[In the following excerpt, Orchard investigates the principal elements of Wulfstan's homiletic style, maintaining that the “essence of Wulfstan's technique is repetition.”]

Archbishop Wulfstan enjoyed a high reputation as a stylist amongst his contemporaries; when he was still bishop of London (996-1002) one correspondent spoke of the ‘very sweet wisdom of [his] eloquence and the richness of [his] composition fittingly organised’, whilst the wide dissemination of his sermons and their susceptibility to imitation bear dual witness to his popularity throughout the eleventh century.¹

For modern readers, however, the problem of defining the origins of Wulfstan's sermon style has proved acute. On the one hand, Dorothy Bethurum pointed to a whole battery of ‘manuals of rhetoric that Wulfstan knew’, by authors such as Alcuin, Isidore and Hrabanus Maurus, and she gave several examples from his sermons of rhetorical figures such as verborum exornatio, gradatio, traductio, interpretatio, conversio, conduplicatio and dubitatio.² But against Bethurum's insistence that Wulfstan was steeped in the Latin rhetorical tradition might be set the comment of Dorothy Whitelock on the Sermo Lupi ad Anglos, that ‘no work smells less of the study’.³ Certainly few scholars have been able to identify direct Latin sources. Whitelock herself indicated that a concluding passage referring to Gildas in one of the three versions of the Sermo Lupi ad Anglos (no. 20c) was drawn from a Latin letter by Alcuin to Archbishop Æthelheard, and highlighted the fact that the relevant sentence of the Latin is copied separately in Cambridge, Corpus Christi College 190, which Bethurum has elsewhere indicated to be drawn from what she described as ‘Archbishop Wulfstan's Commonplace Book’.⁴ But Wulfstan has not simply incorporated this passage unaltered: Whitelock notes that ‘Wulfstan translates very freely, expands and alters to suit his purpose’.⁵ A further passage from the same ‘commonplace book’ (CCCC 190) has been described as a ‘probable’ influence.⁶ Cross and Brown recently published the Sermo ad milites of Abbo of Saint-Germain-des-Prés, found in the manuscript Copenhagen, Royal Library, Gamle Kongelike Samlungen, 1595 (s. xi¹, before 1023, probably Worcester or perhaps York) which appears to have been annotated by Wulfstan himself; the opening and structure of this Latin sermon is similar to that of the Sermo Lupi, but it is significant that Cross and Brown do not claim Abbo's work as a source for Wulfstan's sermon, but instead describe it as a ‘literary impetus’.⁷

Moreover, in a detailed commentary on the last in a series of sermons by Wulfstan on baptism, the so-called “Sermo de baptismate” (no. 8c), Bethurum herself has indicated that the Latin passage from the De ordine baptismi of Theodulf of Orléans upon which the introductory section in Wulfstan seems clearly based is itself a tissue of ornate and rhymed formulae of exactly the sort with which Wulfstan is liable to adorn his vernacular prose, but which he unaccountably eschews here.⁸ Similarly uncharacteristic coyness in the use of rhetorical pyrotechnics is exhibited elsewhere by Wulfstan, for example in his sermon “Be godcundre warnunge” (no. 19), part of which is based on a highly patterned passage from Lev. XXVI.3-12 which in the Latin of the Vulgate seems ripe for Wulfstanization. In short, here and in a number of other sermons, Wulfstan's Old English does not attempt to match the structured syntax of his Latin source in any way.

Wulfstan clearly works at a much greater distance from his sources than, say, Ælfric. This is true even when, as we shall see, that source was Ælfric. The only sources named by Wulfstan are biblical, with the exception of a single passing reference to Gregory (“Sermo 10c,” line 48), and there too Gregory's Latin is transmuted into Wulfstan's idiosyncratic style.⁹ For an initial example of this transmutation we might consider Wulfstan's characteristic treatment of the Lord's Prayer, in contrast with that of Ælfric, as set out below (significant differences are given in italics):¹⁰

The two Ælfrician versions given here could be multiplied, and there are several other Old English renderings of the Lord's Prayer in the vernacular Gospel of Matthew, and in the interlinear glosses of the Lindisfarne and Rushworth Gospels; still further versions were printed by Cook.¹¹ All these extra versions, however, agree broadly with Ælfric, and help to set Wulfstan's anomalous position into still greater relief. Ælfric and the others remain remarkably faithful to their Latin source, even to the extent of often retaining Latin word-order, whilst Wulfstan struggles to make the text his own. Wulfstan expands his source in ways which no other translator does, whether it be the simple addition of and in clauses [3] and [4], presumably to make them parallel to [7] and [10], or the addition of the intensifying adverbs and adverbial phrases that are his stock-in-trade, such as a and ecelice in [2], symble in [3], ealles to swyðe in [10]. Other changes are similarly characteristic, and reflect aspects of Wulfstan's idiolect. The words gemiltsian and miltsian in [7] and [8], or foster in [6], are fairly frequent in Wulfstan's works, whilst forgyfan is rather rare, and hlaf surprisingly absent; similarly Wulfstan shuns the term gylt, which Gneuss has characterized as a word particularly associated with Winchester, and prefers the biblical Amen to Ælfric's etymologising sy hit swa.¹² Most of the changes, however, seem to be rhythmical rather than semantic, and it is striking that of the three extant poetic expansions of the Lord's Prayer in Old English, that found in the Exeter Book, which is the shortest, expands its Latin source in a similar way.¹³ It is also important to point out that Wulfstan prefaces his version of the Lord's Prayer with the exclamatory word eala (for which Bosworth-Toller suggest the intriguing range of translations ‘O!’, ‘Alas!’, ‘Oh!’, ‘Ahah!’); four verses in the translations of The Paris Psalter, and a further three of the so-called Meters of Boethius likewise begin with the interjection eala.¹⁴

Further features which align Wulfstan more closely with the vernacular verse tradition rather than with prose are discussed below. Here we might take a further look at the way in which Wulfstan alters his sources by considering in detail one of a number of cases where his immediate model is the equally idiosyncratic prose of Ælfric himself.¹⁵ A full parallel text of Wulfstan's “Serm. 9” (“De septiformi spiritu”) and its immediate source, a homily by Ælfric most conveniently edited by Napier, is given as an Appendix, below, pp. 259-64. Bethurum prints a Latin preface to Wulfstan's sermon which is not in fact found in any of the three manuscripts containing this work, but has been transferred from Napier, where it introduces Ælfric's rendering. Since Wulfstan never makes use of this Latin preface in preference to Ælfric's Old English version, I omit it in the Appendix. I offer parallel texts of the work of each author, with omissions and additions indicated by gaps in the relevant line, and significant differences highlighted in italics, as before.

The way in which Wulfstan has comprehensively tinkered with Ælfric's prose is highly reminiscent of the way in which he can be observed to alter other sources, such as the Latin text of the Lord's Prayer. McIntosh has noted that: ‘Here is a curious situation in a troubled age; one man produces a special kind of rhythmical writing with a distinct and recognizable texture, then another, heavily burdened with the cares and duties of an enormously responsible position takes the trouble to dissect all this and reconstruct it according to the rules of his own rhythmical practice.’¹⁶ In fact, as we have already seen, Wulfstan constantly revises and reworks his sources, both Latin and vernacular; we shall soon see further that Wulfstan habitually re-uses and recycles his own words also.

Wulfstan expanded the text of the Lord's Prayer, and here too Wulfstan's text is longer than that of Ælfric, as can be seen from the greater number of gaps in the latter's lines. In fact whilst I have given almost all of Ælfric's homily, Wulfstan's sermon continues for some forty-four lines in Bethurum's edition, with a very characteristic reverie on the Antichrist (“Serm. 9,” lines 107-51). Wulfstan is particularly prone to extend Ælfric's prose at the end of sentences, as in my examples [3], [4], [7], [9], [10], [12] and [15]-[21]. Other additions are simply mechanical; they underline Wulfstan's concern for repetition and order, and appear to have been employed by him to highlight the effect of a list on his audience. So in discussing the seven gifts of the Holy Spirit he adds the phrase þurh Godes gyfe in sentences [4]-[9], and in discussing the ‘ungifts’ (to use a Wulfstanism), he adds the phrase þe of Godes agenre gyfe cymð in sentences [16]-[21]. Several other additions are similarly repetitious, rhythmical and emphatic, and he strives to make his sentences more formally regular than those he inherits. So in sentence [2], for example, he omits and four times, to give his list greater regularity (and we might recall how he inserted the word and twice in the Lord's Prayer from similar motives), whilst in sentence [3] he pedantically adds another be, presumably to give shape to the phrase be his mæðe and be his modes geornfulness. In the same way he adds and to the beginning of sentences [4], [9]-[11] and [16]-[21] in the search for similar uniformity. We also have evidence of Wulfstan's concern for intensifying adverbs and adverbial phrases; he adds and witodlice [4], ac bið aa gefædd on æghwylce wisan [7], mid ealle on ælce wisan and dæghwamlice [12], eac for oft [15], swicollice and swiðost [20], on þam þe gyt wyrse is [21] and so on; there are a number of other examples.

Again, Wulfstan will often add alliterating doublets and chiming phrases where Ælfric has none, for example ne mid worde ne mid weorce [9] (and cf. [21]), or to hæbbenne & to healdenne [10], or even the remarkable swa þæt he ne bið on gefean to fægen ne on wean to ormod [7]. It is important to realize that all these phrases just quoted, including this last, can be paralleled in whole or in part elsewhere in Wulfstan's sermons also, and that Wulfstan is expanding Ælfric's work with words and phrases drawn from his own repertoire.¹⁷ In a similar vein he will expand one term in Ælfric to two, often with alliteration. Ælfric has wisne [15], Wulfstan wærne & wisne; Ælfric has modes strengðe [18], Wulfstan modstaðolnysse & modes strencðe; Ælfric has syleð ([19] and [20]), Wulfstan sæwð & sendeð.¹⁸

Still more interesting are the niggling changes of vocabulary which reflect Wulfstan's own idiolect: he will use un- compounds where Ælfric has none, for example unwisdom ([13] and [15], in each case as part of a doublet), or unsoðe ([8], again as part of a doublet). Such nominal and adjectival compounds are a particular hallmark of Wulfstan's characteristic style, as is shown below. The change from gehatene ([2] and [13]) to genamode is again characteristic of Wulfstan; it is surely significant that the word gehaten occurs only four times in the whole of Wulfstan's extant corpus, all four times in the sermon “De falsis deis” (no. 12), adapted from Ælfric, for whom it is the normal form.¹⁹ Again, the variant Drihtne for hælende Criste [3] is significant; Wulfstan never uses hælend, whilst Drihten occurs more than forty times.²⁰ One final point should be made: often Wulfstan tinkers with his own prose. We shall investigate this in detail in a moment, but some notion of this revising of his own work can be gained by comparing, for example, [16] and [17], which are structurally identical. Indeed Wulfstan has ‘ironed out’ some irregularities in Ælfric's presentation, adding his own and and þe of Godes agenre gyfe cymð, and removing the ‘extra’ words his underþeoddum. But in the second half of each sentence, Wulfstan makes slight and quixotic changes to his own words, varying gedeð þæt gyt wyrse is [16] with gemacað þæt gyt wyrse is [17], and se man þurh licetende hiwunge deð [16] with se man þurh lease hiwunge deð [17]. The making of such slight but insistent changes to his own words is again the hallmark of Wulfstan's technique, as we shall see.

It will be appropriate at this point to attempt some description of Wulfstan's characteristic style. Detailed studies have already been made by Kinard, Jost, McIntosh and Bethurum;²¹ here I simply offer a table [not reproduced] detailing the distribution of some of the most popular words and phrases used by Wulfstan. The table refers to the collection of vernacular sermons edited by Bethurum. If we count variant versions of each of the twenty-one extant sermons (signified here by Bethurum's notation a b c), we shall be considering twenty-five compositions, and a total of just over 28,000 words. When we remember that the three variant versions of the Sermo Lupi (20a, 20b, 20c), constitute very nearly a fifth of this total, and that the remaining sermons average little over 1,000 words each, the tiny size of Wulfstan's extant sermon corpus becomes clear. The table … demonstrates Wulfstan's extreme and often-noted tendency to recycle his vocabulary and to re-use the same words and phrases repeatedly.²² Such repetition of vocabulary is notoriously difficult to measure empirically in any meaningful sense; but by the simple manipulation of raw figures it can be calculated that a single form in Wulfstan's sermon corpus occurs on average no fewer than 6.25 times.²³ This is a figure roughly three times higher than that for any corpus of Old English prose that I have analysed, including Ælfrician homilies, Ælfrician saints' lives, the Blickling homilies or the Vercelli homilies; in all Old English literature this rate of repetition is only approached by the verse, with its traditional vocabulary, and even here Wulfstan's figure remains remarkably high. Comparable figures from some of the longer OE poems highlight this fact. Individual forms in Beowulf recur on average 3.12 times (half as frequently as in Wulfstan's sermons), in Daniel 2.62 times, in Andreas 2.7 times.²⁴ To repeat: the essence of Wulfstan's technique is repetition.

… A further characteristic of Wulfstan's style is his repeated fondness for alliterative doublets, many of which have the added adornment of rhyme; some of these we have already seen, and I offer a small selection of the more striking of these below. There are on average roughly ten such doublets in every single sermon.

(a) lagum & larum
                    mærð & myrhð
                    magan & motan
                    wide & side
                    wordes & weorces
(b) blisse & lisse (19, line 81)
                    to hæbbenne & to healdenne (9, line 51)
                    lufian & læran (10a, line 44)
                    saca & wraca (12, line 59 (19, line 61))
                    sacu & clacu (5, line 103)
                    sorgung & sargung (13, line 85)
                    stalu & cwalu (20a, line 52; 20b, line 65; 20c, line 57)
                    wanung & granung (7, lines 153 and 156 (3, line 67))
                    to wlance & ealles to rance (5, line 20)
(c) bewarian & bewerian (16b, line 34)
                    bidde & beodde (8c, line 145)
                    halignesse & gesceadwisnesse (8c, line 62)
                    þæt he heolde & rihtlice weolde (17, line 29)
                    riht lufode & unriht ascunode (21, line 22)
                    swicole & swæslice ficole (5, line 22)
                    wærne & wisne (9, line 75)

Here I have divided these doublets into three representative groups, to give some notion of Wulfstan's usage. In group (a) are found common doublets, which occur at least five times in Wulfstan's sermons and also frequently elsewhere, for example in the Laws, or in the works of other homilists. Olszewsca has demonstrated the vitality and vigour of such self-alliterating doublets in various Germanic languages, by highlighting just how many survive into later Norse sources, and into such later English works as the Ormulum.²⁵ The primary route for the transmission of such self-alliterating phrases is surely oral and traditional, and Wulfstan's remarkably high usage of such forms surely reflects his debt to that native tradition; certainly Wulfstan frequently adds such doublets to his renderings of Latin prose. The second group, (b), represents phrases which occur only rarely in Wulfstan's sermons, but which are found outside his corpus, whilst group (c) represents the numerous examples of one-off phrases based on the pattern which Wulfstan employs; both groups attest to Wulfstan's predilection for this native turn of phrase, to an extent unparalleled by other authors of Old English prose. In a somewhat similar vein, we have already seen how in his reworking of Ælfric's prose Wulfstan is idiosyncratically fond of nominal and adjectival compounds in un-, again an essentially native form.²⁶ Here too Wulfstan seems not so much an innovator as one who is working within an active native tradition. In Wulfstan's sermons such compounds are remarkable not so much for their form as their frequency; in the twenty sermons which contain these compounds, they occur on average nearly ten times.

Several other characteristics of Wulfstan's prose are highlighted …, for example Wulfstan's predilection for the devil and Antichrist, who make an appearance in practically every sermon. Nor are these walk-on parts. In those sermons where the devil is mentioned, his name is invoked more than nine times on average; throughout Wulfstan's sermons (and in strictly numerical terms) the devil is more than twice as popular as Christ (though only a quarter as celebrated as God)!²⁷

Again, many of Wulfstan's sermons follow the same general structural pattern; no less than twenty begin with the address to Leofan men, a phrase once thought to be exclusive to Wulfstan.²⁸ And if many sermons begin in idiosyncratic fashion, a similar number wind up in the same way. Wulfstan's favourite method of conclusion is, characteristically, a rousing exhortation, often including the phrase utan don swa us þearf is. The longest version of the Sermo Lupi concludes with a series of no less than six such exhortations (“Serm. 20c,” lines 174-202). The same theatrical and dramatic strain lies behind fairly frequent use of exclamations such as eala, la, or wala, which punctuate no less than sixteen of the twenty-five sermons under discussion.²⁹ Other authors certainly use such words and phrases, but none to my knowledge employs them with such relentless consistency. Of more relevance to my later argument is Wulfstan's use of stock phrases such as riht geleafa. The word geleafa occurs throughout the sermons some fifty-four times, but on all but eight occasions is preceded by the adjective riht. More striking still is the fact that in four of the eight cases where the collocation riht geleafa is avoided Wulfstan still employs a similar adjective-noun pair (soð geleafa or anræd geleafa), and the ‘correct’ combination riht geleafa is found in close proximity to the new alternative.³⁰ It appears that in such cases Wulfstan is not so much interested in orthodoxy as in stylistic expediency; here and in a number of other cases Wulfstan's prose seems to exemplify the concept of the fixed epithet so beloved by scholars of the oral-formulaic school.³¹

… [There exists] a tightness and consistency to Wulfstan's style which cannot easily be paralleled elsewhere in Old English prose; a similarly consistent list with respect to (say) Ælfric or the homilies in the Blickling and Vercelli collections, for example, would, I suggest, be difficult if not impossible to construct.

The repetition of key words and phrases throughout his corpus is therefore the main hallmark of Wulfstan's style. The overwhelming extent to which this repetition at all levels of discourse forms the key to Wulfstan's technique can be illustrated by considering the repetition of vocabulary and phraseology to be found in any single sermon. I have selected the sermon entitled “De Anticristo” (“serm. 1b”) for ease of reference, since it is short enough to be considered complete (it takes up some forty lines in Bethurum's edition), and has links to an identifiable Latin source which itself draws on the works of Adso, Augustine, Gregory, Isidore and the Vulgate (Bethurum's no. 1a).³² Once again, however, Wulfstan's Old English is far from simple translation. In the numbered list below I have highlighted all repetitions of vocabulary and phraseology which I have detected. Those phrases of two or more words which are repeated in a similar context elsewhere in Wulfstan's sermon-corpus are marked in italics.

[1] Leofan men, understandað swyðe georne þæt ge rihtlice & wærlice þæt healdan þæt eow mæst þearf is to gehealdenne, þæt is, rihtne cristendom.

[2] Forðam ælc þæra þe ongean þæt to swyðe deð oððon oðerne ongean þæt læreð þe his cristendome to gebyreð, ælc þæra bið Antecrist genamod.

[3] ANTICRISTUS is on Læden CONTRARIUS CHRISTO, þæt is on Englisc, Godes wiðersaca.

[4] Se bið Godes wiðersaca þe Godes lage & lare forlæt, & ðurh deofles lare of ðam deð ðe his cristendome to gebyreð, & on synnum hine sylfne to swyðe befyleð oððon oðerne man on synna belædeð.

[5] And ðeah þæt sy þæt fela manna Antecrist sylfne næfre his eagum ne geseð, to fela is þeah his lima þe man wide nu geseon & ðurh heora yfel gecnawan mæg, ealswa hit on þam godspelle geræd is: Surgent Enim Pseudochristi, Et Reliqua.

[6] Wide hit gewyrð þæt up arisað lease leogeras & beoð swæslice swicole, & ða mænigne man amyrrað & on gedwylde gebringað.

[7] And swa mycel earfoðnes gewyrð on mænige wisan gyt wide on worulde, þæs þe bec secgað, þurh deofles bearn þe unriht dreogað, swa næfre ær on worulde ne gewearð; forðam þæt mæste yfel cymð to mannum þonne Antecrist sylf cymð, þe næfre ær on worulde ne gewearð.

[8] And us þincð þæt hi sy þam timan swyðe gehende, forðam þeos woruld is fram dæge to dæge a swa leng swa wyrse.

[9] Nu is mycel neod eac eallum Godes bydelum þæt hy Godes folc warnian gelome wið þone egesan þe mannum is towerd, þe læs þe by unwære wurðan aredode & ðonne to hrædlice ðurh deofol beswicene.

[10] Ac do sacerda gehwylc on his scriftscire þæt hit man gehyre oft & gelome, þe læs ðe hit geweorðe þæt þurh larleste Godes folc losie.

[11] & ðeah þæt geweorðe þæt ure ænig þe nu leofað þonne ne libbe, þeah we agan þearfe þæt we godcunde heorda warnian nu georne hu hy þam deofle Antecriste sylfan wærlicast magan þonne wiðstandan, þonne he his wodscinn widdast tobrædeð.

[12] And utan warnian us eac swa wið his unlara nu swyðe georne & God ælmihtigne georne biddan þæt he us gescylde wið þæne þeodscaðan.

[13] God us gescylde wið þæne egesan, & he us geryme to ðære ecan myrhðe þe þam is gegearwod þe his willan gewyrcað.

[14] Dær is ece blis & æfre byð in ealra worulda woruld a butan ende, amen.

(“Serm. 1b,” “De Anticristo”)

The extent of the repetition is clear; every single sentence contains formulae and phrases found elsewhere in Wulfstan's sermons.³³ Much that remains unparalleled has the stamp of Wulfstan still. I cannot find the exact combination rihtlice & wærlice [1] anywhere else in Wulfstan's sermon-corpus, but it is strongly reminiscent of Wulfstan's fondness for other such doublets, and the phrase Antecrist genamod [2] has the same ring; it was shown above that Wulfstan twice altered Ælfric's gehatene to genamod, and one might invoke further parallel phrases found elsewhere.³⁴ As for the phrase swæslice swicole [6], which cannot be paralleled in exactly that form, one might refer back to the list of rhyming and alliterative doublets found in (a), (b) and (c) on pp. 248-9, where the phrase swicole & swæslice ficole (5, line 22) is found. It is hard to parallel Wulfstan's tendency to recycle his phraseology to such an extent in the works of any other Old English prose authors; such formulaic repetition, characteristic of much of the vernacular verse, has long been thought a mark of an oral and traditional style.³⁵

It is important to stress, however, that this formulaic repetition mostly concerns phrases of five words or less, and predominantly in the short two and three-stress phrases of limited syntactical form that McIntosh demonstrated to be the basis of the characteristic rhythm of Wulfstan's prose.³⁶ Indeed subsequent analyses by such scholars as Hollowell, Sheets and Kubouchi have served to underline the great amount of syntactical and rhythmical repetition to be found in Wulfstan's phrase and sentence-structure.³⁷ It is crucial to recognize that in longer strings of words and phrases Wulfstan appears to take great care to vary the form of his words, occasionally introducing or removing still further formulae. We can see this by considering some of the larger thematic parallels to this short sermon, which can be laid out as follows:

[1] Leofan menn, understandað swyðe georne þæt ge æfre habban rihtne geleafan on ænne ælmihtigne God.

(7, line 26)

[8] And þæt is gesyne, þy is ðeos woruld fram dæge to dæge wyrse & wyrse.

(4, line 78)

Antecristes tima is wel gehende, & ðy hit is on worulde a swa leng swa wacre.

(5, line 46)

[12] Ac utan warnian us georne & gearnian to Gode þæt he us gescylde swa his willa sy.

(4, line 86)

Þonne age we mycle þearfe þæt we God ælmihtigne georne biddan þæt he us gescylde wið þæne egesan & us gestrangie swa his willa sy.

(4, line 68)

[14] Dær is ece blis & æfre byð in ealra worulda woruld a butan ende, amen.

(5, line 119)

From the examples given it will be clear that Wulfstan has not adopted any of these parallels wholesale, but has combined and adapted phrases from a number of sources in his reworking of these formulae. Indeed in the whole of the sermon “De Anticristo,” only sentence [14] can be matched elsewhere verbatim. Once again it appears that Wulfstan is more concerned with varying his own formulaic phrasing within a large fixed framework than with simply repeating large sections of his work unchanged. This willingness to make continuous alterations within what appears to be an oral tradition surely suggests that such a tradition was still vigorous in eleventh-century England. Further evidence to support such a suggestion is offered below.

It was noted earlier that Wulfstan tends to begin and end his sermons in set ways, and it is interesting to see the way in which Wulfstan varies these formulaic passages in particular; parallel endings from eight other sermons can be compared with that seen here in “Serm. 1b,” as follows:

Him sy lof & wuldor aa butan ende, amen.

(2, line 72)

Him symle sy lof & wuldor in ealra worulda woruld a butan ende, amen.

(3, line 79)

þonne geearnige we us ece blisse æt ðam sylfum Gode þe leofað & rixað a butan ende, amen.

(6, line 216)

þe geearnian wile ece myrhðe æt ðam soðan Gode, þe leofað & rixað in ealra worulda woruld a butan ende, amen.

(7, line 23)

þonne geearnige we þæt we habban motan ece lif on ðam toweardum life & ece blisse mid Gode sylfum a butan ende, amen.

(7, line 172)

A sy ecum Gode lof & wyrðmynt in ealra woruld æfre butan ende, amen.

(10c, line 202)

Him symle sy lof & weorðmynt in ealra worulda woruld a butan ende, amen.

(12, line 93)

he geearnað him sylfum witodlice dom þæne betstan æt þam þe on heofenum leofað & ricxað in ealra worolda worold a butan ende, amen.

(18, line 146)

Him simble sy lof & wuldor in ealra worulda woruld a butan ende, amen.

(20a, line 130)

In these cases also, it will be seen that only two sermons (3 and 20a) have exactly identical endings, but that in all the other cases Wulfstan has apparently deliberately varied his phraseology, by mixing and matching the formulae available in his verbal armoury. It might further be noted that, according to this list of parallels, “Serm. 7” (“De fide catholica”) has two separate ‘endings’, at lines 23 and 172. Moreover, immediately after the first of these endings is found a formulaic collection of phrases which matches closely the beginning of a number of other sermons; indeed I have cited this ‘beginning’ as a parallel to the opening of “Serm. 1b.” As printed by Bethurum, “Serm. 7” opens with yet a further formulaic opening, again paralleling the ‘beginning’ found at 7, line 26, and at the begining of 1b, and reading as follows:

Leofan men, doð swa eow mycel þearf is, understandað þæt ælc cristen man ah micle þearfe þæt he his cristendomes gescead wite, & þæt he cunne rihtne geleafan rihtlice understandan.

Purely on the basis of these stylistic and structural parallels, therefore, I should suggest that what Bethurum prints as a single sermon (her “Serm. 7”) is in fact two works, both by Wulfstan, comprising lines 1-25 and 26-179. The further observation that the shorter of these works appears to mirror in microcosm the structure of the larger tends to support such a conclusion; moreover, in two of the three manuscripts containing Bethurum's “Serm. 7” there is a clear break after line 25, with the respective rubrics Item Sermo and Sermo.³⁸

Larger formulaic themes than those examined so far are indeed found in Wulfstan's works, and often. But what should be stressed is that in every case verbatim repetition of the complete theme is studiously avoided, and by recourse to other formulaic diction Wulfstan always strives to vary his themes within the larger constraints of his formulaic technique. This can be illustrated by considering but one example of such a repeated theme, in this case Wulfstan's threefold repetition of his description of Hell:

Dær is ece bryne grimme gemencged, & ðær is ece gryre; þær is granung & wanung & aa singal heof; þær is ealra yrmþa gehwylc & ealra deofla geþring. Wa þam þe þær sceal wunian on wite. Betere him wære þæt he man nære æfre geworden þonne he gewurde. Forðam nis se man on life þe areccan mæge ealle þa yrmða þe se gebidan sceal se ðe on þa wita ealles behreoseð; & hit is ealles þe wyrse þe his ænig ende ne cymð æfre to worulde.

(“Serm. 3,” lines 66-73)

Dær is ece bryne grimme gemencged, & ðær is ece gryre; þær is wanung & granung & a singal sorh. Wa þam þe þær sceal wunian on wite. Him wære betere þæt he æfre on worulde man ne gewurde þonne he gewurde. Nis se man on life þe areccan mæge ealle þa yrmða þe se gebidan sceal se ðe on ða witu ealles behreoseð; & hit is ealles þe wyrse þe his ænig ende ne cymð æfre to worulde.

(“Serm. 7,” lines 122-8)

Dær is ece bryne grimme gemencged, & ðær is ece gryre; ðær is ece æce, & ðær is sorgung & sargung, & a singal heof; ðær is benda bite & dynta dyne; ðær is wyrma slite & ealra wædla gripe; ðær is wanung & granung; ðær is yrmða gehwylc & ealra deofla geðring. Wa ðam ðe ðær sceal wunian on wite. Betere him wære ðæt he man nære æfre geworden ðonne he gewurde. Nis se man on life ðe areccan mæge ealle ða yrmða ðe se gebidan sceal se ðe on ða witu ealles behreoseð; & hit is ealles ðe wyrse ðe his ænig ende ne cymð æfre to worolde.

(“Serm. 13,” lines 84-92; addition from CCCC 419)

I have deliberately selected passages from three sermons from three different categories of address, in Bethurum's reckoning. “Serm. 3” (“Secundum Lucam”) is from her section on eschatology (“Serm. 1-5”), “Serm. 7” (“De fide catholica”), from that on the Christian faith (“Serm. 6-12”), “Serm. 13” (“Sermo ad populum”) from that on archiepiscopal functions (“Serm. 13-18”). Now there is clearly some relationship between these passages, which I present in the order in which Bethurum believes them to have been composed.³⁹ But this relationship is not linear, in the sense that no one passage is clearly the model for any other, and no development of ideas or phraseology can be discerned. Each passage contains both additions and omissions (and variants) with respect to each of the other passages. Furthermore each passage occurs in the context of three quite different sermons on (as Bethurum's categories show) three quite different topics. I suggest that we can best consider this repetition as three quite separate treatments of the same theme.

Now by chance this same theme occurs in three further anonymous homiletic tracts, and here, crucially, the idiosyncrasy of Wulfstan's technique can be underlined. A passage in the anonymous In letania maiore repeats the theme from Wulfstan's “Serm. 3” absolutely verbatim, and is clearly borrowed from the single version of the theme found in this particular sermon, and not from either of the other two.⁴⁰ Another passage from the anonymous De temporibus Antichristi (Napier XLII) repeats the theme from Wulfstan's “Serm. 7” absolutely verbatim, and can likewise be linked specifically to one particular version of the theme.⁴¹ Yet another passage from the anonymous Sunnandæges spell (Napier XLIII) repeats the theme from Wulfstan's “Serm. 13,” as it is represented in the single manuscript Cambridge, Corpus Christi College 419 absolutely verbatim, and can also therefore be said to have drawn on one particular version of this theme found in Wulfstan's works in preference to the other two.⁴² In the last case cited it is interesting to consider that not only has CCCC 419 been described by Bethurum as representing ‘one of the two archetypes of the extant versions’, but that this manuscript actually contains the homily Napier XLIII which borrows its reading, as well as also containing both Wulfstan “Serm. 7” and its companion Napier XLII.⁴³ The difference between Wulfstan's technique and that of other, later authors could not be made more clear. Wulfstan alters and adapts, according to the tenets of his own characteristic style, whilst others simple parrot.

Many, if not most, eleventh-century homilies are almost wholly composed by such verbatim repetition of other works, and there are detailed studies of a number of the composite homilies produced by these cut-and-paste methods by, for example, Godden and Scragg.⁴⁴ It is in every sense a literary technique, and relies on close and sometimes slavish reading of the source text. But it is not, as we have seen, Wulfstan's way; we have seen Wulfstan make minor but characteristic alterations to the Latin text of the Bible and to the vernacular texts by both Ælfric and himself: there always remains a distance and a difference between Wulfstan and his source.

At a level greater than the theme, it is important to note that several of Wulfstan's complete sermons seem to have been revised in similar ways, most importantly the Sermo Lupi itself, which exists in at least three separate versions of quite different length, in five different manuscripts; scholars have been unable to agree about the relative chronology of the three versions.⁴⁵ It may be significant that what Bethurum prints as her “Serm. 17” is found in two overlapping halves ten pages apart in the same manuscript Oxford, Bodleian Library, Hatton 113, whilst the brief “Serm. 21,” which comprises only thirty-four lines in the printed edition, exists in two quite separate versions in the same manuscript, the electic Cambridge, Corpus Christi College 201.⁴⁶ In the case of a number of sermons, including the three just mentioned, there seems, moreover, no good reason to doubt that much extraneous material which Bethurum relegates to the critical apparatus is not the authentic work of Wulfstan, the victim, perhaps, of a modern, later and literary insistence on a single original and fixed text. There is, then, much evidence to suggest that Wulfstan's reworking of both his sources and his own work extended to complete sermons also.

We have seen how Wulfstan employs rhyme, assonance and (above all) alliteration to an extent hard to match in Old English prose. We have seen how repeated phrases form the vast proportion of his work, and how he constantly re-uses what one might call characteristic formulae, again to an extent unparalleled in vernacular prose. We have seen how his reliance on a remarkably small number of syntactical and (as McIntosh has demonstrated) rhythmical patterns provides the framework into which he can mix and match his formulaic phrasing. We have seen how Wulfstan will repeat and rephrase large sections of his corpus, and rework familiar themes in similar ways in different contexts. And we have seen how at times Wulfstan may repeat whole sermons with little variation to suit separate occasions. Wulfstan's repetition can therefore be said to operate at five levels of discourse, namely repetition of sounds and individual words, repetition of formulaic phrases, repetition of sentences and sentence structures, repetition of themes and paragraphs, and repetition of entire compositions.

But exactly these five features are considered by Peabody in his five tests for what he calls oral style, but which I would prefer to designate as oral-traditional style, tests which absorb and supersede the classic work of Parry and Lord.⁴⁷ Creed has argued that these five tests produce positive results when applied to Beowulf, whilst elsewhere I have demonstrated that in the seventh century Aldhelm, uniquely amongst Anglo-Latin authors, seems to have been applying the same traditional techniques of composition in his Latin hexameters.⁴⁸ I suggest that we set Wulfstan's name alongside that of Aldhelm and the Beowulf-poet, all three providing examples of (as I take it) literate Anglo-Saxons who chose to compose in the traditional oral style of vernacular verse. It was inevitably a retrospective style, and the more modern literate and literary methods of authors like Bede and Alcuin and Ælfric would inevitably win the day, bringing with them, I note, the scribblers and copyists and scissors-and-paste scholars of the eleventh-century homilies, who in their combining and copying of so many written sources may even perhaps be mimicking the traditional style they cannot master.

I entitled this paper ‘Crying wolf’ because, like the boy in the story, Wulfstan repeats himself so often that some have ceased to listen. But in the story those who would not listen paid a heavy price. The price we pay is in understanding: Wulfstan may indeed have been fighting a quixotic literary battle against forces he could not defeat. Not to be heard may in the end prove Wulfstan's fate; not to hear him is our folly.⁴⁹

Notes

1. Wulfstan's sermons are ed. D. Bethurum, The Homilies of Wulfstan (Oxford, 1957). This edition supersedes that of A. S. Napier, Wulfstan. Sammlung der ihm zugeschriebenen Homilien nebst Untersuchungen über ihre Echtheit I. Text und Varianten (Berlin, 1883), which, however, contains a number of other anonymous sermons considered below. I have restricted my discussion by taking Bethurum's edition to represent the corpus of Wulfstan's sermons in order to avoid unnecessary complications. There are important studies of Wulfstan's sermon style and his historical context by, for example, D. Whitelock, ‘Archbishop Wulfstan, Homilist and Statesman’, TRHS 24 (1942), 25-45; A. McIntosh, ‘Wulfstan's Prose’, PBA 35 (1949), 109-42; K. Jost, Wulfstanstudien (Bern, 1950); D. Bethurum, ‘Wulfstan’, in Continuations and Beginnings: Studies in Old English Literature, ed. E. G. Stanley (London, 1966), pp. 210-46; M. McC. Gatch, Preaching and Theology in Anglo-Saxon England: Ælfric and Wulfstan (Toronto, 1977). The contemporary comment quoted here is found in a Latin letter to Wulfstan ptd Bethurum, The Homilies of Wulfstan, p. 377, and trans. Bethurum, ‘Wulfstan’, p. 211.

2. Bethurum, The Homilies of Wulfstan, pp. 88-91. For doubts on Bethurum's conclusions, see J. J. Campbell, ‘Classical Rhetoric in Old English Literature’, in Medieval Eloquence: Studies in the Theory and Practice of Medieval Rhetoric, ed. J. J. Murphy (Berkeley, CA, and London, 1978), pp. 173-97, esp. 187-9.

3. Sermo Lupi ad Anglos, ed. D. Whitelock, 3rd ed. (London, 1963), p. 37.

4. D. Whitelock, ‘Two Notes on Ælfric and Wulfstan (ii): Gildas, Alcuin and Wulfstan’, MLR [Modern Language Review] 38 (1943), 125-6; idem, Sermo Lupi, pp. 65-6; D. Bethurum, ‘Archbishop Wulfstan's Commonplace Book’, PMLA 57 (1942), 916-29.

5. Whitelock, ‘Two Notes’, p. 125.

6. Whitelock, Sermo Lupi, p. 35, n. 3.

7. J. E. Cross and A. Brown, ‘Literary Impetus for Wulfstan's Sermo Lupi’, Leeds Stud. in English n.s. 20 (1989), 271-91.

8. Bethurum, The Homilies of Wulfstan, p. 95.

9. Ibid. p. 97.

10. The texts given are as follows: Ælfric 1 is the version edited by B. Thorpe, The Homilies of the Anglo-Saxon Church, 2 vols. (London, 1844-6) I, 258; Ælfric 2 is the version ed. Thorpe, ibid. II, 596; the Wulfstan version is from Serm. 7a, lines 8-13.

11. A. S. Cook, ‘The Evolution of the Lord's Prayer in English’, Amer. Jnl of Philol. 12 (1891). 59-66.

12. For gemiltsian, see Serm. 7, line 138; 14, line 52 and 18, line 31; for miltsian, see Serm. 10c, lines 168 and 191 and 14, line 43; for foster, see Serm. 20a, line 39; for forgyfan, see Serm. 7, line 138 (as a doublet in gemildsian & mycel forgyfan). On gylt, which never occurs in Wulfstan's works, see further H. Gneuss, ‘On the Origin of Standard Old English’, ASE [Anglo-Saxon England] 1 (1972), 63-83, especially 78, and Jost, Wulfstanstudien, p. 156.

13. See The Exeter Book, ASPR 3, 223 (The Lord's Prayer I); The Anglo-Saxon Minor Poems, ASPR 6, 70 (The Lord's Prayer II) and 77 (The Lord's Prayer III). The fact that the last two of these verse renderings are found in manuscripts of the Old English prose translation of the Benedictine Office now assigned to Wulfstan has led naturally to the speculation that the verses, like the prose, are Wulfstan's work. For a rebuttal of such a theory, see J. M. Ure, The Benedictine Office: an Old English Text, Edinburgh Univ. Publ. in Lang. and Lit. 11 (Edinburgh, 1957), 25 and 44. Ure's arguments that a hand other than Wulfstan's was responsible for these verse renderings can be strengthened considerably by comparing their diction with Wulfstan's and Ælfric's as shown above, on pp. 241-2: in every relevant case their wording agrees with the Ælfrician versions against Wulfstan.

14. Paris Psalter 79.5, 114.5, 115.6 and 117.23; Meters of Boethius 18, 19 and 20; see Paris Psalter and Boethius, ASPR 5, 46, 99-100, 103 and 175-7 respectively.

15. On the literary relations between Ælfric and Wulfstan, see particularly P. Clemoes, ‘The Old English Benedictine Office, Corpus Christi College, Cambridge, MS 190, and the Relations between Ælfric and Wulfstan: A Reconsideration’, Anglia 78 (1960), 265-83, and, more recently, H. Ogawa, ‘Revised Syntax in Wulfstan's Rewritings of Ælfric's Prose’, Aspects of English Linguistics: in Memory of Professor Saburo Ohe (Kyushu, 1989), pp. 3-17. Further comparisons of the two texts considered below are given by Bethurum, The Homilies of Wulfstan, pp. 304-6; McIntosh, ‘Wulfstan's Prose’, pp. 121-2; and especially the doctoral dissertation by G. O. Zimmermann, ‘Die beiden Fassungen des dem Abt. Aelfric zugeschreibenen angelsächsischen Traktats uber die siebenfaltige Gabe des Heiligen Geistes’ (Leipzig, 1888).

16. McIntosh, ‘Wulfstan's Prose’, p. 121. The dangers of interpreting Wulfstan's characteristic and compulsive reworking of Ælfric's prose too literally are well illustrated by A. G. Kennedy, ‘Cnut's Law Code of 1018’, ASE 11 (1982), 57-81, at 67, who quotes Wulfstan's recasting in the second Old English letter which Ælfric sent him of Ne drincan æt wynhuse, ne druncengeorn beon to Ne drincan æt winhusum ealles to gelome, ne to druncangeorn wurðan (from Die Hirtenbriefe Ælfrics, ed. B. Fehr (Hamburg 1914; repr. with supplement by P. Clemoes, Darmstadt, 1966), p. 134).

17. See further Bethurum, The Homilies of Wulfstan, pp. 304-6.

18. Cf. further (a), (b) and (c) below, pp. 248-9.

19. The word gehaten occurs in Wulfstan's sermons only at Serm. 12, lines 41, 45, 65 and 72.

20. Wulfstan's preference for drihten over hælend, unique amongst eleventh-century prose authors, can be matched in both earlier prose and, perhaps more significantly, in Old English verse. By reference to A Concordance to the Anglo-Saxon Poetic Records, ed. J. B. Bessinger, Jr and P. H. Smith (Ithaca, NY, 1978), it can be calculated that throughout the extant corpus of Old English poetry the word hælend and its declensional forms occur seventy-two times in all, whilst the word drihten, used both in a secular and (predominantly) religious sense, and its declensional forms is found no less than 1,139 times.

21. J. P. Kinard, A Study of Wulfstan's Homilies: their Style and Sources (Baltimore, MD, 1897); Jost, Wulfstanstudien; McIntosh, ‘Wulfstan's Prose’; Bethurum, The Homilies of Wulfstan, esp. pp. 87-98.

22. Cf. the comments of Kinard, A Study of Wulfstan's Homilies, pp. 19-20; Jost, Wulfstanstudien, pp. 159-76; Bethurum, The Homilies of Wulfstan, pp. 89-90.

23. I count 28,676 words in Wulfstan's extant sermon corpus (as edited by Bethurum), and a vocabulary of 4,608 separate forms. If we exclude variants caused through accidence, it is remarkable that Wulfstan's active vocabulary throughout these twenty-five sermons is something less than 4,000 words.

24. I count 7,365 words in Beowulf, and 2,360 separate forms; 4,467 words in Daniel, and 1,708 separate forms; 9,279 words in Andreas, and 3,438 separate forms.

25. E. S. Olszewska, ‘Alliterative Phrases in the Ormulum: some Norse Parallels’, in English and Medieval Studies Presented to J.R.R. Tolkien on the Occasion of his Seventieth Birthday, ed. N. Davis and C. L. Wrenn (London, 1962), pp. 112-27.

26. Examples of such nominal and adjectival compounds in un- favoured by Wulfstan include: unclænness; uncoþu; uncræft; undæd; ungelimp; ungerim; ungesælig; ungetreow; ungifu; ungeþanc; ungod; ungyld; unhal; unlagu; unlar; unlytel; unnyt; unræd; unriht; unrot; unsælig; unþeaw; unwær; unwæstm; unweder; unwill; and unwisdom.

27. Throughout the sermons, deofol and Antecrist are invoked 151 times; Crist only sixty-eight times; God 591 times.

28. Cf. the remarks of C.I.J.M. Stuart, ‘Wulfstan's Use of leofan men’, ES [English Studies] 45 (1964), 39-42.

29. The interjection eala is by far the commonest, occurring no fewer than twenty-four times in fifteen sermons.

30. So we find anrædne geleafan (4, line 88) and on rihtan geleafan (4, line 91); rihtne geleafan (7, line 21) and soðan geleafan (7, line 22); anrædne geleafan (7a, line 18) and rihtne geleafan (7a, line 19); anrædne geleafan (8c, line 15) and rihtne geleafan (8c, line 19).

31. The classic exposition of the concept of the fixed epithet is still that of M. Parry, ‘Studies in the Epic Technique of Oral Verse-Making I. Homer and Homeric Style’, Harvard Stud. in Classical Philol. 41 (1930), 73-147; idem, ‘Studies in the Epic Technique of Oral Verse-Making II. Homeric Language as the Language of an Oral Poetry’, Harvard Stud. in Classical Philol. 43 (1932), 1-50. Parry's views on the fixed epithet have been re-examined most recently by D. M. Shive, Naming Achilles (Oxford, 1989). The astonishing proliferation of oral-formulaic studies in a large number of literatures can be gauged by the score or more of bibliographies published; see now J. M. Foley, Oral Formulaic Theory and Research: an Introduction and Annotated Bibliography (New York, 1985), which contains a useful historical overview of the subject on pp. 11-77.

32. Bethurum, The Homilies of Wulfstan, pp. 282-6; this same sermon is analysed from a different perspective and translated by Gatch, Preaching and Theology in Anglo-Saxon England, pp. 106-7 and 236-7.

33. For parallels, see particularly the following: [1] 7, line 26; 8b, line 85; 13, line 65; 17, line 34. [2] 8c, line 40; 10c, line 45. [3] 20a, line 99; 20b, line 140. [4] 5, line 115; 20a, line 44; 20b, line 57; 20c, lines 49 and 66. [7] 4, lines 11 and 15; 6, line 203. [8] 4, line 78; 5, line 46. [9] 4, line 2; 5, line 52. [12] 4, lines 68 and 86. [13] 3, line 74; 20a, line 129; 20b, line 176; 20c, line 200. [14] 2, line 72; 3, line 79; 5, line 119; 6, line 216; 7, lines 23 and 172; 10c, line 202; 12, line 93; 18, line 146; 20a, line 130.

34. One might compare, for example, ecclesia genamod (10c, line 43; 18, line 71), Eua genamod (6, line 36), genamod Iuno (12, line 48) and Lucifer genemned (6, line 29). Jon Wilcox points out to me that the phrase rihtlice and wærlice is also found in a similar setting in the anonymous homily In letania maiore, ed. J. Bazire and J. E. Cross, Eleven Old English Rogationtide Homilies (Toronto, 1982), no. 8, lines 42-3. On the further links between this anonymous homily and Wulfstan's sermons, see below, n. 40.

35. For the classic application of this oral-formulaic theory to Old English, see F. P. Magoun, ‘The Oral-Formulaic Character of Anglo-Saxon Narrative Poetry’, Speculum 28 (1953), 446-67. The notion that in Anglo-Saxon England literate authors were also using oral-traditional methods of composition is firmly supported by L. D. Benson, ‘The Literary Character of Anglo-Saxon Formulaic Poetry’, PMLA 81 (1966), 334-41; A. P. McD. Orchard, ‘The Poetic Art of Aldhelm’ (unpubl. PhD dissertation, Cambridge Univ., 1990), pp. 76-151. Further evidence of a modern sermon-tradition by literate authors working within an oral-traditional framework is given by B. A. Rosenberg, The Art of the American Folk Preacher (Oxford, 1970) and ‘The Formulaic Quality of Spontaneous Sermons’, Jnl of Amer. Folklore 83 (1970), 3-20.

36. McIntosh, ‘Wulfstan's Prose’, pp. 114-22.

37. I. M. Hollowell, ‘On the Two-Stress Theory of Wulfstan's Rhythm’, PQ 61 (1982), 1-11; L. A. Sheets, ‘Wulfstan's Prose: a Reconsideration’ (unpubl. PhD dissertation, Ohio State Univ., 1964), esp. pp. 92-120; T. Kubouchi, ‘A Note on Prose Rhythm in Wulfstan's De Falsis Deis’, Poetica 15-16 (1983), 57-106.

38. Bethurum, The Homilies of Wulfstan, p. 158; Bethurum has here preferred the reading of Cambridge, Corpus Christi College 419 (her B) against that of both Cambridge, Corpus Christi College 201 and Oxford, Bodleian Library Hatton 113 (her C and E), despite the evidence of her own stemma, p. 11.

39. Bethurum, The Homilies of Wulfstan, pp. 101-4.

40. The anonymous In letania maiore is ed. in an Appendix to H. P. Tristram, ‘Vier altenglische Predigten aus der heterodoxen Tradition’ (unpubl. dissertation, Freiburg Univ., 1970), pp. 431-7; the passage in question occurs at lines 182-90. For a more accessible text, see Bazire and Cross, Eleven Old English Rogationtide Homilies, pp. 104-14 (lines 131-7).

41. The anonymous De temporibus Antichristi is ed. Napier, Wulfstan, as no. XLII; the passage in question occurs on p. 203, lines 13-20. Wulfstan's own authorship of this sermon is rejected by both Whitelock (Sermo Lupi, p. 14, n. 2) and Jost (Wulfstanstudien, pp. 218-21), but McIntosh (‘Wulfstan's Prose’, p. 142, n. 32) believes ‘on rhythmical evidence’ that the section from p. 202, line 4 to the end, which includes the passage in question, is ‘probably by Wulfstan’. See now, however, J. Wilcox, ‘Napier's “Wulfstan” Homilies XL and XLII: Two Anonymous Works from Winchester?’, JEGP [Journal of English and Germanic Philology] 90 (1991), 1-19.

42. The anonymous Sunnandæges spell is ed. Napier, Wulfstan, as no. XLIII; the passage in question occurs on p. 209, lines 14-25.

43. Cf. the comments of Bethurum, The Homilies of Wulfstan, pp. 1-2.

44. M. R. Godden, ‘Old English Composite Homilies from Winchester’, ASE 4 (1975), 57-65; D. G. Scragg, ‘Napier's “Wulfstan” Homily XXX: its Sources, its Relationship to the Vercelli Book and its Style’, ASE 6 (1977), 197-211. See now further Wilcox, ‘Napier's “Wulfstan” Homilies XL and XLII’.

45. On the various versions and their relationships, see Whitelock, Sermo Lupi, pp. 1-5. Both Bethurum, The Homilies of Wulfstan, pp. 22-3, and Whitelock, Sermo Lupi, pp. 2-5, believe that the shortest version (Serm. 20a) is the earliest, and that the longer versions represent successive stages in the expansion of the sermon; S. Dien, ‘Sermo Lupi ad Anglos: the Order and Date of the Three Versions’, NM [Neuphilologische Mitteilungen] 76 (1975), 561-70, on the contrary, believes that the longest version (Serm. 20c) is original, and that the other versions represent subsequent abridgments.

46. See Bethurum, The Homilies of Wulfstan, pp. 242-5, 276-7, 351-3 and 364-5.

47. B. Peabody, The Winged Word: a Study in the Technique of Ancient Greek Composition as Seen principally though Hesiod's Works and Days (Albany, NY, 1975).

48. R. P. Creed, ‘The Beowulf-Poet: Master of Sound-Patterning’, in Oral Traditional Literature: a Festschrift for Albert Bates Lord, ed. J. M. Foley (Columbus, OH, 1981), pp. 194-216; Orchard, ‘The Poetic Art of Aldhelm’, pp. 119-33.

49. I am grateful to Jon Wilcox and Malcolm Godden for their generous criticism and advice in the composition of this paper.