate on the verb, for example: (Group 1: 'displaced'): wonsseli wer 29

30 31

32

weardode hwile

(105)

'Then the wise son of Wihstan summoned from the troop the king's thegns, seven together, the best . . . Pope, Rhythm, Line Index; Bliss, Metre, Index to t h e Scansion. There are no type E w i t h this syntactic pattern in t h e b-verse either. Presumably there is a different explanation for this. Possible exceptions are 1 5 2 5 b , 1663b and 2 7 1 7 b . In 1525b I take dolode cer to be a class two c o m p o u n d and p u t heavy stress on fela. See ch. 1 3 . Oft has been proposed for of tost in 1663b. For a thorough discussion of the problems connected w i t h 2 7 1 7 b , see Stanley,

'Verbal Stress', pp. 307-21.

54

Displacement

(Group 3: 'undisplaced'): faeger foldan bearm;

/ fundode wrecca

(1137)

(Group 7: 'undisplaced'): beaga bryttan,

/ breac (?onne moste

(1487)

I noted above a difference in behaviour between a-verses and b-verses. If alliteration on a verb in the a-verse were taken as evidence that it had been assimilated to the stressed elements of the clause, then a number of a-verses would violate one or both of Kuhn's laws. But alliteration on a verb in the b-verse never leads to a violation of either law. How can we account for this difference? Two of the most important distinctions between the a-verse and the b-verse are these: first, double alliteration is permitted in the a-verse, but not in the b-verse; second, alliteration is prospective in the a-verse, but retrospective in the b-verse. 33 Let us consider how these distinctions affect a sentence particle which can exhibit extra-metrical alliteration. When we, the audience, hear or see a finite verb, which is such a particle, at the beginning of an a-verse, we have no way of knowing, until we encounter the rest of the line, whether it alliterates, or, if it does alliterate, whether the alliteration is metrical or extra-metrical. But the same verb at the beginning of the b-verse leaves us in no doubt. The alliterative pattern has been established by the a-verse and we know immediately that the verb alliterates or that it does not. Furthermore, if it alliterates, we know that it is the sole alliterating word in the half-line and that it takes the first metrical stress. This is not to suggest that the scansion of one half-line determined that of the other, but rather that the poet had to choose a b-verse with a stressed verb in initial position if he wanted the verb to alliterate. What this means with respect to the metrical grammar is that an alliterating verb in the b-verse (like weardode, fundode, or breac in the verses cited above) is excluded from the initial dip of the clause. Therefore it is a stressed element, and the half-line will conform to all the requirements of Kuhn's laws. That initial verbs in the b-verse are stressed elements rather than sentence particles is confirmed by the fact that no b-verse begins a clause with a verb which is composed of proclitic + stem, whereas a-verses routinely do. In clause-initial position, the proclitic would violate Kuhn's 33

See Hoover, Theory of Old English Meter, pp. 111-12. 55

The Metrical Grammar of Beowulf second law if the verb were a stressed element, but does not as long as it is a sentence particle. One or more sentence particles in the first weak metrical position before or after an alliterating verb in the b-verse signal a clause-initial (I) half-line, in accordance with Kuhn's first law. Such are (where the first weak metrical position (italicized) precedes the verb): wsepen ond gewaedu, feorran cumene,

/ ic eow wisige

(292, also 3103)

/ pcet we fundiaj?

(1819)

and (where the first weak metrical position follows the verb): gryrelicne gist.

/ Gyrede him Beowulf

(1441)

All clause-initial (I) half-lines in the b-verse are nondisplaceable. In some discussions of stress and metre in Old English it is assumed that infinitives and participles, as well as certain adverbs, are stressed elements. 34 Without question, these forms typically alliterate when they are placed in initial position in the half-line, and they regularly receive metrical stress when they follow the first lift. Nevertheless, there is good reason for asserting that many of these forms should be regarded as sentence particles which have been internally or externally displaced. Neither adverbs, nor finite verbs, nor uninflected infinitives and past participles which are used as Verbals' 35 lead stressed elements in the a-verse. That is to say, they never occur where the metre or metrical grammar would require them to be natural stressed elements. Bliss, who regards verbals as stressed elements, nevertheless points out that 'the verbal is never initial [in the verse clause] in Beowulfr, though there are instances of initial verbals in hypermetric lines in other poems'. 36 In my view, since one of the characteristics of a stressed element is that it may be placed anywhere in the verse clause, the avoidance of initial placement seems an additional reason for regarding verbals as sentence particles. Like other sentence particles 34

For i m p o r t a n t and explicit discussion in addition to K u h n , see Bliss, Metre,

§ 9;

Campbell, Old English Grammar, §§ 93-9; and Heusler, Deutsche Versgeschkhte, pp.

105-13. 35

I a m following here D o n o g h u e ' s definition of a verbal as 'a d e p e n d e n t infinitive or past participle; present participles a n d inflected infinitives preceded by to are automatically excluded'. See D o n o g h u e , ' W o r d O r d e r ' , p . 169, n. 6.

36

'Auxiliary and Verbal', p p . 1 6 2 - 3 .

56

Displacement

they become stressed elements when they occupy the first alliterating lift of the b-verse, or when they are internally displaced. Two unstressed, non-alliterating infinitives appear before a first lift in Beowulf:^1 (1) Hyge waes him hinfus, wolde on heolster fleon, / secan deofla gedraeg; ne waes his drohtoS (?aer swylce he on ealderdagum aer gemette. 38 (2)

Beowulfe wearS guShreS gyfe^e; scolde Grendel f>onan feorhseoc fleon under fenhleoSu, / secean wynleas wic; wiste \>e geornor, (?aet his aldres waes ende gegongen . . . 39

(755-7)

(818b-22)

These are clause-initial, nondisplaceable (la) half-lines. In both cases, the infinitive is part of an elliptical construction which is in apposition to another clause with an auxiliary plus dependent infinitive. The auxiliary must be understood from the preceding clause. Normally, however, the auxiliary will appear in the initial dip of the clause and the infinitive or the participle will be displaced. So, for example, the infinitive brecan is dependent on the unstressed auxiliary let which appears in the half-line (2977a) which contains the initial dip of the clause in Let se hearda Higelaces J?egn brad[n]e mece, (?a his broSor laeg, ealdsweord eotonisc entiscne helm brecan ofer bordweal . . . 40

(2977-80a)

Since brecan is the sole sentence particle in 2980a and it alliterates, the halfline is clause-initial, displaceable (Ib) and, of course, it has been displaced. There were two such half-lines in the passage which I analysed in ch. 3: 37

These, and all similar unstressed infinitives in the corpus of O l d English poetry, are the subject of t h o r o u g h investigation by Stanley, in 'Verbal Stress', p p . 3 2 1 - 3 4 .

38

' H i s spirit was eager to get hence, he w a n t e d to flee to his hiding-place, to seek the c o m p a n y of devils; his way of life there was not such as he had m e t before in t h e days of his life.'

39

'Glory in battle was g r a n t e d to Beowulf; G r e n d e l had to flee from there, life-sick, to the lower part of the fen-slopes, to seek his joyless dwelling; he k n e w the more certainly that the end of his life had come . . . '

40

'The brave t h e g n of Hygelac let his broad sword (after his brother lay dead), his old giant's sword, break the h e l m e t m a d e by giants above the shield-wall . . .

57

The Metrical Grammar of Beowulf (1) Pa waes on sande saegeap naca hladen herewaedum hringedstefna . . . (2) drefan deop waeter,

(1896-7)

Gewat him on naoz Dena land ofgeaf.

(1903b-4)

The participle hladen is dependent on wees in 1896a, and the infinitive drefan is dependent on Gewat in 1903b. Both half-lines are displaced. Precisely the same observations may be made of adverbs. 41 Even those adverbs which always either alliterate or take metrical stress are better analysed as sentence particles which are subject to internal or external displacement. So, for example, I mark as extra-metrical (a) the alliteration on the initial adverb up in 2893a: Heht 8a )?aet heaSoweorc a

to hagan biodan

A

up ofer eegclif, J?aer (?aet eorlweorod morgenlongne daeg modgiomor saet . . , 42

(2892-4)

3

This is a clause-initial, displaceable (Ib) half-line (type A ), which has been displaced. Verb + adverb phrases in the a-verse must be stressed by the transformational rule. If the adverb is too far to the right, 43 the verb will receive the alliterating stress (e.g., Ymbeodepa (620a): see above, ch. 2). Otherwise the adverb takes the alliterating stress. But in this case extra-metrical alliteration on the verb appears to be mandatory, and if the verb is the only other sentence particle in the half-line, the half-line will be clause-initial, displaceable (Ib). Such are: sigon aetsomne,

op J>aet hy [s]ael timbred

Snyredon aetsomne,

(?a secg wisode

Jpreatedon J>earle. lemede to lange; 41

See C a m p b e l l , Old English Grammar,

(307) (402)

Ic him )?enode

(560)

he his leodum wearS

(905)

§ 9 4 : 'Adverbs are stressed like verbs: while lower

in stress t h a n nouns and adjectives, they can have fully weak stress only w h e n placed early in the sentence, before or after the first metrical stress . . . ' 42

' T h e n he ordered t h e results of t h e battle to be announced to the company u p above the sea-cliff, where the noble band sat, sad-hearted, all the long m o r n i n g . '

43

T h a t is, if it occupies the final X - p o s i t i o n . For the definition of X-positions, see below, ch. 5.

58

Displacement

Lines 307a and 402a appear at the head of their clauses; lines 560a and 905a have been externally displaced. Compound adverbs like hrcedlice and anunga are best taken as stressed elements. Thus, line 356a is type C with extra-metrical alliteration on the initial verb in Hwearf J?a hraedlice (?aer HroSgar saet eald ond ^nhar mid his eorla gedriht . . . 44

(356—7)

It is clause-initial, non-displaceable (la). 44

'Then he turned quickly to where Hrothgar sat, old and white-haired, with his retinue of warriors . . . "

59

Stressed proclitic adjectives: X-positions and the insertion rule

As heirs of a tradition of accentual—syllabic poetry that is now some six centuries old, we take for granted, first, that poets writing in English normally arrange the natural stress patterns of the language to accord with an abstract metrical pattern and, second, that they often wrest these patterns - either of natural stress or of metre - into other shapes when it suits their purposes. This means that we expect lexically significant words, such as nouns and adjectives, to receive metrical stress, in contrast to proclitics and other function words, which typically do not. In this respect the modern tradition of English poetry differs little from the Old English. But we are not surprised to find distortions of natural stress to fit the metre - little words unexpectedly emphasized, normally significant words subordinated — or to find the metre bent to accommodate the natural rhythms of an oddly placed word. There is a constant tension between the demands of speech and the demands of metre which is variously resolved according to the emphases which poets choose to impose. This licence is alien to the tradition of Old English poetry. When words which are not ordinarily stressed elements and which are neither externally nor internally displaced take metrical stress and alliteration in Beowulf, it will not do to speak casually of the poet's placing rhetorical emphasis on normally unemphatic words. Even here, he is usually working within a describable set of conventions and often with traditional formulae. Possessives, indefinites, demonstratives and definite articles are proclitic adjectives.1 These are forms which are normally weakly stressed and therefore do not alliterate. However, some proclitic adjectives in Beowulf are stressed, even though they have not been displaced or detached from 1

See below, ch. 10. 60

Stressed proclitic adjectives

their proclitic position. Stressed proclitics can be recognized by the fact that they are a necessary part of the metrical cadence: without them the verses could not be scanned. When we first hear of Beowulf, the poet decribes him in heroic terms: se waes moncynnes maegenes strengest on J?aem daege bysses lifes, ael?ele ond eacen.2

(196-8a)

Alliteration in line 197 can only be on the demonstratives pcem and pysses, despite the fact that demonstratives are proclitics which we would not expect to be stressed at all. Stressed proclitics like these, which have not been moved from their regular position immediately before the nouns they modify, always alliterate. The alliterative patterns of the half-lines in which they occur vary, depending on their metrical contours and on whether the half-lines are found in the a- or the b-verse. The purpose of this chapter is to investigate how the metrical grammar accommodates such deviations from customary usage. There are sixty-seven half-lines in Beowulf with what appear to be proclitic adjectives in stressed position. These are (with the stressed proclitics italicized): a-verse

(type A: 8X with double alliteration) A

A

mine magas to metodsceafte (2815; also 5a, 75a, 1178a, 1185a, 1984a, 2001a and 2103a) (type E: IX with double alliteration) A

A

Dys dogor \>u

ge)?yld hafa

(1395)

3

(type A : IX with double alliteration) a

AN

Sicgean ofer pa niht.

PrySswyS beheold

(736)

(type B: 4X with double alliteration; 3X with single alliteration) A

(Ax)

haefde mare maegen. A

Pa hine on morgentid (518; also 337a, 1300a and 2879a) N

Naefre ic cenegum men 2

aer alyfde (655; also 1086a and 1583a)

'He was the strongest in might of mankind in that day of this life, noble and mighty.'

61

The Metrical Grammar of Beowulf

(type C 2 : IX with double alliteration; 3X with single alliteration) a

AN

Meaht 3u, min wine, A

mece gecnawan

(2047)

(N)

on pcem daege

J?ysses lifes

(197; also 790a and 806a) b-verse (type A: 27 X ) A

N

ond f>egna gehwam para leoda (2033; also 197b, 219b, 605b, 634b, 776b, 790b, 806b, 837b, 978b, 1011b, 1673b, 1955b, 2095b, 2131b, 2181b, 2548b, 2667b, 2670b, 2685b, 2797b, 2804b, 2813b, 2837b, 2849b, 2885b and 3101b) (typeD1: 3 X ) A N

maerum J?eodne

min aerende (345; also 270b and 502b)

(type D 4 : I X (actually type E, see ch. 12)) (Ax) N

snotra fengel;

sume worde het

(2156)

(typeE: IX) A

sySSan orwearde

N

cenigne dad

(3127)

(typeB: 11X) A

N

eafe|?um stepte, ofer ealle men (1717; also 353b, 503b, 510b, 534b, 558b, 1099b, 2297b, 2493b, 2532b and 2555b)

(typeC1:2X) A N

J>eoden Scyldinga,

on/^zhealfe

(1675)

A N

l?egnes |?earfe,

swylce py dogore

ecghete eoweS,

ac him eal worold

(1797)

2

(typeC : I X ) A

(N)

(1738)

I begin with the assumption that all the proclitic adjectives in these verses are normally unemphatic words which have become stressed elements by special emphasis (analysis will remove a few from this category). They alliterate in accordance with the fundamental alliterative 62

Stressed proclitic adjectives

principle, which does not, however, account for the distribution and alliterative patterns of the verses. One unusual feature of the distribution of these verses is that all type A, with single alliteration, occur in the b-verse (27 X). Type A appears in the a-verse only with double alliteration (8X). Type B, on the other hand, occurs in the a-verse both with double alliteration (4X) and with single alliteration (3X), as well as with single alliteration in the b-verse (11X). It would appear, to confine ourselves for the moment to these types A and B, that some are limited to the a-verse (types A and B with double alliteration), others to the b-verse (type A with single alliteration), and that still others are unrestricted (type B with single alliteration). That verses with double alliteration are excluded from the b-verse is a basic condition of the metrical grammar of Beowulf. But single alliteration is perfectly acceptable in either half of the line. We might expect to find any of the b-verses listed above in the first half of the line, yet we do not. In this group, there are verse formulae and metrical types which are found exclusively in the b-verse. We have to inquire whether this is merely a statistical quirk or if it reflects limiting conditions for which descriptive rules can be formulated. In order to answer this question, I find it useful to introduce the notion of the X-position. Primary alliteration, that is, the alliteration generated by the fundamental alliterative principle, cannot come too far to the right in the half-line. There is always a minimum number of syllables which must follow the syllable bearing primary alliteration. If any one of this minimum number of syllables is stressed, it may be said to occupy an X-position, where the X-position excludes or limits the possibility of primary alliteration. There are two X-positions. (1) The final X-position (from which primary alliteration is always excluded 3) may be occupied either by a stressed final syllable or by a stressed short penultimate syllable. (2) The non-final X-position may be occupied either by a stressed long penultimate syllable or by a stressed short antepenultimate syllable. Primary alliteration is found on a stressed syllable in the non-final X-position only in the a-verse, and only then when the syllable is preceded by at least one sentence particle (that is to say, in type A 3 ). In all other circumstances primary alliteration is excluded from the non-final Xposition in the a-verse as well as the b-verse. 3

The exceptions to this statement in Beowulf are the six so-called 'short type A 3 ' (262a, 459a, 779a, 1514a, 1728a and 2048a).

63

The Metrical Grammar of Beowulf

In practical terms what this means is that, with the exception noted above, if a stressed syllable occurs in an X-position, there must, be another syllable which can take primary alliteration somewhere to its left. For example, the stressed syllable grund occupies the final X-position, and primary alliteration falls on gynne in the a-verse of under gynne grund,

Geata cempa

(1551)

The stressed syllable wu- occupies the final X-position, and primary alliteration falls on wynleasne in the a-verse of wynleasne wudu;

waster under stod

(1416)

The stressed syllable tru- occupies the non-final X-position, and primary alliteration falls on wi&res in the b-verse of wlonces wigcraeft;

wiSres ne truwode

(2953)

THE ALLITERATIVE RULE OF STRESSED ELEMENTS

With the help of the notion of X-positions, it is possible to formulate an alliterative rule for stressed elements which applies prior to the insertion of the stressed form of a normally unemphatic word: In any half-line the stressed syllable of the leftmost natural stressed element (if any) always receives metrical stress. This syllable will alliterate by the fundamental alliterative principle unless it occupies an X-position. In an X-position in the b-verse it cannot alliterate; in the a-verse alliteration is mandatory in the non-final X-position; it is optional in the

final X-position. Notice that the alliterative requirements of the last sentence in this rule apply only to half-lines which lack a stressed element outside of the X-positions. They do not affect verses like laSum daedum,

J?eah him leof ne waes

(2467)

Here the leftmost stressed element is laftum and dcedum need not alliterate, even though its stressed syllable falls into the non-final X-position in the a-verse. THE INSERTION RULE

We can also formulate at this point a rule for the insertion of proclitic adjectives into metrically stressed positions. The insertion rule may be stated thus: In the absence of a natural stressed element to its left, a proclitic 64

Stressed proclitic adjectives adjective becomes a stressed element when the stressed syllable of its base word occupies an X-position which does not permit primary alliteration. Once insertion takes

place the fundamental alliterative principle applies to the proclitic adjective. It may be asked whether putting stress on a proclitic (or on a sentence particle) alters its nature so that it is just like any other stressed element. The answer seems to be yes and no. Treating the stressed proclitics as stressed elements does not introduce any violations of Kuhn's second law. However the alliterative behaviour of a natural stressed element in the same half-line is affected by the insertion rule. And half-lines which begin with a stressed proclitic, which can in theory be considered unrestricted (III), are not in fact found in clause-initial position. Consider the b-verse of A

J?aes 3e ic moste

N

minum leodum

(2797)

The alliterative stress on minum would seem to make this half-line unrestricted (III). Yet there are twenty-eight such verses with the base word in X-position, and none of them begins a clause. 4 If we had more evidence to go on, it might be that some such verses would be found at the head of their clauses, but it seems likely that the regular placement of verses with initial unstressed proclitics in non-initial position in the verse clause would influence the placement of these verses. Now we can return to the list of half-lines with stressed proclitic adjectives. The base word immediately follows the italicized proclitic and in every case but one it is the leftmost natural stressed element. At least three logical consequences, which have a bearing on the empirical observations set forth above on the distribution and alliterative patterns of these verses, are implicit in the two rules. (1) Half-lines with a single natural stressed element, the stressed syllable of which is in the final X-position, move freely between the a- and the b-verse, unless there is some other limiting factor (such as a type A 3 metrical contour or double alliteration). Twenty-five of the sixty-seven half-lines have base words in the final X-position. They are (with the base word italicized): with alliteration on the base word (final X-position: final syllable) A

A

ofer min gemet 4

maeges helpan (2879; also 337a and 1300a)

See above, b-verse, type A (27X), type E (IX).

65

The Metrical Grammar of Beowulf

without alliteration (final X-position: final syllable) A

N

Naefre ic aenegum men A

aer alyfde

(655)

N

ond oSer swylc A

ut offerede

(1583)

N

Sicgean ofer J?a niht. PrySswyS beheold (736; also 353b, 503b, 510b, 534b, 558b, 1086a, 1099b, 1717b, 2297b, 2493b, 2532b, 2555b and 3127b) with alliteration (final X-position: short penultimate syllable) A

(Ax)

haefde mare mcegen.

Pa hine on morgentid

(518)

without alliteration (final X-position: short penultimate syllable) A

(N)

Meaht 3u, min wine, A

mece gecnawan

(2047)

(N)

on t?2em dcege

|?ysses lifes (197; also 790a and 806a) A

(N)

ecghete eoweS, ac him eal worold (1738) Since ondo&erswylc (1583a) lacks an untransformed stressed element, the rule for stressed elements does not apply to it. 5 The rule does apply to all the other verses and correctly assigns metrical stress to the stressed syllable. Alliteration is not required because the stressed syllables are in the final X-position. The insertion rule applies to all the half-lines, including 1583a, since its operative conditions obtain: there are no stressed elements to the left of the proclitics and the base words occupy an X-position. All the proclitics become stressed elements by the insertion rule and they alliterate by the fundamental alliterative principle. Two rather odd-looking verses in this group call for special comment. (1) Meaht du, min wine, mece gecnawan, J?one \>'m feeder to gefeohte baer under heregriman hindeman si5e, dyre iren, )?aer hyne Dene slogon, weoldon waelstowe, sySSan WiSergyld lseg, aefter haelej>a hryre, hwate Scyldungas?6 5 6

(2047-52)

I take swylc to be enclitic. 'Can you, my friend, recognize the sword that your father under his war-visor carried into battle for the final time, the precious iron, there where the Danes, the bold Scyldings, slew him and possessed the slaughter-field, once Withergyld lay dead after the fall of the

66

Stressed proclitic adjectives

Pope gives full stress to the auxiliary meaht and scans 2047a as a type D* 2. If the auxiliary is in fact a stressed element, which in my view is not possible, then the insertion rule would not apply, and the scansion would seem to be ad hoc. However, the auxiliary is grouped with another sentence particle (du) in the initial dip of the clause. I scan 2047a as a type C 2 (as does Bliss) with extra-metrical alliteration on meaht. The verse is clause-initial, nondisplaceable (la). These are the words which Beowulf puts into the mouth of an old warrior at the court of the Heatho-Bards in his imagined reconstruction of the probable course of events after Freawaru's marriage to Ingeld. One of Freawaru's retainers will wear a sword stripped from the body of a Heatho-Bard warrior in an earlier battle. An old Heatho-Bard will point out the sword to the dead warrior's son. The alliterative crescendo in the old warrior's speech that builds from the lightly stressed meaht to the insinuating mm wine clinched by mece — the sword that belonged to the father - seems grimly appropriate to this classic scene of goading. (2) J?aet he ma moste

Ne waes J?aet wyrd )?a gen, manna cynnes

dicgean ofer pa niht.

maeg Higelaces, under faergripum

PrySswyS beheold

hu se manscaSa gefaran wolde.7

(734b-38)

736a is an unusual verse. Bliss and Pope agree that it is a type A. As with Pope's scansion of 2047a, scanning it this way blocks application of the insertion rule and leaves the second alliterating stress on the proclitic unexplained. For reasons which have already been discussed, I believe that infinitives, like dicgean, are sentence particles rather than stressed elements. 8 Extra-metrical alliteration on picgean, which is the sole sentence particle, permits the half-line to be displaced. Therefore I scan 736a as a type A 3 with primary alliteration on pa and secondary stress on niht. It is a clause-initial, displaceable (Ib) half-line, which has been displaced. The extra-metrical alliteration on picgean probably gives it at least as much prominence as the primary alliteration does to pa: 'it was not his fate that he might consume more of the race of men after that night 7

8

'However it was not [Grendel's] fate, that he might consume more of the race of men after that night. The mighty kinsman of Hygelac beheld how the evil-destroyer intended to proceed with his sudden grips.' See above, ch. 4.

67

The Metrical Grammar of Beowulf Half-lines with double alliteration, either regular (337a, 518a, 1300a and 2879a) or extra-metrical (736a and 2047a), are restricted to the a-verse. So too is any half-line whose primary alliterative syllable would fall into an X-position in the b-verse (namely, type A 3 : 736a if this is properly so-scanned). In terms of the rules set forth here, nothing would prevent any of the remaining half-lines from appearing in either the a- or the b-verse. Most of them have some features in common. Ncefre ic cenegum men (655a), for example, looks like a formulaic variant of Ne inc cenig mon (5 10b). Pending other restrictions (such as might apply to cenigne dcel, 3127b), we can label these half-lines as common a/b-verses. (2) A b-verse which has a stressed element with its stressed syllable in the non-final X-position and which scans by virtue of the insertion rule cannot be moved to the a-verse, because in the a-verse double alliteration would be mandatory. It belongs to the b-verse in the same way that half-lines with double alliteration belong to the a-verse. Examples of half-lines which fall into this category are (with the base word italicized): (non-final X-position: long penultimate syllable (27 X) 9) N

A

J?ass Se ic moste

minum leodum N

A

unbyrnende

(2548)

aenige hwile N

A

o(?res dogores10

oS J?aet ymb antid t?eoden Scyldinga,

N

A

on J?a healfe A N

:swylce J?y dogore11

(?egnes (?earfe,

(2797)

(219) (1675) (1797)

>osition: short antepenultimate syllable (2X)) aeSeling anhydig,

(Nx)

A

ealle magene A

N e gefraegen ic J?a maegj?e

(Nx)

maran weorode

(2667) (1011)

Without the two proposed rules, there is no obvious reason why half-lines like these should not appear in the a-verse. Let us see how the rules account for this restriction. In each case the base words are the 9

10 11

197b, 219b, 605b, 634b, 776b, 790b, 806b, 837b, 978b, 1673b, 1675b, 1797b, 1955b, 2033b, 2095b, 2131b, 2181b, 2548b, 2670b, 2685b,2797b,2804b,2813b, 2837b, 2849b, 2885b and 3101b. Syncopating the medial syllable of dogores. Syncopating the medial syllable of dogore.

68

Stressed proclitic adjectives

leftmost stressed elements. Their stressed syllables all take metrical stress in accordance with the rule for stressed elements. Primary alliteration is blocked because the stressed syllables occupy an X-position in the b-verse, and since the X-position is in the b-verse no secondary alliteration is possible. The conditions which permit the insertion rule to apply obtain. In the absence of any stressed elements to their left, the proclitics are raised to that rank. The fundamental alliterative principle requires that these alliterate. The predicted results correspond to the facts in every instance. Now let us try to analyse the same half-lines as hypothetical a-verses. In the a-verse the non-final X-position requires alliteration. A stressed syllable in these positions must alliterate. In order to determine whether the insertion rule for proclitic adjectives comes into play, we must see whether or not this stressed syllable can take the first metrical stress. If it can, then the half-line scans properly and the rule is unnecesssary. If not, the rule is required. There is only one metrical contour in the a-verse — type A 3 — where the first metrical stress falls on a long penultimate or a short antepenultimate syllable, coinciding with the non-final X-position. Type A 3 , being clause-initial (I), requires at least one sentence particle in its weak onset. With one exception, none of these half-lines has such a sentence particle. Therefore the stressed syllable could not take primary alliteration, and the insertion rule would assign stress and, by the fundamental alliterative principle, primary alliteration to the proclitics. But, since the base words have already been marked for alliteration and none of the proclitics alliterates with them, it follows that the proclitics cannot alliterate. Therefore these half-lines must, by their structure, be confined to the b-verse. The sole exception is swylcepy dogore (1797b). The sentence particle swylce theoretically makes possible a type A 3 metrical contour with alliteration on the d of dogore rather than the p of py.12 Conceivably this half-line could be used as an a-verse. I think this is doubtful on other grounds. 13 (3) An a-verse in which the stressed syllable of a stressed element is in the non-final X-position and which scans by virtue of the insertion rule will exhibit double alliteration. The reasons for this have perhaps become obvious from the preceding discussion, but they should be made explicit. The a-verses in question are these (with the base word italicized): 12

13

Klaeber (following principles enunciated by Sievers, 'Zur Rhythmik', pp. 459—61) syncopates the medial o of dogore; Sievers, 'Zur Rhythmik', p. 233, prefers to read dogor. See the discussion of 1395a, below, p. 71.

69

The Metrical Grammar of Beowulf (non-final X-position: long penultimate syllable) A

A

sinne geseldan (Ax)

in sele J>am hean

(micel) gemeting, (Ax)

(1984; also 2815a)

A

monegum

A

monegum magpum

fira

(2001)

meodosetla ofteah (5; also 75a, 1178a and 2103a)

(non-final X-position: short antepenultimate syllable) A

(Ax)

uncran eaferan,

gif he J?aet eal gemon

(1185)

According to the rule for stressed elements proposed above, the stressed syllable of a stressed element which falls into the non-final X-position in the a-verse must alliterate, if there is no other such word to its left. This is the case in all of these verses. The insertion rule puts stress on the proclitic adjectives which then display primary alliteration according to the fundamental alliterative principle. The result is mandatory double alliteration. We began with a total of sixty-seven half-lines. With the help of the proposed rules for stressed elements and for the insertion of stressed proclitics we have correctly scanned and labelled as a-verse, b-verse or common, without reference to context, all but five. Let us turn to them. They are (with the base word italicized): A

A

Dys dogor \>u

(Ax)

modges merefaran,

N

micel afyunca (Ax)

HabbaS we to J?aem maeran maerum (?eodne

N

micel cerende A

N

m i n cerende (Ax)

snotra fengel;

(1395)

ge^yld hafa

N

sume worde het

(502) (270) (345) (2156)

All share one characteristic: the stressed syllable of the base word does not occupy an X-position. In the absence of a stressed element to the left the base word should alliterate, but it does not. In each case the primary alliteration falls on the proclitic. We cannot rule out the possibility that here the Beowulf-pott deliberately put emphatic rhetorical stress on normally unstressed words instead of making use of verse formulae that supplied stressed proclitics in accordance with the insertion rule. If so, these proclitics would be, exceptionally, the 70

Stressed proclitic adjectives

equivalent of stressed elements and the rule for such words would apply to them rather than to the italicized forms because they are farther to the left. Nor can we rule out the possibility that the rules cannot be formulated so strictly or that they did not operate so consistently as I have implied. All grammars have areas where usage varies or is in the course of transition. However I think the exceptions, if that is what they are, do not really tell against the rules as proposed. The only a-verse among them is Dys dogor pu (1395a), which is the initial half-line of its clause. If primary alliteration were carried by dogor, and Dys were an unstressed proclitic, the half-line would violate Kuhn's second law which requires that the upbeat of the clause, if there is one, contain at least one sentence particle. Dogor derives from dceg with (in Old English) scarcely differentiated meaning, but limited usage. In poetry it is mostly found in a few formulae or formulaic systems, e.g., opres dogores,l4 dogragehwilcne15 and . . . dogra gehwam.16 The

fact that in two of its ten occurrences in Beowulf, dogor follows a stressed demonstrative in the dative or instrumental case suggests that there may have been a quasi-compound pys-dogor lpy-dogor{e) with alliterating stress on its first element. Compare pam-dcege and pa-niht.ll Two of the b-verse exceptions involve micel (270b and 502b). This is one of those forms which seem to lie in a grey area between the weakly stressed proclitics and the fully stressed adjectives (see ch. 10). If the X-positions for the b-verse are correctly defined, then micel leads its base word in these two verses and is here fully stressed. 18 The half-line micel cerende (270b) undoubtedly explains min cerende (345 b). If we consider 345 b in isolation, there is no obvious reason why the alliteration should not be vocalic, with min remaining an unstressed proclitic. We should expect it to scan as a type C 1 : A

*min aerende Compare: J?onne we geheton ussum hlaforde (2634) However it is part of a repetition within the space of seventy-five lines of Beowulf's formal request for an audience with Hrothgar. To the coastguard Beowulf says: 14 17 18

15 16 Beowulf 219b etc. Christ and Satan 24 la etc. Christ 428a etc. Beowulf 197a etc., Beowulf736a. It may of course be fully stressed elsewhere, as I indicate in ch. 10. I prefer to mark it where possible as a proclitic to reflect the fact that it functions like a proclitic.

71

The Metrical Grammar of Beowulf HabbaS we to J?aem maeran Deniga frean . . . 19

micel arende (270-la)

And, once past the coastguard, he uses the same courtly language to Wulfgar: Wille ic asecgan sunu Healfdenes, maerum (?eodne min cerende, aldre(?inum . . . 20

(344-6a)

It is still a great (mice/) errand to the illustrious (mceran) Hrothgar, but now it is also, with a shift to the individual heroic ethos (Beowulf has just named himself), my (min) errand to your (pinum) lord. The stress pattern of the earlier phrase is still echoing in our ears. This is surely a case where the Beowulf-pott chose to put rhetorical stress on the proclitic. The last exception, sume worde het (2156b), can be compared in one respect with 1395a. It is the initial half-line of its clause. If sume were the weakly stressed proclitic, the half-line would violate Kuhn's second law. Curiously, the adjective sum rarely appears in Old English poetry. 21 This is its sole occurrence in Beowulf. Except for the Metres of Boethius (16X), which closely follows an older prose translation, 22 the adjective is found in only ten other verses. What little evidence there is suggests that the adjective, like the pronoun, 23 has two distinct ranges of meaning. As an indefinite it has more or less the meaning of Modern English 'a' or 'an'. The indefinite is almost always left unexpressed; when sum is used it is a weakly stressed proclitic. 24 When it is not the indefinite, sum must mean something like 'great', 'notable', 'special' and be fully stressed. There is a passage in Christ which begins: Eac we J?aet gefrugnon, J?aet gefyrn bi J?e soSfaest ssegde sum woSbora in ealddagum, Esaias . . . 19 20 21

22 23

24

(Christ 301—3)

'We have a great errand to the famous lord of the Danes . . . 'I wish to tell Healfdene's son, the famous chief, your lord, my errand . . . On sum, see Mitchell, Old English Syntax, §§ 3 8 5 - 4 1 0 ; Rissanen, ' "Sum" in Old English Poetry', pp. 197-225. The Paris Psalter and the Meters of Boethius, ed. K r a p p , A S P R 5 , xlvii-xlviii. See W r e n n - B o l t o n , glossary; Barney, Word-Hoard, p p . 3 9 - 4 0 ; Mitchell, Old English Syntax, § 4 0 1 . Cf. Genesis 4 3 2 a , Christ 3 1 8 b , The Fortunes of Men 7 1 b and Solomon and Saturn 2 3 0 b .

72

Stressed proclitic adjectives

This should be translated not, 'we have heard that long ago a prophet, Isaiah, uttered the truth about you . . . ', but, 'we have heard that long ago the great prophet, Isaiah, uttered the truth about you . . . ' Similarly in The Battle of Maldon the poet says of the hostage who fought on the side of the English: aefre embe stunde he seald sume wunde, t?a hwile Se he waepna wealdan moste. (The Battle of Maldon 271-2)

This may be rendered, 'from time to time, as long as he could manage his weapons, he dealt out a great wound'. 25 So, in Beowulf, line 2156b is part of Beowulf's report to Hygelac of Hrothgar's munificence. After commanding Hrothgar's gifts to be brought in, Beowulf says: Me Sis hildesceorp HroSgar sealde, snotra fengel; sume worde het, t?aet ic his aerest 5e est gesaegde . . .

(2155—7)

The stress on sume indicates that we should translate this: 'Hrothgar, the wise prince, gave me this armour; he particularly ['in a special word'] insisted that I should first tell you its history.' The point of formulating the insertion rule is not to deny that the Beowulf-pott sometimes put emphatic stress on normally unimportant words, but rather to insist that, when he did so, he almost always did it in a traditional way, that is, in accordance with the rules of his metrical grammar. This is only another way of saying that his poetic language was at once remarkably ordered and enormously supple. 25

Unfortunately for our purposes, 271 is a rhymed rather than an alliterative line so it is difficult to be certain of the stress on sume. The use of some with full or emphatic stress survives in Modern English. Compare, for example: 'The doctor treated some wounds' (unstressed) with 'Those are some wounds!' (stressed); and see Rissanen, '"Sum" in Old English Poetry', p. 225, nn. 27 and 28, who gives other examples and points out that this usage is 'especially American'.

73

Problems with the identification of clause-non-initial half-lines: the proclitic onset

The three kinds of half-lines, clause-initial (I), clause-non-initial (II) and clause-unrestricted (III), which the Beowulf-poet had in his metrical grammar are for the most part clearly differentiated by their respective onsets. 1 It could not have been otherwise in the tradition of oral composition out of which his metrical grammar developed. The stock of formulae and formulaic systems which comprised the vocabulary of the scop would have had to have been differentially marked as initial, non-initial or unrestricted to be useful in rapid, extemporaneous composition. When, occasionally, we find a half-line which is not readily identifiable, the problem may be in the transmission of the text, or it may be that we are not sufficiently alert to the way proclitics and sentence particles were marked and the way the half-lines were employed. The purpose of this chapter is to investigate several problems that come up in the identification of non-initial (II) half-lines. Because all non-initial half-lines have a weak onset, they correlate with the Sievers types in a predictable way. With a very few exceptions, which will be specifically discussed, they are either type B or type C. The weak onset of a non-initial half-line consists of one or more proclitics. Some examples may be helpful (the weak onset of the non-initial half-line is italicized): TypeBl: Het 3a Hildeburh hire selfre sunu 1 2

set Hnaefes ade sweoloSe befaestan . . . 2

(1114—15)

See above, ch. 3. 'Then Hildeburh ordered her own son to be committed to the flames at Hnaef's funeral pyre

74

Identification of clause-non-initial

half-lines

Type B2: waeron ySgebland eal gefaelsod, eacne eardas, (?a se ellorgast oflet lifdagas ondpas laenan gesceaft.3

(1620-2)

Type C1: Him 3a Scyld gewat to gescaephwile felahror feran on Frean waere . . . 4

(26-7)

TypeC 2 : Swa Wedra helm aefter Herebealde heortan sorge weallinde waeg; wihte ne meahte on 8am feorhbonan faeghSe gebetan . . . 5

(2462b-65)

I find it convenient to refer to the onset of a non-initial half-line as a 'proclitic onset' to distinguish it from the weak onset of an initial half-line which contains one or more sentence particles. A proclitic onset is a w l which precedes the first lift of a half-line and consists entirely of proclitics, as in the above examples. Although it would be possible to construct a proclitic onset five or six syllables in length, the limit in Beowulf is three and the poet's preference is for one or two. 6 Proclitics are not excluded from the weak onset of an initial (I) half-line. The onset of an initial half-line may have syllables which are proclitic on a stressed element or on a sentence particle or both, for example (only the proclitic syllables in the w l are italicized): wergan gastes;

waes \>czt gewin to strang

maeg ^lfheres;

geseah his mondryhten

forleton eorla gestreon 3

4

5

6

eorSan healdan

(133) (2604) (3166)

'The surging waters were entirely cleansed, the mighty dwelling, when the alien spirit departed the days of her life and this transitory world.' 'Scyld departed then at the fated time, very vigorous, to travel to the protection of the Lord 'So the protector of the Weather-Geats endured surging sorrow of heart for Herebeald; he could not in any way settle the feud with the life-slayer . . . ' In Beowulf, 308 half-lines could be said to have a monosyllabic proclitic onset; 219 a disyllabic onset; 33 a trisyllabic. Each of these figures will be reduced slightly in the discussion that follows.

75

The Metrical Grammar of Beowulf These are not proclitic onsets because they each contain a sentence particle (wees, geseah and forleton), and each of these half-lines (133b, 2604b and 3166a) begins a verse clause. Identifying half-lines which have proclitic onsets is generally a straightforward process. The major problem is caused by verse-initial verbs with unstressed prefixes. When such a verb alliterates, its prefix appears to constitute a proclitic onset. But this is not always so. A case in point is the a-verse with an alliterating verb preceding a nominal, such as: uton hra(?e feran7

Am, rices weard, awrcec wintrum frod.

(1390)

Wundor is to secganne

forwrat Wedra helm

wyrm on middan

(1724) (2705)

I scan these as type B with extra-metrical alliteration on the verb. Their onsets therefore are not a-, a- and for-, but arts, awrcec and forwrat (compare 2604b and 3166a cited above). Since they do not have proclitic onsets, these cannot be non-initial (II). On the contrary, they are initial, displaceable (Ib) half-lines. Lines 1390a and 2705a head their clauses; 1724a has been displaced. If we set aside all such a-verses on the ground that the alliteration on the sentence particle is extra-metrical, there are 560 half-lines with nothing but proclitics before the first lift. Based on the punctuation of Klaeber's text, thirty-six are, or appear to be, clause-initial, and thus seem to be in violation of Kuhn's second law. These thirty-six can be grouped according to the nature of the proclitics involved: Unstressed prefix (a) tfledon )?a Gecyste pa.

leofne J?eoden

(34)

cyning ae)?elum god (1870; also 652a, 2516a and 3156a)

0/ereode J?a

aej?elinga beam

(1408; also 2345a)

y ^ e o d e (?a

ides Helminga

(620)

(b) (remunde J?a se goda, Gefeng pa. be eaxle

maeg Higelaces - nalas for faehSe mearn

(758) (1537)

Preposition (c) wi<S wraS werod

wearde healdan

Klaeber underdots the h of hrape; alliteration is assumed to be on R.

76

(319)

Identification of clause-non-initial on sidne sae

half-lines

ymb sund

beadohraegl broden, wi& ord ond wiS ecge

flite

(507)

on breostum laeg

(552)

ingang forstod

(1549)

ofer ealle men

eafejmm stepte, lyftgeswenced,

(1717)

on lande stod

(1913)

Preposition plus prefix (d) For gtfwyrhtum (?u, on geB.it faran

wine min Beowulf fealwe mearas

0# geweald gehwearf

(457) (865; also 2570a)

woroldcyninga

(1684)

Preposition plus demonstrative (e) mid pare waelfylle

wica neosan

Ymb pas helmes hrof

(125)

heafodbeorge

Mt pcem ade wses

(1030)

e^gesyne

JEfterpczm wordum

(1110)

Weder-Geata leod

(1492; also 2669a)

Demonstrative (f) in Caines cynne — ' Done siSfset him

pone cwealm gewraec

(107; also 1202a)

snotere ceorlas

pone yldestan

oretmecgas

Dam wife J?a word

(202) (363; also 801b and 928a)

wel licodon

(639)

Other proclitic adjectives (g) min yldra maeg

unlifigende

(468)

eal bencj?elu

blode bestymed

(486)

idese onlicnes;

oder earmsceapen

(1351)

Proclitic adverb (h) To lang ys to reccenne,

hu i(c S)am leodsceaSan

(2093)

The half-lines in group (a) consist entirely of sentence particles. These are in the first dip of their respective clauses. That is to say, each half-line as a unit forms the clause upbeat. Indeed, any one of these half-lines could

form part of the wl of an initial (I) half-line. Compare: 77

The Metrical Grammar of Beowulf A

Ongunnon pa on beorge A

alegdon tSa tomiddes

baelfyra maest

maerne (?eoden

(3143)

(3141)

A

Gemunde <5a 3a are,

\>e he him aer forgeaf

(2606)

Only when the verb phrase stands alone, as it does in group (a), does the finite verb acquire metrical stress and alliteration through the transformational rule. 8 The half-lines in (a) are properly clause-initial (I) and do not, in fact, have proclitic onsets. Scribal blunders almost certainly account for group (b). Nowhere else in Beowulf does a finite verb lead a nominal in the a-verse. The emendations proposed by Rieger are compelling: 9 a

A

Gemunde J?a se modega, maeg Higelaces, aefenspraece, uplang astod ond him faeste wiSfeng; fingras burston; eoten waes utweard, eorl furfur stop. 10 a

(758-61)

A

Gefeng j>a befeaxe - nalas for faehSe mearn GuS-Geata leod Grendles modor; braegd \>& beadwe heard, J>a he gebolgen waes, feorhgeniSlan, J?aet heo on flet gebeah.11 (1537—40) Kuhn observed that a proclitic 'sometimes . . . stands alone in the upbeat in incomplete (lacking subject or predicate) clauses'. He added: 'Perhaps these clauses were felt only as sentence parts, or perhaps their beginning was not felt as the head of a clause, because important parts 8 10

11

9 See above, ch. 2. 'Die alt- und angelsachsische Verskunst', pp. 24 and 43. 'Then the proud one, the kinsman of Hygelac, remembered his evening-boast, stood upright and gripped him firmly; fingers burst; the giant was headed out, the noble stepped forward.' 'Then he seized Grendel's mother by the hair (not at all did he shrink from the feud), the man of the War-Geats; then brave in battle he flung his deadly foe, now that he was enraged, so that she crashed to the floor.' Chambers's note on this emendation deserves to be quoted in full: 'Rieger's emendation betters the alliteration, and has been adopted by Sweet, and by recent editors. Those who retain the reading goda in 1. 758 would however be justified in quoting that line as a parallel to gefeng pa be eaxle. To mefeaxe appears also to give better sense; but this may be disputed. Mr Wyatt writes: "William Morris agreed with me that it debased Beowulf's character, turning a wrestle into an Old Bailey brawl. Hair-pulling is a hag's weapon."'

78

Identification of clause-non-initial

half-lines

had to be completed from the preceding clause.' 12 All the half-lines in (c) are of this type, although Kuhn did not include 319a, 552b and 1913b — the two latter because these lines had been emended in the edition of Holthausen upon which Kuhn based his analysis. 13 When ellipsis deletes the subject or predicate of one of two adjoining clauses, the boundary between them is obscured, particularly when a medial phrase is appropriate to both. 14 No doubt, as Kuhn suggested, the telescoped clauses were sometimes felt as one. But sometimes a simple change in the punctuation supplied by modern editors removes the difficulty. I have argued in ch. 4 that the comma which Klaeber inserted after lyftgeswenced should be shifted back to follow geprang: ceol up ge(?rang, on lande stod.

lyftgeswenced

(1912b—13)

15

In this case repunctuation improves the sense. Likewise we can read (in each case either deleting or shifting the comma which in Klaeber's text immediately precedes the non-initial half-line): witS wrad werod

Ic to sae wille wearde healdan.16

Waes merefixa mod onhrered; (?aer me wiS laSum licsyrce min heard hondlocen helpe gefremede, beadohraegl broden on breostum Iceg, golde gegyrwed.17

12

13 14

15

16

17

(318b-19)

(549-53a)

'SchlieBlich steht auch in unvollstandigen (subject- oder pradicatlosen) satzen manchmal eine satzteilpartikel allein im auftakt . . . Diese satze sind vielleicht nur als satzteile empfunden, oder ihr anfang wurde nicht als satzspitze empfunden, weil wichtige teile aus dem vorausgehenden satze zu erganzen waren' (Kuhn, p. 44). Beowulf, 6th ed. (Heidelberg, 1929); see Kuhn, p. 4. On this kind of euro xoivou construction, see Meritt, Construction, p. 16; Mitchell, Old English Syntax, §§ 3789-803; Donoghue, Style, p. 50. Sievers, 'Zum Beowulf, p. 141, proposed pat he on lande stod. Holthausen adopted this emendation. Sic Dobbie, Wyatt-Chambers. 'I will return to the sea to stand guard against hostile bands.' Sic Dobbie, Wyatt—Chambers, Wrenn—Bolton. 'The anger of the sea-fishes was aroused; there my mail-shirt, hard, hand-linked, aided me against the hostile ones, the woven battle-garment lay on my breast, adorned with gold.'

79

The Metrical Grammar of Beowulf Him on eaxle laeg breostnet broden; (?aet gebearh feore wid ord ond wiS ecge, ingang forstod. 18

(1547b-49)

breat bolgenmod beodgeneatas, eaxlgesteallan, op J?aet he ana hwearf, maere (?eoden mondreamum from, Seah J?e hine mihtig God maegenes wynnum, eafe(?um stepte ofer ealle men, forS gefremede.19 (1713-18a) Line 507a exhibits the same kind of telescoped construction. Repunctuating it to attach it to the preceding rather than the following clause does not solve the problem here because it creates a new violation of Kuhn's law in 507b: se |?e wi8 Brecan wunne Eart J?u se Beowulf, on sidne sa, ymb sund flite . . . 2 0

(506—7)

With the exception of Thorkelin, the older editors from Kemble to Grein took sund-flite as a compound noun meaning 'swimming contest'. Bugge objected to this reading on the grounds that sund-flite as a compound would have to be a feminine noun in the accusative case, even though^/V and geflit are only found as neuters, and that no language would use the expression 'fight in a swimming contest.' 21 Neither argument is decisive. A return to the older reading, which, after Bugge, editors abandoned, would have the advantage of removing all difficulties. In sum, it appears not to be necessary to label any of the half-lines in (c) as clause-initial. The ambiguity created by certain elliptical, consecutive clauses can be resolved by inspection. A prepositional phrase without any associated sentence particle must normally have been felt as non-initial, and that can be our guide. In group (d) we can, with Kuhn, set aside 457a, because its syntax is the 18

' T h e woven breast-net lay on his shoulder; it protected his life against p o i n t and edge, it prevented p e n e t r a t i o n . '

19

Sic Wyatt—Chambers.

'Angry-minded

he killed his table-companions,

shoulder-

companions, u n t i l he moved alone, t h e illustrious prince, away from t h e joys of m a n k i n d , a l t h o u g h m i g h t y G o d had advanced h i m in t h e joys of s t r e n g t h , of m i g h t , over all m e n , h a d p u s h e d h i m forward.' 20

'Are you that Beowulf w h o contended w i t h Breca on t h e broad sea, competed in s w i m m i n g . . . ?'

21

'Beowulf, p . 4 8 .

80

Identification of clause-non-initial

half-lines

product of editorial reconstruction. 22 Lines 865a and 2570a are infinitive phrases which vary preceding infinitive phrases. These lines could be analysed as consecutive clauses with ellipsis of the governing subject plus finite verb. Analysed this way, they violate Kuhn's law: Hwilum hea(?orofe hleapan leton, / on geflit faran fealwe mearas . . . 23

(864—5)

There are phrases which must clearly be marked in this way (e.g., 439b, 2126b, 2775a etc.). In such constructions, the question of whether the poet felt the successive phrases to be part of two clauses or one (and he must sometimes have felt one way and sometimes the other) can be decided as before by inspection. If the upbeat of the second clause contains a sentence particle, then it must be an elliptical consecutive clause; if not, it can be taken as a variation within a single clause. The latter is the case with 865a and 2570a, and we should read: Hwilum heaj?orofe on geflit faran —

hleapan leton— fealwe mearas . . . 24

Although Kuhn might have given 1684a (on gewealdgehwearf) as an example of a preposition standing alone in the upbeat of a clause without an expressed subject (he does not cite this verse), as I indicated in ch. 3, I find Kuhn's explanation of this class of exceptions dubious. The half-line must stand as a violation of his rule. 25 Groups (e) and (f) involve the demonstrative. Kuhn argued, rightly I think, that the demonstrative in Beowulf is not yet clearly separated from the relative. It refers back to the preceding clause and has something of the force of a conjunction. 26 In these lines, therefore, the demonstratives can be specially marked as sentence particles (however the relative function of the demonstratives in 363a, 639a and 928a is notably weak). 27 The validity of Kuhn's argument about the relative function of the demonstratives can be supported by a piece of indirect evidence. Among the half-lines listed in group (e) is the formula: 22

Kuhn, p. 44, n. 1. The manuscript reads: fere fyhtum \>u.

23

'At times t h e brave warriors let their horses gallop, t h e warriors let their bay steeds engage in contests . . . ' 'At times t h e brave warriors let their bay steeds gallop - engage in contests . . . ' See, however, m y discussion of this passage in ch. 3 . 27 Kuhn, pp. 44-5. 125a begins an elliptical consecutive clause.

24 25 26

81

The Metrical Grammar of Beowulf ^Efter baem wordum

(1492a and 2669a)

This formula appears elsewhere in Old English poetry, and is always clause-initial. 28 In terms of metre, it is a type A 3 . Since type A 3 half-lines are exclusively clause-initial (I), they always contain one or more sentence particles in their upbeats. Lines 1492a and 2669a would be the only exceptions in Beowulf to this statement, if the relatives were not marked as sentence particles. Thus by accepting Kuhn's argument we solve two problems at once. There are several half-lines elsewhere in Anglo-Saxon poetry built on the system after + demonstrative + object which point to some hesitation on the part of other poets as to whether the system was properly clause-initial (by virtue of the relative function of the demonstrative and being type A 3) or non-initial (by virtue of possessing a proclitic onset). In one poem in the Exeter Book the system is used both ways within a few lines: JEfter pare stefne stenc ut cymeS of |?am wongstede . . .

after pare stefne

{The Panther AA-A5&)

deor efne swa some on [>one stenc faraS.29 (The Panther 53b-54)

The relative function of pare in 44a is clearly present and the half-line is clause-initial; in 54a its relative function is not marked and the half-line is clause-non-initial. The weakening of the relative function may have contributed to the use of type A 3 with this syntactical pattern in non-initial position. So, to give one more example, the Lord says to Cain in Genesis A: mundum sinum Gif J?e monna hwelc aldre beneoteS, hine on cymeS after pare synne seofonfeald wracu, wite aefter weorce.30

(Genesis 1040—3a)

In group (g) editors punctuate 468a as though it were the beginning of an elliptical consecutive clause: 28

Exodus 2 9 9 a and 565a; JEfterpam

wordum'. Andreas 1219a; JEfter pyssum wordum:. Andreas

88a, 7 6 l a and 1026a. 29

'After that sound the smell comes out from that place . . .; the beasts likewise after the sound travel toward the s m e l l . ' I owe this example to Professor E. G . Stanley.

30

'If any m a n deprives you of life w i t h his h a n d s , there will come u p o n h i m for that sin sevenfold p u n i s h m e n t , t o r m e n t for the d e e d . '

82

Identification of clause-non-initial half-lines 3a waes Heregar dead, min yldra maeg [sc. waes} unlifigende, beam Healfdenes . . .

(467b-69a)

But everything following 467b is a simple variation which hardly constitutes a separate clause: 'then Heorogar was dead, my elder brother, lifeless, the son of Healfdene'. The form eal in 486a may be, as Kuhn says, a case of a proclitic standing alone in the upbeat of a clause lacking its predicate, 31 but the easier solution, it seems to me, is to take eal as an adverb: Donne waes f?eos medoheal on morgentid, drihtsele dreorfah, J?onne daeg lixte, eal bencj?elu blode bestymed, heall heorudreore . . . 32

(484-7a)

Line 135 lb (oder earmsceapen) is a violation only if we take oder to be an adjective proclitic on earmsceapen. But oder is used as a pronoun in the preceding correlative construction and can be construed in 1351b as a pronoun in apposition with earmsceapen: Daera oder waes, )?aes \>e hie gewislicost gewitan meahton, idese onlicnes; oder earmsceapen on weres waestmum wraeclastas traed, naefne he waes mara Jxmne aenig man o5er . . . 33 (1349b-53) As for the single half-line in (h) (To lang ys to reccenne, 2093a), this appears to be a certain violation. I will suggest in ch. 9 the possibility that the first to is a scribal anticipation of the second. We began by looking at all the verses in Beowulf which have or might be thought to have proclitic onsets. There were 560 of them. Only thirty-six appeared to violate Kuhn's second law. But twenty-two of the thirty-six on closer inspection proved not to have proclitic onsets after all. Another was basically an editorial conjecture and could be disregarded. This left 537 half-lines with proclitic onsets and thirteen apparent violations. In nine of 31 32

33

Kuhn, p. 44. 'Then was this mead-hall in the morning, the splendid hall, blood-stained, when day dawned, the bench planks completely wet with blood, the hall with sword-gore 'One of t h e m was, as far as they could most certainly ascertain, the likeness of a woman; the other one, wretched, trod the exile-paths in the shape of a m a n . . . '

83

The Metrical Grammar of Beowulf the thirteen, clause boundaries could be reassigned, eliminating the violations. Thus only four half-lines out of 537 clearly violated Kuhn's law (758a, 1537a, 1684a and 2093a). All four have been challenged by one scholar or another on grounds that have nothing to do with that law. Their authenticity as they stand must be considered doubtful. Nearly all non-initial (II) half-lines, that is, half-lines with proclitic onsets, correlate with Sievers's types B and C. Only fourteen apparently belong to some other metrical type. 34 These are (with the respective classifications of Pope and Bliss): at fotum saet frean Scyldinga;

gehwylc hiora his fertile treowde (1166a: Pope, Bliss, hypermetric)

|?a se goda, maeg Higelaces (758a: Pope, A with anacrusis; Bliss, A 3 (by emendation)) Gefeng J?a be eaxle — nalas for faehSe mearn (1537a: Pope, A with anacrusis; Bliss, A 3 (by emendation)) To lang ys to reccenne, hu i(c 5)am leodsceaSan (2093a: Pope, D* 1 with anacrusis; Bliss, unclassified remainder) (?aet heo (?one fyrdhom

SurMon ne mihte (1504b: Pope, Bliss, A with anacrusis)

under swegles begong

gesacan ne tealde (1773b: Pope, Bliss, A with anacrusis)

J?aet he |?one breostwylm

/orberan ne mehte (1877b: Pope, Bliss, A with anacrusis)

in maegj?a gehwaere man gej?eon (25a: Pope, A with anacrusis; Bliss, B (by emendation)) in Caines cynne — |?one cwealm gewraec (107a: Pope, A with anacrusis; Bliss, C (original manuscript reading)) ealne utanweardne; ne Saer aenig mon (2297a: Pope, unclassified remainder; Bliss, hypermetric. Pope and Bliss stress ealne) ge at ham ge on herge,

34

ge gehwa^er J?ara (1248a: Pope, Bliss, A with anacrusis)

Type A3 is never non-initial. See the discussion of 1492a and 2669a above.

84

Identification of clause-non-initial half-lines wid ord ond wiS ecge /orgyteS ond forgymeS,

ingang forstod (1549a: Pope, Bliss, A with anacrusis) |?aes )?e him aer God sealde (1751a: Pope, Bliss, A with anacrusis)

forsiteS ond forsworceS; semninga biS (1767a: Pope, Bliss, A with anacrusis) All of the commonly recognized hypermetric lines in Beowulf (1163-8; 1705-7; 2995-6) conform to Kuhn's laws as far as their onsets are concerned. Line 1166a is in fact non-initial as its proclitic onset leads us to expect. Pope describes the onset as monosyllabic anacrusis. 35 Bliss associates it with type C. 36 The argument which I am developing lends support to Bliss's classification. Setting 1166a aside with the observation that hypermetric lines seem to conform to principles that parallel those of normal lines, we can consider the remaining thirteen verses. I want to suggest that, insofar as they exhibit acceptable readings, they should be associated with types B and C rather than A or D. Several of these verses, however, must certainly be regarded as corrupt. Lines 758a, 1537a and 2093a are three of the four verses which have already been pointed out as the only certain violators of Kuhn's second law. Here is further evidence that something is radically wrong. Lines 758a and 1537a are easily emended. 37 They become type A 3 . The problems with 2093a run deeper and cannot be so simply removed. 38 It is perhaps best to leave it, as Bliss does, as an unclassified remainder. I regard 1504b, 1773b and 1877b as varieties of type C rather than as type A with anacrusis. 39 As for 25a and 107a, there is good reason to believe that these are altered from their original states which would have been type B and type C respectively40: *in maeg)?a gehwam 35 36 37 40

(25a)

Rhythm, p . 1 4 5 . Metre, Index t o t h e Scansion of the H y p e r m e t r i c Verses, p . 1 6 2 . 38 39 See above, p . 7 8 . See above, p p . 8 3 . See below, ch. 7 . Gehwcere (25a) is a late analogical formation for gehwcem or gehwam: see Sievers-Brunner, § 3 4 1 . Sievers, ' Z u r R h y t h m i k ' , p . 4 8 5 , regards t h e restoration of gehwam or gehwam as 'sicher'. See also Bliss, Metre, § 4 7 .

85

The Metrical Grammar of Beowulf in Games cynne (107a; the original manuscript reading) Line 2297a has generated much discussion and several proposals for emendations. 41 It has been assumed on the basis of the alliteration linking ealne with utanweardne that ealne must be stressed. I believe, on the contrary, that ealne is weakly stressed here as it is elsewhere, and that the alliteration is incidental. Reducing the level of stress on ealne does not in itself solve the metrical problem. It lends support, however, to at least three of the proposed emendations: eal utanweard (Sievers), ealne utanweard (Trautmann) and ealne utweardne (Sievers).42 Without stress on eal(ne), the first two are type B; the third, type C. Sievers's first emendation seems to me the most plausible. 43 The half-line modifies hlcew in 2296b. If the poet had taken hlcew here, exceptionally, as a neuter, perhaps for the sake of the metre in the next verse, then, since it is commonly a masculine noun, a copyist would have been likely at some stage to normalize the adjectives dependent on it. The remaining four verses (1248a, 1549a, 1751a and 1767a) are not obviously corrupt, yet they are overburdened. Lines 1248a and 1549a almost certainly are drawn from the common phraseology of the language. 44 Possibly the Beowulf-pott borrowed some stereotyped phrases from his everyday speech and employed them without noticing, or bothering about, the irregularities in their metrical contours. But given his normal polish, this seems unlikely. All are built on the syntactic pattern: X + copulative conjunction + Y. If we make up nonsense verses by substituting X for Y , we generate perfectly regular type B: *ge cet ham ge at ham etc. This suggests that the metrical affinity of these verses is with type B rather than with type A. 45 The four verses share another characteristic. 41 42

See Klaeber, textual apparatus. Sievers, ' Z u r R h y t h m i k , ' p . 3 0 6 ; T r a u t m a n n , Das Beowulflied; Sievers, Metrik,

43

Altgermanische

p . 134 (so, H e y n e , Beowulf, rev. Schiicking).

However, Klaeber, p . 2 1 1 , remarks: 'hlcew is normally masc. (one instance of the n e u t . : Sievers, Beitr. ix 237) and appears as such in all t h e passages of our p o e m where t h e gender can be seen ( 2 8 0 3 , 2 8 0 4 , 3 1 5 7 ; 2 4 1 2 ? ) . H e n c e ealne should not be changed to eal. N o r should utanweardne be c o n d e m n e d on metrical g r o u n d s . '

44

1549a seems to derive from p r i m i t i v e G e r m a n i c phraseology. Cf. Icelandic med oddi ok eggju, cited in Bos w o r t h - T o l l e r , s.v. ord.

45

Curiously, three of these verses could have been p r o n o u n c e d in the r h y t h m of type B in the West-Saxon dialect (reading here, forgymS and forsworcd). Since a non-West-Saxon origin is assumed for the p o e m this may be an irrelevant coincidence.

86

Identification of clause-non-initial half-lines All display strict syntactic parallelism (ge <%t . . . ge on, wi& . . . wi& and for. . . for-), which results in supplemental alliteration. This additional formalism in the verses coupled with their metrical irregularity calls attention, perhaps deliberately, to the moments in the poem where they occur. The first (1248a) is when Beowulf's men lie down after the great celebration over Grendel's defeat, ignorant of the terror that is about to burst upon them in the form of Grendel's mother: Wees (?eaw hyra, anwiggearwe,

J?aet hie oft waeron ge cet ham ge on herge,

ge gehwae)?er (?ara

efne swylce maela, swylce hira mandryhtne )?earf gesaelde; waes seo f?eod tilu. 46

(1246b—50)

The second (1549a) is at the moment when Beowulf's armour saves h i m from her knife-thrust: Him on eaxle laeg breostnet broden; J?aet gebearh feore wid ord ond witS ecge, ingang forstod. 47

(1547b-49)

The third and fourth (1751a and 1767a) are found in Hrothgar's thanksgiving sermon to Beowulf, warning h i m against pride: under helm drepen 1745 Ponne biS on hre(?re biteran straele — him bebeorgan ne con — , worn wundorbebodum wergan gastes; J?inceS him to lytel, J?aet he lange heold, gytsaS gromhydig, nallas on gylp seleS ond he \>a. forSgesceaft 1750 fae#e beagas, forgyte& ond forgymeb\ J?aes J?e him aer God sealde, wuldres Waldend, weorSmynda dael . . . Nu is |?ines msegnes blaed ane hwile; eft sona biS, J?aet )?ec adl oSSe ecg eafo(?es getwaefeS, oSSe fyres feng, oSSe flodes wylm, 1765 oSSe gripe meces, o33e gares fliht, oSSe atol yldo; o3Se eagena bearhtm 46

47

'It was their custom that they were normally ready for war, either at home or on campaign, or on whatever of such occasions as their lord had need; that was a good people.' Translated above, p . 80, n. 18. 87

The Metrical Grammar of Beowulf forsited ond forsivorced; \>3£t Sec, dryhtguma,

semninga biS, 48 deaS oferswySeS.

It would be possible to interpret these four verses (1248a, 1549a, 1751a and 1767a) as being hypermetric, though their onsets are unusually short. I choose to mark them as expanded type B (B # ) to indicate what I take to be their metrical affinity. 48

'Then he will be struck in the heart under the helmet with a bitter arrow (he does not know how to protect himself), with dark mysterious promptings of the cursed spirit. It will seem to him too little, what he had held for a long time; angry-minded he will grow avaricious; not at all will he proudly give the plated rings, and he will forget and neglect the future state, since God had given him, Ruler of glory, a share of honours . . . Now is the glorious vigour of your strength for a time; soon enough it will be that sickness or the sword will deprive you of strength, or the fire's grip, or the billows of the flood, or the bite of the sword, or the flight of the spear, or dire old age; or the brightness of your eyes will diminish and darken; presently it will be that death will overpower you, warrior.' Clemoes, 'Style as Criterion', pp. 180—1, argues that 'the coupling of finite verbs was probably not indigenous to Old English verse structure'. He sees in these verses from Hrothgar's sermon (175 la and 1767a; also 174 la) an influence from Old English prose homilies. I discuss these verses with paired verbs in the following two chapters.

88

7 Half-lines with internal clause divisions: the transformational rule (revised)

In the metrical grammar of Beowulf the beginning of a clause typically coincides with the beginning of a half-line. This is a natural consequence of the fact that the basic unit of composition is the half-line formula or formulaic system. As far as independent clauses and whole sentences are concerned, this generalization may be regarded as an iron-clad rule. 1 No independent clause in Beowulf begins in the middle of a half-line. The additive style of the poem makes it difficult to arrive at an unambiguous definition of a sentence. But if for present purposes we define it as a single independent clause together with its dependent clauses, if any, then the sentence consists of a string of one or more half-lines. 2 Sentence boundaries invariably coincide with half-line boundaries. The first three lines of the poem offer a simple illustration (where / = a boundary between clauses within the sentence): Hwaet, we Gar-Dena in geardagum, J?eodcyninga J?rym gefrunon, / hu 3a aej>elingas ellen fremedon!3

(1-3)

Here, the sentence consists of six consecutive half-lines. Within the sentence the rule is nearly as strict. Usually the juncture between an independent clause and a dependent clause or between two dependent clauses, where the second is subordinate to the first, is marked 1

2

3

See Mitchell, Old English Syntax, § 3960, for some cautions, which the discussion here partially addresses. On sentence structure and verse paragraphs and in particular the difference between the behaviour of auxiliaries in principal and dependent clauses, see Donoghue, Style, esp. pp. 48-99. 'Listen! We have heard of the glory of the Spear-Danes, of the people-kings, in days of yore, how the nobles performed deeds of courage.'

89

The Metrical Grammar of Beowulf by a half-line boundary, as it is in the sentence just quoted. However in forty-one instances this juncture does not occur at a half-line boundary but within the half-line. These verses with internal clause divisions are of considerable interest and merit a closer look. 4 Representative examples are: wesan, / J?aes ic wene.

Pu wast, / gif hit is

mynte / J?aet he gedselde, beaga bryttan,

ser (?on daeg cwome

breac / J?onne moste

(272) (731) (1487)

In addition to these forty-one there are eight a-verses with the syntactical pattern: finite verb + conjunction + finite verb, and three a-verses with the pattern: finite verb-adverb phrase / finite verb-pronoun object + conjunction + finite verb. All display double alliteration, e.g., seomade ond syrede (161a; also 600a, 1337a, 1604a, 174la, 1751a, 1767a and 2319a) and Hafa nu ond geheald (658a; also 2057a and 2430a). I choose not to regard these verbs as belonging to separate clauses but rather as simple variations within a half-line expressing a single, sometimes complex, verbal idea. 5 Another kind of half-line which could be considered as having an internal clause division is the one which contains what Bruce Mitchell calls the se'pe type of definite adjective clause. 6 This is the type in which the demonstrative agrees in case with the antecedent in the preceding clause rather than taking the case of the relative in the subordinate clause. Mitchell gives as an example these lines from Beowulf:

4

5

The 41 are: 92a, *272a, 338a, *442a, *731a, 1184a, 1707a, 1810a, *1859a, 1894a, *2158a, 2163a, 2172a, 2939a, *3096a, *3180a; *57b, *90b, *183b, *186b, 199b, *272b, *455b, *469b, *603b, *1003b, *1177b, *1182b, *1224b, #1231b, *1319b, # 1331b, *1379b, #1387b, #1394b, *l487b, 2252b, *2519b, *2766b, *3O55b and * 3062b. The second clause in the verses marked with an asterisk is scanned by the transformational rule (see below). Clemoes, 'Style as Criterion', p. 181, n. 21, cautions that the verbs in 600a {swefetS ond snedep (manuscript sendep)) 'do not jointly express a single idea' and that in 1604a {wiston ond ne wendon) the 'verbs are coupled to point a distinction'. He includes in his list one verb pair which spreads over successive lines: Wit pat gecwcedon cnihtwesende I ondgebeotedon

6

(535-6a). I assign these to separate clauses. See below, ch. 8, for further discussion of these verb pairs. Old English Syntax, §§ 2159-76; the quotation from Beowulf which 1 cite below is in §2175.

90

Half-lines with internal clause divisions sySSan hie gefricgeaS frean userne ealdorleasne, pone 5e aer geheold wiS hettendum hord ond rice . . . 7

(3002-4)

The demonstrative pone is accusative, agreeing in case with frean userne, rather than with the relative particle 6e9 which is the subject of the subordinate clause. Logically, one could put the clause division between pone and &e. However, as Daniel Donoghue points out, although 'pone is formally part of the principal clause, . . . it is placed with tk in the dip at the beginning of the relative clause'. 8 I have not included such verses in this discussion. The alliterative behaviour of the a-verses differs from that of the b-verses. Primary alliteration follows the clause juncture in the a-verses, but usually precedes the juncture in the b-verses. Specifically, the most prominent syllable of the segment following the juncture in every one of the sixteen a-verses alliterates. Six of the sixteen display double alliteration so that in these the chief syllable of the prior segment alliterates too. But in no case is the chief syllable of the prior segment a nominal. 9 The chief syllable of the first segment alliterates in twenty-three of the twenty-five b-verses. This difference in alliterative weight corresponds to a difference in semantic weight between the two clauses. In the a-verse the second clause, though subordinate, typically carries the significant meaning. In the b-verse, the reverse tends to be true, and the second clause is often parenthetical. Thus there is every indication that most of these half-lines are not interchangeable - that most of the ones which appear in the a-verse with single alliteration could not be moved to the b-verse; and that most of those in the b-verse could not be moved to the a-verse. In Beowulf an undisplaced sentence particle always follows the juncture between a clause and a subsequent clause which is subordinate to it. The particle appears in the initial dip of the subordinate clause. Ordinarily the subordinate clause begins with or consists of a clause-initial (I) half-line. In these forty-one half-lines, where the juncture is internal, the subordinate 7

8 9

'. . . after they discover our lord to be dead, him who had defended the treasure and the kingdom against enemies 'Word Order', p. 181. Twice the alliterating form in the prior segment is the infinitive wesan (272a and 1859a). These two-verses resemble b-verses with alliteration on the first segment like 1224b. I discuss them below.

91

The Metrical Grammar of Beowulf clause still has an undisplaced sentence particle in its initial dip. Of course, this is nearly inevitable, since there is almost always a subordinating conjunction following the juncture. However, even in those few cases where the subordinating conjunction is omitted, there is still an undisplaced sentence particle, for example: cwaeS, / he (?one guSwine gesawon seledream.

godne tealde

Nah, / hwa sweord wege

(1810) (2252)

Ordinarily, an undisplaced sentence particle marks its half-line as clause-initial (I). It might appear therefore that these forty-one half-lines would be all of this type. However the situation is more complicated than that. In fact, the presence of a sentence particle following a clause juncture has no bearing on whether the half-line can be displaced. The second segment remains 'clause-initial' by definition no matter how the half-line is positioned. It is the segment preceding the juncture which determines whether the entire half-line is clause-initial (I) or unrestricted (III). 10 If the segment before the juncture contains one or more sentence particles that do not alliterate, the half-line is clause-initial, nondisplaceable (Ia/I). A single, non-alliterating sentence particle is found in the a-verse in cwczd I J?aet se i<£lmihtiga

eorSan worh(te)

(92; similarly 1810a, 1894a, 2158a and 3180a) mynte I J?set he gedselde,

aer J?on daeg cwome

(731)

bced I J?aet ge geworhton aefter wines daedum (3096) An alliterating sentence particle in the b-verse must always be considered a stressed element. But there are seven b-verses with a single, nonalliterating sentence particle (italicized) in the first segment: wuldres Waldend.

Wa bid 1 bsem 5e sceal

wihte gewendan

Wei bid 1 J?aem )?e mot

(186)

godne gegyrwan;

cwad, 1 he guScyning

(199)

Pu wast, / gif hit is

(272)

wesan, (?aes ic wene. 10

(183)

There appears to be no reason in theory why there should be no clause-non-initial (II) half-lines with internal clause divisions. Perhaps there was simply little motivation for their development, since their use would be even more limited than that of the ones we have.

92

Half-lines with internal clause divisions

guj?e gebeodan.

Gae(? eft I se \>e mot

waelgaest wsefre;

ic ne wat / hwae^er

gesawon seledream.

Nah, I hwa sweord wege

(603) (1331) (2252)

There are six half-lines with two non-alliterating sentence particles in the first segment (two in the a-verse; four in the b-verse): wene ic I J?aet he mid gode Hyrde ic I J?aet J?am fraetwum

gyldan wille

(1184)

feower mearas

(2163)

Welandes geweorc.

Gad a wyrd / swa hio seel

beam Healfdenes;

se wees betera / Sonne ic

arum healdan,

(455) (469)

gyfpu ser / bonne he

(1182)

gific wiste / hu

(2519)

waepen to wyrme,

Similarly, there are three a-verses where one of the two sentence particles displays extra-metrical alliteration. These remain clause-initial, nondisplaceable (la): a

A

Wen' ic I J?aet ge for wlenco, a

Wen' ic I pact he wille, a

nalles for wraecsiSum

(338)

A

gif he wealdan mot

(442)

A

Hyrde ic I (?aet he Sone healsbeah

Hygde gesealde

(2172)

Every half-line in these four groups begins a clause. There are two a-verses (272a and 1859a) with a single, alliterating sentence particle in the first segment — in both cases the infinitive wesan. They are clause-initial, displaceable (Ib). One begins a clause (this is an elliptical consecutive clause, with the finite verb understood from sceal in 1855b): Hafast J?u gefered, J?2et (?am folcum sceal, Geata leodum ond Gar-Denum sib gemaene, ond sacu restan, inwitni(?as, J?e hie aer drugon, wesan, I (?enden ic wealde widan rices, magmas gemaene . . . n 11

(1855—60a)

'You have brought it about that peace shall be in common to the folk, to the people of the Geats and to the Spear-Danes, and that strife shall be laid to rest, hostile acts, which they had carried out before, that there shall be, as long as 1 rule the broad kingdom, shared treasures . . . '

93

The Metrical Grammar of Beowulf The other does not begin a clause: HabbaS we to (?aem maeran micel aerende Deniga frean; ne sceal |?aer dyrne sum wesan, I J?aes ic wene.12

(270—2a)

In both cases the subordinate clause which follows is parenthetical rather than consecutive. This is typical of the b-verses where the alliterating stress falls on the chief syllable of the first segment. The two verses are closely akin to b-verse half-lines like Wes penden pu lifige (1224b). What we have here seems to be a variation of the phenomenon that Peter Lucas calls 'convertibility', that is 'the potential for having [the] metrical pattern {of a half-line] altered so as to move the alliterating stress to a different word'. 13 Here a b-verse pattern has been moved into the a-verse with the shift of the alliterating stress to the second segment. The sentence particle wesan retains extra-metrical alliteration, and since in both cases it is the only sentence particle in its clause the half-lines are displaceable. Line 1707a is hypermetric, with double alliteration marking the chief syllables of the two segments: A

A

freode, / swa wit furSum spraecon.

Du scealt to frofre weorf>an (1707)

It is unrestricted (III). All the remaining b-verses have a single stressed element before the clause juncture — either a stressed verb or a nominal. These are, by definition, unrestricted (III). The first segment begins a clause in twelve of them, 14 for example: heah Healfdene;

Heold I )>enden lifde

(57)

But there are two (1231b and 3055b) in which it does not: (1) Her is aeghwylc eorl o(?rum getrywe, modes milde, mandrihtne hol{d], f>egnas syndon gejrwaere, J?eod ealgearo, druncne dryhtguman do&1 swa ic bidde.15 12

13 14

15

(1228—31)

'We have a great errand to the famous lord of the Danes; no important matter shall be secret there, as I think.' 'On Some Breaches', p. 388. 57b, 90b, 1003b, 1177b, 1224b, 1319b, 1379b, 1387b, 1394b, 1487b, 2766b and 3062b. 'Here each warrior is true to the other, mild of mood, loyal to his lord, the thegns are in harmony, the people willing, the retainers having drunk do as 1 ask.'

94

Half-lines with internal clause divisions

(2) J?onne waes (?aet yrfe eacencraeftig, iumonna gold galdre bewunden, J?aet Sam hringsele hrinan ne moste gumena aenig, nefne God sylfa, sigora SoScyning sealde I |?am 3e he wolde - he is manna gehyld hord openian . . . l 6

(3051-6)

Some of the forty-one half-lines with which we are concerned can be scanned in the normal way by the application of the fundamental alliterative principle, which assigns primary alliteration to the first lift. So, those half-lines with nominals in the second segment all display regular metrical contours: type A 3 (with double alliteration) a

A

Wen' ic / J?aet ge for wlenco, a

nalles for wraecsiSum

(338)

Hygde gesealde

(2172)

A

Hyrde ic / )>aet he Sone healsbeah 3

type A (with single alliteration) A

wene ic / J?aet he mid gode A

Hyrde ic / J?aet J?am fraetwum

type C

(1184)

feower mearas

(2163)

eorSan worh(te) 17

(92)

1 A

- cwaeS / l?aet se ^lmihtiga A

cwaeS, / he on mergenne

type C

gyldan wille

meces ecgum

(2939)

godne tealde

(1810)

Wedera leodum

(1894)

2 A

cwaeS, / he (?one guSwine A

cwaeS / (?aet wilcuman godne gegyrwan; gesawon seledream.

A

cwaec"i, / he guScyning A

Nah, / hwa sweord wege

(199) (2252)

In terms of metre, the first segment merely adds additional syllables to the 16

17

'Then was that mighty heritage, the gold of the men of old, wound round with a spell so that no man might enter that ring-hall, unless God himself, true King of victories, gave permission to the one whom he wanted — he is the protector of men — to open the hoard . . . Klaeber suppresses the penultimate syllable of JElmthtiga.

95

The Metrical Grammar of Beowulf

weak onset of the metrical type. The six type C verses are the only ones which are not limited to the a-verse. The remaining half-lines lack stressed elements in the second segment. To scan them it is not necessary to formulate any new rules. The transformational rule serves the purpose nicely. As written it was limited to entire a-verse half-lines. But it can be readily generalized to apply to these segments as well. The modification required is slight: Sentence particles in any clause-initial segment which lacks stressed elements acquire metrical stress from right to left in accordance with the stress and phrase rules of the metrical grammar until the first valid metrical contour emerges. The fundamental

alliterative principle will then operate to assign alliteration to the first lift. Consider three half-lines: A

Hiorogar cyning

cwaeS / )?aet hyt haefde A

heah Healfdene; a

heold / t?enden lifde

A

Pu wast, gif hit is

wesan / J?aes ic wene.

(2158) (57) (272)

The rule applies to the second segment of all three. In 2158a, it puts metrical stress on the first syllable of hcefde. Since this establishes a valid metrical contour for an a-verse (x x x / x: type A 3 ), the fundamental alliterative principle requires alliteration on the first syllable of hcefde. Line 57b superficially resembles 2158a. However, since this is a b-verse and heold alliterates, heold must be regarded as a stressed element. The transformational rule assigns metrical stress to the first syllable of lifde in the segment penden lifde. The contour of the segment (x x / x) is not itself acceptable for the b-verse. But, with alliterating stress on heold it completes a valid metrical contour: / x x / x (type A). In 272a the rule assigns metrical stress to the first syllable of wene. Since this completes a metrical contour (type A 3 ), the fundamental alliterative principle assigns primary alliteration to wene. Alliteration on wesan is extra-metrical. The following scansions result when the transformational rule is applied: type A 3 (with double alliteration) a

A

wesan, J?aes ic wene. a

A

Wen' ic J?aet he wille, a

Pu wast, gif hit is

(272)

gif he wealdan mot

(442)

A

wesan, benden ic wealde

96

widan rices

(1859)

Half-lines with internal clause divisions type A 3 (with single alliteration) A

aer J?on daeg cwome

mynte J?set he gedadde, A

cwaeS J?aet hyt haefde A

baed (?aet ge geworhton

(731)

Hiorogar cyning

(2158)

sefter wines daedum

(3096)

wyruldcyning[a]

(3180)

A

cwaedon J?aet he waere

type A (with single alliteration, b-verse only) A

heah Healfdene;

N

heold J>enden lifde

(57; similarly 90b, 1003b, 1177b, 1224b, 1231b, 1319b, 1379b, 1387b, 1394b, 1487b, 2766b, 3055b and 3062b) type B A

wesan, |?ses ic wene.

A

. N

/elandes geweorc.

GaeS a wyrd swa hio seel

beam Healfdenes;

se waes betera Sonne ic

(Ax)

N

A

arum healdan,

N

gyf t?u aer )?onne he

waepen to wyrme,

type C

N

Pu wast, gif hit is

A

N

gif ic wiste hu

(272)

(455) (469) (1182)

(2519)

2 A

waelgaest waefre;

(N)

ic ne wat hwae^er

(1331)

These last two verses, gif ic wiste hu (2519b) and ic ne wat hwceder (133 lb), are interesting, because they are the only ones with a single sentence particle following the clause juncture. One might argue that hu and hwceder are necessarily stressed, without the aid of the transformational rule, because they follow sentence particles which have been inserted into the first stressed position in the b-verse {wiste, wat) - that they are stressed, that is, as though they had been displaced from the initial dip of the preceding clause {gif ic, ic ne) on the analogy of a verse like ic minne can (1180b). But they have not been displaced, of course. They constitute the initial dip of their own clauses. Both words are nearly always placed at the head of a 97

The Metrical Grammar of Beowulf clause, and therefore nearly always are found as the first word in the half-line. 18 What is unusual here is the fact that the clause boundary is internal. It seems better to treat them exactly like all the other clauseinitial segments which follow an internal clause division and to assign them stress by the transformational rule. Three half-lines present certain difficulties. They are: wuldres Waldend.

Wa biS / (?asm 3e sceal A

wihte gewendan

W e i bi5 / )?aem (?e m o t A

gu(?e gebeodan.

Gae|? eft / se ]>e mot

(183) (186) (603)

Applying the transformational rule to the second segment achieves no result unless secondary stress falls on the second word in the first segment. If it does, then each of these half-lines is a normal type E. Bliss argues for secondary stress on eft in 603b and Pope (1966) agrees. 19 As for the other two half-lines, Bliss puts them among his 'remainders' and Pope stresses / x \ x /. 2 0 My personal feeling is that bid is emphatic in both 183b and 186b. 'There will be woe for anyone who . . . ', and 'It will be well for anyone who . . . ' The stress pattern / \ after all almost always characterizes adverb + verb combinations in Beowulf/even when the verb is a copula, as in A

N

J?aet hie oft waeron 18

19

(1247)

Hu appears 267 X in the corpus of Anglo-Saxon poetry, either as an interrogative adverb or as a proclitic adverb; it is the first word in the half-line in all but twenty-one cases. Of these twenty-one, it functions as an interrogative adverb after a conjunction (ac, ond, oppe) six times; it functions as a proclitic adverb eleven times; and in the remaining four cases, including Beowulf 25 19b, it is the first word (an interrogative adverb) following an internal clause boundary. Only here in 2519b is it the final word in a half-line. The interrogative adverb hwceder ( = hwider I hwyder) appears nine times in Anglo-Saxon poetry. It is the first word in the half-line everywhere except here in Beowulf 133 lb and in sawul on siSfaet,

20

anwiggearwe

Min sceal of lice nat ic sylfa hwider . . .

(Juliana 699b—700)

In the latter case, it is also final and could be analysed as the sole word in an elliptical clause ('My soul must go on a journey from my body, I myself do not know / whither (it will go)'). Bliss, Metre, § 82; Pope, Rhythm, p. 371, n. Bliss, Metre, § 8 6 ; Pope, Rhythm,

p. 371.

98

Half-lines with internal clause divisions I would extend Bliss's scansion of 603b to these two half-lines as well, making all three type E: / \ x x / . No a-verse with an internal clause juncture in which the second segment must be scanned by the transformational rule exhibits a stressed element before the juncture. Every b-verse does. Though this is a small body of evidence to go on, it would appear that the second segment in the a-verse in such cases will always acquire an initial lift by the transformational rule. And since b-verses never require the transformational rule for the assignment of the initial lift, the conclusion must be drawn that all of these half-lines are limited to the a-verse or the b-verse by the presence or absence, respectively, of primary alliteration after the clause juncture.

99

8 The alliterative requirement of unstressed prefixes and the copulative conjunctions

Proclitics serve two special and possibly unexpected functions in the metrical grammar of Beowulf. They help determine the position of the half-line in the verse clause,1 and they impose an alliterative requirement on the forms to which they are attached. The purpose of this and the following two chapters is to describe in analytic terms the way proclitics affect the alliterative patterns of the poem. Proclitics include all those words and prefixes which in ordinary circumstances are not independently stressed within the structure of the clause. Instead they attach themselves to other single words or phrases which do carry independent stress. There are five types of proclitics in the metrical grammar — (1) unstressed prefixes; (2) copulative conjunctions; (3) prepositions; (4) proclitic adverbs and instrumental; and (5) proclitic adjectives and pronouns. The alliterative requirement of the prefixes and conjunctions will be considered first. THE UNSTRESSED PREFIXES

The unstressed prefixes which occur in Beowulfare a-, cet-, be (bi)-, for-, ful(l)-,

ge-, geond (giond)-, of-, ofer-, on (an)-, od-, to-, purh-,

wid- and

2

ymb(e)-. Ten of them are never stressed in the poem. Of these ten, a-, geond-, of-, od-, purh- and wid- are joined only to verbs, for- and to- appear also before adverbs, conjunctions and prepositions and be- andge- are found even with nouns and adjectives. 1 2

See above, ch. 6. The prefix un- is variably stressed. The alliterative consequences of its stressed and unstressed forms are quite different from those of the prefixes under discussion here. See below, ch. 14, and Kendall, 'Prefix un- .

100

Unstressed prefixes and the copulative conjunctions

In general, the remaining five prefixes (cet-, ful-, ofer-, on- and ymb-) are stressed before nouns and adjectives, but remain unstressed before other parts of speech.3 However (as the alliterative patterns of the lines prove) the prefix is stressed in the adverb at-rihte (1657b) and in the verb ful-lczstu (2668b). The unstressed prefixes may be divided into two sets — those that end in a consonant and those that do not. I shall refer to the former as the consonantal and to the latter as the vocalic prefixes. The vocalic prefixes are only four in number (a-, be-, ge- and to-), but high in frequency. In terms of alliterative requirement, they have less force than the consonantal prefixes. The metrical grammar of Beowulf moves stressed elements with an unstressed prefix as far to the right as possible in the half-line. They will be followed only by displaced sentence particles (verbs, adverbs, pronouns etc.), displaced proclitics, or by other postpositive forms (e.g., had, twam etc.). Compare daedum gedefe,

dreamhealdende

Da him HroJ?gar gewat

mid his haele(?a gedryht

(1227) (662)

with to gescipe scyndan.

Scyld wel gebearg

on gesiSes had crseft ond cenSu,

(2570)

be saem tweonum

(1297)

swa him gecynde wees

(2696)

In the second group scyndan and wees are displaced sentence particles and had is a postpositive form. Only four half-lines are or appear to be irregular in this respect: wergan gastes;

waes J?aetg£win to strang

(133)

waes )?aet gewin to swy5

(191)

wean onwendan; geworden in wicun.

Ne waes J?aet gewrixle til

ne garnet mannes,

nefn(e) min anes

(1304) (2533)

The first three have a common syntactical pattern which puts a predicate adjective in final position. A predicate adjective in this position can be added to the list of postpositive forms. The last may be a scribal blunder. It 3

I take up the stressed forms of these prefixes in ch. 13. 101

The Metrical Grammar of Beowulf is part of Beowulf's final boast to his retainers before going into battle against the dragon: Nis \>xt eower siS, ne gemet mannes, nefn(e) min anes, pcet he wiS aglaecean eofoSo daele, eorlscype efne.4

(2532bn-35a)

One expects the order *ne mannes gemet, which is supported by the formulaic system . . . monneslmonna gemet that is found elsewhere in Old English poetry. 5 On the other hand mannes may have been attracted out of its regular position in order to stand nearer to the adjective clause (2534—35a) which modifies it. 6 The rightward movement accounts for the fact that adjectives with an unstressed prefix are postpositive, for example: ond on sped wrecan licegelenge.

spel gerade

(873)

Ic Sas leode heold

(2732)

The rightward movement frequently puts the base word to which the prefix is attached into a non-alliterating lift-position, especially in the b-verse. But where this has not happened, alliteration of the base word is strikingly common. When the prefix is attached to a base word which occupies the first lift of the half-line, the base word must of course alliterate. But in the a-verse, alliteration of the prefixed base word is usual even when it occupies the second lift. Thus we find 174 verses like A

A

wean owwendan;

waes J?aet gewin to swyS

(191)

But we find only seventy-seven verses like A

N

Sonne sweorda gelac

sunu Healfdenes

(1040)

Only when the prefixed base word is preceded by two lifts do we expect it not to alliterate: 4

5

6

'This is not your venture, nor is it fit for a man, except myself alone, that he measure his strength against the terrible monster, perform this heroic deed.' The system is found in three separate poems: ofer monna gemet, maerSa georne (Genesis 1677) Naes 5a monna gemet, ne maegen engla (Christ and Satan 489) Nis J>aet monnes gemet moldhrerendra (Order of the World 27) I owe this suggestion to Professor E . G . Stanley.

102

Unstressed prefixes and the copulative conjunctions A

A

N

atol y3a geswing

eal gemenged

(848)

In the b-verse, on the other hand, alliteration of a base word occupying the second lift is blocked by the basic alliterative rules: A

Sona (?aet onfunde

N

se 3eflodabegong

(1497)

It is possible to give more precision to this description of the alliterative behaviour of prefixed forms. The key is the number and location of the weak metrical position (w) in which the prefix occurs. There is always one, and there are usually two such positions in the half-line. 7 The first weak metrical position (wl) may precede the first strong metrical position or lift, as in wl

feondgrapum faest.

A

Ic gefremman sceal

(636)

Pa J?aet sweord ongan

(1605)

gomela Scylfing

(2487)

anne mannan

(211 A)

It may precede the second lift, as in A

wl

A

selfe gesawon. A

wl

guShelm

A

taglad,

It may follow the second lift, as in A

A

wl

eald enta grweorc

The second weak metrical position (w2) may precede the second lift, as in wl

A

w2

(Nx)

widteondagehwone

flotan

eowerne

(294)

A prefixed base word which occupies the first lift in either the a- or the b-verse can and must alliterate according to the basic alliterative rules. When it falls in any other position in the b-verse, alliteration is blocked by the blanket prohibition against double alliteration. The alliterative patterns of the second lift in the a-verse can be described by a specific rule of the metrical grammar, which may be provisionally expressed this way: Alliteration of a base word which occupies the second lift in the a-verse is obligatory when its prefix falls in the first weak metrical position {wl) and optional when it falls in the second (w2). 7

I describe here only those weak metrical positions which precede a lift. There are weak metrical positions which are final, namely: after the first lift, type A 3 ; after the second lift, types A, C, D 1 " 3 . Proclitics are of course excluded from these positions by definition.

103

The Metrical Grammar of Beowulf What does this mean in practice, and how strictly does it apply? The basic alliterative rule and the fundamental alliterative principle require alliteration in the first lift of both the a-verse and the b-verse. They permit alliteration of the second lift of the a-verse, but forbid it in the b-verse. No other alliteration of a metrically stressed word is allowed. In the five half-lines cited just above, for example, the alliterative rules require alliteration of the base word in 636b, permit it in 1605a, 2487a and 294a and forbidit in 211 Adi. Thus alliteration is blocked when the w l follows the second lift. When the w l follows the first lift but not the second (types A and E), the rule of prefixes as I have stated it requires alliteration in the a-verse. These types should exhibit double alliteration, as do: A wl

A

snude forsended. (Ax) N

Hine sorhwylmas

(904)

sundornytte beheold

(667)

wl (Ax)

seleweard

tfseted;

Among the consonantal prefixes the rule admits of no exceptions (32 of 32 verses8). Zupitza's reconstruction of 2275a violates the rule: 9 (swiSe ondrae)da(S).

He gesecean sceall

(2275)

It is not counted here. When it comes to the weaker vocalic prefixes, the rule still holds (149 of 177 verses) but with sufficient exceptions (28) to show that the vocalic prefixes sometimes failed to exert the alliterative requirement. 10 The norm is represented by verses like 8

9 10

191a, 307a, 321a, 402a, 419a, 548a, 564a, 597a, 684a, 904a, 911a, 956a, 1054a, 1190a, 1577a, 1669a, 1723a, 2126a, 2141a, 2335a, 2469a, 2521a, 2554a, 2601a, 2677a, 2686a, 2748a, 2771a, 2792a, 2872a, 2944a and 3049a. Beowulf in Facsimile, p . 107. The complete list ofa-verses with double alliteration is: 24a, 29a, 55a, 87a, 186a, 199a, 206a, 247a, 250a, 292a, 313a, 315a, 318a, 331a, 369a, 455a, 476a, 477a, 553a, 567a, 633a, 635a, 650a, 654a, 658a, 666a, 667a, 697a, 713a, 721a, 727a, 763a, 775a, 781a, 783a, 811a, 846a, 877a, 884a, 915a, 923a, 929a, 934a, 942a, 975a, 992a, 999a, 1005a, 1014a, 1018a, 1028a, 1031a, 1077a, 1107a, 1113a, 1126a, 1193a, 1194a, 1227a, 1230a, 1264a, 1275a, 1295a, 1333a, 1370a, 1417a, 1439a, 1498a, 1509a, 1530a, 1605a, 1630a, 1643a, 1660a, 1681a, 1691a, 1718a, 1743a, 1755a, 1784a, 1821a, 1839a, 1860a, 1861a, 1885a, 1901a, 1934a, 1951a, 1959a, 1967a, 1984a, 2001a, 2009a, 2029a, 2040a, 2149a, 2164a, 2192a, 2206a, 2249a, 2256a, 2274a, 2293a, 2309a, 2331a, 2332a, 2359a, 2398a, 2457a, 2487a, 2508a, 2518a, 2536a, 2549a, 2566a, 2568a, 2580a, 2595a, 2655a, 2666a, 2676a, 2678a,

104

Unstressed prefixes and the copulative conjunctions gold gegangan, A

wl

A

StiSmod gestod

o36e guS nimeS

(2536)

wi5 steapne rond

(2566)

But we also find verses like A

N

wl

mawe tfcwealde, — A

N

swa he hyra ma wolde

(1055)

swa his faeder ahte

(2608)

N

wl

folcrihta gehwylc,

Since double alliteration is prohibited in the b-verse, the rule of prefixes cannot apply, and we find verses like A

ne (?a wealafe

wl

N

wige /or|?ringan

(1084)

Because they lack double alliteration, such half-lines belong to the b-verse, just as half-lines with double alliteration belong to the a-verse. Could it be that some of the a-verses with vocalic prefixes which violate the rule of prefixes are in fact b-verses transferred from their normal place? The fact that the syllabic, syntactic and metrical patterns of almost all of the twenty-eight exceptions are common in the b-verse suggests that this may be the case. To illustrate: eight of the exceptions display the metrical—syllabic pattern — | x — x (where the vertical bar represents a word boundary), as in -

|

x

-

x

rand geheawe,

(?eah Se he rof sie

(682)

The same metrical—syllabic pattern, where the wl is occupied by an unstressed prefix, occurs 106 times in the b-verse. The syntactical pattern of this verse (direct object + finite verb), which is also found in 870a and 149 la, appears in twenty-seven b-verses with the same metrical—syllabic pattern. The conclusion is inescapable that a half-line like rand geheawe 2680a, 2706a, 2711a, 2732a, 2762a, 2764a, 2784a, 2824a, 2826a, 2886a, 2889a, 2906a, 2917a, 2983a, 3014a, 3016a, 3036a, 3072a, 3073a, 3085a, 3106a, 3114a, 3122a, 3128a, 3139a, 3146a and 3l6la. The twenty-eight exceptions are: {a-) 1055a and 3078a; {be-) 680a; (ge-) 98a, 256a, 603a, 624a, 665a, 682a, 777a, 805a, 870a, 871a, 968a, 996a, 1090a, 1250a, 1375a, 1396a, 149 la, 1658a, 1857a, 1908a, 2094a, 2489a, 2608a, 2859a and 2891a. The lists of verses displaying double alliteration with either consonantal or vocalic prefixes preceding the second alliterating word would have been longer (hence, the percentage of exceptions to the second list less) if I had included verses like Grap pa togeanes (1501a). However, I regard these as properly varieties of metrical type A3 with the base following the prefix constituting the first lift and thus alliterating necessarily.

105

The Metrical Grammar of Beowulf would ordinarily have been perceived as a b-verse. Perhaps the weakness of the vocalic prefix allowed the Beowulf-poet, when it suited his purpose, to borrow these half-lines for the a-verse without offence to his metrical ear. Among the twenty-eight exceptions, however, is one group often which would seem to have moved freely between the a- and b-verses. These are the half-lines ending with the indefinites gehwylc and genog, such as: secga gehwylcum faehSo genoge,

J?ara (?e on swylc staraS feorhsweng ne ofteah

(996) (2489)

Evidently, here it is the word class which overrides the rule of prefixes, and these half-lines should be looked upon as completely unrestricted. 11 Two of the twenty-eight exceptions stand out because in one or more respects they are unique. The half-line ladgewidru (1375a) is the only verse in the poem with a type A metrical contour and the syntactical pattern, adjective + prefixed noun, that does not display double alliteration. The combination is never found in the b-verse. I find the half-line particularly interesting because it may represent a moment when the Beowulf-poet deliberately thwarted the imperatives of his metrical grammar for expressive effect. Such acts must be rare for poets who have learned their craft in an oral culture. The metrical grammar is not so much a set of artificial conventions consciously accepted as it is a description of the language of poetry — the only language the poets had to express themselves in. They certainly would not have thought of rules. A Violation' would simply have sounded unnatural and therefore been avoided. If I am right in thinking that here and there the Beowulf-poet deliberately broke the 'rules', this points to the probability that he was an oral poet who had acquired literacy and composed his poem pen in hand with the leisure for reflection and innovation that that implies. 12 In any case, this half-line appears in an eerie passage describing the haunted mere inhabited by Grendel and his 11

12

The other eight in this group are: 98a, 805a, 1090a, 1396a, 2094a, 2608a, 2859a and 2891a. The corresponding b-verses are: 148b, 412b, 936b, 2057b, 2189b, 2250b, 2450b and 2516b. Whether there can be such a thing as a 'transitional stage' between oral and written poetry is a question that was hotly debated some years ago. There seems now to be a consensus, with which I agree, that there can be, and in Old English poetry certainly was, such a stage. See Watts, Lyre and Harp, esp. pp. A2-A, 50-2 and 182-94, for a summary with references to previous scholarship, and, recently, Opland, Anglo-Saxon Oral Poetrj/, esp. p. 3. For a vigorous defence of the view that Beowulf is the transcription of an oral performance, see Creed, 'The Beoivulf-Poet\ pp. 194-216.

106

Unstressed prefixes and the copulative conjunctions mother. Critics have found in it deliberate echoes of Vergil and of a Visio Pauli,15 which suggest that the poet crafted the passage with special care. The immediate context of the half-line is this: Ponon ySgeblond up astigeS won to wolcnum, bonne wind styrej? lad gewidru, oS J?set lyft drysma)?, roderas reotaS.l4

Q373-6a)

An audience accustomed to hearing double alliteration in the metricalsyntactical pattern of lad gewidru might have experienced its absence as a vaguely sinister corollary of the terror of the place, even if the poet could not resist weaving the W alliteration of the previous line into the alliterative pattern of the half-line in its place. The other exception to the rule of prefixes which is in itself unique is cwen to gebeddan (665a). In this case the prefix is reinforced by the proclitic preposition to. The combination of nominal + prepositional phrase occurs repeatedly in the a-verse in Beowulf, always (17IX), except here, with double alliteration. 15 And 665a is unique in another respect. It is the only instance of the metrical—syllabic pattern — | x x — x in the a-verse without double alliteration. Altogether it has an odd look. Possibly the text is disturbed: the problematic form Kyningwuldor appears in the same line. But perhaps here again the Beowulf-pott deliberately disregarded the rules of his metrical grammar. If we could be sure of the tone of lines 662—5a, it would be easier to judge. Hrothgar has just turned over the great hall Heorot to Beowulf for safekeeping against Grendel, the first time he has ever entrusted his stronghold to another man. And then the poet says, Da him Hrojpgar gewat mid his haele|?a gedryht, eodur Scyldinga ut of healle; wolde wigfruma Wealh(?eo secan, cwen to gebeddan.16 13 14

15

16

(662-5a)

See Klaeber's notes to 1357ff. 'From it the tossing waves mount up black to the clouds, when the wind stirs up hostile storms, until the air becomes gloomy, the skies weep.' I do not count \fuglurri\ to gamene (294la), which is an editorial conjecture. The fact that gamene does not alliterate, however, points to a problem which goes beyond the need to supply an alliterating first lift. 'Then Hrothgar departed with his band of warriors, the protector of the Scyldings, out of the hall; the war-prince wanted to seek Wealhtheow, the queen, for his bed-companion.'

107

The Metrical Grammar of Beowulf Much depends on our estimate of Hrothgar (to which of course this passage contributes). Is he the model of prudential kingship, free from the rash impetuosity of a Beowulf who would risk leaving his people leaderless for the sake of his heroic pride? Or is he in some respects a pathetic figure, impotent in the face of a terrible threat to his people, forced to rely on the heroic spirit of another man, which only casts his own helplessness into strong relief?17 Could there be a hint of mockery in these lines, where the language of heroic action — 'Then Hrothgar departed with his band of warriors, the protector of the Scyldings, out of the hall; the war-prince wanted to seek . . . ' — seems to be deflated by the domestic anticlimax of the last two phrases? Looking at it from another perspective - that of the relationship between Wealtheow and Hrothgar, Helen Damico has called attention to the gratuitous eroticism of this passage and its importance for our understanding of Wealtheow's role in the poem. 18 However we choose to read it, the absence of double alliteration in 665a ruptures the norms of the metrical grammar and may have been deliberately intended to emphasize the unusual nature of the moment. It is possible for the w l to follow two lifts and still be occupied by an unstressed prefix. This happens in type D 4 (/ / x \ ) . Here the possibility of alliteration is blocked by the alliterative rules (in the a-verse the second lift must alliterate in preference to a half-lift) and the rule of prefixes does not apply. When a non-alliterating prefixed base word is inserted into a type D 4 , it goes by preference into the b-verse (47 X). Only on four occasions is it found in the a-verse, and then always with the vocalic prefix ge-\ Fyrst forS gewat;

flota

atol ySa geswing

waes on ySum

(210)

eal gemenged

Metod manna gehwaes.

(848)

Ic eom on mode from

eald enta geweorc

anne mannan

(2527) {211 A)

There is one irregular b-verse of this type: AN

hat ond hreohmod

A

hlaew oft ymbehwearf

(2296)

4

I agree with Pope that this should be scanned as a type D ; Bliss classifies it as a type E. Yet double alliteration involving a half-lift is just as irregular 17

18

For a reading of Beowulf which generally casts Hrothgar in a favourable light, see Goldsmith, Mode and Meaning. For one which emphasizes his relative impotence, see Irving, Reading. Beowulf's Wealhtheow, pp. 139-45.

108

Unstressed prefixes and the copulative conjunctions

in the b-verse as double alliteration on the two principal lifts. Although the verse is clearly anomalous, it does testify to the strength of the alliterative imperative generated by the unstressed prefix. When the unstressed prefix occupies the w2, the rule of prefixes states that alliteration of a base word in the second lift is optional. The principal metrical possibility is type B. There are eighteen instances of double alliteration in the a-verse, as in A

A

Da se gaest ongan

gledum spiwan

(2312)

Set these against the forty-eight instances of single alliteration, A

19

as in

N

hyre beam gewraec,

beorn acwealde

(2121)

There are two irregular verses of this type. The first alliterates on the second lift: N

A

he is manna gehyld —

hord openian

(3056)

The other displays double alliteration in the b-verse: A

HwaeJ?ere me gesaelde,

A

J?aet ic mid sweorde ofsloh

(574)

Twice an unstressed prefix occurs in the w2 in an expanded type E. One is in the a-verse, where alliteration would be optional. It does not in fact generate double alliteration: N

AN

st/3[r]a naegla gehwylc

style gelicost

(985)

The other is in the b-verse, where alliteration is blocked: AN

fuslic f(yrd)leo3.

N

FeJ?a eal gesaet

(1424)

An unstressed prefix is found in one expanded type D with double alliteration: A

A

N

oncyS eorla gehwaem, 19

sy3)?an ^scheres

(1420)

Double alliteration: 79a, 362a, 756a, 760a, 800a, 1360a, 1484a, 1696a, 2045a, 2208a, 2259a, 2287a, 2312a, 2353a, 244la, 2505a, 2767a and 2879a; single alliteration: 25a, 26a, 74a, 80a, 88a, 229a, 294a, 374a, 464a, 662a, 808a, 853a, 860a, 882a, 903a, 907a, 1040a, 1236a, 1365a, 1469a, 1608a, 1662a, 1673a, 1684a, 1705a, 1773a, 1781a, 1872a, 2033a, 2111a, 2121a, 2367a, 2369a, 2387a, 2397a, 2418a, 2484a, 2620a, 2633a, 2685a, 2694a, 2752a, 2773a, 2808a, 2838a, 3068a, 31l6aand 3166a.

109

The Metrical Grammar of Beowulf Since the base word does not occupy the second lift, alliteration is blocked. It also is found in two other verses — one hypermetric and the other of doubtful scansion: A

N

N

arfaest set ecga gelacum.

Spraec 3a ides Scyldinga AN

gesaet on sesse;

(1168)

N

seah on enta geweorc

(2717)

Here too the base word does not occupy the second lift and alliteration is blocked (of course alliteration is blocked in 2717b because it is a b-verse). THE COPULATIVE CONJUNCTIONS

Copulative conjunctions link parallel forms, like nouns, verbs or adverbs, as in (with the conjunctions italicized) brond ne beadomecas wanode ond wyrde.

bitan ne meahton

(1454)

He aet wige gecrang

(1337)

nean ond feorran

nearofages ni3

(2317)

They behave like proclitic particles, in contradistinction to the adverbial conjunctions which join clauses together and invariably move to the initial dip of their clause. The principal copulative conjunctions are ne, ond, o&de and swa. When they link two words within a half-line, as in the three examples above, they give equal metrical status to both words. The resulting phrase must be considered a stressed element, even if the words which have been linked are themselves sentence particles. This proposition is not so self-evident as it may seem. To be sure, when two nouns or two adjectives which are not linked by a conjunction appear in the same half-line, we expect both to receive metrical stress and the first, at least, to alliterate. But when two verbs or two adverbs or a verb and an adverb which are not linked by a conjunction appear together, the case is different. In the a-verse it is often only the second which alliterates, for example: A

Sona )?aet onfunde A

mynte J?aet he gedaelde, A

alegdon Sa tomiddes

fyrena hyrde

(750)

aer (?on daeg cwome

(731)

maerne [?eoden

(3141)

Evidence like this leads to the conclusion that primary metrical stress falls 110

Unstressed prefixes and the copulative conjunctions

on the second element, and that, when double alliteration occurs, alliteration on the first element is extra-metrical, as in a

A

Sona |?aet gesawon

snottre ceorlas

(1591)

However when a copulative conjunction links two sentence particles — two verbs or two adverbs — the first always alliterates, and there are even a few instances when the second does not, despite the general propensity of proclitics to signal alliteration. The formula nean ondfeorran is a case in point: A

N

nean ond feorran

]>u nu hafast

(1174)

If this formula were scanned by the transformational rule it would alliterate on F. The fact that it appears both in the a-verse (1174a) and in the b-verse (2317b) shows that the adverbs are stressed elements. We must conclude that the copulative conjunction has the effect of assigning the same metrical weight to each of the words it joins. The copulative conjunction is not easily distinguished from the adverbial conjunction in certain circumstances. Fortunately the uncertainties lead mainly to problems of classification rather than to fundamental issues in the metrical grammar. When the conjunction joins two verbs as in 1337a quoted above, we might reasonably conclude that each verb belongs by definition to its own separate clause, and thus that the conjunction is adverbial. If so, the conjunction is a sentence particle which is properly in the first dip of its clause. The overall clause structure becomes more complicated but there are no other consequences as far as the metrical grammar is concerned. When the conjunction is found at the beginning of the half-line, it is metrically ambiguous. The half-line may be clauseinitial (I), in which case the conjunction is adverbial and a sentence particle, or it may be clause-non-initial (II), in which case the conjunction is copulative and proclitic. Consider, for example, ondhtelepa beam (1189b) and ondhalig God(1553b). There is no way of determining by inspection of the half-lines alone whether ond is adverbial or copulative in these two verses. Only by looking at the context can we learn which it is. The first is copulative: Hwearf |?a bi bence, J?aer hyre byre waeron, HreSric ond HroSmund, ond haelej?a beam, giogoS aetgaedere . . . 20 20

(1188-90a)

'Then she moved to the bench where her sons were, Hrethric and Hrothmund, and the sons of the warriors, the young men together . . . ' 111

The Metrical Grammar of Beowulf And the second is adverbial: Haefde 3a forsiSod sunu EcgJ?eowes under gynne grund, Geata cempa, nemne him heaSobyrne helpe gefremede, herenet hearde, ondhalig God geweold wigsigor . . . 21

(155O-4a)

Sometimes the context leaves us in doubt. Elliptical constructions in particular may be analysed in two ways, as in Him waes ful boren, wordum bewaegned, estum geeawed . . . 22

ond freondla(?u ond wunden gold (1192-4a)

I prefer to take these as three consecutive clauses with ellipsis of him and the verb to be. But the ellipsis has the effect of making a single action - 'to him three things were done'. Logically one could interpret the ond's as copulatives. Again, it makes little difference except for purposes of classification. Except where the alliterative rules of the metrical grammar prevent it, the base word which follows the copulative conjunction nearly always alliterates. When the conjunction occurs in the wl before the first lift, alliteration of the base word of whatever class is universal as is to be expected (e.g., 1189b, quoted above). When it occurs in the wl after the first lift, alliteration of the base word in the second lift is found in slightly more than 94 per cent of all relevant a-verses (it is of course prohibited in the b-verse). There are 135 such a-verses with double alliteration, and eight with single. The eight exceptions to the alliterative requirement are instructive. 23 They are: A

N

folc ond rice,

21

(?onne 3u forS scyle

(1179)

'The son of Ecgtheow, the c h a m p i o n of t h e Geats, would have perished then beneath the wide earth, if his battle-corslet had not helped h i m , his hard war-net; and holy G o d b r o u g h t about victory in war

22

'The c u p was borne to h i m , and friendship was offered w i t h words, and twisted gold was

23

Bliss, M e t r e , § 1 1 5 , refers t o these as phrases ' w h i c h could n o t be used at all w i t h o u t

bestowed on h i m w i t h good will . . . "

breaking the rule of double alliteration'.

112

Unstressed prefixes and the copulative conjunctions A

N

|?aet is soS Metod 24

saela ond maela; (Ax)

(1611)

(Nx)

duguj?e ond geogoj>e A

dael aeghwylcne

(621; also 1674a)

N

geongum ond ealdum, (Ax)

swylc him God sealde

(72)

N

Iofore ond Wulfe A

mid ofermaSmum

nean ond feorran (Ax)

(2993)

N

Ipu nu hafast

(1174)

o3 Saet deaSes wylm

(2269)

N

daeges ond nihtes,

Word pairs are a common feature of all of the Germanic languages, indeed, of the Indo-European languages generally, both in verse and in prose. 25 In the stereotyped phraseology of Modern English three structural principles account for many, perhaps most, phrases made up of paired elements. Thus we have (1) alliteration: 'time and tide', 'this and that', 'spick and span'; (2) contrast: 'stop and go', 'black and white', 'life and death'; and (3) rhyme: 'stars and bars', 'by hook or by crook', 'lean and mean'. My own casual attempts to come up with examples in each category suggest that alliteration and contrast are now about equally productive principles, with rhyme running a distant third. 26 However that may be, the interesting fact is that the same principles clearly governed Old English phraseology. We would expect to find alliteration of course, 27 but six of the eight exceptions to the rule calling for double alliteration in the a-verse exhibit contrast or rhyme in its place. 'Young and old', 'near and far', 'day and night' were then, as now, commonplaces of everyday speech. 'Seals and meals' survived in the East Anglian dialect at least into the nineteenth century. 28 The contrastive formula 'douth and youth' (the rhyme is an accidental feature of the later forms of the words) died out earlier, though it

24

N o t i c e the cross alliteration.

25

See K o s k e n n i e m i , Repetitive Word Pairs, p p . 9 8 - 1 0 6 .

26

S m i t h , Words and Idioms, p p . 1 7 3 - 5 , gives several lists of paired phrases. H e recognizes, in addition to alliteration, r h y m e and contrast, e m p h a t i c repetition as a structural principle (which could be regarded as c o m b i n i n g alliteration and rhyme).

27

Approximately t h i r t y - e i g h t per cent of the word pairs in the prose of the O l d English and early M i d d l e English periods w h i c h K o s k e n n i e m i sampled alliterated {Repetitive Word Pairs, p . 8 8 ; the percentage includes the alliteration of S w i t h SC, SP and ST).

28

See OED, s.v. 'sele'.

113

The Metrical Grammar of Beowulf must have been common enough in the Old English period. 29 The OED cites what looks to be a thirteenth-century echo of the formula in The Owl and the Nightingale.^0 Contrast and rhyme are means of supplying the emphasis demanded by the proclitic conjunction which the Beowulf-poet apparently felt he could adopt as alternatives to alliteration on a very limited basis. 31 If we take these six half-lines as licensed deviations, there are really only two major exceptions to the generalization that alliteration in the a-verse is obligatory after the copulative conjunction in this position. Iofore ond Wulfe (2993a) illustrates the dictum that proper names make their own rules, if need be, in poetry. 32 The other exception, folc ond rice (1179a), is odd because there are in the tradition at least two more or less synonymous verses (one from Beowulf itself) which supply the missing alliteration: folc o)?3e freoburh, folc and foldan.

|?aer he afeded waes

(693)

Foh hider to me

(Christ and Satan 685; also Battle of Maldon 54)

Neither may be a precise equivalent, but it is hard to see a compelling reason why the Beowulf-pott should have ignored the alliterative requirement of the conjunction in this one case. Though no supporting evidence happens to survive, folc ond rice appears to be a b-verse formula (cf. hord ond rice, 2369k and 3004b), which by design or accident has strayed into the a-verse. When the copulative conjunction occupies the w2 in the a-verse, alliteration of the base word appears to be optional. On the one hand we find fourteen half-lines with double alliteration, 33 for example: 29 30 31

32

In addition to Beowulf 160b, 621a and 1674a, cf. Andreas 152b and 1122b. S.v. 'douth'. The claim is not that contrast and rhyme are formal equivalents of alliteration. They cannot be because they are found, as alliteration is not, in the b-verse. It is only that these features offer an additional degree of markedness which seems to have been sufficient to override the alliterative requirement on these occasions. In view of the fact that Eofor and W u l f are brothers, it may be observed that brothers' names usually alliterate in early G e r m a n i c cultures; hence alliteration would be the n o r m for half-lines of this k i n d : for example, Heorogar ond Hro&gar (6 la), Herebeald

ondHadcyn

(2434a) etc. 33

511a, 642a, 1063a, 1248a, 1549a, 1696a, 1751a, 1763a, 1767a, 1787a, 1864a, 2105a, 243 la and 2472a.

114

Unstressed prefixes and the copulative conjunctions Pa waes eft swa aer J?aet l?ec adl oSSe ecg

inne on healle eafoj?es getwaefeS

(642) (1763)

On the other hand we find five half-lines with single alliteration, but always with either contrast or rhyme, 34 for example: (?aette suS ne norS hu he frod ond god

be saem tweonum

(858)

feond oferswySe)?

(279)

In the b-verse the copulative conjunction rarely appears in the w l before the second lift or in the w2. Nineteen verses fall in the former category; five in the latter. Among the nineteen are four with proper names (498b, 1148b, 2472b and 2946b) and five which display contrast (94b, 160b, 718b, 839b and 2317b). The remaining ten 35 are verses like hroden hiWecumbor,

helm ond byrnan

(1022)

These verses, which resemble folc ond rice (1179a), can be formally labelled as b-verses. Only when the feature of contrast or rhyme (or proper name) is added can this kind of verse be admitted on a limited basis to the a-verse. The five verses with the copulative conjunction in the w2 3 6 include one which displays contrast (1221b) and one with rhyme (1008b). In sum, the copulative conjunctions behave precisely like the unstressed prefixes with respect to alliteration. The rule of prefixes can be readily expanded to incorporate this fact. 34

Contrast: 8 5 8 a , 2 5 0 0 a and 2 8 7 0 a ; r h y m e : 2 7 9 a and 6 5 6 a . It may be purely a m a t t e r of chance t h a t there are no half-lines in this g r o u p w h i c h lack b o t h contrast and r h y m e . Alternatively, it may be t h a t t h e copulative conjunction imposes a stronger alliterative r e q u i r e m e n t t h a n any of t h e other proclitics.

35 36

1022b, 1629b, 2369b, 2775b, 2868b, 2922b, 2943b, 3004b, 3047b and 3163b. 1008b, 1221b, 1700b, 2659b and 3157b.

115

The alliterative requirement of prepositions and the proclitic adverbs and instrumental

PREPOSITIONS

The terminological problems connected with a discussion of prepositions are formidable. Depending on their use, the same forms may be variously described as 'prepositions', 'adverbs', 'prepositional adverbs', 'postpositions', 'separable prefixes' or 'inseparable prefixes'.1 As their name implies, the usual placement of prepositions is before the object they govern. In this position they are normally proclitics, and behave like the other proclitic particles. However, prepositions are subject to internal displacement. That is, they may be moved from their proclitic position and placed after their object within the same half-line. The normal position for prepositions can be illustrated by (with the prepositions italicized): Hwaet, we Gar-Dena

in geardagum

(1)

The same preposition, internally displaced, appears a few lines later in (Ax)

Scyldes eafera

N

Scedelandum in

(19)

Displacement removes the preposition from a weak metrical position and assigns it metrical stress. Displaced prepositions will not be considered here. 2 There are two 'prepositions' in normal position in Beowulf which must be stressed — ongean and samod. The half-lines in which they occur are formulae. After Beowulf's victory over Grendel, Hrothgar presents him with gifts, including a helmet: 1

2

The list of terms, with a mark of interrogation following, is Mitchell's chapter heading for a discussion of these problems: Old English Syntax, §§ 1060-80. For discussion, see below, ch. 11.

116

Prepositions and the proclitic adverbs Ymb (?aes helmes hrof heafodbeorge wirum bewunden waLz utan heold, )?aet him fela laf frecne ne meahte scurheard sce)?5an, bonne scyldfreca A

(A)

gangan scolde.3

ongean gramum

(1030-4)

Here ongean is used so closely with gangan that it comes near to being the equivalent of a separable prefix.4 For want of a better term it can be described as an adverbial preposition and marked as a sentence particle which has been displaced. Samod appears twice in Beowulf (and only in Beowulf) as a stressed 'preposition' in the half-line formula: sigoreadig secg.

(Ax)

Samod aerdaege

(1311; also 2942b)

The formula seems to be identical in meaning to the phrase mid cerdcege, which, however, alliterates as one would expect on cer-. Samod cannot be said to be displaced in 131 la, since it immediately precedes its object and the half-line is clause-initial. Whatever the explanation for it may be, it functions as a stressed element. Unstressed syllables which come between a preposition and its object must themselves be proclitic, for example (with all the proclitic syllables italicized): Beowulf Geata

be pcem gebroSrum twaem

(1191)

But if a stressed element, such as a descriptive adjective or a dependent genitive, comes between the preposition and its object, it will take alliterative priority over the object, as in A

N

ymb bronte ford A

ofer myrcan mor, 3

4

brimliSende

(568)

A

mago]?egna basr

(1405)

'A ridge wound with wires held head-guard on the outside about the crown of the helmet, so that no shower-hardened leaving of files could injure him severely, when the shield-warrior had to advance against hostile foes'; cf. Elene 43a, Juliana 628a, The Battle ofMaldon 100a. Ongean must take alliterating stress, because otherwise the first lift would fall on the short penultimate syllable of gramum and the resulting irregular 'short A 3 ' would neither contain a sentence particle in the w l nor be clause-initial. In Juliana 628a, ongean is closely associated with the verb biseah\ in The Battle ofMaldon 100a, it is associated with stodon. In Elene 43a, it is rather more loosely associated with bannan to beadwe. Campbell, Old English Grammar, §§78—80, gives ongean-pingian 'to speak against' as an example of a stressed 'prepositional adverb' used with a verb to form a 'quasi-compound'.

117

The Metrical Grammar of Beowulf A

N

under ySa gewin

aldre gene|>an

(1469)

When the object of the preposition is a sentence particle, such as a pronoun, the prepositional phrase forms a compound sentence particle. In this case the object will not be metrically stressed if the phrase appears in the initial dip of the verse clause, but if the phrase is internally displaced, then the object will acquire metrical stress. The prepositional phrase is not stressed in this clause-initial (I) half-line: (Ax)

purh hwat his worulde gedal

weorSan sceolde

(3068)

But the prepositional phrase has been internally displaced and is therefore stressed in A

cystum gecyj?ed,

N

ond his cwen mid him

(923)

An object of a preposition which occupies the first lift necessarily alliterates in accordance with the fundamental alliterative principle. Prepositional phrases of this kind occur freely both in the a- and the b-verse, for example: on hand gehwearf;

he geheold tela

heaSolaces hal

to hofe gongan

(2208) (1974)

But the prepositional phrase frequently follows the first lift. This forms one of the Beowulf-poet's favorite metrical—syntactical patterns. It belongs almost exclusively to the a-verse. To determine the alliterative requirement, it is once again necessary to observe whether the weak metrical position in which the preposition occurs is the first or the second of the half-line. There are 210 verses where the preposition occupies the wl between two lifts. 5 Of these 181 are a-verses; twenty-nine are b-verses. Representative half-lines are: (Ax)

A

cyning on cor(?re,

ond seo cwen numen A

sweord swate fah

5

(1153)

N

swin ofer helme

(1286)

I do not include in this count a-verses of the type: alliterating sentence particle + prepositional phrase. In these verses the object of the prepositional phrase always alliterates.

118

Prepositions and the proclitic adverbs Of the 181 a-verses, 179 display double alliteration. 6 The two exceptions are: A

N

cwen to gebeddan. A

Haefde Kyningwuldor

(Nx)

[fuglum] to gamene.

Frofor eft gelamp

(665) (2941)

The unique nature of 665a was discussed in ch. 8. As for 294la, Thorpe's conjectural reconstruction cannot stand as a counter example, despite the fact that a word alliterating on F seems to be required. The problem in lines 2940—1 runs deeper than this single apparent omission. Possibly there is a gap of one or more lines here. 7 The disproportion between a-verses and b-verses of this type is even greater than the ratio 181:29 would suggest. Fifteen of the twenty-nine b-verses have the syntactical pattern: sentence particle + prepositional phrase, 8 as in witena welhwylc

wide geond eor)?an

(266)

As noted above, I have excluded this pattern from the a-verse total, because I believe that the alliteration on the sentence particle in the a-verse is extra-metrical (the object of the preposition always alliterates; the sentence particle sometimes does not; altogether there are seventy-one such a-verses 6

7 8

The 179 verses with double alliteration are: Ba, 14a, 36a, 46a, 48a, 52a, 56a, 81a, 89a, 95a, 110a, 140a, 145a, 152a, 169a, 172a, 195a, 211a, 213a, 216a, 224a, 248a, 249a, 265a, 310a, 342a, 404a, 420a, 426a, 438a, 440a, 473a, 475a, 497a, 534a, 540a, 555a, 580a, 593a, 651a, 672a, 685a, 733a, 747a, 748a, 754a, 802a, 856a, 930a, 931a, 953a, 962a, 970a, 995a, 1025a, 1117a, 1134a, 1138a, 1153a, 1162a, 1168a, 1207a, 1217a, 1237a, 1255a, 1289a, 1302a, 1313a, 1361a, 1366a, 1371a, 1374a, 1388a, 1404a, 1525a, 1574a, 1588a, 1595a, 163 la, 1643a, 1649a, 1654a, 1656a, 1706a, 1709a, 1720a, 1775a, 1790a, 1805a, 1815a, 1830a, 1876a, 1916a, 1924a, 1926a, 1941a, 1961a, 1964a, 1978a, 1983a, 1990a, 1997a, 2005a, 2007a, 2014a, 2021a, 2099a, 2117a, 2243a, 2,262a, 2268a, 2276a, 2281a, 2307a, 2310a, 2314a, 2328a, 234la, 2378a, 2411a, 2415a, 2423a, 2433a, 2446a, 2448a, 2458a, 2459a, 2461a, 2498a, 2506a, 2512a, 2519a, 2530a, 2539a, 2556a, 2559a, 2562a, 2575a, 2578a, 2585a, 2600a, 2625a, 2662a, 2700a, 2709a, 2715a, 2739a, 2743a, 2759a, 2765a, 2768a, 2781a, 2783a, 2790a, 2793a, 2803a, 2816a, 2876a, 2908a, 2914a, 2957a, 2990a, 2996a, 3016a, 3023a, 3025a, 3039a, 3043a, 3060a, 3065a, 3091a, 3095a, 3103a,' 3132a, 3138a, 3145a, 3148a, 3157a and 3167a. See Chambers's note to these lines, Wyatt-Chambers, p. 145. 266b, 642b, 663b, 724b, 1128b, 1418b, 1804b, 1840b, 1909b, 2368b, 2557b, 2717b, 2863b, 2899b and 3099b.

119

The Metrical Grammar of Beowulf

with double alliteration). Thus, only fourteen b-verses, 9 like 1286b quoted above, are directly comparable with the 181 a-verses like 1153a. From these facts two conclusions can be drawn: (1) the twenty-nine b-verses represent patterns which are confined by the rules of the metrical grammar to the second half-line; (2) they, and especially verses like 1286b, are rare because the alliterative requirement of the preposition in the first weak metrical position is so strong — stronger than that either of the unstressed prefix or of the copulative conjunction. The preposition is rarely found in the w2 of the half-line before the second lift (8X in the a-verse; 12 X in the b-verse). When it does appear, alliteration proves not to be mandatory in the a-verse (and of course is prohibited in the b-verse). In more than half of these verses the preposition governs a pronoun (i.e., the prepositional phrase has been internally displaced), 10 for example: A

w^du weallendu.

w2

N

No ic wihtfram \>e

(581)

Three times in the a-verse and once in the b-verse the preposition governs a non-alliterating nominal (in one case an inflected infinitive): A

N

Gefeng |>a be eaxle A

- nalas for faehSe mearn

(1537)

N

To lang ys to reccenne, A

hu i(c 3)am leodsceaSan

N

byrne ne meahte 11

Born bord wid rond,

A

fyrene gefremede

(2093) (2673)

N

— he fag wid God12

(811)

Line 1537a is a nearly unparalled instance in the a-verse of a finite verb alliterating in preference to a nominal. Rieger's emendation, Gefeng pa be feaxe, seems virtually certain. 13 The fact that the Beowulf-poet so rarely exercised the option of making a non-alliterating nominal the object of a preposition, even where the rules of his metrical grammar would seem to permit it, is further proof of the force of the preposition's alliterative 9

10

11 12 13

101b, 295b, 523b, 6l4b, 643b, 649b, 708b, 1049b, 1286b, 1907b, 1968b, 2452b, 2565b and 2705b. 525a, 2948a (neither of the a-verses displays double alliteration; 525a must be scanned by the transformational rule); 137b, 54lb, 581b, 879b, 923b, 1722b, 1723b, 2149b,' 2248b, 2591b and 2845b. Klaeber: bord wid rond. For the reading adopted here, see Pope, Rhythm, p. 320. Klaeber: he [wees] fag wid God. See above, ch. 6.

120

Prepositions and the proclitic adverbs

requirement. There are three cases of double alliteration in the a-verse when the preposition falls in the second weak metrical position. The syntactical pattern is the same in each: two prepositional phrases linked by a copulative conjunction. The three are: A

A

ge cet ham ge on herge, A

A

wid ord ond wits ecge A

A

ge wi6 feond ge wid freond

ge gehwae|?er J>ara

(1248)

ingang forstod

(1549)

fseste geworhte 14

(1864)

The only comparable line I have found elsewhere in Old English poetry is from The Seafarer (the b-verse is defective): A

A

wip leofne ond wiS la(?ne

(112a)

Obviously one cannot generalize on the basis of so small a body of evidence, but I suspect that, had we more, double alliteration might prove to be the norm for this pattern even though the second preposition falls in the w2. Thus the rule formulated for unstressed prefixes and extended to the copulative conjunctions applies equally well to prepositions. THE PROCLITIC ADVERBS AND THE INSTRUMENTALS IN COMPARISON

Certain sentence particles become bound to other forms and thus function as proclitics. They can be described as the proclitic adverbs and the instrumentals in comparison. The proclitic adverbs

One group of proclitic adverbs is made up of the intensifies — swa, pees, pus and to — which are proclitic on a following adjective or adverb. Swa appears four times as an intensifier in the w l before the first lift: J?aet we hine swa godne swa deorlice

daed gefremede

gedeS him swa gewealdene on swa geongum feore 14

gretan mo ton

Note also the internal rhyme in 1864a. 121

worolde daelas guman (?ingian

(347) (585), (1732) (1843)

The Metrical Grammar of Beowulf Once it appears in the w2, where it does not lead an alliterating form: (Nx)

A

J?aet naefre Gre[n]del swa fela

gryra gefremede

(591)

The intensifier cannot always be distinguished from the adverb which modifies the whole clause. If the adverb is the only element in the wl before the first lift, as in 585a, the position of the half-line in the verse clause determines whether or not it is the intensifier. A clause-non-initial half-line can have nothing but proclitics before the first lift. A clauseinitial half-line must have at least one sentence particle in the initial dip of the clause. Compare 585a with Swa manlice (1046a). In isolation the two half-lines are ambiguous. In context 585a is clause-non-initial (II): Breca naefre git aet heaSolace, ne gehwaej?er incer, swa deorlice daed gefremede fagum sweordum . . .

(583b-86a)

Whereas 1046a is clause-initial (I): Swa manlice maere f>eoden, hordweard haele)?a hea^oraeses geald mearum ond madmum, swa hy naefre man Iyh5 . . . (1046-8) Therefore 583b—86a should be translated: 'Neither Breca nor you has ever yet at battle-play performed so bold*, deed with shining swords' (proclitic); and 1046—8 should be translated: "Thus the illustrious prince nobly repaid [Beowulf's] battle-rush with horses and treasures, so that no man will ever fault them' (sentence particle). The intensifier pus can likewise be identified in the clause-non-initial (II) half-line 337a: pus manige men

Ne seah ic elf?eodige modiglicran.15

(336b-37)

However in 238b and 430b pus remains ambiguous. The first is in the coastguard's challenging question: Hwaet syndon ge searohaebbendra, J?e pus brontne ceol byrnum werede, ofer lagustraete laedan cwomon, hider ofer holmas? 15

'I never saw so many foreign men so brave.' 122

(237^Oa)

Prepositions and the proclitic adverbs

The second appears in Beowulf's initial petition to Hrothgar: Ic }>e nu 5a, brego Beorht-Dena, biddan wille, eodor Scyldinga, anre bene, )?aet 5u me ne forwyrne, wigendra hleo, freowine folca, nu ic pus feorran com, J?aet ic mote ana [ond] minra eorla gedryht, |?es hearda heap, Heorot faelsian.

(426b-32)

The first could be translated either: 'What kind of warriors, clothed in corslets, are you who have come hither over the waters bringing so tall a ship over the sea-road', or 'who have come thus bringing hither over the waters a tall ship . The second could be translated either: 'Now then, chief of the Bright-Danes, prince of the Scyldings, I intend to ask you one favour, that you, protector of the warriors, friendly lord of the peoples, not deny me, now that I have come from so far, that I alone and my band of earls, this stout troop, may cleanse Heorot', or 'now that I have thus come from afar . I prefer taking pus as an intensifier (the first alternative in each case) because this gives the more dynamic reading. The adverbial intensifier pees appears four times in the w l , 1 6 as in faeger foldbold;

ac he pees faeste waes

(773)

It is found twice in the w2: waepna gewealdan, forbarn brodenmael;

ac hine wundra/><#r fela waes )?aet blod to pees hat

The intensifier to appears ten times in the w l , faehSe ond fyrene;

17

(1509) (1616)

as in

waes to faest on J?am

(137)

leode mine

(1336)

for|?an he to lange

One of these ten violates Kuhn's second law (the first to is the intensifier): To lang ys to reccenne,

hu i(c 5)am leodsceaSan 18

19

(2093)

This verse is problematic in several respects. Pope and several editors accept Sievers's proposal to emend the last part of the verse to to reccan, which solves one problem (although according to Bliss it creates another). The first to might be a scribal anticipation. Compare: 16 17 18

773b, 968b, 1366b and 1508b. 905a, 1336a, 1748a, 1930a, 2093a; 137b, 694b, 969b, 2289b and 2468b. 19 See Bliss, Metre, §§ 44, 47 and 87. Pope, Rhythm, p. 310. 123

The Metrical Grammar of Beowulf larum lifgan.

Long is to secganne

{Riddle 39, 22)

If we delete to, which eliminates the violation of Kuhn's second law, and accept Sievers's emendation, the result is an unexceptionable b-verse: *Lang is to reccan. However the rule of proclitics now requires double alliteration in the a-verse (because the proclitic to is in the wl), and we are no better off than before. The verse remains a problem. The intensifier to also appears seven times in the w2, always in the b-verse. 20 An example is: wergan gastes;

waes J?aet gewin to strang

(133)

The other proclitic adverbs are various. Most numerous is the negative particle ne, which is always proclitic on a following verb (139X). 21 The verb may appear without alliterating in the initial dip of the verse clause, as in Paer waes madma fela offeorwegum frsetwa gelaeded; ne hyrde ic cymlicor ceol gegyrwan hildewaepnum ond heaSowsedum, billum ond byrnum . . . 22

(36b-40a)

Verses like this are invariably clause-initial. The verb may appear in the a-verse with extra-metrical alliteration (a) in the initial dip, as in a

A

ne gefeah he |?aere faehSe, a

ac he hine feor forwrsec

(109)

feower madmas

(1027)

A

ne gefraegn ic freondlicor

These too are always clause-initial. The verb necessarily occupies the first alliterating lift in the a-verse (as a result of the transformational rule) seventeen times, 23 for example: 20

133b, 191b, 1742b, 2461b, 2684b, 2882b and 3085b.

21

See Bliss, 'Auxiliary and Verbal', p . 1 6 1 , n. 17. W i t h non-verbs and w i t h certain c o m m o n verbs ne contracts, e . g . , nealles, nathwylc,

22

ncebben etc.

'There m a n y treasures from distant places, o r n a m e n t s , were b r o u g h t ; I have not heard of a ship being m o r e splendidly e q u i p p e d w i t h battle-weapons and g a r m e n t s , w i t h swords and corslets

23

429a, 503a, 706a, 751a, 778a, 798a, 937a, 967a, 1082a, 1130a, 1142a, 1465a, 1508a, 2124a, 2665a, 2848a and 3081a.

124

Prepositions and the proclitic adverbs

for(?on J?e he ne uj?e, £aes ne wendon aer

J?aet aenig oSer man

(503)

witan Scyldinga

(778)

In these verses the proclitic ne is always preceded by at least one sentence particle and the verses are all clause-initial. A stressed verb accompanied by the proclitic ne occupies the first alliterating lift in the b-verse in ic ne wat hwaeder

waelgaest waefre; manigra sumne; laSra spella;

(1331)

hyt ne mihte swa

(2091)

he ne leag fela

(3029)

Here again the ne is preceded by a sentence particle and the verses are clause-initial. A verb which is modified by the proclitic ne never appears in the wl following the first or second lift; the verb is always thrust into the second lift or into a half-lift. There are only four such a-verses. In each the ne precedes the second lift and double alliteration is the rule: A

A

lade ne letton. A

Leoht eastan com

A

weana ne wende

(569)

to widan feore (933; similarly 2923a)

A

A

wiston ond ne wendon,

|?aet hie heora winedrihten

B-verses of this type are more numerous (48 X), (Ax)

in modsefan,

A

for his wonhydum

hildebille,

N

Men ne cunnon A

N

waepna ne recceS (Ax)

N

oferhyda ne gym AN

(1604)

for example:

N

Metod hie ne cul?on

murnende mod.

ece raedas;

24

N

hond sweng ne ofteah

(180) (50) (434) (1760)

(1520)

Of these, 180b is necessarily clause-initial because of the sentence particle hie in what becomes the initial dip of the clause. All the rest are unrestricted. 24

50b, 119b, 154b, 162b, 180b, 182b, 185b, 243b, 246b, 434b, 462b, 600b, 862b, 878b, 941b, 1032b, 1100b, 1233b, 1290b, 1377b, 1446b, 1454b, 1468b, 1514b, 1515b, 1520b,1596b, 1757b, 1760b, 1993b, 1995b, 2067b,2184b,2340b,2370b, 2451b, 2464b, 2466b, 2476b, 2489b, 2583b,2600b,2673b,2827b,2904b,2953b, 3053b and 3067b.

125

The Metrical Grammar of Beowulf A verb with proclitic ne never appears in the w2 (this would violate Kuhn's first law); the verb is always thrust into the second lift. There are two a-verses of this type: 25 A

N

He beot ne aleh, A

beagas daelde

(80)

N

forSam me witan ne Searf

Waldend

fira

(27'41)

Neither displays double alliteration. Once again, the b-verses are more numerous (20X). 2 6 An example is: A

mearcaS morhopu;

N

no Su ymb mines ne |>earft

(450)

Four verses in this group are of particular interest: A

N

J?aet heo J?one fyrdhom

Surhfon ne mihte (1504; similarly 1877b)

under swegles begong

gesacan ne tealde

(Ax)

(Ax)

Heald \>u nu, hruse,

N

N

nu haeleS ne mostan

(1773)

(2247)

They appear to have three weak metrical positions, with the ne occupying the second. Since no regular metrical type has three weak positions, the usual interpretation is that the first position is unmeasured. The underlying pattern is said to be type A - with anacrusis (x || / x / x). 27 The problem with this classification is that it obscures the distinction between metrical contours with strong onsets (III) and those with weak or proclitic onsets (I or II). The affinity of these verses is with type C. Compare durhfon ne mihte with a type C verse like purhteon mihte (1140b). The negated form of a verse like 1140b might well have been scarcely distinguished metrically from its affirmative counterpart. The particle could be represented as n in a text edited to suggest the rhythms of performance. I mark these as expanded type C (C*) to indicate what I take to be the essential relationship. The proclitic ne sometimes contracts with the verbs agan, beon (in forms beginning with a vowel or W), habban, willan and witan. These contracted forms never alliterate and are found exclusively either in the initial dip or, 25

26 27

See also 2006a, where the ne is editorially supplied.

450b, 595b, 772b, 1460b, 1504b, 1674b, 1733b, 1739b, 1746b, 1773b, 1877b, 2247b, 2332b, 2448b, 2467b, 2574b, 2682b, 2854b, 3058b and 3064b. So Bliss, Pope.

126

Prepositions and the proclitic adverbs

as a result of internal displacement, in the second lift. When the contracted form appears in the initial dip (35 X), 2 8 the uncontracted form with proclitic ne would appear to function equally well, as in Ntes f?set )>onne maetost

maegenfultuma

(1455)

In some instances the Beowulf-poet may have used the option of the contracted form to avoid unwanted incidental alliteration. This could be the explanation for the contracted form in Ic wat geare, f>aet nceron ealdgewyrht, J?aet he ana scyle Geata duguSe gnorn (?rowian, gesigan aet saecce . . .

(2656b-59a)

The uncontracted ne wceron would have echoed the W alliteration of wat and -gewyrht. Similar instances could be adduced. But of course we have no way of knowing whether or why such alliteration might have displeased him here, when he freely admitted it elsewhere. When the contracted form is displaced to the second lift (10X), 2 9 the uncontracted form would again function equally well in most cases. But twice, the use of the contracted form avoids what would otherwise be an expanded type C (C*): t?aet hie ne moste,

J?a Metod nolde

(706; also 967b )

In addition to its use as a proclitic intensifier, the adverb swa functions as a proclitic in the compound swa peah\ A

N

for5 under fexe. Naes he forht swa 5eh (2967; similarly 972b, 1929b, 2442b and 2878b) It also functions as a proclitic in the correlative constructions (efne) swa . . . (swa): (Ax)

(Ax)

sefa swa searogrim,

28

29

swa \>\x self talast (594; similarly 3168a)

Naron: 2657a; nces: 1455a, 2432a, 3074a, 3126a; 134b, 1299b, 1455a, 1463b, 1575b, 1921b, 1929b, 2141b, 2180b, 2192b, 2262b, 2415b, 2 4 9 3 b , 2 5 5 5 b , 2 5 9 1 b , 2 6 8 7 b , 2733b, 2771b, 2845b, 2967b and 2975b; nah: 2252b; nat: 681a; nelle: 2524b; nis: 249b, 1361b, 1372b, 2458b and 2532b; nolde: 2518b. Ncebben: 1850b; noire: 860b and 1167b; nat: 274b; nelle: 679b; nolde: 706b, 803b, 812b, 967b and 1523b.

127

The Metrical Grammar of Beowulf A

efne swa side

swa sae bebugeS (1223; similarly 1092a and 1283a)

(Ax)

on swa hwaej>ere hond halig Dryhten maerSo deme, swa him gemet J?ince

(686—7)

A

efne swa hwylc maeg)?a,

swa Sone magan cende (943; similarly 3057a)

and (single correlative): A

N

licaS leng swa wel,

leofa Beowulf

(1854)

In each of these verses, alliteration follows the proclitic when it is in the w l , either before or after the first lift. There is no alliteration when the proclitic is in the w2. The adverbs nu and pa are proclitic on gen, gena and gyt, forming compounds. The second element of nu gen and nu gyt always occupies the first alliterating lift, as in A

gode forgylde, swa he nu gyt dyde (956; similarly 1058b, 1134b, 2859b and 3167b) But the compounds pa gen, pa gena and pa gyt either occupy the first dip of the verse clause or are displaced to final position. 30 They occupy the first dip in A

gledum forgrunden. Pa gen guScyning (2677; similarly 47a, 1050a and 1866a; 1164b and 2702b) They are displaced to final position in A

N

laSan liges; ne waes hit lenge pa gen (83; similarly 2081a; 536b, 734b, 1256b, 1276b, 2141b, 2237b, 2975b and 3093b) A

hamas ond heaburh.

N

Hengest da gyt

(1127)

Similarly, the adverb//*/ is proclitic on oft. The compound remains in the first dip in the a-verse (480a and 95 la) and takes the alliterating stress in 30

The compound nu da is stressed on the first element (see 426b and 657b, where putting the stress on the second element would create unacceptable double alliteration in the b-verse). For the stress on these compounds, see Campbell, Old English Grammar, § 84.

128

Prepositions and the proclitic adverbs

the b-verse (1252b). The adverb peer is proclitic in the compound peer inne, which always takes the alliterating stress (118a, 2115a and 3087a; 1617b and 2225b). Although normally nalles is a sentence particle (e.g., 338 and 2145), it is proclitic on the sentence particle aene in a

A

oft nalles aene

elland tredan

(3019)

The adverb ower (= ahweer) appears to be proclitic on a following adverb in the correlative construction ower . . . o&&e . . . The sole example in Beowulf (indeed in all of Old English poetry) is found in Wiglaf's reproach to the cowardly retainers after the dragon fight — men who, as he reminds them, had been hugely rewarded by Beowulf in the past: t?onne he on ealubence oft gesealde healsittendum helm ond byrnan, (?eoden his J?egnum, swylce he |?rydlicost ower feor oSSe neah findan meahte . . . 31

(2867—70)

Compare the interlinear gloss of Ps. CXVIII.8 in the Lambeth Psalter (non me derelinquas usquequaque): ne forlcet &u me ahweer eorpan od&e ceghwanan,

where eorpan is parallel to the adverb aeghwanan 'everywhere'.32 If ower is not proclitic, then 2870a violates Kuhn's first law. The instrumentals in comparison

The instrumentals micle ^ py, pe and pon are proclitic on a following comparative adjective or adverb. This is the construction which survives in such Modern English locutions as 'the more the merrier'. The phrase no py cer is weakly stressed and always appears (6X) in the first dip of the clause. 34 The comparative may occupy the first alliterating lift: A

(?aet me is micle leofre, a

symle waes py saemra, 31

J?aet minne lichaman

(2651)

J?onne ic sweorde drep

(2880)

A

' W h e n on t h e ale-bench he often gave to t h e hall-sitters h e l m e t and corslet, t h e chief to his retainers, the m o s t splendid t h a t he could find anywhere far or near

32

' D o not forsake m e anywhere on earth or anywhere at all': from L a m b e t h Palace 4 2 7 (s. xi 1 ), I 4 9 r , ed. Lindelof, Der Lambeth-Psalter,

33

C a m p b e l l , Old English Grammar,

I, 190.

§ 9 4 , n . 1, gives micle as an example of an 'intensifying

adverb'. I prefer to make a distinction between the intensifying adverbs and the instrumentals in comparison. 34 7 5 4 b ) 1 5 O 2 b , 2081a, 2160a, 2373a and 2466a.

129

The Metrical Grammar of Beowulf A

swegle searogimmas,

J?aet ic dy seft maege (2749; similarly 974a; 2423b)

The comparative may occupy the second lift with the instrumental in the w l . When this occurs in the a-verse, it results in double alliteration: A

A

sundes pe saenra,

3e hyne swylt fornam

(1436)

The instrumental falls in the w2 only once in the a-verse. In this case there is double alliteration: A

A

aefre maerSa pon ma

middangeardes

(504)

In short, the proclitic adverbs and instrumental, like the other proclitic forms so far studied, follow the rule proposed for the unstressed prefixes.

130

10 The alliterative requirement of proclitic adjectives and pronouns: the alliterative rule of proclitics

There are three kinds of proclitic adjectives and pronouns - (1) the possessives, (2) the indefinites and (3) the demonstratives and definite articles. The possessives are the possessive pronouns (adjectives) min, pin, sin, uncer (incer does not occur as a possessive in Beowulf), ure, eower and the genitives his, hire, hira of the third person pronoun used possessively. The possessives of the first and second persons must be distinguished from the genitives of the personal pronouns which have the same forms (though never inflectional endings), but which are sentence particles rather than proclitics and are almost always emphatic. 1 Likewise, the possessive genitives of the third person pronouns need to be distinguished from other non-possessive genitives, though it is not always easy to do so. 2 The indefinites include the indefinite adjectives of quantity studied by Desmond Slay 3 : cenig, eall, \fea],fela, monig, ncenig and sum; together with genog, \lytel, Icessa, Icesest], micel, mara, nicest, nan and ofter. The forms in

brackets are fully stressed adjectives. The demonstratives and definite articles are swylc, pyslic and yica\ se, seo, pcet and pes, peos, pis.

The proclitic adjectives and pronouns are usually weakly stressed when they are in proclitic position, as in (with the proclitic italicized) min yldra maeg 1

2

3

unlifigende

(468)

The genitives of the personal pronouns are min (2084b and 2533b), uncer (2002b and 2532a), user (2074b), ure (1386a), incer (584b) and eower (248b and 596b). I take the non-possessive genitives to be: his (2157a) and hyra {heoralhiora) (69 la, 1055b, 1166b, 1636b, 2599b and 2994a). Klaeber would add his (2579a). 'Some Aspects', p. 2.

131

The Metrical Grammar of Beowulf hwanan eowre cyme syndon

to gecySanne,

ealles moncynnes ac he hafaS onfunden,

mine gefraege

(257) (1955)

J?aet he pa faehSe ne J?earf

(595)

The proclitic adjective is found in four of the six so-called 'short' type A 3 , perhaps by chance: 4 (Ax)

Waes min faeder

folcum gecy^ed (262; similarly 459a and 2048a)

(Ax)

J?aer him ncenig waeter

wihte ne sce^ede

(1514)

Sometimes the proclitic adjective alliterates with the head-stave in the b-verse: (Ax)

Wes (?u mundbora

minum mago(?egnum

(1480)

(Ax)

aldor East-Dena,

f>aet he eower ae|?elu can

(392)

A

forjxm \>e he ne u)?e,

(?aet cenig oSer m a n ( 5 0 3 ; similarly 534b) A

heard swyrd hiked,

ond his helm somod

(2987)

Evidence like this leads to the conclusion that the stress on these proclitic adjectives is scf weak that they play no part in the alliterative scheme of the poem. Otherwise the Beowulf-pott would have avoided the offending pattern of double alliteration in the b-verse. Such alliteration must be incidental.

Proclitics are distinguished as a class from fully stressed adjectives by their alliterative behaviour. The fully stressed adjective 'leads' its base word in alliteration. 5 That is, it must alliterate and the following base word may or may not share in the alliteration. The proclitic adjective either does not alliterate at all, or, if it does, it shares the alliteration of its base word. The alliterative possibilities in the a-verse of the two types of adjectives can be illustrated like this:

4

5

See Pope, Rhythm, A 104-7 (Pope classifies 779a as B 51 rather than short A 3 ; hence he counts five short A 3 in all). For this use of'lead' and important discussion, see Slay, 'Some Aspects', p. 1.

132

Proclitic adjectives and pronouns (a) fully stressed adjectives

(1) sole alliteration A

N

leohtan sweorde;

he me lond forgeaf

(2492)

(2) shared alliteration A

A

beorhtre bote

to barww folmum

(158)

(b) proclitic adjectives

(1) absence of alliteration A

ealles moncynnes

mine gefraege

(1955)

modes bliSe

(436)

(2) shared alliteration A

min mondrihten

A weakly stressed proclitic adjective should never display a pattern like (al); a fully stressed adjective will never display a pattern like (bl). Where only patterns like (a2) and (b2) are found, it may be difficult to be sure whether the forms in question are weakly stressed proclitics or not. The following a-verses exhibit what appears to be double alliteration (the alliterating adjectives are again italicized): A

cenig yrfeweard

aefter wurde

(2731)

A

eal ingesteald eorScyninges (1155; similarly 906a, 998a, 2297a and 2885a) A

min mondrihten

modes bli5e

(436)

A

micel morgensweg. A

monig morgenceald

Maere |?eoden (129; similarly 1348a) mundum bewunden

(Ax)

swylcra searoniSa

secgan hyrde

(3022)

(582)

With the exception of ealne utanweardne (2297a), these verses scan equally well with or without a heavy alliterating stress on the adjective. Bliss assigns heavy stress to the adjective in all eleven verses; Pope stresses all. except min. But is this justified? If any of them were clause-initial, it would be persuasive evidence that the adjective was stressed because, in accordance with Kuhn's second law, a weakly stressed proclitic should not appear in the first dip of a verse clause. But none of them is. 133

The Metrical Grammar of Beowulf We know that eall and min appear in (b 1) patterns, and that cenig and min occur with incidental alliteration in the b-verse. Line 2297a is problematic no matter how ealne is stressed. Pope puts it with his unclassified remainders;6 Bliss suggests that it is an isolated hypermetric verse. 7 Reducing ealne to incidental alliteration neither helps nor hurts. There seems no compelling reason to stress the adjectives cenig, eall and min. I mark them all as exhibiting incidental alliteration. With micel and monig, however, the case is different. Neither of these ever appears in (bl) patterns. That is, in proclitic position they always alliterate. Nor do they ever appear in (al) patterns. They never lead their base words as a fully stressed adjective might. How are we to interpret these facts? The evidence suggests that the-proclitic adjectives were not all stressed alike in the metrical grammar of Beowulf. There are gradations of stress which group them in discrete stages along a continuum ranging from the definite article, which rarely co-alliterates and which, when it does, does so almost always in a way which is so clearly incidental as to go practically unnoticed (e.g., hwcepre him sio swi&re (2098a) etc.), to forms like eall and min which frequently co-alliterate, and yet are found in (bl) patterns, to forms like micel and monig which always co-alliterate, but still never lead a non-alliterating base word in the a-verse. The latter group seems capable of crossing the line which separates the proclitics from fully stressed adjectives. 8 I believe that this has happened in the case of at least two adjectives -fea and lytel, Icessa, lasest, which logically belong with the adjectives of indefinite quantity. The adjective fea appears five times in Beowulf, always with primary alliteration. In the formulayfez worda cwczfS, it leads a stressed element (worda, a dependent genitive) which is not in an X-position: faettan goldes,

fea worda cwaeS

(2246; also 2662b)

In the whole corpus of Old English poetry,/&* is never found in unstressed position, and only twice does it occur without alliteration. 9 Evidently, it had become a stressed element and should be so marked. It is almost certain 6 7 8

9

Rhythm, pp. 525-4 and 'Preface', rev. ed. (1966), p. xxxi. Metre, § 106. Micel twice appears as a fully stressed adjective in the b-verse (270b and 502b). See above, ch. 5. Christ 1170a; Psalms 108.8 lb.

134

Proclitic adjectives and pronouns

that lytel is fully stressed in Beowulf. However lytle hwile (2097a) is perhaps not decisive, despite its (al) pattern, because of the adverbial function of the phrase, nor is pcet he lytel fac (2240a), because fcec occupies a final X-position. But in other Old English poetry there is no question of its status as a fully stressed adjective, e.g., lytle ceastre {Genesis 2520a), lytlum sticcum {Andreas 1488a). I can see no definitive way of deciding whether or not forms like micel and monig had become fully stressed in the metrical grammar of Beowulf. Elsewhere in Old English poetry micel leads its noun in miclum spedum {Genesis 121a); it is weakly stressed in mid hu micle elne {Christ 1317a). Otherwise it either displays double alliteration or is in enclitic position. Monig leads its noun in monige hwile {Genesis 634a, but see comment on lytle hwile above) and, less indicatively, in The Metres ofBoethius 10, 26a and 28, 20a; The Paris Psalter 143.8, 3a. It is found once in a (bl) pattern, in Da aras mcenig goldhladen degn {The Battle ofFinnsburh 13a). Otherwise it too either displays double alliteration or is in enclitic position. 10 Both forms were apparently in the process of becoming fully stressed adjectives. What we see is a combination of poetic conservatism and hesitation in the presence of this change. I mark them as exhibiting incidental alliteration in these three verses {Beowulf 129a, 1348a and 3022a) without prejudice to the possibility that the Beowulf-poet gave them full stress. n Swylc is usually a sentence particle. Swylcra in 582a is the only certain instance in Beowulf of swylc functioning as an adjective in proclitic position. Elsewhere in Old English poetry it does appear as a proclitic with a (bl) pattern. 12 I mark it as a weakly stressed proclitic here. If we set the anomalous 2297'a to one side, the assumption that the proclitic adjective is weakly stressed simplifies the scansion of these verses. They are all either type B (129a, 998a, 1155a, 2731a, 2885a and 3022a), type C 1 (436a, 582a and 906a) or type C 2 (1348a), with incidental alliteration on the proclitic. The alternative - putting metrical stress on the proclitic — increases the number of metrical types to four, including the relatively uncommon expanded types D* 1 and D* 4 : D 1 (436a); D 4 (129a, 998a, 1155a, 2885a and 3022a); D* 1 (582a, 906a and 1348a); D* 4 (2731a). 10 11 12

See the remarks of Fourquet, Ordre des elements, p. 171. Seechs. 5 and 11. E.g., Genesis 283a and 287a, Soul and Body I 102a, Guthlac 1128b (incidental alliteration)

135

The Metrical Grammar of Beowulf When a proclitic adjective is displaced or detached from its normal position before the base word, or when it is used substantively (that is, the base word is omitted), it becomes a sentence particle. 13 Certain of the proclitic adjectives happen always to be displaced or detached in Beowulf. These are fea,lA fela, genog,15 nan, pyslic16 and ylca. And some of the proclitic adjectives have an emphatic form which always alliterates. 17 One problem must be considered separately: (1)

Nu is se raed gelang eft cet pe anum. Eard git ne const, frecne stowe, Saer Jmfindanmiht sinnigne secg . . , 18

(2) Leofa Biowulf, laest eall tela, swa Su on geoguSfeore geara gecwaede, be tk lifigendum t?aet Su ne alaete dom gedreosan . . . 19

(1376b-79a)

(2663-6a)

The personal pronoun Se is a sentence particle. In these two verses it is found in the w l , yet the verses are not clause-initial. In both cases the pronoun is modified by a stressed element and, though grammatically the pronoun functions as the object of the preposition, it behaves as though it were proclitic on its stressed modifier. By marking the pronoun as a proclitic we eliminate the violation of Kuhn's first law (1377a becomes a type A 3 (Ib) and 2665b a C 1 (II)), and as a proclitic it conforms to the alliterative rule which is formulated below. THE POSSESSIVES

The weakly stressed possessives nearly always (125X) occupy the w l preceding the first lift and therefore are proclitic on the first alliterating syllable. The five which do not are: A

beodgeneatas; 13 15 16 18

19

(Nx)

Beowulf is min nama

(343)

l4 See below, ch. 11. Fea is a stressed element (see ch. 11). A postpositive form (see ch. 8). 17 The compound pyslic is a stressed element. See above, ch. 5. 'Now help is again dependent on you alone. You do not yet know the region, the terrible place, where you can find the sinful creature 'Beloved Beowulf, carry everything out well, as you long ago said in the days of your youth, that you would not let your fame diminish as long as you lived.'

136

Proclitic adjectives and pronouns A

A

lond Brondinga

leof his leodum, A

A

swylce he )?rydlicost

J>eoden his (?egnum, A

A

mon mid his (ma)gum A

(Nx)

meduseld buan swa hie a waeron

earn his nefan,

(521) (2869) (3065) (881)

In each of these the possessives are in the w l following the first lift. Alliteration of the base word in 343b, which is type E2 with resolution of the final stress, is blocked by the basic alliterative rules. Lines 52 la, 2869a and 3065a are regular type A. The base words alliterate. Line 881a is curious. It is metrically deficient as it stands. Our only recourse is to take earn as the worn-down reflex of a presumed disyllabic form which could take some emphasis on the second syllable. 20 Then the line can be scanned as a type E1. If this is correct, then 88 la is the only type El in the a-verse with a proclitic before the second lift which does not display double alliteration. 21 Before we conclude that 881a is a transferred b-verse, however, we should note that it belongs to a formulaic system the only other recorded occurrence of which is also in the a-verse: earn ond nefa.

Ealra waeron

fife

{Riddle 46, 6)

Nouns of relationship constitute a special subset of the phrases made up of paired elements in the stereotyped phraseology of English. 22 'Father and mother', 'brother and sister', 'father and son' etc. We might include them under the principle of contrast. Evidently, earn ond nefa belongs to this subset. The phrase must already have been of great antiquity in the Old English period (witness the linguistic change which has disturbed its metrical pattern). It must also have been in very common use, despite the relative blank in the surviving records, because otherwise we should expect it to have been refashioned by analogy into a regular metrical type. The relationship between maternal uncle and nephew was of particular sig20 21

22

See Wyatt-Chambers, note on line 881. With double alliteration in the a-verse (11X): 477a, 548a, 633a, 667a, 1977a, 2469a, 2487a, 2508a, 2566a, 2748a and 2792a; with single alliteration in the b-verse (44X): 259b, 396b, 557b, 624b, 668b, 695b, 736b, 759b, 815b, 874b, 895b, 987b, 1024b, 1044b, 1118b, 1126b, 1127b, 1410b, 1494b, 1501b, 1516b, 1519b, 1521b, 154lb, 1549b, 1564b, 1720b, 1789b, 1890b, 1983b, 2092b,2094b,2100b,2221b,2249b, 2288b, 2302b, 2554b, 2559b, 2584b, 2680b, 2929b, 3144b and 3146b. See above, ch. 8.

137

The Metrical Grammar of Beowulf nificance in Germanic culture generally, as Tacitus noted. 23 This is reflected, for example, in the story of Charlemagne and Roland, 24 or in the special emphasis which the poet places on the relationship between Byrhtnoth and Wulfmaer in The Battle ofMaldon.23 This narrative demand alone would probably suffice to have kept the earn ond nefa formula alive in the tradition. 26 THE INDEFINITES

The weakly stressed indefinite adjectives invariably (18X) occupy the w l before the first lift. 27 One of them (ealle) is proclitic on a sentence particle which is itself unstressed: A

deore maSmas.

Ealle hie dea5 fornam

(2236)

The other seventeen are proclitic on alliterating syllables. THE DEMONSTRATIVES AND DEFINITE ARTICLES

The weakly stressed demonstratives nearly always (383X) occupy the w l before the first lift. 28 More particularly, it may be observed that a demonstrative is virtually never preceded by a natural stressed element in the same half-line. The exceptions to both these observations are instructive. There are nine of them: 23

Germanta,

24

See Brault's discussion of, and reference t o t h e literature on, the relationship between

ch. 2 0 .

maternal uncle and nephew in t h e Chanson de Roland and in the chansons degeste generally, in his edition of The Song of Roland I , 2 0 4 - 5 and n . 2 7 . 25

See Scragg's note t o line 115 i n his edition of t h e p o e m .

26

I n the Exeter Book riddle q u o t e d above, it should be n o t e d , the 'uncle' and 'nephew' are

27

These are: arnig *5O3b, * 5 3 4 b , 1 3 5 3 b , 1 5 6 0 b and * 2 7 3 1 a ; eall 4 8 6 a ,

Lot and either of his sons by either of his d a u g h t e r s . #

906a, *998a,

* 1155a, 1222a, 1955a, 2 2 3 6 b , * 2 2 9 7 a a n d * 2 8 8 5 a ; monig * 3 0 2 2 a ; nanig 1514a; micel #

1 2 9 a a n d * 1348a. Those m a r k e d w i t h a n asterisk display incidental alliteration.

Oder does n o t appear as a weakly stressed proclitic i n Beowulf (ior 1 3 5 1 b , see above, ch. 6). 28

I include here w h a t I consider t o b e type A 3 verses w i t h extra-metrical alliteration o n a sentence particle i n the first d i p , such as a

A

Bruc Sisses beages, a

Beowulf leofa

(1216)

A

lixte se leoma

138

ofer landa fela

(311)

Proclitic adjectives and pronouns A

N

Gemunde )?a se goda,

maeg Higelaces A

weard winegeomor, A

wende/><#J ylcan

(2239)

A

Dam wife pa word (Ax)

(758)

N

wel licodon

(639)

mancynne fram

(110)

A

Metod for py mane

(Ax)

sumne besyrwan

in sele pam hean (713; similarly 919b, 1016b and 1984b) A

in bselstede

A

N

beorh pone hean

(3097)

The evidence is overwhelming that a verb like gemunde should never lead a nominal like goda in the a-verse: 758a is in my view almost certainly corrupt (as it stands it has three weak metrical positions, which makes it metrically suspect as well). 29 It is difficult to say whether 2239b is abnormal in any respect. A verb inserted into the initial stressed position provides the alliteration; the demonstrative pees is proclitic on another proclitic ylcan which is used absolutely and thus behaves like a sentence particle. It is in any case unique. Line 639a combines two demonstrative + noun phrases. It may be a matter of chance that this combination occurs only once in Beowulf. But 639a also violates Kuhn's second law which predicts that a proclitic will not stand alone in the initial dip of a clause. Beowulf has just endured Unferth's taunts; he takes a cup from Wealhtheow and delivers his heroic boast that he will purge Heorot of Grendel or die in the attempt. The poet comments: Dam wife f>a word gilpewide Geates; freolicu folccwen

wel licodon, eode goldhroden to hire frean sittan. 30

(639-41)

(The incidental alliteration linking Dam and/><2 should not be overlooked.) Is there an undercurrent of tension here? Others not so well pleased; other words not so appreciated? Although the line may be corrupt, it gives a kind of abrupt nervous intensity to the scene. I prefer to think that this is 29

See above, ch. 6. Klaeber, p . 1 5 5 , defends t h e 'exceptional alliteration' in this verse as 'permissible, especially in view of the syntactical pause assumed here (comma after goda)'.

30

'To that w o m a n those words were well pleasing, the boasting words of the Geat; t h e noble queen of the people, gold-adorned, w e n t to sit by her lord.'

139

The Metrical Grammar of Beowulf another place where the Beowulf-pott deliberately overrode the rules of his metrical grammar for expressive effect. Although the combination of noun + prepositional phrase is very common, 31 Metodforpy mane (110a) is the only half-line in the poem where the prepositional phrase includes a demonstrative. The verse would scan perfectly well without py. To be sure, the reference is to that crime - Cain's slaughter of Abel which was mentioned two lines before. But no demonstrative would have been needed to make clear to a medieval audience what Cain's crime particularly consisted of. The instrumental py seems to me otiose, a prosaic construction which has crept into the poem. The other five exceptions are perfectly legitimate. They involve a special construction — the adjectival epithet of the kind familiar to us in 'Alfred the Great'. I will return to these in a moment. The alliterative requirement of the demonstratives is clearly very strong. Since they are never preceded by a natural stressed element (the two principal exceptions in the poem are 110a and 639a; the adjectival epithets in 713b, 919b, 1016b, 1984b and 3097b are not really exceptions, as will be shown below), this necessarily means that alliteration of the following stressed element is blocked only when it occupies, in exceptional circumstances, an X-position in the b-verse. 32 And when alliteration is not thus blocked it always occurs (the one exception in the poem being 758a). But this does not exhaust the force of the alliterative requirement of the demonstratives. Unlike the other proclitics, the demonstratives place an alliterative requirement on the underlying base word even when the latter has been intercepted. The most common type of interception is when an adjective intervenes between the demonstrative and its noun. The demonstrative is of course proclitic on the adjective + noun phrase, but its underlying base word may be said to be the noun. The additional alliterative requirement can be put this way: The weakly stressed demonstrative imposes mandatory alliteration on its underlying base word, except where alliteration is blocked because the base word occupies an X-position in the b-verse.

One practical consequence of this requirement is a distinction between ordinary adjective + noun phrases and adjective + noun phrases preceded by a demonstrative. The former frequently do not exhibit double alliter-. ation in the a-verse, for example: 31 32

See above, ch. 9. In t h e six: verses cited above: 7 1 3 b , 9 1 9 b , 1 0 1 6 b , 1984b and 3 0 9 7 b (the adjectival epithets) and 2 2 3 9 b , where a stressed verb takes t h e first lift.

140

Proclitic adjectives and pronouns lange J>rage;

he him Saes lean forgeald A

N

nefne him witig God

(114)

wyrd forstode

(1056)

Grendel haten

(102)

The latter always do: A

A

waes se grimma gaest A

A

Heorot faelsian

pes hearda heap, A

A

Waes pat beorhte bold

tobrocen swiSe

(997)

Brosinga mene

(1199)

(Ax)

A

to pare55 byrhtan by rig A

A

hafelan werede

ac se hwita helm A

A

(hyr)stedgolde

Sceal se hearda helm

(432)

(1448) (2255)

The alliterative requirement operates across the medial caesura which separates the a-verse from the b-verse: A

A

ac hine se modega

maeg Hygelaces

(813)

(Ax)

A

sona me se maera

mago Healfdenes

(2011)

(Ax).

A

(?aet se maera

maga EcgSeowes

(2587)

(Ax)

A

Ne meahte se snella

sunu Wonredes

(2971)

(Ax)

A

Hum se snotra

sunu Wihstanes

(3120)

The significance of this point may not be obvious. Of course, the basic alliterative rules require alliteration on the first lift in the b-verse. What is noteworthy is that the adjective never occupies the second lift in the a-verse, because the alliterative requirement of demonstratives confines the demonstrative to the wl before the first lift. Thus only type A 3 and type C verses can precede the medial caesura in these cases. Only when the underlying base word occupies an X-position of the b-verse is alliteration blocked: oflet lifdagas

A A

geond sael swingeS ,

N

ne se swifta mearh A

wigend weccean, 33

N

ond pas laenan gesceaft N

ac se wonna hrefn

Manuscript: here.

141

(1622) (2264) (3024)

The Metrical Grammar of Beowulf Another consequence is a distinction between the demonstrative + adjective + noun phrases with double alliteration described above and demonstrative + noun + adjective phrases. When the adjective follows the underlying base word no alliterative requirement is put on it. Thus we find: (Ax)

N

ac se maga geonga

under his mseges scyld

(2675)

secggende waes

(3028)

(N)

A

Swa je secg hwata

This brings us to the construction with an epithet following the noun. In these five verses the demonstrative does not occupy the w l before the first lift. But if the epithet (demonstrative + adjective) is considered to be displaced from the more usual word order for adjectives, the demonstrative would be in the w l before the first lift before displacement (the epithet is italicized): #

in pam hean sele (cf.: 713b, 919b, 1016b and 1984b)

*pone hean beorh

(cf.: 3097b)

The result of this displacement is an honorific epithet. Heorot is not a high hall like other high halls, but rather it is distinguished from other halls by its height - it is known as 'Heorot the High'. The construction seems to be the same as the one which appears in such phrases as 'Alfred the Great', 'Pompey the Great', 'Frederick the Great', etc., even though the latter are modelled on Greek and Latin constructions and have parallels in the other European vernaculars. 34 The alliterative requirement of demonstratives alters the alliterative pattern of adjective + noun phrases. Without a demonstrative, these phrases alliterate on the first element - the adjective if it precedes the base word. With a demonstrative, the base word alliterates unless it is in the X-position of the b-verse. Therefore, when the base word is not in the X-position of the b-verse, the phrase must display double alliteration or the adjective will have to be displaced. If this explanation is correct, then the alliterative requirement, as modified, correctly describes every line in the poem with a proclitic demonstrative except 758a, which is certainly 34

OED, s.v., great (sense 12d). The earliest citation of great in this usage is from Wyclifs translation of the Bible (1382).

142

Proclitic adjectives and pronouns

corrupt, and 110a and 639a where, abnormally, the demonstrative is outside of the w l before the first lift. Furthermore, it applies to an analogous construction: the demonstrative + dependent genitive + nominal. Even though the demonstrative modifies the genitive, it apparently is felt as governing the phrase, as being the equivalent of demonstrative + attributive adjective + noun. The final noun alliterates in the a-verse: A

A

Ymb pees helmes hrof heafodbeorge (1030; similarly 1057a, 23l6aand 2348a) The final noun alliterates in the b-verse when the construction is interrupted by the medial caesura: A

A

J?aet he pees gewinnes weorc J?rowade (1721; similarly 562, 2391, 2857 and 3109) But, when the construction is wholly in the b-verse, alliteration on the final noun is blocked: (Ax)

N

rondas regnhearde wiS pees recedes weal (326; similarly 1809b, 2405b, 2546b and 2759b) This construction twice occurs with the demonstrative modifying the whole phrase rather than the dependent genitive. Both are in the a-verse and neither displays double alliteration, although the preceding analysis leads us to expect it: A

N

j>get pcet Seodnes beam A

ge)?eon scolde

(910)

N

J?aet se faemnan )?egn

fore faeder daedum

(2059)

Adriaan Barnouw argued on other grounds that the second pcet should be deleted in 910a, because 'no article stands elsewhere in Beowulfbefore a noun which has the genitive of another noun before it'. 35 Furthermore, he denied that 2059a constituted an exception to his rule, because he considered fcemnan pegn to be a compound. 36 Certainty in these matters is 35

' I m Beowulf s t e h t aber sonst kein art. vor e i n e m s u b s t . , das d e n g e n i t i v eines a n d e r n substantivs neben sich h a t , welcher offenbar schon g e n i i g e n d d e m o n s t r a t i v war; wahrscheinlich ist die conjunction pcet aus versehen d o p p e l t g e s c h r i e b e n . ' B a r n o u w , Textkritische Untersuchungen, p. 2 2 .

36

Ibid., p p . 22—3.

Klaeber, p . 166, remarks t h a t Seodnes beam m i g h t likewise have been

'felt to be a c o m p o u n d ' .

143

The Metrical Grammar of Beowulf not to be had, but the failure of the second noun to alliterate in these two verses lends additional support to the supposition that they are not quite right as they stand. In sum, the proclitic adjectives and pronouns conform entirely to the rule proposed in ch. 8 for the unstressed prefixes. This can now be restated as the alliterative rule of proclitics. The rule supplements the basic alliterative rules of the Beowulf-poet's metrical grammar. The basic rules require alliteration of the first lift in both the a- and b-verse. They cover the case of a first lift which is preceded by a proclitic. No further alliteration is permitted in the b-verse. The alliterative rule of proclitics determines the alliterative status of the second lift in the a-verse when the second lift has a proclitic particle before it. The rule can be stated in terms of the number of the weak metrical position in which the proclitic is located. When a proclitic in the wl of the a-verse directly precedes its base word, its base word must alliterate. Alliteration is optional when the proclitic occupies a w2. Put in terms of the

Sievers's types, the rule predicts that, when a base word directly preceded by a proclitic occupies the second lift in the a-verse, double alliteration will be found in types A and E, and single or double alliteration in type B. As we have seen above, the alliterative requirement of the proclitic demonstratives is still stronger, and double alliteration is mandatory even in type B. Just as all half-lines with double alliteration can be labelled as a-verses by definition, so all type A and type E half-lines of the kind described above with single alliteration can be labelled as b-verses. One class of exceptions to the alliterative rule has been pointed out. It involves two indefinites the pronoun gehwylc and the adjective genog. These do not alliterate when the prefix ge- is in the w l before the second lift. 37 Both of these indefinites have been internally displaced from their normal, weakly stressed positions. 38 Evidently, displacement is subsequent to the operation of the alliterative rule, or, to put it in terms of the historical development of the metrical grammar which the Beowulf-pott inherited, these verses may have been formed more recently than the great bulk of the verse formulae which embody the requirements of the rule. 39 They can be labelled as common to the a- and b-verse. 37

10X in t h e a-verse: 9 8 a , 8 0 5 a , 996a, 1090a, 1396a, 2 0 9 4 a , 2489a, 2 6 0 8 a , 2859a and

2891a. 38

39

T h e indefinite p r o n o u n gehwylc (which never alliterates in Beowulf) is a sentence particle. It is weakly stressed as long as it remains in t h e first d i p of the clause. Compare the adjustment to t h e alliterative requirement of demonstratives necessitated by t h e displaced epithets discussed above.

144

Proclitic adjectives and pronouns

Twenty-eight other exceptions to the rule have been noted. In eighteen the proclitic is a prefix - always vocalic, almost always (15 X) ge-. Fourteen of the eighteen prefixed forms are verbs or verbal adjectives. These, like the indefinites gehivylc and genog, can be regarded as displaced from their normal, weakly stressed positions. However, since the great majority of similar verses conform to the rule — those with single alliteration going into the b-verse; a-verses displaying double alliteration — these are better marked as b-verses which the poet has allowed for one reason or another into the first half-line. Two of the four prefixed nominals may be expressive violations (665a and 1375a); the other two (256a and 1857a) seem clearly to be misplaced b-verses. Of the eight exceptions where the proclitic is a conjunction, six offer an alternative - contrast or rhyme - to the expected alliteration. Of the two exceptions where the proclitic is a preposition, one is an editorial conjecture, the other (665a) has already been mentioned as one of the prefixed nominals. 40 The sole exception among the possessives (881a) is a variant of the alternative mentioned above - contrast. Three other apparent exceptions should be noted. They involve the prefix un-\ (Ax)

N

dugu5 unlytel

Dena ond Wedera (498; similarly 833a and 3012a)

I have discussed elsewhere my reasons for thinking that the prefix un- is not heavily stressed in these three verses. 41 But this represents a shift from the normal stress pattern. Neither un- nor any similar form which may lose stress as a result of such a shift belongs to the class of weakly stressed proclitics. 40

See above, ch. 8.

41

See Kendall, 'Prefix un-\ pp. 39- 52.

145

11 Displaced and detached proclitics

Unstressed proclitics are bound to an immediately following base word or phrase. Unbound proclitics, that is, proclitics which have been separated from their regular proclitic positions, may become stressed. They are found in two types of structures. One is illustrated in the narrator's comment that God punished Cain for the murder of Abel (the proclitic is italicized): (a) Metod for \>y mane

ac he hine feor forwraec, mancynne/nz/w.1

(109b-10)

The normal order would of course befram mancynne. The other appears in Beowulf's pledge to Wealhtheow: (b)

Ic gefremman sceal eorlic ellen, o)?Se endedaeg minne gebidan!2 on (?isse meoduhealle

(636b-38)

In this case the normal position for the proclitic would be before endedceg. In (a) the proclitic has been moved to the right of the base word within the same half-line. This is an example of internal displacement. The displaced proclitic, like the displaced sentence particle, becomes a stressed element. In (b) the proclitic has been detached from its base word. A detached proclitic behaves like a sentence particle and may be regarded as such. Prepositions, possessives, indefinites and demonstratives can all be displaced. The data that are presented in the next several paragraphs on dis1 2

'But he drove him far for that crime, the Ruler, from mankind.' 'I shall perform a deed of noble courage, or experience myfinalday in this meadhall!'

146

Displaced and detached proclitics

placed proclitics may be better comprehended by reference to the following text figure: prep. 9

a-verse b-verse

27

poss.

ind./dem.

11 27

44 7 0 0

20

displacement around pronouns into:

1st lift 2nd lift half-lift

17 0

0 0 0

displacement around stressed elements into:

1st lift 2nd lift half-lift

0 14 0

0

0

35 3

43 14

5

Fig. 1 Displacement of prepositions, possessives and indefinites and demonstratives

THE DISPLACED PROCLITICS

There are thirty-six displaced prepositions in Beowulf, nine in the a-verse; twenty-seven in the b-verse. 3 When the object of the preposition is a pronoun, the displaced preposition usually (17 X) occupies the first lift, as in A

J>aet hire an daeges

eagum starede

(1935)

Exceptionally (5X), displacement moves it to the second lift: (Ax)

frecne daede, A

bote gelyfde A

weana gehwylces, feSewiges,

(909) N

swa ic J?e wene to (Ax)

(1396)

N

]>e him foran ongean

(2364)

N

J?e us seceaS to 3

(889)

N

se J?e him bealwa to

A

N

ne waes him Fitela mid

Sweona leoda

(3001)

A-verse: 666a, 671a, 909a, 1626a, 1935a, 2903a, 3001a, 3047a and 3114a; b-verse: 19b, 41b, 110b, 313b, 564b, 681b, 689b, 889b, 1396b, 1542b, 1625b, 1654b, 1715b, 1893b, 1924b, 2242b, 2290b, 2357b, 2364b,241lb, 2523b,2547b,2796b, 2816b, 2831b, 2853b and 2866b.

147

The Metrical Grammar of Beowulf Displacement around a stressed object invariably (14X) puts the preposition in the second lift, as in A

freawine folca

N

Freslondum on

(2357)

Displaced prepositions which themselves are formed with a proclitic element (on efn, ongean, togeanes) obey the alliterative rule of proclitics. That is, they alliterate unless they occupy an X-position of the b-verse. An example in the a-verse is: A

A

Grendle togeanes,

swa guman gefrungon (666; similarly 3114a)

But alliteration is blocked when the displaced preposition is in an X-position in the b-verse: (Ax)

feSewiges,

N

J?e him foran ongean

(2364)

It should be observed that certain prepositions — especially prepositions of more than one syllable — are usually or always displaced, at least in poetry (so: neah, ongean, togeanes).

There are thirty-eight displaced possessives in Beowulf: eleven in the a-verse; twenty-seven in the b-verse. 4 Displacement invariably moves these possessives to the right of a stressed base word. For the most part, therefore, they occupy the second lift in type A, B, C, D 1 or E verses, as in A

N

eSel sinne; —

(1960a: type A)

J?onon Eom^r woe (Ax)

Herebeald ond HaeScyn

N

oSSe Hygelac min (2434b: type B) (Ax) N

ond him HroJ?gar gewat

to hofe sinum (1236b: type C1; also 1507b) A

feorran gefricgean

fleam (Ax)

Eow het secgan

N

eowerne

(2889b: type D 1 )

N

sigedrihten min

(391b: type E)

However, in three cases the displaced possessive moves around an initial resolved disyllabic The result is that it occupies the first (and weaker) of 4

A-verse: 346a, 365a, 1960a, 2095a, 2160a, 2251a, 2283a, 2722a, 2729a, 2789a and 3093a; b-verse: 178b, 267b, 294b, 391b, 415b, 457b, 473b, 530b, 550b, 592b, 698b, 1169b, 1226b, 1236b, 1246b, 1336b, 1345b, 1507b, 1704b, 1708b, 1776b, 1848b, 2434b, 2537b, 2889b, 3002b and 3107b.

148

Displaced and detached proclitics

two metrically stressed positions after the alliterating lift syllable (type A 2a ) 5 : (Ax)

geond widwegas,

N

N

wine min Beowulf (1704; also 457b, and similarly 530b)

There is one instance of double alliteration (optional) in the a-verse: (Ax)

A

no Sy aer suna sinum

syllan wolde

(2160)

There are sixty-four displaced indefinites and demonstratives in Beowulf: twenty in the a-verse; forty-four in the b-verse. 6 Where the proclitic modifies a stressed base word, displacement usually (43 X) moves it to the second lift, for example: A

(N)

Paet waes wraec micel

wine Scyldinga A

j?ses J?e (?incean maeg

(170)

(Nx)

)?egne monegum (1341; also 1419b)

However, when its base word is stressed, the indefinite eal is invariably (13X, always in the b-verse) displaced to the first (and weaker) of two metrically stressed positions after the alliterating lift in a type E, 7 as in (Ax)

wan under wolcnum.

N

N

Werod eall aras

(651; also 3030b)

The indefinite oper is found in the same construction: AN

worn gemunde, 5

6

7

N

word oper fand

(870)

This is evidence that type A2a should be differentiated from type A1. I treat the resulting phrase as a class I compound (see ch. 12). A-verse: (anig) 2772a; (eal) 94 la and 1608a; (fela) 530a; (genog) 2489a; (mara) 136a, 753a and 2016a; (micel) 67a, 69a, 170a, 771a, 1167a and 1778a; (monig) 776a, 854a, 1015a and 3077a; (se) 2588a and 2959a; b-verse: (anig) 1851b and 3080b; (eal) 523b, 651b, 767b, 823b, 1185b, 1417b, 1424b, 1567b, 1790b, 2005b, 2017b, 2080b, 2427b, 2691b, 2739b, 3030b, 3087b and 3094b; (fela) 1783b; {mast) 459b; (micel) 146b, 922b and 958b; (monig) 399b, 838b, 908b, 918b, 1112b, 1289b, 1341b, 1419b, 1510b and 2762b; (nanig) 859b; (oder) 870b, 1353b and 1560b; (se) 2007b, 2334b, 2969b and 3081b; (sum) 271b. That is, I interpret the phrase werod eall as a class I compound (see ch. 12). When a compound governs a verb, the first element of the compound takes the alliterating stress and the vert occupies the second lift. Thus the stress on eall is necessarily less than that on

149

The Metrical Grammar of Beowulf Displacement around an unstressed pronoun puts an indefinite in the first lift seven times: A

tie we eatle uncran eaferan,

aer ne meahton

(941)

A

gif he J?aet eat gemon (1185; similarly 2005b, 2427b and 3087b)

A

J?aet hit eat gemealt

ise gelicost

(1608)

A

aSa on unriht.

Ic 5aes ealles maeg

(2739)

Ten of the twenty a-verse displacements of indefinites and demonstratives result in double alliteration. Eight of these displacements move the indefinite around a fully stressed compound, 8 as in (A)

A

medubenc monig

mine gefraege

(776)

Another such displacement occurs around a base word with a stressed prefix: A

A

onsyn anig,

ac hyne ecg fornam

(2772)

The final case of double alliteration involves a hypermetric verse: A

(A)

\>2£t he haefde mod micel,

peah \>e he his magum naere

(1167)

THE DETACHED PROCLITICS

A proclitic which does not have a base word in the same half-line is 'detached'. I include in this category proclitics which are used absolutely as nouns or pronouns. Detached proclitics are sentence particles. Like other sentence particles their behaviour depends on their location within the verse clause. A detached proclitic which appears in the first dip of the clause is unstressed unless affected by the transformational rule; when externally displaced it will alliterate; when internally displaced it will take metrical stress.

8

See below, ch. 12. Twice, grundwong pone (2588a) and freoSowong pone (2959a), displacement around a fully stressed compound in the a-verse does not result in double alliteration. In both cases the demonstrative occupies a final X-position.

150

Displaced and detached proclitics A-verse

It is best to separate the discussion of the a-verse from the b-verse here, because of the special characteristics of each half-line. In the a-verse, the detached proclitic may exhibit extra-metrical alliteration which creates the possibility of external displacement. When the detached proclitic (italicized) appears with one or more other sentence particles, with or without alliteration, in the first weak metrical position (wl) before a stressed element, the half-line is unambiguously clause-initial, non-displaceable (la). 9 There are five such half-lines without alliteration: (Ax)

Fela ic on giogoSe

guSraesa genaes

(2426)

(Ax)

ne J?aer ncenig witena wenan J?orfte (157; similarly 691a, 1197a and 1933a) Fela appears once with extra-metrical alliteration: 1 0 (Ax)

a

Swa fela fyrena

feond mancynnes

(164)

If the detached proclitic is the only sentence particle in the w 1 and it alliterates, then the half-line is clause-initial, displaceable (Ib). The verses in question are: A

Eal \>u hit gej>yldum healdest (1705: hypermetric)

din ofer |?eoda gehwylce. A

irenna cyst

cenig ofer eorj?an A

(802; also 2007a)

Paet waes y l d u m cuj?

ealle buton anum. A

forS onsended

(2266)

geond bysne middangeard

(1771)

fela feorhcynna A

manigum maegba A

maegenbyrSenne

micle mid mundum A

otter to yldum. 9

10

(705)

Pa waes eft hraSe

(3091) (2117).

So also would be a half-line with a non-alliterating detached proclitic and no other sentence particles. There is none in Beowulf. Bliss and Pope assign full stress to fela in 164a.

151

The Metrical Grammar of Beowulf All of these half-lines are, in fact, displaced. Micel and monig may be fully stressed adjectives, 11 in which case 1771a and 3091a would be type A rather than type A 3 verses. Micel is also detached in A

A

micelne ond maerne,

swa he manna waes

(3098)

However, since the copulative conjunction gives equal metrical status to both elements within a half-line, 12 micelne can be taken as a stressed element with mcerne in a quasi-compound phrase, and the verse is a type A. Detached proclitics sometimes appear with other sentence particles in a-verses which lack stressed elements. Consider the following: (a) J?aet heo on aenigne

eorl gelyfde

hwaej?er him cenig waes Da waeron monige, J?aet ge genoge

(1356) (2982)

neon sceawiaS

(3104)

)?onne aenig man oSer

(1353; similarly 1560a)

ealdre gebohte

(?onne he swylces hwaet he fear a sum

aer acenned

]>e his maeg wriSon

naefne he waes mara |?eah 5e otkr his

(627)

secgan wolde

beforan gengde

13

(2481) (880) (1412)

(b) sumne besyrwan

in sele (?am hean

(713)

(c) manigra sumne\

hyt ne mihte swa

(2091)

All the verses in (a) are clause-initial and the transformational rule correctly assigns metrical stress to the stressed syllable of the (leftmost) detached proclitic in every case. These are, without exception, clauseinitial, non-displaceable (la). As for (b), the transformational rule assigns metrical stress to besyrwan. Since this produces a valid metrical contour (type A 3 ), the alliteration on sumne must be extra-metrical. When there is a single sentence particle with extra-metrical alliteration before the alliterating lift syllable, the half-line is clause-initial, displaceable (Ib). 14 In fact, this is the third half-line in a clause consisting of four verses: 11 13 14

u See above, chs. 5 and 10. See above, ch. 8. Actually,'I believe feara must be a stressed element. See above, ch. 10. Bliss and Pope assign full stress to sumne in 713a.

152

Displaced and detached proclitics mynte se manscaSa sumne besyrwan

manna cynnes in sele |?am hean.15

(712—13)

The half-line in (c) looks as though it might belong to (a). However, if we were to apply the transformational rule, the result would be: #

A

manigra sumne

But in fact it is manigra which alliterates. This means that manigra must be a stressed element and that this is a type A verse which is unrestricted (III). I have pointed out in ch. 10 that monig falls in a grey area between the proclitics and the fully stressed adjectives. Here it is the latter. A detached proclitic which follows a stressed element in the a-verse always (19*) occupies the second lift and is therefore stressed. All happen to be indefinites. 16 The alliterative rule of proclitics generates double alliteration in A

A

ana wiS eallum, A

A

an aefter eallum, A

A

eorl ofer o&rum

o3 J?aet idel stod

(145)

unbliSe hwe(arf)

(2268)

unlifigendum

(2908)

Double alliteration is also found in the hypermetric verse: A

A

aeghwylc odrum try we.

Swylce J?ser Unfer\> Jpyle

(1165)

In 591a the intensive proclitic swa does not generate double alliteration because it is in the w2: 17 A

(Nx)

t?aet naefre Gre{n]del swa. fela

gryra gefremede

(591)

B -verse

In the b-verse, any detached proclitic appearing in the w l before the first stressed element signals a clause-initial (I) half-line. If alliteration occurs, 15 16

17

'The evil-doer intended to entrap a certain one of the race of men in the high hall.' In addition to the verses discussed below these are: 474a, 988a, 1235a, 1312a, 1412a, 1461a, 2171a, 2301a, 2401a, 2440a, 3054a and 3123a. Fakundiny, 'Art of Verse Composition', p. 134, points out that this is the only instance in Old English poetry of 'an independent noun [which] precedes the indefinite adjective and leads it in alliteration'.

153

The Metrical Grammar of Beowulf

it must be incidental rather than extra-metrical. There are fifteen verses of this kind. 18 Examples without alliteration include: A

lungre gelimpe

Fela ic la(?es gebad

(929)

A

Ic (?aet gehyre,

J?aet pis is hold weorod

(290)

In 4 0 0 b sume is followed by a stressed adverb + v e r b : A

J?rySlic )?egna heap;

sume J?ser bidon

(400)

Incidental alliteration is found in Sigon j?a to slsepe.

A

Sum sare angeald

(1251)

A

sibaeSelingas;

swylc sceolde secg wesan

(2708)

As long as it is not a case of incidental alliteration, which can be determined by inspection, an alliterating sentence particle in initial position in the b-verse is a stressed element. Half-lines in which such particles appear behave like any other verses with initial stressed elements. Kuhn's laws apply to them in the normal fashion: that is, a proclitic before the stressed element signals a clause-non-initial (II) half-line; a sentence particle in the first dip after the stressed element signals a clause-initial (I) half-line; the absence of both signals an unrestricted (III) half-line. There are fifty-two detached proclitics in initial stressed position in the b-verse in Beowulf.19 Most of these are followed by a noun or a displaced sentence particle which occupies the second lift. None is preceded by a proclitic. With the exception of 1060b, these are all unrestricted (III) half-lines, for example: A

Nolde eorla hleo

N

cenige f>inga A

J?aet he for mundgripe 18

19

(791)

N

minum scolde

(965)

In addition to those cited, they are: {eat) 1705b (hypermetric) and 2814b; (otkr) 135 lb; (sum) 1113b, 1432b, 2940b and 3124b; (swylc) 178b, 1940b and 254 lb. Preposition: (ymbe) 2597b; possessives: (min) 255b, 293b, 418b, 638b, 965b, 2012b, 2479b, 2750b and 2799b; (pin) 1477b; indefinites and demonstratives: (cenig) 791b, 2374b, 2449b, 2734b and 2905b; (eat) 71b, 111b, 699b, 744b, 830b, 848b, 1086b, 1593b, 1620b, 1796b, 2087b and 3170b; (fela) 153b, # 992b, # 1060b and 2620b;' (mket) 533b, 1823b and 2185b; (monig) *171b, 349b, *857b, 1023b, * 1860b and 2001b; (namig)*598b;(oder) 859b, 1029b, 1228b, 1874b, 1945b, 2167b, *2198band # 245 lb; (iswylc) 1347b and 2798b. The eight half-lines marked with an asterisk stand at the head of their clauses.

154

Displaced and detached proclitics (Ax)

fyrene ond faehSe

N

fela missera

(153)

There are three half-lines in which the initial detached proclitic is followed by two internally displaced sentence particles: 20 (Ax)

folmum gefraetwod;

(Ax) N

modes brecSa.

N

N

fela l?aera waes

(992)

N

Monig oft gesaet (171; similarly 857b)

Not surprisingly, given their syntax, all three begin a clause, although in theory they, like the others, should be unrestricted (III). One verse with an initial detached proclitic has an undisplaced sentence particle (sceal) in the initial dip: (Ax)

N

ferhSes forej>anc. Fela sceal gebidan (1060) This is a clause-initial (I) half-line. Half-lines like these could not (and do not) appear in the a-verse with these stress patterns, because there the transformational rule would put stress on a sentence particle further to the right. Nor does the Beowulf-poet ever use the same half-line with one stress pattern in the a-verse and another in the b-verse. This is what I call the principle of linear integrity. 21 Every b-verse with a detached proclitic bearing alliterating stress which is not in initial position has at least one sentence particle in its weak onset. All such verses are by definition clause-initial (I) half-lines. If another sentence particle follows the detached proclitic it will be a stressed element because it has been displaced from the w l before the first lift. Just as some a-verses which scan by the transformational rule would display the same contour if they were considered as b-verses with alliterating stress on a detached proclitic, some of these half-lines, if considered as a-verses, would display the same metrical contour by the transformational rule. 22 This would be true of (with the detached proclitic italicized): 20

21

22

In 992b para is dependent on fela with which it forms a quasi-compound, 'many-ofthem (singular) was'. This pre-empts the w l position after the first stressed element. The combination of adverb + verb which is found in 17 lb and 857b regularly forms a class II compound (see ch. 13). See Kendall, 'Displacement', pp. 26—7. I should emphasize that this principle applies to the individual poet, not to the tradition.

They are: 247b, 251b, 445b, 450b, 913b, 932b, 949b, 972b, 996b, 1079b, 1133b, 1180b, 1338b, 1349b, 1471b, 1598b, 1613b, 1727b, 1772b, 1815b, 1887b, 2042b, 2061b, 2149b, 2231b, 2461b, 2727b, 2737b and 2742b.

155

The Metrical Grammar of Beowulf A

aescum ond ecgum,

J?aet ic me cenige A

aenlic ansyn.

(1772)

N

Nu ic eower sceal A

secga gehwylcum

(251)

(N)

J?ara |?e on swylc staraS

(996)

These bear a close resemblance to the type B and C verses in group (a) above. Nevertheless, the two kinds of half-lines must be kept distinct, and these are by definition b-verses (there are in fact no formulae which are common to the a-verse and the b-verse and which scan either by the transformational rule or as a result of alliterating stress on a detached proclitic). Three half-lines with a detached proclitic bearing alliterative stress which is not in initial position can be shown to be limited to the b-verse, because, if they were to be shifted to the a-verse, the transformational rule would generate a different metrical contour. These are: A

maegen mid modes snyttrum.

N

Ic pe sceal mine gelaestan (1706: hypermetric) (Ax)

ac hie haefdon gefrunen,

N

J?aet hie aer to fela micles (694; similarly 809b)

As in the a-verse, a detached proclitic frequently (52 X) occupies the second lift after a stressed element. Where the stressed element is natural (that is, not the result of the insertion of a stressed sentence particle into the first lift) and there is nothing that would require double alliteration if the half-line were moved to the a-verse, the half-line may be presumed to be common to the a- and b-verse. Examples are: (Ax)

to gegangenne

N

gumena cenigum (2416; cf. 474a etc.) A

maerne be maeste.

(Nx)

Paer waes madma/^z (36; cf. 988a etc.) A

waepna gewealdan,

(Nx)

ac hine wundra \>desfela (1509; cf. 591a) (Ax) N

[Gejgrette J?a guma operne (652; cf. 2440a) When the detached proclitic is in the second lift after a stressed element under conditions that would normally generate double alliteration in the 156

Displaced and detached proclitics

a-verse, we may take the half-line to be limited to the b-verse. One of these conditions is the alliterative rule of proclitics. Thus in two half-lines the detached proclitic is preceded by a preposition: A

o\> Se nipende

N

niht ofer ealle

A

bealohycgendra

N

broga fram otkum

(649)

(2565)

Half-lines like these are not found in the a-verse. The first stressed element in the b-verse may be a sentence particle inserted into the alliterating position. An initial stressed verb marks the half-line as being limited to the b-verse, as in A

sunu Healfdenes

N

- swigedon ealle A

ac he soSlice

N

saegde ofer ealle A

weard winegeomor,

(1699) (2899)

N

wende l?aes ylcan

(2239)

In 689b the first stressed element is a displaced preposition: (N)

A

eorles eorles andwlitan, andwlitan,

on< ond hine ymb monig

(689)

Other possibilities can be illustrated by: (Nx)

A

niehstan siSe:

Ic geneSde fela A

eorla ofer eor(?an,

(2511) N

Sonne is eower sum

(248)

Finally, there is a group of b-verses in which the detached proclitic is one of two stressed elements following the first lift: AN

laene licgan;

N

lyt cenig mearn

(3129)

breostgewaedu.

Bruc ealles well

(2162)

Ic Sara fraetwa

Frean ealles Sane

(2794)

fremeS on folce,

feor eal gemon

(1701)

laest eall tela

(2663)

Me tod eallum weold

(1057)

Leofa Biowulf, ond 5aes mannes mod. laSbite lices.

Lig ealle forswealg (1122; similarly 1080b)

Seodne <et (?earfe;

Solode derfela

(1525)

faege gefealleS; fehS oper to (1755) These have no a-verse equivalents. It is not immediately apparent which of the two stressed elements following the first lift should be more heavily 157

The Metrical Grammar of Beowulf stressed. Bliss gives priority to the final element in every case. Pope prefers to stress the first of the two elements in 1057b, 1080b, 1122b and 1525b. I stress the first of the two elements in 1057b, 1080b, 1122b, 1701b and 3129b. 23 Line 2794b stands apart from the others in this group because it is the only one with a noun {danc) rather than a sentence particle in one of the two final positions. It is part of Beowulf's dying speech: Ic Sara fraetwa Frean ealles danc, Wuldurcyninge wordum secge, ecum Dryhtne, \>e ic her on starie, \>&s 5e ic moste minum leodum aer swyltdaege swylc gestrynan.

(2794-8)

Its syntax is ambiguous. Ealles is either the genitive object of secge danc ('I give thanks in words to the Lord, King of Glory, eternal Lord, for all of the treasures, which I am looking on here, that I might acquire such for my people before my death-day') or dependent on Frean ('I give thanks in words to the Lord of All for the treasures . . .'). 2 4 I incline to the latter, because the word order is less involved. It would make the verse a type E. 23

For further discussion, see below, ch. 1 3 .

24

For ealles as genitive object, see, e . g . , Beowulf, trans. Crossley-Holland, p . 109; as d e p e n d e n t on Frean, see e . g . , Beowulf, trans. Donaldson, p . 4 9 . For further discussion, see below, c h . ' 12.

158

12 The three classes of compounds: the alliterative requirement of class I compounds

The alliterative patterns of Beowulf"are limited by three sets of conditions the basic rules of alliteration and metre, the placement of sentence particles and proclitics within the half-line and the nature of the simplex or compound which takes metrical stress. A reader who knows these conditions can predict the potential placement in the a- or b-verse of practically any metrical—syntactical pattern, can say of any given half-line of the poem whether its alliterative behaviour is normal or not and can determine whether, if double alliteration is involved, the double alliteration is a reflex of the metrical grammar or a spontaneous option exercised by the poet. One of the principal objects of this study has been to provide this knowledge. There remain to be discussed the alliterative requirements of compounds.' The compounds which are the subject of this and the following chapter are all stressed elements. Notwithstanding this fact, the alliterative behaviour of these compounds is remarkably varied. Some compounds alliterate no matter where they are found and therefore are not found where the alliterative rules prohibit alliteration. Other compounds alliterate whenever they appear in the a-verse, but occur freely in the second lift of the b-verse without alliteration. And still others are found in the second lift of both the a-verse and the b-verse without alliteration. In short, three classes of compounds can be distinguished on the basis of their alliterative behaviour. l It seems virtually certain that the three classes were broadly related to 1

Hoover, 'Evidence for Primacy of Alliteration', pp. 77-8, divides compounds into two classes. His 'Class One' corresponds roughly to my class I. His 'Class Two' includes some forms which I would put into class I, but for the most part corresponds to my classes II and III.

159

The Metrical Grammar of Beowulf different levels of stress in the compounds. Whether these variations in stress corresponded to natural levels of stress in the language of everyday speech or whether they were the product of the special constraints of the metrical grammar, however, is a question which I do not intend to address. Moreover, I do not believe that all the compounds in any one class necessarily had the same level of stress, nor do I believe that stress is the only factor that could have played a part in determining which class a compound belonged to. As Arthur Brodeur so well observed, the special richness of the art of Beowulf is largely owing to the poet's creative use of compound nouns and adjectives. Brodeur's words are worth repeating: The richest and most meaningful content-words in the poetic vocabulary are the substantival and adjectival compounds; they not only express concepts, often very forcefully or imaginatively; they often contain or imply partial description of concepts as well. Moreover, they play a very important part in the rhetorical devices of variation, enumeration, and progression. From these compounds, more than from other parts of speech, we must form our estimate of the language of Beowulf.2

In this and the following chapter I explore the formal, and especially the alliterative, properties of the three classes of compounds in the epic. These properties are part of the metrical grammar of the Beowulf-pott. They are, no doubt, inherited from the poetic tradition, but the poet's heightened sensitivity to them, the way he has shaped what may have been tendencies into an intricate formal system of great consistency can perhaps be regarded as a signature of his style. CLASS I COMPOUNDS

The strongest alliterative requirement in the metrical grammar of the Beowulf-pott is that of the fully stressed compounds and their equivalents. These are the compounds which belong to 'class I'. Fully stressed compounds are ones in which both elements retain their full semantic value. 3 Their alliterative equivalents include compounds with variably 2 3

Brodeur, Art ofBeowulf\ p. 8; see also Niles, Beowulf, pp. 138-51. Campbell, Old English Grammar, § 87. Hoover, 'Evidence for Primacy of Alliteration', pp. 76—7, points out the significance of self-alliterating compounds for interpreting the role of the fully stressed class I compounds in the metrical grammar. Class I self-alliterating compounds actually form a 'super-class'. They are, by definition,

160

Class I compounds

stressed prefixes when the stress is on the prefix, probably compounds with disyllabic fixed stressed prefixes, compounds with certain suffixes, possibly a few compounds in which the two elements are no longer semantically distinct and certain quasi-compound phrases. Wherever a fully stressed compound appears, whether in the a-verse or the b-verse, its first stressed syllable will alliterate. 4 Since the fundamental alliterative principle assigns alliteration in any case to the first stressed element of either half-line, the presence of alliteration in compounds which take the first lift needs no special comment. Representative examples are (with the compounds hyphenated and italicized): frea-wine folca

aet minum faeder genam

mid his freo-dryhtne (?aet hio Beowulf,

fremman sceolde beag-hroden cwen

J>ast hire on hafelan

hring-mcel agol

(2429) (2627) (623) (1521)

There is only one instance in Beowulf of a compound in initial position which lacks proper alliteration: brun-fagne helm,

hringde byrnan

(2615)

It is not easy to say whether the difficulty here is in the a-verse or the b-verse. The first half-line is supported by gold-fahne helm (2811b), which alliterates, as one would expect, on the initial syllable. Rieger's solution, which is attractive in its simplicity, is to reverse the order of the words in the b-verse.5 This would seem to produce a normal line with cross alliteration:

4 5

excluded from the b-verse, as ordinary class I compounds are not. Hoover, ibid., pp. 79-80, lists thirty class I self-alliterating compounds in Beowulf: wig-weorpunga (176a), wi& peod-preaum (178a), after deaS-dage (187a and 885a), in hyra gryre-geatwum (324a), heard-hicgende (394a. and 799a), pone cwealm-cuman (792a), geond wid-wegas (840a and 1704a), in fen-freodo (851a), of bryd-bure (921a), bearn-gebyrdo (946a), Swylce ferhd-frecan (1146a), after sce-si&e (1149a), geosceaft-gasta (1266a), mil-gemearces (1362a), heoro-hocyhtum (1438a), GutS-Geata leod (1538a), fela-fricgende (2106a), hilde-hlammuml-hlemma (2201a, 235 la and 2544a), eall-irenne (2338a), wid dam gryre-gieste (2560a), mid minne gold-gyfan (2652a), in Hrefnes-holt (2935a), Was sio swat-swa&u (2946a) and swylce giomor-gyd (3150a); in eowrum gufi-geatawum (395b). The only b-verse among them (395b: which Klaeber emends to in eowrum gud-getawum - thus eliminating the offending double alliteration) Hoover properly regards as an anomaly. See Kendall, 'Prefix un-\ pp. 49-52. Rieger, 'Die alt- und angelsachsische Verskunst', p. 2 1 .

161

The Metrical Grammar of Beowulf *brun-fagne helm,

byrnan hringde

However it must be observed that hrinde byrnan is a formula. Compare hea|?o-steapa helm,

hringed byrne

(1245)

Whatever the explanation for it may be, there is some textual difficulty with the line. 6 There are two practical consequences of the alliterative requirement of class I compounds: (1) all a-verses with a fully stressed compound in the second lift display double alliteration; and (2) fully stressed compounds are excluded from the second lift in the b-verse. (1) The first consequence follows logically from the fundamental alliterative principle which requires that the first lift syllable alliterate. Therefore, if a fully stressed compound is placed in the second lift, the alliterative requirement of class I compounds will result in double alliteration (no fully stressed compound is placed after the second lift). There are 209 a-verses with a fully stressed compound in the second lift. 7 All 209 display double alliteration, for example: 6

7

Some formulae in the Anglo-Saxon poetic tradition are reversible. A case in point in Beowulf is feond mancynnes (164b) versus mancynnes feond (1276a) (See Fry, 'Old English Formulas', p. 202; I owe this observation to Lawrence Moe). Thus Rieger's solution cannot be dismissed as inherently improbable. 21a, 31a, 33a, 54a, 58a, 65a, 82a, 103a, 127a, 160a, 165a, 193a, 216a, 218a, 223a, 263a, 275a, 277a, 288a, 298a, 305a, 308a, 320a, 322a, 326a, 380a, 392a, 398a, 408a, 427a, 445a, 485a, 487a, 502a, 551a, 554a, 558a, 594a, 596a, 606a, 608a, 609a, 619a, 641a, 690a, 693a, 755a, 769a, 770a, 775a, 780a, 823a, 826a, 868a, 912a, 936a, 938a, 986a, 990a, 1022a, 1023a, 1060a, 1069a, 1087a, 1096a, 1109a, 1112a, 1127a, 1157a, 1162a, 1200a, 1212a, 1231a, 1259a, 1277a, 1298a, 1307a, 1339a, 1359a, 1400a, 1409a, 1410a, 1424a, 1426a, 1430a, 1440a, 1444a, 1447a, 1454a, 1468a, 1489a, 1505a, 1522a, 1527a, 1533a, 1546a, 1559a, 1564a, 1568a, 1575a, 1617a, 164 la, 1646a, 1668a, 1678a, 1698a, 1702a, 1719a, 1729a, 1747a, 1785a, 1790a, 1793a, 1800a, 1806a, 1816a, 1845a, 1847a, 1863a, 1881a, 1890a, 1895a, 1921a, 1927a, 1969a, 2025a, 2037a, 2042a, 2074a, 2079a, 2082a, 2090a, 2112a, 2118a, 2123a, 2173a, 2175a, 2196a, 2198a, 2205a, 2210a, 2226a, 2239a, 2263a, 2271a, 2273a, 2286a, 2289a, 2292a, 2296a, 2315a, 2368a, 2370a, 2396a, 2408a, 24l4a, 2420a, 2442a, 2451a, 2462a, 2476a, 2478a, 2493a, 2496a, 2517a, 2558a, 2563a, 2603a, 2613a, 2625a, 2642a, 2648a, 2660a, 2670a, 2674a, 2682a, 2689a, 2691a, 2704a, 2710a, 2719a, 2725a, 2749a, 2755a, 2760a, 2800a, 2810a, 2811a, 2819a, 2825a, 2827a, 2829a, 2847a, 2863a, 2929a, 2931a, 2950a, 2953a, 2957a, 2996a, 3021a, 3041a, 3055a, 3063a, 3099a, 3111a, 3132a, 3136a and 3155a. I exclude from the category of fully stressed two-element compounds all two-element compounds beginning with af~, cer-, an-, and-, (e)al(t)-, in-, on-, or-, un- or ud-. In the case of a-verse half-lines

162

Class I compounds

l?a waes sund liden

side sa-nassas;

(223)

geatolic ond gold-fah

ongyton mihton

(308)

manna maegen-crceft

on his mundgripe

(380)

on hreon mode

har hilde-rinc

nallas beagas geaf

(1719)

wende J?aes ylcan

(2239)

breosthord blod-reow\ weard wine-geomor,

(1307)

Among these 209 are certain a-verses in which the compound in the second lift follows an undisplaced sentence particle. The compound alliterates according to the same principle, as in stig wisode8

Street waes stan-fah,

(320)

(2) The second consequence - that fully stressed compounds will not be found in the second lift in the b-verse — also follows logically. The fundamental alliterative principle assigns alliteration to the first lift syllable. The other major alliterative rule, the one which permits double alliteration in the a-verse and prohibits it in the b-verse, rules out the possibility of alliteration on the second lift in the b-verse. Therefore, if a compound must alliterate, it cannot occupy the second, non-alliterating lift in the b-verse. There are only two significant exceptions to this prohibition in Beowulf:9 N

A

Swa fela fyrena

feond man-cynnes (Ax)

se \>e on handa baer

(164)

(Nx)

hroden ealo-wczge

(495)

I think it is probable that mancynnes was variably stressed in Old English poetry as its reflex is in Modern English. There is a line in Christ and Satan which violates the alliterative requirement of compounds in the a-verse unless the second element is stressed, in which case the a-verse displays double alliteration:

8

9

consisting of finite verb + compound and the like, I regard the compound as occupying the first metrical lift. Fourquet, Ordre des elements, p. 191, concludes that double alliteration is required in verses of this type (noun + copula + compound adjective) because the attribute is 'heavier' in weight and thus receives obligatory alliteration. I believe these are part of the broader generalization 1 am making. See Kendall, 'Prefix un-\ pp. 49—52, for a fuller discussion of the possible exceptions to the alliterative demand of compounds.

163

The Metrical Grammar of Beowulf cyning moncynnes.

Cer Se on baecling {Christ and Satan 697)

Both Beowulf 164b and Christ and Satan 697a could be scanned as type A. This leaves only hroden ealo-wcege (495 b) which cannot be satisfactorily explained and must stand as an exception to the rule. The alliterative requirement of fully stressed compounds has another dimension. Not only do all such compounds alliterate, but they impose an alliterative requirement on any stressed element which follows them. When a compound is followed by a stressed element, double alliteration is the rule in the a-verse unless the stressed element occupies the final X-position; the use of a natural stressed element in the non-final Xposition is generally avoided in the b-verse. There are 483 fully stressed compounds which occupy the first lift and are followed by another stressed element. Of these 142 are followed by a stressed element which is not in the final X-position. In the a-verse double alliteration is obligatory, for example: (IX with second stressed element not in any X-position) fyrd-searu fuslicu;

hine fyrwyt braec

(232)

(13IX with second stressed element in the non-final X-position) fea-sceaft funden; wis-fast wordum

he J?aes frofre gebad l?aes Se hire se willa gelamp

gomen-wudu greted,

gid oft wrecen

heard-ecg habban; breost-hord blodreow;

(7) (626) (1065)

ic me mid Hruntinge

(1490)

nallas beagas geaf

(1719)

In the b-verse, where double alliteration is prohibited, the second stressed element is found only ten times in the non-final X-position and only once is it a natural stressed element (Hro&gar): 5inra gegncwida,

glad-man HroSgar

(367)

In the remaining nine the second stressed element is a displaced sentence particle: morj?re gemearcod

man-dream

(1264)

- heal-wudu dynede (1317; similarly 1906b)

mid his handscale heorot hornum trum 10

fleon10

holt-wudu sece

(1369)

Klaeber marksfleonand sired (2436b) as contracted forms which are metrically disyllabic.

164

Class I compounds

ferh(?um faegne,

fold-weg maeton (1633; similarly 1889b)

hyldo gehealdej?

Hord-weard sohte

(2293)

maeges daedum

morpor-bed stred

(2436)

mon mid his (ma)gum medu-seld buan (3065) The absence of double alliteration in these ten verses marks them as restricted to the b-verse. However, when a fully stressed compound is followed by a stressed element in the final X-position, the alliterative requirement weakens considerably and the number of b-verses exceeds the number of a-verses. There are 341 half-lines with this combination: 133 a-verses and 208 b-verses. Double alliteration is preferred to single alliteration in the a-verse by a ratio of nearly four to one, for example: (106X in the a-verse with double alliteration) won-smi wer

weardode hwile sonahaefde11

syn-sncedum swealh;

Samod aerdaege

(1311)

Sumne Geata leod

(1432)

sigor-eadig secg. guS-horn galan. sorh-leas swefan

(105) (743)

mid J?inra secga gedryht

(1672)

(27 X in the a-verse with single alliteration) pryS-word sprecen, mago-rinca heap.

Seod on sadum

(643)

Pa his mod ahlog

(730)

12

(1538)

Grendles modor

GuS-Geata leod

waetere gelafede

(2722)

for# ofereodon

(2959)

Eofor-lic scionon13

(303)

symbel ond seleful,

sige-rof kyning

(619)

Da waes on healle

heard-ecg togen

(1288).

wine-dryhten his freo&o-wong )?one

The 208 b-verses include: on ancre faest.

11

12 13

Line 743a displays irregular triple alliteration. Hoover, 'Evidence for Primacy of Alliteration', p. 81, n. 16, argues that syn- should be regarded as a stressed prefix. With irregular double alliteration in the self-alliterating compound. The second element -lie 'figure' must be distinguished from the suffix -lie.

165

The Metrical Grammar of Beowulf hyS-weard geara 14

Hraf>e waes set holme faehSo genoge,

(1914)

feorh-sweng ne ofteah

(2489)

FULLY STRESSED COMPOUNDS AND CLASHING STRESS

In half-lines which combine a monosyllabic simplex with a fully stressed trisyllabic compound the Beowulf-pott does not place the compound after the simplex unless the compound has short stress on its second element. The combination of trisyllabic compound with long stress on its second element plus monosyllabic simplex, forming a type E, is possible: A

N

A

won-sceli wer

weardode hwile

(105)

1

But there are no type D verses with clashing stress like *wer won-sceli. On the other hand, a trisyllabic compound with short stress on its second element can precede a monosyllabic simplex, forming a type E 2 : 15 A

t>aet hio Beowulfe,

(N)

N

beag-hroden cwen

(623) 2

Or it can follow a monosyllabic simplex, forming a type D : A

A

(N)

leof leod-cyning

longe J>rage

(54)

It is also a fact that the poet never places a fully stressed disyllabic compound after a monosyllabic simplex. Therefore fully stressed compounds are not found in final position in type C 1 . There are no verses with clashing stress like *paet wees godgudrinc, although there is one type C 1 verse with a reduced stress compound in the second lift: A

An

Oft Scyld Scefing

scea)?ena )?reatum

(4)

On the other hand, in half-lines which combine a disyllabic simplex with a disyllabic compound the poet places fully stressed compounds with long stress on the second element in either initial or final position. We find both (type A 2b ): A

AN

feondes fot-last\ 14 15

he to forS gestop

(2289)

The second element -weard 'guard' must be distinguished from the suffix -weard. It must be noted, however, that type E 2 is quite rare. There is only one a-verse (with double alliteration): 1584a. There are five b-verses: 463b, 623b, 783b, 1009b (with resolved stress on the final simplex) and 2779b.

166

Class I compounds

and (type A 2a): AN

A

wcel-gmt waefre;

ic ne wat hwae^er

(1331)

The crucial factors are stress and syllabic length. A fully stressed compound with two adjacent long stressed syllables will not be immediately preceded by a stressed monosyllabic simplex. It appears that the Beowulf-poet went out of his way to avoid three consecutive long stressed syllables with even or descending levels of stress. The stress pattern - | - - , as in *wer won-sceli, seems to be objectionable, whereas the pattern — — | —, which avoids clashing stess, as in wcel-gcest wcefre, is not. Furthermore, in this respect at least, resolved stress on a disyllable with a short first element is not the equivalent of a long stressed monosyllable. If any one of the first two stressed syllables in the sequence - | - - is replaced by resolved stress (wx), the pattern becomes acceptable. l6 It may be asked how many verses with the type D 1 contour — | — — x we should expect to find, if there were no prohibition against them. We can obtain an estimate by considering verses with the same elements in reversed order, namely type E verses with the contour — — x | — with or without resolution in any of the three stressed positions. Having determined what proportion of these lack resolution, we may postulate the same ratio for the type D 1 verses under consideration. There are forty-eight type E half-lines without resolution, like won-scelt wer (105a), in the a-verse in Beowulf.17 There are thirty-two very similar verses with resolved stress in one of the three stressed positions, 18 for example: (Ax) N

A

sigor-eadig secg. 16 17

18

Samod aerdaege

(1311)

Resolution of the half-lift in this sequence can not occur. 78a, 105a, 167a, 512a, *528a, 542a, 636a, 722a, 734a, 743a, *803a, *829a, 850a, 908a, 1042a, 1128a, 1160a, *1276a, 1278a, 1299a, 1429a, *1538a, 1567a, 1613a, # 1645a, 1889a, * 1918a, 1991a, 1993a, 2068a, *2189a, *2285a, 2393a, 2543a, 2671a, *2740a,(2843a, 2890a, *2894a, 2904a, 3052a, 3154a and *3l65a. To these may be added two half-lines with numerical compounds: 147a and 545a; and three half-lines with compounds with stressed prefixes: 573a, 1500a and 2695a. The ratio of double to single alliteration is 36:12 (75 per cent:25 per cent). The monosyllabic word does not alliterate in the half-lines with asterisks. 93a, 142a, *209a, 329a, 335a, 453a, 644a, *73Oa, 787a, 920a, 924a, 1001a, 1136a, 1187a, 1245a, 1271a, 1311a, 1416a, 1974a, 2153a, 2182a, 2317a, 2352a, 2379a, 2547a, *2551a, 2650a, *2722a, 276la, 3119a, 3l6Oaand *3l62a. The ratio of double to single alliteration is 27:5 (84 per cent: 16 per cent). The final word, monosyllabic or resolved, does not alliterate in the half-lines with asterisks.

167

The Metrical Grammar of Beowulf (Nx)

A

A

gled-egesa grim A

God wat on mec

(2650)

(?e him God sealde

(2182)

(Ax)

N

gin-fastan gife,

Resolved stress is obviously a routine variation, although even with three possible places for it to happen the number of a-verses without resolution exceeds the number with resolution by a ratio of 60 per cent to 40 per cent (48:32). The same patterns of distribution are found in the b-verse. There are ninety-five type E b-verses without resolution. 19 And there are fifty-three type E b-verses with resolution in one of the three positions. 20 The number of b-verses without resolution exceeds the number with resolution by a ratio of 64 per cent to 36 per cent (95:53). The ratio compares closely with that of the a-verses. The sequence — | — — appears ten times in the a-verse in Beowulf with resolution of one of the stressed syllables. These are in type D 1 verses,21 such as: (Ax)

A

N

fromum feoh-giftum

on faeder (bea)rme

(21)

weard wine-geomor•, wende )?aes ylcan

(2239)

(Ax) N

A

And there is one a-verse with a compound beginning with the prefix unand resolution of the initial simplex: (Ax) 19

20

21

N

N

eafoS uncupes. UJ?e ic swij?or (960) 8b, 24b, 104b, 110b, 131b, 161b, 193b, 207b, 24lb, 302b, 305b, 314b, 321b, 322b, 550b, 564b, 675b, 719b, 846b, 921b, 998b, 1051b, 1119b, 1169b, 1194b, 1195b, 1229b, 1240b, 1316b, 1332b, 1352b, 1365b, 1401b, 1459b, 1489b, 1582b, 1623b, 1631b, 1642b, 1715b, 1752b, 1782b, 1838b, 1881b, 1882b, 1908b, 1917b, 1924b, 1958b, 1982b, 2028b, 2097b, 2136b, 2245b,2270b,2279b,2281b,2357b, 2360b, 2377b, 2385b, 2411b, 2414b, 2510b, 2768b,2783b,2811b,2820b,2826b, 2831b, 2833b, 2886b, 2894b, 2901b, 3037b, 3043b, 3072b, 3143b and 3149b; similarly with stressed prefix: 130b, 150b, 276b, 410b, 602b, 928b, 955b, 1072b, 1269b, 1657b, ,1774b, 1776b, 2115b, 2188b, 2268b and 2938b (79 + 16). 19b, 82b, 152b, 236b, 391b, 422b, 659b, 849b, 9 3 5 b , 9 9 5 b , 1007b, 1028b, 1047b, 1200b, 1405b, 1 4 l i b (double resolution), 1425b, 1492b, 1511b, 1557b, 1577b, 1612b, 1634b, 1643b, 1669b, 1740b, 1878b, 1 9 0 5 b , 1 9 6 5 b , 1 9 8 0 b , 2 0 2 1 b , 2 0 8 6 b , 2 2 2 4 b , 2 2 2 7 b , 2 2 4 2 b , 2 2 4 3 b , 2 2 8 2 b , 2 3 1 3 b , 2 3 2 6 b , 2 3 2 8 b , 2 3 5 8 b , 2539b, 2576b, 2631b, 2661b, 2678b, 2714b, 2757b, 2763b, 2780b and 3039b; similarly with' stressed prefix: 876b and 2140b (51 + 2). 21a, 165a, 398a, 487a, 1109a, 1847a, 2239a, 2263a, 2408a and 2950a. Resolution of the second element of the compound does not occur. If magen Hre&manna (445a) were stressed - | - - , it would not display proper double alliteration: see below, ch. 14.

168

Class I compounds If the ratio of type E half-lines without resolution to ones with resolution can be taken as a valid indicator of the frequency of resolution, we should expect to find from sixteen to twenty type D 1 half-lines without resolution. 22 But we do not. One a-verse appears to have three consecutive long stressed syllables: A

A

N

wis wel-pungen,

J?eah Se wintra lyt

(1927)

There is also one a-verse with a compound beginning with the prefix unwhich might be said to have three consecutive long stressed syllables: A

AN

seghwylc gecwaeS23

egl unheoru\

(987)

The alliterative requirement of compounds precludes a fully stressed compound from appearing in second position in the b-verse. Two exceptions to this rule have already been noted, but they are worth looking at again in this connection. One of them, hroden ealo-wcege (495b), has double resolution. It does not therefore exhibit three consecutive long stressed syllables. The other is A

N

N

feond man-cynnes (164b) The only explanation for the almost total absence of half-lines without resolution is the ppet's avoidance of three consecutive long stressed syllables. And it may be questioned whether the three half-lines which I have cited as exceptions (987a, 1927a and 164b) really do have three consecutive stresses. Welpungen (1927a) seems to be a class II compound with reduced stress on the stem. 24 If the other two compounds in these verses were variably stressed, they would scan as type A. There is no question but that the prefix un- is variably stressed in the metrical grammar of Beowulf.25 And I have already argued on other grounds that man-cynnes is variably stressed. The fact that if man- were stressed we would have not 22

Since there are three places for resolution to occur in type E and only two in type D 1 , the ratios are not precisely comparable. If a n y t h i n g , this discrepancy should increase the expected n u m b e r of type D 1 half-lines w i t h o u t resolution.

23

Klaeber e m e n d s to egl[u] unheoru. W r e n n - B o l t o n defend the m a n u s c r i p t reading, t a k i n g egl as a feminine n o u n . L e h m a n n , 'Post Consonantal Imnr',

does not include this form

a m o n g the words he discusses w i t h final /, m, n, r. It looks to m e like a case of a 'light syllable'

4- consonantal / ( <

Worterbuch,

eg + / ) .

See H o l t h a u s e n , Altenglisches

etymologisches

s.v. egl(e).

24

See b e l o w , c h . 1 3 .

25

I assumed'Stress o n un- in 9 8 7 a in 'Prefix un-', b u t I also n o t e d ( p . 51) t h a t t h e unstressed form m a y be overlooked in t h e a-verse because of t h e a m b i g u i t i e s of alliteration.

169

The Metrical Grammar of Beowulf only an exception to the rule that compounds alliterate but also a pattern of stress which the Beowulf-poet regularly avoided seems to me to clinch the argument against its being stressed. Compound proper names in which the two elements remain semantically distinct behave for the most part just like other fully stressed class I compounds. Prime examples of such names are the sets of appellations for the principal tribes of the poem — the Danes and the Geats. The Danes are variously referred to as the Beorht-Dene, the Gar-Dene, the Hring-Dene, the East-Dene, the Nord-Dene, the Sud-Dene, the West-Dene, the Ar-Scyldingas, the Here-Scyldingas, the Sige-Scyldingas and the Peod-Scyldingas. In all these cases it is the second element which identifies the tribe and the first element which provides an alliterative variant. When the first element is crucial to the identification, the compound may be reduced in stress. It is not always easy to tell. Take these two half-lines, for example: (Ax)

(N)

A

hxl&Healf'Dena, (Ax)

Hncef Scyldinga

(1069)

gehyrde on Beowulfe

(609)

(N)

A

brego Beorbt-Dena;

The Half-Danes were apparently a distinct tribe, though this has been disputed. 26 Is Healf-Dene a fully stressed compound, as Beorht-Dena certainly is? The fact that it alliterates in second position in the a-verse is suggestive, but not decisive (I mark it as fully stressed). 27 Another name for the Danes is Ingwine ('the^riends of Ing'). This is evidently not a fully stressed compound because it appears both with and without alliteration in second position in the a-verse: (Ax)

(n)

A

eodor Ingwina AN

(n)

frean Ingwina,

on weald geteah

(1044)

fraegn gif him waere

(1319)

There are two family or tribal names with three elements, Merewioing (?) and the Wcegmundingas. I mark these as fully stressed (2607b, 2814a and 2921a). In the speech of the Messenger who reports the fight of Beowulf and Wiglaf against the dragon (lines 2900—3027), there occur two forms of the name Ravenswood, the place where Hygelac cornered Ongentheow and 26 Klaeber, p. 130; von Schaubert, pt 3, 2 7 2 - 3 . 27 Healfdene, the king of the Danes, is not a class I compound.

170

Class I compounds

defeated the Swedes. It is first given as Hrefnawudu (2925b) and then as Hrefnesholt (2935a). The variation proves that both elements of the compound were fully stressed. Since Hrefnes-holt is a self-alliterating compound, it is excluded from the b-verse. On the other hand Hrefna-wudu could be placed in either half-line. It seems likely that the place was commonly referred to as Hrefnesholt and that the alternate form was chosen in order to fit the name into the b-verse. 28 On the basis of this evidence, I am inclined to regard similar forms, like Earna-nces ('Eagle-point'), as class I compounds (see 3031b). 29 The alliterative behaviour of fully stressed compounds provides a way of testing for equivalents. Any compound or quasi-compound phrase which always alliterates in the second lift in the a-verse and never appears in the second lift in the b-verse and which imposes an alliterative requirement can be considered the alliterative equivalent of a fully stressed compound. Compounds with variably stressed prefixes behave in this way when the stress falls on the prefix. Un- certainly and man- probably are variably stressed, and therefore form class I compounds when they take initial stress. Three disyllabic fixed stress prefixes - ofer-, wider- and ymb(e)— likewise seem to form class I compounds. There are a few compounds the two elements of which may no longer have been semantically distinct, like aglceca {ceglceca), which almost certainly still belong to class I. Compounds formed with the disyllabic suffix -scipe and with the suffixes -dom, -fast, -feald, -full, -leas, -lie, -sum and -weard when they become disyllabic through inflection behave like fully stressed compounds. Because these various types of compounds resemble class II or class III compounds or may become class II or III compounds as a consequence of variations in stress or inflection fuller treatment of them will be reserved for the following chapters. Two types of quasi-compound phrases meet the conditions set for class I. (1) Qualitative {and numerical) adjective + base phrases and base + stressed

adjective (of whatever kind) phrases. Such phrases appear five times in the second lift of the a-verse, always with double alliteration: 30 28

Ravenholt is said to be 'a very c o m m o n place-name in D e n m a r k ' . Klaeber, p . xlii, n. 3 .

29

So: Fres-lond ( 1 1 2 6 b and 2357b); Fres-wal (1070a); Hreosna-beorh (2477b); Hrones-nces

30

T h e hypermetric verse gan under gyldnum

( 2 8 0 5 b and 3136b); S cede-land {I9b)\ Sceden-ig (1686a); Weder-mearc (298b). beage (1163a) exhibits double alliteration.

However, I regard t h e initial alliteration o n t h e infinitive gan as extra-metrical. T h e phrase gyldnum beage therefore occupies t h e first lift.

171

The Metrical Grammar of Beowulf hond ond heard sweord ymb hord wigan (2509; similarly 2638a, 2964a, 2995a and 3105a) They do not appear in the second lift of the b-verse. They put an alliterative requirement on a following simplex in the non-final X-position in the a-verse (2 X): seofon niht swuncon;

he J>e set sunde oferflat

heardswyrd hiked,

ond his helm somod

(517) (2987)

There are only three cases of a simplex in the non-final X-position in the b-verse, always a proper name after the phrase wine min\ geond widwegas,

wine min Beowulf (1704; also 457b; similarly 530b)

A simplex appears in the final X-position in the a-verse only three times, twice with alliteration, but there are sixteen cases of a simplex in the final X-position in the b-verse. 31 All phrases of this kind are the alliterative equivalents of fully stressed compounds in Beowulf. (2) Dependent genitive!dative -f base phrases (in either order). The case for treating these phrases as the alliterative equivalents of fully stressed compounds is strong, though not conclusive. I will review the evidence for each of the two orders. Base + dependent genitive/dative: (a-verse: double alliteration (3X)) swutol sang scopes.

Saegde se |?e cu(?e

beorht beacen Godes, set fotum s&tfrean Scyldinga;

brimu swa(?redon

(90) (570)

gehwylc hiora his fertile treowde (1166)

(b-verse: failure of compound to alliterate (IX)) geong goldhroden,

31

gladum suna Frodan

(2025)

A-verse, double alliteration: 147a and 545a; single alliteration: 985a; b-verse: 523b, 651b, 767b, 823b, 999b, 1417b, 1424b, 1567b, 1790b, 2017b, 2080b, 2156b, 2610b, 2691b, 3030b and 3094b. The final simplex is never a natural stressed element in the b-verse.

172

Class I compounds

Dependent genitive/dative + base: (a-verse: compound in second lift, with double alliteration (23 X) 32) faeger foldan bearm; atol cese wlanc Metod manna gehwces.

fundode wrecca eftsiSas teah Ic eom on mode from

(1137) (1332) (2527)

(b-verse: failure of compound to alliterate (2X)) Gecyste J?a

cyning czpelum god

gesaet on sesse;

seah on enta geweorc

(1870) (2717)

There are no simplexes in the non-final X-position following this kind of compound phrase. Nor are there any simplexes in the final X-position in the a-verse, but there are nine in the b-verse. 33 In all there are twenty-six a-verse half-lines with the compound phrase in second position. All of them display double alliteration (and the sequence of monosyllabic stressed simplex plus disyllabic compound does not occur). But there are also three b-verse half-lines with the compound phrase in second position. These constitute exceptions to the alliterative rule if the phrases are the equivalent of class I compounds. Despite the fact that three exceptions seems excessive in a small body of evidence, I do not think it is necessary to treat these dependent genitive/dative + base phrases as a separate group with its own alliterative rules. One of the exceptions is seah on enta geweorc (2717b). This is a much discussed verse. 34 It is highly irregular in that the alliteration falls on the finite verb seah rather than on the noun enta. It is, as E. G. Stanley says, an 'unsolved' problem. Whatever the solution may be, it does not seem a persuasive counter example. 32

3 4 1 a , 4 0 0 a , 8 4 8 a , 9 7 8 a , 1038a, 1137a, 1165a (hypermetric), 1286a, 1332a, 1364a, 1369a, 1420a, 1594a, 1627a, 1706a (hypermetric), 1906a, 1932a, 2 1 7 8 a , 2 2 1 7 a , 2 5 1 3 a , 2 5 2 7 a , 2 6 8 7 a and 2 7 7 4 a .

33

711b, 1904b, 2246b, 2290b, 2469b, 2662b, 2794b, 2836b and 2853b.

34

See esp. Stanley, 'Verbal Stress', p p . 3 0 6 - 2 1 . H e calls t h e p r o b l e m of t h e verse 'unsolved' on p . 2 1 . R u s s o m , Old English Meter, p p . 9 1 - 2 , argues t h a t enta geweorc is a 'genitive c o m p o u n d ' . This does not g e t around t h e present difficulty, since w h a t he calls genitive c o m p o u n d s would be class I c o m p o u n d s in m y terminology and t h e alliterative irregularity remains. D o n o g h u e , Style,

p p . 37—40, proposes t a k i n g seah

on as a

corruption or alternate spelling of t h e infinitive seon or *seohon and scans t h e verse as a type E 2 ( p u t t i n g t h e second element of a class I q u a s i - c o m p o u n d phrase into t h e second lift instead of t h e first - a construction w i t h o u t parallel in

173

Beowulf).

The Metrical Grammar of Beowulf Another of the exceptions is gladum suna Frodan (2025b). But this is only an exception on the assumption that the adjective gladum modifies the noun phrase suna Frodan, 'gracious Froda's-son' {gladum of course agrees with suna, not Frodan). In fact, it seems clear that Frodan is dependent on the phrase gladum suna, 'Froda's gracious-son'. That is, what we have is an initial compound phrase followed by a proper noun in the non-final X-position. If this analysis is correct, the best way to determine the underlying syntax of adjective + noun + dependent genitive is by inspection. 35 There are two possibilities. Either the adjective modifies the noun + dependent genitive phrase, or the adjective + noun phrase is followed by a dependent genitive. The former is clearly the case in beorht beacen Godes (570a) where beacen Godes is a metaphor for 'sun', and double alliteration confirms that the adjective modifies the phrase. In 2025b, as we have just seen, the absence of alliteration supports the latter interpretation of the syntax. However in the case of swutol sang scopes (90a), where there does not seem to be a clear choice between 'the sweet scop's-song' and 'the scop's sweet-song', it is only the double alliteration which indicates that the former is the correct phrasing. The third exception then, cyning cepelum god (1870b), is the only unquestionable counter example to the proposition that dependent genitive/dative + base phrases should be treated as alliterative equivalents to fully stressed compounds. There are nine other half-lines with the same syntax (noun '4- dependent dative + adjective). 36 All are a-verses with double alliteration on the first two nouns. I see no reason to doubt the authenticity of this verse, but it is plainly an uncommon one. On balance, it seems to me best to include these phrases as the alliterative equivalents of fully stressed compounds and to consider 1870b to be a violation for whatever reason of the poet's metrical grammar. 35

M a l i n g , 'Sentence Stress', sets o u t t h e principles for d e t e r m i n i n g stress subordination w h e n there are m o r e t h a n t w o stressed elements in the half-line. However, she sees 'no motivated internal structure' in these particular verses. See her discussion of 90a and 570a on p . 3 8 5 . Unfortunately, unless I have overlooked it, she does not m e n t i o n 2 0 2 5 b . Keyser's solution ('Old English Prosody', p . 3 3 9 ) , which is to bracket (adjective + (noun + noun)), works for 9 0 a and 570a, b u t not for 2 0 2 5 b .

36

9 7 8 a , 1038a, 1286a, 1364a, 1369a, 1594a, 1906a, 2 1 7 8 a and 2 6 8 7 a .

174

13 The alliterative behaviour of class II compounds

Mandatory alliteration marks fully stressed class I compounds. Class I compounds always alliterate in the second lift in the a-verse; they are excluded from the second lift in the b-verse. Furthermore, they impose an alliterative requirement on a following simplex in the a-verse. Finally, a class I compound whose first two syllables are long and stressed will not be preceded by a stressed monosyllabic simplex. The alliterative behaviour of the various compounds and compound phrases which have not already been described provides the evidence for their classification as class II or class III compounds. CLASS II COMPOUNDS

Class II compounds alliterate in the a-verse (except when the stressed element has a prefix which falls in the second weak metrical position (w2)), but they are not excluded from the second lift in the b-verse (where, of course, they do not alliterate), and they impose no alliterative requirement on a following stressed element. They include compounds with fixed stress prefixes, most compounds in which one or both elements have not retained their full semantic value, compound numerals, pronominal and adverbial compounds in ag-9 the pronouns nathwylc and welhwylc, simplexes which are preceded by a proclitic and their alliterative equivalents — compound phrases in which one or more of the lift positions is occupied by a verb or an adverb.

175

The Metrical Grammar of Beowulf Nouns, adjectives, verbs and adverbs with fixed stress prefixes The fixed stress prefixes in Beowulf are: cef-, cet-, and- (ond-), ed-, ful-, in-, ofer-, on- (an-), or-, u<5-, wider- and ymb(e)-.l With the probable exceptions of ofer-, wider- and ymb(e)-, these form class II compounds. They alliterate in the a-verse and yet are found in second position in the b-verse. The second element of compounds with these prefixes presumably loses some degree of stress. 2 The relevant examples are these: (a-verse: compound in second lift, with double alliteration) aenlic ansyn.

Nu ic eower sceal

eald ond anhar

(357)

ond hine ymb monig

(689)

mid his eorla gedriht

eorles andwlitan,

No J>aet ySe byS

(1002; also 1565a)

4

oSer earmsceapen

(1351)

oS J?set hine yldo benam

(1886)

aldres orwena. idese onlicnes; aeghwaes orleahtre,

(251) 3

(b-verse: compound in second lift, without alliteration) searonet seowed modges merefaran, gryreleoS galan

smiles orpancum

(406) (502)

micel cefpunca Godes andsacan

HroSgar ma|?elode fremu folces cwen, Paer waes Hondscio

(786; also

1682b)

h i m on andsware

(1840)

firen' ondrysne

(1932)

hi\d onscege

(2076; similarly 2483b) frodan fyrnwitan Pa waes aet Sam geongan 1

2

feorh uSgenge grim andswaru

(2123) (2860)

JEt-, ful-, ofer-, on- and ymb- are stressed before nouns and adjectives, but remain unstressed before other parts of speech. See above, ch. 8. Bi- is the stressed form of the prefix be- and fra- is the stressed form of the prefix for-. These stressed forms are etymologically the first elements of beot (from *bi-hat) and gebeotian and of fracod (from *fra-cu8) and fratwe and the weak verbs frcetwan and gefrcetwian. See Sievers-Brunner, § 43, n. 4, and Campbell, Old English Grammar, § 74, n. 6. I treat them as simplexes, although it may be noted that in Beowulf these forms always either alliterate or are found in the second lift of the b-verse, and therefore may be indeed class II compounds. I take an-hydig and an-radto be compounded with an- ('one') and thus to be class I compounds. 3 4 Bliss, Metre, § 32 (6). Manuscript: un hdr. Manuscript: onlic nces.

176

Class II compounds

If these were fully stressed class I compounds they would not only violate the rule excluding such compounds from the second lift of the b-verse, but several of the verses would also be exceptions to the prohibition against a stressed monosyllabic simplex preceding a compound with two consecutive long, stressed syllables (i.e., 2076b, 2123b and 2483b). The disyllabic (or potentially disyllabic) stressed prefixes ofer-, wifterandymb(e)- probably form fully stressed class I compounds. The evidence is slight. No class II compound in Beowulf is found with resolved stress on its second element, unless we take ofermcegene as class II: (Ax) (Nx)

elne geeodon

mid ofermcegene

(2917)

The medial vowel is sometimes syncopated when mcegen is inflected. It could be argued that the poet intended *mid ofermcegne. But I suspect that the answer is that no stress reduction was associated with these disyllabic prefixes. With the exception of wider- in the adverb wifterrcehtes (3039b), every one of the disyllabic prefixes is carefully separated from its stem in the Beowulf manuscript. 5 Perhaps the slight trochaic rhythm of the disyllabic prefix tended to preserve the stress on the stem. The adverbial prefixes cer-, eal- (pel-, al-, eall-) and wel- appear to have the same effect on the stem as the monosyllabic prefixes listed above. We find these compounds in the second lift of the b-verse: Mael is me to feran;

Faeder alwalda

swatfah syrce,

(316)

swyn ealgylden (1111; similarly 2767b)

f>egnas syndon ge|?waere, sarigmodum

)?eod ealgearo (1230; similarly 2241b)

somod cerdcege

(2942; also 1311b)

The compound welpungen appears in the second lift of the a-verse following a stressed monosyllable: wis welpungen,

J?eah 3e wintra lyt

(1927)

If it were a class I compound, it would be an exception to the prohibition against a stressed monosyllabic simplex preceding a compound with two consecutive long, stressed syllables, as would ealgylden (1111b and 2767b). 5

Stressed o/er-: 1740b, 1760b, 2917b and 2993b; stressed ymb{e)-: 9b, 1827a and 2734a. This is not a decisive argument: class II compounds are often similarly separated. 177

The Metrical Grammar of Beowulf I mark all these compounds as class II, with the exception of the whole-verse compound eallirenne (2338a) and compounds consisting of cer+ unstressed prefix + stem which must be class I. 6 Compounds in which one or both elements have not retained their full semantic value

(a) Compounds with the separate meanings of the elements apparently obscured. It is difficult to be certain where or whether such loss of separate meanings has occurred. Phonological weakening of one or both elements is one independent clue. Another is the placement of the compound where fully stressed compounds are generally excluded (that is, in a sequence of three long stresses or in the second lift of the b-verse). Neither of these factors is decisive. Compounds that might fall into this category have to be investigated one at a time, and the conclusions should be treated with caution. The compound aglceca (ceglceca) behaves in all respects like a fully stressed class I compound. It always alliterates (19X), even when it appears in second position in the a-verse. It never appears in second position in the b-verse. Furthermore, in line with the Beowulf-pott's avoidance of three consecutive long stressed syllables with descending or even levels of stress, it is always (4 X) preceded by a disyllabic word with resolved stress when it is in second position in the a-verse immediately following a stressed element: atol ceglaeca ealdre [?inum7 (592; also 732a and 816a; similarly 2557a) The compound fyrwet (fyrwyt) appears three times in Beowulf always in the same traditional half-line formula: 6

7

A type A l or D 1 with the second element of a class II compound in the second lift seems doubtful. If cergestreona (2232b) and eallirenne (2338a) were class II compounds, they would be the only examples of either. The weak prefix ge- coming between ter- and the stem doubtless preserves the stress on -streona (as also in 1679a and 1757a). The double alliteration in eall-irenna is another indicator of its class I status. (These considerations need not apply to personal proper names like Eormenrk (see 1201a) and Yrmenlaf (see 1324a), which remain class III compounds.) Lehmann, 'Post Consonantal / m n r\ p. 155, argues that atol should always be read metrically as a monosyllable, atl. In view of the poet's use of resolved stress to avoid three consecutive long stressed syllables, I think the situation is more complicated than Lehmann allows.

178

Class II compounds N

A

fyrdsearu fuslicu; hine fyrxvyt braec (232; also 1985b and 2784b; cf. Juliana 27b) Elsewhere in Old English poetry it always alliterates, and once in the a-verse it is followed by an alliterating verb: fyrwit frineS,

fus gewiteS {Solomon and Saturn 58)

The evidence is not conclusive. Fyrwet may have behaved like a fully stressed compound. On the other hand, fully stressed compounds rarely, if ever, appear with the second element in unstressed position in type B verses as fyrwet does in the formula hine fyrwet brcec for this reason I mark it as a class II compound. Garsecg and sidfat can for the same reason be labelled as class II compounds in Beowulf even though the evidence again is slight. 8 Garsecg appears twice in the a-verse, both times in a clause-initial, displaceable (Ib) half-line, preceded by a finite verb with extra-metrical alliteration: a

A

geafon on garsecg; a

him waes geomor sefa

(49)

A

glidon ofer garsecg;

geofon yj?um weol

(515)

But in its one other occurrence in the poem, it is found with its second element in unstressed pQsition in a type B half-line: A

on geogoSfeore -

N

)?set wit on garsecg ut

(537)

Sitifcet alliterates both times it appears (202a and 2639a), but in the former its second element is in unstressed position: -AN

Done sidfat him

snotere ceorlas

(202)

The compound hlaford always alliterates in Beowulf — five times in uninflected form, once in the form hlaforde and once in the form hlafordes. It fails to impose an alliterative requirement in A

N

hlaford sinne.

Da waes hord rasod

(2283)

It would appear to be a class II compound, although it may well have behaved like a class I compound in inflected form. The compound oretta occurs only twice in Beowulf. The two verses in 8

Hoover, 'Evidence for Primacy of Alliteration', p. 77, includes these among his Class Two.

179

The Metrical Grammar of Beowulf which it appears are: A

A

yrre oretta,

|?aet hit on eorSan laeg

(1532)

A N

Aras 3a bi ronde tot oretta (2538) Its appearance in the second lift in a b-verse is enough to establish that this is not a fully stressed compound. It appears to belong to class II. The compound missere occurs four times in Beowulf, always in second, non-alliterating position in the b-verse.9 It is not >a fully stressed compound. It also appears to belong to class II. Ombeht is a worn-down form of obscure origin. 10 It appears in two a-verses with double alliteration; once in the first lift, and once in the second: ombeht unforht: ^ghwae)?res sceal (287) ar ond ombiht. Ne seah ic el)?eodige (336) Elsewhere in Old English poetry it always alliterates, with the exception of one verse in the Paris Psalter {Psalm 115.6,2), where it occupies the second lift of the b-verse. I mark it, doubtfully, as a class II compound. The two elements of wilcuma would appear to be semantically distinct, yet the compound, which occurs three times in Beowulf, always as the complement of the subject, is found in the second lift in the b-verse in hider wilcuman (394b). 11 Possibly the use of the noun as or like a predicate adjective in more or less the modern sense of 'welcome' had begun to obscure its root meaning of'desired-guest'. Here Wulfgar reports to the Geats what Hrothgar has just said in the hall: Eow het secgan sigedrihten min, aldor East-Dena, )?2et he eower ae(?elu can, ond ge him syndon ofer saewylmas heardhicgende

hider wilcuman.12

(391-4)

I mark it as a class II compound. 9 10

II 12

Fela missera 153b and 2620b; hund missera 1498b and 1769b. According to Holthausen, Altenglisches etymologisches Worterbuch, s.v. ambiht, it derives from Gallo-Latin ambactu-s. Similarly in leofwilcuma (Maxims I 94b). 'My victorious lord, the chief of the East-Danes, ordered me to say to you that he knows your lineage, and that you stout-hearted men are welcome to him hither over the sea-waves.' Wilcuma is in the first lift in pcet hie sint wilcuman (388b), where it is the complement of hie, and in cwced pcet wilcuman (1894a), where it is the complement of scapan scirhame (1895a).

180

Class II compounds (b) Compounds in which the second element has been reduced in semantic force to the status of a suffix. These include compounds with the monosyllabic suffixes -dom, -fast, -feald, -full, -leas, -lie, -sum, -weard.13 When the suffix becomes disyllabic through inflection, the compound is fully stressed; otherwise it belongs to class II. No compound with a disyllabic suffix appears in the second lift in the b-verse, but we find the following compounds with a monosyllabic suffix in that position: (Ax) (Nx)

fyre gefysed.

Waes se fruma egeslic A

Da waes haten hre)?e

(2309)

N

Heort innanweard (991; similarly 1976b)

Furthermore, the alliterative requirement of compounds fails in the a-verse when the suffix is monosyllabic: A

N

earmlic wurSan, A

ond se ellorgast

(807; similarly 842a)

N

dryhtlic iren;

draca morSre swealt (892; similarly 1809a)

When a disyllabic suffix is added to a fully stressed compound, the suffix is evidently reduced in stress. Otherwise in a verse like 3138b: A

ad on eorSan

N n

unwaclicne

(3138)

there would be three consecutive long stressed syllables. The suffix -lie presents several problems. Editors mark the vowel long or short, depending on a combination of metrical and phonological considerations. It is etymologically long, but, according to Campbell, was subject to Early Old English shortening even in inflected forms. l4 This shortening explains the occasional retention of final -u which would otherwise be lost after a long syllable under stress. Klaeber marks the vowel long except in the form -lieu (£32a, 64 la, 1584a and 2637a; 194 lb) and once in the form -lice (1426a; cf. -lice with a long vowel in 585a and 1158a). The behaviour of compounds with monosyllabic -lie supports Campbell's See Campbell, Old English Grammar, § 88. When it is the second element of a compound with the meaning 'guardian', or 'watch', -weard is not a suffix in this sense, nor is -lie when it means 'body' or 'figure'.

Ibid., §356.

181

The Metrical Grammar of Beowulf contention that the 'practice of adding marks of length to the vowels of final syllables of reduced meaning (e.g., Hropgar, hetelic) has nothing to recommend it'. 15 Disyllabic (and presumably stressed) -lieu for the most part seems to vary freely with monosyllabic (and presumably unstressed) -lie. The formula fyrdsearo fuslie (2618a), which scans as type A2, implies that -lieu might be resolved in fyrdsearu fuslicu (232a), although resolution of the final drop in any type A is to my mind doubtful. (I do not resolve the short stressed syllable of a disyllabic second element of a fully stressed class I compound unless it is immediately preceded or followed in the same half-line by a syllable of lesser stress. l6 ) For similar reasons, I am inclined to doubt that -lieu is resolved in freolicu folccwen

to hire frean sittan

(641)

In peah &e hio cenlicu sy (194 lb), the suffix falls in the drop of what has been assumed to be a type B verse, and resolution (to say nothing of stress) creates difficulties. On the basis of the last two verses, we might posit that the suffix -lieu was weakly stressed. David Hoover points out that in the Old English poetic tradition fully stressed self-alliterating compounds are excluded from the b-verse and from the a-verse when they would generate triple alliteration. 17 Therefore ladlieu should not be fully stressed in laSlicu lac.

He him J?ass lean forgeald

(1584)

Based on this evidence he puts the suffixes -dom, -fast, -full, -lie, -sum and -weard{zs well as the stressed prefixes) into his Class Two. His argument is unassailable, and it is tempting to label these as class II compounds in Beowulf. The problem is that the Beowulf-poet avoids the use of selfalliterating compounds composed of stem + suffix (with this single exception) or of stressed prefix + stem. In Beowulf the inflected forms of these suffixes as ,well as variably stressed prefixes with stress on the prefix pattern like class I compounds. Here is another illustration of the fact that the Beowulf-poet's metrical grammar exhibited features which were not necessarily shared by the poetic tradition. 15 16

17

Ibid., §90, n. 4. See the headnote to the Index of Alliteration, Scansion and Metrical Clause Structure, p. 231. 'Evidence for Primacy of Alliteration', pp. 77 and 88.

182

Class II compounds

A more radical solution would be to argue that, since -lieu varied freely with -lie, final -u should be dropped as a probable scribal corruption wherever it leads to an unacceptable metrical contour. Although this is the solution that I favour, I have refrained here, as elsewhere, from emending the text in order to suit my assumptions, and I therefore have marked all the compounds with -lieu as fully stressed. In the case of 194lb, taking -lieu as stressed leads to the conclusion that the line in which it appears is hypermetric: idese to efnanne,

J?eah Se hio aenlicu sy

(1941)

The first half-line scans HC 1 and the second hE 2 . 1 8 Compound numerals

The combination of numeral + numeral should probably be considered as a class II compound. There is one example in Beowulf that is relevant: Swa se SeodsceaSa

preo hund wintra

(2278)

This is a formula in the Old English tradition. It appears in the a-verse in Soul and Body I, where the compound numeral obviously does not impose an alliterative requirement on the simplex. 19 Nor can there be double alliteration in most compound numeral phrases of the type tu hundondpreo, which nevertheless appears in the a-verse in Elene.20 None of this is decisive, but it seems likely that compound numerals behave like other compounds with reduced stress. Pronominal and adverbial compounds in aeg-, the pronoun nathwyle and

welhwylc Compounds in ag- and the pronouns nathwyle and welhwylc are probably class II rather than class III compounds, although the evidence is inconclusive. Nathwyle and welhwylc are found in the second lift in the a-verse, both with double alliteration (2X): 18

W h e r e , in accordance w i t h Pope's system for labelling h y p e r m e t r i c verses, H stands for a s e g m e n t w i t h a stressed element before a regular metrical contour, and h stands for a s e g m e n t w i t h o u t a stressed element before a regular metrical contour.

19

Soul and Body I 12a; Soul and Body II 12 (the corresponding half-line is missing).

20

Elene 2a.

183

The Metrical Grammar of Beowulf witena welhwylc niS[S]a nathwylc,

wide geond eorj?an (se (?e ne)h gefe(al)g

(266) (2215)

JEghwa, ceghwylc and nathwylc are found in the second lift of the b-verse (6X): dugu)?e ond geogo(?e dsel ceghwylcne (621; similarly 984b, 1384b, 2053b, 2223b and 2233b)

Simplexes which are preceded by a proclitic

When a proclitic in the w 1 of the a-verse directly precedes its base word which is in the second lift, the base word must alliterate (with certain exceptions) according to the alliterative rule of proclitics. 21 This is precisely in accordance with the alliterative behaviour of the class II compounds. Therefore, if the base word is a simplex, the combination proclitic + simplex evidently forms a compound or quasi-compound phrase which belongs to class II. Subject/object + verb phrases

Phrases in which a verb is preceded by a subject or object other than a personal pronoun appear to form the equivalent of class II compounds. These phrases behave differently from the compound phrases described in ch. 12. The verses marked (°) contain two nominals. There are twenty-four of them. These, we might intuitively suppose, would be most likely to attract alliteration to themselves. Yet only two are in the a-verse, and in one of these, Flod blode weol (1422a), the object + verb phrase still fails to alliterate (although the internal rhyme flod I blod- and the closely related initial consonant clusters FL and BL are suggestive). All of the verses that do not contain two nominals are found in the b-verse. The tendency of the finite verb to move to the final position of the b-verse is of course well known. 22 This may explain the comparative absence of a-verses in this group. Despite the lack of evidence, it seems best to associate these phrases with class II compounds, and to note 1422a as a (partial) exception to the requirement that class II compounds alliterate in the a-verse. 21

See above, chs. 8-10.

22

See Lehmann, 'Proto-Germanic Syntax', p. 244. 184

Class II compounds (a) Subject + verb: (b-verse only: phrase in second lift, without alliteration (4X)) feondes

fingras;

siSode sorhfull;

foran ceghwylc wees

(984)

°sunu dead fornam (2119; similarly °2706b and 3129b)

(b) Genitive/dative/accusative/instrumental object + verb: (a-verse: phrase in second lift, with double alliteration (IX)) °holmheolfre weoll,

ond ic heafde becearf

(2138)

(a-verse: phrase in second lift, without alliteration (IX)) °Flodblode weol

- folc to saegon

(1422)

(b-verse: phrase in second lift, without alliteration (24X) 23 ) hringedstefnan, —

°holmstorme weol

fremeS on folce,

feor eal gemon

(1131) (1701)

Adverb + verb,phrases

When an adverb is immediately followed by a verb, the phrase forms a class II quasi-compound in which the adverb takes the heavier stress. 24 The alliterative behaviour of these compounds parallels that of the subject- or object-verb phrases just discussed. There are sixty-eight half-lines with an adverb-verb phrase or adverb + adverb-verb phrase in second position. Sixty-six of the sixty-eight are b-verses. Both a-verses display double alliteration (one of them, freode, swa wit furfSum sprcecon (1707a), is a hypermetric verse). (a) Adverb + verb (in this configuration there seems to be no significant variation in behaviour between a lexical verb and an auxiliary): (a-verse: phrase in second lift, with double alliteration (2X)) Fyrst ford gewat\ flota waes on ySum freode, swa witfurdum sprcecon. 23

24

(210)

Du scealt to frofre weor|?an (1707)

O °494b, °515b, °742b, °892b, 1057b, 1080b, 1122b, °1131b, °1132b, 12l4b, O l 4 2 3 b , ° 1 5 2 0 b , ° 1 5 6 9 b , 1 7 0 1 b , ° 1 9 9 7 bO ,2 2 l 6 b , ° 2 3 9 6 b , ° 2 5 9 3 b , ° 2 6 0 9 b , ° 2 6 7 2 b , °2693b, °2966b, °3118b and °3155b. The verses marked (°) contain two nominals. Campbell, Old English Grammar, § 9 4 and n. 1.

185

The Metrical Grammar of Beowulf (b-verse: phrase in second lift, with long stress on penult (2X)) A

Flod blode weol

N N

- folc to sagon (1422; similarly 1650b)

(b-verse: phrase in second lift, with short stress on penult (3X)) (Ax)

guSsearo geatolic;

N

(N)

guman ut scufon (215; similarly 572b and 3131b)

(b-verse: phrase in second lift, with long stress on final (resolved stress does not occur) (53 X) 25) AN

Beowulf waes breme

N

- blaed wide sprang

(18)

Bliss consistently puts heavier stress on the verb than the adverb in these b-verses. In effect, he would argue that here the adverb—verb phrases constitute quasi-compounds with the stress patterns \ / x and \ x /, although when they occur by themselves without a preceding stressed element or when they occur in the a-verse with a preceding stressed element they form quasi-compounds with the stress patterns / \ x and / x \ , as their alliterative patterns indicate. Although Bliss's claim cannot be disproved, it is hard to avoid the conclusion that he puts it forward in order to be able to assert, as he does, that metrical types D 1 , D 2 and D 4 with secondary stress belong exclusively to the a-verse and require double alliteration. But it is precisely this assertion which he uses to argue that the heavier stress in all these verses belongs on the verb. 26 I prefer to assume that the stress pattern on these compounds remains the same and to argue that it is the nature of the compound rather than the metrical type which determines whether or not double alliteration is obligatory. 25

18b, 55b, 9 7 b , 121b, 171b, 315b, 3 4 l b , 390b, 4 7 8 b , 4 9 6 b , 519b, 569b, 761b, 782b, 8 5 7 b , 871b, 8 9 7 b ,

1017b,

1031b,

1146b,

1160b,

1377b,

721b, 1416b,

1503b, 1526b, 1570b, 1588b, 1615b, 1619b, 1800b, 1 9 1 2 b , 2 1 1 3 b , 2 1 5 4 b , 2 2 1 3 b , 2 2 5 4 b , 2 2 9 6 b (with irregular double alliteration on final verb), 2 3 1 9 b , 233 l b , 2365b, 2552b, 2 5 6 2 b , 2 5 7 0 b , 2 5 7 5 b , 2 6 0 0 b , 2 7 6 4 b , 2 7 7 6 b , 2 7 7 7 b , 2824b, 2 9 3 7 b , 2941b, 2 9 7 3 b and 3060b. Pope scans 9 7 b , 1377b and 2 7 7 6 b as type E. H e hesitates between D 4 and E in the cases of 121b and 8 9 7 b . All the rest he scans as D 4 . The adverbs in question in these five verses are eac (twice), sona, hat (adjective used adverbially: see Pope, Rhythm, p . 364) and git. I do not see any compelling reason to treat these differently from any of the others. 26

Metre, §§ 6 1 - 3 , 7 7 . In his text at this p o i n t Bliss uses Sievers's typology which differs from his own in A p p e n d i x C. D 2 in § 77 is the same as D 3 in A p p e n d i x C etc.

186

Class II compounds The reduction of stress on the second element of these class II compound phrases appears to permit the sequence of stressed monosyllabic simplex plus stressed compound with two consecutive long syllables, because we find the verse folc to scegon (1422b). (b) Adverb + adverb + verb: (b-verse only: type D 4 (?)) \>ritig )?egna; fet ond folma.

panon eft gewat Ford near cetstop

freode to friclan.

(123) (745)

From arest cwom

(2556)

nihtes hwilum,

nyder eft gewat

(3044)

hordgestreona,

hider ut cetbasr

(3092)

eft sona bid

(1762)

(b-verse only: type E (?)) ane hwile; handgewri)>ene;

hrape seoptSan wees

J?aet he )?one niSgaest

nio&or hwene sloh

(1937) (2699)

Since all eight half-lines in this group are in the b-verse, there is no way of determining from inspection whether it is the second adverb or the verb which is more heavily stressed. 27 Bliss puts the stress on the verb in every case; Pope prefers to stress the second adverb except in 745b, 1762b, 1937b and 2699b. 28 The problem resembles that of the adjective + noun + dependent genitive phrases. 29 The two adverbs could be analysed as a compound phrase which governs the verb or the first adverb could be taken as governing a compound adverb—verb phrase. I do not pretend to offer a definitive solution. My inclination is to assume that the compound adverb-verb phrasing prevails wherever possible. Only in 2699b is it certain that the second adverb modifies the 27

T h e t h i r d possibility is t h a t neither was elevated in stress over the other and therefore that the stress contour of these verses was 1/2/2. O f course it could be argued t h a t the distinction between all types D 4 and E and between A 2 a / k and D 1 / 2 is meaningless in t h e b-verse. However, accepting this a r g u m e n t does n o t , as one m i g h t expect t h a t it w o u l d , lead to a simpler description of the metrical p a t t e r n s .

28

In t h e original edition of Rhythm,

Pope stressed t h e second adverb in 7 4 5 b and 1 9 3 7 b ; in

the Preface to t h e 2 n d e d . , p . xxiii, he hesitantly reassigns verses like 7 4 5 b and 1937b to type E (that is, p u t t i n g stress on the final verb). 29

See above, ch. 12.

187

The Metrical Grammar of Beowulf preceding adverb rather than the following verb. Wiglaf struck the dragon nio&or hwene, 'a little lower down'. In view of the subsequent development in the language of an adverb 'eftsoon(s)', it might seem to be beyond doubt that eft sona is a compound in 1762b. The problem here is that as a compound the phrase is accented on its second element in its one other occurrence in Old English poetry, just as it was later to be in Middle English. 30 Apparently the Beowulf-pott stressed the phrase eft sona on its first element, although possibly we ought to take sona with fe'#as an adverb + verb phrase. The phrase hra&e seopdan (in 1937b) appears eight times elsewhere in Old English poetry, sometimes by itself and sometimes with a following verb, but always in the b-verse with alliteration on hrade.51 I classify these three verses as type E and the others as type D 4 . Other verb phrases

When the verb precedes a pronominal subject or object or an adverb, the phrase is normally found in the dip, where it may receive metrical stress by the transformational rule. 32 However, there are four b-verses where the verb is a stressed element. In these cases the combination of verb + pronominal subject/object or verb + adverb probably forms a class II compound after the insertion of the verb into the initial stressed position in the b-verse. The four verses are: (a) Verb + pronominal subject/object: faege gefealleS; breostgewaedu.

fehd oper to

(1755)

Bruc ealles well

(2162)

Leofa Biowulf,

Last eall tela

(2663)

(b) Verb + adverb: Seodne aet )?earfe;

30

32

(1525)

T h e other occurrence of eft sona in O l d English is in Soul and Body II 6 2 :

synnum gesargad, 31

(Solode cer fela

ond eft sona from 5e

By itself: Riddle 71 7 b , Psalm 69.1 2 b , Psalm 77.59 2 b , Psalm 90.15 2 b , Psalm 118.71 2 b , The Metres of Boethius 13 3 l b and Fragment of Psalm 69.1 2 b ; w i t h a following verb: Andreas 1111b; once t h e order is reversed: Psalm 118.132 l b . In Beowulf 1937 cross alliteration links t h e final verb w i t h t h e a-verse. See above, ch. 4 .

188

Class II compounds Other adverb phrases (a) Adverb + participle: (a-verse: phrase in second lift, with double alliteration) heard her cumen,

sohte holdne wine

(376)

(b-verse: phrase in second lift, without alliteration) )?a waes aefter wiste wop up ahafen

(128)

gomenwudu greted,

gid oft wrecen

(1065)

(b) Adverb + infinitive: (b-verse: phrase in second lift, without alliteration) bealuwa bisigu

bot eft cuman

secean on gesyntum, Weder-Geata leod

(281)

snude eft cuman

(1869)

word ut faran

(2551)

(c) Adverb + adjective: (a-verse: phrase in second lift, with double alliteration) waes se iren(?reat33

aescholt ufan grag; sigel sudan fus.

Hi siS drugon

(330) (1966)

It is likely that the two quasi-compounds in (c) are qualitatively different from the ones with the intensive adverbs switk and ungemete which appear in the b-verse, and that they should be marked as the equivalent of fully stressed compounds. (b-verse: phrase in second lift, without alliteration) hea{h on] healle,

Hygd swide geong (1926; similarly 1743b and 2170b)

wsefre ond waelfus,

wyrd ungemete neah (2420; similarly 2721b and 2728b)

(d) Adverb + adverb: (b-verse: phrase in second lift, without alliteration) billa brogan.

Breca ncefre git

Geat waes glaedmod, gamela Scylding. 33

geong sona to

(583) (1785)

Geat unigmetes wel

See Campbell, Old English Grammar, § 94, n. 1, for stress on ufan grceg.

189

(1792)

The Metrical Grammar of Beowulf sto[n]dan stanbogan, beam ond bryde;

stream ut ponan beah eft ponan

(2545) (2956)

Pope takes 2545b and 2956b as type D 2 and 583b and 1785b as type E. Once again I do net see any strong reason not to assume stress on the first adverb in all of them. I scan them as type D 2 or D 4 . All quasi-compound phrases described above which contain a sentence particle as the second of their constituent elements are class II compounds. Adverb-adjective phrases are also class II compounds with the probable exceptions of ufan grceg (330a) and suSan fus (1966a), which belong to class I.

190

14 The alliterative behaviour of class III compounds and simplexes

CLASS III C O M P O U N D S

Class III compounds may appear without alliteration in the a-verse, and impose no alliterative requirement. That is to say, their alliterative behaviour is indistinguishable from that of any stressed simplex that is not specially marked for alliteration. They include compounds with variably stressed prefixes when the stress is on the base, proper names composed of two elements and at least one reduced stress quasi-compound phrase. Likewise, simplexes with an inflected suffix or other long final syllable plus inflectional ending do not acquire any special alliterative status. They behave like other simplexes and class III compounds. Nouns, adjectives, verbs and adverbs with variably stressed prefixes

At least one prefix in Beowulf twists in two forms, one stressed, the other unstressed, either of which may be employed at the poet's option. This is the prefix un-.l When the prefix un- is stressed, the compound of which it is a part behaves like any other class I compound. However, when the stress is shifted to the base, as I believe must always be the case when the prefix does not alliterate, 2 the alliterative requirement of the unstressed prefixes3 fails and the compound falls into class III. This stress shift occurs at least three times in the a-verse: (Ax)

N

duguS unlytel

A

N

torn unlytel. A

N

gold unrime

See Kendall, 'Prefix un-. un-.

2

Dena ond Wedera

(498)

Paet waes tacen sweotol

(833)

grimme gecea(po)d

Ibid., p. 46.

191

3

See above, ch. 8.

(3012)

The Metrical Grammar of Beowulf One of the restrictions which the metrical grammar places on class I compounds is that a compound with two consecutive long stressed syllables will not be preceded by a monosyllabic stressed simplex. 4 If we were to assume that un- is always stressed except when alliteration falls on the base, then we would find ten possible violations of this restriction, three in the a-verse5 and seven in the b-verse: (a-verse) torn unlytel.

Paeit waes tacen sweotol

(833)

egl unheoru\

6

aeghwylc gecwaeS

(987)

gold unrime

grimme gecea(po)d

(3012)

(b-verse) wonsceaft wera.

Wiht unhcelo

(120)

leoht unfceger

(727)

slat unwearnum

(741)

aefter deaSdaege

dom unlytel

(885)

wighete Wedra.

Wif unhyre

(2120)

Weard unhiore

(2413)

brun on bane,

bat unswiftor

(2578)

ferhSgeniSlan,

fyr unswidor

(2881)

ligge gelicost slaependne rinc,

wraetta ond wira.

One thing to be noticed about this group is that it is hardly representative of the general run of verses with un- compounds. It is skewed toward the b-verse, although the prefix is evenly distributed between the a-verse and the b-verse in the poem (34:36), and it includes only one case of vocalic alliteration (987a). Furthermore, the seven b-verses seem an unacceptably large number of exceptions to the rule that class I compounds are excluded from the second lift of the b-verse. Of course, if I am right that the prefix un- is not stressed when it does not alliterate, then these anomalies disappear. Assuming that the stress falls on the base when the prefix does not alliterate eliminates all of the exceptions 4 5

6

See above, ch. 12. I do not include the problematic verse {ea]l unhlitme (1129a). In this reconstruction, the alliteration on eal would have to be regarded as extra-metrical. Klaeber emends to egl[u] unheoru.

192

Class HI compounds and simplexes

to the rule excluding compounds from the b-verse and leaves only one potential violation of the prohibition against three consecutive stresses. I have discussed this verse (987a) in ch. 12. When un- is stressed, the compound belongs to class I; when it is unstressed, the compound falls into class III. It seems unlikely that a linguistic phenomenon of this sort would be confined to a single example. No other prefix in Beowulf offers incontrovertible evidence of being variably stressed, 7 but there is at least one other initial morpheme whose alliterative behaviour and positioning in the half-line might be explained on the assumption that it is variably stressed — the form: mon- {man-). The verse in question is this: Swa fela fyrena

feond mancynnes

(164)

If mancynnes were a fully stressed class I compound, it would not only violate the rule excluding such compounds from the second lift of the b-verse, but it would also be an exception to the prohibition against a stressed monosyllablic simplex preceding a compound with two consecutive long, stressed syllables. Words compounded with the variably stressed prefix un- and with mon(if it is in fact variably stressed) behave like fully stressed compounds when the prefix is stressed. The prefix necessarily alliterates and the word may be classed as a class I compound. When the prefix is unstressed, the alliterative requirement is lost. The stem necessarily alliterates only if it occupies the first lift; it may appear without alliteration in the second lift position in the a-verse8 as well as in the b-verse. When the prefix is unstressed, the word may be regarded as a class III compound (that is, the unstressed prefix does not impose the normal alliterative requirement of proclitics).

7

8

The editors have been curiously reluctant to consider the merits of the manuscript reading orfeorme (2385a). See the notes of Klaeber, p. 212, and Chambers, in Wyatt-Chambers, respectively. If it were accepted in the sense 'forsaken' (which Chambers thinks 'does not give very satisfactory sense'), then the prefix or-, like the prefix un-, could be regarded as variably stressed. Moncyn does not appear in the second lift in the a-verse in Beowulf. It appears in that position without alliteration in other Old English poems, however (e.g., wra& money nne {Genesis 1254a), feond moneynnes {Juliana 630a) etc.).

193

The Metrical Grammar of Beowulf Compound proper nouns Proper names made up of two elements which are not semantically distinct do not behave like fully stressed compounds. They appear freely in non-alliterating position both in the a-verse and in the b-verse, as in )?eoden Hrodgar,

J>aet ic J?e sohte

geond widwegas,

(417)

wine min Beowulf

(1704)

The compound Hre&manna is not irregular if it is taken as a name for the Geats with reduced stress (and incidental alliteration) on -manna: (Ax)

N

n

maegen Hredmanna.

Na (?u minne (?earft

(445)

It seems best to treat the tribal name Hea&o-Beardan the same way. 9 Otherwise there would be a class I compound in non-alliterating position in the b-verse in (Nx)

A

Maeg J?ses bonne of)?yncan

n

3eod«£ Heafto-Beardna

(2032)

Compound proper names put no alliterative requirement on a following simplex, for example: A

n

N

Beowulf"Geata, A

n

aer he on bed stige

(676)

N

Hrodgar leofa,

Higelace onsend

(1483)

There is no attempt to avoid three consecutive long syllables, for example: A

HroSgar maj?elode,

A

Beowulf ma[?elode, A

N

n

helm Scyldinga N

(371)

n

beam Ecgpeowes

(529)

An

Oft Scyld Scefing

scea(?ena )?reatum

(4)

The double alliteration in line 4a probably reflects the tendency of family names to alliterate rather than any alliterative requirement of the metrical grammar. There are only three other half-lines in the poem in which a 9

Two other proper names in Heado- are, or appear to be, class I compounds: Heapo-Rcemas and Hea&o-Scilfingas. The distinction probably turns on the question of whether the Beowulf-pott and his audience ever spoke of these peoples as the Beardan, the Rcemas and the Scylfingas as opposed to the Heado-Beardan, the Heapo-Rcemas and the Heado-Scilfingas. In the case of the Scylfingas, we know that they did. The failure of Hea&o-Beardna to alliterate in 2032b suggests that in this case they did not. The evidence for the Heapo-Rcemas is inconclusive (it is the initial lift in 519a).

194

Class III compounds and simplexes

proper name is accompanied by a patronymic. Two are type A; one is a D 1 . All three display double alliteration: Higelac Hrepling,

]>3dr aet ham wunaS

(1923)

HaeScen Hrepling

wi3 Hrefnawudu

(2925)

WulfWonreding

waepne geraehte

(2965)

This evidence must be set against Campbell's statement that inflected forms of compound proper names take half-stress on the second element. 10 Campbell, I take it, would consider the correct scansion of beam Ecgpeowes to be: / / \ x. Bliss, on the other hand, scans this verse: / / x x (1D1 as opposed to 1D2). 11 The alliterative patterns of the poem clearly support (and indeed must have been the evidence for) Bliss's contention. However, I myself am not persuaded that the three classes of compounds which I have described necessarily correlate absolutely and in every case with presence or absence of half-stress, although there certainly must have been broad correspondence between those classes and levels of stress. In this instance, it seems to me quite possible that inflected compound proper names might have been stressed like class I compounds and still have alliterated like class III compounds - the poet's need to use the names overriding the restrictions imposed by their stress contours. One quasi-compound phrase

The adverb + adverb combination nu da appears with the stress on the first syllable: nu-$a.u The two verses in which the phrase appears are both type C: A N

Sing wiS J?yrse.

Ic J>e nu tSa

(426)

A N

5ry(?aern Dena

buton ]>e nu da

(657)

No other class I or class II compound appears in the second lift of a type C. The phrase is best taken as the equivalent of a disyllabic simplex.

10 12

n Old English Grammar, § 8 8 . Metre, Index to the Scansion. If the stress fell on the second element there would be unacceptable double alliteration in the b-verse in the two verses from Beowulf, as there would elsewhere in Old English poetry in such half-lines as preamforprycte, I
195

The Metrical Grammar of Beowulf Nouns, adjectives, verbs and adverbs with derivative suffixes or long final syllables, with the addition of inflectional endings1** All such simplexes can appear in the second lift of either the a-verse or the b-verse without alliteration, and they are not restricted from appearing in any sequence of three long syllables. Some examples are: (a-verse: compound in second lift, without alliteration) Weard mapelode, broSor oSerne

Saer on wicge saet

(286)

blodigan gare

(2440)

(b-verse: compound in second lift, without alliteration) feorhbealo feorran,

fea pingian

(156)

(a-verse: three consecutive long syllables) A

An

wig-iveorpunga,

wordum baedon

(176)

SIMPLEXES

The alliterative requirements of class I and class II compounds are structural — that is, the alliterative requirements are independent of the lexical content of the compounds. Class III compounds and simplexes do not have any structural alliterative requirement beyond the important one that, as stressed elements, they must alliterate when they appear in the first lift. Nevertheless, some of these are specially marked for alliteration in the metrical grammar of Beowulf. Many of these specially marked words are found only in poetry. They form part of an inherited poetic vocabulary that was heavily weighted toward the interests of a 'heroic' society. Dennis Cronan has shown that these words can be grouped into three categories on the basis of their alliterative behaviour: (1) words that always alliterate (and therefore never occur in the second lift of the b-verse); (2) words that alliterate except when they occur in the second lift of the b-verse; (3) words with non-alliterating occurrences in both the second lift of the a-verse and the second lift of the b-verse. l4 That is to say, there are some simplexes 13 14

See Campbell, Old English Grammar, §§ 89-90. 'Alliterative Rank', p. 153. Cronan's study of the poetic vocabulary is based on the entire corpus of Old English poetry.

196

Class III compounds and simplexes

within the total vocabulary of Beowulf which seem to replicate the alliterative behaviour of class I and class II compounds. For example, of the words for 'man', 'warrior', 'retainer' or 'hero' which appear in Beowulf, ten are poetic (in the sense that their use in this meaning is confined to poetry) and seven are non-poetic (in the sense that they occur regularly in this meaning in prose as well as verse). The poetic terms (marked with an asterisk for easy identification) are * beorn, *freca, *guma, *hcelel heeled, *oretta, *rinc, *scealc, *sceotend, *secg and *wzgend {oretta is a

class II compound). The non-poetic terms are cempa, ceorl, eorl, monn, pegn, wer and wiga. By the strict criteria given above, and on the basis of their alliterative behaviour in all surviving Old English poetry, * beorn, *freca, *sceotend, wiga and *wigend could be assigned to the first category (100 per cent alliteration), * oretta, *scealc and *secg to the second category (100 per cent alliteration in the a-verse), and cempa, ceorl, eorl, *guma, *hcele(8), monn, *rinc, pegn and wer to the third category (non-alliteration in the second lift in both the a-verse and the b-verse). 15 In the end, however, these categories cannot be proved to be an aspect of the metrical grammar, since they do not correlate with any metrical or syntactical structure. In the absence of any objective criterion for predicting which words go into which category, we have no way of estimating whether or not more evidence might result in a lower classification. The Beowulf-poet employs the simplex * beorn (biorn) ten times, always with alliteration. When he uses it in the first lift or when he uses it in the second lift under conditions that call for double alliteration, we have no way of judging whether the word itself is specially marked for alliteration. Consider, for example: beorn acwealde hyre beam gewraec, bealoniS biorna; him to bearme cwom

(2121) (2404)

Only once does he put it in the second lift in a metrical—syntactical pattern that does not require double alliteration: blaedfaestne beorn.

Nses Beowulf Saer

(1299)

Was he at this point exercising an option - using *beorn with alliteration where alliteration was not necessary? Or was *beorn so alliteratively charged that it would not have occurred to him to insert it into any half-line pattern 15

Alliteration rates for these words and the number of their occurrences in all Old English poetry in this and the following paragraphs are taken from Cronan, 'Poetic Simplexes', Appendix B.

197

The Metrical Grammar of Beowulf without alliteration? The fact that *beorn occurs 128 times in the corpus of Old English poetry and alliterates every single time strongly suggests the latter. But a single counter-example would disprove the thesis. The simplex *secg (not to be confused with secg, 'sword', which is not poetic) appears thirty-six times in Old English poems other than Beowulf, always with alliteration. It alliterates twenty-eight times in Beowulf — including four times in the second lift in the a-verse, where double alliteration is not mandatory: Da waes swigra secg, sunu Ec{g]lafes (980; similarly 1311a, 1379a and 2352a) But it also appears twice in Beowulf in the second lift of the b-verse, without alliteration: meces ecge;

(?aet waes modig secg

Se waes on Sam Sreate

breotteoSa secg

(1812) (2406)

Despite an overall alliteration rate of 97 per cent (64/66), these two verses drop the simplex from category one to category two. What are we to make of this? Was *secg marked for alliteration, but less strongly than *beorn? Did the Beowulf-poet deliberately violate the alliterative requirements of his metrical grammar in 1812b and 2406b? Finally, we might compare the non-poetic simplexes ceorl and eorl. Both are in category III. Despite the fact that 'churls' and 'earls' fight side by side in The Battle of Maldon, their alliterative rankings seem to mirror their social distance from each other. Eorl has nearly the same alliterative rate in Old English poetry (186/195: 95 per cent) as *secg. In Beowulf its alliteration rate is one hundred per cent (45/45). Three times it appears in the second lift in the a-verse under conditions where double alliteration is not mandatory: unfaegne eorl,

J?onne his ellen deah (573; similarly 1228a and 2695a)

Does eorl belong in category I in the metrical grammar of Beowulf but in category III elsewhere? Or is it pure chance that the Beowulf-poet never used it without alliteration in either the second lift of the a-verse or of the b-verse? Ceorl by contrast has an overall alliteration rate in Old English poetry of 23 per cent (3/13). It never appears in the first lift of an a-verse, and only three times in the first lift in the b-verse. In Beowulf its alliterative 198

Class III compounds and simplexes rate is 0 per cent (0/6). It fails to alliterate even when it refers to the king of the Swedes, Ongentheow: Ne meahte se snella sunu Wonredes ealdum ceorle ondslyht giofan . . . 1 6

(2971-2)

There does not appear to be any satisfactory way of answering these questions. Clearly, simplexes are differentially marked for alliteration. Clearly, poetic simplexes are more likely than non-poetic simplexes to have a high alliterative rate. But it is impossible to assign words to one of the three categories with any degree of confidence. The best that can be said is that the rate of alliteration of words which occur frequently in Old English poetry is an important indicator of their poetic worth in the tradition, and that the descending scale of alliterative rates of simplexes mirrors, imperfectly, the threefold classification of compounds. As Cronan concludes: In Old English the distinction between words of high alliterative rank and those of low rank is not always clear-cut. The alliteration rate of many words is, strictly speaking, neither high nor low, but simply average. Despite this gradation of rank, the connection between a high rate of alliteration and words which are marked by some other characteristic — restriction to poetry, possible descent from a common Germanic poetic tradition, use in a figurative or extended meaning — is consistent. This connection is not absolute, for there are words found only in poetry which have a low alliteration rate, and words that occur in both poetry and prose that have a high rate. But the consistency of the connection is sufficient to demonstrate that in Old English poetry a high alliteration rate is a reflection of a stylistic quality inherent in certain words. This knowledge enables us to use alliterative frequency to rank synonyms according to their poetic worth, and also to define more precisely the poetic vocabulary.17 With the poetic vocabulary and the relative alliterative ranking of simplexes we have moved out of the domain of the metrical grammar and into the area of stylistics. 16

17

'Nor could the bold son of Wonred [Wulf} give a return blow to the old man [Ongentheow}.' Curiously (?), if ceorle were thought to be a scribal error for eorle, because of Ongentheow's rank (cf. eorl Ongenpio (295 la)), eorle would alliterate, as it always does in Beowulf. 'Alliterative Rank', pp. 154-5.

199

15 Metrical typology and the metrical grammar

Students of Old English metrics are well aware of the importance of grammar for understanding the nature of metre. The systematic analyses of Sievers, Pope and Bliss, however much they differ in emphasis and detail from each other, are based on careful investigations of the patterns of stress and syntax revealed by Old English verse. Their work has established beyond a doubt the general validity of the metrical classifications which are usually identified by the Sievers types (Bliss's typology, which is derived from Sievers's, has been gaining in acceptance), even though the significance of the classifications, especially for performance, remains very much an open question. Nevertheless the Sievers typology is limited in its usefulness by its rather too narrow focus on metre, which has tended to mask some of the evidence of the metrical grammar. The study of the metrical grammar which is the subject of this book puts this evidence in a new light. The level of abstraction represented by the five types is misleading when, for example, it leads to the quite valid generalization that the 'types may occur in any combination or sequence'. l When properly analysed, the metrical types are not completely interchangeable. And I do not refer merely to the well-known fact that double alliteration limits any half-line to the a-verse. The onset of the half-line determines which of the three basic types it belongs to. If the onset is weak and contains a sentence particle, the verse is clause-initial (I). If the onset is proclitic, the verse is clause-non-initial (II). If the onset is strong, the verse is clause-unrestricted (III). 2 A metrical 1 2

Pearsall, Poetry, p. 15 There is a kind of verse with a strong onset which is followed by a sentence particle in the second allowable position. There are few of these in Beowulf"and I have treated them as a variety of the clause-initial (I) half-line (see ch. 4). I suspect that if more material were

200

Metrical typology and the metrical grammar

typology ought to be based on this threefold distinction. It is misleading to rely upon one metrical contour - one of the Sievers types - by itself to describe verses which belong to different basic types (I, II or III). The metrical grammar generates these three basic types. Each of the types is then differentiated into specific metrical contours, which are subtypes of I, II or III. The Sievers types (A (which includes A 3 ), B, C, D and E) should be reinterpreted as reflexes of these basic types. Type A 3 has nothing to do with type A; it is related to B and C. A 3 , B and C are labels for metrical contours which begin with a weak onset. One disadvantage of the Sievers typology is that it does not distinguish between the basic types I and II. Since we cannot be sure what practical effect an undisplaced sentence particle had on the rhythm of the verse, it may seem convenient to lump onsets with sentence particles together with proclitic onsets. It is certainly interesting that at this level of abstraction, where the onset is simply 'weak', the same metrical contours turn up in type I and type II. Yet there must have been a difference between the feel or weight of a type I and a type II verse that prevented them from being confounded. In hypothesizing a rhythmical distinction between type B and C verses with 'anacrusis' and B and C verses with initial weak syllables filling all or part of the first measure, Pope comes very close to making this difference part of his theory of rhythm. 3 Although he mixes type I and type II verses in both categories, almost all of the verses he chooses to read with 'anacrusis' have proclitic onsets, and he puts most onsets with sentence particles into the first measure. It would be instructive to see if his theory could accommodate a strict separation of the two onsets, which would have the advantage of providing an objective criterion by which 'anacrusis', in the sense that he uses it here, could be identified.

3

available by way of evidence it would be useful to treat these as a separate type. A large majority of type IJI half-lines in Beowulf are not initial in their verse clause. This suggests the possibility that at some earlier stage the traditional metrical grammar may have displayed four basic types - two with weak onsets and two with strong onsets - and that type III, like type II, was clause-non-initial, while the fourth type, like type I, was clause-initial. Alternatively, the Beowulf-poet himself may have intuitively grasped a potential for this kind of distinction in his metrical grammar, and moved in that direction. Other poets made more use of half-lines with a sentence particle in the second allowable position. They are prominent, for example, in The Wife's Lament and Wulfand Eadwacer (see ch. 16). Rhythm, pp. 38-79.

201

The Metrical Grammar of Beowulf In any case, metrical contours that belong to type I should be distinguished from those that belong to type II. In the Appendix (pp. 219-20), I use a chart to make the relationship between the Sievers types and the basic clausal types I, II and III clear. The basic type I metrical contours are A 3 (la), B (I/la) and C (I/la). The basic type II contours are B (II) and C (II). The basic type III contours are A (III), D (III) and E (III). When there is only one sentence particle in the weak onset, and the particle alliterates, a sub variety of type I is created - the clause-initial, displaceable half-line (Ib) which is restricted to the a-verse. Another variety of type I occurs when a sentence particle goes into the second allowable position. Type I must then be understood to be the category of half-lines with one or more undisplaced sentence particles in either the first or the second allowable positions and to have as its principal subtypes A 3 (Ia/Ib), B (I/Ia/Ib) and C (I/Ia/Ib), with the sentence particles in the first allowable position (the weak onset) and to have as a secondary subtype A (Ic), with the sentence particle (never more than one) in the second allowable position. The notion of'anacrusis' is one which comes up repeatedly in discussions of Old English metre. The word is customarily used in two different senses. In a theory of performance it can refer to an upbeat before a rhythmical measure. In a theory of metre it refers to one or more preliminary, unmeasured syllables before a regular metrical contour. When Pope speaks of 'anacrusis' in a type B or C verse, he uses the word in the former sense. This conception of anacrusis is necessary to his theory of performance, with which I am not here concerned. The other sense is the more common one, and it has been a source of confusion. When prosodists speak of anacrusis they have in mind a subset of type A or D verses with extra unstressed syllables before the first stress. 4 Since the extra syllables form a weak onset, the classification implies that types A and D can be either clause-initial (I) or clause-non-initial (II), as well as unrestricted (III). This classification must be rejected, because it completely obscures the essential differences that mark the three types. As I have argued in ch. 4, many of the verses in question are simply mislabelled types A 3 , B or C. When these are removed from consideration, there remain in Beowulf nineteen verses at most that could be said to display

4

See the important discussion in Cable, Meter and Melody, pp. 32-44.

202

Metrical typology and the metrical grammar

anacrusis.5 Four of these verses (1504b, 1773b, 1877b and 2247b) have the proclitic ne in the second of three weak metrical positions; three more (93b, 666b and 1223b) have a vocalic prefix {be- or ge-) in the same position. These may best be regarded as expanded type C (C*). 6 The problem in the half-line in Caines cynne (107a) may result from the difficulties attendant on inserting an unusual proper name into a traditional half-line formulaic system. The form Caines itself is altered in the manuscript from cames, which testifies to some confusion about the line. In the half-line in mcegpa gehwcere (25a), the form gehtvcere is a later, analogical formation' for the regular dative feminine singular gehwcemP Editors frequently emend to the latter form, which gives a type B. 8 One verse (2592b) involves a violation of Kuhn's first law. Klaeber prints the passage as follows: t?set 6a aglaecean

Naes 3a long to Son, hy eft gemetton.9

(2591b—92)

The sentence particle hy has been displaced from its normal'position in the initial dip (with/?<#/ in 2592a 10 ), but it does not receive alliterating stress (nor can it be scanned as the last word of 2592a, 11 if aglcecean is a fully stressed class I compound 12 ). Whatever the solutions to these problems (and others) may be, I do not intend to suggest that all of the half-lines in this group must be blunders of one kind or another. One of the ways that Middle English alliterative verse differs from Old English is in its preference for metrical patterns with a 'rising-falling rhythm' which appear to be combinations of types B and C. 13 Although the Beowulf-pott routinely avoided these rhythms, the few that he admitted into the poem (to the extent that they are authentic) may 5

6 7

8 9 10

11 13

25a, 107a, 758a, 1248a, 1537a, 1549a, 1751a, 1767a, 2093a and 2297a; 93b, 402b, 666b, 1223b, 1504b, 1773b, 1877b, 2247b and 2592b. I do not include the commonly accepted hypermetric verses, 1163a, 1164a, 1166a and 1705a; 1163b, 1164b, 1165b, 1166b, 1167b, 1168b, 1705b, 1706b, 1707b, 2995b and 2996b. See above, ch. 9. Klaeber, Appendix III, § 11 (however, Klaeber says that 25a is 'metrically above criticism'). On 25a and 107a, see also ch. 6. 'Then it was not long until the terrible adversaries faced each ether again.' Holthausen, review of Holder, Beowulf, 61, proposed restoring it to this position. So, Holthausen, Beowulf, 7th and 8th ed., Sedgefield, 3rd ed. 12 So, Dobbie, ASPR 4. See above, ch. 13. See Borroff, Sir Gawain, pp. 183-9 and nn. 18, 19 and 30.

203

The Metrical Grammar of Beowulf be early manifestations of this development. They have the same anomalous standing in his metrical grammar as the handful of 'short' A 3 . The essential point, however, is that their affinities, both with respect to their onsets and their overall syntax, are with types B and C rather than A and D . l 4 Nothing is gained by pasting the label 'anacrusis' on their onsets and associating them with unrelated metrical types. All subvarieties of metrical types with anacrusis can be eliminated from the typology of metrical contours in Beowulf.15 Another consequence of this study for the classification of metrical contours concerns the percentage of so-called 'expanded' verses to be found in the poem. Expanded verses are type III half-lines which have what would appear to be an 'extra' weak metrical position after the first lift. 16 They are for the most part classified as varieties of type D (D*). There can be no doubt that expanded verses constitute regular and legitimate subtypes. These are half-lines like Ax

A N x

side saenaessas; A

x

A

N

J>a wses sund liden

rondas regnhearde A

n

wiS (?ass recedes weal

(326)

- J?set waes Wendla leod

(348)

|?eodnes gefegon

(1627)

J?egne gesealde

(2810)

(NxXn)x

Wulfgar mat>elode A

n

A

x

N

SrySlic |?egna heap, Ax

(223)

x

A

N x

pioden J?risthydig,

However, many half-lines which are commonly said to be expanded are in reality type I verses with a single alliterating sentence particle in their weak onset. Reclassifying these substantially reduces the number of expanded half-lines. Where, for example, Pope counts 208 expanded type D, not including those with 'anacrusis', 17 I count 143, a reduction of more than 30 per cent. Some of these are susceptible of alternative scansions that would, if adopted, reduce this number still further. 18 14 15 16

17 18

See above, ch. 6. Hoover, Theory of Old English Meter, pp. 138-44, comes to a similar conclusion. They are 'extra' only if one accepts Sievers's theory that the normal half-line in Old English contains four metrical positions. Rhythm, Catalogue of Rhythmic Variations. The 143, including those with alternative scansions, are: (expanded types D* 1 and D*2) 103a, 223a, 232a, 263a, 275a, 326a, 348a, 360a, 371a, 392a, 405a, 456a, 499a, 502a, 529a, 596a, 606a, 631a, 689a, 768a, 769a, 770a, 913a, 925a, 957a, 986a,

204

Metrical typology and the metrical grammar

With the typology which I propose, it is no longer true to say that the various metrical types are completely interchangeable. A type B (II) cannot be replaced by a type B (I) and vice versa. Though a type A (III) could replace either; a type B (I) can replace a type A (III) only if the type A (III) happens to stand at the head of a verse clause, and a type B (II) can replace a type A (III) only if it does not, and so forth. These are important facts which must be recognized. The variety of stress patterns within the three basic types (I, II and III) is nearly as great as the variety between and among the Sievers types A, B, C, D and E. If we are justified in speaking of the 'metre' of Beowulf, it is because the half-lines of the poem are precisely measured. The number of metrical contours that are acceptable is large if subvarieties are included in the count. 19 But it is not unlimited, and the contours are not those that inevitably derive from any selection of phrases in a representative passage of Old English prose. 20 Most, but not all, prosodists explicitly or implicitly impose some theory of isochronic rhythm upon the metrical contours. Such theories can neither be proved nor disproved. What is certain is that the formal art of the Beowulf-poet as reflected in his metrical grammar exhibits constant variation within strictly defined limits. This applies both to metre and to alliteration which must finally be regarded as separate, but interdependent, systems, mutually structured on the verse syntax generated by the metrical grammar.

19

20

990a, 1002a, 1096a, 1097a, 1154a, 1157a, 1162a, 1212a, 1215a, 1231a, 1298a, 1321a, 1339a, 1351a, 1383a, 1410a, 1426a, 1440a, 1454a, 1468a, 1473a, 1505a, 1532a, 1565a, 1568a, 1651a, 1668a, 1675a, 1678a, 1687a, 1710a, 1727a, 1790a, 1793a, 1817a, 1840a, 1863a, 1865a, 1871a, 1874a, 1886a, 1890a, 1969a, 1999a, 2079a, 2123a, 2136a, 2173a, 2205a, 2396a, 2425a, 2440a, 2442a, 2443a, 2462a, 2471a, 2496a, 2510a, 2562a, 2603a, 2613a, 2631a, 2646a, 2648a, 2667a, 2674a, 2689a, 2710a, 2719a, 2724a, 2725a, 2734a, 2749a, 2751a, 2760a, 2800a, 2810a, 2811a, 2819a, 2825a, 2829a, 2847a, 2862a, 2931a, 3045a, 3055a, 3076a and 3099a; 53b, 840b, 1125b, 1573b, 1663b, 1840b, 1869b, 2020b, 2032b, 2432b, 2671b and 3032b; (expanded type D*4) 400a, 938a, 1023a, 1359a, 1420a, 1447a, 1617a, 1627a, 1679a, 1729a, 1757a, 2286a, 2451a, 2478a, 2687a and 2964a; 2717b. Two verses could, very doubtfully, be regarded as expanded type E: stitSra ncegla gehwylc (985a) and Fepa eal gescet (1424b). I class 194 la, very doubtfully, as hypermetric. See the introduction to the Index of Alliteration, Scansion and Metrical Clause Structure (pp. 233-4), which lists twenty-eight. Hypermetric lines and anomalous (F) patterns are not included in this count. Despite the impression given by D a u n t , ' O l d English Verse', esp. p p . 6 2 - 3 . See the reply of Cable, Meter and Melody, p . 1 1 .

205

16 Conclusions

I began with the assumption that the half-lines of Beowulfare 'formulaic', not in the sense that any particular half-line must have been traditional or a variant of a traditional verse, though many of them are, but rather in the sense that the alliteration, the metre and the syntax of the half-lines are interrelated in precise and predictable ways which form recurring patterns. This study has both confirmed and given new precision to that assumption. The half-lines are formulaic because they are the product of the poet's metrical grammar. The Beowulf-poet did not impose metre or alliteration on a phrase he wished to use, nor did he distort or rearrange the phrase in order to secure a particular metrical contour or alliterative pattern. He composed with the half-lines which his metrical grammar provided (even if the half-line was entirely his creation), and which came already marked for metre and alliteration. The exercise of his genius came in his choice or creation of half-lines which conformed to his metrical grammar and the way he linked them together. Whether art is conservative or iconoclastic, all artists inherit a tradition which they interpret and express in their individual manners. I do not want to leave the impression, which I may have given, that the metrical grammar was some kind of invisible book of unchanging rules which the Beowulf-poet received from the tradition and over which he had no control. He must have shaped his metrical grammar in the act of composition to what he felt were the standards of the tradition. The tradition expressed an aesthetic which he made his own. What I have described as 'rules' or 'principles' must have been for him matters of taste and propriety — his realizations of the possibilities of the poetic tradition. The formal properties to which he obviously responded in poetry and which he developed out of the potential of his metrical grammar imposed 206

Conclusions

tight constraints on the shape of the half-line. Working within these constraints, he generated intricately linked and endlessly varied, but nevertheless predictably regular, patterns of sound, rhythm and syntax. As a result we can specify which half-lines are clause-initial (I), which are clause-non-initial (II) and which are unrestricted (III). We can also specify with some precision which half-lines belong to the a-verse, which to the b-verse, and which may be placed in either. The following observations sum up the main points which have emerged from this study. The onset of the half-line determines whether it is initial (I), non-initial (II) or unrestricted (III). A strong onset (that is, the absence of any unstressed syllables before the first lift) marks an unrestricted half-line which may be placed anywhere in the verse clause. A proclitic onset (that is, a weak onset without any sentence particles) marks a non-initial half-line which may not be placed at the beginning of a verse clause. A weak onset with one or more sentence particles marks an initial half-line which can only be placed at the beginning of a verse clause. Two qualifications to these statements must be made. First, a weak onset with a single alliterating sentence particle marks a half-line (Ib) which may be displaced from the initial position in the verse clause. Second, a strong onset which is immediately followed by a sentence particle in the first weak metrical position marks a half-line (Ic) which is restricted to the beginning of the verse clause. All these features are marked in the Index of Alliteration, Scansion and Metrical Clause Structure. All half-lines which display double alliteration are of course restricted to the a-verse. Double alliteration may take the form AA (two alliterating lifts) or aA (extra-metrical alliteration + alliteration on the first lift). When double alliteration takes the form aA, alliteration on the second lift is prohibited. Double alliteration in the form aA is a necessary condition for external displacement, and in this sense is mandatory when the half-line has been displaced. All such displaced half-lines can be recognized in the Index by the absence of the symbol (11) before an alliterating sentence particle (a). Initial dips which constitute entire half-lines and are scanned by the transformation rule are restricted to the a-verse, whether or not they display double alliteration. These can be recognized in the Index by the underlining of the entire half-line and the absence of the symbol (11) within the half-line.

207

The Metrical Grammar of Beowulf / Conditions for mandatory double alliteration {and therefore restriction of the half-line to the a-verse)

i Class I compounds must alliterate (class I compounds are therefore excluded from the second lift in the b-verse). A class I compound in the second lift generates mandatory double alliteration, and the half-line is restricted to the a-verse. ii Class I compounds impose an alliterative requirement on a following lift in the a-verse which results in mandatory double alliteration, provided the second lift is not in the final X-position. When a class I compound is followed by a non-alliterating second lift which is not in the final X-position the half-line is restricted to the b-verse (see below, II. i). Hi Class II compounds must alliterate in the a-verse when the stressed syllable of the compound is not in the final X-position. Therefore double alliteration is mandatory when the compound occupies a second lift which is not in the final X-position. When a non-alliterating class II compound occupies a second lift which is not in the final X-position, the half-line is restricted to the b-verse (see below, II. ii). iv When a class II compound composed of proclitic + simplex occupies the second lift in the a-verse, (double) alliteration is mandatory if the proclitic is in the w l , even if the second lift is in the final X-position. When such a compound does not alliterate under these conditions, the half-line is restricted to the b-verse (see below, II. Hi). (This condition partially overlaps I. iii.) v When the leftmost natural stressed element occupies the non-final X-position in the a-verse, and the initial lift is filled by the insertion rule, double alliteration is mandatory. When the leftmost natural stressed element occupies the non-final X-position and does not alliterate, and the initial lift is filled by the insertion rule, the half-line is restricted to the b-verse (see below, II. v). vi In half-lines with an internal clause juncture followed by a clause-initial segment which acquires metrical stress by the transformational rule, placement in the a-verse is mandatory when the first lift follows the juncture (i.e., the absence of a stressed element before the clause juncture marks the half-line as an a-verse). l When the first lift precedes the juncture, the half-line is restricted to the b-verse (see below, II. vi). 1

This condition is in a sense vacuous, because the seven a-verses to which it applies (272a, 442a, 731a, 1859a, 2158a, 3096a and 3180a) are all type A 3 . I keep it, both because it confirms the restriction of A 3 to the a-verse, and because it reflects what I take to be the

208

Conclusions II Conditions for mandatory placement of the half-line in the b-verse

i When a class I compound is followed by a non-alliterating second lift which is not in the final X-position the half-line is restricted to the b-verse. ii When a non-alliterating class II compound occupies a second lift which is not in the final X-position, the half-line is restricted to the b-verse. iii When a class II compound composed of proclitic + simplex does not alliterate when the proclitic is in the w 1, the half-line is restricted to the b-verse. iv When the initial lift is supplied by a stressed sentence particle, the half-line is restricted to the b-verse. v When the leftmost natural stressed element occupies the non-final X-position and does not alliterate, and the initial lift is filled by the insertion rule, the half-line is restricted to the b-verse. vi In half-lines with an internal clause juncture followed by a clauseinitial segment which acquires metrical stress by the transformational rule, placement in the b-verse is mandatory when the first lift precedes the juncture (i.e., a stressed element before the clause juncture marks the half-line as a b-verse). Each of these conditions is marked in the Index. The Beowulf-poet's metrical grammar is an instrument of expression which he inherited from the Anglo-Saxon poetic tradition, and which he modified and shaped out of his personal taste and experience. In order to reveal the formal regularities to which it gives rise, it has been necessary to compare and analyse large numbers of half-lines. Similar procedures could be used to see to what extent other poems share, or depart from, the features of this particular metrical grammar. A poem of a thousand lines or more would be long enough to permit a useful summary of its metrical grammar to be made. Genesis A from the Junius Manuscript, Andreas and Elene from the Vercelli Book and Christ and Guthlac from the Exeter Book, all offer sufficient scope for analysis. If the metrical grammars of these poems were thoroughly described, it might be possible to place shorter poems in some kind of relationship to them. Even assuming, as I think is evidently the case, that there was a good deal of continuity and overlap among the various individual metrical grammars of the Anglo-Saxon scops, certain features might prove to be essential fact, that the first lift in the b-verse is never stressed by the transformational rule.

209

The Metrical Grammar of Beowulf peculiar to a particular time or region of the country or to one poetic tradition as opposed to another. Some genres, like the elegies of the Exeter Book, seem to be products of markedly different traditions from that represented by Beowulf. Their metrical grammars may vary accordingly. An intriguing possibility in my judgement is that there could have been an independent tradition of women's poetry which is not adequately represented among the poems which happen to survive on parchment. Do we have any women's poems? The veil of anonymity conceals the sex along with every other personal characteristic of all Anglo-Saxon poets except Caedmon and Cynewulf. The Wife's Lament and Wulf and Eadwacer, with their female voices and their common Verbal emphasis on suffering, separation, and passive endurance', 2 would be leading candidates in a search for poems which might have been authored by women. 3 Let us look at them with this possibility in mind. Both The Wife's Lament and Wulf and Eadwacer happen to display an unusually low percentage of type A verses (21 per cent and 20.6 per cent, respectively, compared to the 36.6 per cent of type A verses in Beowulf, which may be taken as typical). 4 The total number of lines involved is only seventy-two - too few to attach any great statistical importance to these percentages, much less to guess what they might signify about the metrical grammars of the poems. Still it may be worth noting that the percentage of type A verses in The Husband's Message, which is often taken as a companion piece to The Wife's Lament, is approximately the same as in Beowulf. Both The Wife's Lament and Wulf and Eadwacer exhibit half-lines which seem irregular if they are judged by the standards of Beowulf. The Wife's Lament begins: Ic pis giedd wrece bi me ful geomorre, minre sylfre siS. Ic (?aet secgan mgeg, hwaet ic yrm)?a gebad, si^an ic up weox, niwes o\>\>e ealdes, no ma (?onne nu. A ic wite wonn minra wraecsij?a.5 (1—5) 2 3

4 5

Renoir, 'Reading Context', pp. 2 3 9 ^ 0 . Chance prudently observes that female authorship of these two poems is 'a point which cannot be proven' (Woman as Hero, p. 81). I owe this observation to Professor Dennis Cronan. 'I tell this tale about me, very sorrowful, the adventure of myself. I can say what miseries I experienced, after I grew up, recently and long ago, no more than now. Ever I have gained pain for my journeys into exile.'

210

Conclusions In the first line we find a sentence particle (the prepositional phrase bi me) which has been displaced, but which does not alliterate or take metrical stress. In terms of the metrical grammar of Beowulf\ bi me ful geomorre is a clause-initial (I) half-line which is out of place. It would not be hard to propose a solution. The line could be rearranged to read: #

Ic bi me ful geomorre

\>is giedd wrece

Maeg ic be me sylfum

soSgied wrecan

Compare: (The Seafarer 1)

But it makes perfectly good sense as it is, and when some lines later we come upon a similar construction, again with a displaced but nonalliterating and metrically unstressed sentence particle, and this time one that could not be so easily rearranged, we may begin to suspect that we are not dealing with an error of transmission: wesan geomormod, A scyle geong mon heard heortan gef>oht . . . 6

(42—3a) 3

In the same opening lines of The Wife's Lament we find an A half-line sippan ic up weox (3b) - in the b-verse, and an odd pattern of double (?) alliteration on N in the following b-verse — no ma ponne nu (where we would expect alliteration on M). Similarly in Wulfand Eadwacer we find an uncertain pattern of alliteration (W/HW?) and a displaced but non-alliterating sentence particle (eac) in waes me wyn to J?on,

waes me hwaej?re eac Ia3

(12)

There are other displaced but non-alliterating sentence particles in Sindon waelreowe

weras peer on ige

Uncerne earne hwelp hired wulf to wuda

(6) (16b- 17a)

These examples, which could be multiplied, suggest that both poems might be the products of metrical grammars which were quite different from the Beowulf-poet's.7 It would be instructive to compare them with 6 7

'Ever must a young man be sad in mind, hard must be the thought of his heart Lehmann, 'Metrics and Structure', pp. 152-4, points out the numerous metrical problems in Wulfand Eadwacer. Pearsall, Poetry, pp. 55-6, in distinguishing between the two poems, remarks: 'It is possible that Wulfand Eadwacer, with its irregular alliteration, 211

The Metrical Grammar of Beowulf other poems in the Exeter Book to see whether any of these features contrast against that background as well. Stability and continuity in the poetic tradition, however, are more evident than change and diversity. Two short poems from opposite ends of the Old English period - Caedmon's Hymn, which, as reported by Bede and transcribed in Old English in several versions of Bede's Historia ecclesiastica, must go back to the second half of the seventh century, and the poem Durham, which was written at the beginning of the twelfth century — illustrate the extent to which the metrical grammar of Beowulf may be taken as normative for more than four centuries in Anglo-Saxon England. One Northumbrian version of Caedmon's Hymn reads as follows:8 Nu scylun hergan metudaes maecti uerc uuldurfadur, eci dryctin,

end his modgidanc,

Al (III) : Bl (II)

sue he uundra gihuaes,

B1 (la) : A1 (III)

haleg scepen;

tha middungeard

D41 (III) : B2 (I) Al (III) : Al (III)

aelda barnum

heben til hrofe,

firum foldu, '

A3 (la) : El (III)

or astelidae.

5 He aerist scop

eci dryctin,

hefaenricaes uard,

Al (III) : Al (III)

moncynnaes uard

aefter tiadae

B1 (la) : E1 (III) Al (III) : Al (III)

frea allmectig.9

Al (III) : D 1 (III)

Five of the eighteen half-lines of the poem have weak onsets. There are undisplaced sentence particles in four of them: nu scylun (la), sue he (3b), he (5a) and tha (7a). These four should be (in terms of the metrical grammar of Beowulf) clause-initial (I) half-lines, and, in fact, the editor places each of them at the head of a verse clause. None of the three a-verses has a single alliterating sentence particle, and therefore none is clause-initial, displacesemi-strophic form, end-stopped lines and simple repetitive syntax, represents a relic of a genuine popular form . . . The Wife's Lament, however, seems to be deliberately obscure 8 9

Text from ASPR 6, 105. 'Now we must praise the guardian of the heavenly kingdom, the might of the ruler and the thought of his mind, the work of the glorious father, as he established the beginning of each of wonders, eternal lord. Hefirstcreated for the children of men the sky for a roof, the holy creator; then the guardian of mankind, the eternal lord, afterwards fashioned middle-earth, the land for men, the almighty lord.'

212

Conclusions able (Ib). The other weak onset is proclitic: end his (2b). A proclitic onset should mark the half-line as clause-non-initial (II), which this one is. The remaining thirteen half-lines have strong onsets and are clause-unrestricted (HI). At least five fully stressed class I compounds appear in the Hymn: hefaenricaes (lb), modgidanc (2b), uuldurfadur (3a), middungeard (7a) and money nnces (7b). One other compound might be considered fully stressed: allmectig (9b). All but allmectig alliterate on their first stressed syllable. None of them is followed by a stressed element which is not in the final X-position; therefore none of them imposes an alliterative requirement on another word. The compound uuldurfadur is itself in second position, and, as the alliterative rule of compounds requires, the a-verse in which it appears displays double alliteration (condition I. i) The same rule excludes fully stressed compounds from second position in the b-verse. If allmectig were a fully stressed compound, it would violate this rule. However, forms with this prefix were excluded from the category of fully stressed class I compounds in the metrical grammar of Beowulf.10 There is a verse in Beowulf -which is exactly comparable to line 9b: Mael is me to feran;

Fceder alwalda

{Beowulf 316)

I would argue that allmectig is a class II compound and therefore that the rule does not apply. The only verb which appears in an initial clausal dip is the auxiliary scylun (la). Two finite verbs appear in adverb + verb phrases which take alliterating stress on the adverb and are class II compounds. These are: aerist scop (5a) and after tiadce (8b). Another finite verb has been internally displaced and is part of an object + verb, class II compound: or astelidce (4b). The only preposition in Caedmon's Hymn is til (6a) and it follows the first lift. Since the preposition is in a wl position, by the rule of proclitics in the metrical grammar of Beowulf, its object should display (double) alliteration (condition I. iv), as in fact it does. The metrical grammar also moves stressed elements with an unstressed prefix as far to the right as possible in the half-line. 11 There are two simplexes with unstressed prefixes: gihuaes (3b) and astelidce (4b). Both are in final position. All of the metrical contours of the poem are regular. There are, however, no type C and a rather high percentage of type A (9/18 or 50 per cent). A 10 n See above, ch. 13. See above, ch. 8. 213

The Metrical Grammar of Beowulf poem nine lines in length can provide only the sketchiest idea of the metrical grammar of its poet, but it seems likely that in its main lines it did not differ substantially from that of the Beowulf-poet. Furthermore, none of the numerous variants in the manuscript copies of Caedmon's Hymn significantly affects the description of its metrical grammar which I have given, even though there is evidence that the text was not regarded as fixed and that in the course of transmission it was influenced by oral readings. 12 The distinctive characteristics generated by the metrical grammar may have been a powerful agent of conservation in the fluid conditions of an essentially oral culture. Durham was written some four hundred and fifty years later than Caedmon's Hymn, shortly after the translation of St Cuthbert's relics into the newly constructed Durham Cathedral in 1104. 13 The poem exists in two closely related versions which go back to the end of the twelfth and the beginning of the thirteenth centuries. 14 The poem as printed in ASPR is as follows:15 Is Seos burch breome

geond Breotenrice,

steppa gestaSolad,

stanas ymbutan

wundrum gewaexen.

Weor ymbeornad,

ea ySum stronge, 5 feolafiscakyn

in deope dalum

wudafaestern micel;

13

14

? : Bl (I) Bl (Ib) : ? Bl (la) : El (III) A3 (Ib) : ?

wilda deor monige,

deora ungerim

Is in Sere byri eac

12

A2 (III) : Al (III)

onflodagemonge.

wuniad in Sem wycum

and 5es clene

A2 (III) : A2 (III)

and Ser inne wunaS

And Saer gewexen is

10 Se arfesta

C1 (la) : C1 (II)

bearnum gecySed

B1 (II) : ? 3

A (la) : A2 (III) C1 (II) : Al (III)

eadig Cudberch cyninges heafud,

Osuualdes, Engle leo,

and Aidan biscop,

Eadberch and EadfriS,

aeSele geferes.

Is Ser inne midd heom

^Selwold biscop

A3 (?) : Al (III) ?: ? A2 (III) : A2 (III) B2 (la) : Al (III)

See O'Keeffe, 'Orality and Developing Text', esp. pp. 19-20. I have discussed the date of this poem and the circumstances of its composition in 'Let Us Now Praise', pp. 507-8, and in 'Dry Bones', pp. 6-7 and 11. 15 Dobbie, ASPR 6, xliii- iv. ASPR 6, 27.

214

Conclusions 15 and breoma bocera Beda, 5e clene Cudberte lerde lustum,

on gecheSe

in in 5em minstre

A 3 (Ib) : ? C 2 (Ib) : A2 (III) C 2 (II) : ?

reliquia,

20 Saer monia wundrum gewurSaS, midd Sene drihnes wer

?: ? ?: ?

and he his lara wel genom.

EardiaeS set 3em eadige unarimeda

and Boisil abbot,

Ses 5e writ seggeS,

domes bideS. l6

? : C 1 (I) Bl (II) : A l (III)

Durham is quite similar to Beowulf m its metrical clause structure. Four half-lines, feola fisca kyn (5a), wuniad in dem wycum (7a), lerde lustum (17a) and EardicetS at dem eadige (18a), have extra-metrical alliteration on a single sentence particle (feola; wuniad; lerde; EardiceS) and thus are clause-initial, displaceable (Ib). Lines 5a and 17a have been displaced; 7a and 18a have not. Another nine half-lines, la, 4b, 6a, 9a, 14a, 16a, 17b, 20a and 20b, should be clause-initial, nondisplaceable (la), by virtue of having one or more non-alliterating sentence particles in their weak onsets. All nine do in fact head their clauses. Eleven half-lines, Ib, 5b, 8a, 10a, l l a , 12b, 15a, 15b, 16b, 19a and 21a, have proclitic onsets. Most of these are unambiguously non-initial (II), but several require a closer look. One apparent Violation' of the metrical clause structure occurs in the last line where a clause-non-initial (II) half-line, midd dene drihnes wer (2 la), heads a clause. But this violation may be only the result of the editor's punctuation. I see no reason why a full stop could not be placed after wer. Then the last two lines might be translated: 'where many things won16

'This borough is famous throughout Britain, steeply established, stones round about wondrously grown. The Wear runs around it, the stream, strong with waves, and therein dwell the kin of many fish in the mingling of the flood. And there has grown there a great secure wood; there dwell in the place many wild animals, in deep dales a countless number of animals. There is in the city also, as is known to the sons, the virtuous blessed Cuthbert and the clean king's head, Oswald's, the protector of the English, and Bishop Aidan, Eadbert and Eadfrith, the noble companions. With them there within is i€thelwold the bishop, and the famous bookman Bede, and Boisil the abbot, who gladly taught clean Cuthbert in youth [?], and he took well his teachings. There dwell beside the blessed one within in the minster a countless number of relics, where many things miraculously come to pass, as the book says, [and] with the man of God await the Judgement.' 215

The Metrical Grammar of Beowulf drously come to pass, as writ says, by means of [midd] the man of God. It [the city?} awaits the Day of Judgement.' 17 The half-line anddes dene (I la.) appears to scan as a type A 3 , even though it lacks a sentence particle in its w l and is non-initial (II). In the metrical grammar of Beowulf type A 3 is exclusively clause-initial (I). This use of the A 3 would have seemed strange to the Beowulf-poet.18 Another problem is presented by the apparently corrupt half-line: on geche&e (16b). Dobbie interprets this as 'evidently a late spelling of geogo&e.19 If so, the problem is manifold. Alliteration breaks down; an A 3 appears in the b-verse; and the A 3 lacks a sentence particle (and a syllable) in its w l . Holthausen proposed the emendation to cildhade, 'for metrical reasons'. 20 This would eliminate the problem, although it is paleographically somewhat extreme and does not correspond to what was known in twelfth-century Durham of Cuthbert's life. 21 The Durham-poet may have had an alliterative rule for proclitics in his metrical grammar rather like the Beowulf-poet's. There are several halflines in Durham with a proclitic in a w l which is followed by a second lift. 22 All those in which the second lift does not alliterate are b-verses (2b, 3b, 5b, 8b (?), 9b, 13b and 17b (?)) (condition II. iii). Every proclitic in a w l in the a-verse is followed by an alliterating lift. This results in mandatory double alliteration in lines 2a, 3a, 13a and also in 20a, if, as I think, wundrum gewurdad is a separate half-line (condition I. iv). There is one a-verse (14a) where a proclitic occupies a w2, and in this case the following lift does not alliterate. As in Beowulf all forms with an unstressed prefix have moved as far to the right in the half-line as possible, with the exception of gewexen in And (Seer gewexen is (6a) where there is a displaced sentence particle (is) further to the right. This was one of the regular exceptions in the metrical grammar of Beowulf Compare swa him gecynde wees (Beowulf 2 6 9 6 b ) . 2 3

There are proportionally more weakly stressed proclitic demonstratives in Durham (eight in twenty-one lines) than in Beowulf (e.g., nine in the first hundred lines, counting the reading of the manuscript, para, in 9b) or, for 17

Cf. the translation in A S P R 6, 1 5 3 .

18

T h e nearest parallel would be buton pone hafelan ( I 6 l 4 a ) ; see ch. 3 .

19

A S P R 6, 152.

21

Bede reports in his prose Vita S. Cuthberti, ch. 6, t h a t C u t h b e r t was a g r o w n m a n w h e n

20

22

See above, ch. 10.

' Z u altenglischen D i c h t u n g e n ' , p . 2 9 .

he entered the monastery and received instruction from Boisil. 23

See above, ch. 8.

216

Conclusions

that matter, in Caedmon's Hymn (none in nine lines). In the absence of any other evidence for its date, this might lead us to suspect that Durham was written late in the Old English period. With respect to the placement and alliterative requirement of these demonstratives, the rules of the Durhampott's metrical grammar differed in some particulars from the Beowulfpott's. In line 18b the demonstrative &em does not occupy the w l before the first lift. And in line 21a, where the base word of the demonstrative dene has been intercepted by the possessive drihnes, the base word fails to alliterate. The comparative absence of fully stressed compounds in Durham is noteworthy. There are only three: wudafastern (6b), wilda deor (7b — assuming that the poet intended wildeor) and arfesta (10a). Since all three occupy an initial lift, there is no way of knowing what the alliterative requirement, if any, of fully stressed compounds might have been in the Durham-pott's metrical grammar. The number of verses in Durham whose metrical contours are doubtful is too high (14 : 42 or 33 per cent) to be explained entirely on grounds of textual corruption. 24 Some of these verses may be hypermetric. Then, too, five of them contain proper names (12a, 12b, 15a, 15b and 16a), which the poet may have felt he had to fit into the traditional verse patterns as best he could. Nevertheless it seems evident that the old Germanic metre no longer held absolute sway. Certainly the Durham-pott was comfortable with metrical contours, especially ones with a 'rising-falling' rhythm, that would have seemed unacceptable to the Beowulf-pott. In general, Durham is sufficiently close to the norms of the metrical grammar of Beowulf to show that it was composed by a poet who had inherited or was familiar with the old Anglo-Saxon poetic techniques, even though there are aspects of his metrical grammar that suggest change or innovation. Since Durham must have been written in Northumbria, presumably at Durham itself, it testifies to the continuance in the north of England of the ancient poetic traditions right up to the beginning of the twelfth century. These few examples can do no more than suggest the complexity of the task of differentiating the metrical grammar of the tradition from the metrical grammars of the various poets who worked within it. The poems which have survived from the Old English period are not the homogeneous 24

Cable puts the percentage of verses 'falling out of the classical patterns' at 36 per cent. See his discussion in 'Metrical Style', p. 79.

217

The Metrical Grammar of Beowulf products of a tradition which never varied. It is important to respect the differences as well as the similarities in their metrical grammars. The assumption that any deviation from the norm of Beowulf indicates some kind of rot in the Anglo-Saxon poetic tradition can safely be discarded. If this study of the metrical grammar o£ Beowulf contributes finally to a better understanding of the rich particularities of that tradition, it will have achieved its purpose.

218

APPENDIX

A typology of the metre of Beowulf in relation to the metrical grammar

Sievers Types:

A°

Clausal Types I (b-verse only)

B

c

X

X

la (a-verse only)

X

X

X

Ib (a-verse only)

X

X

X

D

E

X

X

X

Ic II

A

X

X X

III

Type I (b-verse only): half-lines with one or more sentence particles in the first allowable position (clause-initial) Type la (a-verse only): at least one non-alliterating sentence particle in the first allowable position (clause-initial, nondisplaceable) Type Ib (a-verse only): single alliterating sentence particle in the first allowable position (clause-initial, displaceable) Type Ic: single non-alliterating sentence particle in the second allowable position (clause-initial, nondisplaceable) Type II: half-lines with proclitic onsets (clause-non-initial) Type III: half-lines with strong onsets and no sentence particles in the second allowable position (clause-unrestricted) 219

Appendix Beyond this level of specification, there is a range of specific metrical contours, whose patterns of single or double alliteration and placement in the a-verse or b-verse or both are determined by the various factors (e.g., class of compound, location of proclitic etc.) discussed in this study. Hypermetric lines can also be classified as type I, II or III.

220

Glossary of technical terms

Additional alliterative requirement of demonstratives. In addition to following the alliterative rule of proclitics, the weakly stressed demonstrative imposes mandatory alliteration on its underlying base word, except where alliteration is blocked because the base word occupies an X-position in the b-verse. See ch. 10. Alliteration. See cross alliteration; extra-metrical alliteration; incidental alliteration; primary alliteration; secondary alliteration; transverse alliteration. Alliterative rule of proclitics. When a proclitic in the w l of the a-verse directly precedes its base word, its base word must alliterate. Alliteration is optional when the proclitic occupies a w2. In the case of the demonstratives, the rule is even stronger: see additional alliterative requirement of demonstratives. See ch. 10. Alliterative rule of stressed elements. In any half-line the stressed syllable of the leftmost natural stressed element (if any) always receives metrical stress. This syllable must alliterate unless it occupies an X-position. A stressed syllable in an X-position in the b-verse cannot alliterate; in the a-verse alliteration is optional in the final X-position, it is mandatory in the non-final X-position. See ch. 5. Base word. A word modified by an adjective or similar form or by a dependent genitive or dative. Basic alliterative rule. At least one and no more than two syllables in the a-verse must alliterate with one and no more than one syllable in the b-verse. See ch. 2. Class I compound. Class I compounds include fully stressed compounds and their alliterative equivalents — compounds with variably stressed prefixes when the stress is on the prefix, compounds with certain suffixes, a few compounds in which the two elements may no longer be semantically distinct and certain quasi-compound phrases. See ch. 12. Class II compound. Class II compounds include compounds with fixed stress prefixes, most compounds in which one or both elements have not retained their full semantic value, compound numerals, pronominal and adverbial compounds in ag-, the pronouns nathwylc and welhwylc, simplexes which are preceded by a

221

Glossary of technical terms proclitic and their alliterative equivalents - compound numerals and other compound phrases in which one or more of the lift positions is occupied by a verb or an adverb. See ch. 13. Class III compound. Class III compounds include compounds with variably stressed prefixes when the stress is on the base, proper names and at least one quasi-compound phrase. See ch. 14. Consonantal prefix. An unstressed prefix which ends in a consonant. The consonantal prefixes impose an alliterative requirement on the forms to which they are attached. See ch. 8. Cross alliteration. Alliteration (B) linking the second lift or a half-lift of the a-verse with the second lift or a half-lift of the b-verse (in the pattern AB:AB). Cross alliteration is marked in the Index of Alliteration, Scansion and Metrical Clause Structure. Detached proclitic. A proclitic which does not have a base word in the same half-line. Seech. 11. Dip. The first string of syllables in the verse clause not containing a stressed element. The dip is located either before or immediately after the first stressed element. See upbeat and weak metrical position. Displacement. The movement of a sentence particle from its normal position in the first dip, or the movement of a proclitic from its normal proclitic position before a base word. See external displacement; internal displacement. External displacement. The movement of a clause-initial, displaceable (Ib) halfline from initial position in the verse clause. See ch. 4. Extra-metrical alliteration. Alliteration on a sentence particle in the a-verse which lies outside (to the left of) the metrical contour of the half-line. Extra-metrical alliteration is functional and follows the basic alliterative rule which limits the number of alliterating syllables in the a-verse to a maximum of two. Extrametrical alliteration is to be distinguished from incidental alliteration. See ch. 3. Final X-position. See X-position. Fully stressed compound. A compound in which both elements retain their full semantic value. Fully stressed compounds are class 1 compounds. Fundamental alliterative principle. In any half-line prior to the insertion of the half-line into the verse clause the stressed syllable of the leftmost stressed element (or of the leftmost sentence particle marked for metrical stress by the transformational rule) always bears the first metrical stress, which is marked by alliteration. See ch. 2. Half-lift. A syllable taking a metrical stress which is subordinate to two other lifts in the same half-line. Head-stave. The first, alliterating, lift of the b-verse. Incidental alliteration. Non-functional alliteration on a sentence particle or proclitic. I regard transverse alliteration as incidental, even though transverse

222

Glossary of technical terms alliteration may link lexically significant parts of speech and be consciously sought for by the poet. Incidental alliteration may be found both in the a-verse and in the b-verse. It is not part of the regular alliterative pattern of the line. Insertion rule. In the absence of a natural stressed element to its left, a proclitic adjective becomes a stressed element when the stressed syllable of its base word occupies an X-position which does not permit primary alliteration. Once insertion takes place the fundamental alliterative principle applies to the proclitic adjective. See ch. 5. Internal displacement. The movement of a sentence particle or proclitic from its normal position in the w l of a half-line to a subsequent lift position in the same half-line. See ch. 4. Lift. A syllable, or two resolved syllables, taking a metrical stress. Metrical contour. One of the acceptable patterns of metrical stress and unstress in the half-line. Metrical stress. One of the stressed positions in a metrical contour. Natural stressed element. See stressed element. Non-final X-position. See X-position. Onset. The first syllable or syllables of a half-line. See proclitic onset, strong onset, weak onset. Primary alliteration. Alliteration generated by the fundamental alliterative principle. The syllable displaying primary alliteration also takes the first metrical stress.

Proclitic. A word or bound morpheme which is not independently stressed and which is tied to a following form. Proclitics include unstressed prefixes, copulative conjunctions, prepositions, proclitic adverbs and instrumentals and proclitic adjectives and pronouns. See chs. 8, 9 and 10. Proclitic onset. A weak onset, consisting entirely of one or more proclitics, of a clause-non-initial half-line. See ch. 6. Quasi-compound. A phrase which, for the purpose of analysis, is treated as the equivalent of a class I or class II compound. Secondary alliteration. Alliteration on the second lift. Sentence particle. Any part of speech which is neither a stressed element nor a proclitic. See proclitic; stressed element. Simplex. An uncompounded stressed element. Stressed element. Natural stressed elements include initially stressed compounds with two fully meaningful elements; nominals (nouns and attributive adjectives, including participles used attributively); initially stressed compound verbs and adverbs; most quasi-compound phrases; and a few other classes and specially marked words. A natural stressed element always takes metrical stress. Other parts of speech and some phrases may become the equivalents of

223

Glossary of technical terms stressed elements by displacement, emphatic stress or insertion into alliterating position in the b-verse, but they must be distinguished from natural stressed elements. Strong onset. An initial lift which is not preceded by any unstressed syllables. Transformational rule. Sentence particles in any clause-initial segment which lacks stressed elements acquire metrical stress from right to left in accordance with the stress and phrase rules of the metrical grammar until the first valid metrical contour emerges. The fundamental alliterative principle will then operate to assign alliteration to the first lift. See chs. 2 and 7 . Transverse alliteration. Incidental alliteration linking an initial sentence particle in the a-verse with the second lift of the b-verse (in the pattern bA:AB). Upbeat. A dip which comes before the first stressed element in the verse clause. Vocalic prefix. An unstressed prefix which ends in a vowel. The alliterative requirement of the vocalic prefixes is weaker than that of the consonantal prefixes. See ch. 8. Weak metrical position. An unstressed position in a metrical contour. A weak metrical position may be filled by one or more syllables. The first weak metrical position (wl) may either precede or follow the first lift. The second weak metrical position (w2) may follow either the first or the second lift. Weak onset. A string of one or more unstressed syllables preceding the initial lift. The string consists of sentence particles and/or proclitics. If the string contains no sentence particles, it is a proclitic onset. X-position. A position in the half-line from which primary alliteration is usually or always excluded. There are two X-positions. (1) The final X-position (from which primary alliteration is always excluded) may be occupied either by a stressed final syllable or by a stressed short penultimate syllable. (2) The non-final X-position (from which primary alliteration is usually excluded) may be occupied either by a stressed long penultimate syllable or by a stressed short antepenultimate syllable. See ch. 5.

224

Bibliography

Alcuin, Epistulae, ed. E. Duemmler, MGH Epist. IV: Epistolae Karolini Aevi II (Berlin, 1895) Andrew, S. O., The Old English Alliterative Measure (Croy'don, 1931) ASPR 1-6, see E. V. K. Dobbie and G. P. Krapp Barney, S. A., Word-Hoard: an Introduction to Old English Vocabulary (New Haven, CT, 1977) Barnouw, A. J., Textkritische Untersuchungen nach dem Gebrauch des bestimmten Artikels und des schwachen Adjectivs in der altenglischen Poesie (Leiden, 1902) Baum, P. F., The Meter of the Beowulf, MP 46 (1948-9), 73-91 and 145-62 Bede, VitaS. Cuthberti, ed. and trans. B. Colgrave, in Two Lives ofSaint Cuthbert: a Life by an Anonymous Monk of Lindisfarne and Bede's Prose Life (Cambridge, 1940) Benson, L. D., 'The Literary Character of Anglo-Saxon Formulaic Poetry', PMLA 81(1966), 334-41 Bessinger, J. B., Jr, ed., A Concordance to Beowulf programmed by P. H. Smith, Jr (Ithaca, NY, 1969) A Concordance to ASPR, programmed by P. H. Smith, Jr (Ithaca, NY, 1978) Bessinger, J. B., Jr, and R. P. Creed, ed., Franciplegius: Medieval and Linguistic Studies in Honor of Francis Peabody Magoun, Jr (New York, 1965) Bliss, A. J., The Metre of Beowulf rev. ed. (Oxford, 1967) 'Auxiliary and Verbal in Beowulf, ASE 9 (1981), 157-82 Borroff, M., Sir Gawain and the Green Knight: a Stylistic and Metrical Study (New Haven, CT, 1962) Bosworth, J., and T. N. Toller, An Anglo-Saxon Dictionary (Oxford, 1898); with T. N. Toller, Supplement (1921); rev. and enlarged by A. Campbell (1972) Brault, G. J., The Song of Roland: an Analytical Edition, 2 vols. (University Park, PA, 1978) Brodeur, A. G., The Art of Beowulf'(Berkeley and Los Angeles, CA, 1959) Brooks, K. R., ed., Andreas and the Fates of the Apostles (Oxford, 1961) 225

Bibliography Brown, P. R., G. R. Crampton and F. C. Robinson, ed., Modes of Interpretation in Old English Literature: Essays in Honour of Stanley B. Greenfield (Toronto, 1986) Brunner, K., Altenglische Grammatik nach der angelsdchsischen Grammatik von Eduard Sievers neubearbeitet (Halle, 1942) Bugge, S., 'Spredte iagttagelser vedkommende de oldengelske digte om Beowulf og Waldere', Tidskrift for Philologi og Pcedagogik 8 (1868-9), 40-78 and 287-307 Burlin, R. B., and E. B. Irving, Jr, ed., Old English Studies in Honour of John C. Pope (Toronto, 1974) Cable, T., The Meter and Melody of Beowulf\ Illinois Studies in Language and Literature 64 (Urbana, IL, 1974) 'Metrical Style as Evidence for the Date ofBeowulf, in The Dating of Beowulf, ed. C. Chase, pp. 77-82 Campbell, A., Old English Grammar (Oxford, 1959) 'The Old English Epic Style', in English and Medieval Studies Presented to J. R. R. Tolkien, ed. N . Davis and C. L. Wrenn, pp. 13—26 'Verse Influences in Old English Prose', in Philological Essays: Studies in Honour of Herbert Dean Meritt, ed. J. L. Rosier, pp. 93-8 Cassidy, F. G., and R. N . Ringler, ed., Bright's Old English Grammar & Reader, 3rded. (New York, 1971) Chance, J., Woman as Hero in Old English Literature (Syracuse, NY, 1986) Chase, C , ed., The Dating of Beowulf"(Toronto, 1981) Clark Hall, J. R., and H. D. Meritt, A Concise Anglo-Saxon Dictionary (Cambridge, 1966) Clemoes, P., 'Style as the Criterion for Dating the Composition ofBeowulf, in The Dating of Beowulf, ed. C. Chase, pp. 173—85 Coetsem, F. van, and H. L. Kufner, ed., Toward a Grammar of Proto-Germanic (Tubingen, 1972) Cosijn, P. J., Aanteekeningen op den Beowulf (Leiden, 1892) Cosmos, S., 'Kuhn's Law and the Unstressed Verbs in Beowulf, Texas Studies in Literature and Language 18 (1976-7), 306-28 Creed, R. P., '". . . wel-hwelc gecwaej? . . . ": The Singer as Architect', Tennessee Studies in Literature 11 (1966), 131-43 'The Beowulf-Poet: Master of Sound-Patterning', in Oral Traditional Literature, ed. J. M. Foley, pp. 194-216 Cronan, D. J., 'Alliterative Rank in Old English Poetry', Studia Neophilologica 58 (1986), 145-58 'Old English Poetic Simplexes' (unpubl. PhD dissertation, University of Minnesota, 1986) Crossley-Holland, K., trans., Beowulf (New York, 1968) 226

Bibliography Curschmann, M., 'Oral Poetry in Mediaeval English, French, and German Literature: some Notes on Recent Research', Speculum 42 (1967), 36-52 Damico, H., Beowulf s Wealhtheowandthe Valkyrie Tradition (Madison, WI, 1984) Daunt, M., 'Old English Verse and English Speech Rhythm', TPS (1946), 56-72 Davis, N . , andC. L. Wrenn, ed., English and Medieval Studies Presented to J. R. R. Tolkien on the Occasion of his Seventieth Birthday (London, 1962) Dobbie, E. V. K., ed., The Anglo-Saxon Minor Poems, ASPR 6 (New York, 1942) Beowulf andJudith, ASPR 4 (New York, 1953) Donaldson, E. T., trans., Beowulf(New York, 1966) Donoghue, D., 'Word Order and Poetic Style: Auxiliary and Verbal in The Metres ofBoethius\ ASE 15 (1986), 167-96 Style in Old English Poetry: the Test of the Auxiliary, Yale Studies in English 196 (New Haven, CT, 1987) Fakundiny, L., 'The Art of Old English Verse Composition', RES, n.s. 21 (1970), 129-42 and 257-66 Foley, J. M., Oral Traditional Literature: a Festschrift for Albert Bates Lord (Columbus, OH, 1980) Fourquet, J., L'Ordre des elements de la phrase en germanique ancien (Paris, 1938) Fry, D. K., 'Old English Formulas and Systems', ES 48 (1967), 193-204 Fry, D. K., ed., The Beowulf Poet: a Collection of Critical Essays (Englewood Cliffs, NJ, 1968) Fulk, R. D., 'Contraction as a Criterion for Dating Old English Verse', JEGP 89 (1990), 1-16 Goldsmith, M. E., The Mode and Meaning of 'Beowulf (London, 1970) Heusler, A., Deutsche Versgeschichte mit Einschluss des altenglischen undaltnordischen Stabreimverses I, Paul's Grundriss der germanischen Philologie 8 (Berlin and Leipzig, 1925) Heyne, M., ed., Beowulf, mit ausfuhrlichem Glossar, 13th ed., rev. L. L. Schiicking (Paderborn, 1929) Holthausen, F., review of Holder, Beowulf, 2nd ed., Literaturblatt fur germanische und romanische Philologie 21 (1900), 60-2 'Zu altenglischen Dichtungen', Beiblatt zur Anglia 31 (1920), 25-32 Altenglisches etymologisches Wbrterbuch, 2nd ed. (Heidelberg, 1963) Holthausen, F., ed., Beowulf nebst dem Finnsburg-Bruchstuck, 2 vols.; 6th ed., pt 1 only (as Beowulf nebst den kleineren Denkmdlern der Heldensage) (Heidelberg, 1929); 7th ed. (Heidelberg, 1938); 8th ed., pt 1 only (Heidelberg, 1948) Hoover, D. L., 'Evidence for Primacy of Alliteration in Old English Metre', ASE 14(1985), 75-96 A New Theory of Old English Meter (New York, 1985) Irving, E. B., Jr, A Reading of Beowulf (New Haven, CT, 1968)

227

Bibliography Kellogg, R. L., 'The South Germanic Oral Tradition', in Franciplegius, ed. J. B. Bessinger, Jr and R. P. Creed, pp. 66-74 Kendall, C. B., T h e Prefix un- and the Metrical Grammar ofBeowulf, ASE 10 (1982), 39-52 'The Metrical Grammar of Beowulf: Displacement', Speculum 58 (1983), 1—30 'Dry Bones in a Cathedral: the Story of the Theft of Bede's Relics and the Translation of Cuthbert into the Cathedral of Durham in 1104', Mediaevalia 10 (1988 (for 1984)), 1-26 'Let Us Now Praise a Famous City: Wordplay in the OE Durham and the Cult of St Cuthbert', JEGP 87 (1988), 507-21 Keyser, S. J., 'Old English Prosody', College English 30 (1969), 331-56 Klaeber, F., ed., Beowulf and The Fight at Finns burg, 3rd ed. with first and second supplements (Boston, 1950) Koskenniemi, I., Repetitive Word Pairs in Old and Early Middle English Prose: Expressions of the Type Whole and Sound and Answered and Said, and Other Parallel Constructions, Annales Universitatis Turkuensis, Series B 107 (Turku, 1968) Krapp, G. P., ed., Thejunius Manuscript, ASPR 1 (New York, 1931) The Paris Psalter and the Meters ofBoethius, ASPR 5 (New York, 1932) The Vercelli Book, ASPR 2 (New York, 1932) Krapp, G. P., and E. V. K. Dobbie, ed., The Exeter Book, ASPR 3 (New York, 1936) Kuhn, H., 'Zur Wortstellung und -betonung im Altgermanischen', BGDSL 57 (1933), 1-109 Lehmann, R. P. M., 'The Metrics and Structure of "Wulf and Eadwacer"', PQ 48 (1969), 151-65 Lehmann, W. P., review of Bliss, The Metre of Beowulf, JEGP 59 (I960), 137-42 'Post-consonantal Imnr and Metrical Practice in Beowulf, in Nordica et Anglica, ed. A. H. Orrick, pp. 148-67 'Proto-Germanic Syntax', in Toward a Grammar of Proto-Germanic, ed. F. van Coetsem and H. L. Kufner, pp. 39-68 Lehmann, W. P., and T. Tabusa, The Alliterations of the Beowulf (Austin, TX, 1958) Leyerle, J., T h e Interlace Structure ofBeowulf', University of Toronto Quarterly 37 (1967), 1-17 Lindelof, U., ed., Der Lambeth-Psalter, 2 vols., Acta Societatis Scientiarum Fennicae 35.1 and 43.3 (Helsinki, 1909-14) Lord, A. B., The Singer of Tales (Cambridge, MA, I960) Lucas, P. J., 'On Some Breaches of Kuhn's Law of Particles and Genesis A 2745', PQ 64 (1985), 386-91 228

Bibliography Magoun, F. P., Jr, 'The Oral-Formulaic Character of Anglo-Saxon Narrative Poetry', Speculum 28 (1953), 446-67 Maling, J. M., 'Sentence Stress in Old English', Linguistic Inquiry 2 (1971), 379-400 Mayr-Harting, H., The Coming of Christianity to Anglo-Saxon England (London, 1972) Meritt, H. D., The Construction &JEO XOLVOIJ in the Germanic Languages, Stanford University Publications in Language and Literature 6.2 (Palo Alto, CA, 1938) Mitchell, B., Old English Syntax, 2 vols. (Oxford, 1985) Nicholson, L. E., and D. W. Frese, ed., Anglo-Saxon Poetry: Essays in Appreciation for John C. McGalliard (Notre Dame, IN, 1975) Niles, J. D., Beowulf: The Poem and Its Tradition (Cambridge, MA, 1983) O'Keeffe, K. O., 'Orality and the Developing Text of Caedmon's Hymn , Speculum 62(1987), 1-20 Opland, J., Anglo-Saxon Oral Poetry: a Study of the Traditions (New Haven, CT, 1980) Orrick, A. H., ed., Nordica et Anglica: Studies in Honor of Stefan Einarsson (The Hague, 1968) Orton, P. R., 'Verbal Apposition, Coordination and Metrical Stress in Old English', NM 86 (1985), 145-58 Pearsall, D., Old English and Middle English Poetry, The Routledge History of English Poetry I (London, 1977) Pope, J. C., The Rhythm of Beowulf: an Interpretation of the Normal and Hypermetric Verse-Forms in Old English Poetry, rev. ed. (New Haven, 1966) Renoir, A., 'A Reading Context for The Wife's Lament', in Anglo-Saxon Poetry: Essays in Appreciation forJohn C. McGalliard, ed. L. E. Nicholson and D. W. Frese, pp. 224-41 Riedinger, A., 'The Old English Formula in Context', Speculum 60 (1985), 294-317 Rieger, M., 'Die alt- und angelsachsische Verskunst', Zeitschrift fur deutsche Philologie 7 (Halle, 1876), 1-64 Rissanen, M., '"Sum" in Old English Poetry', in Modes of Interpretation in Old English Literature, ed. P. R. Brown, et al., pp. 197-225 Robinson, F. C , 'Lexicography and Literary Criticism: a Caveat', in Philological Essays in Honour of Herbert Dean Meritt, ed. J. L. Rosier, pp. 99-110 Beowulf and the Appositive Style (Knoxville, TN, 1985) Rosier, J. L., ed., Philological Essays: Studies in Old and Middle English Language and Literature in Honour of Herbert Dean Meritt, Janua Linguarum series maior 37 (The Hague, 1970) Russom, G., Old English Meter and Linguistic Theory (Cambridge, 1987)

229

Bibliography Schaubert, E. von, ed., Heyne—Scheckings Beowulf, in 3 pts; 17th ed., pts 2 and 3 (Paderborn, 1961); 18th ed., pt 1 only (Paderborn, 1963) Scragg, D. G., ed., The Battle of Maldon (Manchester, 1981) Sedgefield, W. J., ed., Beowulf 2nd ed. (Manchester, 1913); 3rd ed. (Manchester, 1935) Sievers, E., 'Zum Beowulf, BGDSL 9 (1884), 135-44 and 370 'Zur Rhythmik des germanischen Alliterationsverses', BGDSL 10 (1885), 209-314 and 451-545 Altgermanische Metrik (Halle, 1893) Slay, D., 'Some Aspects of the Technique of Composition of Old English Verse', TPS (1952), 1-14 Smith, L. P., Words and Idioms: Studies in the English Language (Boston, MA, 1925) Stanley, E. G., 'Some Observations on the A3 Lines in Beowulf, in Old English Studies in Honour of John C. Pope, ed. R. B. Burlin and E. B. Irving, Jr, pp. 139-64 'Verbal Stress in Old English Verse', Anglia 93 (1975), 307-34 Tolkien, J. R. R., 'Beowulf: The Monsters and the Critics', PBA 22 (1936), 245-95; rpt in D. K. Fry, ed., The Beowulf Poet, pp. 8-56Trautmann, M., ed., Das Beowulflied. A Is Anhang das F inn-Bruchstuck und die Waldhere-Bruchstucke, Bonner Beitrage zur Anglistik 16 (Bonn, 1904) Watts, A. C., The Lyre and the Harp: a Comparative Reconsideration of Oral Tradition in Homer and Old English Epic Poetry (New Haven, CT, 1969) Willard, R., and E. D. demons, 'Bliss's Light Verses in the Beowulf JEGP 66 (1967), 230-44 Wrenn, C. L., ed., Beowulf with The Finnes burg Fragment, rev. W. F. Bolton (New York, 1973) Wyatt, A. J., ed., Beowulf, with The Finns burg Fragment, rev. R. W. Chambers (Cambridge, 1914) Zupitza, J. Beowulf: Reproduced in Facsimile from the Unique Manuscript British Museum Ms. Cotton Vitellius A. XV, 2d ed., Containing a New Reproduction of the Manuscript with an Introductory Note by N . Davis, EETS 245 (London, 1959)

230

Index of alliteration, scansion and metrical clause structure

Each half-line is scanned to indicate alliteration, cross alliteration, and extrametrical alliteration. Stressed elements, sentence particles and proclitics are distinguished. Word and clause boundaries are marked. The index employs the following symbols: IN STRESSED ELEMENTS

A = a stressed syllable with alliteration (including the second stressed syllable in a fully stressed two-element compound) N = a stressed syllable without alliteration (including the second stressed syllable in a fully stressed two-element compound); B = N with cross alliteration; -N = the second element in a class II compound or quasi-compound phrase; -B = -N with cross alliteration in a class II quasi-compound phrase only (i.e., cross alliteration in class II compounds is not marked) (A), (N) = a short syllable, as above (Ax), (Nx) = a short syllable + another syllable, as above (resolved stress) N.B. Syllables are resolved where possible with one exception: the short stressed syllable of a disyllabic second element of a fully stressed class I compound is not resolved unless it is immediately preceded or followed in the same half-line by a syllable of lesser stress (this implies that the short stressed syllable of a trisyllabic second element of such a compound is resolved). The lack of resolution in the 'short A 3 ' is an indication of its irregularity. The second element of a class II compound (-N) is never resolved; the second element of a class II quasicompound phrase may be resolved (only once does this occur when the second element falls in a half-lift ( = 128b). {N} = a contracted syllable marked A by Klaeber (half-lines with contracted syllables are classified by metrical type as though they were uncontracted; for the scansion of 681b and 1048b (C 2 (I)), see Fulk, 'Contraction', pp. 2 - 3 and 6) 231

Index n = a variably stressed prefix without stress; a long syllable of a formative element when followed by another syllable; the uninflected formative elements and suffixes -end, -ing (n) = a short syllable of a formative element when followed by another syllable {n} = a contracted syllable marked A by Klaeber IN SENTENCE PARTICLES

a = a stressed syllable (long or short) of an untransformed sentence particle in the a-verse, displaying extra-metrical alliteration w = a stressed syllable (long or short) of an untransformed sentence particle, either lacking, or displaying merely incidental, alliteration (such a syllable will sometimes display transverse alliteration (e.g. 2152a): these are not specially marked) IN STRESSED ELEMENTS, SENTENCE PARTICLES AND PROCLITICS

x = a proclitic syllable (long or short), or an unstressed enclitic'syllable (long or short), or any syllable (long or short) which can be resolved with a preceding short stressed syllable [x] = a syllable underdotted (suppressed) by Klaeber OTHER SYMBOLS

Underlined letters = (1) a dip with sentence particles stressed by the transformational rule; or (2) sentence particles or detached proclitics inserted into the first, alliterating, stressed position in the b-verse; or (3) proclitic adjectives stressed by the insertion rule. In (1) the entire dip is underlined; in (2) and (3) the underlining is confined to a single sentence particle or proclitic; the underlining in (3) is always followed by a second A or N with no intervening vertical bar (|) The symbol (=) is used between the stressed elements of fully stressed (class I) compounds, including compounds with variably stressed prefixes with the stress on the prefix. It is omitted between the words in a class I quasi-compound phrase, except when the phrase is preceded or followed by another stressed element. When a compound with a variably stressed prefix is stressed (in my judgement) on the second element, the prefix is marked n, and the compound is to be understood as class III (unless, of course, the stressed element itself is the first element of a compound, e.g., un-murn-lice) The symbol (-) is used between the stressed elements of class II compounds. It is omitted between the words in a class II quasi-compound phrase, except when

232

Alliteration, scansion and metrical clause structure the phrase is preceded or followed by another stressed element, and it is omitted between a proclitic prefix and the following stressed element Klaeber marks fully stressed compounds in his glossary with a hyphen. In general I accept his classification, but note these suffixes and other forms: Suffixes forming fully stressed compounds: -sceaft, -scipe Suffixes forming fully stressed compounds when followed by inflectional endings: -dom, -fast, -feald, -ful(l), -leas, -lie, -sum, -weard All other suffixes when followed by inflectional ending (n) Forms in ceg- and -hwyle are class II compounds = word boundary (omitted after proclitics) 11 = initial and internal clause boundary; also final boundary of parenthetical clauses where the next half-line does not begin a clause These symbols assume an implicit hierarchy of metrical stress within the half-line. The first A takes priority over any following A or N . Any second A takes priority over an N , if there is one. When there are two Ns, the one which is not preceded by the symbol ( = ) or (-) takes priority over the one which is. Following the scansion, the half-line is classified according to'its metrical type and clausal type. I employ the traditional Sievers typology, with slight modifications. An unraised 1 following the basic letter type indicates a monosyllabic drop (weak metrical position) between two lifts. An unraised 2 indicates a two or three syllable drop. A superscript numeral or numeral + small letter indicates a particular configuration of lifts and half-lifts distinguishing this type from others of the same basic letter type. An asterisk following the basic letter type indicates a one or two syllable expansion between adjacent lifts (or in one case after the second lift).

METRICAL TYPES

Al A2 A 2a A 2k A 2ab

2b

A A3 A3' Bl B2 B* C1

/x/x / (x) x x / x (including / \ x / x and / w x / x) /-/x /\ :x /

^

/ (x x) x / - (including / \ x / \) ( . . . ) x x / x (including . . . x x / - ) ( . . . ) x x vS x (the 'short A 3 ') (...)x/x/ (. . . ) x / ( x ) x x / (. . . ) x / ( x x ) x / x ( )x//x 233

Index (. . . ) x - ~ x C2 C# (. . . ) x / x / x D1 //\x D2 / / wx D3 -wxx D41 //x\ D42 //xx \ D* 1 /(xx)x/\x D* 2 / (x) x / w x (including / w x / ~ x) D* 4 1 / x / x \ D* 4 2 / x / xx \ El /\x/ E2 / \ (x) x x / E2 /w x/ E*l /x\x/ E*2 /x\xx/ F any verse which does not fit one of the above patterns H a hypermetric verse with an alliterating stressed syllable in'the onset before the regular metrical contour (which is labelled with one of the letters above) h a hypermetric verse without an alliterating stressed syllable in the onset before the regular metrical contour (NB, a syllable with extra-metrical alliteration in the onset is marked h, not H) NB / or \ = a long or a resolved stressed syllable; - = a long stressed syllable only; C or ~ = a short stressed syllable. CLAUSAL TYPES

I la Ib Ic II III

a clause-initial half-line, nondisplaceable (b-verse) a clause-initial half-line, nondisplaceable (a-verse) a clause-initial half-line, displaceable (a-verse) a clause-initial half-line, nondisplaceable, with sentence particle in 2nd allowable position (a- or b-verse) a clause-non-initial half-line an unrestricted half-line ADDITIONAL COMMENTS

Where the metrical clause structure requires double alliteration (and therefore restriction to the a-verse), the half-line is preceded by the symbol ( + + ) . Where the metrical clause structure requires the half-line to be restricted to the b-verse (and therefore to display single alliteration), the half-line is preceded by the symbol (oo).

234

Alliteration, scansion and metrical clause structure Certain special cases (i.e. double alliteration mandated by a proclitic preceding the first lift and base word in second lift; restriction to the b-verse mandated by a proclitic preceding the first lift and lack of alliteration on base word in second lift; single alliteration in the a-verse where the second lift is governed by a vocalic prefix in the w l ; single alliteration in the a-verse or the b-verse where the second lift is occupied by gehwylc or genog; single alliteration in the a-verse or the b-verse where contrast, rhyme or proper name substitutes for alliteration) are preceded by the symbol (°). Violations of the metrical clause structure are marked with an asterisk. Other problems are marked with two asterisks and listed here by line number: 274b: This is a very odd looking verse. It appears to be a parenthetical clause which is in apposition to the subject, deogol dcedhata (275a). Deletion of the ic would turn it into a regular A1 (III): sceaSona nathwylc, without change of meaning. Cf. gumena nathwylc (2233b). 363b: Demonstrative with a relative function. 4l4a: Reading hador with a short a for hador. see Pope, Rhythm, p. 323, and Bliss, Metre, §§ 46-7. 457a: Violation of the metrical clause structure in a partially reconstructed line. 495b: Class I compound in second position in b-verse. 525a: Consists entirely of sentence particles and proclitics and is stressed by the transformational rule. Hence the apparent placement of a sentence particle in a w2 is in reality caused by the fact that the entire verse is a w 1. 570a: Or, if beacen is read as disyllabic, D 4 1 . 574b: Double alliteration in b-verse. 596b: In the same formula in 634b eowra is a possessive adjective (proclitic). 603b: Though I mark gaep eft as a class II compound, the stress on the verb must be the result of insertion into the initial stressed position in the b-verse. 639a: Demonstrative with a relative function. 665a: Failure of alliterative rule of proclitics. 673a: Taking the penultimate syllable of irena as long; see Klaeber, p. 153. 686a: Taking hwcepere as a fully stressed adjective. 743a: Triple alliteration in a-verse. 758a: Verb leads noun in a-verse. 801b: Demonstrative with a relative function. 845a: Failure of alliterative rule of proclitics. 910a: Failure of alliterative requirement of demonstratives. 928a: Demonstrative with a relative function. 947b: Violation of the metrical clause structure: the sentence particle appears to be redundant.

235

Index 984b: JEghwylc is a class II compound. I mark it Nn here because it is embedded in a class II compound phrase. 992b: Though I regularly consider class II compounds and their equivalents to be stressed elements, in this case it would appear that the quasi-compound phrase fela paera behaves like a class II compound as a result of the insertion otfela into the initial stressed position in the b-verse. 1030a: Demonstrative with a relative function. 1110a: Demonstrative with a relative function. 1166a: The verb sat is displaced and presumably receives metrical stress in the hypermetric verse, but it does not alliterate. There is too little evidence to determine whether this constitutes an irregularity in the metrical grammar. 1179a: Failure of alliterative rule of proclitics. 1180a: The uncontracted infinitive sehan is assumed to have had a short penult. Hence it occupies the final X-position and double alliteration is not mandatory (see ch. 12). With the uncontracted form inserted, the scansion is A 2k . 1275b: The uncontracted infinitive sehan is assumed to have had a short penult. With the uncontracted form inserted, the scansion is A 2k . 1327b: Finite verb leads a noun. The authenticity of the verse is confirmed by the fact that it is a formula (see 2544b). 1377a: See 2665b. 1390b: Proper alliteration depends on the assumption that the h of hrape was not pronounced. 1422a: Failure of class II compound to alliterate in a-verse. 1492a: Demonstrative with a relative function. 1537a: Verb leads noun in the a-verse. The emendation of eaxle to feaxe is surely right, and would convert this to a normal type A 3 (la). 1538a: Abnormal double alliteration. 1567b: Or, if eal is taken as an adverb, D 4 . 1583a: The transformational rule, which operates first, makes the insertion rule redundant. 1697a: Taking the penultimate syllable of irena as long; see Klaeber, p. 153. 1727b: Incidental alliteration on the finite verb ah. 1828a: The first syllable of hetende ( = hettende) is long. 1828b: Possibly An|(N)x: see Pope. 1870b: Class I compound in b-verse. 194la/194lb: It is doubtful whether these are hypermetric verses. 1975a: Klaeber underdots the initial h of hra&e (see 1390b). If the h is allowed to stand, the pattern is ||wx|w|xAx. 2059a: Failure of alliterative requirement of demonstratives. 2259b: Taking the penultimate syllable of irena as long; see Klaeber, p. 153. 2275 a: Failure of alliterative rule of proclitics: the verse is partially reconstructed.

236

Alliteration, scansion and metrical clause structure 2296b: Irregular double alliteration in b-verse. 2430b: Evidently modelled on the pattern of the A 2k , but the second syllable of Hredel is hardly heavy enough to justify so marking it. Cf. 1828b. 2475a: Other editors mark the penultimate syllable of Ongenckowes as long. If the antepenultimate syllable were underdotted (cf. 2387b), the verse could be marked as C 1 . 2544b: See 1327b. 2585b: The adverb no does not ordinarily take stress in an adverb + verb phrase. Hence it is here marked as being inserted into the initial stressed position in the b-verse. 26l5a/26l5b: These are normal verses when taken separately, but fail to alliterate properly when joined together. I mark 2615a as F; it would be an El if the b-verse alliterated on B. 2665b: Pronoun behaving as if proclitic on following stressed participial adjective, similarly, 1377a. 2669a: Demonstrative with a relative function. 2717b: Verb leads noun. Cf. 1327b. Class I compound in b-verse. 294 la: Violation of the alliterative rule of proclitics: partially reconstructed verse. 3019a: This half-line is a formula: see Elene 1252a, Christ 1194a. Nalles is here a proclitic adverb with quasi-conjunctive force. Although the verse could be scanned by the transformational rule, yielding an A 3 (Ib), I think it better to take it as parallel to such conjoined adverbial phrases as Pa wees eft swa cer (642a, 1787a) axxdpczt mec cer ondsid(2500a). Nalles (nealles) is also virtually proclitic in 1076a and 3089a, without, however, being conjunctive. In these verses, I mark it as a sentence particle. 3056a: This would be a regular B2 if the b-verse alliterated on M.

INDEX

la 2a 3a 4a 5a 6a 7a 8a 9a 10a lla 12a 13a

||w|w|A=(B)x A=(N)nx | w|x(Ax)nx ||w|A|An ++ (Ax)xAx ||axx|Ax ++ A=N|Ax ||a|xxAx ||w|w|w|A-N xxA=Nx ||Ax|Ax ||w|(Ax)x|N ++ A|xAx

C2 (la) D3 (III) C1 (la) C1 (la) Al (III) A3 (Ib) A2a (III) A3 (Ib) A3 (la) C1 (II) Al (III) Bl (la) Al (III)

Ib xA=(B)x oo 2b A|xNx

Ax|(Nx)x (Ax)x|Nx oo (Ax)=Nx|xN ||wx|Ax|N ww|Ax|xN ||A=Nx|N A=Nnx Ax|Nx l i b ||w|w|A|(N)n 12b Ax|Nx 13b ||wx|A|Nx 3b 4b 5b 6b 7b 8b 9b 10b

237

C2 (II) Al (III) Al (III) Al (III) E2 (III) B1(I) B2(I) El (III) D 1 (III) Al (III) C2(I) Al (III)

Index 14a 15a 16a 17a 18a 19a 20a 21a 22a 23a 24a 25a 26a 27a 28a

29a 30a 31a 32a 33a 34a 35a 36a 37a 38a 39a 40a 41a 42a 43a 44a 45a 46a 47a 48a 49a 50a 51a 52a 53a 54a 55a 56a 57a 58a

\x|xAx ||w|w|A|(N)x Ax|Nx Ax An ||An|w|Ax Ax|(Bx)x llw w|A|(A)x (Ax)|A=Nx 1 l wwx|xAx A=xNx ++ ||Ax|xAx xAx|xNx ||w|w|A|xN ++ (Ax)=N|Ax wx w|xAx llw ++ Ax|xAx ||wx|Ax|A ||A|A=(N)x ||w|xAx|B ++ Ax|xA=B ||xAx|B Ax|Ax ++ Ax|xAx xA=(N)x | xwx|w|A=Nx Ax=Bx ++ Ax|xAx Ax|(Ax)x xAx|N ||wx w|wx|Ax A=xNx ||w|wx|xA=Nx ++ Ax|xxAx | xw|w|w xAx ++ A|xxAx ||ax|xA-N Anx A

ax|xAx ++ (Ax)|xx(Ax)x ||w|w|xAx A|A=(N)n ++ AxxAx ++ AxxAx A|A(n)x ++ (Ax)|xA=N

oo ||(Ax)=Nx|xN Ax=Nx ||w w|A={N} oo (Ax)=Nx|xN oo ||A|Nx|-N|| (Ax)=Nx|B oo Ax|xNx x(Ax)|Nx oo A|x(Nx)x 23b ||wx|A|(N)x 24b ||A=Nx|N 25b ooA|x{N} 26b xxA=Nx 27b xA|Nx 28b x(Ax)|(Nx)x 29b llw w|Ax|N 30b (Ax)|Nnx 31b Ax|Nx 32b Ax=Bx 33b (Ax)nx|B 34b Ax|Bx 35b xA|(N)x 36b llw w|Ax|(Nx) 37b ooAx|xNx 38b ooA|xNx 39b x(Ax)=Bx 40b ||w|xAx|N 41b ||w|w|A|Nx 42b ooA|xNx 43b Ax|Nx 44b ||w|A|(N)x 45b ooA|xNx 46b A[x}=(N)nx 47b (Ax)|Nnx 48b ||wx|A|(N)x 49b llw w|Ax|(Nx) 50b oo||A|xNx 51b (Ax)=Nnx 52b ||w xAx xN 53b An|Nnx 54b Ax|Nx 55b oo||(Ax)|Nx|-N 56b llw w|w|A|xN 57b oo||A||wx]Nx|| 58b (Ax)|Nnx

A2 (III) C2 (la) Al (III) Al (III) A2 (Ic) Al (III) C2 (la) D 1 (III) A3 (la) Al (III) A2 (III) B* (II) Bl (la) A2a (III) A3 (la) A2 (III) Bl (la) D2 (III) Bl(Ia) A2b (III) Bl (la) Al (III) A2 (III) C2 (II) C1 (la) Al (III) A2 (III) Al (III) Bl (II) A3 (la) Al (III) C1 (la) A2 (III) A3 (la) A2 (III) A3 (Ib) El (III) A3 (Ib) A2 (III) A3 (la) D2 (III) A2 (III) A2 (III) D2 (III) A2b (III)

14b 15b 16b 17b 18b 19b 20b 21b 22b

238

E2 (III) Al (III) CX(I) E2 (III) D41 (III) El (III) A2 (III) C1 (II) Al (III) C2(I) El (III) Al (III) C1 (II) C1 (II) C1 (II) Bl (I) D 1 (III) Al (III) Al (III) El (III) Al (III) C2 (II) Bl (I) A2 (III) Al (III) C1 (II) Bl (I) C^I) Al (III) Al (III) C2(I) Al (III) D3 (III) D 1 (III) C2(I) Bl (I) Al (III) D 1 (III) B2(I) D #1 (III) Al (III) D41 (III) Bl (I) A2 (III) D 1 (III)

Alliteration, scansion and metrical clause structure 59a ||w|Ax|N 60a x(Ax)|Ax ++ 61a (Ax)n|xAn 62a defective

63a (Ax)=Nnx 64a ||w|w|Anx 65a ++ Ax|A=B 66a Ax|Nx 67a (Ax)=N|(A)x 68a ||w|A=(N)x 69a ||(Ax)=N|(N)x 70a ||wx|Ax|N 71a ||w w|xAx 72a °Ax|xNx 73a xxA=(N)x 74a ||w|w|Ax|xN ++ 75a (Ax)xAx ++ A=(N)x|Ax 76a 77a axxAx 78a A=Nx|N 79a ||w w|xAx xA 80a HwAxxN ++ 81a A|xAx 82a + + A|xA=N 83a Ax Ax 84a ||w|xA=(N)x 85a xxA=Nx 86a ||w|xAx=N ++ Ax|x(Ax)x 87a 88a ||w|w|A[x}x|xB ++ Ax|xAx 89a 90a (Ax)|A|=(N)x ++ A=N|Ax 91a 92a ||w||w|xA-N[x}x 93a (Ax)=BxA 94a ||xax|(Ax)=Nx 95a ++Ax|xAx 96a ||w|xA(n)x ++ 97a Ax|xAx 98a °Ax|xBx 99a ||w|xA=(N)x 100a Ax=Nx 101a (Ax)x|Ax 102a °||w|xAx|A 103a ++ Ax|A=(N)x

Bl (la) C1 (II) A2 (III) D 1 (III) C1 (la) A2b (III) Al (III) A2k (III) C2 (la) A2k (III) Bl (la) A3 (la) A2 (III) C2 (II) B2 (la) Al (III) A2 (III) A3 (Ib) El (III) B2 (la) B2 (la) Al (III) A2b (III) Al (III) C2 (la) C1 (II) Bl(Ia) A2 (III) B2 (la) A2 (III) D2 (III) A2a (III) C1 (la) El (III) C1 (Ib) A2 (III) C2 (la) A2 (III) A2 (III) C2 (la) Al (III) Al (III) Bl(Ia) D*2 (III)

59b A=xNx 60b (Ax)x|Nx 61b xAx|N 62b x(Ax)x|N 63b A=xNx 64b (Ax)=N|(N)x 65b ||w|w|x(Ax)=Bx 66b ||w|w|x(Ax)|xN 67b ||w|xA|xN 68b Ax|Nx 69b ooA|xNx 70b ooAx|xNx 71b ooA|xNx 72b ||w|w|A|Nx|| 73b x(Ax)|(Nx)x 74b ooA|xNx 75b xxxAx=N 76b ||w|xAx|xN 77b ||w|w|w|A-(N)x 78b ||w|w|A|(N)x 79b Ax|Nx 80b ||Ax|Nx 81b ||(Ax)|N(n)x 82b ||(Ax)=Nx|N 83b ||xw|w|Ax|xN 84b Ax=Nx 85b Ax|Nx 86b Ax=Nx 87b ||w|w|xAx|N 88b ooA|xBx 89b ||w|w|Ax|N 90b oo||Ax||wlw|Nx 91b Ax|Nx 92b Ax|Nx 93b ||w|(Ax)|xBx 94b °Ax|xNx 95b A=Nnx 96b Ax|Nx 97b oo||A|N|-xN 98b ||wx|w|(Ax)|Bx . 99b Ax|Nx 100b ||w|w|A|xN 101b ooA|xNx 102b Ax|Nx 103b ||w|w|Ax|N

239

A l (III) A l (III) B 1 (II) B 1 (II) A l (III) A 2 k (III)

A l (III) A l (III) A 2 (III) A l (III) C^I) C 1 (II) A l (III) B 1 (II) B2(I) C 2 (I) C 2 (I) A l (III) A l (III) D 2 (III) El (III) B2(I) A l (III) A l (III) A l (III) B1(I) A l (III) B1(I) A 2 (III) A l (III) A l (III) C*(I) A 2 (III) D 1 (III) A l (III) A l (III) B1(I) A l (III) A l (III)

Index 104a + + A|xAx 105a A=Nx|A 106a ||wx|w|An 107a xA(n)x|Ax 108a Ax|Nx 109a ||xxa|w xxAx 110a ++ (Ax)|xxAx Ilia ||wx|A=Nx 112a ++ (Ax)x|xAx 113a ||wx|Anx 114a Ax|Nx 115a ||xw|w|A(n)x 116a {A}|Ax 117a xxA=(N)x 118a l|w|w[xAx 119a ax | xxAx 120a A=N|(A)x 121a + + A|xAx 122a + + A|xAx 123a Ax|Ax 124a Ax|Ax 125a ||xwx|A=Nx 126a |w|w|xAx 127a + + Ax|A=N 128a ||w w|xxAx 129a xxAx=N 130a + + (Ax)n|A-N 131a ||axx|A=B 132a ||wx wxAx 133a Ax|Nx 134a + + A|xA-N 135a ||w|xAx|N 136a + + A=(N)x|Ax 137a ++ Ax|x(Ax)x 138a ||w w|A=Nx 139a xA=Nx 140a + + A|xxAx 14 la ||xa|A=Nx 142a A=Nx|(Ax) 143a + + A|xAx 144a Hw|A(n)x 145a + + Ax|xAx 146a Ax|Nx 147a ||A|=Nx|A 148a (Ax)|Nnx

Al (III) El (III) A 3 (la) F(II) Al (III) A 3 (la) A2 (III) C 1 (la) A2 (III) C 1 (la) Al (III) C2 (la) Al (III) C2 (II) A 3 (la) A 3 (Ib) A 2k (III) Al (III) Al (III) Al (III) Al(III) C 1 (la) A 3 (la) A 2b (III) A3 (la) Bl (II) Al (III) A 3 (Ib) A 3 (la) Al (III) Al (III) Bl(Ia) A2 (III) A2 (III) C 1 (la) C 1 (II) A2 (III) C 1 (Ib) El (III) Al (III) C 2 (la) A2 (III) Al (III) El (III) D 1 (III)

104b ||A[x}=Nx|N 105b ooA(n)xlNx 106b x(Ax)|Nx 107b ||wx|A|xN 108b ||w|w w|Ax|N 109b ||w|w w x A x N 110b A=Nx|N 111b Ax|xNx 112b xA{n} 113b ||w|x(Ax)|Nx 114b ||w w|w|A|xN 115b ||wx|A|xN|| 116b | w w|A=(N)x 117b xA|Nx 118b oo(Ax)nx|xN 119b oo ||Ax|xNx 120b ||A|nNx 121b oo(Ax)|Nx -N 122b llw xAx xN 123b oo ||(Ax)!N|-xN 124b xA|(N)x 125b Ax|Nx 126b xA-(N)x 127b A|nNx 128b oo A|N|-x(Nx) 129b ||Ax|Nx 130b A=Nx|N 131b ||A=Bx|N 132b A|N(n)x 133b ||w|xxA xN 134b ||w|w|Ax|N 135b ooA|x(Nx)x 136b oo ||w|w|A|(N)x|| 137b oo ||w|xA|xN 138b llw w|Ax|N 139b Ax|Nx 140b ||w w|xAx|N 141b (Ax)x|Nx 142b oo ||A|wx|Nx 143b ||w xAx xN 144b ||w|xAx|N 145b llw w|Ax|N 146b llw xA|(N)x 147b ooA|x(Nx)x 148b °Ax|xNx

240

El (III) A2 (III) C 1 (II) B1(I) B1(I) B1(I) El (III) A2 (III) C 1 (II) Oil) B1(I) B1(I) C 2 (I) C 1 (II) E2 (III) A2 (III) Al (III) D 4 1 (III) B2(I) D 4 1(III) C2 (II) Al (III) C2 (II) Al (III) D 4 1 (III) Al (III) El (III) El (III) D 2 (III) B1(I) B1(I) Al (III) C 2 (I) B1(I) B1(I) Al (III) B1(I) Al (III) A2 (Ic) B2(I) B1(I) B1(I) C 2 (I) Al (III) A2 (III)

Alliteration, scansion and metrical clause structure 149a Ax | Ax 150a Ax|Nx 151a Ax|A[x]x 152a ++Ax|xAn 153a + +(Ax)x|xAx 154a Anx|(Ax) 155a xAx|(Nx) 156a + + ||A=(N)x|Ax 157a ||w|w|wx|(Ax)x 158a Ax|Ax 159a ||w|xA=Nx 160a A|A=(N)x l 6 l a + + ||(Ax)x|x(Ax)x 162a A(n)x|Ax 163a ||wx|A=Nx 164a ||w|ax|(Ax)x 165a (Ax)|A=Nx 166a Ax|Ax 167a A=Nx|A 168a ||w|w|xxA=N 169a ++Ax|x(Ax)x 170a ||w|w|A|(N)x 171a Ax|Nx 172a ++Ax|xAx 173a ||w|A=Nx 174a xA=(N)x 175a llax[w|xAx 176a A=Anx 177a ||w|w|A=(N)x 178a xA=Ax 179a Anx|A 180a xA=(N)x 181a Ax|An 182a ||w|w|wx|(Ax)x|A 183a Ax|An 184a xAx|N 185a xAx|A 186a + + ||Ax|xAx 187a xxA=(A)x 188a ||w|x(Ax)|Ax 189a ||w|xA=(N)x 190a Anx|A 191a ++ A|xAx 192a ++ A|xA-N 193a + + A=(N)x|A=N

149b ||xw|Ax|N 150b A=Nx|N 151b ||wx|Ax|N 152b ||(Ax)=Nx|N 153b oo(Ax)|N-Nx 154b oo||Ax|xNx 155b (Ax)x|(Nx)x 156b ||A|N(n)x 157b Ax|Nx 158b x(Ax)|Nx 159b Anx|N 160b °(Ax)x|x(Nx)x 161b ||A=Nx|N 162b oo||A|xNx 163b Ax|Nx 164b A|nNx 165b ooA|x(Nx)x 166b ||(Ax)|N(n)x 167b Ax|Nx 168b Ax|Nx 169b ||w|x(Ax)|Nx 170b (Ax)|Nnx 171b oo||(Ax)|N|-xN 172b ||A|N(n)x 173b Ax|Nx 174b xxAnx 175b xA=(N)x 176b ||Ax|Nx 177b ooAx|x(Nx)x w|A|(N)x 178b 179b >o||Ax|xNx 180b >o||(Ax)|w|xNx 181b |xwx|w|Ax|N 182b )O(Ax)x|xNx 183b >oHA|-N|lwlw|N 184b >oAx|xNx 185b )0 ||Ax|xNx 186b 187b Ax|Nx 188b (Ax)|N(n)x 189b (Ax)|N(n)x 190b ||xwx|A[x}|(N)x 191b ||w|xxA|xN 192b ||w|xxAx|xN 193b A=Nx|N

Al (III) Al (III) Al (III) A2 (III) A2 (III) El (III) Bl (II) A2 (III) A3 (la) Al (III) C1 (la) D2 (III) A2 (III) A2 (III) C^Ia) A3 (la) D1 (III) Al (III) El (III) A3 (la) A2 (III) C2 (la) Al (III) A2 (III) C1 (la) C2 (II) A3 (la) D1 (III) C2 (la) C1 (II) C2 (II) Al (III) Bl (la) Al (III) Bl (II) Bl (II) A2 (III) C2 (II) C1 (la) C2 (la) El (III) Al (III) Al (III) A2b (III)

241

El (III) B1(I) El (III) D 1 (III) A2 (III) Al (III) D2 (III) Al (III) C1 (II) El (III) A2 (III) El (III) Al (III) Al (III) Al (III) Al (III) D2 (III) Al (III) Al (III) C^I) D 1 (III) D41 (III) D2 (III) Al (III) C1 (II) C2 (II) Al (III) A2 (III) C2(I) A2 (III) A2 (Ic) B1(I) A2 (III) E2 (III) A2 (III) A2 (III) E2 (III) Al (III) D2 (III) D2 (III) C2(I) B1(I) B2(I) El (III)

Index 194a ||w|xA|xN 195a ++ A|xAx 196a ||w|w|A=Nx 197a xA(N)x 198a + +(Ax)x|xAx 199a ++Ax|xAx 200a xxA=Nx 201a Ax|Bx 202a ||wx|A-N|N ++ A=N|Ax 203a 204a ||ax|(Ax)=Nx 205a ||wx|xAx ++ 206a Ax|x(Ax)x 207a Ax|Nx 208a ++ A=(B)x|Ax 209a (Ax)=Nx|B 210a + + ||A|A|-xN 211a + + ||A|xxAx 212a xA|(A)x 213a ++ A|xAx 214a xA|(N)x 215a + +A=(N)x|(Ax)-N 216a ++ (Ax)|xA=N 217a ||xa|w|xxA=N ++ 218a (Ax)|Ax=N 219a |w|w|xA=N 220a Ax=Nx 221a ||w|xAnx 222a + +A=(N)x|Ax 223a ++ Ax|A=Nx 224a ++(Ax)x|xAx 225a (Ax)x|Nx 226a + + ||A=(N)x|Ax 227a A=xNx 228a ||w|w|w|A=Nx 229a ||w|xAx|xN 230a ||w|w|A=(N)x 231a ax|xxAx 232a ++ A=(N)x|A=(N)x 233a A=xNx 234a ||xa|w|w|x(Ax)x 235a A|Nnx 236a ++(Ax)=(B)x|Ax 238a Ax|(Nx)x

Bl (la) Al (III) C1 (la) C2 (II) A2 (III) A2 (III) C1 (II) Al (III) Bl (la) A2a (III) C1 (Ib) A3 (la) A2 (III) Al (III) A2 (III) El (III) D41 (III) A2 (III) C2 (II) Al (III) C2 (II) A2 (III) A2b (III) A3 (la) D41(III) A3 (la) Al (III) C1 (la) A2 (III) D*1 (III) A2 (III) Al (III) A2 (III) Al (III) C1 (la) B2 (la) C2 (la) A3 (Ib) D*2 (III) Al (III) A3 (la) D 1 (III) A2 (III) Bl(Ia) Al (III)

194b (Ax)nx|N 195b Ax|Nx 196b (Ax)x|Nx 197b ooAxNx 198b ||w|w|A=(N)x 199b ||w||w|A=(N)n 200b Ax|Nx 201b ||w|w|w|Ax|B 202b (Ax)x|Nx 203b ||w|w|w|A|Nx 204b ||A|N(n)x 205b Ax|Nx 206b ||wx|w|w|Anx 207b ||A=Nx|N 208b ||A|B(n)x 209b A=xBx 210b oo||(Ax)|w|xNx 211b ||Ax|Nx 212b ||Ax|Nx 213b ||Ax|Nx 214b Ax|Nx 215b oo||(Ax)|N|-(N)x 216b (Ax)|Nnx 217b ooAx|xNx 218b ooAx|xNx 219b ooAxN[x}x 220b x(Ax)|Nx 221b ooA|xNx 222b Ax|Nx 223b ||w|w|A|(N)x 224b ||wx|A|(N)x 225b xA|(N)x 226b ||Ax|(Nx)x 227b ||(Ax)|Nnx 228b Ax|Nx 229b A|Nnx 230b Ax|Nx|| 231b Ax|Nx 232b ||wx|An|N 233b ||w|xA|Nx 234b Ax|Nx 235b ||Ax|Nx 236b ||(Ax)=Bx|N 237b (Ax)=Nnx 238b ||w|xAx|N

242

El (III) A l (III) A l (III) A l (III) C 2 (I) C 2 (I) A l (III) B1(I) A l (III) Cl{\) D 2 (III) A l (III)

O(D

El (III) D 2 (III) A l (III) A 2 (Ic) A l (III) A l (III) A l (III) A l (III) D 2 (III) D 1 (III) A 2 (III) A 2 (III) A l (III) C 1 (II) A l (III) A l (III) C 2 (I) C 2 (I) C 2 (II) A l (III) D 1 (III) A l (III) D 1 (III) A l (III) A l (III) A l (III) A l (III) El (III) D 1 (III)

Alliteration, scansion and metrical clause structure 239a 240a 241a 242a 243a 244a 245a 246a 247a 248a 249a 250a 251a 252a 253a 254a 255a 256a 257a 258a 259a 260a 261a 262a 263a 264a 265a 266a 267a 268a 269a 270a 271a 272a 273a 274a 275a 276a 277a 278a 279a 280a 281a 282a 283a

xx(Ax)=Nx ax | xxAx Ax=Nx ||w|xA|(N)x xA=Nx j|w|w|A=Nx A=Nnx A=Nnx + + Ax|xAx ++ Ax|xxAx ++ A|xAx + + Ax|xAx + + A-N|A-N A=N|(N)x A=N(n)x Ax|Ax (Ax)=Nnx °A=Nx|xN xxAnx |w|xAnx (Ax)xAx |w|w|A=Nx xAnx ||w|x(A)x ++ (Ax)x|A=(N)x ||xw|Ax|A ++ (Ax)|xAx + + (Ax)x|A-N ||w|xAx|(Ax) (Ax)|Nnx A=xNx ||wx w xxAx

(Ax)x|{N} + +

ax||\v|wjAx ||w w|A=Nx ||w|xAnx ++ Ax|A=(N)x ax|xAx + + Ax|xA=N xAx|(Nx) llw w|A|xN ||w|w A-Nx (Ax)x|(Ax)x ||w|xA=Bx wxlAlNx '

C 1 (II) A3 (Ib) A l (III) C2 (la) G1 (II) C 1 (la) D 1 (III) D 1 (III) A2 (III) A2 (III) A l (III) A2 (III) A l (III) A 2k (III) D 2 (III) A l (III) D 1 (III) E2 (III) C 1 (II) C 1 (la) A l (III) C 1 (la) C 1 (II) A3" (la) D* 2 (III) Bl(Ia) A l (III) A l (III) Bl(Ia) D 1 (III) A l (III) A3 (la) A l (III) A 3 (Ib) C 1 (la) C 1 (la) D* 2 (III) A 3 (Ib) A 2b (III) Bl (II) Bl(Ia) C 1 (la) A l (III) C 1 (la) C 1 (la)

239b 240b 241b 242b 243b 244b 245b 246b 247b 248b 249b 250b 251b 252b 253b 254b 255b 256b 257b 258b 259b 260b 261b 262b 263b 264b 265b 266b 267b 268b 269b 270b 271b 272b 273b 274b 275b 276b 277b 278b 279b 280b 281b 282b 283b

243

Ax|Nx ||w|w|Ax|N A=Nx|N Ax|Nx oo Ax|xNx oo (Ax)|xNx ||w w|Ax=N oo Ax|xNx oo ||wx|w|Ax|xN oo ||wx|w|Ax|N ||w|w|A=(N)x | wx|w x(Ax)|Nx oo ||w w|Ax|N ||w|w|A|(N)x xA|(N)x ||w|w A=Nn oo Ax|xNx ||(Ax)|w|Nx ||wx|xx(Ax)|Nx A=(N)(n)x oo ||A=N|xN Ax Nx

A=xNx oo Ax|xNx An|Nx llw w|xA|Nx wx|Ax xN Ax|xNx A-NNx Ax|Nx w|w w|A(n)x|N (Ax)|Nnx | xw|w Ax|N oo ||w|A||w|w|N Ax|Nx oo ||(Ax)x|w|N|-N|| Ax|Nx A=Nx|N ||w|w|An|N oo A|xNx oo A|xxNx Ax|Nx oo A|N|-(N)x Ax|Bx Ax=Nx

Al (III) B1(I) El (III) Al (III) A2 (III) Al (III) Bl (I) A2 (III) B2(I) Bl (I) C2(I) C^I) Bl (I) C2(I) C2 (II) C1 (I) A2 (III) Al (Ic) C1 (I) D3 (III) El (III) Al (III) Al (III) A2 (III) Al (III) C^I) B2(I) A2 (III) Al (III) Al (III) B2(I) D 1 (III) Bl (I) B2(I) Al (III) A2 (Ic)** A1 (III) El (III) Bl(I) Al (III) A2 (III) Al (III) D2 (III) Al (III) Al (III)

Index 284a A=N|(A)x 285a xA=(N)x 286a ||A|(Nx)(n)x 287a + + A-N|A=N 288a A|A=(B)x 289a ++Ax|xAx 290a ||w|w|xAx 291a A|Nnx 292a + +Ax|xxAx 293a ||wx|w|(Ax)=Nx 294a xAx|x(Nx) 295a A=Nnx 296a Ax|Nx 297a xx(Ax)=Nx 298a + +(Ax)|Ax=N 299a ||A=Nnx 300a ||w|xxAx=N 301a ||xwx|w|w|Ax 302a ||axx|xAx 303a xAx|N 304a xxA=Nx 305a + + A|xA=B 306a + + A=N|Ax 307a ||ax|xAx 308a + +(Ax)-N|xA=N 309a ||w|w|(Ax)=Nx 310a ++(Ax)x|xx(Ax)x 311a ||ax|xAx 312a ||w|w|Ax=N 313a + +A|xAx 314a Ax|Ax 315a ++ A|xAx 316a ++ ||N|w|w|xAx 317a xA=(N)x 318a + +Ax|xAx 319a xA|(A)x 320a + + ||A|w|A=N 321a ++(Ax)|x(Ax)x 322a A|A=(N)x 323a a|xAx 324a xxx(Ax)=Ax 325a ||ax|A=Nx 326a ++ Ax|A=Nx 327a ||ax|w|xAx 328a + +A=(N)x|(Ax)x

A2k (III) C2 (II) D2 (HI) A2b (III) D2 (III) A2 (III) A3 (la) D 1 (III) A2 (III) C1 (la) B2 (II) D 1 (III) Al (III) C1 (II) D41 (III) D 1 (HI) Bl (la) A3 (la) A3 (Ib) Bl (II) C1 (II) A2b (III) A2a (III) A3 (Ib) A2b (III) C1 (la) A2 (III) A3 (Ib) Bl (la) Al (III) Al (III) Al (III) F(lc) C2 (II) A2 (III) C2 (II) A2b (Ic) Al (III) D2 (III) A3 (Ib) C1 (II) C1 (Ib) D #1 (III) A3 (la) A2 (III)

284b ||wx|A|(N)x 285b Ax|Nx 286b ||w|xAx|N|| 287b ||A-Nx|N 288b xA|(B)x 289b ||w w|A|Nx 290b llw w|wA|(N)x 291b ||xwx|A|(N)x 292b oo lh N w|A(n)x 293b ooAx|Nx 294b (Ax) | Nnx 295b oo(Ax)|xNx 296b llw w|A|(N)x 297b Ax|Nx 298b x(Ax)=Nx 299b wx|(Ax)x|N 300b ooA|xNx 301b ||(Ax)|=Nx|N 302b A=Nx|N 303b ||A=N|(N)x 304b x(Ax)|Nx 305b ||A=Nx|B 306b ||(Ax)|Nnx 307b ||w w|w|A|Nx 308b x(Ax)|Nx 309b A=Nnx 310b ||xw|xAx|N 311b xxAx|(Nx) 312b A Nnx 313b | w w|w|A|Nx 314b A=Nx|N 315b oo||A|Nx|-N 316b 00 (Ax)|N-Nx 317b : xNx 318b ||w|xA|Nx 319b Ax|Nx 320b ||A|N(n)x 321b ||A=Nx|N 322b A = N x N 323b |w ^sv|w|(Ax)|Nx 324b Ax|Nx 325b Ax|Nx 326b °xx(Ax)x|N 327b ||Ax|Nx 328b ||Ax|Nx

244

C2(I) Al (III) Bl(I) El (HI) C2 (II) Oil) C2(I) C2(I) C2(I) Al (III) D 1 (HI) Al (III) C2(I) Al (III) C1 (II) Bl(I)* Al (III) El (III) El (III) A2k (III) C1 (II) El (III) D 1 (III) O{1) C1 (II) D1 (III) Bl(I) B 1 (II) D 1 (III)

O(D

El (III) D41 (III) D 1 (III) A2 (III) C ! (I) Al (III) D2 (III) • El (III) El (III) CX(I) Al (III) Al (III) B1 (II) Al (III) Al (III)

Alliteration, scansion and metrical clause structure 329a 330a 331a 332a 333a 334a 335a 336a 337a 338a 339a 340a 341a 342a 343a 344a 345a 346a 347a 348a 349a 350a 351a 352a 353a 354a 355a 356a 357a 358a 359a 360a 361a 362a 363a 364a 365a 366a 367a 368a 369a 370a 371a 372a 373a

A=Nx|(Ax) ++ A=N|(Ax)|=N ++ A>: xAx Ax=Nx ||wx|(Ax)x|N Ax|Nx (Ax)=Nx|A ++ A|xA-N x(Ax)xA lla w||w|w xAx ||w|x(Ax)=Nx ||w w|Ax=N ++ A|(Ax)x|=N ++ A|xxAx A=xBx wxwxAx Ax|Nx Ax|Nx llw w wx|xAx ||An|(Nx)(n)x ||w|xA=(N)x ++ A|xA-N A|Nnx Ax Ax Ax|Nx ||w|w w|A-(N)x ||w|w xAx Ha w|A=Nx ++ A : «A-N ||ax|Ax=N (Ax)x|{N} ||An|(Nx)(n)x ||w|wx|x(Ax)x xx(Ax)x xA |wx|Anx An|Nx ||w|w|Ax|B Ax|Ax xxA=(N)x ||w|xA=x(Nx) ++ Ax|xAx ||w|x(Ax)=Nx ||An|(Nx)(n)x ||w|wx|Ax ||w|xA=(N)x

329b oo(Ax)|x(Nx)x 330b ||w|xAx=N 331b ||w|w|A|(N)x 332b xx(Ax)x|N 333b Ax|Nx 334b xA=Nx 335b ||w|w Anx 336b ||xw|w|A=N[x}x 337b Ax=Nx 338b | wx|xA=Nx 339b An|Nx 340b A=(N)(n)x 341b oo ||A|Nx|-N 342b ||w w|(Ax)nx 343b oo||An|w|x(Bx) 344b (Ax)|N(n)x 345b A|Nnx 346b ||w|w w|xAx|(Nx) 347b Ax|Nx 348b ||w w|Ax|N 349b oo(Ax)x|xNx 350b w w|(Ax)|(Nx)x 351b Ax|Nx 352b llw w|Ax|N|| 353b xAxN 354b ooAx KNX 355b x(Ax)|Nx 356b ||w|An|N 357b xxAx|xN 358b llw w|xAx|xN 359b ||wx|w|(Ax)x|N 360b xx(Ax)=Nx 361b Ax|(Nx)x 362b ||Ax|Nx 363b Ax=Nx 364b ||w|Ax|N 365b ooxA|Bx 366b ||w w w|Ax|xN 367b o o A=N|Nn 368b Ax|Nx 369b | wx|xAx|N 370b (Ax)|N(n)x 371b A|Nnx 372b A=(N)nx 373b An|Nx

El (III) A 2ab (III) A2 (III) Al (III) Bl (la) Al (III) El (III) Al (III) Bl (II) A 3 (la) C 1 (la) Bl (la) D 4 1 (III) A2 (III) Al (III) A3 (la) Al (III) Al (III) A 3 (la) D* 2 (III) C 2 (la) Al (III) D 1 (III) Al (III) Al (III) C2 (la) A 3 (la) C 1 (la) Al (III) Bl(Ib) Al (III) D* 2 (III) A3 (la) B2 (II) C 1 (la)** Al (III) Bl (la) Al (III) C2 (II) Bl(Ia) A2 (III) C 1 (la) D* 2 (III) A 3 (la) C2 (la)

245

Al (III) B1(I) C 2 (I) B 1 (II) Al (III) C 1 (II) O{1) Oil) Al (III) Oil) Al (III) D 3 (III) D 4 1 (III) C'd) E2 (Ic) D 2 (III) D 1 (III) B1(I) Al (III) B1(I) A2 (III) Oil) Al (III) B1(I) Bl (II) A2 (III) C1 (II) B1(I) B2 (II) B2(I) Bl(I) C1 (II) Al (HI) Al (III) Al (III) B1(I) C1 (II) B2(I) A2a (III) Al (III) Bl(I) D2 (III) D 1 (III) D3 (III) Al (III)

Index 374a 375a 376a 377a 378a 379a 380a 381a 382a 383a 384a 385a 386a 387a 388a 389a

||w|xA|xB Ax|Nx ++ A|A|-(N)x ||wx|Ax|N ||w|w|A=Nx ax|xAx ++ Ax|(Ax)=N ++ (Ax)=N|Ax xA=(N)x xA=(N)x xAx|(Ax) xxA=(N)x ||w|w|x(Ax)x a|Ax=xN ||xwx|w|w|Ax (Ax)x|Nx

374b 375b 376b 377b 378b 379b 380b 381b 382b 383b 384b 385b 386b 387b 388b

Bl (la) Al (III) D2 (III) Bl(Ia) C1 (la) A3 (Ib) A2b (III) A2a (III) C2 (II) C2 (II) B 1 (II) C2 (II) A3 (la) B2 (Ib) A3 (la) Al (III) 3

A (la) D*2 (III) A3 (la) D 1 (III) A3 (la) C1 (II) Bl (la) D 1 (III) A3 (la) D*41(III) A2a (III) A3 (Ib) Bl (II) A2 (III) D*2 (III) A2k (III) Bl(Ia) A2b (III) A3 (Ib) B 1 (II) C1 (Ib) D 1 (III) A2b (III) Bl(II)** A3 (la) C1 (II) Al(III) Bl(Ia)

Al (III) D 1 (III) Al (III) C2(I) C2 (II) B1(I) Al (III)

Al (III) O{1) Al (III) C2(I)

389b supplied

390b 391b 392b 393b 394b 395b 396b 397b 398b 399b 400b 401b 402b

390a supplied

391a ||w|w|Ax 392a ++ Ax|A=(N)x 393a Hwlwlw|Ax 394a A=Anx 395a [|w|w|wx|Ax 396a xx(Ax)=Nx 397a ||wx|Ax=B 398a (Ax)|A=Nx 399a ||xa|w|xAx 400a + + A-N|Ax|=N 401a ++ (Ax)=N|Ax 402a Haxx[xAx 403a xx(Ax)x|A 404a ++A|xxAx 405a ||An|(Nx)(n)x 406a (Ax)=N|(A)x 407a ||w|w|An|A 408a ++ A|x(Ax)=N 409a xax|x(Ax)x 410a xxxAx=N 411a ||ax|A=Nn 412a (Ax)|Nnx 413a ++ Ax|xA=N 414a xx(Ax)x|(Ax) 415a ||w|w|w|xAx 416a xAnx 417a Ax|Nn 418a ||xw|w|(Ax)x|B

Ax|Bx ||w|x(Ax)x|N ||wx|Ax|(Nx) A=Nnx Ax|(Nx)x ||w|w|A(n)x xxA=(N)x ||wx|Ax|N oo A|xNx ||w|w|A|Nx|| ||w|xAx|N Ax|Nx ||w|A|{N} oo (Ax)|x(Nx)x ||w|w|w|A-(N)x oo

||A|Nx|-xN (Ax)=Nx|N ||w|w|xx(Ax)x|N xxA=Nx oo (Ax)|N-(N)x xxxA=x(Nx) oo An|xN o0 A|xBx oo Ax|xNx ||xwx|A|(N)x ||wx|A|(N)x ||w|w|xAx|xN ||w|A|N(n)x

D42 (III) El (III) Bl(I) C1 (II) D2 (III) B1 (II) El (III) Al (III) A2 (III) C2(I) C2(I) B2(I)

403b supplied

404b 405b 406b 407b 408b 409b 410b 411b 412b 413b 414b 415b 416b 417b 418b

246

||w|w|xAx|xN ||xw|Ax|N oo (Ax)|N-Nx ||w|w|(Ax)nx ||wx|w|Ax|(Nx) ||w|w|Ax|N A=Nx|N ||w|x(Ax)|Nx °Ax|xNx ||wx|Ax=N x(Ax)|Nx Ax|Nx (Ax)x|Nx oo ||w|w|A|Nx oo Ax|Bx

B2(I) B1(I) D 1 (III) O(l) B1(I) B1(I) El (III) C^I) A2 (III) B1(I). C1 (II) Al (III) Al (III) O(l) Al (III)

Alliteration, scansion and metrical clause structure 419a + Hk||Ax|xxAx 420a + +A|xAx 421a ||ax|(Ax)x|N 422a (Ax)x|Ax 423a ||a|(Ax)x|N 424a ||xa|(A)x 425a xxA=Nx 426a ++ A|xAx 427a (Ax)|A=(N)x 428a (Ax)Nnx 429a ||w|w w|xxAx 430a ++ A=(N)x|Ax 431a ||w|w|wx|Ax 432a °xAx|A 433a ||wx|w|w|xAx 434a xxA=Nx 435a ||\*i w|wx|xAx 436a xA=Nx 437a II*r w|A|(N)x 438a ++ (Ax)=N|xAx 439a a x Ax 440a + +A|xAx 441a Ax|Ax 442a + + | a|w||wJwjAx 443a xxA=(N)x 444a ax A=Nx 445a (Ax)|Nnx 446a (Ax)x|Ax 447a Ax|Nx 448a ||ax|Ax|N 449a ||ax|A=Nx 450a ||ax|A=(N)x 451a Ax|Nx 452a ||xw|(Ax)nx 453a (Ax)=Nx|A 454a Ax|Nx 455a + +Anx|xA 456a ||An|(Nx)(n)x 457a ||xxAx|N 458a ||w|xA=(N)x 459a ||xw|x(A)x 460a ||w|w|(Ax)nx 461a xAnx 462a x(Ax)=Nx 463a ||wx|w|xAx

A2 (III) Al (III) Bl(Ib) Al (III) Bl (Ib) A3" (Ib) C1 (II) Al (III) D2 (III) D 1 (III) A3 (la) A2 (III) A3 (la) Bl (II) A3 (la) C1 (II) A3 (la)

419b llw w|xAx|N 420b ||w w|Ax|xN 421b ||w|xAx|N 422b ||(Ax)=Nx|N 423b ||A|N(n)x 424b ||w|w xAx|N 425b ooAx|xNx 426b oo||w|A|Nx 427b Ax|Nx 428b Ax|Nx 429b Anx|N 430b llw w|xAx|N 431b xxxAx xN 432b (Ax)|N(n)x 433b ||w|xA=Nx 434b Ax|xNx 435b ||w|w|(Ax)n|N 436b Ax|Nx 437b xxAx|N 438b 1 l ww|xAx|N 439b ||w|xA|(N)x 440b ||w|xAx|N 441b ||w w|wx|A|(N)x 442b ||w|w|Ax|N|| 443b A(n)x|Nx 444b ||w|w|A|(N)x|| 445b oo||w|w|Ax|N 446b l|w|w w|Ax|(Nx) 447b iiw w|A|(N)x 448b ||Ax|Nx 449b A=Nnx 450b oo||w|wxAx|xN 451b A|N(n)x 452b ||w|w|A|(N)x|| 453b ||w|xxA|(N)x 454b |]w|w|Ax|N 455b oo||w w|A||w|w|N 456b ANnx 457b oo(Ax)|=N|Nn 458b ooAx|Nx 459b Ax|Nx 460b xA=(N)x 461b ||w|wx|(Ax)x|N 462b ooAx|xNx 463b A=(N)x|N

C 1 (ID

C2 (la) A2 (III) A3 (Ib) Al (III) Al (III) A3 (la) C2 (II) C1 (Ib) D 1 (III) Al (III) Al (III) Bl(Ib) C1 (Ib) C2 (Ib) Al (III) C1 (la) El (HI) Al (III) E2 (III) D*2 (III) Bl (II)** C2 (la) A3" (la) C1 (la) C1 (II) C1 (II) A3 (la)

247

Bid) B2(I) Bl(I) El (III) D2 (III)

Bid) A2 (III) C'd)

Al (III) Al (III) El (III) Bl (I) B2 (II) D2 (III) Cl(l) A2 (III) Bl (I) Al (III) Bl (II) Bl(I) C2(I) Bl(I) C2(I) Bl(I) A2 (III) C2(I) Bl(I) Bl(I) C2(I) Al (III) D 1 (III) B2(I) D2 (III) C2(I) C2 (I) Bl(I) B2(I) D 1 (III) A2a (III) Al (III) Al (HI) C2 (II) Bl(I) A2 (III) E2 (III)

Index 464a xxAx|xN 465a ||w|w|Ax|N 466a ||w|x(Ax)x|N 467a ++ A=N|(Ax)x 468a xAx|N 469a A|N(n)x 470a ||wx|xAx w|Anx 471a 472a Ax|Nx 473a + + ||A|w|w|xAnx 474a (Ax)x|Nx 475a ++Ax|x(Ax)x 476a ++ A=Nx|x(Ax) 477a ++ A=N|x(Ax) 478a xAx|(Ax) 479a xxA=(N)x 480a Hxw[xA(n)x 481a xx(Ax)=Nx 482a ||w|w|xA=(N)x 483a Ax|Ax 484a ||wx|w|x(Ax)=N 485a + + A=(N)x|A=N 486a ||w|A=(N)x 487a A|(Ax)=Nx 488a Ax|(Ax)x 489a ||ax|w|xAx 490a ++ (Ax)=B|Ax 491a ||w|w|A=Nx 492a xA=(N)x 493a ||w|A=Nx 494a Ax|Nx 495a ||w|w|xAx|N 496a ||ax|A|(N)x 497a ++Ax|x(Ax)x 498a (Ax)|nNx 499a ||An|(Nx)(n)x 500a |jw|xAx|N 501a ||xa|(Ax)=Nx 502a ++Ax|(Ax)=(N)x w[w|xAx 503a 504a wx|Ax|xA 505a xax|xx(Ax)x 506a ||w|w|xAn 507a xAx|A 508a ||w|w|xAx

B2 (II) Bl(Ia) Bl (la) A2a (III) B 1 (II) D2 (III) A3 (la) C1 (la) Al (III) D #1 (Ic) Al (III) A2 (III) E2 (III) El (III) Bl (II) C2 (II) C2 (la) C1 (II) C2 (la) Al (III) A3 (la) A2b (III) C2 (la) D 1 (III) Al (III) A3 (la) A2a (III) C1 (la) C2 (II) C1 (la) Al (III) Bl(Ia) C2 (Ib) A2 (III) Al (III) D*2 (III) Bl(Ia) C1 (Ib) D*2 (III) A3 (la) B2 (la)* A3 (Ib) A3 (la) Bl (II)A A3 (la)

464b A=Nnx 465b Ax|(Nx)x 466b Ax|Nx 467b ||w|w|(Ax)n|N 468b A=(Nx)nx 469b oo||w|w|(Ax)x||wxlN 470b A|N(n)x 471b xx(Ax)x|N 472b ||w|w|Ax|N 473b x(Ax)|Nx 474b ||w|w|Ax|(Nx) 475 b xx(Ax)=Nx 476b ||w|xA=(N)x 477b ||w|A|xN 478b oo||A|Nx|-N 479b ooAx|xNx 480b Ax|Nx 481b Ax=Nx 482b Ax|Nx 483b x(Ax)|Nx 484b xAx=N 485b ||wx|A|Nx 486b ooAx|xNx 487b ||wx|w|Ax|xN 488b ||w|w|A|xN 489b ||w|xA|(N)x 490b ||w|x(Ax)|Bx 491b oo(Ax)|xNx 492b ooA|xNx 493b Ax|Nx 494b oo||A|Nx|-xN 495b (Ax)|(Nx)=Nx 496b oo||A|Nx|-N 497b ||w|w|(Ax)x|N 498b °(Ax)|x(Nx)x 499b Anx|N 500b A|Nnx 501b ||w|w|Anx|N 502b oo(Ax)|N-Nx 503b ||w|xxAxN 504b Ax=Nx 505b oo||wx|A|Nx 506b ||w|w|x(Ax)|Nx 507b xA=(N)x 508b (Ax)|N(n)x

248

D 1 (III) Al (III) Al (III) B1(I) D 1 (III) B2(I) D2 (III) B 1 (II) B1(I) C1 (II) B1(I) C1 (II) C2(I) B1(I) D41 (III) A2 (III) Al (III) Al (III) Al (III) C1 (II) Bl (II) Cl(D A2 (III) B2(I) B1(I) C2(I) Cl(D Al (III) Al (III) Al (III) D42 (III) D 1 (III)** D41 (III) B1(I) Al (III) El (III) D 1 (III) B2(I) D 1 (III) B1(I) Al (III)

CM) Cl(l)

C 2 (II) D 2 (III)

Alliteration, scansion and metrical clause structure 509a 510a 511a 512a 513a 514a 515a 516a 517a 518a 519a 520a 521a 522a 523a 524a 525a 526a 527a 528a 529a 530a 531a 532a 533a 534a 535a 536a 537a 538a 539a 540a 54 la 542a 543a 544a 545a 546a 547a 548a 549a 550a 551a 552a 553a

||w|xA=Nx Ax|Nx xA|xA A=Nx|A ||w|w|Ax=N ||ax|(Ax)=Nx ||ax|xxA-N Ax|Ax ++ (Ax)|=N|Ax ||wx|Ax(Ax) x(Ax)=Nx l|wx[w|xAx ++ A|xAx ++ (Ax)=N|(Ax)x ++ A|xAx (Ax)|Nnx HwxlAxlwlxB ||w|w|(Ax)=Nx Ax|Ax A=Nx|N ||An|(Nx)(n)x ||w|w|A|(N)x Ax|Nx ||ax|xxAx ||w|w|(Ax)=Nx ++ A(n)x|xAx [|w|w[xAx ||w[xA(n)x x(Ax)=Nx Ax|Nx ||wx|A|(N)x ++ A|xAx (Ax)x|Nx A=Nx|A ax|xAx ||w|w|xAx A|=Nx|A (Ax)|Anx Anx|A ++ (Ax)=N|xA ||w|(Ax)=Nx ||w|w|xAx A|A=(N)x ++ ||(Ax)=N|Ax ++ Ax|xAx

C1 (la) Al (III) B 1 (II) El (III) Bl (la) C1 (Ib) A3 (Ib) Al (III) A2a (III) Bl(Ia) C1 (II) A3 (la) Al (III) A2a (III) Al (III) D 1 (III) B2 (la)** C1 (la) Al (III) El (III) D*2 (III) C2 (la) Al (III) A3 (Ib) C1 (la) A2 (III) A3 (la) C2 (la) C1 (II) Al (III) C2 (la) Al (III) Al (III) El (III) A3 (Ib) A3 (la) El (III) D 1 (III) El (III) El (III) C1 (la) A3 (la) D2 (III) A2a (III) A2 (III)

509b 510b 511b 512b 513b 514b 515b 516b 517b 518b 519b 520b 521b 522b 523b 524b 525b 526b 527b 528b 529b 530b 531b 532b 533b 534b 535b 536b

537b 538b 539b 540b 541b

542b 543b 544b 545b 546b 547b 548b 549b 550b 551b 552b 553b

249

xA|(N)x ||w|w|AxN xA|Nx ||w|w|xA|{N} Ax|Nx ||Ax|Nx oo ||(Ax)|Nx|-N ||w|x(Ax)x|N ||w|w|xAx|xxN ||w|wx|xAx=N oo A|N|-xN Ax|Nx A|Nnx ||w|w|A|Nx oo ||A|=N|xN oo Ax|xNx oo Ax|xBx xA|Nx ||w|w|Ax|N {A}|Nx A|Nnx oo (Ax)|=N|Nn x(Ax)|Nx ||A|w|(Nx)x oo Ax[Nx ||wx|xxAxN A=(N)nx ||wx|Ax|xN w|w|xA-N|N 3 ||w|w|xAx|N w|w|xA|{N}|| w|xA=Nx w|w|A|xN Ax|Nx °°||w|w|xA|Nx xA|Nx ||w|w|w|A|xN (Ax)x|Nx ||w|Ax=N ||A|wx|Nx oo A|xNx A=Nx|N oo Ax|x(Nx)x xAx|N ||w|xAx|N

C2 (II) B1(I) C1 (II) C^I) Al (III) Al (III) D41(III) B1(I) B2(I) B1(I) D41 (III) Al (III) D 1 (III) C^I) El (III) A2 (III) A2 (III) C1 (II) B1(I) Al (III) D1 (III) A2a (III) C1 (II) Al (Ic) Al (III) B1(I) D3 (III) B2(I) B1(I) Bl(I)

Bl(I) Al (HI)

Oil) c 1 (ii) Bid) Al (III) Bl(I) A2 (Ic) Al (III) El (III) A2 (III) B 1 (II) Bl(I)

Index 554a A|A=(N)x 555a ++ A|xAx 556a ||w|w|A=Nx 557a Ax=Nx 558a ++ Ax|(Ax)=N 559a ||w|w|xAx 560a axx|Ax 561a Ax|Nx 562a ||w|w|xxAx 563a A=xNx 564a ++ ||Ax|xAx 565a ||w|xAnx 566a xA=Bx 567a ++Ax|x(Ax)x 568a xAx|N 569a ++Ax|xAx 570a A|A[x}|=(N)x 571a ||w|w|A=Nx 572a A(n)x|Ax 573a A=Nx|A 574a [|wxx[wlxAx 575a (Ax)x|Ax 576a xx(Ax)x|A 577a ||w|xA=Nx 578a ||wxx|w|Ax|A 579a Ax|Nx 580a ++A|xx(Ax)x 581a (Ax)|Anx 582a xx(Ax)=Nx 583a Ax|Ax 584a x(Ax)=Nx 585a xA=Nx 586a Ax|Nx 587a ||w|w|xxAx 588a Ax=Nx 589a Ax|Bx 590a ||ax|w|w|xAx 591a ||w|wx|Ax|x(Bx) 592a (Ax)|A=Nx 593a ++Ax|x(Ax)x 594a + +(Ax)|x(Ax)=N 595a ||w[w[wx|xAx 596a ++(Ax)x|A=(N)x 597a ++Ax|xAx 598a ||ax|A=Nx

D2 (III) Al (III) C1 (la) Al (III) A2b (III) A3 (la) A3 (Ib) Al (III) A3 (la) Al (III) A2 (III) C1 (la) C1 (II) A2 (III) B 1 (II) A2 (III) D2 (III)** C1 (la) A2 (III) El (III) A3 (la) Al (III) Bl (II) C1 (la) Bl(Ia) Al (III) A2 (III) D 1 (III) C1 (II) Al (III) C1 (II) C1 (II) Al (III) A3 (la) Al (III) Al (III) A3 (la) B2 (la) D 1 (III) A2 (III) A2b (III) A3 (la) D*2 (III) A2 (III) C1 (Ib)

554b ||Ax|Nx 555b ||wx|w|(Ax)x|N 556b ooAx|xNx 557b oo||(Ax)=N|xN 558b xAxN 559b A=xNx 560b ooHwlwlA(n)x 561b ||w|w|xAx|N 562b xA|Nx 563b oo||w|w|A|Nx 564b A=Nx|N 565b Ax|Nx 566b Ax|Bx 567b ||w|Ax|N 568b A=Nnx 569b oo||A|Nx|-N 570b ||(Ax)|N(n)x 571b xA|Nx 572b oo||A|N|-(N)x 573b ||wx|xAx|N 574b ||w|w|xAx|A 575b ||w|w|xA|xN 576b Ax|Nx 577b Ax|Nx 578b ooAx|xNx 579b ||w|w|A|xN 580b xAx|N 581b ||w|w|A|xN 582b Ax|Nx 583b oo||(Ax)|Nx|-N 584b ||w|x(Ax)|Nx|| 585b ooA|x(Nx)x 586b ||w|w|w|(Ax)|Nx 587b x(Ax)|Nx 588b ||w|w|xAx|N 589b ||w|xA|(B)x 590b (Ax)|Nnx 591b oo(Ax)|x(Bx)x 592b Ax|Nx 593b ||w|x(Ax)|Nx 594b ||w|w|A|(N)x 595b ||w|w|xAx|xN 596b ooAx|Nx 597b (Ax)=Nnx 598b ooHA(n)xlNx

250

Al (III) B1(I) A2 (III) El (III) B 1 (II) Al (III) C2(I) B1(I) C1 (II)

On) El (III) Al (III) Al (III) Bl(I) D1 (III) D41 (III) D2 (III) C1 (II) D2 (III) B1(I) Bl (I)** B1(I) Al (III) Al (III) A2 (III) B1(I) Bl (II) Bl(I) Al (III) D41 (III) Oil) Al (III) O{1) O (II) Bl(I) C2(I) D 1 (III) Al (III) Al (HI) Cl{l). Oil) B2(I) Al (III)** D 1 (III) A2 (III)

Alliteration, scansion and metrical clause structure 599a 600a 601a 602a 603a 604a 605a 606a 607a 608a 609a 610a 611a 612a 613a 614a 615a 616a 617a 618a 619a 620a 621a 622a 623a 624a 625a 626a 627a 628a 629a 630a 631a 632a 633a 634a 635a 636a 637a 638a 639a 640a 641a 642a 643a

Ax|(Nx)x ++ ||(Ax)|xAx xA=(N)x ++ (Ax)|xAx °Ax|xNx x(Ax)|Ax xxAx|N ++ Ax|A=(N)x ||w|w|xAx ++ (Ax)=N|xA=N (Ax)|A=(N)x Ax|Nx ||w|w|(Ax)x|A[x] ++ ||Awx|A=(N)x A|Nnx ||ax|A=(B)x ||w|w|A-N|N ax|A=(N)x Ha wx|Ax Ax Ax ++

Ax|x(Ax)=N ||xAx]N °(Ax)x|x(Nx)x ++ ||A=(N)x|Ax ||w|w|Anx °Ax|xNx ||ax|Ax|N A-N|Ax llw w|xAnx (Ax)x|Nx A=N|(A)x llw w|A(n)x ||An|(Nx)(n)x ||w w|(Ax)x ++ ||A=N|xA ||w|w|Anx ++ Ax|xAx A=Nx|A A-N|Ax xxx(Ax)=Nx ||w|Ax|xA ++ A=(N)x|Ax ++ A=(N)x|A=N ||w|w|A|xA A=N|(N)x

Al (III) Al (III)

599b 600b 601b 602b 603b 604b 605b 606b 607b 608b 609b 610b 611b 612b 613b 614b 615b 616b 617b 618b 619b 620b 621b 622b 623b 624b 625b 626b 627b 628b 629b 630b 631b 632b 633b 634b 635b 636b 637b 638b 639b 640b 641b 642b 643b

C 2 (ID

Al (III) A2 (III) C1 (II) Bl (II) D*2 (III) A3 (la) A2b (III) D 2 (III) Al (III) Bl (la) D*2 (Ic) D 1 (III) C2 (Ib) Bl(Ia) C2 (Ib) A3 (la) Al (III) A2b (III) Bl(Ia) A2 (III) A2 (III) C1 (la) A2 (III) Bl (Ib) Al (III) C1 (la) Al (III) A2k (III) C2 (la) D*2 (III) A3 (la) El (III) C1 (la) A2 (III) El (III) Al (III) C 1 (II) B2 (la)** A2 (III) A2b (III) Bl (la) A2k (III)

251

||w|w|A|(N)x A> c xNx ||w w|w|Ax|N A=Nx|N oo ||Al-N||w|wlN|| |wx|Ax=N oo AxN[x}x Ax|Nx Ax|Nx oo ||Ax|xNx ||xwx xAnx oo A=Nx|xN ||A|N(n)x ||wx|An|N oo Ax|xNx oo (Ax)|xBx oo A|xNx Ax=Nx xxxA=(N)x ||w xA|xN (Ax)=N|(N)n oo (Ax)|Nnx OO A|N-Nx l|w w|A|xN A=(N)x|N oo (Ax)=N|xN ||(Ax)|N(n)x ||w|w wx|xAx xN oo A|xNx w|xA|xN xA{n} AxxNx A|Nnx ||w|w|xA|xN xxxAx|xN oo AxNx ||wx|xA|Nx ||w|xAx|N ||wx|Ax=N oo Ax|xNx A|N(n)x ||wx A=(N)x xxxA|Nx oo Ax|xNx oo A|xNx OO

C 2 (I) A2 (III) Bid) El (III) E2 (III)** Bl(I) Al (III) Al (III) Al (III) A2 (III) C^I) E2 (III) D2 (III) Bl(I) A2 (III) Al (III) Al (III) Al (III) C2 (II) Bl(I) A2k (III) D 1 (III) D 1 (III) Bl(I) E2 (III) El (III) D2 (III) B2(I) Al (III) Bl(I) C1 (II) A2 (III) D 1 (HI) Bl(I) B2 (II) Al (III) C^I) Bl(I) Bl(I) A2 (III) D2 (III) C2(I) C 1 (ID

A2 (HI) Al (III)

Index 644a (Ax)=Nx|A 645a (Ax)|N(n)x 646a Ax=Nx 647a xxA=(N)x 648a ||wx|w|Ax|N 649a ||w|w|Anx 650a ++(Ax)=Nx|x(Ax) 651a ++A|xxAx 652a ||xAx|N 653a An|Bn 654a + +A=Nx|xA 655a ||wx|w|AnxN 656a ||wx|w|A|xN 657a A=N|(N)x 658a + + ||(Ax)|-N|xxA 659a ||xax|Ax 660a ||ax|xAx 661a ||w|w|xAx=N 662a ||w|w|An|xN 663a (Ax)|Nnx 664a ||ax|A=(N)x 665a °A|xxNx 666a ++Ax|xAx 667a ++(Ax)=N|x(Ax) 668a xAx|(Nx) 669a ||wx|Ax|N 670a Ax|Ax 671a ||w|w|w|A|(N)x 672a ++A|x(Ax)x 673a Anx|N 674a ||w|xAx|A 675a w|xAx 676a An|Nx 677a ||w|w|w|x(Ax)=Nx 678a A=xNx 679a ||xw|w|wx|Ax 680a °||Ax|xNx 681a ||w|w|xxAx 682a °||A|xNx 683a A=xNx 684a ++Ax|xxAx 685a ++A|xxAx 686a xx(Ax)x|A 687a Ax|Nx 688a ||ax|wx|w|(Ax)=N

644b ||w|w|Anx 645b Ax|Nx 646b ||wx|xA=Nx 647b ooAx|xNx 648b xA|Nx 649b ooA|xxNx 650b Ax|Nx 651b oo||(Ax)|=N|xN 652b (Ax)|Nnx 653b ||w|w|A|xB 654b ||w|xA|xN 655b ooA|xNx 656b Ax|Nx|| 657b ||wx|A|Nx 658b Ax|Nx 659b ||(Ax)=Nx|N 660b ||xw|w|Ax|N 661b ooAx|xNx 662b xx(Ax)x|xN 663b ooA|xNx 664b An|Nx 665b ||wx|(Ax)=Nx 666b ||w|(Ax)|xNx|| 667b oo||A[x}=Nx|xN 668b oo||(Ax)=N|xN 669b Ax|(Nx)x 670b (Ax)x|Nx 671b Ax=Nx 672b ||wx|xAx|N 673b An=Nx 674b Ax=Nx 675b A=Nx|N 676b ||w|w|xA(N)x 677b Ax|(Nx)x 678b ||wx|Ax|(Nx) 679b Ax|Nx 680b w|A|(N)x 681b w|w|xA|{N} 682b |w|w|w|A|(N)x 683b |w|w|xA|(N)x 684b |w|w|xAx|N 685b |w|wx|Ax|N 686b Ax|Nx 687b |w|w|xA|Nx 688b °||A=Nx|xN

El (III) D2 (III) Al (III) C2 (II) Bl(Ia) C1 (la) E2 (III) A2 (III) Bl (la) Al (III) E2 (III) B2 (la) Bl(Ia) A2k (III) E2 (III) A3 (Ib) A3 (Ib) Bl(Ia) B2 (la) D 1 (III) C2 (Ib) A2 (III)** A2 (III) El (III) Bl (II) Bl (la) Al (III) C2 (la) Al (III) El (III)** Bl(Ia) A3 (la) Al (III) C1 (la) Al (III) A3 (la) A2 (III) A3 (la) Al (III) Al (III) A2 (III) A2 (III) Bl(II)** Al (III) A3 (la)

252

C1® Al (III) Cl(l) hi (III) C1 (II) A2 (III) Al (III) El (III) D 1 (III) B1(I) Bl(I) Al (III) Al (III) Cl(l) Al (III) El (III) B1(I) A2 (III) B2 (II) Al (III) Al (III)

Cl(D

C*(I) E2 (III) El (III) Al (III) Al (III) Al (III) Bl(I) Al (III) Al (III) El (III) C2(I) Al (III) Bl(I) Al (III) C2(I) C2(I) C2(I) C2(I) Bl(I) Bid) Al (III) Cl(l) E2 (III)

Alliteration, scansion and metrical clause structure 689a + + Ax|A-(N)x 690a + + A-N|A=B 691a ||wx|wx|Ax 692a a|A=(N)x 693a + + A|xxA=N 694a ||w w|wx|xAx 695a xxA=(N)x 696a (Ax)x|Nx 697a ++ A=Nx|x(Ax) 698a + +Ax|xAx 699a xAx|B 700a Ax|Nx 701a ||w|Ax|N 702a a|Ax=N 703a ax|(Ax)=Nx 704a ||w|xA=(N)x 705a ax xxAx 706a ||w w|xAx 707a xA=(A)x 708a ||w w Anx 709a a|Ax=N 710a w w xAx 711a Ax|Ax 712a ||ax|xA=(N)x 713a ax|xAx 714a ||a|xxAx 715a + +A=(N)x|(Ax)x 716a Ax|Ax 717a |w|w|Anx 718a wx|w|xAx=(Nx) 719a Ax Ax 720a llw w x(Ax)x 721a ++Ax|xAx 722a A=Nx|A 723a xa w|(Ax)=Nx 724a (Ax)x|Nx 725a xAx A 726a ||ax|Ax=N 727a ++Ax|xAx 728a ||xw w|x(Ax)x 729a ax Ax=xN 730a (Ax)=Nx|B 731a + + ||wx||w|w|xAx 732a (Ax)|A=Nx 733a + +A|xAx

D*2 (III) A2b (III) A3 (la) C2 (Ib) A2b (III) A3 (la) C2 (II) Al (III) E2 (III) A2 (III) B 1 (II) Al (III) Bl (la) Bl(Ib) C1 (Ib) C2 (la) A3 (Ib) A3 (la) C2 (II) C1 (la) Bl(Ib) A3 (la) Al (III) C2 (Ib) A3 (Ib) A3(Ib) s A2 (III) Al (III) C1 (la) Bl (la) Al (III) A3 (la) A2 (III) El (III) C1 (la) Al (III) Bl (II) Bl(Ib) A2 (III) A3 (la) B2 (Ib) El (III) A3 (la) D 1 (III) Al (III)

689b ||w wx|A|(N)x 690b oo(Ax)=Bx|xN 691b ||w|w|(Ax)|Nx 692b ooAx|xNx 693b ||w|w|xAx|N 694b oo||w|w|wx(Ax)|Nx 695b oo A=N|xN 696b | w wAxxN 697b (Ax)x|Nx 698b ||w w|A|(N)x 699b ooAx|xxBx 700b oo||A|w|xNx 701b Ax|Nx 702b ||w|xAx|N 703b ||An|Nx 704b Ax|Nx|| 705b llw w|Ax|N 706b ||w|(Ax)|Nx|| 707b xx(Ax) Nx 708b ooAx|xNx 709b ooAx|xNx 710b xxA=(N)x 711b ||(Ax)|=Nx|N 712b Ax|Nx 713b x(Ax)|xN 1 W w|A=(N)x 714b 715b Ax|Nx 716b ||x*^w|Ax|N 717b ooA|xNx 718b °A|xNx 719b A=Nx|N 720b A|N(n)x 721b oo||(Ax)|Nx|-xN 722b | |wx w|wx|Ax xN 723b ||w w|xAx|N|| 724b oo||(Ax)|xxN 725b A|N(n)x 726b wxAxN 727b A|nNn 728b Ax|(Nx)x 729b oo(Ax)|x(Nx)x 730b ||w xA xB 731b llw w|A|Nx|| 732b ooAx|xNx 733b ||w w|xAx|N

253

C2(I) E2 (III) Cl(l) A2 (III) B1(I)

Cl(D

El (III) B2(I) Al (III) C2(I) A2 (III) A2 (Ic) Al (III) B1(I) Al (III) Al (III) B1(I)

Cl(D

C1 (II) A2 (III) A2 (III) C2 (II) El (III) Al (III) B 1 (II) C2(I) Al (III) B1(I) Al (III) Al (III) El (III) D2 (III) D42 (III) B2(I) B1(I) F (III) D2 (III) B1(I) Al (III) Al (III) Al (III) B1(I) Cl(l) A2 (III) B1(I)

Index 734a A=Nx|A 735a ||w|w|A|Ax 736a ax|xxAN 737a A|(Nx)nx 738a xxA=(N)x 739a ]|w|w|xA=Nx 740a ||w|w|xA|(N)x 74 la Anx|N 742a ||a|A=(N)x 743a ||A=Ax|A 744a A=(Nx)nx 745a ++ A|xAx 746a ||w|w|xAx 747a ++ A|xAx 748a ++ A|xAx 749a An=Nx 750a ||wx|w[xAx 751a [|w|w|xAx 752a Ax|Nx 753a ++ A=(N)x|Ax 754a ++ A|xAx 755a + + ||(Ax)|w|w|A=B 756a ||wx|Ax|xA 757a ||wx|w|xAx=(Nx) 758a ||xax|w|xNx 759a Ax=Nx 760a ||w|w|Ax|xA 76 la + + ||(Ax)|w|A-N 762a ||ax|xAx 763a ++Ax|xAx 764a a|xA=(N)x 765a x(Ax)|Ax 766a ||w|xA=(N)x 767a + + ||A=(N)x|(Ax)x 768a Ax=Nnx 769a + +Ax|(Ax)=Nx 770a ++ Ax|A=Nx 771a ||w|w|Ax|(Nx) 772a xax|(Ax)=Nx 773a + + Ax|A=N 774a ++Ax|xAx 775a ++(Ax)=Nx|x(Ax) 776a (Ax)=N|(A)x 777a °Ax|xNx 778a ||w|xAx|N

734b ||xw|xA|xN 735b Ax|Nx 736b oo ||A=N|xN 737b ||w|xA=(N)x 738b x(Ax)|Nx 739b Ax|Nx 740b Ax|Nx 741b oo||A|nNx 742b oo||A|Nx|-N 743b ||Ax|Nx 744b ooA|xNx 745b oo||A|N|-xN 746b (Ax)=Nnx 747b oo||Ax|xN 748b oo||w|xA|(N)x 749b ||w|xA|xN 750b (Ax)x|Nx 751b Ax=Nx 752b xAx|N 753b ||w|xAx|N 754b ||w|xw|A|Nx 755b ||wx|xAx|B 756b ||xw|xAx|N 757b ooA|xNx 758b A|(Nx)nx 759b 0OA=N|xN 760b ||Ax|Nx 761b oo||A|Nx|-N 762b oo||w|w|Ax|N|| 763b ||w|xA|(N)x 764b ||wx|xAx|xN 765b ||w|w|Ax|N 766b x(Ax)x|xN 767b ||(Ax)|=Nx|N 768b ooAx|xNx 769b ||Ax|wx|Nx 770b ||(Ax)|N(n)x 771b ||w|xA=(N)x 772b ||w|w|xAx|xN 773b ||w|w|xAx|N 774b Ax=Nx 775b ||w|xAx|xN 776b °°AxxNx 777b ||w|x(Ax)|Nx 778b (Ax)|Nnx

El (III) C1 (la) A3(Ib) D 1 (III) C2 (II) C1 (la) C2 (la) El (III) C2 (Ib) El (III)* D 1 (III) Al (III) A3 (la) Al (III) Al (III) Al (III) A3 (la) A3 (la) Al (III) A2 (III) Al (III) A2b (Ic) B2 (la) Bl(Ia) F (la)** Al (III) B2 (la) Al (Ic) A3 (Ib) A2 (III) C2 (Ib) C1 (II) C2 (la) A2 (III)

Bl(Ia) C1 (Ib) A2b (III) A2 (III) E2 (III) A2k (III) A2 (III) Bl(Ia)

254

Al (III) El (III) C2(I) C1 (II) Al (III) Al (III) Al (III) D41 (III) Al (III) Al (III) D41(III) D1 (III) F (III) C2(I) B1(I) Al (III) Al (III) B 1 (II) B1(I)

Al (III) D 1 (III) El (III) Al (III) C2(I) B2(I) B1(I) B2 (II) El (III) A2 (III) A2 (Ic) D2 (III) C2(I) B2(I) B1(I) Al (III) B2(I) A2 (III)

cM)

D 1 (III)

Alliteration, scansion and metrical clause structure 779a 780a 781a 782a 783a 784a 785a 786a 787a 788a 789a 790a 791a 792a 793a 794a 795a 796a 797a 798a 799a 800a 801a 802a 803a 804a 805a 806a 807a 808a 809a 810a 811a 812a 813a 814a 815a 816a 817a 818a 819a 820a 821a 822a 823a

llwhv w|xx(A)x

+ +

A-N|xA=N ++ ||Ax|xAx ax|x(Ax)x ++ Ax|xAx (Ax)-N|(Ax)x ||wx w|xAx ||(Ax)=N|(A)x (Ax)=Nx A Ax|Ax ||w|w|Ax|N xA(N)x ||wx|Ax|N xxA=(A)x ||w|xA=(N)x Ax|Nx A|Nnx ||wx A=Nx Ax|Nx ||w|w|xAx A=Anx |w|xAx|x(Ax) ||Ax|Ax ax xxAx A=Nx|B ||w|w|(Ax)=Nx °Ax|xNx • xA(N)x A-N|Nx xAx|xN ||w|w|xAx Ax|Ax ++ (Ax)x|x(Ax)x ||w|w|xA=(N)x ||w|wx|xA[x}x ax xAx (Ax)nx|A (Ax)|A=Nx A=N|(A)x ||ax|A=(N)x ++ A=N|(Ax)x ++

A=N|{A}

||wx|A-N|A ||w|xAx|N A(n)x|A=N

A 3 ' (la) A2b (III) A2 (III) A3 (Ib) A2 (III) Al (III) A3 (la) A2k (III) El (III) Al (III) Bl (la) C2 (II) Bl (la) C2 (II) C2 (la) Al (III) D 1 (III) C1 (la) Al (III) A3 (la) D 1 (III) B2 (la) Al (III) A3 (Ib) El (III) C1 (la) A2 (III) C2 (II) Al (III) B2 (II) A3 (la) Al (III) A2 (III) C2 (la) A3 (la) A3 (Ib) El (III) D 1 (III) A2k (III) C2 (Ib) A2a (III) A2a (III) Bl (la) Bl (la) A2b (III)

779b 780b 781b 782b 783b 784b 785b 786b 787b 788b 789b 790b 791b 792b 793b 794b 795b 796b 797b 798b 799b 800b 801b 802b 803b 804b 805b 806b 807b 808b 809b 810b 811b 812b 813b 814b 815b 816b 817b 818b 819b 820b 821b 822b 823b

255

Ax|Nx x(Ax)|Nx ||wx|Ax|N oo ||A|N|-xN ||A=(N)x|N oo Ax|xNx oo A|xNx oo (Ax)|N-(N)x ||A|N(n)x oo ||A|wx|Nx (Ax)x|Nx oo AxNx oo A(n)x|Nx oo Ax|xNx Ax|N[x}x ||w xAx N Ax|Nx A|N(n)x oo ||ww|Ax|N ||w|w|xA|(N)x Ax=Nx Ax|Nx ||wx|A=(N)x Anx|N Ax|Bx x(Ax)|Nx ||wx|xAx=xN oo AxNx ||w|xAx=N A|N(n)x oo ||w|w|(Ax)|Nx Ax|Nx ||w|A|xN Ax|Nx A|(Nx)nx ||w|x(Ax)|Nx oo ||A=N|xN ||w|xAx|N oo ||(Ax)x|xNx ||Anx|N ||wx|Ax|(Nx) xxA=(N)x oo ||Ax|xNx oo Ax|xNx ||(Ax)|=Nx|N

Al (III) C1 (II) Bl (I) D41(IIJ E2 (III) A2 (III) Al (III) D 2 (III) D 2 (III) A2 (Ic) Al (III) Al (III) A2 (III) A2 (III) Al (III) Bl (I) Al (III) D 2 (III) Bl (I) C 2 (I) Al (III) Al (III) C2 (I)** El (III) Al (III) C1 (II) B2(I) Al (III) Bl (I) D 2 (III)

Cl(D

Al (III) Bl(I) Al (III) D 1 (III) C^I) El (III) Bl(I) A2 (III) El (III) Bl(I) C2 (II) A2 (III) A2 (III) El (III)

Index 824a xxxA=Nx 825a ||wx|w|xAx 826a + + (Ax)|xA=N|| 827a xax xAx 828a Ax=Nx 829a A=Nx|B 830a ||wx|A-Nx 831a An=Nx 832a ||w|xA=Nx 833a A|nNx 834a ||wx|Ax=N 835a ++ A|xAx 836a Ax|Ax|| 837a ||w|w|xAx 838a xxA=Nx 839a ||ax|A=(N)x 840a xA=(A)x 841a Ax|Ax 842a A-N|Nx 843a llwx w|A=Nx 844a ||w|w|Ax=N 845a Ax|xx(Nx) 846a + +Ax|xxAx 847a ||w|w|xAx 848a + + ||(Ax)|Ax|=xN 849a Ax|Ax 850a ||A=Nx|A 851a xA=(A)x 852a A(n)x|Nx 853a ||wx|A|x(Nx) 854a ||wx|A|(N)x 855a x(Ax)|Ax 856a ++ Ax|xAx 857a Ax|Ax 858a ||wx|A|xN 859a xxAx=N 860a xxAx|xN 861a A=Nnx 862a llwhv wx|(Ax)=Nx 863a Ax|Nn 864a ||ax|(Ax)=Nx 865a xxA|(A)x 866a ||w|w|A=(N)x 867a Ax^ ix 868a (Ax)|A=(N)x

C1 (II) A3 (la) A2b (III) A3 (Ib) Al (III) El (III) C1 (la) Al (III) C1 (la) Al (III) Bl (la) Al (III) Al (III) A3 (la)

824b ooAx|xNx 825b ||w|w|w|Ax|N 826b (Ax)|Nnx 827b oo ||A=Nx|xN 828b ||wx|A=(N)x 829b ooA|xBx 830b ooAx|xNx 831b ||w w|A|(N)x 832b (Ax)x|Nx 833b ||w|w|Ax|(Nx) 834b ooA|xNx 835b ||w|w|A|(N)x 836b xxAx|N 837b ooAxxNx 838b =N|(N)x 839b °Ax|x{N} 840b Ax|N(n)x 841b ||w|xA=xN 842b Ax|N[x}x 843b (Ax)|N(n)x 844b xA|(N)x 845b x(Ax)x|(Nx) 846b A=Nx|N 847b A|Nnx 848b ooA|xNx 849b (Ax)=Nx|N 850b ||wx|Ax|N 851b ooA|xNx 852b ||w|w|A|xN 853b A=xNx 854b x(Ax)=Nx 855b Ax|Nx 856b ||w|w|Anx 857b oo||(Ax)|N|-xN 858b xA|Nx 859b ooAx|Nx 860b Ax|Nx 861b Ax|Nx 862b ooA|xNx 863b ||w w|w|A|(N)n 864b Ax|Nx 865b Ax|Nx 866b (Ax)x|Nx 867b ||wx|(Ax)x|N 868b ooAx|xNx

C 1 (ID

C2 (Ib) C2 (II) Al (III) Al (III) C1 (la) Bl(Ia) F (HI)** A2 (III) A3 (la) D42 (III) Al (III) El (III) C2 (II) A2 (III) Bl(Ia) C2 (la) C1 (II) A2 (III) Al (III) Bl(Ia) Bl (II) B2 (II) D 1 (HI) C1 (la) Al (III) C1 (Ib) C2 (II) C2 (la) Al (III) D 2 (III)

256

A2 (III) Bl (I) D 1 (III) E2 (III) C 2 (I) Al (III) A2 (HI) C 2 (I) Al (III) Bl(I) Al (HI) C 2 (I) B 1 (II) A2 (III) A2k (III) A2 (HI) D*2 (III) Bl(I) Al (III) D 2 (III) C2 (II) B 1 (II) El (III) D 1 (III) Al (III) El (HI) Bl(I) Al (III) Bid) Al (HI) C1 (II) Al (III)

CM)

D41(IH) C1 (II) Al (HI) Al (III) Al (HI) Al (III) C 2 (I) Al (III) Al (HI) Al (III) Bl(I) A2 (III)

Alliteration, scansion and metrical clause structure 869a II*|w|A-(N)x 870a °A|xNx|| 871a °Ax|xNx 872a A|Nnx 873a ||w|xA|(N)x 874a ||Ax|Ax 875a ||w|w x(Ax)nx 876a Ax=Nx 877a + +Anx|xA 878a ||wx|w|(Ax)x|N 879a + +Ax|x(Ax)x 880a ||wxw|Ax|N 881a °{A}|x(Nx) 882a xAx|xN 883a ||wx|A-(N)x 884a + +Ax|xAx 885a xxA=(A)x 886a ||wx|Ax|N 887a Ax|Ax 888a (Ax)nx|N 889a Ax|Nx 890a ||wx|w|xAx 891a A=Nx|A 892a A-N|Nx 893a ||wx|A=Nx 894a ||w w|A=Nx 895a Ax|Nx 896a ||a|xA|(N)x 897a Ax|(Nx)x 898a ||w|w|A(n)x 899a xxA=Nx 900a Ax=Nx 901a ||wx|(Ax)nx 902a + +(Ax)|xAx 903a xAx|xN 904a ++ ||Ax|xAx 905a axx|xAx 906a xx(Ax)nx 907a ||wx|A|xB 908a A=Nx|A 909a llw w|w|Ax|N 910a ||w|xAx|N 911a + + ||(Ax)=(Nx)x|xA 912a + + A|xA=N 913a Ax|Nnx

C2 (la) Al (III) A2 (III) D 1 (III) C2 (la) Al (III) C1 (la) Al (III) E2 (III) Bl (la) A2 (III) Bl(Ia) F (III) B2 (II) C2 (la) A2 (III) C2 (II) Bl(Ia) Al (III) El (III) Al (III) A3 (la) El (III) Al (III) C1 (la) C1 (la) Al (III) C2 (Ib) Al (III) C2 (la) C1 (II) Al (III) C1 (la) Al (III) B2 (II) A2 (III) A3 (Ib) C1 (II) Bl(Ia) El (III) Bl(Ia) Bl (la)** E2 (III) A2b (III) D^dll)

869b A=x(Nx)x 870b A|=Nx|N 871b oo||A|N|-xN 872b Ax|(Nx)x 873b ooA|xNx 874b oo||A-N|xN 875b Ax|Nx 876b A=Nx|(Nx) 877b Ax|Nx 878b ooAx|xNx|| 879b | wx|(Ax)x|x(Nx) 880b Ax|Nx 881b ||w|w|A|Nx 882b A=xNx 883b (Ax)x|Nx 884b oo||(Ax)nx|xN 885b A|nNx 886b °°A|xNx 887b ||w|xxAx|N 888b ooAx|xNx 889b ||xw[w|(Ax)x|N 890b | w|xA|xN 891b ||w|w|xAx|xN 892b oo||(Ax)|Nx|-N 893b ooAx|xNx 894b Ax|Nx 895b oo ||A=N|xN 896b Ax|Nx 897b oo||A|N|-xN 898b Ax|Nx 899b Anx|N 900b ||w|w|A|xN 901b A|N(n)x 902b ||w|xA(n)x|N 903b ooA|xNx 904b ||wx|A=Nx 905b ||w|xAx|N 906b xAx=(Nx) 907b Ax|Bx 908b (Ax)|=N|(N)x 909b ooAx|xNx 910b xA|Nx 911b oo||A|xNx 912b (Ax)x|Nx 913b oo||w|w|Ax|N

257

Al (III) El (III) D41 (III) Al (III) Al (III) El (III) Al (III) El (III) Al (III) A2 (III) B2(I) Al (III) O{\) Al (III) Al (III) E2 (III) Al (III) Al (III) B1(I) A2 (III) B1(I) B1(I) B2(I) D41 (III) A2 (III) Al (III) •El (III) Al (III) D41(III) Al (III) El (III) Bl(I) D2 (III) B2(I) Al (III) C^I) B1(I) Bl (II) Al (III) A2k (III) A2 (III) C1 (II) Al (III) Al (III) Bl(I)

Index 914a 915a 916a 917a 918a 919a 920a 921a 922a 923a 924a 925a 926a 927a 928a 929a 930a 931a 932a 933a 934a 935a 936a 937a 938a 939a 940a 941a 942a 943a 944a 945a 946a 947a 948a 949a 950a 951a 952a 953a 954a 955a 956a 957a 958a

A|(Nx)nx ++ Ax|xAx ||wx|Anx Ax|Ax ++ (Ax)|xAx A=Bnx (Ax)=Nx|A xA=Ax axx|A-N ++ Ax|xAx (Ax)-Nx|A ||An|(Nx)(n)x ||a|x(Ax)x Ax|Nx ||wx|A-Nx ++ Ax xAx ++ Ax|xAx ++ Ax|xxAx ||w|w|A=Nx ++ Ax|xAx ++ AxxAx Ax|Nx ||A|A=(N)x ||wx|w|xAx +T Ax|A=xB ++ AxxAx xAx|N |.|w|w|Ax ++ Ax|xAx ||wx|xw|Ax xxA=Nx ||wwx|A=(N)x A=xAx A|Nx ax|xAx Ax|Nx (Ax)x|Ax ||xw|w|xAx A=Nnx ++ Ax|xAx °Ax|x(Nx) ax|xAx ++ Ax xAx ||An|(Nx)(n)x ||w|xAx=N

D 1 (III) A2 (III) C1 (la) Al (III) Al (III) D 1 (III) El (III) C1 (II) A3 (Ib) A2 (III) El (III) D*2 (III) A3 (Ib) Al (III) C1 (la)** A2 (III) A2 (III) A2 (III) C1 (la) A2 (III) A2 (III) Al (III)

914b Ax|Nx 915b ||wx|(Ax)|xN 916b Ax|Nx 917b ||w|w|Ax=N 918b ||wx|A|(N)x 919b x(Ax)|xB 920b ||wx|A|(N)n 921b A=Nx|N 922b x(Ax)|Nx 923b ||w|xA|xN 924b Ax|Nx 925b ||w|xAx|N 926b ||xw|Ax|N 927b xAx|N 928b A-Nx|N 929b ||wx|w|Ax|xN 930b ||w w|A|Nx 931b Ax|Nx 932b oo||w|w|Anx|N 933b xAx|(Nx) 934b ||wx|Ax|N 935b (Ax)=Nx|N 936b °(Ax)x|xNx 937b ||w|w|Ax=N 938b ooAx|x(Bx)x 939b ||w|A|(N)x 940b ooA|x(Nx)x xNx 941b 942b llw w|Ax|N 943b | w|xx(Ax) Nx 944b ||w|w|A|(N)x 945b Ax|Nx 946b ||w|w|An|N 947b w|x(Ax)|Nx 948b ||w|A|(N)x 949b oo||wx|w|Anx|N 950b ||w wxANx 951b A|N(n)x 952b Ax|Nx 953b ||w|w|A|(N)x 954b ||w|xA|(N)x 955b ||A-Nx|N 956b ||w|wxA|(N)x 957b A|Nnx 958b AxlNx

D 2 (III)

A3 (la) D*41 (III) A2 (III) B 1 (II) A3 (la) A2 (III) A3 (la) C1 (II) C2 (la) Al (III) F (III) A3 (Ib) Al (III) Al (III) A3 (la) D 1 (III) A2 (III) F (III) A3 (Ib) A2 (III) D*2 (III) Bl(Ia)

258

Al (III) B1(I) Al (III) B1(I) C 2 (I) B 1 (II) C 2 (I) El (III) C1 (II) B1(I) Al (III) B1(I) B1(I) B 1 (II) El (III) B2(I) C^I) Al (III) B2(I) B 1 (II) B1(I) El (III) A2 (III) B1(I) A2 (III) C 2 (I) Al (III) Al (III) Bl(I) C^I) C 2 (I) Al (III) Bl(I) C1 (I)** C 2 (I) B2(I) Cl(l) D2 (III) Al (III) C 2 (I) C 2 (I) El (III) C 2 (I) D 1 (III) Al (III)

Alliteration, scansion and metrical clause structure 959a Ax|(Ax)x 960a (Ax)|A=Nx 961a ||w|w|wx|Ax ++ 962a A|xAx 963a ||w|wx|A=Nx 964a" xA=Nx 965a ||w|w|xA=(N)x 966a ax|A=(N)x 967a ||w|wx|xAx 968a °Ax|xNx 969a A—xNx 970a + +A|xAx 971a xA=(B)x 972a + +A|xAx 973a A=N|(N)x 974a ||w|xA|(A)x 975a + +Ax|xAx 976a xA=(N)x 977a Ax|Ax 978a ++ (Ax)|Ax|=N 979a ||w|w|A|(N)x 980a ||w|w|Ax|A 981a xA=Nx 982a ||wx|(Ax)nx 983a xxAx|A 984a Ax|Ax 985a Ax =Nx|xN 986a + +A[x}x|A=(N)x 987a A|nNx 988a |w w|Ax|N 989a + +Ax|A-N 990a ++ Ax|(Ax)=Nx 991a ||w|w|Ax|(Ax) 992a + +Ax|xAx 993a + +(Ax)|xAx 994a + +A=(N)x|(Ax)x 995a ++A|xxAx 996a °Ax|xNx 997a °||w|xAx|A 998a xAx-N 999a + +Ax|x(Ax)x 1000a ax|A-N 1001a (Ax)=Nx|A 1002a + +Ax|A-Nx 1003a xx{A}x

959b 960b 961b 962b 963b 964b 965b

oo

||Ax|xNx ||Ax|w|Nx xA|Nx A=Nnx Ax|Nx Ax|Nx oo Ax|Nx 966b ||wx|xA|(N)x 967b ||w|(Ax)|Nx|| 968b ||w|w|w|xAx|xN 969b ||w|x(Ax)=Nx 970b ||wxx|w|xAx|xN 971b A|B(n)x 972b oo||w|w|A[x]x|xN 973b ooAx|xNx 974b A=xNx 975b ||w|wx|A|(N)x 976b ooAx|xNx 977b ||w xAx|N 978b ooAxNx 979b Ax|Nx 980b (Ax)|Nnx 981b A=xNx 982b Ax|Nx 983b A|N(n)x 984b oo||(Ax)|Nn|-N 985b ooAx|xNx 986b Ax=Nx 987b oo||A-N|xN 988b Ax|Nx 989b ||w xA=Nx 990b x(Ax)|Nx 991b ooA|Nx-N 992b oo||(Ax)|-Nx|N 993b ||w|xA=(N)x 994b ||A=N|(N)x 995b A[x}=Nx|(Nx) 996b oo||wx|wxA|(N)x 997b x(Ax)|Nx 998b A[x]=Nx|N 999b oo ||A|=Nx|xN 1000b ||w|xA=Nx 1001b xA|xN 1002b ||w|w|Ax|N 1003b oo| |Ax||ww|Nx

Al (III) D1 (III) A3 (la) Al (III) C1 (la) C1 (II) C2 (la) C2 (Ib) A3 (la) A2 (III) Al (III) Al (III) C2 (II) Al (III) A2k (III) C2 (la) A2 (III) C2 (II) Al (III) D41 (III) C2 (la) Bl(Ia) C1 (II) C1 (la) Bl (II) Al (III) E*2 (III) D*2 (III) Al (III) Bl(Ia) Al (III) D #1 (III) Bl (la) A2 (III) Al (III) A2 (III) A2 (III) A2 (III) Bl (la) B 1 (II) A2(III) A3 (Ib) El (III) D #1 (III) C1 (II)

259

oo

A2(III) A2(Ic) Cl(U) D 1 (III) Al (III) Al (III) Al (III) C2(I) Cl(l) B2(I) Cl(l) B2(I) D2 (III) B2 (I) A2 (III) Al (III) C2(I) A2 (III) Bl (I) Al (III) Al (III) D1 (III) Al (III) Al (III) D2 (III) D41(III)** A2 (III) Al (III) El (III) Al (III)

c1®

C1 (II) D41 (III) El (III)** C2(I) A2k (III) El (III) C2(I) C1 (II) El (III) E2 (III) Cl(l)

B 1 (II) Bl (I) A2 (III)

Index 1004a ||w|xAx|N 1005a + + Ax xAx 1006a A=Nnx 1007a ||w|xA=(N)x 1008a ax xxAx 1009a ||w|xAx|N 1010a ||wx|A|(N)n 1011a ||xxwx w|xAx 1012a xxxA=(N)x 1013a ||ax|w|xAx 1014a + + Ax|xAx 1015a ||(Ax)=N|(A)x 1016a A=Bnx 1017a + + An|xAn 1018a + + Ax xAx 1019a A=Nnx 1020a ||xw w|Anx 1021a (Ax)|Nnx 1022a ++ (Ax)|Ax=N{x} 1023a + + ||Ax|Ax=B 1024a xxxA|(A)x 1025a + +A|xAx 1026a xAnx 1027a |xxa|w|A=Nx 1028a ++Ax|x(Ax)x 1029a x(Ax)=Nx 1030a °||xw|Ax|A 1031a + +Ax|xAx 1032a ||w|w|Ax|N 1033a + + A=N|Ax 1034a xA|(A)x 1035a | w w|Ax|N 1036a Ax=Nx 1037a a|xx(Ax)x 1038a + +(Ax)|Ax = N 1039a ! w|w|Ax=N 1040a ||wx Ax|xN 1041a Ax|Nx 1042a A=Nx A 1043a ||w wAnx 1044a (Ax)|A(n)x 1045a ++ Ax|xAx 1046a ||w|A=Nx 1047a ++ A=N|(Ax)x 1048a + + Ax|xAx

1004b A=nx 1005b Ax|Nx 1006b AxNx 1007b (Ax)=Nx|N 1008b ||w|w|A|xN 1009b A(n)x|(Nx) 1010b Ax|Nx 1011b ooAx(Nx)x 1012b oo A|xNx 1013b A = N n x 1014b oo||(Ax)x|xNx 1015b Ax|Nx 1016b x(Ax)|xB 1017b oo ||(Ax)|Nx|-N 1018b ||wx|Ax=(Nx) 1019b Ax|(Nx)x 1020b A|N(n)x 1021b oo(Ax)x|xNx 1022b oo A|xNx 1023b oo(Ax)x|xBx 1024b oo ||An|xN 1025b ||w w|xxA=Nx 1026b (Ax)x|Nx 1027b Ax|Nx 1028b A=Nx|(Nx) 1029b oo Ax|xNx 1030b Ax=Nx 1031b oo (Ax)|Nx|-N 1032b ooAx xNx 1033b wx A=(N)x 1034b Ax|Nx 1035b AxNx 1036b xA|{N} 1037b ||wx|Ax|N oo 1038b Ax|xNx 1039b A=(N)nx 1040b (Ax)|N(n)x 1041b ||wx|xAx|N 1042b ||wx|(Ax)|Nx oo 1043b Ax|xNx oo 1044b A-N|xN 1045b w wx|A|Nx 1046b Ax|Nx 1047b (Ax)=Nx|N 1048b ||w|w|wx|A|{N}

Bl(Ia) A2 (III) D 1 (III) C 2 (la) A 3 (Ib) Bl(Ia) C2 (la) A3 (la) C2 (II) A 3 (la) A2 (III) A 2k (III) D 1 (III) A2 (III) A2 (III) D 1 (III) C 1 (la) D 1 (III) D 4 1 (III) D* 4 1 (III) C 2 (II) A l (III) C 1 (II) C 1 (la) A2 (III) C 1 (II) Bl (la)** A2 (III) Bl(Ia) A2a (III) C 2 (II) Bl (la) Al (III) A 3 (Ib) D 4 1 (III) Bl (la) B2 (la) Al (III) El (III) C 1 (la) D 2 (III) A2 (III) C 1 (la) A2a (III) A2 (III)

260

D 3 (III) Al (III) Al (III) El (III) Bl (I) E2 (III) Al (III) Al (III) Al (III) D 1 (III) A2 (III) Al (III) B 1 (II) D41(111 Bl (I) Al (III) D 2 (III) A2 (III) Al (III) A2 (III) El (III) C^I) Al (III) Al (III) El (III) A2 (III) Al (III) D 4 1 (III) A2 (III) C 2 (I) Al (III) Al (III) C1 (II) Bl (I) A2 (III) D 3 (III) D 2 (III) Bl (I) C1 (I) A2 (III) El (III) C^I) Al (III) El (III) C 2 (I)

Alliteration, scansion and metrical clause structure 1049a ||w|w|Ax|(Nx) 1050a ||xw|A-Nx 1051a ||wx|w|xAnx 1052a xxx(Ax)=Nx 1053a Ax=Nx 1054a ++Ax|xAx 1055a °Ax|xNx 1056a ||wx|w|Ax|N 1057a °||w|xAx|A 1058a (Ax)x|Nx 1059a ||xw|w|A-N 1060a ++ Ax|(Ax)=N 1061a ++Ax|xAx 1062a xxxA=(N)x 1063a ||w|w|A|xA 1064a xxAnx 1065a + + ||(Ax)=(B)x|Ax 1066a ||wx|A=(N)x 1067a xx(Ax)=Nx 1068a ||Ax|(Nx)x 1069a (Ax)|A=(N)x 1070a xA=(N)x 1071a ||w|wx|Axn 1072a A(n)x|Nx 1073a xax|Ax 1074a ++Ax|xAx 1075a Ax|Nx 1076a ||wx|Anx 1077a ++ (Ax)=N|xA 1078a ||w|w|xxAx 1079a ++A[x}=(N)x|Ax 1080a (Ax)x|Ax 1081a Ax|Nx 1082a HwlwlxAx 1083a A|Nnx 1084a ||w|xA=Nx 1085a Ax|Ax 1086a ||w|w|w|AxN 1087a ++ Ax|xA=N 1088a xA(n)x|N 1089a ||w|xA=Nx 1090a °Ax|xNx 1091a ||Anx|A 1092a HwxlxAx 1093a Ax|Nx

1049b ooA|xxNx 1050b Ax|Nx 1051b A=Nx|N|| 1052b ooAx|xNx 1053b ||w|xxAx|N 1054b ||wx|w|Ax|N 1055b ||w|w|wx|A|Nx 1056b ooA|xNx 1057b oo||(Ax)|Nx|-N 1058b ||w|w|xA|{N} 1059b A-N|Nx 1060b oo||(Ax)|w|xNx 1061b ||w|w|Ax|N 1062b (Ax)x|Nx 1063b oo(Ax)|x(Nx)x 1064b Ax=Nx 1065b oo||A|N|-(B)x 1066b Anx|N 1067b Ax|Nx 1068b ||w|w|xA|xN|| 1069b A|Nnx 1070b Ax|Nx 1071b (Ax)x|Nx 1072b ||A=Nx|N 1073b xxA=(N)x 1074b ||w|xxA|(N)x 1075b ||w|w|A(n)x|(Nx) 1076b Ax|Nx 1077b ||wx|Ax|N 1078b xA|Nx 1079b oo||w|w|w|Ax|N 1080b oo||A|Nx|-xN 1081b xxA[x}|Nx 1082b xxA{x} = (N)x 1083b ooA|xNx 1084b ooAx|xNx 1085b ||w|w|w|xAx|(Nx) 1086b ooA|xNx 1087b ||w|w|Ax|xN 1088b Ax|Nx 1089b Anx|(Nx) 1090b (Ax)|N(n)x 1091b Ax|(Nx)x 1092b A=xNx 1093b ||w|w|A(n)x|N

Bl(Ia) C1 (la) C1 (la) C1 (II) Al (III) A2 (III) A2 (III) Bl(Ia) Bl(Ia) Al (III) A3 (la) A2b (III) A2 (III) C2 (II) Bl (la) C1 (II) A2 (III) C2 (la) C1 (II) Al (III) D2 (III) C2 (II) Bl(Ia) A2 (III) A3 (Ib) A2 (III) Al (III) C1 (la) El (III) A3 (la) A2 (III) Al (III) Al (III) A3 (la) D 1 (III) C1 (la) Al (III) Bl(Ia) A2b (III) B2 (II) C1 (la) A2 (III) El (III) A3 (la) Al (III)

261

A2 (III) Al (III) El (III) A2 (III)

Al (III) D41 (III) C1® Al (III) A2 (Ic) B1(I) Al (III) Al (III) Al (III) D2 (III) El (III) Al (III) B1(I) D 1 (III) Al (III) Al (III) El (III) C2 (II) C2(I) B2(I) Al (III) B1(I) C1 (II) Bl(I) D42 (III) C1 (II) C2 (II) Al (III) A2 (III) Bl(I) Al (III) B2(I) Al (HI) El (III) D2 (III) Al (III) Al (III) B2(I)

Index 1094a >cA=(N)x 1095a | Iw w|x(Ax)x 1096a "h+ Ax|(Ax)=Nx 1097a *l " + AxA=Nx 1098a | |w w|xA=Nx 1099a Ax|Nx 1100a "h+Ax|xAx 1101a |w|xAn=(Nx) 1102a | |w|w|xxA=(N)x 1103a Ax=Nx 1104a | |w|wx|Ax|N 1105a xAx=(Nx) 1106a | |w:K w|Ax|N 1107a H h+ ||A|w|xAx 1108a xax|xAx 1109a A|(Ax)=Nx 1110a | |xw|Ax|N 1111a H h + A=N|Ax 1112a H h+ (Ax)|Ax=N 1113a H h+Ax|xAx 1114a | | a i sv|Axn 1115a xxAx|(Ax) 1116a H h+ ||A=(N)x|Ax 1117a H h+Ax|xAx 1118a | axx|Ax 1119a | a|xAx 1120a | axx|xAx 1121a + + ||A=(N)x|Ax 1122a + + A=(N)x|Ax 1123a Ax|Ax 1124a Ax|Nx 1125a | xax|w|w|An 1126a ++ Ax|xAx 1127a + + Ax|xA=N 1128a A=Bx|A[x} 1129a a A=Nx 1130a | w|w|xAx 1131a Ax=Bx 1132a ||a|xAx 1133a A=xNx 1134a ++ A|xAx 1135a | w w|Anx 1136a A[x]=Nx|(Ax) 1137a 4 " + (Ax)|Ax|=N 1138a 4 \|xAx

C2 (II) A3 (la) D # 1 (III) D # 1 (III) C1 (la) Al (III) A2 (III) Bl(Ia) C2 (la) Al (III) Bl(Ia) Bl (II) Bl(Ia) A2 (Ic) A3 (Ib) D 1 (III) Bl(Ia)** A2a (III) D 4 1(III) A2 (III) Bl(Ia) Bl(II) A2 (III) A2 (III) A3 (Ib) A3 (Ib) A3 (Ib) A2 (III) A2 (III) Al (III) Al (III) A3 (la) A2 (III) A 2b (III) El (III) C1 (Ib) A3 (la) Al (III) A3 (Ib) Al (III) Al (III) C1 (la) El (III) D 4 1 (III) Al (III)

1094b 1095b 1096b 1097b 1098b 1099b 1100b 1101b 1102b 1103b 1104b 1105b 1106b 1107b 1108b 1109b 1110b 1111b 1112b 1113b 1114b 1115b 1116b 1117b 1118b 1119b 1120b 1121b 1122b 1123b 1124b 1125b 1126b 1127b 1128b 1129b 1130b 1131b 1132b 1133b 1134b 1135b 1136b 1137b 1138b

262

Ax|Nx xA|Nx ||A|Nnx oo Ax|xNx (Ax)x|Nx ||w|w|AxN oo Ax|xNx oo Ax|xNx (Ax)|N(n)x llw w|w|xAx|N Ax|Nx A(n)x|Nx Ax|Nx ||w|Ax|N | (Ax)=Nnx w|xA|(N)x A=xNx OO A|N-Nx ||(Ax)n|(N)x ||wx|x(Ax)|Nx x(Ax)|Nx oo (Ax)x|xNx

Al (III) C1 (II) D 1 (III) A2 (III) Al (III) Bl (I) A2 (III) A2 (III) D 2 (III) Bl (I) Al (III) A2 (III) Al (III) B1(I) D 1 (III) C2 (I)* Al (III) D 1 (III) A2k (III) C^I) C1 (II) A2 (III)

||w|xA|{N}

CM)

||(Ax)|N(n)x oo ||A=N|xN A=Nx|N ||(Ax)x|Nx ||wx|A|xN oo ||A|Nx|-xN wxw w|A|xN ||w wx|A|(N)x Ax|N(n)x oo ||A=N|xN oo ||An|xN °°(Ax)x|xBx oo ||A|xNx x(Ax)|Nx oo ||A|Bx-N 00 A[x}|Nx|-xN oo ||ww|Ax|N

D 2 (III) El (III) El (III) Al (III) Bl (I) D42 (III) Bl (I) C 2 (I) D*2 (III) El (III) El (III) A2 (III) Al (III) C1 (II) D 4 1 (III) D42 (III) B1 (I). Cl(l) A2 (III) Bl (I) A2 (III) C 2 (I)

||w|xA|{N} oo

Ax|x(Nx)x ||w|w|Ax|(Nx) oo HA(n)xlNx ||w|xA=(N)x

Alliteration,

scansion and metrical clause structure 1139b ||w>: xA=Nx 1140b xA|Bx 1141b ooAx|xNx 1142b (Ax)=Nnx 1143b Ax=Nx 1144b xA|(N)x 1145b Ax]Nfx 1146b ooA|N|-xN 1147b xxAx|N 1148b °An|xNn 1149b Ax|Nx 1150b ||xwx|Ax|N 1151b ||w|w|A|(N)x 1152b ||wx|A|(N)x 1153b llw xA|(N)x 1154b x(Ax)|(Nx)x 1155b A=(N)nx 1156b Ax|Nx 1157b ||w|xA=Nx 1158b x(Ax)|(Nx)x 1159b oo||A|w|xNx 1160b oo||(Ax)|N|-xN 1161b ||(Ax)x|Bx 1162b l|w 5v|An|B 1163b ||w|xAx|Nx 1164b ||xw w|xxA|x(Nx)x 1165b ||wx w|An|(Nx) 1166b ||xw|wx|xAx|Nx 1167b llwhN w|xAx|Nx 1168b ||w, *|(Ax)|Nnx 1169b A=Nx|N 1170b ||w|xAx|N 1171b ||w|xAx|N 1172b Hw|w|A|{N} 1173b oo||(Ax)x|xNx 1174b ||w|A|(N)x 1175b ||w.v w|x(Ax)|Nx 1176b oo||(Ax)|w|xNx 1177b ooHA|lwx|w|Nx 1178b ||w|xxAx|N 1179b ||wx w|A|(N)x 1180b oo||w|Ax|N 1181b ||w|w|x(Ax)x|(Nx) 1182b oo||w w|A||wx|B|l 1183b oo(Ax)|xNx

1139a Ax|Nx Al (III) Bl (la) 1140a Ik w|A=xB B2 (la) 1141a llw w|A(n)x|N 1142a ||w|w|xxAx A3 (la) 1143a ||wx|w|Anx C1 (la) 1144a Ax|Nx Al (III) 1145a ||w|wx|xA(n)x C2 (la) 1146a ||wx|A=(A)x C2 (la) ++ 1147a A=(N)x|Ax A2 (III) 1148a ||wx|Ax|(Ax) Bl(Ia) 1149a xxA=Ax C1 (II) 1150a ||xax|Ax|N Bl(Ib) 1151a xax|xAx A3 (Ib) Al (III) 1152a Ax|Ax 1153a + +(Ax)|xAx Al (III) 1154a ||An|Anx D* 1 (III) 1155a xA=xN Bl(II) Bl (la) 1156a ||wx|w|xAx|N 1157a + + Ax|(Ax)=Nx D* 1 (III) El (III) 1158a A=Nx|N 1159a ax|xAx A3 (Ib) 1160a A=Nx|A El (III) 1161a ||axx|A=B A3 (Ib) ++ 1162a A|xA[x]=(B)x D*2 (III) 1163a a|xxAx|Nx hAl (Ib) 1164a ax|Anx=(Nx)x hA2 (Ib) 1165a + + ||A-N|Ax|=Nx HAl (III) 1166a + + xAx|=N|A|=NnxHD 1 (II)** 1167a ||w|w|wx|A|(A)x hC2 (la) ++ 1168a A-N|xAx|=xNx HA2 (III) A3 (Ib) 1169a ||xa|xxAx 1170a Ax|Nx Al (III) 1171a ++ A=(N)x|(Ax)x A2 (III) 1172a Ax|Nx Al (III) Bl(Ia) 1173a ||w|xAx|A 1174a °Ax|xNx A2 (III) 1175a [|w|w|Ax A3 (la) ++ 1176a (Ax)=N|Ax A2a (III) 1177a + +A=(N)xAx A2 (III) 1178a + +(Ax)xAx Al (III) Al (III)** 1179a A|xNx 1180a (Ax)=N|{N} A2k (III)** 1181a Ax|Nn Al (III) 1182a Ax|Bx Al (III) D 1 (III) 1183a (Ax)|Nnx

263

C(I) C1 (II) A2 (III) D 1 (III) Al (III) C2 (II) Al (III) D41 (III) Bl (II) A2 (III) Al (III) B1(I) C 2 (I) C 2 (I) C 2 (I) C1 (II) D 3 (III) Al (III)

Cl(D

C1 (II) A2 (Ic) D 4 1 (III) Al (III) B1(I) hAl (I) hAl (I) hBl(I) hAl (I) hAl (I) hD 1 (I) El (III) Bl(I) B1(I)

c1®

A2 (III) C 2 (I)

Cl(D

hi (Ic) A2 (III) B1(I). C2 (I) Bl(I) Bl(I) B2(I) Al (III)

Index 1184a ||wx w||w|w|xAx 1185a + +Ax(Ax)x 1186a ||w|w|xAx 1187a A[x}=(Nx)x|A 1188a ||w|w|xAx 1189a + +An|xAn 1190a + +(Ax)|x(Ax)x 1191a An|Nx 1192a ||w|w|A|(N)x 1193a + +Ax|xAx 1194a + +Ax|xAx 1195a + +A|xAx 1196a ||wx w|w|xAx 1197a ||wxx w|xxAx 1198a + +A=N[x}|(Ax)x 1199a °xxxAx|(Ax) 1200a + + Ax|xA=B 1201a Axnx 1202a ||wx|A|Ax 1203a (Ax)|Bnx 1204a ||wx|w|xxAx 1205a ++ ||A=N|(Ax)x 1206a ||wx w|xAx 1207a + +Ax|xAx 1208a Ax=Nx 1209a Ax|Nx 1210a ||xw|w|xAx|A 1211a A=xNx 1212a + + ||Ax|A=(N)x 1213a xxA=(N)x 1214a + + A=N|Ax 1215a ||An|(Nx)(n)x 1216a ||a|xxAx 1217a ++(Ax)|xAx 1218a A=xNx 1219a ||a|w|xAx 1220a Ax|Ax 1221a ||wx|w|xAx 1222a xxAx=N 1223a ||wx|xAx 1224a + + A=N|Ax 1225a (Ax)n|Ax 1226a A=xNx 1227a ++ A>: xAx 1228a ||w|w|A-N|A

A3 (la) Al (III) A3 (la) El (III) A3 (la) A2 (III) Al (III) Al (III) C2 (la) A2 (III) A2 (III) Al (III) A3 (la) A3 (la) A2a (III) Bl (II) A2b (III) Al (III) C1 (la) D 1 (III) A3 (la) A2a (III) A3 (la) A2 (III) Al (III) Al (III) Bl(Ia) Al (III) D*2 (III) C2 (II) A2a (III) D* 2 (III) A3 (Ib) Al (III) Al (III) A3 (la) Al (III) A3 (la) Bl (II) A3 (la) A2a (III) Al (III) Al (III) A2 (III) Bl (la)

1184b Ax|Nx 1185b ||w w|xA|xN 1186b xxA=Nx 1187b ooAx|x(Nx)x 1188b ||w|xx(Ax)|Nx 1189b x(Ax)x|N 1190b ||w|xAx|N 1191b xxxAx|N 1192b |w|A=(N)x 1193b ||w|Ax|N 1194b A=Nx|N 1195b A=Nx|N 1196b x(Ax)|Nx 1197b Ax|Nx 1198b ||wx|Ax|xN 1199b Anx|(Nx) 1200b ||(Ax)=Nx|B 1201b ||xw|Ax|N 1202b (Ax)n|Nx 1203b Ax|Bx 1204b A|N(n)x 1205b ||wx|A|xN 1206b A|N(n)x 1207b ||w|xAx|N 1208b xxAx|N 1209b ||w|xxAx|xN 1210b A|(N)nx 1211b ||w xA|(N)x 1212b A|N(n)x 1213b ||Ax|Nx 1214b oo ||A|Nx -xN 1215b ||w|xxx(Ax)x|N 1216b An|Nx 1217b ||w|xxAx|N 1218b oo||w|xA|(N)x 1219b ||w|xxAx|N 1220b ||w w|w|A|xN 1221b ||w|w|A|xN 1222b (Ax)|N(n)x 1223b ||w|A|xNx 1224b °°l|A||wxw|(Nx)x|| 1225b oo ||ww|A|(N)x 1226b ||w|w|(Ax)|Nx 1227b A=Nnx 1228b ooAx|xNx

264

Al (III) B1(I) C1 (II) A2 (III) Oil) Bl (II) B1(I) Bl (II) C 2 (I) B1(I) El (III) El (III) C1 (II) Al (III) B2(I) El (III) El (III) B1(I) Al (III) Al (III) D 2 (III) B1(I) D 2 (III) B1(I) B 1 (II) B2(I) D 3 (III) C 2 (I) D 2 (III) Al (III) D42 (III) Bl(I) Al (III) Bl(I) C 2 (I) Bl(I) Bl(I) Bl(I) D 2 (III) C*(D A2 (III) C 2 (I) Oil) D 1 (HI) A2 (III)

Alliteration, scansion and metrical clause structure 1229a 1230a 1231a 1232a 1233a 1234a 1235a 1236a 1237a 1238a 1239a 1240a 1241a 1242a 1243a 1244a 1245a 1246a 1247a 1248a 1249a 1250a 1251a 1252a 1253a 1254a 1255a 1256a 1257a 1258a 1259a 1260a 1261a 1262a 1263a 1264a 1265a 1266a 1267a 1268a 1269a 1270a 1271a 1272a 1273a

Ax|Ax ++ ||Ax|wx|xAx ++ Ax|A=(N)x ||wx|w|xAx ||wx|A|(A)x ++ A=N|Ax Ax|(Nx)x ||w w|An|xN + + Ax|xAx ++ A=N|Ax ++ ||A=(N)x|(Ax)x + + Ax|xAx + + A|xAx ||wx|w xAx ++ A=(B)x|Ax xx(Ax)nx (Ax)=Nx A + + A=(N)x|A-N

llw w|A|Nx xxA|xxAx |wx|wx|Ax °A|xNx ax w|xAx Ax=Nx ||wx|A=(N)x ++ ||A=N|Ax ++ A|xxAx A=N|(A)x ax | xxAx xxA=(N)x ++ (Ax)|An=N

llw w|(Ax)=(Nx)x Ax|Nx xA=Bx (Ax)x=Nx + + Ax|xAx ||Ax|(Ax)x An=Ax + + (Ax)=N|(Ax)-N Anx|A ||w|w|A=Nx ||wx|w|xAx A=Nx|(Ax) ||w w|xA-Nx ++

\xlxAx

Al (III) A2 (Ic) D*2 (III) A3 (la) C2 (la) A2a (III) Al (III) B2 (la) A2 (III) A2a (III) A2 (III) A2 (III) Al (III) A3 (la) A2 (III) C1 (II) El (III) A2 (III) C1 (la) B* (II) A3 (la) Al (III) A3 (la) Al (III) C2 (la) A2a (III) A2 (III) A2k (III) A3 (Ib) C2 (II) D41 (III) C1 (la) Al (III) C1 (II) Al (III) A2 (III) Al (III) Al (III) A2a (III) El (III) C1 (la) A3 (la) El (III) C1 (la) A2 (III)

1229b A=Nx|N 1230b oo||A|N-(N)x 1231b ooA||w|w|Nx 1232b ||w|w|Ax|N 1233b oo||A|xNx 1234b | w w|xAx|N 1235b ||wx|Ax|N 1236b x(Ax)|Nx 1237b ||(Ax)|N(n)x 1238b ||w|ww|A|(N)x 1239b wxAx|N 1240b ||A=Nx|N 1241b oo A=Nx|xN 1242b Ax=Nx 1243b ||w|xAx|B 1244b A=xNx 1245b Ax|Nx 1246b ||w|A|(N)x 1247b An=Nx 1248b xx( Ax) Nx 1249b ||wx|xxA=Nx 1250b ||w|xA|(N)x 1251b ||w Ax|xN 1252b ||w|w|xA|xN 1253b Ax|(Nx)x 1254b ||w|w|Ax|xN 1255b | w|xAx|N 1256b ||wx|(Ax)|xN 1257b Ax|Nx 1258b ||Ax|Nx 1259b ooAx|xNx 1260b (Ax)x|Nx 1261b ||wx|Ax|N 1262b Ax|Bx 1263b ||w|w|A|xN 1264b OOA=N|{N} 1265b ||wx|A|(N)x 1266b ||w|wx|Ax|N 1267b ||wx(Ax)x|N 1268b Ax|Nx 1269b A-Nx|N 1270b (Ax)x|Nx 1271b ||w|w|A|Nx 1272b ooAx|xNx 1273b ||w|w|xxA|xxN

265

El (III) D2 (III) A2 (III) B1(I) Al (III) B1(I) B1(I) C1 (II) D2 (III) C2(I) B1(I) El (III) E2 (III) Al (III) B1(I) Al (III) Al (III) C2(I) Al (III) C1 (II) C^I) C2(I) B2(I) B1(I) Al (III) B2(I) B1(I) B1(I) Al (III) Al (III) A2 (III) Al (III) B1(I) Al (III) B1(I) A2a(III) C2(I) B1(I) B1(I) Al (III) El (III) Al (III) O{\) A2 (III) B2(I)

Index 1274a ||xax|Ax|N 1275a ++Ax|xAx 1276a A=Nx|N 1277a ++ Ax|xA=N 1278a A=Nx|A 1279a ||w|w|x(Ax)x 1280a xxA|Ax 1281a A-N|Ax 1282a Ax|Nx 1283a HwxlxAx 1284a ++ A=(N)x|Ax 1285a ||wx|(Ax)|Nx 1286a + + A|Ax|=N 1287a Ax|Nx 1288a ||w|w|xAx 1289a ++A|xxAx 1290a ax|Ax|N 1291a Ax|Nx 1292a ||w|w|xAx 1293a ||Ax|Nx 1294a ||wx|w|(Ax)nx 1295a ++Ax|xAx 1296a ||w|w|Anx 1297a xxAx|N 1298a ++ Ax|A=(N)x 1299a A=Nx|A 1300a ||w|w|AxA 1301a xxAx=(Bx) 1302a + + ||A|w|x(Ax)x 1303a Ax|Nx 1304a ||xax|xAx 1305a ||w|w|xA|Nx 1306a Ax|Ax 1307a + + A|Ax=N 1308a ||wx|w|Ax=N 1309a xxAnx 1310a ||wx|w|xAx 1311a (Ax)=Nx|A 1312a ax|Ax|N 1313a ++A|xxAx 1314a ||wx|w|A-Nx 1315a xxA=Nx 1316a ||w|w|xxAx 1317a xxA=(N)x 1318a ||w|w|xxAx

Bl(Ib) A2 (III) El (III) A2b (III) El (III) A3 (la) C1 (II) Al (III) Al (III) A3 (la) A2 (III) C1 (la) D41 (III) Al (III) A3 (la) A2 (III) Bl(Ib) Al (III) A3 (la) Al (III) C1 (la) A2 (III) C1 (la) Bl(II) D*2 (III) El (III) Bl(Ia) Bl(II) A2 (Ic) Al(III) A3 (Ib) C1 (la) Al (III) D41(III) Bl(Ia) C1 (II) A3 (la) El (III) Bl (Ib) A2 (III) C1 (la) C1 (II) A3 (la) C2 (II) A3 (la)

1274b ||w|w|A|xN 1275b A=N|{N} 1276b ||w|xAx|xN 1277b xA|Nx 1278b ||(Ax)|=N|(N)x 1279b ||w|A=(N)x 1280b ||w|w|Ax|N 1281b ||wx|Ax|N 1282b ||w|x(Ax)|Nx 1283b ||w|w|Ax|N 1284b xAx=N 1285b oo(Ax)x|xNx 1286b ooA|xxNx 1287b A-N|(N)x 1288b A=N|(N)x 1289b ||A=N|(N)x 1290b oo||A|xxNx 1291b ||w|wx|xAx|xN 1292b ||wx|A|(N)x 1293b ||w|w|xAx|N 1294b Ax|Nx 1295b ||w|w|xAx|N 1296b (Ax)x|Nx 1297b xA|Nx 1298b ||wx|w|w|xAx|xN 1299b ||w|An|N 1300b ooA|xNx 1301b Ax|Bx 1302b ||w|xxAx|xN 1303b oo||(Ax)|w|xNx 1304b ||xw|xxAx|N 1305b Ax|Nx 1306b ||w|w|A|(N)n 1307b xA|Nx 1308b A=(Nx)nx 1309b Ax|Nx 1310b An|(N)x 1311b oo||(Ax)|N-(N)x 1312b (Ax)x|Nx 1313b ||w|x(Ax)x|N 1314b Ax|Nx 1315b ooAx|xNx 1316b A=Nx|N 1317b oo||A=(N)x|(Nx)x 1318b Ax|Nx

266

Bl(I) A2k (III)** B2(I) C1 (II) A2k (III) C2(I) Cl{\) Bl(I) Bl (II) A2 (III) A2 (III) A2k (III) A2k (III) A2k (III) A2 (III) B2(I) C2(I) B1(I) Al (III) Bl(I) Al (III) C1 (II) B2(I) B1(I) Al (III) Al (III) B2(I) A2 (Ic) Bl(I) Al (III) C2 (I) C1 (II) D 1 (III) Al (III) A2k (III) D2 (III) Al (III) Bl(I). Al (III) A2 (III) El (III) A2 (III) Al (III)

Alliteration, scansion and metrical clause structure 1319a A|N(n)x 1320a xxA=(N)x 1321a ||An|(Nx)(n)x 1322a l|xw|\si xxAx 1323a (Ax)x|Nx 1324a Axnx 1325a xA=(N)x 1326a A=xNx 1327a (Ax)x|(Nx)x 1328a ||(Ax)x|(Nx)x 1329a ++ (Ax)n|A-N 1330a llw w|x(Ax)x 1331a + + A=N|Ax 1332a + + (Ax)|Ax|=N 1333a + +Ax|xAx 1334a ||w|w|Ax|N 1335a xAx|A 1336a ||xw|w|xAx 1337a + +(Ax)x|xAx 1338a Ax|Nx 1339a + + Ax|A=(N)x 1340a ||w|A|(N)x 1341a ||w|w|Ax|B 1342a ||w|w|xxA=(B)x 1343a + +(Ax)=(N)x|Ax 1344a llw w w|A-Nx 1345a ||w|w|A=Nn 1346a (Ax)=Nnx 1347a ||w|w|xAx 1348a xxA=(N)x 1349a Ax=Nx 1350a ||w|w|w|xA=Nx 1351a + +(Ax)x|A-Nx 1352a x(Ax)|Ax 1353a ||wx|w w|Ax 1354a ||wx|xA=(N)x 1355a A=Nnx 1356a llw, w|Ax|N 1357a Ax|Nx 1358a axx|A=(N)x 1359a + + Ax A=xN 1360a xxAx|x(Ax) 1361a + +A|xxAx 1362a A=xAx 1363a ||xxw|A(n)x

D 2 (III) C2 (II) D* 2 (III) A3 (la) Al (III) Al (III) C2 (II) Al (III) Al (III) Al (III) Al (III) A3 (la) A2a (III) D 4 1(III) A2 (III) Bl (la) B 1 (II) A3 (la) A2 (III) Al (III) D*2 (III) C2 (la) Bl (la) C2 (la) A2 (III) C1 (la) C1 (la) D 1 (III) A3 (la) C2 (II) Al (III) C1 (la) D^III) C1 (II) A3 (la) C2 (la) D 1 (III) Bl(Ia) Al (III) C2 (Ib) D* 4 1(III) B2 (II) A2 (III) Al (III) C2 (la)

1319b oo||A||w|w]Nx 1320b ooA|xNx 1321b A|Nnx 1322b oo ||A|w|xNx 1323b ||A|w|N(n)x 1324b Ax|Nx 1325b xxA=(N)x 1326b ||wx|w xA-(N)x 1327b oo||wx|(Ax)|Nx 1328b ||w|wx|A|(N)x 1329b ||w|A(n)x|N 1330b xA=(N)x 1331b oo||w|xA||(N)x 1332b A=Nx|N 1333b ||w|xAx|N 1334b Ax|Nx 1335b AxNx 1336b Ax|Nx 1337b ||w|xAx|xN 1338b oo ||w|w|Ax|N 1339b ||wx|xxA|(N)x 1340b ooAx|xNx 1341b Ax|(Bx)x 1342b x(Ax)|Bx|| 1343b ||w|xA|(N)x 1344b Ax|Nx 1345b Ax|Nx 1346b Ax|Nx 1347b ooAx|Nx 1348b Ax|Nx 1349b oo||wx|Ax|N 1350b x(Ax)|Nx|| 1351b ||wx|A=(N)x 1352b A=Nx|N 1353b ||wx|xxA|Nx 1354b Ax|Nx 1355b ||w|w|(Ax)|Nx 1356b ooA|xNx 1357b ||w|Ax|N 1358b A(n)x|Nx 1359b ||w|Ax=N 1360b oo(Ax)|xNx 1361b ||w|w|A|(N)x 1362b ||w|x(Ax)|Nx 1363b Ax|Nx

267

A2 (III) Al (III) D 1 (III) A2 (Ic) D* 2 (Ic) Al (III) C2 (II) C 2 (I) C1 (I)** C 2 (I) B2(I) C2 (II) C 2 (I) El (III) B1(I) Al (III) Al (III) Al (III) B2(I) B1(I) C 2 (I) A2 (III) Al (III) C1 (II) C 2 (I) Al (III) Al (III) Al (III) Al (III) Al (III) B1(I) C1 (II) C 2 (I) El (III)

CM) Al (III) C^I) Al (III) Bl(I) A2 (III) B1(I)' Al (III) C 2 (I) C^I) Al (III)

Index 1364a + + ||(Ax)|Ax|=N 1365a ||w|w|Ax|xN 1366a ++ A|xAx 1367a (Ax)x|Nx 1368a ||w|w|A=(N)x 1369a ++ (Ax)|Ax|=N 1370a ++Ax|xAx 1371a ++Ax|xAx 1372a ||(Ax)x|Nx 1373a ||wx|A=xN 1374a ++ A|xAx 1375a °A|xNx 1376a ||(Ax)x|Ax 1377a a|xxAx 1378a Ax|Nx 1379a Anx|A 1380a ||w|w|xAx 1381a A=xNx 1382a Ax|Nx 1383a ||An|(Nx)(n)x 1384a ||xax|(Ax)|(N)x 1385a ||w|w|xA|(N)x 1386a ||ax|A-N|N 1387a (Ax)x|Nx 1388a ++Ax|xAx 1389a A=Nnx 1390a ||xa|Ax|N 1391a Ax|Nx 1392a Hw|w|w|xAx 1393a ||w|xAx|A 1394a ||w|x(Ax)x|A 1395a ||A|=Nx|A 1396a °Ax|xNx 1397a ||xw|w|x(Ax)x 1398a A(n)x|Nx 1399a ||w|w|Anx 1400a ++ A|Ax=B 1401a (Ax)-N|Ax 1402a A=Nnx 1403a xxA=(B)x 1404a ++A|xxAx 1405a xxAx|A 1406a xxAnx 1407a ||wx|w|xAnx 1408a HxxAxlN

D41 (III) B2 (la) Al (III) Al (III) C2 (la) D41(III) A2 (III) A2 (III) Al (III) Bl (la) Al (III) Al (III) Al (III) A3 (Ib)** Al (III) El (III) A3 (la) Al (III) Al (III) D*2 (III) C2 (Ib) C2 (la) Bl(Ib) Al (III) A2 (III) D 1 (III) Bl(Ib) Al (III) A3 (la) Bl(Ia) Bl(Ia) El (III) A2 (III) A3 (la) A2 (III) C1 (la) D41 (III) Al (III) D 1 (III) C2 (II) A2 (III) Bl(II) C1 (II) C1 (la) Bl(Ia)

1364b oo(Ax)|xxNx 1365b A=Nx|N 1366b ||w|xA|(N)x 1367b .||w|xxA|(N)x 1368b ooAx|xNx 1369b ooA=(N)x|Nx 1370b 1371b 1372b 1373b 1374b 1375b 1376b 1377b 1378b 1379b

||w|w|A|(N)x

||w|w|Ax|N °°A|xNx ||wx|A|(N)x ||w|w|A|Nx ||w|w|xA|xN 3O ||A|N|-xN ||w|w|Ax|N '°l|AHwlwlNx 1380b A|N(n)x 1381b ||w|w|A|(N)x|| 1382b ||w|w|xA|(N)x 1383b A|Nnx 1384b oo||Ax|w|N-N 1385b ||wx|w|(Ax)|Nx 1386b ooAx|xNx 1387b ooHAxllwlwlNxll 1388b ||w|w|A=(N)x 1389b Ax|Nx 1390b ||wx|(Ax)|Nx 1391b A|N(n)x 1392b ||w|w|xA|(N)x 1393b ||w|xAx=N 1394b oo||A||wlw[Nx 1395b xA|(N)x 1396b oo||w|w|w|Ax|N 1397b ||(Ax)|N(n)x 1398b ||w|xA|xN 1399b ooA|xNx 1400b ||Ax|Bx 1401b ||A=Nx|N 1402b ||Ax|Nx 1403b ooAx|xBx 1404b ||w|Ax|N 1405b ||(Ax)=Nx|N 1406b Ax=Nx 1407b A|N(n)x 1408b (Ax)nx|N

268

A2 (III) El (III) C2(I) C2(I) A2 (III) A2 (III) C2(I) C2(I) B1(I) Al (III) C2(I) D41 (III) B1(I) A2 (III) D2 (III) C2(I) C2(I) D 1 (III) A2 (Ic)

CM) A2 (III) A2 (III) C2(I) Al (III) C1 (I)** D2 (III) C2(I) B1(I) A2 (III) C2 (II) B1(I) D2 (III) Bl(I) Al (III) Al (III) El (III) Al (III) A2 (III) B1(I) El (III) Al (III) D2 (III) El (III)

Alliteration, scansion and metrical clause structure 1409a 1410a 1411a 1412a 1413a 1414a 1415a 1416a 1417a 1418a 1419a 1420a 1421a 1422a 1423a 1424a 1425a 1426a 1427a 1428a 1429a 1430a 1431a 1432a 1433a 1434a 1435a 1436a 1437a 1438a 1439a 1440a 1441a 1442a 1443a 1444a 1445a 1446a 1447a 1448a 1449a 1450a 1451a 1452a 1453a

A|A=(N)x ++ Ax|A=(N)x Ax|Ax ||w|Ax|N Ax|Nx ||w|w w|Anx xxAx N A=Nx|(Ax) ++ Ax|xxAx (Ax)|Nnx xx(Ax)nx ++ A-N|Ax|=xN xxA=(N)x ||A|Nx|-N Ax Ax ++ A-N|A=N | xwxw|xx(Ax)x ++ A=(N)x|A=(N)x ||wx|xA=(N)x ||w|xAx=N A=Nx|A ++ Ax|xA=N ++ Ax|xxAx ||A=N|(A)x xA=(N)x A=xNx ++ (Ax)=N|Ax ++ A x xAx ||wx|w|xAx (Ax)=Anx ++ ||AxxAx ++ A[x]-N|A=(N)x (Ax)=Nx|A A=xNx ||wx|(Ax)=Bx ++ A|x(Ax)=N ||w|w|A=(B)x ||w|w Ax=N ++ Ax|An=N °||w|xAx|A ||ww|(Ax)=Nx ||ax|A=xN ||xax|A=Nx ax|Ax|N ||xax|A=Nx

D 2 (III) D* 2 (HI) Al (III) Bl(Ia) Al (III) C 1 (la) B 1 (II) El (III) A2 (III) D 1 (III) C 1 (II) D* 4 2 (III)

1409b Ax|Nx 1410b o o A=N|xN 1411b (Ax)=Nx|(Nx) 1412b x(Ax)|Nx 1413b A|N(n)x 1414b Ax=Nx 1415b (Ax)x|Nx 1416b oo ||(Ax)|Nx|-N 1417b ||(Ax)|=Nx|N 1418b ooAx|xNx 1419b Ax|(Nx)x 1420b ||wx|A(n)x 1421b (Ax)x|Nx 1422b OO ||A|N|-Nx|| 1423b 0O ||A|Nx|-N 1424b |Ax = N | x N 1425b A=Nx|(Nx) 1426b A|N(n)x 1427b Ax|Nx 1428b DOA|x(Nx)x 1429b xA=Nx 1430b |w|xA|(N)x 1431b DO||A|xNx 1432b |wx|Ax|N 1433b 30Ax|xNx 1434b w w|xAx|N 1435b w|xAx|N 1436b |w|wx|A|xN 1437b sc(Ax)=Nx 1438b 3OAx|xNx 1439b |w|xA|(N)x 1440b |(Ax)|Nnx 1441b ( 5O||(Ax)x|wx|Nn 1442b |wx|xAx|N 1443b c 5OAx|xBx 1444b ,&|N(n)x 1445b ,^x|Bx 1446b c >oAx|xNx 1447b c >oAx|xNx 1448b ( Ax)x|(Nx)x 1449b A\x|Nx 1450b c >oAx|xNx 1451b |w wx A=(N)x 1452b |Ax|Nx 1453b |w wx|Ax|N

C 2 (ID

D 4 1(III)** Al (III) A 2b (III) A 3 (la) D* 2 (III) C2 (la) Bl(Ia) El (III) A 2b (III) A2 (III) A 2k (III) C 2 (ID

Al (III) A2a (III) A2 (III) A 3 (la) D 1 (HI) A2 (III) D* 2 (III) El. (Ill) Al (III) C 1 (la) A 2b (III) C2 (la) Bl(Ia) D* 4 1 (III) Bl(Ia) C 1 (la) Bl(Ib) C 1 (Ib) Bl(Ib) C 1 (Ib)

269

Al (III) El (III) El (III) C 1 (II) D 2 (III) Al (III) Al (HI) D 4 1 (III) El (III) A2 (III) Al (III) C 2 (I) Al (HI) D 1 (III) D 4 1 (III) E*1(III) El (III) D 2 (III) Al (III) Al (III) C 1 (II) C 2 (I) Al (III) Bl(I) A2 (III) Bl(I) Bl(I) Bl(I) C 1 (ID

A2 (III) C2(I) D1 (III) A2 (Ic) Bl(I) A2 (III) D2 (III) Al (III) A2 (III) A2 (HI) Al (III) Al (HI) A2 (III) C2(I) Al (III) B1(I)

Index ++

A|x(Ax)=Bx 1454a 1455a ||w|w|wx|Ax 1456a ||w|w|xAx|N 1457a ||w|xA=Nx 1458a ||w|w|A|(N)x 1459a l|A|w|Ax 1460a ||xax|(Ax)=Bx 1461a Ax|Nx 1462a ||w|w|(Ax)=Nx 1463a + +A=(N)x|Ax 1464a ||w|w|Ax=N 1465a ||wx|xxAx 1466a (Ax)x|Nx 1467a Ax|Nx 1468a + +Ax|A=(N)x 1469a xxAx|xN 1470a + + ||A=(N)x|Ax 1471a Ax=Nx 1472a ||wx|w|wx|xAx 1473a ||An|(Nx)(n)x 1474a ||xw|w|xAx 1475a Ax|Bx 1476a + +A=(N)x|(Ax)x 1477a ||w|w|xAx 1478a Ax|Nx 1479a A=x(Nx)x 1480a ||w|w|A=(N)x 1481a A=xNx 1482a ||wx|w|xAx 1483a An|Nx 1484a ||w|wx|xxAx|x(Ax) 1485a ||xa|(Ax)|Nx 1486a ||w|w|A=Nx 1487a Ax | Ax 1488a ||w|w|An|B 1489a + + A-N|A=N 1490a + + A=N|Ax 1491a °A|xNx 1492a ||xxw|Ax 1493a ax | xAx 1494a Ax|Bx 1495a Ax=Nx 1496a ||w|w|xxA=N 1497a l[wx[w|xAx 1498a + +(Ax)=Nx|xA

1454b 1455b 1456b 1457b 1458b 1459b 1460b 146 Ib 1462b 1463b 1464b 1465b

oo

Ax|xBx (Ax)=N(n)x (Ax)|Nnx An|(N)x A=xNx ||A[x]=Nx|N ||wx|w|xAx|xB ||wx|w|w|xAx|xN xA|Nx ||w|w|Ax|N Ax|Nx (Ax)|Nnx 1466b w|A|xN 1467b |w|w|xAx|xN 1468b >o||Ax|xNx 1469b )OAx|xNx 1470b |w|w|Ax|xN 1471b >o||xw|xAx|N 1472b x(Ax)|Nx 1473b A|Nnx 1474b (Ax)|N(n)x w|w|Ax|B|| 1475b 1476b w|A|Nx 1477b ooAx|Nx 1478b ||w|w|w|A|Nx 1479b x(Ax)|Nx 1480b xx(Ax)=Nx 1481b ||w|w|A|(N)x 1482b oo||w|w|A|Nx|| 1483b oo(Ax)nx|xN 1484b Ax|Nx 1485b ||wx|w|xxA|(N)x 1486b Ax|Nx 1487b oo]]AHwxlNx 1488b Ax|Bx 1489b A=Nx|N 1490b ||w|w|xAnx 1491b ||wx|w|A|(N)x 1492b (Ax)=Nx|N 1493b ||wx|A-(N)x 1494b oo ||A=B|xN 1495b ||w|w|A|(N)x 1496b x(Ax)|Nx 1497b ||w|w|Ax|xN 1498b OOA|N-Nx

3

A (la) Bl (la) C1 (la) C2 (la) Al (Ic) C1 (Ib) Al (III) C1 (la) A2 (III) Bl(Ia) A3 (la) Al (III) Al (III) D*2 (III) B2 (II) A2 (III) Al (III) A3 (la) D*2 (III) A3 (la) Al (III) A2 (III) A3 (la) Al (III) Al (III) C2 (la) Al (III) A3 (la) Al (III) B2 (la) C1 (Ib) C1 (la) Al (III) Bl (la) A2b (III) A2a (III) Al (III) A3 (la)** A3 (Ib) Al (III) Al (III) A3 (la) A3 (la) E2 (III)

270

A2 (III) D2 (III) D 1 (III) A2k (III) Al (III) El (III) B2(I) B2(I) C1 (II) B1(I) Al (III) D1 (III) B1(I) B2(I) A2 (III) A2 (III) B2(I) B1(I) C1 (II) D 1 (III) D2 (III)

Al (III)

c'a)

C1 (II) C1 (II) C2(I) C^I) E2 (III) Al (III) C2(I) Al (III) A2 (III) Al (III) El (III) C^I) C2(I) El (III) C2(I) El (III) C2(I) C1 (II) B2(I) D 1 (III)

Alliteration, scansion and metrical clause structure 1499a ++ A|xAx 1500a A=Nx|A 1501a ||a|w|xAx 1502a (Ax)x|Nx 1503a Ax|Nx 1504a ||w|w|xxA=N 1505a ++(Ax)x|(Ax)=Nx 1506a ||a|w|xA=N 1507a Ax|Nx 1508a ||w|w|xAx 1509a ++Ax|xAx 1510a ax|xAx 1511a Ax=Nx 1512a ||ax|A=Nx 1513a j|w|w|xA=(N)x 1514a ||w|w|xx(A)x 1515a ||w|w|xA=(N)x 1516a ++ A=(N)x|Ax 1517a Ax|Nx 1518a ||xa|w|xAx 1519a ++ (Ax)=N|Ax 1520a Ax=Nx 1521a ||w|wx|x(Ax)x 1522a + + Ax|A=N 1523a ||w|x(Ax)=Nx 1524a ||Ax|Nx 1525a ++Ax|xAx 1526a A=xNx 1527a + + Ax|A=N 1528a Ax|Nx 1529a ||a|w|A=N 1530a + +Ax|xAx 1531a ||a|w|Ax=N 1532a ++ Ax|A-Nx 1533a + + A|xA=N 1534a ++A=(N)x|(Ax)x 1535a ||wx|w|xAx 1536a A=Nx|A 1537a ||xa|w|xNx 1538a A=Ax|N 1539a ||a|w|Ax|N 1540a A=xNx 1541a ||w|w|A|(N)x 1542a Ax|Ax 1543a ||xxa|w|Ax=N

Al (III) El (III) A3 (la) Al (III) Al (III) A3 (la) D*1 (III) A3 (la) Al (III) A3 (la) A2 (III) A3 (Ib) Al (III) C1 (Ib) C2 (la) A3" (la) C2 (la) A2 (III) Al (III) A3 (la) A2a (III) Al (III) A3 (la) A2b (III) C1 (la) Al (III) A2 (III) Al (III) A2b (III) Al (III) A3 (la) A2 (III) Bl (la) D*1 (III) A2b (III) A2 (III) A3 (la) El (III) F (la)** El (III)** Bl(Ia) Al (III) C2 (la) Al (III) Bl(Ia)

1499b ||w|w|(Ax)x|N 1500b (Ax)|N(n)x 1501b oo ||A=N|xN 1502b ||w|xw|A|xN 1503b oo||A|Nx|-xN 1504b xA|xNx 1505b Ax|Nx 1506b ||w|w|xAx|N|| 1507b x(Ax)|Nx 1508b ||w|w|xAx|N|| 1509b ||w|wx|Ax|x(Nx) 1510b ||A=N|(N)x 1511b (Ax)=Nx|N 1512b ||w|xA|xN 1513b A-Nx|N 1514b ooAx|x(Nx)x 1515b ooAx|xNx 1516b oo ||A=N|xN 1517b Ax|Nx 1518b A=Nnx 1519b oo||(Ax)=N|xN 1520b oo||A|N|-xxN 1521b oo A=N|xN 1522b ||w|xA|xN 1523b Ax|Nx 1524b ||w|xA|xN 1525b ooH(Ax)xl-Nl(Nx) 1526b oo||A|N|-xN 1527b ||w|w|Ax|N 1528b ||w|xA|xN 1529b ||wx|Ax|N 1530b A|Nnx 1531b ooAx|xNx 1532b ||w|w|xAx|N 1533b oo||Ax|x(Nx)x 1534b ||w|w|A|{N} 1535b xA|Nx 1536b ||w|xxA|(N)x 1537b ||wx|xAx|N|| 1538b Ax|Nx 1539b ||w|w|xAx|N|| 1540b ||w|w|xA|xN 1541b ooA-N|xN 1542b ||w|w|xAx|N 1543b (Ax)x|Nx

271

D2 (III) El (III) B1(I) D42 (III) C* (II) Al (III) Bl(I) C1 (II) B1(I) B2(I) A2k (III) El (III) B1(I) El (III) A2 (III) A2 (III) El (III) Al (III) D 1 (III) El (III) D42 (III) El (III) B1(I) Al (III) Bl(I) F (III) D41(IH) Bl(I) Bl(I) Bl(I) D 1 (HI) A2 (III) Bl(I) A2 (HI) O{\) C1 (II) C2(I) Bl(I) Al (III) Bl(I) B1(I) El (III) B1(I) Al (III)

Index 1544a Ax=Nx 1545a ||xa w|xx(Ax)=N 1546a + + A|xA=N 1547a Ax|(Ax)x 1548a + + A=N|Ax 1549a ||xA|xxAx 1550a ||wx|w|xAx 1551a xxAx A 1552a ||wx|w|(Ax)=Nx 1553a ++ (Ax)=N|Ax 1554a xa|A=(N)x 1555a (Ax)x|An 1556a Ax=Nx 1557a xa w|xAx 1558a + +A=N|(Ax)x 1559a + +(Ax)x|A=N 1560a ||wx|w|w|Ax 1561a x(Ax)=Nx 1562a ++A|x(Ax)-N 1563a llw xa|x(Ax)=N 1564a ++ A|x(Ax)=N 1565a + +Ax|A-Nx 1566a ||w wx|xAx 1567a ||A=Nx A 1568a + +Ax A=(N)x 1569a HA|w|Ax 1570a ||ax|xAx 1571a ||wx w|x(Ax)x 1572a (Ax)x|Nx 1573a ||w|w|xAx 1574a ++ A|xAx 1575a + + Ax|xA=N 1576a Ax=Nx 1577a + +Ax|xAx 1578a ||wx w|w|xAx 1579a Ax|Nx 1580a ||wx w|Anx 1581a a x(Ax)x 1582a Ax|(Nx)x 1583a ||w|AxN 1584a A=(N)x|A 1585a Ax|Nx 1586a A=Nnx 1587a Ax=Nx 1588a + +A|x(Ax)x

Al (III) Bl(Ia) A2b (III) Al (III) A2a (III) B* (II)* A3 (la) Bl (II) C1 (la) A2a (III) C2 (Ib) Al (III) Al (III) A3 (la) A2a (III) A2b (III) A3 (la) C1 (II) Al (III) A3 (la) A2b (III) D #1 (III) A3 (la) El (III) D*2 (III) Al(Ic) A3 (Ib) A3 (la) Al (III) A3 (la) Al (III) A2b (III) Al (III) A2 (III) A3 (la) Al (III) C1 (la) A3 (Ib) Al (III) Bl (la)** E2 (III) Al (III) D 1 (III) Al (III) Al (III)

1544b l|w|w|xAx|N 1545b ||w|xxA|xN 1546b | |W5c xxA|(N)x 1547b ||w|xAx|N 1548b ||w|xA|Nx 1549b ooA-N|xN 1550b (Ax)|Nnx 1551b Ax|Nx 1552b ooAx|x(Nx)x 1553b ||w|Ax|N 1554b Ax|Nx 1555b ||w|xA|xN 1556b ||wx|w|A|xN 1557b (Ax)=Nx|N 1558b Ax|Nx 1559b ||w w|Ax|N 1560b |wx xxA Nx 1561b x(Ax)|Nx 1562b ooAnx|xN 1563b (Ax)|Nnx 1564b oo ||A=N|xN 1565b ||Anx|N 1566b A|N(n)x 1567b oo ||A|=N|xN 1568b ||w|xA|xN 1569b oo||A|Nx|-xN 1570b oo||A|Nx|-N 1571b Ax|Nx 1572b |w xx(Ax)x|N 1573b ||Ax|(Nx)(n)x 1574b (Ax)nx|N 1575b ||w|xA|(N)x 1576b ||w|w|(Ax)|Nx 1577b ooA=Nx|xN 1578b xA=(N)x 1579b ||wx|xAx|N 1580b A=xNx 1581b ||Anx|N 1582b A=Nx|N 1583b ooA|x(Nx)x 1584b ||w|w|w|A|xN 1585b |xw w|w|xAx|xN 1586b Ax|Nx 1587b ||w|w|A|xN 1588b oo||A|Nx|-N

272

B1(I) B1(I) C2(I) B1(I) O{l) El (III) D 1 (III) Al (III) A2 (III) B1(I) Al (III) B1(I) B1(I) El (III) Al (III) B1(I) C^I) C1 (II) E2 (III) D 1 (III) El (III) El (III) D2 (III) El (III)** B1(I) D42 (III) D41 (III) Al (III) B1(I) D*2 (III) El (III) C2(I) O(l) E2 (III) C2 (II) Bl(I) Al (III) El (III) El (III) Al (III) Bl(I) B2(I) Al (III) Bl(I) D41(III)

Alliteration, 1589a ||wx|w|xxAx 1590a ++ (Ax)=N|Ax 1591a HaxjwlxAx 1592a ||w|w|xAnx 1593a i|w|w|A=xN 1594a ++ A|Ax|=B 1595a ++(Ax)x|xAx 1596a ||w|w|x(Ax)nx 1597a ||w|w|(Ax)=Nx 1598a Ax|Nx 1599a ||w|wx|xA=N 1600a ||w|w|A|(N)x 1601a (Ax)|Nnx 1602a ++A=(N)x|(Ax)x 1603a Ax|Nx 1604a ++||Ax|xxAx 1605a ++Ax|xAx 1606a xx(Ax)=Nx 1607a + + A=N|(Ax)x 1608a Hw|w|A|xN 1609a ||wx|Ax|N 1610a ||xax|A=Nx 1611a DAx|xBx 1612a ||w|w|w|xxAx 1613a A=Nx|A 1614a ||wx|xx(Ax)x 1615a Ax|Nx 1616a ||xa|An=N 1617a ++ Ax|Ax=N 1618a |ax|w|xAx 1619a +A=(N)x|Ax 1620a |wx|A=xN 1621a Ax|Ax 1622a xa|A=(N)x 1623a ||w|w|xAx 1624a + + A=N|Ax 1625a (Ax)=Nnx 1626a ||wx|w|w|xAx ++ 1627a A-N|Ax|=N 1628a ||w|w|w|wx|xAx 1629a ||w|w|xxAx ++ 1630a Ax|xAx ++ 1631a (Ax)|xxAx 1632a ||ax|A|(N)x 1633a Ax|Ax

scansion and metrical clause structure

A 3 (la) A 2 a (III) A 3 (la) C 1 (la) Bl(Ia) D 4 1 (III) A 2 (III) C 1 (la) C 1 (la) Al (III) A 3 (la) C 2 (la) D 1 (III) A 2 (III) Al (III) A 2 (III) A 2 (III) C 1 (II) A 2 a (III) Bl(Ia) Bl(Ia) C 1 (Ib) A 2 (III) A 3 (la) El (III) A 3 (la) Al (III) Bl(Ib) D* 4 1 (III) A 3 (la) A 2 (III) Bl(Ia) Al (III) C 2 (Ib) A 3 (la) A 2 a (III) D 1 (III) A 3 (la) D*41(III) A 3 (la) A 3 (la) A 2 (III) A 2 (III) C 2 (Ib) Al (III)

1589b (Ax)|N(n)x 1590b ||w|wx|w|Ax|xN 1591b Ax|Nx 1592b xA|(N)x 1593b ooA|xNx 1594b ||Ax=Bx 1595b x(Ax)|Nx 1596b ooA|xNx 1597b Ax|Nx 1598b oo||w|w|(Ax)x|xN 1599b x(Ax)|Nx 1600b oo||A|xNx 1601b ||xw|w|A|(N)x 1602b ||Ax|Nx 1603b ||w|x(Ax)|(Nx)x 1604b ||w|w|xx(Ax)=Nx 1605b ||w|xA|xN 1606b Ax=(Nx)x 1607b ||w|w|Ax|N 1608b ooAx|xNx 1609b oo(Ax)|xNx 1610b ||w|xA|(N)x 1611b ||w|w|A|(B)x 1612b (Ax)=Nx|N 1613b oo||w|w|w|(Ax)x|xN 1614b xxA|(N)x 1615b oo||A|N|-xN 1616b ||w|xA|xxN 1617b ||w|xAx|N 1618b ||w|w|w|xAx|xN 1619b 0o||(Ax)|N|-xN 1620b ooA|xNx 1621b ||w|xAx=N 1622b °xxAx|xN 1623b A=Nx|N 1624b oo ||A=Nx|xN 1625b ||wx|w|w|w|A|Nx 1626b ||(Ax)|N(n)x 1627b oo||Ax|xNx 1628b xA|Nx 1629b ooA|xNx 1630b ||(Ax)|N(n)x 1631b A=Nx|N 1632b Ax=Nx 1633b oo ||A=N|Nx

273

D2 (III) B2(I) Al (III) C2 (II) Al (III) Al (III) C1 (II) Al(III) Al (III) B2(I) C1 (II) Al (III) C2(I) Al (III)

CM) CM) Bl(I) Al (III) Bl(I) A2 (III) Al (III) C2(I) C2(I) El (III) B2(I) C2 (II) D41 (III) B2(I) Bl(I) B2(I) D41 (III) Al (III) Bid) B2 (II) El (III) E2 (III)

CM)

D2 (III) A2 (III) C1 (II) Al (III) D2 (III) El (III) Al (III) A2a (III)

Index

1635a 1636a 1637a 1638a 1639a 1640a 1641a 1642a 1643a 1644a 1645a 1646a 1647a 1648a 1649a 1650a 1651a 1652a 1653a 1654a 1655a 1656a 1657a 1658a 1659a 1660a 1661a 1662a 1663a 1664a 1665a 1666a 1667a 1668a 1669a 1670a 1671a 1672a 1673a 1674a 1675a 1676a 1677a 1678a

Ax|Nx xxA=(N)x Ax=Nx (Ax)=Nnx xxA=Nx xxA=(N)x ||w|w|Anx (Ax)|A=(N)x Ax|Ax ++ Ax|xxAx ||w|w|A|{N} A=Nx|N ++ (Ax)|Ax=N ||w|w|xAx Ax|Nx ++ (Ax)-N|xAx ++ (Ax)=B|A-N ||An|(Nx)(n)x ||w|w|w|xA=N A|Nnx ++ Ax|xAx ||w|w|A=Nx ++ Ax|xx(Ax)x Ax=Nx °A|xNx ||xwx|w|xAx ++ A|xAx ||w[w|xAx ||w|w|xAx|xN ++ A=N|Ax (Ax)x|Nx ||xa|w|xxxAx Ax|Ax xa|Ax=N ++ Ax|A=Nx ++ Ax|x(Ax)x ++ A=N|(Ax)x llw[w|w|wx|xAx A-N|(A)x xAx|xN °(Ax)x|x(Nx)x Ax|Nnx ++ A[x]=(N)x|Ax ||w|w|Ax|N ++ Ax|A=(N)x

1634b ||(Ax)=Nx|N 1635b (Ax)x|Nx 1636b wx|A-Nx 1637b ||Ax|Nx 1638b ooAx|x(Nx)x 1639b Ax|Nx 1640b x(Ax)|Nx 1641b Ax=Nx 1642b ||A=Nx|N 1643b (Ax)=Nx|N 1644b Ax|Nx 1645b ooAx|xNx 1646b An|Nx 1647b xA|(N)x 1648b ||w|(Ax)|Nx|| 1649b xxx(Ax)x|N 1650b oo||(Ax)|N|-Bx 1651b A|Nnx 1652b (Ax)|N(n)x 1653b Ax|Nx 1654b ||w|w|w|A|Nx 1655b ooAx|xNx 1656b oo||A|xNx 1657b ||A-Nx|N 1658b ||wx|w|A|Nx 1659b xAnx 1660b ||w|xAx|(Nx) 1661b Ax|Nn 1662b (Ax)|N(n)x 1663b ||Ax|N(n)x 1664b ||w|w|xAx|xN 1665b ||w|w|A|xN 1666b ||w|xAx=N 1667b ||w|xA|xN 1668b ||w|xA|(N)x 1669b ||(Ax)=Nx|N 1670b ||w|w|xAx|N 1671b ||w|w|x(Ax)x|N 1672b xxxAx | xN 1673b ooAxNx 1674b ||w|w|w|xAx|xN 1675b ooxANx 1676b ||w|w|A|(N)x 1677b (Ax)x|Nx 1678b xA|(N)x

Al (III) C2 (II) Al (III) D 1 (III) C1 (II) C2 (II) C1 (la) D2 (III) Al (III) A2 (III) C1 (la) El (III) D41 (III) A3 (la) Al (III) A2 (III) A2a (III) D*2 (III) A3 (la) D 1 (III) A2 (III) C1 (la) A2 (III) Al (III) Al (III) A3 (la) Al (III) A3 (la) B2 (la) A2a (III) Al (III) A3 (la) Al (III) Bl(Ib) D*1 (III) A2 (III) A2a (III) A3 (la) A2k (III) B2 (II) A2 (III) D #1 (III) A2 (III) Bl (la) D*2 (III)

274

El (III) Al (III) C1 (I)* Al (III) A2 (III) Al (III) C1 (II) Al (III) El (III) El (III) Al (III) A2 (III) Al (III) C2 (II) B 1 (II) D 1 (III) D 1 (III) D2 (III) Al (III) C^I) A2 (III) Al (III) El (III) Cl(l) C1 (II) B1(I) Al (III) D2 (III) D*2 (III) B2(I) Bl(I) C2(I) El (III)

B2 (II) Al (III) B2(I) C1 (II) C2(I) Al (III) C2 (II)

Alliteration, scansion and metrical clause structure + +

1679* Ax|A=xN 1680a xxAx|(Nx) 1681a + + A[x]=(Nx)|xA 1682a A=N|(A)x 1683a Ax|Nx 1684a ||xxA|xN 1685a xAnx 1686a l|wx w|x(Ax)=Nx 1687a ||An|(Nx)(n)x 1688a Ax|Nx 1689a A=xNx 1690a (Ax)|Anx ++ ||Ax|xAx 1691a 1692a Ax|Nx 1693a x(Ax)x|A 1694a llw w|xxAx 1695a xA=(N)x 1696a x(Ax)|xxA 1697a Anx|N ++ 1698a (Ax)=N|xA=N 1699a (Ax)|N(n)x 1700a ||w|ur w|Ax 1701a ax|xAx ++ A|Ax=B 1702a 1703a xax|(Ax)x 1704a xA=(A)x 1705a a|xxAx|xBx ++ (Ax)|xAx|=Nx 1706a ++ 1707a Ax||w w|Ax|-Nx 1708a w|A=Nx ++ 1709a (Ax)x|xAx 1710a (Ax)x|A(n)x 1711a ||xxa w|w xAx 1712a xxA=(N)x 1713a ||a|Ax=N 1714a A=xNx 1715a Ax|Nx 1716a llw w|wx|Ax|N 1717a (Ax)x|Nx ++ A|x(Ax)x 1718a ++ 1719a A=N|A=N ++ 1720a (Ax)|xxAx 1721a ||w|w|xxAx 1722a + + A=(N)x|A-N 1723a + + |A=Nx|xA

D*41 (III) Bl (II) El (III) A2k (III) Al (III) B1 (II)* C1 (II) C1 (la) D*2 (III) Al (III) Al (III) D 1 (III) A2 (III) Al (III) Bl(II) A3 (la) C2 (II) B2 (II) El (III)** A2b (III) D2 (III) A3 (la) A3 (Ib) D41 (III) A3 (Ib) C2 (II) hA2 (Ib) HAl (III) HAl (III) C1 (la)* A2 (III) D*2 (III) A3 (la) C2 (II) Bl(Ib) Al (III) Al (III) Bl (la) Al (III) Al (III) A2ab (III) A2 (III) A3 (la) A2 (III) E2 (III)

1679b ||w|xA|xN 1680b (Ax)x|{N} 1681b ||w|w|x(Ax)|xN 1682b oo(Ax)|N-(N)x 1683b ||w|xAx|N 1684b (Ax)=(N)nx 1685b xA|Nx 1686b Ax|Nx 1687b ||A|N(n)x 1688b ||xw|w|A|(N)x 1689b ||wx|A|xN 1690b Anx|N 1691b ||w w|Ax|N 1692b llw w|Ax=N 1693b An|Nx 1694b Ax|Nx 1695b ooAx|xNx 1696b ||w|xA|xN 1697b Ax|Nx 1698b ||w|xAx|N 1699b oo||A(n)x|Nx 1700b llw w|A|xN 1701b oo||A|N|-xN 1702b ||w|xA|Bx 1703b oo||A|w|xNx 1704b oo(Ax)|=N|Nn 1705b ||w|w|w|xAx|Bx 1706b oo||w w|w|Ax|xNx 1707b ||w|w|xAx|Nx 1708b Ax|Nx 1709b ||xw|(Ax)n|N 1710b A=Nnx 1711b llw xA=Nx 1712b (Ax)x|Nx 1713b A=xNx 1714b ||w w|w|Ax|N 1715b A=Nx|N 1716b (Ax)x|Nx 1717b xxAxN 1718b ||wxx|w|xAx|N 1719b ||wx|Ax N 1720b oo||A-N|xN 1721b A|N(n)x 1722b oo||w|w|A|xN 1723b ||w|xA|xN

275

Bl (I) Al (HI) Bl (I) D2 (HI) Bid) D3 (III) C1 (II) Al (III) D2 (HI) C2(I) Bl (I) El (III) Bl (I) Bl (I) Al (III) Al (III) A2 (III) Bl (I) Al (III) Bl(I) A2 (III) Bl(I) D41(111

Oil)

A2 (Ic) A2a (III) hAl (I) hA2 (I) hAl (I) Al (III) Bid) D 1 (III) C1 (I) Al (III) Al (III) Bid) El (III) Al (III) Bl (II) Bl(I) Bl(I) El (III) D2 (III) Bl(I) Bl(I)

Index 1724a xa|Ax|N 1725a ||w|Ax|N 1726a xAx|(Ax) 1727a + + A | x A = ( N ) x 1728a ||wx|w|x(A)x 1729a + + A x | A = x N 1730a ||wx|w|xAx 1731a xAnx 1732a ||xw w|xxA[x}x 1733a Ax|Nx 1734a x A = N x 1735a ||ax|w|xAx 1736a + + A|xAx 1737a x(Ax)|Ax 1738a + + A = ( N ) x | A x 1739a ax|xAx 1740a ||w|w|w|xAx 1741a + +A>c xAx 1742a A(n)x|Nx 1743a + + Ax|xAx 1744a ||w|w|xA=(N)x 1745a ||wx|w|xAx 1746a (Ax)x|Nx 1747a + + A|A[x}=x(Nx) 1748a ||wx|w|xAx 1749a ||ax|A=Nx 1750a Ax|Nx 1751a x(Ax)|xxAx 1752a Ax|An 1753a ||w|xAx=N 1754a ||w|xA=(N)x 1755a + + Ax|xAx 1756a llw w | n A = N x 1757a + + A x | A = x N 1758a ||xa w|xx(Ax)=N 1759a A|Nx 1760a Ax|Nx 1761a A x N x 1762a Ax|Nx 1763a ||w|w|A|xxA 1764a xxAx|A 1765a xx(Ax)|Nx 1766a xx(Ax) Ax 1767a x(Ax)|xxAx 1768a IN w A=(N)x

1724b oo||Axw|xNnx 1725b Ax|Nx 1726b Ax|Nx 1727b oo||w|w|Ax|xN 1728b ooAx|Nx 1729b Ax|Nx 1730b AxNx 1731b A=N|(N)x 1732b (Ax)x|Nx 1733b ||w|wlxAx|xN 1734b ooAx|xNx 1735b ||w|wx|A|(N)x 1736b llw w|An=N 1737b ||w x(Ax)|Nn 1738b ||w w|A(N)x 1739b ||w|xAx|xN 1740b (Ax)=Nx|N 1741b ||wx|xA|(N)x 1742b ||w|xA|xN 1743b 00||(Ax)|Nx|-N 1744b (Ax)x|Nx 1745b xxA|(N)x 1746b ||w|xAx|xN|| 1747b Ax|Nx 1748b ||w|w|Ax|N 1749b ||wx|xA|(N)x 1750b ||w|w|xA=xN 1751b ||w|w|w|w|A|Nx 1752b A=Nx|N 1753b ooA|xNx 1754b ooAx|xNx 1755b OOA -Nx|N 1756b Ax|Nx 1757b oo||(Ax)x|xNx 1758b An|Nx 1759b ||w|w|xAx|xN 1760b oo||(Ax)=Nx|xN 1761b ||w|w|xxAx|N 1762b | A|-Nx|N 1763b oo(Ax)x|xNx 1764b xxAx|N 1765b xxAxN 1766b ||wx|A(n)x|N 1767b ||Anx|N 1768b o o A xxNx

Bl(Ib) Bl (la) Bl(II) D*2 (III) A3" (la) D*41 (III) A3 (la) C1 (II) A3 (la) Al (III) C1 (II) A3 (la) Al (III) C1 (II) A2 (III) A3 (Ib) A3 (la) A2 (III) A2 (III) A2 (III) C2 (la) A3 (la) Al (III) D41 (III) A3 (la) C1 (Ib) Al (III) B* (II) Al (III) Bl(Ia) C2 (la) A2 (III) C1 (la) D*41(III) A3 (la) F (III) Al (III) Al (III) Al (III) B2 (la) B 1 (II) C1 (II) C1 (II) B* (II) C2 (la)

276

D*1 (Ie) Al (III) Al (III) B2 (I)** Al (III) Al (III) Al (III) A2k (III) Al (III) B2(I) A2 (III) C2(I) Bl (I)

Oil) Oil)

B2(I) El (III) C2(I) B1(I) D41 (III) Al (III) C2 (II) B2(I) Al (III) Bl(I) C2(I) Bl (I)

Oil)

El (III) Al (III) A2 (III) El (III) Al (III) A2 (III) Al (III) B2(I) E2 (III) B1(I) El (III) A2 (III) Bl (II) Bl (II) B2(I) El (III) A2 (III)

Alliteration, scansion and metrical clause structure 1769a ||w|w|A=(N)x 1770a a | xxAx 1771a axx|Ax 1772a + +Ax|xAx 1773a xxAx|xN 1774a ||w|w|w|xAx 1775a + +A|xx(Ax)x 1776a A=xNx 1777a ||w|xxAx 1778a ++ A=(N)x|Ax 1779a Ax|Nx 1780a ||w|w|xxx(Ax)x 1781a xxA|xN 1782a ||w|w|xAx 1783a A=xAx 1784a ++Ax|xAx 1785a + + ||A|w|A=N 1786a Ax|Nx 1787a ||w|w|A|xA 1788a A=Nnx 1789a Ax|Nx 1790a ++ A|xxA=(N)x 1791a ||wx|Ax=N 1792a (Ax)x|Nn 1793a + +Ax|A=(N)x 1794a ||ax|w|(Ax)=N 1795a Ax=Nx 1796a ||w|xA-Nx 1797a Ax | Ax 1798a (Ax)=Nnx 1799a ||ax|wx|w|A=B 1800a + + A|xA=N 1801a ||w|w|A|(N)x 1802a ++ A=N|(Ax)x 1803a supplied 1804a ||wx|(Ax)nx 1805a ++Ax|x(Ax)x 1806a ++ (Ax)|Ax=N 1807a ||a|w|xAx 1808a (Ax)|Nnx 1809a A-N|Nx 1810a ||w||w|wx|A=(N)x 1811a A=Nnx 1812a Ax|Nx 1813a ||w|xA=(N)x

C2 (la) A3 (Ib) A3 (Ib) A2 (III) B2 (II) A3 (la) A2 (III) Al (III) A3 (la) A2 (III) Al (III) A3 (la) Bl (II) A3 (la) Al (III) A2 (III) A2b (Ic) Al (III) Bl (la) D 1 (III) Al(III) D*2 (III) Bl(Ia) Al (III) D*2 (III) A3 (la) Al (III) C1 (la) Al(III) D 1 (III) A3 (la) A2b (III) C2 (la) A2a (III) C1 (la) A2 (III) D41(III) A3 (la) D 1 (III) Al (III) C2 (la) D 1 (III) Al (III) C2 (la)

1769b OOA|N-Nx 1770b ||w|w|Ax|xN 1771b xxxAx=N 1772b oo||w|w|w|Anx 1773b x(Ax)|xNx 1774b A-Nx|N 1775b ||wx|Ax|N 1776b A-Nx|N 1777b Anx|N 1778b w|(Ax)x|N 1779b w w|w|xAx|xN 1780b (Ax)=Nnx 1781b Ax|(Nx)x 1782b A[x}=Nx|N 1783b ||w|w|A|(N)x 1784b ||wx|Ax|N 1785b oo||A|Nx|-N 1786b ||w|xAx|N 1787b Ax=Nx 1788b oo(Ax)x|xNx 1789b oo ||A=N|xN 1790b oo||(Ax)|=N|xN 1791b Ax|Nx 1792b oo||A|nx(Nx)|-N 1793b Ax|Nx 1794b Ax|Nx 1795b A|N(n)x 1796b ooAx|x(Nx)x 1797b oo||wx|AN[x]x 1798b Ax|Nx 1799b ||(Ax)|B(n)x 1800b oo||A|Nx|-N 1801b (Ax)x|Nx 1802b ||w|w|A|(N)x 1803b ||(Ax)|Nnx 1804b ooA|xNx 1805b ||wx|A|(N)x 1806b Ax|Nx 1807b An|(N)x 1808b ||w|xA|(N)x 1809b °||wx|w|xAx|N 1810b Ax|Nx 1811b ||wx|Ax|N 1812b ||w|w|Ax|N 1813b Ax|Nx

277

D 1 (III) B2(I) B 1 (II) Cl(l) C * (II) El (III) B1(I) El (III) El (III) B1(I) B2(I) D 1 (III) A l (III) El (III) C 2 (I)

A l (III) A 2 (III) El (III) El (III) A l (III) F (III) A l (III) A l (III) D 2 (III) A 2 (III) C^I) A l (III) D 2 (III) D41(III) A l (III) C 2 (I) D 1 (III) A l (III) C 2 (I) A l (III) A 2 k (III) C 2 (I) B1(I) A l (III) A l (III)

Index 1814a An|Ax 1815a ++(Ax)n|xAx 1816a + + ||(Ax)|Ax=N 1817a ||An|(Nx)(n)x 1818a ||w|w|A=Nn 1819a Ax|(Nx)x 1820a (Ax)n|Nx 1821a + +Ax|x(Ax)x 1822a ||w|w|wx|xAx 1823a xxA=(N)x 1824a (Ax)x|Bx 1825a A=xNx 1826a ||w|w|w|xAx 1827a ||w|w|A=Nn 1828a ||w|w|Anx 1829a ||w|w|Anx 1830a ++(Ax)x|xAx 1831a Ax|Nx 1832a Ax|Nx 1833a + +Ax|xAx 1834a ||w|w|xAx 1835a (Ax)x|Nx 1836a ||w|w wx|An 1837a xax|Ax|N 1838a Ax|Ax 1839a + +Ax|xAx 1840a ||An|(Nx)(n)x 1841a ||w|xA=(N)x 1842a x(Ax)|Ax 1843a xxAx|(Nx) 1844a ||w|w|(Ax)x|N 1845a A|A=(N)x 1846a ||w|w[xAx 1847a A|(Ax)=Nx 1848a ++A|xxAx 1849a Ax|Bx 1850a ||w|w|A=Nx 1851a xxAnx 1852a ++ A=B|(Ax)x 1853a Ax|Nx 1854a ax|A|xN 1855a ||wx|w|xAx 1856a Ax|Nx 1857a °A|xNx 1858a An=Nx

1814b ||wx|A|(N)x 1815b oo||w|xAx|N 1816b An|Nx 1817b A|Nnx 1818b Ax|Nx 1819b oo||w|w|A(n)x 1820b ||wx|A|(N)x 1821b llw w|A|Nx 1822b Anx|N 1823b ooAx|(Nx)x 1824b ||wxw|A|(B)x 1825b ||w|w|CAx)|Nx 1826b xxAx|xN 1827b (Ax)x|Nx 1828b Ax|(N)x 1829b Ax|Nx 1830b ||w|x(Ax)nx|N 1831b ||w|w w|A|{N} 1832b ||w w|w|Ax|(Nx) 1833b ||w w|w|A|(Nx)x 1834b A=N|(N)x 1835b llw w|w|Ax|N 1836b x(Ax)|Nx 1837b oo||wiw|A|(N)x 1838b ||A=Nx|N 1839b ||w|w w|Ax|N 1840b ooA|xN-(N)x 1841b Ax|Nx 1842b ||xwx|w|(Ax)=Nx 1843b (Ax)|N(n)x 1844b ||w|xAx|N 1845b ||A|w|(Nx)x 1846b ||w|w|A|(N)x 1847b Ax|(Nx)x 1848b Ax|Nx 1849b ||w w|xA|(B)x 1850b Ax|Nx 1851b (Ax)|Nnx 1852b llw w|Ax|B 1853b ||w|xA=(N)x 1854b Ax|Nn 1855b ||w|xAx|N 1856b xA=(N)x 1857b ||w|(Ax)|Nx 1858b llwl w|A|(N)x

Al (III) A2 (III) D41 (III) D*2 (III) C1 (la) Al (III) Al (III) A2 (III) A3 (la) C2 (II) Al (III) Al (III) A3 (la) C1 (la) C1 (la)** C1 (la) A2 (III) Al (III) Al (III) A2 (III) A3 (la) Al (III) A3 (la) Bl(Ib) Al (III) A2 (III) D*2 (III) C2 (la) C1 (II) Bl (II) Bl(Ia) D2 (III) A3 (la) D 1 (III) A2 (III) Al (III) C1 (la) C1 (II) A2a (III) Al (III) Bl(Ib) A3 (la) Al (III) Al (III) Al (III)

278

C2(I) B1(I) Al (III) D 1 (III) Al (III) C2(I) C2(I) C^I) El (III) Al (III) C2(I) C^I) B2 (II) Al (III) F (III)** Al (III) B2(I) O{1)

B1(I)

Cl{\) A2k (III) B1(I) C1 (II) C2(I) El (III) B1(I) D*2 (III) Al (III) Cl{\) D2 (III) Bl(I) Al (Ic) C2(I) Al (III) Al (III) C2(I) Al (III) D 1 (III) B1(I) C2(I) Al (III) B1(I) C2 (II) C^I) C2(I)

Alliteration, scansion and metrical clause structure 1859a + + ||ax||wx|w|Ax 1860a + +Ax|xAx 1861a + + AxxAx 1862a ||w|A=(N)x 1863a + + A|xA=Nx 1864a xxA|xxA 1865a + + A-N|A=Nx 1866a ||xw|w|Ax|N 1867a (Ax)|N(n)x 1868a llw wx xxAx 1869a ax xxAx 1870a HxAxlN 1871a An|Nnx 1872a ||w|xAx|xN 1873a Ax=Nx 1874a ++ Ax|A-Nx 1875a llw w|Ax|N 1876a ++A(n)x|xAx 1877a ||w w|xxA=N 1878a ||w|w|xAx 1879a xxAx|N 1880a a|xAx 1881a + + A = N | A = N 1882a Ax|Nx 1883a An={N} 1884a ||w|w|xAx 1885a + +A|xAx 1886a ++ A-N|A-Nx 1887a (Ax)x|Nx 1888a llw w|xAx 1889a A=Nx|A 1890a + +(Ax)x|(Ax)=Nx 1891a + + A=N|Ax 1892a ||w|w|xAx 1893a Ax|Ax 1894a ||w||w|A-(N)x 1895a (Ax)|A=(N)x 1896a llw w|xAx 1897a ax|(Ax)=Nx 1898a + +Ax|xAx 1899a xxAnx 1900a ||w|xA=Nx 1901a ++ A|xAx 1902a x(Ax)=Nx 1903a Ax=Nx

A3 (Ib) A2 (III) A2 (III) C2 (la) D* 1 (III) B2 (II)A D* 1 (III) Bl(Ia) D 2 (III) A3 (la) A3 (Ib) Bl(Ia) D* 1 (III) B2 (la) Al (III) D* 1 (III) Bl(Ia) A2 (III) A3 (la) A3 (la) Bl (II) A3 (Ib) A2ab (III) Al (III) Al (III) A3 (la) Al (III) D* 1 (III) Al (III) A3 (la) El (III)

1859b Ax|Nx|| 1860b oo||(Ax)|Nnx 1861b xx(Ax)x|N 1862b xx(Ax)|Nx 1863b ||w|xAx|N 1864b ooAx|xNx 1865b Ax Nx 1866b ooAx|xNx 1867b Ax|Nx 1868b Ax|Nx 1869b oo||Ax|N|-(N)x 1870b (Ax)|(Nx)x|=N 1871b A|Nx 1872b oo||(Ax)|w|Nx 1873b ||w|w|Ax|N 1874b ooAx|Nx 1875b xA Nx 1876b llw w|xA|xxN 1877b x(Ax)|xNx 1878b (Ax)=Nx|N 1879b Ax|Nx 1880b ||w|An|(Nx) 1881b A=Nx|N 1882b ||A=Nx|N 1883b ||w|w|xAx|N 1884b (Ax)|Nnx 1885b llw w|A|(N)n 1886b llw w|wx|Ax|xN 1887b oo||w|ww|(Ax)xlN 1888b (Ax)|Nnx 1889b o o ||A=N|Nx 1890b o o ||A=N|xN 1891b ||w|w|A|(N)x 1892b xAx|(Nx) 1893b ||w|w|xAx|N 1894b (Ax)x|Nx 1895b x(Ax)|Nx 1896b A=N|(N)x 1897b Ax=Nx 1898b ||A|N(n)x 1899b A=xNx 1900b Ax|Nx 1901b ||w|w|Ax|N 1902b ooAx|xNx 1903b oo lh cw|w|A|(N)x

D^dll) A2a (III) A3 (la) Al (III) C2 (la) D 2 (III) A3 (la) C1 (Ib) A2 (III) C1 (II) C1 (la) Al (III) C1 (II) Al (III)

279

Al (III) D 1 (III) Bl (II)

C 1 (ID Bl(I) A2 (III) Al (III) A2 (III) Al (III) Al (III) D*2 (III) D41(III)** F (III) Al (Ic) B1(I) Al (III) C1 (II) B2(I) C* (II) El (III) Al (III) Bl(I) El (III) El (III) Bid) D 1 (III) C2(I) B2(I) Bl(I) D 1 (III) A2a (III) El (III) C2(I) Bl (II) Bid) Al (III) C1 (II) A2k (III) Al (III) D2 (III) Al (III) Al (III) Bl(I) A2 (III) C2(I)

Index 1904a ax|A|(N)x 1905a ||w|w|xAx 1906a + + A|Ax|=N 1907a ||w|w|A=(N)x 1908a °Ax|xNx 1909a ||a|Ax=N 1910a Ax=Bx 1911a llw w|Ax|(Nx) 1912a Ax|Nx 1913a ||A=xNx 1914a ax w|xAx 1915a ||w|w|w|Ax|N 1916a ++A|x(Ax)x 1917a ||ax|xAx 1918a A[x]=Nx|N 1919a (Ax)|A=(N)x 1920a llw w|A|(N)x 1921a + + Ax|xA=N 1922a xxAnx 1923a (Ax)n|An 1924a + +Ax|xxAx 1925a + + ||A|w|A-N 1926a ++ A|xAx 1927a A|A-Nx 1928a xxA=(N)x 1929a (Ax)x|Nx 1930a ||w|xA|(A)x 1931a A=xNx 1932a ++ (Ax)|Ax|=N 1933a ||wx|w|Ax 1934a + +Ax|xAx 1935a llw wx|A|(N)x 1936a ||w|w|A=Nx 1937a A=x(Bx)x 1938a xxA=(N)x 1939a ||w|w|Ax=B 1940a ++ A=(N)x|Ax 1941a ++(Ax)|xAnx 1942a ||wx|(Ax)=Nx 1943a xx(Ax)=Nx 1944a ||wx|w|xA(n)x 1945a (Ax)=Nnx 1946a ||w|w|A=(Nx)x 1947a An=Nx 1948a ax|A=(N)x

C2 (Ib) A3 (la) D41 (III) C2 (la) A2 (III) Bl(Ib) Al (III) Bl(Ia) Al (III) Al (III) A3 (la) Bl(Ia) Al (III) A3 (Ib) El (III) D 2 (III) C2 (la) A2b (III) C1 (II) Al (III) A2 (III) Al (Ic) Al (III) D 1 (III) C2 (II) Al (III) C2 (la) Al (III) D41 (III) A3 (la) A2 (III) C2 (la) C1 (la) Al (III) C2 (II) Bl(Ia) A2 (III) HC1 (III)** C1 (la) C1 (II) C2 (la) D 1 (III) C1 (la) Al (III) C2 (Ib)

1904b oo ||(Ax)|=N|xN 1905b (Ax)=Nx|N 1906b oo||A=(N)x|(Nx)x 1907b ooA|xxNx 1908b ||A=Nx|N 1909b ooA|xxNx 1910b xxA=Bx 1911b x(Ax)|Nx 1912b oo ||A|N|-xN 1913b xAx|N 1914b A=N|(N)x 1915b Ax|Nx 1916b A|N(n)x 1917b A=Nx|N 1918b llw w|w|Ax|N 1919b x(Ax)|Nx 1920b oo(Ax)nx|xN 1921b ||w|w|A|(N)x 1922b Ax|Nx 1923b ||w|xA|(N)x 1924b A=Nx|N 1925b ||(Ax)=N|(N)n 1926b oo ||A|Nx|-N 1927b llw w|Ax|N 1928b x(Ax)|Nx 1929b ||w|w|A|xN 1930b Ax|Nx 1931b ||Anx|N 1932b oo(Ax)|N-Nx 1933b ooA|xNx 1934b ||wx|A={N} 1935b Ax|(Nx)x 1936b (Ax)x|Nx 1937b ||(Ax)|-Nx|B 1938b oo Ax|xNx 1939b Ax|Bx 1940b ||xw|w|A-N|N 1941b ||w|ww|A=(N)x|N 1942b oo Ax|xNx 1943b Ax|Nx 1944b Anx|N 1945b oo Ax|Nx 1946b ooA|x(Nx)x 1947b ||wx|Ax|N 1948b Ax|Nx

280

El (III) El (III) A2 (III) A2 (III) El (III) A2 (III) C1 (II) C1 (II) D41(111 Bl (II) A2k (III) Al (III) D 2 (III) El (III) B1(I) C1 (II) E2 (III) C 2 (I) Al (III) C 2 (I) El (III) A2k (III) D 4 1(III) Bl (I) C1 (II) B1(I) Al(III) El (III) D 1 (III) Al (III) C^I) A l (III) Al (III) El (III) A2 (III) Al (III) Bl(I) E2 (I)** A2 (III) Al (III) El (III) Al (III) A l (III) Bl (I) Al (III)

Alliteration, scansion and metrical clause structure 1949a (Ax)x|Nx 1950a xx(Ax)x|A 1951a ++ Ax|xAx 1952a xA=Nx 1953a A=xNx 1954a ||a|A=(N)x 1955a xxA=Nx 1956a xxAnx 1957a Ax=Nx 1958a + +(Ax)|xAx 1959a ++ Ax|xAx 1960a Ax|Nx 1961a ++(Ax)x|xAx 1962a (Ax)|Nnx 1963a ||xw w|w|xAx 1964a ++ A|xxAx 1965a Ax|(Ax)x 1966a ++ (Ax)|Ax = N 1967a ++ Ax|xAx 1968a (Ax)|Bxn[x} 1969a + + Ax|A=(N)n 1970a Ax Nx 1971a A|Nnx 1972a ||w w|xAx 1973a A=xNx 1974a (Ax)=Nx|A 1975a l[ax|w|xAx 1976a Ax=Nx 1977a ||xa|w|xAx 1978a ++ A|xAx 1979a xAx=(Nx) 1980a Ax|Nx 1981a xxA=(N)x 1982a ||axx|xAx 1983a ++(Ax)x|xAx 1984a ++ AxxAx 1985a Ax|Ax 1986a ||wx|A=Nx 1987a ||w|a|w|xAx 1988a llw w|Anx 1989a Ax|Ax 1990a ++Ax|x(Ax)x 1991a A=Nx|A 1992a Ax|Nx 1993a A=Nx|A

Al (III) Bl(II) A2 (III) C1 (II) Al (III) C2 (Ib) C1 (II) C1 (II) Al (III) Al (III) A2 (III) Al (III) A2 (III) D 1 (III) A3 (la) A2 (III) Al (III) D 4 1 (III) A2 (III) D41 (III) D* 2 (III) Al (III) D 1 (III) A3 (la) Al (III) El (III) A3 (la)** Al (III) A3 (la) Al(III) Bl(II) Al (III) C2 (II) A3 (Ib) A2 (III) A2 (III) Al (III) C1 (la) A3 (la) C1 (la) Al (III) A2 (III) El (III) Al (III) El (III)

1949b ||wx|w|Ax|N 1950b x(Ax)|Nx 1951b ||w|w|Ax|N 1952b Ax|Nx 1953b (Ax)nx|N 1954b x(Ax)x|(Nx) 1955b ooAxxNx 1956b xA|Nx 1957b ||xw|Ax|N 1958b A=Nx|N 1959b | A=Nx|N 1960b wx|An|N 1961b Anx|N 1962b Ax|Nx 1963b xxA=(N)x 1964b A=N|(N)x 1965b ||(Ax)=Nx|N 1966b ||w|A|(N)x 1967b ||ww|w|Ax|N 1968b ooAx|xBx 1969b oo AxxNx 1970b ||(Ax)nx|N 1971b ooAx|xNx 1972b Anx|N 1973b (Ax)nx|N 1974b x(Ax)|Nx 1975b ||w|xAx|xN|| 1976b oo A|Nx-N 1977b ||w|xAx|xN|| 1978b ||wx|A=Nx 1979b ooAx|xNx 1980b ||(Ax)=Nx|N 1981b (Ax)x|Nx 1982b ||A=Nx|N 1983b oo||(Ax)n|xN 1984b x(Ax)|xN 1985b ||wx|An|N 1986b Ax|Nx 1987b Ax|Nn 1988b ooA|x(Nx)x 1989b xxA|(N)x 1990b ||w|w|Anx 1991b ooAx|xNx 1992b ||w|w|A=(N)x 1993b oo||Ax|x(Nx)x

281

Bl (I) C1 (II) Bl (I) Al (III) El (III) B 1 (II) A2 (III) C1 (II) Bl (I) El (III) El (III) Bl(I) El (III) Al (III) C2 (II) A2k (III) El (III) C2(I) Bid) A2 (III) A2 (III) El (III) A2 (III) El (III) El (III) C1 (II) B2(I) D41 (III) B2(I) Cl(l) A2 (III) El (III) Al (III) El (III) El (III) Bl (II) Bid) Al (III) Al (III) Al (III) C2 (II) CX(I) A2 (III) C2(I) A2 (III)

Index 1994a Ax|Nx 1995a ||w|w|xxA=B 1996a ||wx|A=(N)x 1997a + +Ax|xAx 1998a ||w|w|w|w|xAx 1999a ||An|(Nx)(n)x 2000a ||w|w|nAx 2001a + +(Ax)xAn 2002a ||w|An=N 2003a a | xxAx 2004a (Ax)=Nnx 2005a ++Ax|xAx 2006a ||w|xAx|xN 2007a ax | xxAx 2008a ||w|w|Ax|(Ax) 2009a + +Ax|xAx 2010a xxA=(N)x 2011a ||wx|w|xAx 2012a ||wx|w|A=(N)x 2013a xxAx|(Ax) 2014a + + ||(Ax)|w|xAx 2015a xx(Ax)x|A 2016a + +(Ax)=N|Ax 2017a + +(Ax)=N|Ax 2018a ||ax|(Ax)|Nx 2019a Ax | Ax 2020a ||wx|x(Ax)x 2021a + +Ax|xAx 2022a ||w|w|A(n)x 2023a Ax|Nx 2024a (Ax)x|Nx 2025a A|A=(N)x 2026a llwx[wlxAx 2027a Ax|Nx 2028a ||w|w|xxAx 2029a ++Ax|xAx 2030a xxA=(B)x 2031a + + A=N|Ax 2032a llw[w|wx[xAx 2033a xAx|xN 2034a ||wx|w|xAx 2035a ||A=N|(A)x 2036a llxw[(Ax)x 2037a + + A|xA=N 2038a ||wx|w|xAx

1994b ||w|w|Ax|N 1995b ooAx|xBx 1996b ooAx|xNx 1997b oo||(Ax)|w|N|-Nx 1998b xA|Nx 1999b A|Nnx 2000b Ax|(Nx)n 2001b oo(Ax)xlNx 2002b Ax|Nx 2003b ||w|w|Ax|(Nx) 2004b ooAx|x(Nx)x 2005b ||w|xA|xN 2006b A(n)x|Nx 2007b A=N|(N)x 2008b Ax|Nx 2009b ||w|w|Ax|N 2010b An|Nx 2011b (Ax)|N(n)x 2012b ooAx|Nx|| 2013b ooA|xNx 2014b ||w|w|w|Ax|N 2015b A=Nnx 2016b ||wx|Ax|N 2017b oo A|=N|xN 2018b ||w|w|A=(N)x 2019b ||w|w|xAx|N 2020b Ax|Nnx 2021b (Ax)=Nx|N 2022b A=Nnx 2023b ||w|w|Ax|N 2024b jjw|xAx|N 2025b oo(Ax)|=(N)x|Nx 2026b (Ax)|Nnx 2027b ||w|w|A|(N)x 2028b A=Nx|N 2029b ||w|Ax|N 2030b Ax|Bx 2031b ||w|xA|(N)x 2032b Ax|(Nx)=Nx 2033b ooAxNx 2034b xA|{N} 2035b oo(Ax)x|x(Nx)x 2036b (Ax)x|Nx 2037b oo(Ax)nx|xN 2038b Ax|Nx

Al (III) A3 (la) C2 (la) A2 (III) A3 (la) D*2 (III) A3 (la) Al (III) Bl(Ia) A3 (Ib) D 1 (III) A2 (III) B2 (la) A3 (Ib) Bl (la) A2 (III) C2 (II) A3 (la) C2 (la) Bl (II) A2 (Ic) Bl(II) A2a (III) A2a (III) C1 (Ib) Al (III) A3 (la) A2 (III) C2 (la) Al (III) Al (III) D2 (III) A3 (la) Al (III) A3 (la) A2 (III) C2 (II) A2a (III) A3 (la) B2 (II) A3 (la) A2k (III) A3 (la) A2b (III) A3 (la)

282

A2 (III) A2 (III) D #1 (Ic) C1 (II) D 1 (III) Al (III) Al (III) Al (III) Bl(I) A2 (III) B1(I) A2 (III) A2k (III) Al (III) B1(I) Al (III) D2 (III) Al (III) Al (III) B1(I) D 1 (III) B1(I) El (III) C2(I) B1(I) D*1 (III) El (III) D 1 (III)

A2 (III) D 1 (III) C2(I) El (III) B1(I) Al (III) C2(I) D #1 (III) Al (III) C1 (II) A2 (III) Al (III) E2 (III) Al (III)

Alliteration, scansion and metrical clause structure 2039a ||w|w|w|xAx 2040a + +Ax|xAx 2041a ||wx|w|xAx 2042a ||A|A=(N)x 2043a ++ A=N|(Ax)x 2044a ||xax|Ax=N 2045a xAx|xA 2046a ++ ||A=(N)x|Ax 2047a ||a|w|A(N)x 2048a ||wx|x(A)x 2049a xx(Ax)=Nx 2050a Ax|Nx 2051a ||ax|A=Nx 2052a xx(Ax)x|(Ax) 2053a ||w|w|xx(Ax)x 2054a Ax|Nx 2055a ||Ax|Nx 2056a ||wx|w|w|xAx 2057a + + ||(Ax)|w|xAx 2058a Ax|Nx 2059a ||w|xAx|N 2060a xxAx|(Ax) 2061a Ax|Nx 2062a ax|(Ax)nx 2063a ||wx|w|x(Ax)x 2064a + + A=N|Ax 2065a ax|A=Nx 2066a xxA=Bx 2067a ||ww|(Ax)=Nx 2068a A = N x A 2069a ++ A=(N)x|Ax 2070a a|xxAx 2071a Ax|Nx 2072a + + A=N|(Ax)x 2073a a xxAx 2074a ++ (Ax)|Ax=N 2075a ||w|w|xAx 2076a ||w|w|A{n} 2077a + + A=(N)x|Ax 2078a Ax|Nx 2079a + + Ax|(Ax)=Nx 2080a ||Ax|Nx 2081a ||w|xw|A|xN 2082a ++ (Ax)|Ax=N 2083a xxA=(N)x

A3 (la) A2 (III) A3 (la) D 2 (III) A2a (III) Bl(Ib) B2 (II) A2 (III) C2 (la) A3" (la) C1 (II) Al (III) C1 (Ib) Bl (II) A3 (la) Al (III) Al (III) A3 (la) A2 (Ic) Al (III) Bl (la)** Bl(II) Al (III) C1 (Ib) A3 (la) A2a (III) C1 (Ib) C1 (II) C1 (la) El (III) A2 (III) A3 (Ib) Al (III) A2a (III) A3 (Ib) D 4 1(III) A3 (la) C1 (la) A2 (III) Al (III) D* 1 (III) Al (III) Bl(Ia) D 4 1 (III) C2 (II)

2039b xxA=(N)x 2040b xxxAx|N 2041b ||w|w|A|xN|| 2042b oo lh iv w|A|xN 2043b ||w|w|A|(N)x 2044b Ax|Nx 2045b (Ax)|N(n)x 2046b ||w|xA|xN 2047b ooAx|xNx 2048b xxAx|N 2049b A(n)x|Nx|| 2050b ||w|wx|(Ax)|Nx 2051b ||wx|(Ax)n|N 2052b (Ax)|Nnx 2053b oo(Ax)|Nnx 2054b xA|{N} 2055b ||w|xxAx|(Nx) 2056b Ax|Nx 2057b °Ax|xNx 2058b ||w|w|A|(N)x 2059b xx(Ax)|Nx 2060b A=N|(N)x 2061b oo||w|xAx|(Nx) 2062b ||w w|A|(N)x 2063b xA|Nx 2064b ||WJ : Anx 2065b ||w|w|A=(N)x 2066b Ax|Bx 2067b ooAx|xNx 2068b (Ax)|nNx 2069b ||w w|A|(N)x 2070b llw w|(Ax)|Nx 2071b ||xw|Ax|N 2072b ||wx|(Ax)x|N 2073b oo ||A|Nx-N 2074b ooAx|Nx 2075b A|N(n)x 2076b OOA|N-Nx 2077b ||w|Ax|N 2078b ||w|Ax|N 2079b xA=(N)x 2080b oo A|=N|xN 2081b Ax=Nx 2082b oo(Ax)x|xNx 2083b Axl S[x

283

C2 (II) Bl (II) B1(I) Bl(I) C2(I) Al (III) D2 (III) Bl(I) A2 (III) Bl(II) A2 (III) CX(I) Bl(I) D 1 (III) D 1 (III) C1 (II) Bl(I) Al (III) A2 (III) C2(I) C1 (II) A2k (III) Bl(I) C2(I) C1 (II) CX(I) C2(I) Al (III) A2 (III) Al (III) C2(I) CX(I) Bl(I) Bl(I) D41(III) Al (III) D2 (III) D 1 (III) Bid) Bid) C2 (II) El (III) Al (III) A2 (III) Al (III)

Index 2084a ||w|w|Ax|N 2085a axx|(Ax)=N 2086a ++ A|xA-N 2087a ||w|w|A-Nx 2088a Ax|Nx 2089a ||w|w|w|xAx 2090a A|A=(N)x 2091a (Ax)x|Bx 2092a ||wx|w|xAx 2093a ||xA|w|xNnx 2094a °Ax|xNx 2095a ||w|w|Ax|N 2096a axx|Ax 2097a ||Ax|Nx 2098a ||wx|w|xAx 2099a ++ A|xAx 2100a Ax|Bx 2101a ||w|xxA=N 2102a Ax|Nx 2103a ++(Ax)xAx 2104a ||w|w|xAx 2105a ||w|w|A|xA 2106a (Ax)=Anx 2107a ||ax|Ax=N 2108a ++(Ax)=(B)x|Ax 2109a ++ A|xA-N 2110a ax|xxAx 2111a ||wx|A|xN 2112a (Ax)|A=(N)x 2113a Ax=Nx 2114a ||wx|w|Ax|N 2115a ||w|w|xAx 2116a Ax|Ax 2117a ax|xAx 2118a (Ax)|A=(N)x 2119a axx|A-N 2120a ++ A=(B)x|Ax 2121a xxA|xN 2122a Ax=Nx 2123a ++ Ax|A=(N)x 2124a |lwx|w|wx[xAx 2125a A=Nnx 2126a ++Ax|xAx 2127a Ax|Nx 2128a AxlAx

2084b A|N(n)x 2085b ||A|N(n)x 2086b (Ax)=Nx|N 2087b ooA|xNx 2088b x(Ax)|Nx 2089b A=Nnx 2090b xA|Nx 2091b oo||w|xAx|B 2092b oo A=N|xN 2093b ||w|w|xA=(N)x 2094b ooA-N|xN 2095b ooAxNx 2096b ||w|xA|(Nx)x 2097b A=Nx|N 2098b (Ax)|N(n)x 2099b ||w|w|A|(N)x 2100b oo(Ax)=B|xN 2101b (Ax)|Nnx 2102b oo(Ax)|N(n)x 2103b ||wx|Ax|N 2104b x(Ax)|Nx 2105b ||(Ax)x|Nn 2106b Ax|Nx 2107b Ax|Nx 2108b ||wx|A|xB 2109b ||wx|An|N 2110b A=N|(N)n 2111b ooAx|xNx 2112b (Ax)x|Nx 2113b oo||(Ax)|Nx|-N 2114b ooA|xNx 2115b A-Nx|N 2116b ||w|w|A|xN 2117b ||w|w|A|(N)x 2118b A(n)x|Nx 2119b oo||(Ax)|N|-xN 2120b ||A|nBx 2121b oo||A|xNx 2122b ||w|w|A(n)x 2123b OOA|N-Nx 2124b ||wx|Ax|N|| 2125b (Ax)x|Nx 2126b ||w|xA|(N)x 2127b ||w|xA|xN 2128b xxAx=N

Bl(Ia) A3 (Ib) Al (III) C1 (la) Al (III) A3 (la) D2 (III) Al (III) A3 (la) F (II)* A2 (III) Bl(Ia) A3 (Ib) Al (III) A3 (la) Al (III) Al (III) A3 (la) Al (III) Al (III) A3 (la) Bl(Ia) D 1 (III) Bl(Ib) A2 (III) Al (III) A3 (Ib) Bl(Ia) D2 (III) Al (III) Bl(Ia) A3 (la) Al (III) A3 (Ib) D2 (III) A3 (Ib) A2(III) Bl(II) Al (III) D*2 (III) A3 (la) D 1 (III) A2 (III) Al (III) Al (III)

284

D2 (III) D2 (III) El (III) Al (III) C1 (II) D 1 (III) C1 (II) B1(I) El (III) C2(I) El (III) Al (III) O{\) El (III) D2 (III) C2(I) El (III) D 1 (III) D2 (III) B1(I) C1 (II) Al (III) Al (III) Al (III) Bl(I) Bl(I) A2k (HI) A2 (III) Al (III) D41 (III) Al (HI) El (III) Bl(I) C2(I) A2 (III) D41 (III) Al (III) Al (III) C2(I) D 1 (III) Bl(I) Al (III) C2(I) Bid) Bl(II)

Alliteration, scansion and metrical clause structure 2129a ||w|w|Anx 2130a wx|w|A=(N)x 2131a ||w|xAx|N 2132a axx|A=N 2133a + + A=(N)x|Ax 2134a ||Ax|(Nx)x 2135a ||w|w|xAx 2136a ++ Ax|(Ax)=Nx 2137a ||w|w|Ax|N 2138a + + ||A|Ax|-N 2139a xxA=(N)x 2140a Ax|Ax 2141a ++A|x(Ax)x 2142a ||w|w|Ax|N 2143a Ax|(Ax)x 2144a ||w|xA=(N)n 2145a ||wx|w|xAx 2146a (Ax)x|Ax 2147a (Ax)|N(n)x 2148a ||w|w|w|A=(N)n 2149a ++ ||Ax|xAx 2150a + + Ax xA 2151a Ax=Nx 2152a ||w|w|A|(N)x 2153a (Ax)=Nx A 2154a + + A=N|(Ax)-N 2155a ||w|xAx=N 2156a Ax|Nx 2157a ||w|w|w|Ax|N 2158a + + ||w||w|w|Ax 2159a A|Nnx 2160a ||w|xw|(Ax)|Ax 2161a (Ax)|(Ax)nx 2162a A=xBx 2163a ||wx|w||w|xAx 2164a + + Ax xAx 2165a Ax=(Nx)x 2166a ++ Ax|xAx 2167a |[wx|An=N 2168a Ax|Nx 2169a A=xNx 2170a Ax|Bx 2171a ||w|x(Ax)|Nx 2172a |ax|w||w|w|xxA=N 2173a + + A=Nx|A[xJ=Nx

C1 (la) C2 (la) Bl (la) A3 (Ib) A2 (III) Al (III) A3 (la) D #1 (III) Bl(Ia) D 4 1(III) C2 (II) Al (III) Al (III) Bl(Ia) Al (III) C2 (la) A3 (la) Al (III) D 2 (III) C2 (la) A2 (III) F (III) Al (III) C2 (la) El (III) A2a (III) Bl(Ia) Al (III) Bl(Ia) A3 (la) D 1 (III) C1 (la) D 1 (III) Al (III) A3 (la) A2 (III) Al (III) A2 (III) Bl(Ia) Al (III) Al (III) Al (III) C 1 (la) A3 (la) D* 1 (III)

2129b Ax|Nx 2130b ooAx|xNx 2131b ooAxNx 2132b ||w|w|xAx|xN 2133b oo||Ax|xNx 2134b ||w|w|Ax|xN 2135b ||w|w|Ax|N|| 2136b A=Nx|N 2137b ooA|xNx 2138b ||w|w|Ax|xN 2139b A(n)x|Nx 2140b ||A=Nx|(Nx) 2141b ||w|w|Ax|xN 2142b oo AixNx 2143b (Ax)|N(n)x 2144b Ax|Nx 2145b x(Ax)|Nx 2146b ||w|w|w|Ax|N 2147b xxxAx|N 2148b Ax|Nx 2149b oo ||w|w|A|xN 2150b ||w|A|(N)x 2151b ||wx|(Ax)n|N 2152b (Ax)=Nxn 2153b Ax|Nx 2154b oo ||A|Nx|-N 2155b An|Nx 2156b ||(Ax)|=Nx|N 2157b ooA|xNx 2158b (Ax)n|(N)n 2159b Ax|Nx 2160b Ax|Nx 2161b ||w|w|w|A|Nx|| 2162b OO||A|-Nx B 2163b Ax|Nx 2164b A|N(n)x 2165b llw w|A|xN 2166b ||w|w|A|{N} 2167b ooAx|Nx 2168b ||A|N(n)x 2169b ||(Ax)nx|N 2170b oo(Ax)|Nx -B 2171b ooAx|xNx 2172b ooAx|xNx 2173b llwx wlwlAnlN

285

Al (III) A2 (III) Al (III) B2(I) A2 (III) B2(I) Bl (I) El (III) Al (III) B2(I) A2 (III) El (III) B2(I) Al (III) D2 (III) Al (III) C 1 (II) Bl (I) B 1 (II) Al (III) Bl (I) C 2 (I) Bl (I) D41(111 Al (III) D41(111 Al (III) El (III) Al (III) A2k (III) Al (III) Al (III) C1 (I) El (III) Al (III) D 2 (III) Bl (I) C^I) Al (III) D 2 (III) El (III) D 4 l(IIi; A2 (III) A2 (III) Bl (I)

Index 2174a Ax|Nx|| 2175a ++ Ax|x(Ax)=N 2176a xxA=(N)x 2177a l|w[A(n)x 2178a + + (Ax)|Ax|=N 2179a ||a|xxAx 2180a A=xNx 2181a ||w|w|A=Bx 2182a A=Nx|(Ax) 2183a a|Ax=N 2184a ||w|wx|Ax|N 2185a llw wx x(Ax)=Nx 2186a Ax|(Bx)x 2187a ||Ax|Bx 2188a + + (Ax)n|A=N 2189a A=N[x}x|N 2190a ||w|w|Ax|N 2191a (Ax)=N|(N)n 2192a ++Ax|x(Ax)x 2193a ++ A=N[x}|Ax 2194a ||w|w|xAnx 2195a Hwlw[xAx 2196a ++ A|x(Ax)=N 2197a xxA=(N)x 2198a + + A | A x = N 2199a Ax|Nx 2200a l[a|wlxAx 2201a Ax=Ax 2202a ||w|Anx 2203a xxA=Nx 2204a ||w|wx|xAx 2205a ++ Ax|A=(N)x 2206a ++ ||Ax|xAx 2207a ||wx|Anx 2208a xA|xN 2209a Ax|Nx 2210a + + A Ax=N 2211a Ax|Nx 2212a ||w|w|xA{x]|Ax 2213a + + A = N | A x 2214a + + A x | A = N 2215a + + A x A - N 2216a Ax|Ax 2217a + + A|Ax|=N 2218a llw w|w|Anx

Al (III) A2b (III) C2 (II) C2 (la) D 4 1 (III) A3 (Ib) Al (III) C1 (la) El (III) Bl(Ib) Bl(Ia) C1 (la) Al (III) Al (III) A2b (III) El (III) Bl(Ia) A2k (III) A2 (III) A2a (III) C1 (la) A3 (la) A2b (HI) C2 (II) D 4 1(III) Al (III) A3 (la) Al (III) C 1 (la) C1 (II) A3 (la) D* 2 (III) A2 (III) C1 (la) Bl (II) Al (III) D 4 1 (III) Al (III) C1 (la) A2a (III) A2b (III) Al (HI) Al (III) D 4 1 (III) C1 (la)

2174b A|=N|(N)x 2175b ||wx|Ax|N 2176b ooA|xNx 2177b A|Nnx 2178b Ax|Nx 2179b ||wx|Ax|N 2180b ||w|w|A|(N)x 2181b ooAxBx 2182b ||w|w|A|Nx|| 2183b ||A|w|Nx 2184b ooAx|xNx 2185b ooAx|Nx 2186b xA|Bx 2187b ||w w|A|Bx 2188b ||A-Nx|N 2189b °Ax|xNx 2190b ooA|x(Nx)x 2191b Ax|Nx 2192b ||w|xAx|N 2193b xAx|N 2194b ooA|xNx 2195b (Ax)|Nnx 2196b ||w|w|A|(N)x 2197b ooA|xNx 2198b oo ||Ax|Nx 2199b ||w|w|Ax|N 2200b (Ax)x|Nx 2201b ||wx|(Ax)n|N 2202b Ax=Nx 2203b x(Ax)|Nx 2204b x(Ax)=Nx 2205b (Ax)=Nnx 2206b (Ax)|(Nx)nx 2207b Ax|Nx 2208b oo||w|xA|(N)x 2209b ||w|w|A|(N)n 2210b ||w|wAxN 2211b (Ax)|N(n)x 2212b ooA|x(Nx)x 2213b oo ||A|Nx|-N 2214b ||w w|Ax|N 2215b ||w w|A|xN 2216b oo ||A|Nx|-N 2217b ||w|w|w|Ax|xN 2218b x(Ax)|Nx

286

A2k (III) B1(I) Al (III) D 1 (III) Al (III) Bl (I) C 2 (I) Al (III) Cl(l) Al (Ic) A2 (III) Al (III) C1 (II) Cl(l) El (III) A2 (III) Al (III) Al (III) Bl(I) Bl (II) Al (III) D 1 (III) C 2 (I) Al (III) Al (III) Bl (I) Al (III) Bl (I) Al (III) C1 (II) C1 (II) D 1 (III) D 1 (III) Al (III) C 2 (I) C 2 (I) Bl (I) D 2 (III) Al (III) D 4 1(III) Bl(I) Bl(I) D 4 1 (III) B2(I) C1 (II)

Alliteration, scansion and metrical clause structure 2219a 2220a 2221a 2222a 2223a 2224a 2225a 2226a

Ax|Nx ++ A=N|Ax ||wx|xxAx Ax|Nx ||w|xA=Bx (Ax)x|Nx Ax|Nx A|A=(N)x

2227a defective

2228a ||wx|A=(N)x

Al (III) A2a (III) A3 (la) Al (III) C1 (la) Al (III) Al (III) D2 (III)

2219b 2220b 2221b 2222b 2223b 2224b 2225b

2256a 2257a 2258a

2259a 2260a 2261a 2262a

xxA=Nx ||w|w|xA=(N)x Ax=Nx A=Bnx Ax|Nx Ax|Nx Ax|(Nx)x A|(Ax)=Nx|| ||w|w|Ax|N Ax|Nx axx|xAx ++ Ax|xAx ||w|xAx|N Ax|Ax Ax|Nx ||a|w|w|Ax Ax|Ax ++ Ax|xAx ++ A=(N)x|Ax Ax|Nx ||xax|(Ax)=N Hwx[A(n)x ++ A=N|Ax °||w|xAx|A ++ Ax|xAx |w|w|(Ax)=Nx |w|wx|x(Ax)=N xxAx|xA ax|xxAx xxA=(B)x ++ (Ax)x|xAx

2263a (Ax)|A=Nx

B1(I) Bl (I) El (III) B2 (I) D 1 (III) El (III) Bl (I)

2226b defective

2227b (Ax)=Nx|N

C2 (la)

El (III)

2228b defective 2229b defective

2229a defective 2230a defective 2231a defective

2232a 2233a 2234a 2235a 2236a 2237a 2238a 2239a 2240a 2241a 2242a 2243a 2244a 2245a 2246a 2247a 2248a 2249a 2250a 2251a 2252a 2253a 2254a 2255a

||w|xA|xN ||w|w|xAx|N oo A=N|xN ||w|w|w|Ax|xN OO A|B-Nx (Ax)=Nx|N ||w|xAx|N

2230b ||w|wx|xA|xN 2231b oo||w|w|Ax|(Nx) 2232b A=xNx 2233b oo(Ax)x|N-N 2234b (Ax)x|Nx 2235b ooA|xBx 2236b ||xxw|A|xN 2237b ||w|xA|xN 2238b ||w|w|Ax|N 2239b ooAx|xNx 2240b A=xNx 2241b oo||A|N-(N)x 2242b (Ax)=Nx|N 2243b (Ax)=Nx|N 2244b A=xNx 2245b A=Nx|N 2246b ||A|=Nx|N 2247b ||w|(Ax)|xNx|| 2248b oo||w|w|A|xN 2249b oo ||A=N|xN 2250b °Ax|xNx 2251b ||wx|w|xA|xN 2252b ||w||w|A|(N)x 2253b Ax|Nx 2254b oo||(Ax)|Nx|-N 2255b Ax=Nx 2256b ||An|(N)x 2257b Ax|Nx 2258b ||w|xAx|xN 2259b (Ax)|Nnx|| 2260b ||xw|Ax|N 2261b Ax|Bx 2262b ||w|Ax|N 2263b ||w|A|(N)x

C1 (II) C2 (la) Al (III) D 1 (III) Al (III) Al (III) Al (III) D 1 (III) Bl(Ia) Al (III) A3 (Ib) A2 (III) Bl (la) Al (III) Al (III) A3 (la) Al (III) A2 (III) A2 (III) Al (III) A3 (Ib) C2 (la) A2a (III) Bl(Ia) A2 (III) C1 (la) A3 (la) B2 (II) A3 (Ib) C2 (II) A2 (III) D 1 (III)

287

Bl (I) Bl (I) Al (III) Al (III) Al (III) Al (III) Bl (I) Bl (I) Bl (I) A2 (III) Al (III) D2(III) El (III) El (III) Al (III) El (III) El (III) C* (I) Bl (I) El (III) A2 (III) Bl (I) C2 (I) Al (III) D41 (III) Al (III) A2k (III) Al (III) B2 (I) D 1 (III)** Bl (I) Al (III) Bl (I) C2 (I)

Index 2264a xA|Ax 2265a + + A=(N)x|Ax 2266a ax A = N x 2267a ||w|Ax=B 2268a + + A|xxAx 2269a °(Ax)|xNx 2270a a|xAx 2271a A|A=(N)x 2272a ||w|w|Anx 2273a (Ax)|A=(N)x 2274a + + Ax|xAx 2275a Ax|xNx 2276a + + A|xAx 2277a ax|Ax|N 2278a ||w|xA=(N)x 2279a a|xAx 2280a Ax=Nx 2281a + + A x A x 2282a Ax|Bx 2283a A-N|Nx 2284a ||xax|Ax|N 2285a A=Bx|N 2286a + + A x | A = x N 2287a ||w|xA|xA 2288a Ha w|xxAx 2289a + + A x | A = N 2290a Ax|Nx 2291a ||w|w|A=Nx 2292a + + A | x A = N 2293a + + Ax|xAx 2294a ax | xxAx 2295a ||wx|w|w|x(Ax)x 2296a + + ||A|xA=N 2297a xxAx=Nx 2298a xxxAnx 2299a (Ax)x|Nx 2300a + + A=B|Ax 2301a ||w|wx|(Ax)x|N 2302a A=xNx 2303a Ax=Nx 2304a |[w|w|xAx 2305a ||wx|xAx 2306a + + A = N | A x 2307a + + Ax|xAx 2308a AxlNx

C 1 (II) A2 (III) C 1 (Ib) Bl(Ia) A2 (III) Al (III) A 3 (Ib) D 2 (III) C 1 (la) D 2 (III) A2 (III) A2 (III)** Al (III) Bl(Ib) C2 (la) A 3 (Ib) Al (III) A l (III) Al (III) Al (III) Bl(Ib) El (III) D* 4 1 (III) Bl(Ia) A 3 (la) A 2b (III) Al (III) C 1 (la) A 2b (III) A2 (III) A 3 (Ib) A 3 (la) A 2b (HI) F(II) C 1 (II) Al (III) A2a (III) Bl(Ia) A l (III) Al (III) A 3 (la) A 3 (la) A2a (III) A2 (III) Al (III)

2264b 2265b 2266b 2267b 2268b 2269b 2270b 2271b 2272b 2273b 2274b 2275b 2276b 2277b 2278b 2279b 2280b 2281b 2282b 2283b 2284b 2285b 2286b 2287b 2288b 2289b 2290b 2291b 2292b 2293b 2294b 2295b 2296b 2297b 2298b 2299b 2300b 2301b 2302b 2303b 2304b 2305b 2306b 2307b 2308b

288

°||w|xAx|N ||(Ax)=N|(N)x oo A|xNx Ax|Bx A=Nx|N ||w|w|Ax|N A=Nx|N (Ax)x|Nx Ax|Nx ||Ax|Nx ||wx|A=(N)n ||w|xAx|N w|w|Ax|N ||xw|w|Ax|xN A|-N|Nx A=Nx|N ||w|w|wx|A|xN | A=Nx|N ||(Ax)=Bx|N ||w|w|A|Nx oo ||Ax|xNx ||A|B(n)x Ax|Nx oo ||A|.w|xNx oo ||A=N|xN ||w|xA|xN (Ax)|=Nx|N oo Ax|xNx ||w|w|Anx oo ||A=N|Nx ||wx|(Ax)|Nx oo Ax|xNx oo A|N|-xxA ||w|w|AxN ||wx|Ax|xN ||wx|xA|xN ||w|w|Ax|xB oo Ax|xNx oo ||A=N|xN ||ww|Ax|N Ax|Nx oo Ax xNx ||w|w|A|(N)x ||w|xAx|N 1 wxAxN

Bl(I) A2k (III) Al (HI) Al (III) El (III) Bl(I) El (III) Al (HI) Al (III) Al (III) C2(I) Bl(I) Bl(I) B2(I) Al (III) El (III) Bid) El (III) El (III) C^I) A2 (III) D2 (III) Al (III) A2 (Ic) El (III) Bid) El (III) A2 (III) C^I) A2a (III) O{1) A2 (III) D42 (III)** B1(I) B2(I) Bid) B2(I) A2 (III) El (III) Bl(I) Al (III) A2 (III) C2(I) Bl(I) Bl(I)

Alliteration, scansion and metrical clause structure 2309a ++Ax|xAx 2310a ++Ax|xAx 2311a xxxA=(N)x 2312a ||w|xA|xA 2313a + + ||A|=(N)x|Ax 2314a ++Ax|xAx 2315a A|A=(N)x 2316a °||w|xAx|A 2317a (Ax)=Bx|A 2318a ||w|xA=(N)x 2319a ++(Ax)x|xAx 2320a ++ A=(N)x|Ax 2321a ||wx|A=(N)x ++ 2322a Ax|xAx ++ 2323a Ax|xAx 2324a ||w|w|Anx 2325a axxAx 2326a Ax|Nx 2327a + + A=N|Ax 2328a ++ A|xAx 2329a ||ax|xAx 2330a xxAx|N 2331a ++Ax|xAx 2332a ++Ax|xAx 2333a ||wx|A=(N)x 2334a + + A=N|Ax 2335a ++Ax|xAx 2336a (Ax)x|Nx 2337a ||w|w|w|xAx 2338a A=Anx 2339a + + A=N|A-N 2340a ||w|w|A=(N)x 2341a ++ A|xAx 2342a ++(Ax)n|A-N 2343a (Ax)x|Nx 2344a ||w|w|A=(N)x 2345a Hxx(Ax)x[N 2346a ||w|w|xxA=(N)x 2347a Ax|Nx 2348a °||w|w|xAx|A 2349a ++(Ax)|xAx 2350a (Ax)|Anx 2351a Ax=Ax 2352a (Ax)=Nx|A 2353a ||w|xAx|xA

2309b oo||w|x(Ax)|(Nx)-N 2310b ||w|w|Ax|N 2311b ooAx|xNx 2312b Ax|Nx 2313b ||(Ax)=Nx|N 2314b ||w|w|A|(N)x 2315b Ax|Nx 2316b ooAx|xNx 2317b °A|xBx 2318b Ax|Nx 2319b oo||A|N|-xN 2320b x(Ax)|Nx 2321b ooAx|xNx 2322b oo||Ax!x(Nx)x 2323b ||w|xA|xN 2324b ooAx|xNx 2325b ||w|xAx|N 2326b (Ax)=Nx|N 2327b ||w|xAx|N 2328b (Ax)=Nx|N 2329b ||w|w|Anx 2330b Ax|Nx 2331b oo||A|Nx|-N 2332b ||w|w|xAx|xN 2333b Ax|Nx 2334b A=N|(N)x 2335b ||w|w|A=(N)n 2336b (Ax)|N(n)x 2337b Anx|N 2338b Ax|Nx 2339b oo||Ax|w|Nx 2340b ooAx|xNx 2341b ||wx|A=(N)x 2342b ooAx|xNx 2343b ||w|xA|(N)x 2344b ooAx|Nx 2345b Ax|Nx 2346b oo(Ax)x|xNx 2347b ||w|w|w|xAx|xN 2348b xA|(N)x 2349b oo||xw|w|A|(N)x 2350b ooAx|xNx 2351b ||wx|w|Anx 2352b (Ax)|N(n)x 2353b A(n)x|Nx

A2 (III) A2 (III) C2 (II) Bl(Ia) A2 (III) A2 (III) D2 (III) Bl(Ia) El (III) C2 (la) A2 (III) A2 (III) C2 (la) A2 (III) A2 (III) C1 (la) A3 (Ib) Al (III) A2a(III) Al (III) A3 (Ib) Bl (II) A2 (III) A2 (III) C2 (la) A2a (III) A2 (III) Al (III) A3 (la) D 1 (III) A2a (III) C2 (la) Al (III) Al (III) Al (III) C2 (la) Bl(Ia) C2 (la) Al (III) Bl (la) Al (III) D 1 (III) Al (III) El (III) B2 (la)

289

A2 (III) Al (III) El (III) C2(I) Al (III) A2 (III) Al (III) Al (III) D41(III) C1 (II) A2 (III) A2 (III) B1(I) A2 (III) B1(I) El (III) B1(I) El (III) Al (III) D41 (III) B2(I) Al (III) A2k (III) C2(I) D2 (III) El (III) Al (III) A2 (Ic) A2 (III) C2(I) A2 (III) C2(I) Al (III) Al (III) A2 (III) B2(I) C2 (II) C2(I) A2 (III) C^I) D2 (III) A2 (III)

Index 2354a Ax|Nx 2355a A = x N x 2356a ||wx|Ax|(Nx) 2357a + + A = ( N ) x | A x 2358a Ax|(Nx)x 2359a' + +A> : xAx 2360a Ax|Nx 2361a ||wx|w|xAx 2362a A x = N x 2363a ||wx|A(n)x 2364a A x = N x 2365a Ax|Nx 2366a xxA=(N)x 2367a ||xxa w|(Ax)x|xN 2368a A A=(N)x 2369a llw w|A|xN 2370a + +Ax|x(Ax)=N 2371a llw w|xA=Nx 2372a Ax|Nx 2373a ||w xw|A=Nx 2374a xx(Ax)nx 2375a ||w|w|Anx 2376a ||wx|xx(Ax)-N 2377a ||wx|w|wx|xAx 2378a + +Ax|xAx 2379a ||(Ax)=Nx|A 2380a xxA|Ax 2381a ||ax[w|xAx 2382a xxAnx 2383a ||wx|w|xA=Nx 2384a Ax|Nx 2385a ||w|w|xAx 2386a Ax|Ax 2387a ||w|w|A|xN 2388a Ax|Nx 2389a ||w|xx(Ax)=N 2390a ||Ax|Nx 2391a w|xA=(N)x 2392a (Ax)x|Nx 2393a A=Nx|A 2394a xxA|Ax 2395a ++(Ax)|xAx ++ Ax A=Nx 2396a 2397a ||w|w|Ax|x(Bx) ++ Ax|xAx 2398a

2354b ||w|w|Ax|N 2355b ww(Ax)n|N 2356b Ax|Nx 2357b A=Nx|N 2358b (Ax)=Nx|N 2359b ||wx|An|N 2360b ||A=Nx|N 2361b Ax|Nx 2362b ||w|w|xAx|N 2363b Ax|Nx 2364b oo||w|w|(Ax)|xN 2365b oo||A|N|-xN 2366b Ax|Nx 2367b (Ax)|Nnx 2368b ooA|xNx 2369b ooAxNx 2370b oo||Ax|x(Nx)x 2371b Ax=Nx 2372b ||w|w|(Ax)n|N 2373b Ax|Nx 2374b ooA(n)x|Nx 2375b A-N|Nx 2376b Ax|Nx 2377b A=Nx|N 2378b ||w|w|w|Ax|N 2379b |wx|A=Nx 2380b (Ax)|N(n)x 2381b A|Nnx 2382b A=(N)nx 2383b A|N(n)x|| 2384b ||w|w|xAx|N 2385b A=Nx|N 2386b (Ax)|(Nx)nx 2387b A[x]nx|N 2388b ||wx|An|N 2389b An|Nx 2390b ||w|w|A|(N)n 2391b ooA|xNx 2392b ||Anx|N 2393b oo||Ax|xNx 2394b (Ax)|N(n)x 2395b oo||w|xA|Nx 2396b oo||(Ax)|Nx|-xN 2397b x(Ax)|Bx 2398b (Ax)|Nnx

Al (III) Al (III) Bl (la) A2 (III) Al (III) A2 (III) Al (III) A3 (la) Al (III) C2(Ia) Al (III) Al (III) C2 (II) B2 (la) D2 (III) Bl(Ia) A2b (III) C1 (la) Al (III) C1 (la) C1 (II) C1 (la) A3 (la) A3 (la) A2 (III) El (III) C1 (II) A3 (la) C1 (II) C1 (la) Al (III) A3 (la) Al (III) Bl(Ia) Al (III) A3 (la) Al (III) C2 (la) Al (III) El (III) C1 (II) Al (III) D*1 (III) B2 (la) A2 (III)

290

Bl (I) Bl (I) Al (III) El (III) El (III) Bl (I) El (III) Al (III) Bl (I) Al (III) Bl(I) D41 (III) Al (III) D 1 (HI) Al (III) Al (III) A2 (III) Al (III) Bl(I) Al (III) A2 (III) Al (III) Al (III) El (III) Bl (I) C ! (I) D2 (HI) D 1 (III) D3 (III) D2 (III) Bid) El (III) D 1 (III) El (III) Bid) Al (III) C2(I) Al (III) El (III) A2 (III) D2 (III) C1 (I) D42 (III) C1 (II) D 1 (III)

Alliteration, scansion and metrical clause structure 2399a Ax=Nx 2400a ||w|w|xxAx 2401a |xw|w|Ax|N 2402a Ax|Nx 2403a ||wx|w|xAx 2404a' ++ (Ax)=N|Ax 2405a ++ A[x}=N|Ax 2406a ||w|w|xxAx 2407a ||w|xA-(N)x 2408a A|(Ax)=Nx 2409a A|A(n)x 2410a ||xw|w|w|A=(N)x 2411a + +A|xxAx 2412a A=xNx 2413a + +Ax|xAx 2414a (Ax)|A=(N)x 2415a + +A|xxAx 2416a xxAnx 2417a ||xw|w|xAx 2418a wx|Ax|xN 2419a ++ A=(N)x|Ax 2420a ++ Ax|xA=N 2421a ||w|xx(Ax)x 2422a ax|Ax|N 2423a + +A|xAx 2424a A|(Nx)nx 2425a ||An|(Nx)(n)x 2426a ||wx|w|x(Ax)x 2427a An=Nx 2428a ||w|w|(Ax)=Nx 2429a ++ A=(N)x|Ax 2430a ++ ||A|w|xAx 2431a ||w|w|A|xA{x} 2432a ||w|w|w|xAx 2433a + +A|xAx 2434a + +(Ax)n|xAn 2435a ||w|xAnx 2436a Ax|Nx 2437a ||wx|wx|An 2438a xA=(N)x 2439a ||ax|Anx 2440a Ax|Nnx 2441a ||w|w|A-N|xA 2442a ++ Ax|(Ax)=Nx 2443a + +(Ax)n|A=(N)x

2399b ||w|xxAx|N 2400b x(Ax)|Nx 2401b ooAx|xNx 2402b (Ax)|N(n)x 2403b ||wx|xA|xN 2404b w|xAx|N 2405b °xxAx|N 2406b Anx|N 2407b ooA|xNx 2408b ||wx|A|(N)x 2409b ||w|xxAx|N 2410b Ax|Nx 2411b A=Nx|N 2412b ||w w|Ax|N 2413b ||A|nNx 2414b A=Nx|N 2415b ||w|w|Ax|N 2416b (Ax)x|N[x]x 2417b A=N|(N)n 2418b A=xNx 2419b ||w|w|Ax|(Nx) 2420b oo||A|nx(Nx)|-N 2421b Ax|Nx 2422b oo||Ax|xNx 2423b ||w w|Ax|N 2424b ooAx|xNx 2425b A|Nnx 2426b oo A=Nx|xN 2427b ||w|xA|xN 2428b llw w|Ax|N[x] 2429b xxx(Ax)|xN 2430b Ax|(N)n 2431b oo||Ax|xNx 2432b Ax|Nnx 2433b ||wx|xAx|N 2434b xx(Ax)n|N 2435b A=xNxnx 2436b ooA[x}=N|{N} 2437b xA=(N)x 2438b ooAx|xNx 2439b ||w|xA|xN 2440b A(n)x|Nx 2441b (Ax)x|xNx 2442b ||wx|Ax|xN 2443b AxlNx

Al (III) A3 (la) Bl(Ia) Al (III) A3 (la) A2a (III) A2a (III) A3 (la) C2 (la) D 1 (III) D2 (III) C2 (la) A2 (III) Al (III) A2 (III) D2 (III) A2 (III) C1 (II) A3 (la) B2 (la) A2 (III) A2b (III) A3 (la) Bl(Ib) Al (III) D 1 (III) D*2 (III) A3 (la) Al (III) C1 (la) A2 (III) A2 (Ic) Bl (la) A3 (la) Al (III) A2(III) C1 (la) Al (III) A3 (la) C2 (II) C1 (Ib) D^dll) B2 (la) D^dll) D*2 (III)

291

B1(I) C1 (II) A2 (III) D2 (III) Bl(I) Bl(I) B 1 (II) El (III) Al (III) C2(I) Bid) Al (III) El (HI) Bid) Al (III) El (III) Bl(I) Al (III) A2k (III) Al (III) Bid) F (III) Al (III) A2 (III) Bl(I) A2 (III) D 1 (III) E2 (III) Bl(I) Bl(I) Bl (II) F (III)** A2 (III) D*1 (III) B1(I) B 1 (II) F (III) A2a (HI) C2 (II) A2 (III) Bl(I) A2 (III) A2 (III) B2(I) Al (III)

Index 2444a ||w w|Ax-N 2445a xxAnx 2446a ++ A|xAx 2447a Anx A 2448a + +Ax|xAx 2449a + + A|xA=B 2450a llwx w|xAx 2451a ++ (Ax)x|Ax=N 2452a xxAnx 2453a Ax=Nx 2454a xAx|N 2455a ||xa|A=(N)x 2456a ++A=(N)x|Ax 2457a + +Ax|x(Ax)x 2458a ++(Ax)|xAx 2459a ++(Ax)|xAx 2460a ||xwx wx xAx 2461a + +A|xxAx 2462a + + Ax|xA=(N)x 2463a xx(Ax)nx 2464a Anx|A 2465a xxA=(B)x 2466a ||w|xw|w xx(Ax)=N 2467a Ax|Nx 2468a ||w|w|xxxAx 2469a + + A=N|xA 2470a U(Ax)x|Nx 2471a + + A|xA=(N)x 2472a ||w|w|A|x(Ax) 2473a xxA|(A)x 2474a ++ (Ax)=N|Ax 2475a ||wx|w|Ax(n)x 2476a (Ax)|A=(N)x 2477a xx(Ax)|Ax 2478a ++ (Ax)x|An=N 2479a l|w> ^=(B)x 2480a + +Ax|x(Ax)x 2481a ||w|w|Ax|N 2482a Ax|Nx 2483a Ax|Nx 2484a l|w|.v xAx xN 2485a Ax|Nx 2486a llwi"ben 2487a + + ||A=N|xA 2488a a|(Ax)=N

2444b (Ax)x|Nx 2445b ||w|x(Ax)|Nx 2446b \\xv v w|A|(N)x 2447b ||wx|x(Ax)|Nx 2448b ||w|w w Ax|xN 2449b ooA(n)x|xBx 2450b °Ax|xNx 2451b oo||Ax|xNx 2452b ooAx|xNx 2453b ||wx|xA|(N)x 2454b ooAx|xNx 2455b xx(Ax)|Nx 2456b Ax|Nx 2457b ||An|(N)x 2458b ||w w|Ax|N 2459b Hw>cw|A|Nx 2460b ||A=N|(N)x 2461b oo||wx w|A|xN 2462b ||w|Ax|N 2463b Ax|Nx 2464b oo||Ax|xNx 2465b ooAx|xBx 2466b oo(Ax)x|xNx 2467b ||w|w|A|xN 2468b ||w|w|xA|xN|| 2469b oo ||(Ax)|=N|xN 2470b ||w w|Ax|N 2471b ||w w|xAx|xN 2472b °Ax xNx 2473b ooA|xNx 2474b ||wx|Ax|N 2475 b (Ax)x|Nx 2476b oo||Ax|xNx 2477b | w xAx=N 2478b ooA|x(Nx)x 2479b ooAx|xBx 2480b ||w w|xAx|N 2481b ooAx|xNx 2482b ||Anx|N 2483b OOA|N-Nx 2484b A|Nnx 2485 b x(Ax)|Nx 2486b (Ax)x|Nx 2487b ||(Ax)x|Nn 2488b oo||A|xNx

Bl(Ia) C1 (II) Al (III) El (III) A2 (III) A2b (III) A3 (la) D* 4 1(III) C1 (II) Al (III) B 1 (II) C2 (Ib) A2 (III) A2 (III) Al (III) Al (III) A3 (la) A2 (III) D # 2 (III) C1 (II) El (HI) C2 (II) A3 (la) Al (III) A3 (la) El (III) Al (III) D* 2 (HI) Bl (la) C2 (II) A 2a (III)

F (la)** D 2 (III) C1 (II) D*41 (III) C2 (la) A2 (III) Bl (la) Al (III) Al (III) B2 (la) Al (III) Bl(Ia) El (III) F(Ib)

292

Al (III) C1 (I) C 2 (I) Cl(l) B2(I) A2 (III) A2 (III) A2 (III) A2 (III) C 2 (I) A2 (III) C1 (II) Al (III) A2k (III) Bid) C^DOd) A 2k (III)

Bid) Bid) Al (III) A2 (III) A2 (III) A2 (III) Bl(I) Bl(I) El (III) Bl(I) B2(I) A2 (III) Al (HI) Bl(I) Al (III) A2 (III) Bl(I) Al (III) A2 (III) Bid) A2 (III) El (III) D1 (III) D1 (III) C1 (II) Al (III) Al (III) Al (III)

Alliteration, scansion and metrical clause structure 2489a 3Ax|xNx 2490a ||w|w|xAx 2491a xAx 2492a Ax|Nx 2493a + + A|Ax=N 2494a ||w|w|xAx 2495 a xxxA=Nx 2496a ++ Ax|A=(N)x 2497a ||wx|w|w|xAx 2498a ++Ax|xAx 2499a Ax|Nx 2500a ||w|w|A|xN 2501a ||wx|w|x(Ax)x 2502a xA=(N)x 2503a ||wx|w|xAx 2504a A=Nnx 2505a ||w|xAx|xA 2506a ++(Ax)n|xAx 2507a ||w|w|Ax=N 2508a + + A=N|xA 2509a + + A|xA|=N 2510a ||An|(Nx)(n)x 2511a Ax|Nx 2512a ++Ax|x(Ax)x 2513a + + A|Ax|=N 2514a ||Ax|Nx 2515a xA=(B)x 2516a HxAxjN 2517a (Ax)|A=(N)n 2518a ++Ax|xAx 2519a ++Ax|xAx 2520a xxA=Nx 2521a ++Ax|xAx 2522a ||w|w|w|(Ax)=Nx 2523a ++(Ax)x|xAx 2524a ++ A|xAx 2525a xxa|Ax|N 2526a ax|xAx 2527a ++ (Ax)|Ax|=xN 2528a ||w|w|xxxA=(N)x 2529a ||xax|w|xAx 2530a ++Ax|xAx 2531a xxA=Nx 2532a Ax|Nx 2533a ||w|xA|Ax

A2 (III) A3 (la) A3 (Ib) Al (III) D41 (III) A3 (la) C1 (II) D*2 (III) A3 (la) A2 (III) Al (III) Bl (la) A3 (la) C2 (II) A3 (la) D 1 (III) B2 (la) A2 (III) Bl (la) El (III) A2b (III) D*2 (III) Al (III) A2 (III) D41 (III) Al (III) C2 (II) Bl(Ia) D2 (III) A2 (III) A2 (III) C1 (II) A2 (III) C1 (la) A2 (III) Al (III) Bl(Ib) A3(Ib) D42 (III) C2 (la) A3 (la) A2 (III) C1 (II) Al (III) C1 (la)

2489b oo ||A=N|xxN 2490b oo||w|w|A|Nx|| 2491b ||w|w|(Ax)x|N|| 2492b ||w|w|A|xN 2493b ||w|w|AxN 2494b xxxA=(N)x 2495b Ax|Nx 2496b oo||Ax|xNx 2497b x(Ax)|Nx 2498b ||w|w|xAx|N 2499b ||wx|xA|(N)x 2500b ooA|xNx 2501b Anx|N 2502b Ax|Nx 2503b A=(N)nx 2504b Ax|Nx 2505b Ax|Nx 2506b ||xw|A|(N)x 2507b Ax|Nx 2508b ||w|w|Ax|N 2509b xA|Nx 2510b ||A=Nx|N 2511b oo||w|xAx|(Nx) 2512b oo||A|w|Nx 2513b Ax|Nx 2514b ||w|w|xA=(N)x 2515b ooA|xBx 2516b °(Ax)x|xNx 2517b A(n)x|Nx 2518b ||wx|w|A|(N)x 2519b oo||w|w|Ax||N 2520b Ax|Nx 2521b ||w|w|w|xAx|(Nx) 2522b Ax|Nx 2523b ||xw|w|w|A|(N)x 2524b ||wx|w|Ax|N 2525b ||w|w|Ax|N 2526b ||w|w|A|xN 2527b ||w|w|xAx|N 2528b ooA|xxNx 2529b Ax|(Nx)x 2530b ||wx|A|(N)x 2531b ooAx|xNx 2532b oo||w|w|AxN 2533b ||wx|A|Nx

293

E2 (III)

Bl(I) C 2 (II) A l (III) A 2 (III) C 1 (II) B1(I) C 2 (I) A l (III) El (III) A l (III) D 3 (III) A l (III) A l (III) C 2 (I) A l (III) B1(I) C 1 (II) El (III) B1(I) A l (Ic) A l (III) C 2 (I) A l (III) A 2 (III) A 2 (III) C 2 (I) Bl(I) A l (III) Bl(I) A l (III) C 2 (I) Bid) Bl(I) Bl(I) Bl(I) A 2 (III) A l (III) C 2 (I) A 2 (III) Bl(I)

Index 2534a ||w|w|xA=Nx 2535a + +A=(N)x|Ax 2536a + +A|xAx 2537a + +A=(N)x|Ax 2538a ||xa|w|xAx 2539a ++A|xxAx

2540a xxA=(N)x 2541a Ax|Nx 2542a ||xw|w|xAx 2543a A=Nx|A 2544a Ax=Ax 2545a ax|A=(N)x 2546a ax|xAx 2547a (Ax)=Nx|A 2548a A=Nnx 2549a ++ A|xAx 2550a ||w|w|xAx 2551a (Ax)=Nx|N 2552a + + ||A=N|Ax 2553a ++ (Ax)=N|Ax 2554a + + ||(Ax)|w|xAx 2555a Ax|Nx 2556a ++Ax|xAx 2557a (Ax)|A=Nx 2558a + + A|Ax=N 2559a + + ||A|xxAx 2560a xx(Ax)=Ax 2561a ||w|w|A=(N)x 2562a Ax|xAnx 2563a A|A=(N)n 2564a + + Ax|A=N 2565a (Ax)=Nnx 2566a + + ||A=N|xA 2567a (Ax)x|Bx 2568a ++Ax|xAx 2569a ||xw|w|Anx 2570a xx(Ax)|Ax 2571a ++Ax|xAx 2572a Ax|Nx 2573a ||w|w|xAx 2574a Ax|Nx 2575a ++ A|xAx 2576a Ax|Nx 2577a Ax=Nx 2578a ++ A|xAx

C1 (la) A2 (III) Al (III) A2 (III) A3 (la) A2 (III) C2 (II) Al (III) A3 (la) El (III) Al (III) C2(Ib) A3 (Ib) El (III) D 1 (III) Al (III) A3 (la) El (III) A2a (III) A2a (III) A2 (Ic) Al (III) A2 (III) D 1 (III) D4 (III) A2 (III) C1 (II) C2 (la) D*1 (III) D2 (III) A2b (III) D 1 (III) El (III) Al (III) A2 (III) C1 (la) C1 (II) A2 (III) Al (III) A3 (la) Al (III) Al (III) Al(III) Al (III) Al (III)

2534b (Ax)x|Nx 2535b ||w|xAx|N 2536b ||wx|A|(N)x 2537b A|Nnx 2538b OOA|N-Nx 2539b ||(Ax)=Nx|N 2540b oo||Ax|x(Nx)x 2541b ||xw|w|Ax|N 2542b ||w|w|Ax|(Nx) 2543b ooAx|xNx 2544b oo||wx|(Ax)|Nx|| 2545b oo||A|N|-(N)x 2546b °||w|xxAx|N 2547b ||xwx|Ax|N 2548b ooA(n)xNx 2549b x(Ax)|Nx 2550b ||w|w|xAx|N|| 2551b ooA|N|-(N)x 2552b oo||A|N|-xN 2553b xxAx|N 2554b oo ||A=N|xN 2555b ||w|w|AxN 2556b oo||A|Nx|-N 2557b ooA|xNx 2558b ||Ax|(Nx)x 2559b oo A=N|xN 2560b Ax|Nx 2561b ooAx|xNx 2562b oo||A|N|-xN 2563b (Ax)x|Nx 2564b ||A-Nx|N 2565b ooAx|xNx 2566b xAx|N 2567b ||w|xA|xB 2568b ||w|xAx|N 2569b x(Ax)|Nx 2570b oo||A|N|-xN 2571b Ax|Nx 2572b ||wx|x(Ax)|Nx 2573b Ax|N{x}x 2574b ||w|w|A|xxN 2575b oo||A|N|-xN 2576b ||(Ax)=Nx|N 2577b ||w|xA|xN 2578b oo||A|nNx

294

Al (III) Bl (I) C2 (I) D 1 (III) D1 (III) El (III) A2 (III) B1(I) Bl (I) A2 (III) C 1 ©** D2 (III) Bl (I) Bl (I) A2 (III) C1 (II) Bl (I) D2 (III) D41 (III) Bl (II) El (III) Bl (I) D41 (III) Al (III) Al (III) El (III) Al (III) A2(III) D41 (III) Al (III) El (III) A2 (III) Bl (II) Bl (I) Bl (I) C1 (II) D41 (III) Al (III) C1 (I) Al (III) B2 (I)' D41 (III) El (III) Bl (I) Al (III)

Alliteration, scansion and metrical clause structure 2579a ||wx|xA=(N)n 2580a ++(Ax)x|xAx 2581a xx(Ax)=Nx 2582a ||a|A=Nx 2583a Ax=Nx 2584a ++ A=(N)x|Ax 2585a ++(Ax)|xAx 2586a ++ Ax|A-N 2587a ||w|xAx 2588a A=B|(N)x 2589a ||wx|xxAx 2590a Ax|Nx 2591a xax|A=(N)x 2592a ||w|xA=Nx 2593a ||ax|wx|A=B 2594a Ax|Nx 2595a ++Ax|xAx 2596a ||wx|w|xAx 2597a (Ax)nx|N 2598a Ax=Nx 2599a ||Ax|Nx 2600a ++(Ax)|xAx 2601a ++ A|xAx 2602a ||Ax|w|Nx 2603a ++ A-N|A=(N)x 2604a A|N(n)x 2605a xx(Ax)=Nx 2606a ||xwx|w|xAx 2607a ++A=(N)x|(Ax)x 2608a °A=Nx|xN 2609a 2610a Ax|Nx 2611a ||w|w|xAx 2612a (Ax)|N(n)x 2613a ++ Ax|(Ax)=Nx 2614a Ax|Nx 2615a B=Nx|A 2616a ++ A=N|(Ax)x 2617a x(Ax)nx 2618a ++ A=(N)x|A-N 2619a ||w|w|w|xAx|A 2620a ||w|w|Ax|xN 2621a ++ A|xAx 2622a ++ A=(N)x|Ax 2623a ||a|w|w|xAx

C2 (la) A2 (III) C1 (II) C1 (Ib) Al (III) A2 (III) Al (III) Al (III) A3 (la) A2k (III) A3 (la) Al (III) C2 (Ib) C1 (la) A3 (la) Al (III) A2 (III) A3 (la) El (III) Al (III) Al (III) Al (III) Al (III) A2 (Ic) D*2 (III) D2 (III) C1 (II) A3 (la) A2 (III) E2 (III) A3 (la) Al (III) A3 (la) D2 (III) D*1 (III) Al (III) F (III)** A2a (III) C1 (II) A2 (III) Bl (la) B2 (la) Al (III) A2 (III) A3 (la)

2579b Ax|Nx 2580b ||w|w|Ax|N 2581b xA[x]|Nx 2582b ||Ax|Nx 2583b oo||A=(Nx)x|xN 2584b oo ||A=N|xN 2585b oo||w|w|A|Nx 2586b ||xw|w|Ax|N 2587b (Ax)|Nnx 2588b x(Ax)|Bx 2589b A]N(n)x 2590b ||w|w|A-N|N 2591b ||w|w|A|xN 2592b w|A|xNx 2593b oo||(Ax)|Nx|-B 2594b ||(Ax)|N(n)x 2595b ||w|w|w|Ax|N 2596b A=xNx 2597b ooAx|xNx 2598b ||w|w|xA|(N)x 2599b ||wx|xAx|N 2600b oo||A|Nx|-xN 2601b ||w|w|A|Nx 2602b Anx|(Nx) 2603b A|Nnx 2604b ||xw|xA=Nx 2605b A|N(n)x 2606b ||w|w|w|A|xN|| 2607b A=Nnx 2608b ||w|x(Ax)|Nx 2609b oo||A|N|-xN 2610b oo||(Ax)|=N|xN 2611b Anx|N 2612b ||w|xAx|N 2613b An|(N)x 2614b ||w|xAx|xN 2615b Ax|Bx 2616b ||w|w|(Ax)x|xN 2617b A=xNx 2618b ||w|xxxAx|N 2619b xA(n)x 2620b oo(Ax)|N-Nx 2621b ||w|w|x(Ax)|Nx 2622b ||w|xA-(N)x 2623b A=xNx

295

Al (III) Bl(I) C1 (II) Al (III) E2 (III) El (III) C1 (I)** B1(I) D 1 (III) C1 (II) D2 (III) Bl(I) Bl(I) F(D* D41(III) D2 (III) Bl(I) Al (III) A2 (III) C2(I) Bid) D42 (III)

CM) El (III) D 1 (HI) C1© D2 (III) Bid) D 1 (HI) El (III) El (III) Bl(I) A2k (III) B2(I) Al (III)** B2(I) Al (HI) Bl(I) C2 (II) D 1 (III) C1© C2(I) Al (HI)

Index 2624a + + A-N|A=N 2625a + + A|xA=N 2626a Ax|Nx 2627a xxA=Nx 2628a ||xxa|w|xA=(N)x 2629a xa|xAx 2630a llwx|w[xAx 2631a ||An|(Nx)(n)x 2632a ax | xAx 2633a ||w|xA|xA 2634a Hwx|w|xAx 2635a xA=(N)x 2636a | |w|w|w|xA=x(Nx) 2637a ||w|w|A=(N)x 2638a ++ Ax|xA|=N 2639a xxxA-(N)x 2640a ||xax|wx|Ax 2641a ||w|w|wx|A=Nn 2642a (Ax)|A=(N)n 2643a xAx=N 2644a xxAnx 2645a ||xw|w|Ax|A 2646a + + Ax|A=Nx 2647a ||w|xxA=Nx 2648a + + Ax|A=Nx 2649a ax|A=(N)x 2650a A=(Nx)x|A 2651a ||w|w|w|xxAx 2652a xxxA=(A)x 2653a ||xwx|w|xAx 2654a a|xAx 2655a + +Ax|xAx 2656a Ax|Nx 2657a ||w|wx|A=xN 2658a Ax|(Nx)x 2659a ||xax|xAx 2660a ++ Ax|x(Ax)=N 2661a ||a|w|xxxA=N 2662a + + A|xAx 2663a ||Ax|Nx 2664a ||w|w|x(Ax)=Nx 2665a Hw|wlxxAx 2666a + + A|xAx 2667a + + (Ax)n|A=Nx 2668a A|N(n)x

A 2 b (III) A 2 b (III) Al (III) C 1 (II) C 2 (la) A 3 (Ib) A 3 (la) D* 2 (III) A 3 (Ib) Bl(Ia) A 3 (la) C 2 (II) Bl(Ia) C 2 (la) A 2 b (III) C 2 (II) A 3 (la) C 1 (la) D 2 (III) Bl (II) C 1 (II) Bl(Ia) D#1(III) C 1 (la) D* 1 (III) C 2 (Ib) El (III) A 3 (la) C 2 (II) A 3 (la) A 3 (Ib) A 2 (III) Al (III) Bl(Ia) Al (III) A 3 (Ib) A 2 b (III) A 3 (la) Al (III) Al (III) C 1 (la) A 3 (la) A l (III) D* 1 (III) D 2 (III)

2624b ||w|w|xAx|xN 2625b ||w|w|Ax|N 2626b ||w|w|Ax|N 2627b Ax|Nx 2628b ||w|xAx|N 2629b ||w|xA|xN 2630b xA|Nx 2631b ||A=Nx|(Nx) 2632b ||w|w|(Ax)|Nx 2633b ||w|w|(Ax)|Nx 2634b xxA-Nx 2635b ||w|w|xAx|N 2636b Ax|Nx 2637b ooA|xNx|| 2638b | |w|w|wx|xAx|xN 2639b Ax|Nx 2640b ||w|w|xAx|N 2641b Ax|Nx 2642b ||w|w|A-N|N 2643b ooAx|xNx 2644b Ax|Nx 2645b ooAx|x(Nx)x 2646b ||w|w|xA|(N)x 2647b oo(Ax)x|xNx 2648b ||wx|Ax|N 2649b Hwx|A|{N} 2650b oo ||A|=N|xN 2651b ||w|xxA=(N)x 2652b A|N(n)x 2653b ||w|w|Ax|(Nx) 2654b ||wx|w|Ax|N 2655b l|A|N(n)x 2656b oo||w|A|(N)x 2657b ||w|w|Ax|(Nx) 2658b A|N(n)x 2659b w|A|xN 2660b ooA|xNx 2661b ||A=(Nx)x|N 2662b ||A|=Nx|N 2663b ooA|-N|(N)x 2664b ooAx|xNx 2665b xx(Ax)nx 2666b ||w|w|Ax|N 2667b ooAx(Nx)x 2668b ||w|w|A-Nx

296

B2(I)

Al (III) C1 (II) El (III) C1® Cl{l) C1 (II) B1(I) Al (III) Al (III) B2(I) Al (III) Bl(I) Al (III) B1(I) A2 (III) Al (III) A2 (III) C2(I) A2 (III)

El (III) C2(I) D2 (III) D2 (III) C2(I) B1(I) D2 (III) B1(I) Al (III) El (III) El (III) A2k (III) A2 (III) C1 (II)** B1(I) Al (III)

C1®

Alliteration, scansion and metrical clause structure 2669a ||xxw Ax 2670a ++ (Ax)|An=N 2671a A=Nx|A 2672a Ax|Nx 2673a Ha|A|xN 2674a ++ Ax|A=(N)x 2675a ||w|x(Ax)|Nx 2676a + +Ax|xAx 2677a + + AxxAx 2678a + + AxxAx 2679a Ax=Nx 2680a + + Ax xAx 2681a ||ax|xAx 2682a ++ (Ax)|xA=N 2683a ||w|w Anx 2684a ax xAx 2685a llw w|Ax|x(Nx) 2686a + +Ax|xxAx 2687a + + Ax Ax = N 2688a ||w|w|A=(N)x 2689a + + Ax A=(N)x 2690a ||ax|xxxAx 2691a ++ A|x(Ax)=N 2692a (Ax)x|Ax 2693a Ax=Nx 2694a llw w|xAx|xN 2695a A-Nx A 2696a + +A|xAx 2697a ||xax|"vv x(Ax)x 2698a A(n)x|Ax 2699a ||w|w|xxA=N 2700a ++ A|xAx 2701a + + AxAx 2702a axx|Ax 2703a xa|xxAx 2704a ++ (Ax)|x(Ax)=N 2705a ||xa|Ax|N 2706a + + ||A|xAx 2707a ||w|w|wx|w|Ax 2708a A=(Nx)nx 2709a ++ A|xAx 2710a + + Ax|(Ax)=Nx 2711a ++ Ax|xAx 2712a ||w|w|xA=(N)x 2713a + + :AX )|xAx

A^ (la)** D 4 1 (III) El (III) Al (III) Bl(Ib) D*2 (III) C1 (la) A2 (III) A2 (III) A2 (III) Al (HI) A2 (III) A3 (Ib) A2b (III) C1 (la) A3 (Ib) B2 (la) A2 (HI) D*41 (III) C2 (la) D*2 (III) A3 (Ib) A2b (III) Al (III) Al (HI) B2 (la) El (III) Al (HI) A3 (la) A2 (III) A3 (la) Al (III) Al (III) A3 (Ib) A3 (Ib) A2b (III) Bl(Ib) Al (HI) A3 (la) D 1 (III) Al(III) D #1 (III) A2 (III) C2 (la) Al (III)

2669b A|=Nx|N 2670b ooAxNx 2671b Ax|N(n)x 2672b oo ||A|Nx|-N 2673b 00 Ax|xNx 2674b ooAx|xNx 2675b xxxAx|N 2676b | w|xAx|N 2677b ||xw A=(N)n 2678b (Ax)=Nx|N 2679b ||w|w|x(Ax)x|N 2680b oo ||An|xN 2681b ANnx 2682b ||w|w|(Ax)x|xN 2683b Ax|Nx 2684b ||w|xA|xN 2685b AxxNx 2686b ||wx w|xAx|N 2687b ||w|w|Ax|xN 2688b Ax|Nx 2689b ooAx|xNx 2690b Hw|w|A|xN|| 2691b oo ||A|=Nx|xxN 2692b ||w|xA(n)x|N 2693b oo ||A|Nx|-N 2694b A=(N)nx 2695b Ax|Nx 2696b ||w w|xAx|N 2697b w|xA|xN 2698b ||w w|xAx|N 2699b (Ax)|-Nx|N 2700b ||w|xA|xN 2701b wxAxN 2702b ||xw|A|(N)n 2703b oo ||A=Nx|xN 2704b ||w w|xAx|N 2705b ooA|xNx 2706b oo ||A|Nx|-N 2707b x(Ax)|Nx 2708b ||w|wx|A|(N)x 2709b ||wxAx|N 2710b Ax|Nx 2711b ||w|xA|xN 2712b ooA|xNx|| 2713b llw w|Ax|xN

297

El (III) Al (III) D*2 (III) D 4 1(III) A2 (III) A2 (III) Bl (II) Bl(I) C 2 (I) El (III) Bl (I) El (III) D 1 (III) B2(I) Al (III) Bid) A2 (III) Bl(I) B2(I) Al (III) A2 (III) Bid) E2 (III) B2(I) D41 (III) D3 (III) Al (III) Bid) Bl(I) Bl(I) El (III) Bl(I) Bl(I) C2(I) E2 (III) Bid) Al (III) D41 (III) C1 (II) C2(I) Bl(I) Al (III) Bid) Al (HI) B2(I)

Index 2714a ||w|w|xAx 2715a ++Ax|xAx 2716a ||w|w|xAx 2717a ax|xAx 2718a ||w|xA=(N)x 2719a ++ Ax|A=(N)x 2720a ||wx|w|xAx 2721a Ax|Bx 2722a (Ax)=Nx|N 2723a Ax|Nx 2724a ||An|(Nx)(n)x 2725a ++ Ax|A=Nx 2726a ||w|w|A=Bx 2727a Ax|Nx 2728a Ax=xNx 2729a ||w|w|(Ax)|Nx 2730a A=xNx 2731a xxAx=B 2732a ++Ax|xAx 2733a Ax|Nx 2734a Ax=Nnx 2735a ||w|w|A=(N)x 2736a ||(Ax)x|{N} 2737a A=xNx 2738a ||xax|(Ax)=Nx 2739a ++ Ax|xA=N 2740a A=Nx|N 2741a ||xw|w|Ax|xN 2742a ++A[x}=(N)x|Ax 2743a ++ A|xAx 2744a A|N(n)x 2745a An|Bx 2746a ||ax|Ax|N 2747a |w|w|x(Ax)x 2748a ++ A=N|x(Ax) ++ 2749a Ax|(Ax)=Nx 2750a xxAx=(Nx) 2751a ++ A|xA=(N)x 2752a ||w|w|Ax|xN 2753a xxA=(N)x 2754a ax|(Ax)=Nx 2755a ax|(Ax)=Nx 2756a ||xa|w|(Ax)=Nx 2757a ++ (Ax)=N|Ax 2758a A|(Ax)(n)x

A3 (la) A2 (III) A3 (la) A3 (Ib) C2 (la) D # 2 (III) A3 (la) Al (III) El (III) Al (III) D*2 (III) D*1 (III) C1 (la) Al (III) A2 (III) C1 (la) Al (III) Bl (II) A2 (III) Al (III) D*1 (III) C2 (la) Al (III) Al (III) C1 (Ib) A2b (III) El (III) B2 (la) A2 (III) Al (III) D2 (III) Al (III) Bl(Ib) A3 (la) El (III) D*1 (III) Bl(II) D*2 (III) B2 (la) C2 (II) C1 (Ib) C1 (Ib) C1 (la) A2a (III)

2714b (Ax)=Nx|N 2715b ||w|x(Ax)n|N 2716b A=Nnx 2717b oo ||A|xNx|=xN 2718b (Ax)x|Nx 2719b Ax|Nx 2720b (Ax)=Nnx 2721b ooA|nx(Bx)|-N 2722b oo(Ax)x|x(Nx)x 2723b ||w|xA|xN 2724b ||w|xxAx|N 2725b oo||Ax|w|Nx 2726b x(Ax)|Bx 2727b oo||w|w|A|(N)x 2728b oo||A|nx(Nx)|-N 2729b Ax|Nx 2730b ||w|w|(Ax)x|N 2731b Ax|Bx 2732b ||w|xAx|N 2733b ||w|xA=(N)n 2734b ooAx|Nx 2735b Ax|Nx 2736b ||w|xAx|N 2737b oo||w|A|(N)x 2738b ||w|w|A|(N)x 2739b ||w|xAx|N 2740b xA|Nx 2741b An|Nx 2742b oo||wx|A|(N)x 2743b ||w|w|Ax|N 2744b xxAx|N 2745b ||w|xA|(B)x 2746b ooAx|xNx 2747b ||w|w|A-(N)x 2748b ||(Ax)|N(n)x 2749b ||w|w|xA|(N)x 2750b ooA|xNx 2751b ||wx|w|Ax|N 2752b (Ax)|Nnx 2753b Ax|Nx 2754b ||A=N|(N)x 2755b xxAx|N 2756b ||w|w|xAx|N|| 2757b A[x}=Nx|(Nx) 2758b ooAx|xNx

298

El (III) B1(I) D 1 (III) D*42 (II Al (III) Al (III) D 1 (III) F (III) A2 (III)

A2 (Ic) C1 (II) C2 (I) F (III) Al (III) B1(I) Al (III) Bl(I) C2(I) Al (III) Al (III) B1(I) C2(I) C2(I) Bl(I) C1 (II) Al (III) C2(I) B1(I) Bl (II) C2(I) A2 (III) C2(I) D2 (III) C2(I) Al (III) B1(I) D 1 (III) Al (III) A2k (III) B 1 (II) B1(I) El (III) A2 (III)

Alliteration, scansion and metrical clause structure 2759a + +Ax|xAx 2760a ++ Ax|A=(N)x 2761a A=Nx|(Ax) 2762a ++Ax|x(Ax)x 2763a + +A|xAx 2764a + +Ax|xAx 2765a ++ A|xAx 2766a xxA(n)x 2767a ||wx|w|(Ax)x|xA 2768a ++ A|xxAx 2769a xax|(Ax)=Nx 2770a ||w|w|xxA=N 2771a + + ||Ax|xAx 2772a A-N|Ax 2773a ||w|w|xAx|xN 2774a + + A|Ax|=xN 2775a ||w|xA|(N)x 2776a Ax|Nx 2777a Ax Ax 2778a HA|w|Ax|| 2779a ||w|xxAx 2780a Ax|Nx 2781a ++ Ax|xAx 2782a Ax=Nx 2783a + + ||A|w|x(Ax)x 2784a ++ Ax|xAx 2785a ||wx|Ax=N 2786a xxA=(N)x 2787a Ax=Nx 2788a ||w|w|xxAx 2789a Ax|Nx 2790a + +Ax|xAx 2791a (Ax)x|Ax 2792a + + A = N | x A 2793a ++(Ax)|xAx 2794a ||w|xxAx 2795a Ax=(Nx)x 2796a Ax|Nx 2797a ||w|w|w|Ax 2798a xA=(N)x 2799a ||w|w|xAx|N 2800a + + Ax|A=(N)x 2801a Ax|Nx 2802a ||ax|(Ax)=Nx 2803a ++ Ax|xxAx

A2 (III) D*2 (III) El (III) A2 (III) Al (III) A2 (III) Al (III) C2 (II) B2 (la) A2 (III) C1 (Ib) A3 (la) A2 (III) Al (III) B2 (la) D42 (III) C2 (la) Al (III) Al (III) Al (Ic) A3 (la) Al (III) A2 (III) Al (III) A2 (Ic) A2 (III) Bl (la) C2 (II) Al (III) A3 (la) Al (III) A2 (III) Al (III) El (III) Al (III) A3 (la) Al (III) Al (III) A3 (la) C2 (II) Bl(Ia) D* 2 (III) Al (III) C1 (Ib) A2 (III)

2759b °xxAx|N 2760b ||Ax|Nx 2761b An=Nx 2762b ]|w|w|A|(N)x 2763b ||A=Nx|(Nx) 2764b oo ||A|Nx|-N 2765b oo A=Nx|x(Nx) 2766b oo||Ax||wlw|Nx 2767b OOA|N-Nx 2768b A=Nx|N 2769b ||xw|Ax|N 2770b x(Ax)|Nx 2771b ||w|xAx|N 2772b |w|wx|A|xN 2773b A|N(n)x 2774b Ax|Nx 2775b oo(Ax)|xNx 2776b oo ||A|N|-xN 2777b oo ||A|N|-xN 2778b A=Nnx 2779b A=(N)x|N 2780b ||A=(Nx)x|N 2781b (Ax)=Nnx 2782b ||w|w w|Ax|N 2783b A=Nx|N 2784b ||wx|An|N 2785b ooAx|xNx 2786b Ax|Nx 2787b ||w|w|wx|A|xN 2788b Ax|Nx 2789b Anx|N 2790b ||w|wx|A|xN 2791b ||w|w|AxN 2792b supplied 2793b ||A|N(n)x 2794b A|=Nx|N 2795b Ax|Nx 2796b ||w|w|w|A|(Nx)x 2797b ooAxNx 2798b ooA|xNx 2799b ooAx|xNx 2800b oo ||Ax|Nx 2801b ||xw|w|w|A|(N)x 2802b ooA|xNx 2803b x(Ax)|(N)x

299

Bl (II) Al (III) Al (III) C 2 (I) El (III) D 4 1(III) E2 (III) A2 (III) D 1 (III) El (III) B1(I) C1 (II) B1(I) B1(I) D 2 (III) Al (III) Al (III) D 4 1 (III) D 4 1 (III) D 1 (III) E2 (III) El (III) D 1 (III) B1(I) El (III) Bl(I) A2 (III) Al (III) B1(I) Al (III) El (III) B1(I) Bl(I) D 2 (III) El (III) Al (III) CX(I) Al (III) Al (HI) A2 (III) Al (III) C 2 (I) Al (III) C2 (II)

Index 2804a llwiv|xxAx 2805a A|A(n)x 2806a ||w|w A=Nn 2807a Anx A 2808a xxAx|x(Nx) 2809a ||wx w|xAx 2810a + + Ax|A=Nx 2811a + + Ax|A=(N)x 2812a ++ A|xAx 2813a | w \v|Ax=N 2814a A=Nnx 2815a + +AxAx 2816a ++ Ax|xAx 2817a ||w|vv x(Ax)x 2818a A=xNx 2819a + +Ax|(Ax)=Bx 2820a Ax|Ax 2821a ||w|w|xAx 2822a Ax=Nx 2823a xxAnx 2824a + +Ax|xAx 2825a ++ (Ax)-N|A=(N)x 2826a + +Ax|xAx 2827a A|A=(N)x 2828a ||w|v/ Anx 2829a + + Ax|(Ax)=Nx 2830a ||w|xA=(N)x 2831a a|xAx 2832a ||wx xxAx 2833a Ax=Nx 2834a A-N|Ax 2835a xxA=(N)x 2836a |wx w|xAx 2837a (Ax)=Nnx 2838a ||w|*i w Ax|xN 2839a ||w|w|xAx=N 2840a ||wx|A=(N)x 2841a | w|w Anx 2842a ax|xAx 2843a A=Nx|A 2844a ||wx|A-(N)x 2845a Ax|Ax 2846a ||w|xA=(N)x 2847a + + Ax|A=(N)x 2848a NwlxAxlN

A3 (la) D 2 (III) C1 (la) El (III) B2 (II) A3 (la) D* 1 (III) D*2 (III) Al (III) Bl(Ia) D 1 (III) Al (III) A2 (III) A3 (la) Al (III) D* 1 (III) Al (III) A3 (la) Al (III) C1 (II) A2 (III) D*2 (III) A2 (III) D 2 (III) C1 (la) D # 1 (III) C2 (la) A3 (Ib) A3 (la) Al (III) Al (III) C2 (II) A3 (la) D 1 (III) B2 (la) Bl(Ia) C2 (la) C1 (la) A3 (Ib) El (III) C2 (la) Al (III) C2 (la) D*2 (III) Bl (la)

2804b 2805b 2806b 2807b 2808b 2809b 2810b 2811b 2812b 2813b 2814b 2815b 2816b 2817b 2818b 2819b 2820b 2821b 2822b 2823b 2824b 2825b 2826b 2827b 2828b 2829b 2830b 2831b 2832b 2833b 2834b 2835b 2836b 2837b 2838b 2839b 2840b 2841b 2842b 2843b 2844b 2845b 2846b 2847b 2848b

300

oo AxNx x(Ax)=Nx Ax|Nx . w|w|Anx Ax|Nx A|Nnx oo ||Ax|xNx A=Nx|N 1 l wwxAxN oo AxNx |wx|A|xN x(Ax)=Nx w w|Ax N Anx|N ||w|w|A|(N)x ||w|xAx xB A=Nx|N (Ax) | nNx ||w w xAx|xN oo Ax|xNx oo ||(Ax)|Nx|-N oo Ax|xNx ||A=Nx|N oo Ax|xNx oo Ax|xNx (Ax)x|Nx Ax|Nx A=Nx|N Anx|N ||A=Nx|N ||w w Ax|xN A=xNx A|=Nx|N oo AxxNx Ax|Nx oo (Ax)x|xNx Ax|(Nx)x oo A|xNx ||Anx|N oo Ax|xNx oo Ax|xNx llw w|A|xN oo A|xNx oo Ax|xNx (Ax)x|Nx

Al (III) C 1 (II) Al (III) C1 (I) Al (III) D 1 (III) A2 (III) El (III) Bl (I) Al (III) Bl (I) C1 (II) Bl (I) El (III) C 2 (I) B2(I) El (III) Al (III) B2(I) A2 (III) D41(111 A2 (III) El (III) A2 (III) A2 (III) Al (III) Al (III) El (III) El (III) El (III) B2(I) Al (III) El (III) A2 (III) Al (III) A2 (III) Al (III) Al (III) El (III) A2 (III) A2 (III) Bl (I) Al (III) A2 (III) Al (III)

Alliteration, scansion and metrical clause structure 2849a xxxA=Nx 2850a ||w|w|(Ax)nx 2851a A=xNx 2852a ||ax|xAn 2853a Ax=Nx 2854a ||ax|wx|Ax 2855a ||xwx|w|xAx 2856a xxA=Nx 2857a ||w|xAnx 2858a ||wx|A|(N)x 2859a °(Ax)x|xNx 2860a ||w|w|xxAx 2861a A=xNx 2862a ||An|(Nx)(n)x 2863a + + A|Ax=N 2864a ||w|w|w|Ax 2865a ||w|xA=Nx 2866a Ax=Nx 2867a ||wx|w|x(Ax)=Nx 2868a A=Nnx 2869a ++Ax|xAx 2870a xxA|xxN 2871a ||w|w|Anx 2872a fAx|xAx A=(N)n 2873a 2874a Ax|Nx 2875a (Ax)x|Bn 2876a ++Ax|xAx 2877a ||w|w|A=(N)x 2878a xax|xAx 2879a xxAxA 2880a ||ax|w|xAx 2881a A=xNx 2882a a|xxAx 2883a a|xxAx 2884a ||w|w|A=(N)x 2885a xAx=N 2886a ++(Ax)|xAx 2887a xxA=Nx 2888a Ax|Nx 2889a ++Ax|xAx 2890a A=Nx|A 2891a °Ax|xNx 2892a ||a|w|x(Ax)=N 2893a a|xxA=N

C1 (II) C1 (la) Al (III) A3 (Ib) Al (III) A3 (la) A3 (la) C1 (II) C1 (la) C2 (la) A2 (III) A3 (la) Al (III) D*2 (III) D41 (III) A3 (la) C1 (la) Al (III) C1 (la) D 1 (III) A2 (III) B2 (II) C1 (la) A2 (III) C2 (la) Al (III) Al (III) A2 (III) C2 (la) A3 (Ib) Bl (II) A3 (la) Al (III) A3 (Ib) A3 (Ib) C2 (la) B 1 (II) Al (III) C1 (II) Al (III) A2 (III) El (III) A2 (III) A3 (la) A3 (Ib)

2849b ooAxNx 2850b Ax|Nx 2851b ||w|x(Ax)x|N 2852b ||w|xAx|N 2853b A|=Nx|N 2854b ||w|A|xN 2855b oo||w|w|Ax|N|| 2856b ooA|xNx 2857b ooA|xNx 2858b Ax|Nx 2859b ||w|w|xA|{N} 2860b ooA|N-(N)x 2861b ||w|w|w|xAx|xN 2862b Anx|(Nx) 2863b oo||A|xnNx 2864b ||w|w|wx|A|(N)x 2865b ||w|w|xAx|N 2866b ||w|w|w|A|Nx 2867b ooA|xNx 2868b ooA|xNx 2869b ||wx|w|A=Nx 2870b Ax|Nx 2871b A=xNx 2872b ||w|wx|A|xN 2873b A=xNx 2874b ||wx|w|A|Nx 2875b ||w|w|wx|Ax|xB AxN 2876b 2877b Ax|Nx 2878b oo||w|xA|xN 2879b Ax|Nx 2880b ||wx|w|Ax|N 2881b ||A|nNx 2882b oo||Anx|xN 2883b ||w|wx|xA|xN 2884b xA=(N)x 2885b ooAxNx 2886b ||A=Nx|N 2887b ooAx|N-N 2888b ||wx|(Ax)nx 2889b A|Nnx 2890b ||A|w|Nx 2891b ||wx|An=N 2892b x(Ax)|Nx 2893b ||w|xA=(N)x

301

A l (III) A l (III) El (III) A l (III) A l (III) A l (III)

CM)

D 2 (III) B2(I) El (III) A 2 (III) C 2 (I) A l (III) A l (III) 0(1)

A l (III) A l (III) B1(I) A l (III) Cl(l) B2(I) B1(I) A l (III) B1(I) A l (III) B1(I) A l (III) E2 (III) B1(I) C 2 (II) A l (III) El (III) A l (III) C^I) D 1 (III) A l (Ic) B1(I) C 1 (II) C 2 (I)

Index 2894a A[x}=Nx|N 2895a A=Nnx 2896a Ax=N[x}x 2897a Ax|Nx 2898a AxNx 2899a- ||w w|A=Nx 2900a llw w|A=(N)x 2901a Ax|Nx 2902a ||ax|A=Nx 2903a ||w|xA|(N)x 2904a A=Nx|A 2905a xxA=Nx 2906a + +Ax|xAx 2907a xxAnx 2908a + +A|xxAx 2909a ||ax|(Ax)=Nx 2910a ++Ax|xAx 2911a An=Nx 2912a + +Ax|xAx 2913a Ax|Ax 2914a ++ A|xAx 2915a ax|A=Nx 2916a ||w|wx|A(n)x 2917a + + ||Ax|xAx 2918a ||w|xA=(N)x 2919a ||a|xAx 2920a Ax|(Nx)x 2921a (Ax)=Nnx 2922a ||w|w|xA=Nx 2923a ++Ax|xAx 2924a ||wx|Axn 2925a An|An 2926a ||w|xA-Nx 2927a Ax|Nx 2928a ||wx|w|xAx 2929a + +A|x(Ax)-N 2930a ||xa|A=Nx 2931a + +(Ax)x|A=Nx 2932a (Ax)x|Nx 2933a ||w|w|A(n)x 2934a ||w|w w|xAx 2935a xAx=A 2936a ||xa|w|A=Nx 2937a Ax|Ax 2938a AxlNx

2894b A=Nx|N 2895b ooAx|xNx 2896b xA=(N)x 2897b ||A|N(n)x 2898b ||w|w|A|xN 2899b ooAx|xxNx 2900b Ax|Nx 2901b A=Nx|N 2902b Ax|Nx 2903b Ax=xNx 2904b oo||Ax|xNx 2905b ooA(n)x|Nx 2906b ||An|(N)x 2907b (Ax)|Nnx 2908b A=(Nx)nx 2909b Ax=Nx 2910b ||w w|Ax|N 2911b ||wx|A=Nx 2912b A|(N)nx 2913b ||w|xA|(N)x 2914b ||wx|(Ax)n|N 2915b xAx|N 2916b ooAx|xNx 2917b x(Ax)=(Nx)x 2918b Ax|Nx 2919b ||wx|Ax|N 2920b |w|w|A|Nx 2921b A|n(Nx)x 2922b ooAx|xxNx 2923b ||w|w|Ax|N 2924b ooAx xNx 2925b xAx=(Nx) 2926b ooAx|xNx 2927b A=Nnx 2928b (Ax)|N(n)x 2929b ooA-N|xN 2930b oo||A|xNx 2931b ooAx|x(Nx)x 2932b xA(n)x 2933b A=xNx 2934b Ax=Nx 2935b An=Nx 2936b Ax|Nx 2937b oo||A|N|-xN 2938b A-Nx|N

El (III) D 1 (III) Al (III) Al (III) Al (III) C1 (la) C2 (la) Al (III) C1 (Ib) C2 (la) El (III) C1 (II) A2 (III) C1 (II) A2 (III) C1 (Ib) A2 (III) Al (III) A2 (III) Al (III) Al (III) C1 (Ib) C2 (la) A2 (III) C2 (la) A3 (Ib) Al (III) D 1 (III) C1 (la) A2 (III) Bl (la) Al (III) C1 (la) Al (III) A3 (la) Al (III) C1 (Ib) D*1 (III) Al (III) C2 (la) A3 (la) B 1 (II) C1 (la) Al (III) Al (III)

302

El (III) A2 (III) C2 (II) D2 (III) B1(I) A2 (III) Al (III) El (III) Al (III) A2 (III) A2 (III) A2 (III) A2k (III) D 1 (III) D 1 (III) Al (III) B1(I) O(l) D3 (III) C2(I) B1(I) Bl (II) A2 (III) C1 (II) Al (III) Bl(I) O(l) Al (III) A2 (III) B1(I) A2 (III) Bl (II) A2 (III) D 1 (III) D2 (III) El (III) Al (III) A2 (III) C2 (II) Al (III) Al (III) Al (III) Al (III) D41(III) El (III)

Alliteration, scansion and metrical clause structure 2939a ||w||w|xAnx 2940a Ax|Nx 2941a Ax|x(Nx)x 2942a Ax=Nx 2943a ||wx|w|(Ax)nx 2944a' ++Ax|xAx 2945a Ax|(Nx)x 2946a ||w|xA=(A)x 2947a A=N|(A)x 2948a ||w|xA|xN 2949a ||xw|w|w|xAx 2950a A|(Ax)=Nx 2951a ||A|Axn 2952a ||ax|(Ax)nx 2953a + + Ax|A=N 2954a ||w|w|A=Bx 2955a ||(Ax)=Nnx 2956a ++ A|xAx 2957a ++ A|xxA=N 2958a Ax|Nx 2959a (Ax)=N|(N)x 2960a ||wx|Anx 2961a ||w|w|Axn 2962a Ax=Nx 2963a ||w|xA=(N)n 2964a ++ (Ax)x|Ax|=N 2965a A|Anx 2966a ||w|w|xAx 2967a a|xxAx 2968a (Ax)x|Nn 2969a Ax|Ax 2970a ||wx|A=(B)n 2971a ||xwx|xAx 2972a Ax|Nx 2973a ||w|w|w|xAx 2974a ||w|w|Ax|N 2975a ||a|xAx 2976a | |w|w|wx|xAx 2977a ||w|xAx 2978a Ax|Nx 2979a ++ A=N|(Ax)x 2980a ax|xxA=N 2981a Ax|Nx 2982a Hwlwxl(Ax)x 2983a + + ||(Ax)x|xAx

C1 (la) Al (III) A2 (III)** Al (III) C1 (la) A2 (III) Al (III) C2 (la) A2k (III) Bl (la) A3 (la) D 1 (III) D41(III) C1 (Ib) A2b (III) C1 (la) D 1 (III) Al (III) A2b (III) Al (III) A2k (III) C1 (la) Bl (la) Al (III) C2 (la) D*41 (III) D 1 (III) A3 (la) A3 (Ib) Al (III) Al (III) C2 (la) A3 (la) Al (III) A3 (la) Bl(Ia) A3 (Ib) A3 (la) A3 (la) Al (III) A2a (III) A3 (Ib) Al (III) A3 (la) A2 (III)

2939b Ax|Nx 2940b ||wx|xA=Nx 2941b oo||A[x}|N|-xN 2942b oo(Ax)|N-(N)x 2943b ooA|xNx 2944b ||w|xAx|N 2945b xA|(N)x 2946b °Ax|xNx 2947b ooAx|xNx 2948b ooAx|xNx 2949b xx(Ax)nx 2950b Ax|Nx 2951b oo(Ax)|xNx 2952b ooAx|xNx 2953b oo||Ax|x(Nx)x 2954b x(Ax)|Bx 2955b ooA|xNx 2956b oo||A|N|-(N)x 2957b ||w|w|A|(N)x 2958b ||A|(Nx)nx 2959b ooA|xxNx 2960b x(Ax)|Nx 2961b Ax|Nx 2962b xA|(N)x 2963b (Ax)x|Nx 2964b ||wx|Anx 2965b ooAx|xNx 2966b ooA|Nx|-N 2967b ||w|w|A|xN 2968b oo||w|xA|(N)x 2969b A=N|(N)x 2970b oo(Ax)|xBx 2971b (Ax)|Nnx 2972b A-N|(N)x 2973b 00A|N|-xN 2974b Ax|Nx 2975b ||w|w|Ax|xN 2976b ||w|w|w|A|(N)x 2977b (Ax)nx|N 2978b ||w|xAx|N|| 2979b Anx|N 2980b oo||w|xA|(N)n 2981b ||w|xA|(N)x 2982b ||w|xA|(N)x 2983b ||w|w|xAx|N

303

Al (III) D2 (III) Al (III) B1(I) C2 (II) A2 (III) A2 (III) A2 (III) C1 (II) Al (III) Al (III) A2 (III) A2 (III) C1 (II) Al (III) D2 (III) C2(I) D 1 (III) A2 (III) C1 (II) Al (III) C2 (II) Al (III) Cl(l) A2 (III) D41 (III) B1(I) C2(I) A2k (III) Al (III) D 1 (III) A2k (III) D41 (III) Al (III) B2(I) C2(I) El (III) B1(I) El (III) C2(I) C2(I) C2(I)

Index 2984a ||w w|A=Nx 2985a ||wx|A(n)x 2986a ||w|xAxn 2987a + + A|=B|Ax 2988a ||Ax|Ax 2989a |w|w|Ax|A 2990a ++ Ax|xAx 2991a ||a|xxA=N 2992a Ax|(Nx)x 2993a °(Ax)x|xNx 2994a ||wx|wx|xAx 2995a ++ Ax|x(Ax)x|=Nx 2996a + + A|xAx=Nx 2997a ||w|w|(Ax)x|xN 2998a A=Bnx 2999a ||w|w|xAx 3000a A=N|(A)x 3001a ||w w|Ax|N 3002a llwic w|xAx 3003a Ax=Nx 3004a xAnx 3005a xx(Ax)x|(Ax) 3006a + + ||A=N|(Ax)x 3007a + + A=(N)x|Ax 3008a ||w|w|A=(N)n 3009a llw wx xAx 3010a xA=(N)x 3011a ax|xxA[x]x 3012a A|nNx 3013a ||w|w|xAnx 3014a + +Ax|xAx 3015a ||Ax|Nx 3016a ++ Ax|xxAx 3017a ax|xAx 3018a ||w|w|Ax=N 3019a + +A|xxAx 3020a llw x(Ax)=Nx 3021a + + (Ax)|xA=N 3022a xxAx=N 3023a ||ax|xAx 3024a An|Ax 3025a + +A|xxAx 3026a ||Ax|Nx 3027a ||wx|w|xA 3028a ||w|xA|(N)x

C1 (la) C2 (la) Bl(Ia) A2a (III) Al (III) Bl(Ia) A2 (III) A3 (Ib) Al (III) A2 (III) A3 (la) HA1 (III) HA1 (III) B2 (la) D 1 (III) A3 (la) A2k (III) Bl(Ia) A3 (la) Al (III) C1 (II) Bl(II) A2a (III) A2 (III) C2 (la) A3 (la) C2 (II) A3 (Ib) Al (III) C1 (la) A2 (III) Al (III) A2 (III) A3 (Ib). Bl(Ia) A2 (III)** C1 (la) A2b (III) Bl (II) A3 (Ib) Al (III) A2 (III) Al (III) F(Ia) C2 (la)

2984b Ax|Nx 2985b A|Nnx 2986b Ax=Nx 2987b xxA|(B)x 2988b (Ax)nx|N 2989b ||w|w|Ax|xN 2990b ||w|xAx|N 2991b Ax|Nx 2992b ||w|w xA|xN|| 2993b x(Ax)=Nx 2994b A|Nnx 2995b ||xwx w|xA|xNx 2996b ||wx|w|xAx|xNx 2997b Ax|Nx 2998b ooAx|xBx 2999b xxA=(N)x 3000b ||w|w w|A|(N)x 3001b Ax|Nx 3002b A|Nnx 3003b ||wx w|A|xN 3004b ooA|xNx 3005b (Ax) | Nnx 3006b ||wx|Ax|N 3007b ||w|w|A[x}|(N)x 3008b A|N(n)x 3009b ||w|w|Ax|N|| 3010b ||xw|Ax|N 3011b ||w|w w|Ax|N 3012b ooAx|xNx 3013b Ax|Nx 3014b ||ww|A|(N)x 3015b l|wx|A|(N)x 3016b ||w|A|Nx 3017b A=Nnx 3018b ooAx|xNx 3019b A=N|(N)x 3020b ooAx|xNx 3021b ||xw|w|A|(N)x 3022b ooAx|xNx 3023b ||wx|Ax|N 3024b °||w|xAx|N 3025b oo(Ax)|N(n)x 3026b ||w|w|xAx|N 3027b A|N(n)x 3028b Anx|]N1

304

Al (III) D 1 (III) Al (III) C2 (II) El (III) B2(I) B1(I) Al (III) B1(I) C1 (II) D 1 (III) hAl (I) hA2 (I) Al (III) A2 (III) C2 (II) C 2 (I) Al (III) D 1 (III) Bl(I) Al (III) D 1 (III) Bl(I) C 2 (I) D 2 (III) B1(I) B1(I) B1(I) A2 (III) Al (III) C 2 (I) C 2 (I) C ! (I) D 1 (III) A2 (HI) A2k (III) A2 (III) C 2 (I) A2 (III) Bl(I) Bl(I) D 2 (III) Bl(I) D2 (III) El (III)

Alliteration, scansion and metrical clause structure 3029a 3030a 3031a 3032a 3033a 3034a 3035a 3036a 3037a 3038a 3039a 3040a 3041a 3042a 3043a 3044a 3045a 3046a 3047a 3048a 3049a 3050a 3051a 3052a 3053a 3054a 3055a 3056a 3057a 3058a 3059a 3060a 3061a 3062a 3063a 3064a 3065a 3066a 3067a 3068a 3069a 3070a 3071a 3072a 3073a

Ax|Nx ++ Ax|xAx ||ax|A=Nx Ax=Nx ||wx|w|xAx ++ A=N|Ax Ax Nx ++

Ax|xAx Ax|Nx ||w|w|w|xAx ++ A|xAx Ax|Ax ++ A-N|(Ax)=N ||w|w|A(n)x ++ A|xAx Ax|Nx AxN(n)x ||wx|A=(N)x w|A|Nx | Ax|Nx ++ A(n)x|x(Ax)x An|Nx ||w>: w|xAx A=Nx|A ||w|xA=(N)x (Ax)x|Nx ++ (Ax)x|A=(N)n ||w|w|Nx|xA|| ||wx|xwx|Ax ||w w|xAx llw w A = N x ++ Ax|xxAx AxlNTx

xax|A=Nx ++ A|Ax=N A=xNx ++ J ^. xxAx ||w|w|Anx ax|(Ax)=Nx ||xw x(Ax)x|xN ||w|w|xAx|A Ax|Nx ||w|xA|Nx ++ Ax|x(Ax)x ++ AxxAx

Al (III) A2 (III) C 1 (Ib) Al (III) A 3 (la) A2a (III) Al (III) A2 (III) Al (III) A3 (la) Al (III) Al (III) A 2b (III) C2 (la) Al(III) Al (III) D* 2 (III) C2 (la) C 1 (la) Al (III) A2 (III) Al (III) A3 (la) El (III) C2 (la) Al (III) D* 2 (III) F (la)** A3 (la) A3 (la) C 1 (la) A2 (III) Al (III) C 1 (Ib) D4 (III) Al (III) A2 (III) C 1 (la) C 1 (Ib) B2 (la) Bl(Ia) Al (III) C 1 (la) A2 (III) A2 (III)

3029b 3030b 3031b 3032b 3033b 3034b 3035b 3036b 3037b 3038b 3039b 3040b 3041b 3042b 3043b 3044b 3045b 3046b 3047b 3048b 3049b 3050b 3051b 3052b 3053b 3054b 3055b 3056b 3057b 3058b 3059b 3060b 3061b 3062b 3063b 3064b 3065b 3066b 3067b 3068b 3069b 3070b 3071b 3072b 3073b

305

°°llw|xA|(N)x oo ||(Ax)|=N|xN xxAx=N Ax|N(n)x Ax=Nx ||wx|w|w|Ax|N ||w|w|Ax=N | |w|xA=(N)n A[x]=Nx|N A=Nx|N (Ax)=Nx|N ||w|xA=(N)x oo Ax|xNx A=xNx ||A=Nx|N oo ||(Ax)|N|-xN ||w w|Ax|N oo Ax|xNx oo (Ax)|xNx xAx|N ||w|w|xAx|N A|N(n)x Ax=Nx oo Ax|xNx oo AxxNx ||wx|A|Nx oo Ax||ww|w|Nx A|N(n)x ||w|w|xA|Nx ||w|xA|xN oo Ax|xNx oo ||A|N|-xN ||w|xA|xN oo HAxl]wlNx oo Ax|xNx oo ||wx|A|xN oo (Ax)=N|Nx ||w|w|Ax|N oo ||Ax|xNx Ax|Nx oo Ax|xNx ||w w|A|(N)x Ax|Nx A=Nx|N ||w|xxA|(N)x

C 2 (I) El (III) Bl (II) D* 2 (III) Al (III) B1.(D Bl(I) C2(I) El (III) El (III) El (III) C2(I) A2 (III) Al (III) El (III) D41 (III) Bl(I) A2 (III) Al (III) Bl (II) Bl(I) D2 (III) Al (III) A2 (III) A2 (HI) C^I) A2 (III) D2 (III) OQ)

Bl(I) A2 (III) D41 (III) Bl(I) A2 (III) A2 (III) Bl(I) A2a (III) Bl(I) A2 (III) Al (III) A2 (III) C2(I) Al (III) El (III) C2(I)

Index 3074a ||w|w|A=(B)x 3075a Anx|A 3076a ||An|(Nx)(n)x 3077a ||w|w|A|(N)x 3078a °A|xNx 3079a' [[xwxlwlxAx 3080a Ax|Nx 3081a HwlwlxAx 3082a ||ax|wx|Ax 3083a ||Ax|(Ax)x 3084a ||a|xA=xB 3085a ++ ||Ax|xAx 3086a ||w|xxA=(N)n 3087a HwlwlxAx 3088a (Ax)x|Nx 3089a ||wx|A=Nx 3090a a|xxA=N 3091a ax|xAx 3092a A=xNx 3093a (Ax)x|Nx 3094a ++A|xxAx 3095a ++(Ax)|xAx 3096a + + ||w||w|wJxAx 3097a xA=(N)x 3098a ++(Ax)x|xAx 3099a + + An|A=Nn 3100a ||wx|w|A=(N)x 3101a [|ax|w|Ax 3102a ++ A|xAx 3103a ++Ax|xxAx 3104a [|w|w|xAx 3105a ++ Ax|xA|=B 3106a + + ||Ax|xAx 3107a Hw|wxlx(Ax)x 3108a Ax|Nx 3109a xxAnx 3110a HwlwlxAx 3111a ++ (Ax)|Ax=N 3112a A=Nnx 3113a Ax|(Ax)x 3114a ++Ax|xAx 3115a ||ax|Ax|N|| 3116a ||wx|w|A|xN 3117a ||wx|Ax|A 3118a axxA=N

C2 (la) El (III) D*2 (III) C2 (la) Al (III) A3 (la) Al (III) A3 (la) A3 (la) Al (III) Bl (Ib) A2 (III) C2 (la) A3 (la) Al (III) C1 (la) A3 (Ib) A3 (Ib) Al (III) Al (III) A2 (III) Al (III) A3 (la) C2 (II) A2 (III) D*1 (III) C2 (la) A3 (la) Al (III) A2 (III) A3 (la) A2b (III) A2 (III) A3 (la) Al (III) C1 (ID A3 (la) D41(III) D 1 (III) Al (III) A2 (III) Bl(Ib) Bl(Ia) Bl(Ia) A3 (Ib)

3074b Ax|Bx 3075b ooA|xNx 3076b Anx|(Nx) 3077b Ax|Nx 3078b ||w|w|xAx|N 3079b Ax|Nx 3080b A|Nnx 3081b A=N|(N)x 3082b ||w|w|Ax|N 3083b x(Ax)=Nx 3084b oo||A|w|xBx 3085b ||w|x(Ax)x|xN 3086b oo(Ax)|xNx 3087b ||w|xA|xN 3088b ||w|w|xAx|N 3089b ooA|xNx 3090b ||w|x(Ax)x|xN 3091b (Ax)=Nnx 3092b oo||(Ax)|N|-xN 3093b oo||(Ax)|w|xNx 3094b oo ||A|=N|xN 3095b ||w|wx|Ax|N 3096b xx(Ax)|Nx 3097b ooA|xx{N} 3098b ||w|w|Ax|N 3099b ooAx|xNx 3100b Ax|Nx 3101b ooAxNx 3102b ooA=Nx|xN 3103b oollw[wlA(n)x 3104b A|N(n)x 3105b ||w|xA|(B)x 3106b ||wx|w|A|(N)x 3107b A|Nnx 3108b ||w|w|Ax|N 3109b ooAx|x(Nx)x 3110b (Ax)|Nnx 3111b (Ax)x|(Nx)x 3112b ||w|w|A=(N)x 3113b A=Nnx 3114b ||w|w|A|(N)x 3115b (Ax)x|Nx 3116b Ax=Nx 3117b ooAx|xNx 3118b oo||A|Nx|-N

306

Al (III) Al (III) El (III) Al (III) B1(I) Al (III) D 1 (III) A2k (III) B1(I) C1 (II) A2 (Ic) B2(I) Al (III)

Al (III) B2(I) D 1 (III) D41(III) A2 (Ic) El (III) B1(I) C1 (II) A2 (III) B1(I) A2 (III) Al (III) Al (III) E2 (III) C2(I) D2 (III) C2(I) C2(I) D 1 (III) B1(I) A2 (III) D 1 (III) Al (III) C2(I) D 1 (III) C2(I) Al (III) Al (III) A2 (III)

Alliteration, scansion and metrical clause structure 3119a 3120a 3121a 3122a 3123a 3124a 3125a 3126a 3127a 3128a 3129a 3130a 3131a 3132a 3133a 3134a 3135a 3136a 3137a 3138a 3139a 3140a 314 la 3142a 3143a 3144a 3145a 3146a 3147a 3148a 3149a 3150a 3151a 3152a 3153a 3154a 3155a 3156a 3157a 3158a 3159a 3160a 3161a 3162a 3163a

||(Ax)=Nx|A ||wx|xAx xax xAx ++

(Ax)x|xAx ||ax|Ax|N Ax=Nx Ax=Nx ||w|w|xAx ||wx|A-Nx ++ Ax xAx ||Ax|Ax ||w|w|(Ax)=Nx Ax|Nx ++ A|xxA=N ||A|A(n)x ||w w|Ax|N ++ ||A-N|A=N ++ A|Ax=N ||w|w x(Ax)x ++ A|xAx ++ Ax|xAx Ax|Ax ||xwx|w|xAx (Ax)|Anx ||xwx|w|xAx An Ax ++

A|xx(Ax)x Ax|xAx ||w|w|w|xA=B ++ A|xAx ++ A=(B)x|Ax ||wx|Ax=A ++

supplied

a A=(N)x ||w|w xxA=(B)x A=Nx|A ++ Ax|xA=N HxAxjN ++ A|xAx ||A=Nnx [|w|xA(n)x (Ax)=Nx|A ++ Ax|xAx (Ax)=Nx|B ||w|xA|(N)x

3119b ooAx|xNx 3120b (Ax)|Nnx 3121b (Ax)x|Nx 3122b xAnx 3123b xxAn=N 3124b ||w|xAx|N 3125b ||w|w|xAx|N 3126b l|w|xA|(N)x 3127b AnxN 3128b x(Ax)|(Nx)x 3129b oo ||A|Nx-N 3130b ooA|x(Nx)x 3131b oo||(Ax)|N|-(N)x 3132b ||wx|A|(N)x 3133b Ax|Nx 3134b xA|(N)x 3135b ||(Ax)n|(N)x 3136b x(Ax)=Nx 3137b Ax|Nx 3138b A=Nnx 3139b Ax=Nx 3140b ||w|w|Ax|N 3141b Ax|Nx 3142b A-N|Nx 3143b A=Nx|N 3144b oo ||(Ax)=N|xN 3145b Anx|N 3146b oo ||A=N|xN 3147b x(Ax)|Bx 3148b J|(Ax)|nNx 3149b A=Nx|B 3150b Ax|Nx 3151b Ax=Nx 3152b oo||Ax|xNx 3153b ooAx|xBx 3154b (Ax)x|(Nx)x 3155b oo ||(Ax)|Nx-N 3156b Ax|Nx 3157b ||w|w|A|xN 3158b ooAx|xNx 3159b xA|(N)x 3160b ||Ax|Nx 3161b ||ww|A=Nx 3162b Ax|Bx 3163b ooA|xNx

El (III) A3 (la) A3 (Ib) A2 (III) Bl(Ib) Al (III) Al (III) A3 (la) C1 (la) A2 (III) Al (III) C1 (la) Al (III) A2b (III) D 2 (III) Bl(Ia) A2b (III) D 4 1 (III) A3 (la) Al (III) A2 (HI) Al (III) A3 (la) D 1 (III) A3 (la) Al (III) A2 (III) A2 (HI) A3 (la) Al (III) A2 (III) Bl (la) C2 (Ib) C2 (la) El (III) A2b (III) Bl(Ia) Al (HI) D 1 (III) C2 (la) El (III) A2 (HI) El (III) C2 (la)

307

A2 (III) D 1 (III) Al (III) C1 (II) Bl(II) Bl(I) Bl(I) C 2 (I) El (III) C1 (II) D41(111 Al (III) D 2 (III) C 2 (I) Al (III) C2 (II) A2k (III C1 (II) Al (III) D 1 (III) Al (III) Bl(I) Al (III) Al (III) El (III) El (III) El (III) El (III) C1 (II) Al (III) El (III) Al (III) Al (III) A2 (III) A2 (III) Al (III) D41(111 Al (HI) Bl(I) A2 (III) C2 (II) Al (III)

O(l) Al (III) Al(III)

Index 3164a llw wx|Ax 3165a A=B[x}x|N 3166a ||xwx|Ax xN 3167a ++ A|xAx 3168a + + Ax|w|A=N 3169a llw xxA|(N)x 3170a (Ax)nx|N 3171a ||wx|(Ax) Ax 3172a ||A=N|(A)x 3173a ||axx|A=(N)x 3174a (Ax)x|Ax 3175a ||w w|x(Ax)=Nx 3176a II A ,: Ax 3177a xA=(N)x 3178a ||w[xA(n)x 3179a A-Nx|(Ax) 3180a + + ||wxllwlw|Ax 3181a Ax|Ax 3182a Ax Ax

A3 (la) El (III) B2 (la) Al (III) A2b (Ic) C2 (la) El (III) C1 (la) A2k (III) C2 (Ib) Al (III) C1 (la) Al (III) C2 (II) C2 (la) El (III) A3 (la) Al (III) Al (III)

3164b 3165b 3166b 3167b 3168b 3169b 3170b 3171b 3172b 3173b 3174b 3175b 3176b 3177b 3178b 3179b 3180b 3181b 3182b

308

||wx|xAx|N x(Ax)|Bx Ax|Nx ||w|w|xA|(N)x ||w|w|Ax|N Ax=Nx oo Ax|Nx ||w|(Ax)|Nx ||wxA|(N)x ||w|xAn=N ||w|w|xAx|N Ax|Nx ||wx|w|A|(N)x Ax|Nx Ax|Nx A=xNx (Ax)|(N)nx xA=Nx xA=Nx

B1(I) C1 (II) Al (III) C2(I) B1(I) Al (III) Al (III)

O(T)

C2(I) B1(I) B1(I) Al (III) C2(I) Al (III) Al (III) Al (III) D3 (III) C1 (II) C1 (II)

Index of verses specially discussed

The index lists all verses which are quoted for discussion in the text or notes. It does not include verses which are cited by number as supporting evidence. References give the verse in question followed by a colon and then the page or pages on which the verse is quoted. Andreas 88: 36n21 88a: 82 n28 761a: 82 n28 1026a: 82 n28 1219a: 82 n28 1488a: 135 1636: 13 Battle ofFinnsburh 13a: 135 Battle of Maldon 112a: 15 n5 271-2: 73 Beowulf 1-11:6-7 1-3: 29, 89 lb: 30, 116 2:31 3b: 31 4a: 7, 166, 194 5a: 7, 70 6a: 7 7a: 7, 164 7h^8: 50 8a: 7 11:7-8 18a: 47 18b: 186

19b: 116 21a: 168 25a: 84, 85, 203 26-7: 75 34a: 76 36b-40a: 124 36b: 156 47-8a: 20 49a: 179 50b: 125 54a: 166 57b: 94, 96, 97 61a: 114 n32 72a: 113 80a: 126 83b: 128 90a: 172, 174 92a: 92, 95 102a: 141 105a: 165, 166, 167 105b: 54 107a: 84, 86, 203 107b: 77 109a: 124 109b-10: 146 110a: 139, 140 114a: 141 120b: 192

309

Index of verses specially discussed Beowulf cont. 123b: 187 125a: 77 128b: 189 129a: 133 129b-30: 29 133b: 75, 101, 124 137b: 123 145a: 153 153b: 155, 180 n9 156b: 196 157a: 151 158a: 133 161a: 90 164a: 151 164b: 162 n6, 163, 169, 193 170a: 149 171b: 155 176a: 161 n3, 196 178a: 161 n3 180b: 125 182a: 33 183b: 92, 98 186b: 92, 98 187a: 161 n3 191a: 102 191b: 101 196-8a: 61 197: 11 197a: 62, 66 199b: 92, 95 202a: 77, 179 210a: 108, 185 215b: 186 217-18:22 217a: 23, 24, 34 219b: 68 223a: 163, 204 232a: 164, 182 232b:179 237-40a: 122 237a: 26 248b: 157 251a: 176 251b: 156

257b: 132 262a: 132 266a: 184 266b:119 270-2a: 94 270-la: 72 270b: 70, 71 272: 90 272a: 96 272b: 92, 97 279a: 115 281b: 189 286a: 196 287a: 180 290b:154 292b: 56 294a: 103 303b: 165 307a: 58 308a: 163 311a: 138 n28 316b: 177, 213 318b-19: 79 319a: 76 320a: 163 321b-23a: 50 323a: 49, 53 324a: 161 n3 326a: 204 326b: 143 330a: 189 336a: 180 336b-37: 122 338a: 93, 95 343b: 45 n4, 47, 136 344-6a: 72 345b: 62, 70, 71 347a: 121 348a: 204 356-7: 59 357a: 176 363a: 77 367b: 164 371b: 194 376a: 189

310

Index of verses specially discussed 537b: 179 549-53a: 39, 79 552b: 77 560a: 58 568a: 117 569a: 125 570a: 172, 174 573a: 198 574b: 109 581b: 120 582a: 133 583b-86a: 122 583b: 189 585a: 121 591a: 122, 153 592a: 178 594a: 127 595b: 132 600a: 90 n5 603b: 93, 98 609a: 170 612a: 33 619b: 165 620-1: 20 620a: 23, 24, 45, 58, 76 621a: 113 621b: 184 623b: 161, 166 626a: 164 627a: 152 636b-38: 37, 146 636b: 103 639-41: 139 639a: 77, 139 641a: 182 642a: 115 643a: 165 649b:157 651b: 149 652b: 156 655a: 61, 66, 68 657b: 195 658a: 90 659a: 21 662-5a: 107

380a: 163 388b: 180 nl2 391-4: 180 391b: 148 392b: 132 394a: 161 n3 394b: 180 395b: 161 n3 397a: 53 400b: 154 402a: 58 406b: 176 417a: 194 426b: 195 426b-32: 123 432a: 141 433a: 33 434b:125 436a: 133 442a: 93, 96 445a: 168 n21, 194 450b: 126 455b: 93, 97 457a: 77 467b-69a: 83 468a: 77, 131 469b: 93, 97 484-7a: 83 486a: 77 495b: 163, 164, 169 498a: 145, 191 502b: 70, 176 503a: 125 503b: 132 504a: 130 506-7: 80 507a: 77 510b: 68 515a: 179 517a: 172 518a: 61, 66 521a: 137 525a: 26 529b: 194, 195 535-6a: 90 n5

311

Index of verses specially discussed Beowulf cont.

851a: 161 n3 858a: 115 864-5:81 865a: 77 867b-87a: 1-2 870b: 149 873b: 102 880a: 26, 152 881a: 137 885a: 161 n3 885b: 192 889b: 147 892a: 181 904a: 104 905a: 58 909a: 147 910a: 143 921a: 161 n3 923b: 118 929b: 154 933a: 125 941a: 150 942b-46a: 35 943a: 128 946a: 161 n3 956b: 128 960a: 168 965b: 154 980a: 198 984b: 185 985a: 109, 205 nl8 987a: 169, 169 n23, 192, 192 n6 991b: 181 992b: 155 996a: 106 996b: 156 997a: 141 1002a: 176 1011b: 68 1022b: 115 1027a: 124 1030^: 117 1030a: 77, 143 1040a: 102 1044a: 170

662b: 101 665a: 107, 119 666a: 148 667a: 104 676a: 194 682a: 105 686-7: 128 689a: 176 689b: 157 693a: 114 694b: 156 705a: 151 706b: 127 708-9: 51 712-13: 153 713a: 152 713b: 139 727b: 192 730a: 165 731a: 90, 92, 97, 110 734b-38: 67 736a: 61, 66 74lb: 192 743a: 165 745b:187 750a: 24, 110 755-7: 57 758-61: 78 758a: 76, 84, 139 773b: 123 776a: 150 778a: 125 786b:176 791b: 154 792a: 161 n3 799a: 161 n3 802a: 151 807a: 181 811b: 120, 120 nl2 813: 141 818b-22: 57 833a: 191, 192 840a: 161 n3 848a: 103, 108

312

Index of verses specially discussed 1224b: 94 1227a: 101 1228-31:94 1230b: 177 1232a: 18 1236b: 148 1239a: 47 1245: 162 1246b-50: 87 1247a: 98 1248a: 84, 121 1251b: 154 1253: 18 1258t>-59: 19 1264b: 164 1266a: 161 n3 1276a: 162 n6 1286b: 118 1288b: 165 1290b-9 la: 46 1296-8a: 19 1297a: 101 1299a: 197 1302a: 18, 33 1304b: 101 1307a: 163 1311a: 165, 167 1311b: 117 1317b: 164 1319a: 170 1331a: 167 1331b: 93, 97, 125 1332a: 173 1336a: 123 1337a: 110 1341b: 149 1349h^53: 83 1351a: 176 1351b: 77 1353a: 152 1356a: 26, 152 1362a: 161 n3 1369b: 164 1373-76a: 107 1375a: 106

1046-8: 122 1046a: 122 1055a: 105 1056a: 141 1057b: 157 1060b: 155 1065a: 164 1065b: 189 1069a: 170 1084b: 105 1108b-13a: 39 1110a: 77 1111b: 177 1114-15: 74 1122b: 157 1127h^29a:42 1127b: 128 1129a: 192 n5 1131b: 185 1136b-38a: 42 1137a: 173 1137b: 55 1140b: 126 1146a: 161 n3 1149a: 161 n3 1153a: 118 1155a: 133 1163a: 171 n30 1165a: 153 1166a: 84, 172 1167a: 150 1168a: 110 1174a: 111, 113 1179a: 112, 114, 115 1182b: 93, 97 1184a: 93, 95 1185a: 70 1185b: 150 1188-90a: 111 1189b: 111 1191b: 117 1192-4a: 112 1199a: 141 1216a: 138 n28 1223a: 128

313

Index of verses specially discussed 1553b: 111 1583a: 26, 66 1584a: 182 1591a: 24, 111 1604a: 90 n5, 125 1605a: 103 1608a: 26, 150 1611a: 113 I6l2-15a: 35 1614a: 216 nl8 I6l5b-17: 52 1616a: 52 1616b: 123 1620-2: 75 1622b: 141 1627a: 204 1633b: 165 I666b-68a: 53 1672a: 165 1675b: 62, 68 l679t>-86: 40 1684a: 77, 81 1699b: 157 1701b: 157, 185 1704a: 161 n3 1704b: 149, 172, 194 1705a: 151 1706b: 156 1707a: 94, 185 1713-18a:80 1713-15: 52 1717b: 62, 77 1719a: 163, 164 1721: 143 1724a: 76 1732a: 121 1738b: 62, 66 1745-52: 87 1751a: 85 1755b: 157, 188 1760b: 125 176lb-68: 87-8 1762b: 187 1763a: 115 1767a: 85

Beowulf cont.

1376b-79a: 136 1390a: 76 1395a: 61, 70, 71 1396b: 147 1405a: 117 1408a: 76 1412a: 152 1416a: 64 1420a: 109 1422a: 184, 185 1422b: 186, 187 1424b: 109, 205 nl8 1432a: 165 1435b: 44 1436a: 130 1438a: 161 n3 1441b: 56 1448a: 141 1454a: 110 1455a: 127 1469a: 118 1480b: 132 1483a: 194 1487b: 55, 90 1490a: 164 1492a: 35, 38, 77, 82 1497b: 103 1498b: 180 n9 1501a: 105 nlO 1504b: 84, 126 1509b: 123, 156 1514a: 132 1520b: 125 1521b: 161 1525b: 157, 188 1532a: 180 1537-^0: 78 1537a: 76, 84, 120 1538a: 161 n3, 165 1547b-49: 80, 87 1549a: 77, 85, 121 155O-4a: 112 1551a: 64

314

Index of verses specially discussed 1937b: 187 1941b: 182, 183 1955a: 132, 133 1960a: 148 1966a: 189 1974b: 118 1984a: 70 2001a: 70 2011: 141 2025b: 172, 174 2032b: 194 2033b: 62

1769b: 180 n9 1771a: 151 1772b: 156 1773b: 84, 126 1785b: 189 1791a: 53 1792b: 189 1797b: 62, 68, 69 1810a: 92, 95 1812b: 198 1819b: 56 1840b: 176 1843a: 121 1853b: 33 1854a: 128 1855-6Oa: 93 1859a: 96 1864a: 121 1869b: 189 1870a: 76 1870b: 173, 174 1877b: 84 1886a: 176 1888-1913: 31-2 1892b: 37 1894a: 95, 180 n 12 1895a: 180 nl2 1895b: 37 1896-7: 58 1896a: 32 1899a: 37 1902a: 37 1903b-4: 58 1905a: 32 1909a: 51 1910b: 37 1911b: 37 1912b-13:40, 79 1913b: 37, 77 1914b: 166 1923a: 195 1926b: 189 1927a: 169, 177 1932b: 176 1935a: 147

2047-52: 66

2047-5 0a: 44-5 2047a: 62, 66

2059a: 143 2076b: 176 2091a: 152 2091b: 125 2093a: 77, 83, 84, 120, 123 2097a: 135 2098a: 134 2106a: 161 n3 2117a: 151 2119b: 185 2120b: 192 2121a: 109 2121b: 197 2123b: 176 2138a: 185 2155-7: 73 2i56b:62, 70, 72 2158a: 96, 97 2160a: 149 2162b:157, 188 2163a: 93, 95 2171a: 15 n5 2172a: 93, 95 2180b-83a: 52 2182a: 168 2201a: 161 n3 2208a: 118 2215a: 184 2232b:178 n6 2236b: 138

315

Index of verses specially discussed Beowulf cont. 2239a: 163, 168 2239b: 139, 157 2240a: 135 2246b: 134 2247b: 126 2252b: 92, 93, 95 2255a: 141 2264b: 141 2266a: 151 2268a: 153 2269a: 113 2275a: 104 2278b: 183 2283a: 179 2288a: 21 2289a: 166 2293b: 165 2296b: 108 2297a: 84, 86, 133 2309b: 181 2312a: 109 2317b: 110 2338a: 161 n3, 178, 178 n6 2345a: 26 2351a: 161 n3 2357b: 148 2364b: 147, 148 2367a: 34 2369b: 114 2404a: 197 2406b: 198 2413b: 192 2416b: 156 2420b: 189 2426a: 151 2429a: 161 2431a: 34 2434a: 114 n32 2434b: 148 2436b: 165 2440a: 196 2462b-65: 75 2467a: 64 2481a: 26, 152

2487a: 103 2489a: 106 2489b: 166 2492a: 133 2509a: 172 2511b: 157 2519b: 93, 97 2527a: 108, 173 2532b^35a: 102 2533a: 101 2536a: 105 2538b: 180 2544a: 161 n3 2545b: 190 2548b: 68 2551b: 189 2556b: 187 2560a: 161 n3 2565b: 157 2566a: 105 2570a: 101 2578b: 192 2587: 141 2588a: 150 n8 2591^92:203 2602a: 47 2604b: 75 2606a: 21, 78 2608a: 105 2615: 161 2618a: 182 2619a: 37 2620b: 180 n9 2627a: 161 2634b: 71 2650a: 168 2651a: 129 2652a: 161 n3 2656b-59a: 127 2662b: 134 2663-6a: 136 2663b: 157, 188 2667b: 68 2669a: 35, 38, 82 2673a: 120, 120 nil

316

Index of verses specially discussed 2675a: 142 2677b: 128 2696b: 101, 216 2699b: 187 2705a: 76 2708b: 154 2717b: 110, 173 2722a: 165 2731a: 133 2732a: 102 2739b: 150 2741a: 126 2749b: 130 2772a: 150 2774a: 103, 108 2777h^80a: 35 2794-8: 158 2794b: 157 2797b: 65, 68 2810a: 204 2811b: 161 2815a: 61 2860b: 176 2867-70: 129 2869a: 137 2879a: 65 2880a: 129 2881b: 192 2889b: 148 2892-4: 58 2899b: 157 2908a: 153 2917b: 177 2925a: 195 2935a: 161 n3 2939a: 95 2941a: 107 nl5, 119 2942b: 177 2946a: 161 n3 2951a: 199 n 16 2953b: 64 2956b: 190 2959a: 150 n8, 165 2965a: 195 2967b: 127

2971-2: 199 2971: 141 2977-80a: 57 2982a: 152 2987a: 172 2987b: 132 2993a: 113, 114 3001a: 147 3002^:91 3004b: 114 3012a: 191, 192 3019a: 129 3022a: 133 3024b: 141 3028a: 142 3029b: 125 3044b: 187 3051-6:95 3056a: 109 3065a: 137 3065b: 165 3068a: 118 3091a: 151 3092b: 187 3096a: 92, 97 3097b: 139 3098a: 152 3104a: 152 31l4b-18: 47-8 3120-2: 54 3120: 141 3127b: 62, 68 3129b: 157 3138b: 181 3141a: 78, 110 3143a: 78 3150a: 161 n3 3166a: 75 3169-82: 8-9 3180a: 97 3181a: 9 n22

Ccedmon's Hymn 1-9: 212 1: B 4b: 213

317

Index of verses specially discussed Juliana 19b-22a: 36 520b: 195 nl2 630a: 193 n8 699b-700: 98 nl8

Ccedmon's Hymn cont.

8b: 213 Christ

301-3: 72 1317a: 135

Maxims I

Christ and Satan

94b: 180 n i l

489a: 102 n5 685a: 114 697: 164

Order of the World

27a: 102 n5 Riddle 39

Durham

22b: 124

1-21: 214-15 5a: 215 6a: 216 7a: 215 lla: 216 16b: 216 17a: 215 18a: 215 21a: 215

Riddle 46 6SL: 137 Panther

44-45a: 82 53b-54: 82 Seafarer 1: 211 112a: 121 Seasons for Fasting 39: 13 Solomon and Saturn

Elene

2a: 183 271: 13 591a: 44 Exodus

299a: 82 n28 565a: 82 n28

58a: 179 Soul and Body I

12a: 183 Soul and Body II

62b: 188 n30

Genesis

121a: 135 634a: 135 1040-3a: 82 1254a: 193 n8 1677a: 102 n5 2520a: 135 Guthlac

328a: 44

Wife's Lament

1-5: 210 1: 211 3b: 211 4b: 211 42-3a: 211 Wulfand Eadwacer 6: 211 12: 211 I6b-17a: 211

318