Selim 20.indb ISSN: 1132–631X Rafael J. Pascual, SELIM 20 (2013–2014): 49–79 THREE-POSITION VERSES AND THE METRICAL PRACTICE OF THE BEOWULF POET Abstract: This article assesses the authenticity of the three-position SS verse type in Beowulf on the basis of its unambiguous incidence both in Beowulf and in a larger corpus of Old English poetry. The fi rst part of this essay examines the metrical confi guration of thirteen verses fr om Beowulf that have recently been identifi ed as instances of the SS pattern. In doing so, it demonstrates that nearly all of them furnish a standard four-position metrical structure. The second part discusses the empirical obstacles to accepting the formal legitimacy of the three-position SS pattern in Old English verse, thereby reaffi rming the validity of the stricture of traditional Sieversian metrics against verses consisting of less than four metrical positions. Keywords: Old English Metre, Textual Criticism, Beowulf, Old English Literature, Early Germanic Poetry. Resumen: Este artículo analiza la autenticidad del verso de tres posiciones SS en Beowulf en base a su incidencia inequívoca tanto en Beowulf como en un corpus más grande de poesía inglesa antigua. La primera parte de este ensayo examina la confi guración métrica de trece versos de Beowulf que han sido identifi cados recientemente como ejemplos del patrón métrico SS. Así demuestra que casi todos ellos tienen una estructura convencional de cuatro posiciones métricas. La segunda parte discute los obstáculos empíricos que impiden aceptar la legitimidad formal del patrón de tres posiciones SS en la poesía inglesa antigua, reafi rmando así la validez de la prohibición de la métrica tradicional sieversiana contra versos de menos de cuatro posiciones métricas. Palabras clave: Métrica inglesa antigua, crítica textual, Beowulf, literatura inglesa antigua, poesía germánica inicial. Probabilistic reasoning governs the study of the metrical practice of the Beowulf poet. Considerations of relative probability enable editors and metrists to identify scribal corruptions, recover authorial readings, and understand the metrical regularities that the poet meticulously imposed upon his work.1 The role of probability in these matters can readily be illustrated by the scholarly response to the words hrēas blāc, which appear in line 15 on folio 188r of the Beowulf manuscript. At this point, Beowulf is describing Ongentheow’s death, and the transmitted text of the poem would have him state that the Swedish king hrēas 1 For in-depth discussions of the role of probabilism in Old English philology, see Fulk 2003 and 1992: §§8–23; on the balancing of metrical probabilities in textual criticism, see Fulk 1996. 50SELIM 20 (2013–2014) Rafael J. Pascual blāc, “fell pale,” aft er Eofor’s blow. If the manuscript evidence were taken at face value, hrēas blāc should constitute a verse by itself and an apparent two-position verse pattern SS should then be regarded as the genuine outcome of the poet’s metrical practice. Although the sense, syntax, and alliteration exhibited in hrēas blāc are sound, it is improbable that the Beowulf poet composed a verse of this sort. In the surviving corpus of approximately 30,000 lines of Old English poetry, verses unambiguously featuring the SS stress contour are virtually non-existent. If Old English poets considered the SS pattern an authentic verse type, we should expect to fi nd more evidence for the authenticity of this type than a few dubious attestations. To regard hrēas blāc as an authentic verse generates a gross improbability: it forces one to believe that the systematic avoidance of a legitimate metrical pattern in so large a corpus of poetry is entirely due to accident. The most probable explanation for the apparent existence of a handful of verses exhibiting the SS pattern is that these verses are the products of scribal corruption, not authorial practice. Editors of Beowulf unanimously regard hrēas blāc as a corrupt verse requiring emendation; metrists rightly conclude that this verse does not refl ect the metrical practice of the Beowulf poet. By the same token, verses exhibiting a rarely attested three- position SS pattern have traditionally been considered unmetrical and regarded as the consequences of scribal error. According to the tenets of Sieversian formalism, the rationale behind the unmetricality of the SS pattern is its failure to comply with the most basic rule of Old English metre, the four-position principle (Sievers 1885: 220–222, 270; 1893: §§8, 14.2).2 In the word-foot theory, the SS pattern corresponds to a foot, and hence it 2 The four-position theory of Old English metre is laid out in Cable 1974: 84– 93. For some qualifi cations to Cable’s original analysis, see Cable 1991: 39. For a concise summary of the four-position principle, see Stockwell & Minkova 1997: 67–69; Fulk 2002: 337–340; 2012a: 558; and Pascual forthcoming. Three-position verses and the metrical practice of the Beowulf poet 51 SELIM 20 (2013–2014) cannot stand as a verse by itself (Russom 1987: 13, 28–29).3 This traditional stance has recently been questioned by Eric Weiskott in his essay “Three-Position Verses in Beowulf ” (2013). He gathers thirteen verses that purportedly feature the three-position SS pattern fr om the poem and contends that they furnish suffi cient evidence for metrists to accept it as a genuine metrical type. Of these thirteen verses, he focuses the body of his essay exclusively on one of them, Beowulf 2150a lissa ġelong (a longstanding crux in Old English metrical studies), and relegates the remaining twelve to a list in a footnote without detailed commentary. Further, he tries to overcome the diffi culty posed by the nonconformity of the SS pattern to the four-position principle by proposing an analogy with the expanded type D verse (i.e., type D*). Since Sieversian metrics accepts type D* verses, whose metrical structure apparently fails to comply with the four-position principle, the inability of the SS pattern to conform to that principle would not constitute suffi cient grounds for being considered unmetrical. Rather, he maintains, although the poets would have perceived the SS pattern as anomalous when it was fi rst developed, it would have been reinterpreted as a regular type over the course of the history of Old English metre. The present article subjects Weiskott’s case for the authenticity of the three-position SS pattern in Beowulf to critical scrutiny. The fi rst part examines the metrical structure of the twelve verses that supposedly feature the SS pattern, which Weiskott summarily consigned to a list in a footnote. Close analysis demonstrates that these verses either genuinely feature a four-position metrical confi guration or are corrupt manuscript readings. Consequently, it becomes clear that Weiskott’s case for the authenticity of the SS type is predicated exclusively on the evidence aff orded by a single verse, lissa ġelong—an untenable position, for reasons made 3 On the word-foot theory and the explanatory power it brings to Old English metre, see Russom 1987 and 1998. 52SELIM 20 (2013–2014) Rafael J. Pascual clear below. The second part assesses his comparison of the SS pattern with type D* and the supposed reinterpretation to which the catalectic SS type would have been subjected over the course of Old English metrical history, along with other methodological issues raised by Weiskott’s essay. The conclusion is that his argumentation fails to make a convincing case for the authenticity of the SS pattern, which, judging by its virtually non-existent incidence in the surviving corpus of Old English poetry, must have been considered unmetrical throughout the Anglo-Saxon period. 1 Manuscript evidence and metrical structure 1.1 Syntactically uncommon four-position verses Due to their syntactic complexity, the two parallel gnomic verses 183b and 186b, Wā bið þǣm ðe sceal and wēl bið þǣm þe mōt,4 have presented certain diffi culties to metrists, who have sometimes scanned them as instances of the three-position SS pattern.5 This scansion is at odds, however, with Hans Kuhn’s fi rst metrico- syntactic rule, the law of Germanic sentence particles (1933: 8),6 the operation of which reveals that these two verses have a standard four-position metrical confi guration. The three-position analysis must then be disallowed on that basis, since Beowulf faithfully conforms to the regularities observed by Kuhn.7 4 “Wrong to one who must” and “well to one who is permitted.” 5 So, for example, Bliss (1967: §86), Momma (1989: 425), and Hutcheson (1995: 254). Sievers (1885: 267; 1893: §85, n. 10), Pope (1966: 371), and Getty (2002: 225), on the contrary, would scan them as atypical four-position verses with a half-stress on þǣm. 6 Summaries of Kuhn’s fi rst law can be found in Campbell (1970: 94), Lucas (1990: 294), Kendall (1991: 17–18), Hutcheson (1992: 129), Momma (1997: 56–64), Orton (1999: 289 n. 11), Fulk (see Pope 2001: 136–138; 2012a: 558–559; 2012b: 389), and Terasawa (2011: 95–96). 7 On the regular compliance of Beowulf with Kuhn’s fi rst law, see for example Lucas (1990: 294), Orton (1999: 289), and Fulk (2012b: 389–390; 2014). Three-position verses and the metrical practice of the Beowulf poet 53 SELIM 20 (2013–2014) Kuhn’s fi rst law states that all the unstressed sentence particles of a verse clause must be placed together in the fi rst drop of that clause, either directly before the fi rst lift , or between the fi rst lift and the second;8 the direct implication is that a sentence particle not found in that clausal position is stressed. For example, in Beowulf 2134b hē mē mēde ġehēt,9 which constitutes a clause by itself, the sentence particles hē and mē, two personal pronouns, should be unstressed because they appear in the fi rst drop of the clause immediately preceding the fi rst lift , the root syllable of the noun mēde. That the root syllable of mēde is the fi rst lift of the verse clause and that therefore hē and mē are unstressed is confi rmed by the participation of mēde in the alliterative scheme of the line.10 The remaining sentence particle, the fi nite verb -hēt, should then take stress, since it is found outside its prescribed place in the clause besides hē and mē. This must indeed be the case, since according to traditional Sieversian metrics no verse ends in more than one unstressed syllable (see, for example, Pope 2001: 141; and Terasawa 2011: 35). Thus, the application of Kuhn’s fi rst law reveals that the stress contour of this verse is SS, corresponding to a standard type B. The metrical contour of hē mē mēde ġehēt is transparent because that verse is syntactically simple. In Wā bið þǣm ðe sceal and wēl bið þǣm þe mōt, on the contrary, there is a verse-internal clause boundary, as is indicated by the presence of two verbs in each of them (bið and sceal in 183b, and bið and mōt in 186b). In both instances, the indirect object of the main clause, the dative pronoun þǣm, is modifi ed by a dependent relative clause. The 8 Sentence particles are semantically independent words that, unlike stress-words, usually fail to receive rhythmic stress. Particles include fi nite verbs, personal and demonstrative pronouns, demonstrative adverbs, and some conj unctions (see Pope 2001: 136–137; and Terasawa 2011: 27–28). 9 “He promised me reward.” 10 The on-verse is mǣrðo fr emede. In Old English poetry, the fi rst lift of the off - verse must participate in the alliteration of the line, while the second must not. 54SELIM 20 (2013–2014) Rafael J. Pascual occurrence of a clause boundary within a verse is an infr equent syntactic feature in Beowulf, which has obscured how Kuhn’s law of sentence particles applies in these two verses.11 As a result of this syntactic complexity, the stress contour of these two parallel verses has not always been obvious to metrists. If the position of the clause boundary within the verse is correctly established, however, Kuhn’s fi rst law can be seen to operate regularly in these two verses, which leads to the recognition of their four-position metrical confi guration. The establishment of the clause boundary can in turn be achieved by means of a comparison with, for example, Beowulf 2600b–2601b sibb’ ǣfr e ne mæġ / wiht onwendan / þām ðe wēl þenċeð.12 These three verses accommodate a sentence that also consists of a main clause with its indirect object, þām, modifi ed by a dependent relative clause, ðe wēl þenċeð. One might well suppose that the clause boundary falls between þām and the relative particle ðe. Nevertheless, the metre and the alliteration of these verses, in conj unction with Kuhn’s fi rst law, show that it falls between onwendan and þām, and that the Beowulf poet must have regarded the pronoun þām as part of the relative clause by which it is modifi ed. The alliteration of l. 2061, which is on /w/, indicates that wēl is the fi rst lift of the off -verse, so that þām and ðe, which immediately precede it, must be unstressed (like hē and mē in hē mē mēde ġehēt). This means that þām must be part of the clause- initial drop of the relative clause. If it were part of the main clause, it would receive stress for being outside its prescribed position in the fi rst drop besides the unstressed syllables -fr e and ne, thereby spoiling both the metre of the verse and the alliteration of the line. The lack of stress of þām thus indicates that it must have become associated with the relative particle ðe, both of which are 11 According to Kendall, there are forty one instances of verse-internal clause boundaries in Beowulf (1991: 89–90). 12 “Nothing can ever change ties of kinship for one who thinks rightly.” Three-position verses and the metrical practice of the Beowulf poet 55 SELIM 20 (2013–2014) treated by the Beowulf poet as an unstressed integral unit at the onset of the relative clause.13 The treatment of the particles þām and þe as a clause-initial unit in 2601b furnishes compelling evidence that the Beowulf poet must have also regarded them as such in both Wā bið þǣm ðe sceal and wēl bið þǣm þe mōt. If þǣm þe is a clause-initial unit, then the immediately preceding word, bið, must be clause-fi nal. The verse-internal clause boundary between the main clause and its dependent relative clause can then be established between bið and þǣm, which allows us to observe how Kuhn’s fi rst law operates. In regular compliance with the law, the clause-initial unit þǣm ðe must be unstressed, since it is placed in the fi rst drop of the relative clause, immediately before the fi rst lift , which in both instances is occupied by a verse-fi nal fi nite verb that has been promoted to a stressed position (sceal and mōt).14 The fi nite verb bið, being at the end of the main clause, must then receive stress, since it fails to adhere to Kuhn’s fi rst law: it is not either directly before the fi rst lift (wā and wēl), or between the fi rst lift and the second, since there is no second lift besides bið itself. As we can see, then, the workings of Kuhn’s fi rst law suggest that the two gnomic verses 183b and 186b have the stress contour of a standard type E, SsS, and that therefore they regularly comply with the four-position rule of Old English metre.15 Given the demonstrable reality of 13 The same situation can be appreciated, for example, in Beowulf 1838b–1839b feorcȳþðe bēoð / sēlran ġesōhte / þǣm þe him selfa dēah, where the alliteration of selfa indicates that þǣm þe must be unstressed and hence clause-initial. 14 Kendall’s transformational rule states that in a clause-initial segment which lacks stress-words (as in þǣm ðe sceal and þǣm þe mōt), sentence particles acquire metrical stress fr om right to left in accordance with the stress and phrase rules of Old English until the fi rst valid metrical contour emerges (1991: 96; cf. Fulk’s comment in Pope 2001: 138, n. 18). 15 This explanation is endorsed by the editors of Klaeber IV (Fulk, Bjork & Niles 2008: 129). Russom also scans these two verses as instances of the SsS pattern (1987: 120). 56SELIM 20 (2013–2014) Rafael J. Pascual Kuhn’s law of Germanic sentence particles (Donoghue 1997), the three-position scansion of 183b and 186b is untenable. These two verses possess a four-position metrical confi guration and cannot be adduced as evidence for the existence of the SS pattern in Beowulf. 1.2 Four-position verses with an unresolved lift in the coda Weiskott’s SS scansion of another three verses in Beowulf must also be rejected. The verses in question are 845a nīða ofercumen,16 954a dǣdum ġefr emed,17 and 2430b Hrēðel cyning.18 Metrists have traditionally held that these atypical verses feature an unresolved second lift ,19 and that they are therefore type A verses that comply with the four-position principle (see, for example, Pope 1966: 272; Russom 1987: 46, 51 and 117; and Fulk 1992: §§207–209).20 The three-position analysis of these verses requires us to assume that their second short stressed syllable and its unstressed successor undergo resolution. Although these three verses are exceptional under any of the two interpretations, the three-position analysis is demonstrably less probable for a number of reasons. Most saliently, it demands credence in an improbable coincidence. Verses with an unresolved lift in the coda can be found in other poems as well.21 Some examples are tempel Gode (Exodus 391b);22 16 “Overcome by violence.” 17 “Performed with deeds.” 18 “King Hrethel.” 19 These verses are atypical because a short second lift is usually preceded by a monosyllabic lift or half-lift , not a drop. See, for example, Pascual forthcoming. 20 Bliss scans 845a and 954a as instances of the SS pattern. Fulk has demonstrated, however, that Bliss’s acceptance of three-position verses is misguided (see Fulk 1992: §210). 21 The coda of the verse comprises the last full lift and all subsequent syllables. The linguistic material preceding the coda of a verse is its onset (Fulk 1992: 201, n. 60). 22 “Temple for God.” Three-position verses and the metrical practice of the Beowulf poet 57 SELIM 20 (2013–2014) on ġenimeð;23 Loth wæs āhreded;24 eorlum bedroren;25 Scēawa heofon26 (Genesis A 1209a, 2085b, 2099a, and 2191a); scride and færelt (Metres of Boethius 28.11b);27 wildne fugol (Solomon and Saturn 315a),28 among several others.29 As we can appreciate, it is the second stressed syllable that is systematically short, never the fi rst. This regular distributional pattern suggests that these verses have a four-position metrical structure with an unresolved lift in the coda of the verse. If the metrical confi guration of all these verses were SS with a resolved second lift , as the three-position analysis requires, it would be remarkable that this verse type is never realized with resolution occurring in the fi rst lift . The three-position analysis of these verses would thus compel one to believe that the absence of three- position SS verses with a resolved fi rst lift , like the hypothetical *guma ġehēt or *guma mē ġehēt,30 is accidental. It seems far more probable that the non-occurrence of verses like *guma ġehēt is an indication of their unmetricality. And since the only possible scansion for the non-occurring *guma ġehēt is SS,31 it follows that SS is not a valid metrical analysis for attested verses like nīða ofercumen. Thus, unless one is ready to give credence to extreme 23 “Away takes.” 24 “Loth was liberated.” 25 “Bereft of warriors.” 26 “Behold the sky.” 27 “Course and orbit.” 28 “Wild bird.” 29 Sievers lists twelve such verses (1885: 458). For a few more examples, see Schabram 1960. 30 “A man promised;” “a man promised me.” 31 Verse-initial resolvable sequences must necessarily undergo resolution (see Suzuki 1995: 26; Pascual forthcoming). 58SELIM 20 (2013–2014) Rafael J. Pascual coincidences, it is necessary to regard these verses as relatively atypical instances of a regular type A metrical confi guration with a short second lift . Another reason to credit the traditional four-position analysis for verses like nīða ofercumen is that it receives independent support fr om a well-known fact of all Indo-European metrical systems, namely that they tend to demand more fi xed structures toward the end of the verse.32 A clear expression of this tendency can be appreciated, for example, in the ability of non-verse-fi nal drops to accommodate a variable number of syllables, while only one unstressed syllable is allowed to occupy a verse-fi nal drop, as has been stated above.33 This characteristic feature of Indo-European metres manifests itself even more evidently in the application of Fulk’s law, according to which the metrical value of disyllabic sequences with a short penultimate syllable under tertiary stress is determined by their position within the verse.34 In the onset, they occupy a single metrical position; but if they are in the coda,35 then each of their two syllables must constitute a single metrical position on its own. Although an exhaustive explanation of exceptional verses like nīða ofercumen falls beyond the scope of the present essay, the preceding discussion should have suffi ced to make it clear that there are good metrical reasons to expect an unresolved lift in the coda of the verse. Thus, the SS scansion is not tenable for verses like nīða ofercumen, and hence they cannot be off ered in support of the authenticity of the SS verse pattern in Beowulf. 32 Foley refers to this phenomenon as “right justifi cation” (1985: 12; see also Fulk 1992: §226). 33 See p. 53. 34 Fulk names his law “Rule of the Coda” (1992: §§221–245). Notice that although “tertiary stress” is here retained as a useful concept, the application of Fulk’s law demonstrates that ictus at the tertiary level is exclusively predicated on syllable quantity (Fulk 1992: §268). 35 See fn. 21 above. Three-position verses and the metrical practice of the Beowulf poet 59 SELIM 20 (2013–2014) 1.3 Miscellaneous four-position verses Weiskott’s SS scansion for 881a ēam his nefan,36 1728a Hwīlum hē on lufan,37 736a ðicgean ofer þā niht38 and 940b dǣd ġefr emede39 must likewise be rejected. In the fi rst instance, 881a, it is probable that the poet regarded ēam as a non-contracted disyllabic word with the stress contour Ss, refl ecting prehistoric Old English *ēa-am, descended in turn fr om the Proto-Germanic compound *awa- haim (Holthausen 1963: 84, s.v. éam).40 The stress pattern of 881a would therefore be SsS, corresponding to a rhythmical type E with resolution of its second lift . Indeed, the circumfl ex diacritic above ēam printed in Klaeber IV indicates that its editors endorse this scansion. Further, the traditional four-position interpretation for this verse is predicated upon compelling philological evidence that the three-position analysis neglects. In the prehistoric Old English form *ēa-am, a hiatus separates a diphthong ending in a back vowel, ēa-, fr om an unstressed vowel, -a. Hiatuses of this kind underwent contraction at some point between the late seventh and the early eighth century (Campbell 1964: §235.2; Hogg 2011: §5.131). Since the composition of Beowulf can be reliably dated to the period 685–725,41 36 “Uncle to his nephew.” 37 “Sometimes he in delight.” 38 “Consume beyond that night.” 39 “Deed accomplished.” 40 See also Fulk, Bjork, and Niles 2008: 330, n. 3. 41 Beowulf is the only poem in the Old English corpus with regular and extensive adherence to both parts of Kaluza’s law. This means that the poet was aware of the distinction between etymologically short and long desinences that became indistinct ca. 725 in Mercia and ca. 825 in Northumbria. Because the language of Beowulf is less conservative than that of the Épinal-Erfurt glossary (ca. 685) and because dialectal indications point to Mercian composition, Fulk has concluded that Beowulf was most likely composed between ca. 685 and ca. 725 (1992: §§406– 421; 2007a: 268; 2007b: 317–323; see also Clark 2014 and Neidorf & Pascual forthcoming [2015]). The law originates in the observations of Max Kaluza (1896). 60SELIM 20 (2013–2014) Rafael J. Pascual it follows that the non-contracted form must still have been within easy reach for the poet at the time he composed Beowulf. Therefore, the four-position scansion for 881a not only is metrically regular, but also preferable on a philological basis. With respect to 1728a Hwīlum hē on lufan, the three-position analysis is based on the assumption that the entire line features a transverse alliterative scheme of the type AB : BA, which would involve promotion of hwīlum to a stressed position.42 This assumption is open to doubt, however, not only because transverse alliteration is very infr equent in Old English poetry (Terasawa 2011: 18),43 but also because it would entail a breach of Kuhn’s fi rst law, an extremely uncommon situation in Beowulf. Since hwīlum is a sentence particle that appears in the fi rst drop of its clause, in complete obedience to Kuhn’s fi rst law, it must be unstressed. Accordingly, this verse scans as an A3 type with its alliterating lift occupied by the short stressed syllable lu-.44 Remarkably, even in the improbable case that ēam scanned as a monosyllable and that hwīlum bore a stress, verses 881a and 1728a would not unambiguously feature the SS pattern, since the SS pattern with an unresolved lift in the coda would still be a more probable explanation for them, as has been argued in the previous section. Neither of these two verses therefore carries any conviction as an authentic instance of the SS pattern. The stress pattern traditionally posited for Beowulf 736a ðicgean ofer þā niht, is the four-position SSs, corresponding to a heavy type A with primary stress on þā, secondary stress on niht, and featuring double alliteration. The three-position scansion of this verse is based on the premise that þā is unstressed. This premise, 42 The off -verse is lǣteð hworfan. 43 R. B. Le Page considers that this line features transverse alliteration (1959: 435), but see the criticism raised by Terasawa (2011: 25). 44 Type A3 verses with a short lift are occasionally found in Beowulf and elsewhere (Fulk, Bjork & Niles 2008: 330). Three-position verses and the metrical practice of the Beowulf poet 61 SELIM 20 (2013–2014) however, is not supported by the evidence. There is a similar verse elsewhere in Old English poetry, Judith 306a þeġnas on ðā tīd,45 and in both instances the alliteration of the line is on /þ/.46 The three-position interpretation thus neglects the fact that in these two verses promotion of þā to a stressed position results not only in a metrically regular verse, but also in an acceptable alliterative scheme. Furthermore, the assignment of primary stress to þā and secondary stress to niht in 736a is supported by the occurrence in Beowulf and elsewhere of a signifi cant number of parallel verses in which niht takes secondary stress and is preceded by an alliterating monosyllable (Hutcheson 1995: 159, n. 3). This can be transparently appreciated in verses like Beowulf 517a seofonniht swuncon,47 Exodus 63a Hēht þā ymb twā niht or Andreas 185a Nū bið fore þrēo niht.48 Thus, in the absence of a substantial number of unambiguous instances of the SS pattern, it is not justifi able to adduce Beowulf 736a as authentic evidence of that pattern. With regard to 940b dǣd ġefr emede, this verse as it stands in the manuscript scans as a regular four-position type A verse with resolution of the second lift . In order to make it conform to the SS pattern, Weiskott assumes that the form ġefr emede is a scribal substitution of authorial ġefr emed, which would in turn undergo resolution. The emendation of ġefr emede to ġefr emed, which has never been proposed by any editor in the history of Beowulf textual criticism, has no rational basis and cannot be accepted. The manuscript reading ġefr emede is grammatically unquestionable: it is a past participle declined as an accusative singular feminine adjective, in perfect agreement with the noun it modifi es, the feminine i-stem 45 “Warriors at that time.” Griffi th scans it as a regular four-position verse with alliteration on þā (1997: 141). 46 Beowulf 736b and Judith 306b read Þrȳðswȳð behēold and þearle ġelyste respectively. 47 Disyllabic seofon- is resolved and counts therefore as a single syllable. 48 See also Fulk 1992: §199 for further commentary on verses of this sort. 62SELIM 20 (2013–2014) Rafael J. Pascual dǣd, the direct object of its clause. The single reason adduced by Weiskott in support of his emendation is that there is one verse in the whole poem, 476a, in which the past participle of ġefr emman is uninfl ected: fǣrnīða ġefr emed.49 If this line of reasoning were accepted, we would be obliged to change, for example, Beowulf 216b wudu bundenne to wudu bunden, since this is the only occurrence in the whole poem in which the participle of bindan is infl ected. But we would then be altering a metrically regular verse to a verse with the unmetrical SS confi guration. Indeed, the poet’s choice of the infl ected form of the participle of bindan in this single instance is most probably motivated by the demands of metre: since resolution of verse-initial wudu is unavoidable, the uninfl ected form bunden would make the verse fall short of a syllable (cf. Mitchell 1985: §36; and Terasawa 2011: 80). A similar metrical rationale is most likely behind the poet’s use of the infl ected participle of ġefr emman in dǣd ġefr emede: he infl ected the participle in this instance precisely to avoid the unmetrical SS pattern. Weiskott’s emendation of dǣd ġefr emede to dǣd ġefr emed refl ects an antiprobabilistic mode of reasoning: it gratuitously corrupts an authorial four-position verse, for no reason other than to increase the apparent evidence for the authenticity of the SS contour. 1.4 Corrupt manuscript readings The previous three sections have accounted for the four-position metrical confi guration of nine out of the thirteen instances that Weiskott adduced as evidence for the SS pattern. The present section considers the remaining four items, which are in actuality corrupt manuscript readings. One of them is rǣhte ongēan, at the beginning of line 7 on folio 149r (corresponding to l. 747b). Taken 49 The past participle ġefr emed in 476a must depend upon the accusative singular neuter pronoun hwæt in 474b, since fǣrnīða is genitive plural. The uninfl ected form ġefr emed is therefore the only grammatically correct possibility. Weiskott’s line of reasoning, however, seems to be based on the false assumption that ġefr emed ought to be infl ected. Three-position verses and the metrical practice of the Beowulf poet 63 SELIM 20 (2013–2014) at face value, it should constitute a verse featuring the three- position SS stress contour. But this reading is suspect on text- critical grounds, since it is immediately preceded by an erasure of fi ve letters at the end of line 6 (Zupitza 1959: 36). Since the fi rst of the damaged letters is an h, Weiskott argues that the scribe copied the word handa by eye-skip to 746a, and that he then erased it intentionally to achieve the purportedly correct three-position reading (2013: 483, n. 4). Several problems present themselves. First, if rǣhte ongēan were an authentic verse by itself, the verbal form rǣhte would lack an object (Robinson 1996: 56). Consequently, it would have to be assumed that the noun phrase hiġeþīhtiġne rinċ is used ἀπὸ κοινοῦ by the Beowulf poet as the grammatical object of two distinct verbs, the preceding nam and the following rǣhte. This interpretation must be rejected, since ἀπὸ κοινοῦ constructions are not a genuine feature of Old English verse (Fulk 2003: 3–9). Second, and even more important, the three-position analysis neglects the well-known fact that the erasure preceding ræhte coincides with an erasure at exactly the same place on the following leaf (Zupitza 1959: 37).50 This clearly suggests that it was something spilt on the vellum that obscured the words preceding ræhte, not the deliberate hand of the scribe. The reading ræhte ongean can then be reliably considered defective. Indeed, this is the stance adopted by the editors of Klaeber IV, who fi ll the fi ve-letter gap in the manuscript by adding hē him, two sentence particles that not only restore the syntax and sense of the passage, but also make a standard four- position type B verse. Another two readings that Weiskott presented as authorial SS verses in Beowulf are grētte þā (corresponding to 652a),51 and ġeġnum fōr (corresponding to 1404b).52 But surely these two readings are corrupt, since trisyllabic verses of any kind are virtually 50 See also Pope (1966: 372), and Fulk (1992: §209). 51 “Addressed then.” 52 “Had gone forward.” 64SELIM 20 (2013–2014) Rafael J. Pascual non-existent in the Old English poetic corpus. The evidential force of their absence is so compelling that the stricture against verses with less than four syllables is regarded by practically all metrists as “the most basic and universal of the metrical rules” (Amos 1980: 15),53 and even the most conservative editors of Old English poetry emend trisyllabic verses on that single basis. Further, plausible sources of scribal confusion for these two readings have been readily identifi ed. In regard to 652a, the source of error suggests itself clearly aft er comparison with 2516a Ġegrētte þā: the similarity between ge- and gre- likely led the scribe to overlook ge- as he copied fr om his exemplar (Andrew 1948: 141). This is in fact the position uniformly endorsed by editors of Beowulf, who emend the text at this point by adding the prefi x ġe-. With regard to ġeġnum fōr, to consider it authentic would neglect the fact that trisyllabic sequences with an SS stress contour and with an alliterating fi rst lift occurring within the second half of the line are invariably either preceded by a minimum of one unstressed syllable or followed by exactly one unstressed syllable both in Beowulf and elsewhere in Old English poetry. Since ġeġnum fōr does not admit any other element aft er fōr, at least one unstressed element seems to have been dropped accidentally by the scribe immediately before ġeġnum. The editors of Klaeber IV supply þǣr, a relative conj unction that not only makes for a regular type B verse, but also improves the syntax of the passage. There remains only one supposed instance of the SS pattern in Beowulf to be considered: 2150a lissa ġelong. Although its source of error is debatable, the three-position analysis for this verse has been nonetheless repeatedly questioned. Geoff rey Russom, for example, has proposed an attractive four-position interpretation, according to which the letter a should be construed not as the infl ectional ending for genitive plural, but as the lexically prominent adverb ā 53 See also Sievers 1885: 268ff ., 312; Bliss 1962: §15; Pope 1966: 318–320; 371–372; Fulk 1996: 5; 1997: 39; Terasawa 2009; 2011: 49–52. Three-position verses and the metrical practice of the Beowulf poet 65 SELIM 20 (2013–2014) “always” (1987: 117–118). That way, the verse would read liss ā ġelong, with the stress contour of a rhythmical type E. Additionally, the editors of Klaeber IV have suggested that ġelong is perhaps a scribal substitute for an authorial dialect form that would have resulted in an original four-position verse (Fulk, Bjork, and Niles 2008: 234). Nevertheless, this verse is clearly unique in that none of the plausible sources of error has achieved clear consensus among metrists. The diffi culty that lissa ġelong has traditionally posed to experts in Old English metre is used by Weiskott as a lever for his argument in support of the metricality of the SS verse type. He focuses the entire body of his essay exclusively on the resistance shown by lissa ġelong to consensual emendation, while relegating the other supposed instances of the SS pattern to a list in a footnote, as if the mere lack of a universally accepted source of error for lissa ġelong somehow validated the dubious three-position interpretation of the other twelve verses. The strategy followed by Weiskott might be rhetorically eff ective, but it is unwarranted in proper metrical argumentation, where, as has been argued at the beginning of this study, the authenticity of a verse type is established not on the basis of an isolated and relatively ambiguous manuscript reading, like hrēas blāc or lissa ġelong, but on the strength of a statistically signifi cant incidence of unambiguous instances in the surviving corpus of Old English poetry. This and other related methodological issues are treated more extensively in the next part of the present article. 2 Incidence and authenticy When Weiskott’s ambiguous corpus of thirteen verses is taken fr om its marginal location in the footnotes and is examined carefully, it becomes noticed that only two of the supposed instances of the SS pattern are trisyllabic. Such an insignifi cant incidence in an already insignifi cant corpus indicates that the SS type is inauthentic, since the occurrences of ideal realizations of a genuine metrical pattern ought to outnumber those of marked 66SELIM 20 (2013–2014) Rafael J. Pascual realizations. For example, four-syllable type A verses like hringa fenġel,54 in which each syllable constitutes a single metrical position, were regarded by poets as ideal realizations of the four-position SS pattern, as is indicated by their outstanding incidence in Old English poetry. On the other hand, fi ve-syllable type A verses like monegum mǣġþum55 or swǣse ġesīþas56 were perceived as acceptable, marked variants of the same metrical type. Consequently, their incidences, though substantial, are not as high as those of the four-syllable realization. The situation is exactly the opposite with regard to Weiskott’s corpus: it consists of only a handful of verses, the majority of which could not count as ideal realizations. According to his scansion, fi ve verses would show protracted drops (183b, 186b, 736a, 747b and 2150a); another six would have a resolved second lift (845a, 881a, 940b, 954a, 1728a and 2430b); and only the remaining two would be ideal trisyllabic realizations of the three-position pattern (652a and 1404b). Had the SS pattern been an authentic metrical type at the disposal of Old English poets, the body of verses that could have been gathered would be much larger, and the proportion between ideal and marked realizations would be the converse. Thus, it is precisely the character of Weiskott’s own corpus of evidence that betrays the inauthenticity of the pattern it aims to validate. Trisyllabic verses featuring an SS pattern are vanishingly rare in the surviving corpus of Old English poetry. In fact, some syllabic sequences with the SS stress contour for which there are good linguistic reasons to be expected are never found as independent verses. Such absence is indicative of the unmetricality of the SS pattern, because an authentic verse type would inevitably have resulted in a signifi cant number of linguistically probable ideal 54 “Prince of rings.” 55 “To many nations.” 56 “Dear comrades.” Three-position verses and the metrical practice of the Beowulf poet 67 SELIM 20 (2013–2014) realizations. For example, trisyllabic verses like *swǣs ġesīþ,57 which consists of a monosyllable followed by an iambic disyllable, are systematically absent fr om the records. Words with the stress contour of swǣs, however, are extremely common in both Old English language and verse; and words with the stress contour of ġesīþ, though not as fr equent, are also common in the language and easily found in other metrical contexts. Clearly, the nature of the stricture against the occurrence of the sequence *swǣs ġesīþ in verse must be other than linguistic. The recursive incidence of similar verses like swǣse ġesīþas (Beowulf 29a, 2040a, 2518a; cf. 1934a) suggests that the restriction is purely metrical: verses like *swǣs ġesīþ do not occur because they are prohibited by the metrical system. That the ideal realization of the SS pattern is unmetrical is perhaps the clearest indication that the entire pattern must be inauthentic. To argue that the SS type is authentic in Beowulf is thus to fi ght an inevitable defeat against the virtually non-existent incidence of its ideal trisyllabic realizations both in Beowulf and in the rest of Old English poetic monuments. The problems with Weiskott’s argument do not end here. He states that the traditional prohibition of Sieversian metrics against three-position verses is unwarranted, given the legitimacy accorded to type D* verses like Beowulf 770a rēþe renweardas,58 whose metrical confi guration (SSs) apparently consists of fi ve positions (2013 passim). Or, to put it another way, Weiskott maintains that since an apparently non-four-position pattern like type D* is regarded as authentic by Sieversian metrics, then the restriction traditionally held by metrists against another non-four-position pattern like SS must be arbitrary. But, on the contrary, it is Weiskott’s charge against Sieversian metrics that is demonstrably baseless. The foundation on which the edifi ce of Sieversian metrics is constructed is essentially empirical. Thus, an apparently problematic verse type 57 “Dear comrade.” 58 “Fierce guardians of the house.” 68SELIM 20 (2013–2014) Rafael J. Pascual like D* is nonetheless considered authentic in Sieversian formalism on account of its statistically signifi cant incidence. According to A. J. Bliss’s scansion, there are 146 unambiguous instances of that type in Beowulf.59 Since the appearance of that type is restricted to the on-verse, its incidence in the whole poem is of approximately 4.6%. The regularity indicated by such a substantial fi gure cannot be ascribed to scribal corruption. The presence of 146 verses unambiguously featuring the same metrical structure must necessarily refl ect the metrical practice of the Beowulf poet. The authenticity that Sieversian metrics accords to type D* is therefore supported by strong empirical evidence. This point can be further illustrated by reference to another well-known verse pattern that also seems not to comply with the four-position principle: type A3, as in Beowulf 106a siþðan him scyppen,60 which apparently comprises three metrical positions. Its statistical incidence is once again the reason why it is accepted as an authentic verse type by Sieversian metrics. Calvin B. Kendall counts 315 unambiguous instances of type A3 in Beowulf (1983: 14). Since instances of type A3 can occur only in the on-verse, its incidence in the whole poem amounts to approximately 10%. It cannot reasonably be doubted that this fi gure results fr om the application of the principles governing the metrical practice of the poet. The authenticity of this verse type, like that of type D*, is therefore founded on its statistically substantial number of occurrences. The genuine existence of types D* and A3, however, does not compromise the authenticity of the four-position principle, which is still able to account for the metricality of more than 85% of the remaining verses in the poem. It is on the basis of the empirically demonstrable reality of types D* and A3, on the one hand, and of the four-position principle, on the other, that the Sieversian theory 59 This fi gure has been calculated using Vickman 1990. 60 “Aft er him the lord.” On the loss of fi nal d in scyppen, see Klaeber IV, “Language and Poetic Form,” §20.7. Three-position verses and the metrical practice of the Beowulf poet 69 SELIM 20 (2013–2014) of Old English metre is constructed. Thus, although type A3 has sometimes been analysed as a three-position pattern, traditional metrists categorize it as an acceptable variant of the basic four- position metrical confi guration (Cable 1974: 20–31; Fulk 2002: 339, n. 10). The theoretical harmonization between type D* and the four-position principle has been provided by both Tom Cable and Seiichi Suzuki, who convincingly account for type D* as a surface manifestation of an underlying four-position pattern (Cable 1991: 37, 138–139, 143; Suzuki 1992; 1995: 21–22). The case for the authenticity of the SS pattern, on the contrary, lacks empirical justifi cation. A cursory glance at the incidences of types D* and A3, on the one hand, and at the incidence of verses that according to Weiskott feature the SS pattern, on the other, suffi ces to substantiate this claim. Since according to him there is no metrical restriction on SS verses preventing them fr om occurring in the off -verse, the incidence of his thirteen verses in Beowulf amounts to 0.2%, a fi gure that starkly contrasts with the 4.6% and 10% incidences of types D* and A3. These fi gures demonstrate that the incidences of the three verse types are not comparable. If the SS type had been considered authentic by the poet, it is improbable that he would have produced such an insignifi cant quantity of three-position verses. As we can see, then, the mere calculation of the pertinent statistics confi rms that the allegation of arbitrariness levelled by Weiskott against Sieversian metrics is fr ivolous. The Sieversian acknowledgement of the authenticity of types D* and A3 is as empirically justifi ed as its rejection of the genuineness of the SS pattern, since the two positions are determined exclusively by their diff erential incidences in the corpus of Old English verse. With an incidence slightly above 0%, the plea for the authenticity of the SS pattern seems to be based on nothing but wishful thinking. Yet Weiskott’s argument is faulty in a more fundamental way. If the results of metrical research are to be trusted, the incidence 70SELIM 20 (2013–2014) Rafael J. Pascual of a certain verse type can be calculated only according to the number of unambiguous instances that occur in the corpus under study. As has been argued in the fi rst part of the present article, however, at least twelve out of Weiskott’s thirteen instances do not unambiguously feature the SS pattern. Weiskott himself seems to be aware of this fact when he states that the verses he gathers “are always diff erently explained or emended” (2013: 483). That the traditional explanations of these verses are summarily dismissed by Weiskott is unsurprising, given his belief that the restriction of Sieversian metrics against the SS pattern is arbitrary. But genuine metrical studies do not proceed that way. The authenticity of a verse type can be gauged only by the number of its unambiguous instances, since an authentic verse type would have resulted in a signifi cant number of verses of that type for which no coherent alternative explanations could be proposed. Since the only instance that could be unambiguously adduced in support of the authenticity of the SS pattern in Beowulf is lissa ġelong, the actual incidence of unambiguous occurrences in the poem is 0.01%. Thus, the SS pattern is far too infr equent in Beowulf to admit its metricality. As Fulk has put it, “scribal transmission is too uncertain to permit a single example of a metrical type to carry much weight” (1992: §209). Another argument advanced by Weiskott in support of the authenticity of the SS type in Beowulf is that the pattern in question is authentic in Old Norse verse (2013: 484, n. 4). Yet once again, his argument is contradicted by the evidence. The SS pattern has traditionally been considered a genuine metrical type in fornyrðislag due to its statistically signifi cant presence in Eddic poetry, especially in some poems. For example, according to Suzuki’s count, the heroic Sigurðarqviða in scamma, at 568 verses, contains 29 instances of the SS pattern; and the mythological poems Hyndlolióð and Rígsþula, at 294 and 296 verses respectively, Three-position verses and the metrical practice of the Beowulf poet 71 SELIM 20 (2013–2014) contain 21 and 59 occurrences (2009: 31).61 Therefore, the respective incidences of the SS verse type in these Eddic poems are 5.1%, 7.14%, and 20%. Faced with statistics such as these, Eduard Sievers naturally accepted the type as formally legitimate in fornyrðislag, since such substantial incidences cannot have accidentally resulted fr om the scribes’ unstable practice (1893: §45.2). These fi gures starkly contrast with the trivial 0.2% incidence to which Weiskott’s corpus of thirteen instances would amount if they unambiguously featured the SS pattern. Indeed, Weiskott seems not to have calculated these statistics before advancing his argument: it is diffi cult to see how he could have ever proposed the purported parallel between the presence of the SS type in Old Norse and in Beowulf had he reckoned them. Weiskott also argues that the SS pattern would have been subjected to a gradual process of regularization over the history of Old English metre. Although this is certainly the case with the SS pattern in Old Norse, Weiskott’s argument runs counter to the course of Old English metrical history, which is inextricably linked to the history of the Old English language. In Old Norse, the unstressed short vowels of many words had been dropped by the beginning of the ninth century, as a result of which many four-syllable verses became trisyllabic (Gordon 1957: 276; Russom 1998: 34). For example, a trisyllabic verse like Þrymsquiða 17/2 þrúðugr áss,62 whose stress contour is SS, might conceivably have originated in a regular four-position type A verse with the Proto-Norse disyllabic u-stem *ansuʀ in the place of monosyllabic áss. Aft er the loss of unstressed short vowels, the poets would have reinterpreted these trisyllabic verses as regular, thereby composing new verses with a three-position metrical 61 On the development of the SS type in Old Norse verse, see also Suzuki 2011 and 2014. 62 “Mighty god.” 72SELIM 20 (2013–2014) Rafael J. Pascual confi guration, like Rígsþula 43/5 meirr kunni hann,63 a four- syllable verse featuring the three-position SS pattern.64 This interpretation is substantiated by the abovementioned incidences of the SS type in some Eddic poems composed in fornyrðislag. Many of the changes that took place in the Old English language, however, were conducive to the converse tendency (Lehman 1956: 88–93). For one, the development of epenthetic vowels produced an increase in the number of syllables of a signifi cant amount of words. Further, as a consequence of the reduction in the number of compounds that came about with the decline of the poetic tradition, the language of poetry was patterned on that of prose, which substantially increased the number of unstressed function words in the line. Hence, contrary to Weiskott’s argument, non-early Old English poetry was an unlikely context for the regularization of the catalectic SS metrical type to occur.65 The incidence of the SS pattern in a larger corpus of Old English poetry also contradicts the notion that it became regularized later in the period. In his monumental A History of Old English Meter, Fulk endeavoured to determine the chronological signifi cance of a set of metrical and linguistic archaisms by studying their distribution throughout a corpus of more than fourteen thousand lines, containing poems that can be externally dated to both the early period, like Cædmon’s Hymn, and the late period, like The Battle of Maldon or Durham. Excluding Beowulf fr om this corpus, the number of verses with an unambiguous SS stress 63 “He knew more.” 64 Winfr ed P. Lehman lists a series of phonological and morphological changes in the Old Norse language that contributed to the appearance of the catalectic SS type in fornyrðislag metre (1956: 80–84). 65 The increase in the number of unstressed function words in late Old English verse has long been recognized. See, for example, Russom 2002, where the expansion of the line is treated as an indication of late composition. See also Cable 1991: 41–65; Fulk 1992: §§290–317; Russom 2004: 292–297, 2012; and Hartman 2014. Three-position verses and the metrical practice of the Beowulf poet 73 SELIM 20 (2013–2014) contour is three: Genesis A 2217b ǣniġ ne wearð,66 2695 lāre ġebearh67 and Elene 534a fr īġneð ymb ðæt trēo68 (Fulk 1992: §§210–211). Or, to put it another way, the incidence of unambiguous occurrences of the SS type in such an enormous body of verse is of approximately 0.01%, exactly the same as the incidence in Beowulf. This fi gure unequivocally indicates that the poets perceived the three-position SS pattern as unmetrical throughout the history of Old English metre, and that therefore the four unambiguous instances found in the large corpus of verse analysed by Fulk, including Beowulf, are not authorial. As Fulk has put it, “the underlying four-position pattern remains unchanged over the history of Old English verse, fr om Cædmon’s Hymn to Durham. Even poems like Maldon that diff er widely fr om the standard of Beowulf in numerous details do not violate the four-position pattern” (1992: §208). 3 Conclusion At the beginning of his essay, Weiskott asserts that “verses of the form SS occur in Beowulf ” (2013: 483). Of course, SS verses occur in the transmitted text of the poem, alongside many corrupt forms requiring emendation.69 Their occurrence in Beowulf, however, is a hypothesis to be tested, not an indisputable fact upon which an argument can be constructed. The present essay has demonstrated that the SS metrical type fails to pass the test of authenticity. As we have seen, there are four unambiguous occurrences in a corpus of approximately 28,364 verses. This means that its overall incidence in such a substantial body of poetry, at 0.01%, is almost non-existent. To regard it as authentic, therefore, would compel us to believe that the poets’ systematic avoidance of that type is an accident—an 66 “No [son] was.” 67 “By cunning protected.” 68 “Asks about that tree.” 69 See, for example, the errors surveyed in Lapidge 2000 and Neidorf 2013. 74SELIM 20 (2013–2014) Rafael J. Pascual extremely improbable coincidence. It is far more probable that the reason why the poets systematically avoided the SS pattern is that they considered it unmetrical on account of its three positions, and that consequently the few unambiguous instances of the SS type that have happened to be recorded are the products of scribal error. This hypothesis has the complementary virtue of accounting for the metricality of the vast majority of verses found in the surviving poetic manuscripts. Weiskott’s claim (2013: 485) that treating SS verses as authentic would be a gain for textual criticism and metrical study is plainly mistaken. In actuality, crediting his untenable hypothesis would cause textual critics to regard scribal errors as authorial readings and lead metrists to misapprehend the principles that govern the metrical practice of the Beowulf poet. Rafael J. Pascual Universidad de Granada References Amos, A. C. 1980: Linguistic Means of Determining the Dates of Old English Literary Texts. Cambridge (MA), Medieval Academy of America. Andrew, S. O. 1948: Postscript on Beowulf. Cambridge, Cambridge University Press. Bliss, A. J. 1962: An Introduction to Old English Metre. Oxford, Blackwell. Bliss, A. J. 1967: The Metre of Beowulf. Rev. ed. Oxford, Blackwell. Cable, T. 1974: The Meter and Melody of Beowulf. Urbana, University of Illinois Press. Cable, T. 1991: The English Alliterative Tradition. Philadelphia, University of Pennsylvania Press. Campbell, A. 1964: Old English Grammar. Rev. ed. Oxford, Clarendon Press. Three-position verses and the metrical practice of the Beowulf poet 75 SELIM 20 (2013–2014) Campbell, A. 1970: Verse Infl uences in Old English Prose. In J. L. Rosier ed. Philological Essays: Studies in Old and Middle English Language and Literature in Honour of Herbert Dean Meritt. The Hague and Paris, Mouton: 93–98. Clark, G. 2014: Scandals in Toronto: Kaluza’s Law and Transliteration Errors. In L. Neidorf ed. The Dating of Beowulf: A Reassessment. Cambridge, D. S. Brewer: 219–234. Donoghue, D. 1997: Language Matters. In K. O’Brien O’Keeff e ed. Reading Old English Texts. Cambridge, Cambridge University Press: 59–78. Foley, J. M. 1985: The Scansion of Beowulf in Its Indo-European Context. In A. Hernández & A. Renoir eds. Approaches to Beowulfi an Scansion. Berkeley, University of California: 7–17 Fulk, R. D. 1992: A History of Old English Meter. Philadelphia, University of Pennsylvania Press. Fulk, R. D. 1996: Inductive Methods in the Textual Criticism of Old English Verse. Medievalia et Humanistica 23: 1–24. Fulk, R. D. 1997: Textual Criticism. In R. E. Bjork & J. D. Niles eds. A Beowulf Handbook. Lincoln, University of Nebraska Press: 35–54. Fulk, R. D. 2002: Early Middle English Evidence for Old English Meter: Resolution in Poema morale. Journal of Germanic Linguistics 14: 331–355. Fulk, R. D. 2003: On Argumentation in Old English Philology, with Particular Reference to the Editing and Dating of Beowulf. Anglo- Saxon England 32: 1–26. Fulk, R. D: 2007a: Archaisms and Neologisms in the Language of Beowulf. In C. M. Cain & G. Russom eds. Studies in the History of the English Language III. Berlin, Mouton de Gruyter: 267–287. Fulk, R. D. 2007b: Old English Meter and Oral Tradition: Three Issues Bearing on Poetic Chronology. Journal of English and Germanic Philology 106: 304–324. 76SELIM 20 (2013–2014) Rafael J. Pascual Fulk, R. D. 2012a: Germanic Prosody. In R. Greene & S. Cushman eds. The Princeton Encyclopedia of Poetry and Poetics. Princeton, Princeton University Press: 557–559. Fulk, R. D. 2012b: Old English: Literary Language. In A. Bergs & L. Brinton eds. English Historical Linguistics: An International Handbook. Vol. 1. Berlin, Mouton de Gruyter: 385–398. Fulk, R. D. 2014: Beowulf and Language History. In L. Neidorf ed. The Dating of Beowulf: A Reassessment. Cambridge, D. S. Brewer: 19–36. Fulk, R. D., R. E. Bjork & J. D. Niles eds. 2008: Klaeber’s Beowulf and the Fight at Finnsburg. 4th ed. Toronto, University of Toronto Press. Getty, M. 2002: The Metre of Beowulf: A Constraint-Based Approach. Berlin, Mouton. Gordon, E. V. 1957: An Introduction to Old Norse. 2nd ed. rev. by A. R. Taylor. Oxford, Oxford University Press. Griffi th, M. ed. 1997: Judith. Exeter, University of Exeter Press. Hartman, M. E. 2014: The Limits of Conservative Composition in Old English Poetry. In L. Neidorf ed. The Dating of Beowulf: A Reassessment. Cambridge, D. S. Brewer: 79–96. Hogg, R. M. 2011: A Grammar of Old English. Vol. 1: Phonology. Chichester, Wiley-Blackwell. Holthausen, F. 1963: Altenglisches etymologisches Wörterbuch. Heidelberg, Carl Winter Universitätsverlag. Hutcheson, B. R. 1992: Kuhn’s Laws, Finite Verbs, and the Critics. Studia Neophilologica 64: 129–139. Hutcheson, B. R. 1995: Old English Poetic Metre. Cambridge, D. S. Brewer. Kaluza, M. 1896: Zur Betonungs- und Verslehre des Altenglischen. In Festschrift zum siebzigsten Geburtstage Oskar Schade. Köningsberg, Hartung: 101–133. Kendall, C. B. 1983: The Metrical Grammar of Beowulf: Displacement. Speculum 58: 1–30. Three-position verses and the metrical practice of the Beowulf poet 77 SELIM 20 (2013–2014) Kendall, C. B. 1991: The Metrical Grammar of Beowulf. Cambridge, Cambridge University Press. Kuhn, H. 1933: Zur Wortstellung und -betonung im Altgermanischen. Beiträge zur Geschichte der deutschen Sprache und Literatur. Vol. 57. Halle, Max Niemeyer: 1–109. Lapidge, M. 2000: The Archetype of Beowulf. Anglo-Saxon England 29: 5–41. Le Page, R. B. 1959: Alliterative Patterns as a Test of Style in Old English Poetry. Journal of English and Germanic Philology 58: 434–441. Lehman, W. P. 1956: The Development of Germanic Verse Form. Austin, University of Texas Press and Linguistic Society of America. Lucas, P. J. 1990: On the Role of Some Adverbs in Old English Verse Grammar. In S. Adamson, V. A. Law, N. Vincent & S. Wright eds. Papers fr om the 5th International Conference on English Historical Linguistics. Amsterdam, John Benj amins: 293–312. Mitchell, B. 1985: Old English Syntax. 2 vols. Oxford, Clarendon. Momma, H. 1989: The ‘Gnomic Formula’ and Some Additions to Bliss’s Old English Metrical System. Notes and Queries 4: 423–426. Momma, H. 1997: The Composition of Old English Poetry. Cambridge, Cambridge University Press. Neidorf, L. 2013: Scribal Errors of Proper Names in the Beowulf Manuscript. Anglo-Saxon England 42: 249–269. Neidorf, L. & R. J. Pascual forthcoming [2015]: The Language of Beowulf and the Conditioning of Kaluza’s Law. Neophilologus. Orton, P. R. 1999: Anglo-Saxon Attitudes to Kuhn’s Laws. Review of English Studies 50: 287–303. Pascual, R. J. forthcoming: Ælfr ic’s Rhythmical Prose and the Study of Old English Metre. English Studies. Pope, J. C. 1966: The Rhythm of Beowulf. Rev. ed. New Haven, Yale University Press. 78SELIM 20 (2013–2014) Rafael J. Pascual Pope, J. C. ed. 2001: Eight Old English Poems. 3rd ed. rev. by R. D. Fulk. New York and London, Norton. Robinson, F. C. 1996: Beowulf in the Twentieth Century. Proceedings of the British Academy 94: 45–62. Russom, G. 1987: Old English Meter and Linguistic Theory. Cambridge, Cambridge University Press. Russom, G. 1998: Beowulf and Old Germanic Metre. Cambridge, Cambridge University Press. Russom, G. 2002: Dating Criteria for Old English Poems. In D. Minkova & R. P. Stockwell eds. Studies in the History of the English Language: A Millennial Perspective. Berlin, Mouton de Gruyter: 245–265. Russom, G. 2004: The Evolution of Middle English Alliterative Meter. In A. Curzan & K. Emmons eds. Studies in the History of the English Language II: Unfolding Conversations. Berlin, Mouton de Gruyter: 279–304. Russom, G. 2012: What Explanatory Metrics Has to Say about the History of English Function Words. In D. Denison, R. Bermúdez-Otero, C. McCuly & E. Moore eds. Analysing Older English. Cambridge, Cambridge University Press: 15–27. Schabram, H. 1960: The Seasons for Fasting 206f. Mit eine Beitrag zur ae. Metrik. In W. Iser & H. Schabram eds. Britannica: Festschrift fü r Hermann M. Flasdieck. Heidelberg, Carl Winter: 233–237. Sievers, E. 1885: Zur Rhythmik des germanischen Alliterationsverses. Beiträge zur Geschichte der deutschen Sprache und Literatur. Vol. 10. Halle, Max Niemeyer: 209–314, 451–545. Sievers, E. 1893: Altgermanische Metrik. Halle, Max Niemeyer. Stockwell, R. P. & D. Minkova 1997: Prosody. In R. E. Bjork & J. D. Niles eds. A Beowulf Handbook. Lincoln, University of Nebraska Press: 55–84. Suzuki, S. 1992: On Reducing Type D* to Type D in the Meter of Beowulf. Neuphilologische Mitteilungen 93: 257–269. Three-position verses and the metrical practice of the Beowulf poet 79 SELIM 20 (2013–2014) Suzuki, S. 1995: In Defense of Resolution as a Metrical Principle in the Meter of Beowulf. English Studies 76: 20–33. Suzuki, S. 2009: Three-Position Verses in Old Norse Fornyrðislag Meter: Statistical and Comparative Perspectives. NOWELE 2009: 3–40. Suzuki, S. 2011: Catalexis, Suspension of Resolution, and the Organization of the Cadence in Eddic Meters. In M. Lotman and M. K. Lotman eds. Frontiers in Comparative Prosody: In Memoriam Mikhail Gasparov. Bern, Peter Lang: 373–400. Suzuki, S. 2014: The Meters of Old Norse Eddic Poetry. Berlin and Boston: de Gruyter. Terasawa, J. 2009: Beowulf における3音節以下半行について [Notes on Half- lines of Three or Fewer Syllables in Beowulf ]. In K. Karasawa ed. ベーオウルフとその周辺 : 忍足欣四郎先生追悼論文集 [Essays on Beowulf and Other Medieval English Literature in Memory of Kinshiro Oshitari]. Yokohama, Shumpusha: 278–293. Terasawa, J. 2011: Old English Metre: An Introduction. Toronto, University of Toronto Press. Vickman, J. 1990: A Metrical Concordance to Beowulf. With a Preface by R. D. Fulk. Binghamton, Center for Medieval and Early Renaissance Studies, State University of New York at Binghamton (=Old English Newsletter Subsidia 16). Weiskott, E. 2013: Three-Position Verses in Beowulf. Notes and Queries 60 (4): 483–485. Zupitza, J. 1959: Beowulf: Reproduced in Facsimile fr om the Unique Manuscript British Museum MS. Cotton Vitellius A. xv. 2nd ed. rev. by N. Davis. Oxford, Oxford University Press. • Received 12 May 2014; accepted 21 May 2014