The vast majority of the approximately 60,000 verses that constitute the surviving corpus of Old English poetry can be shown to have, despite their differing number of syllables, a basic structure of exactly four metrical positions.Footnote ¹ Indeed, four-position verses are so numerous in the texts, and obvious scribal corruptions are so frequent, that Sieversian metrists have traditionally regarded the few verses that apparently consist of less than four positions as the mere products of scribal interference, even if those verses are unobjectionable in terms of grammar, sense, and alliteration.Footnote ²

The communis opinio against catalectic verses was recently challenged by Weiskott (Reference Weiskott2013). On the basis of a few verses from Beowulf, Weiskott postulated the rule that the three-position / x / contour is an authentic pattern of Old English meter. As I demonstrated in a subsequent response (Pascual Reference Pascual2013–Reference Pascual2014), however, of all the verses that Weiskott gathered in support of his case, only l. 2150a, lissa ġelong ‘of joys dependent on’, can be unambiguously analyzed as an instance of the / x / contour. Such empirical insufficiency is a clear indication of the unreliability of Weiskott's argument. If it were credited, one would be compelled to believe that, in the more than 6,000 verses of Beowulf, there is only one unambiguous example of a verse type that was purportedly licensed by the metrical system. As mentioned above, scribes obviously have tampered with their exemplar during the process of transmission.Footnote ³ Given this, it seems more reasonable to regard lissa ġelong as a product of scribal interference (despite its appropriate sense and correct alliteration), rather than to introduce an ad hoc metrical rule to account for this instance alone. Thus, the scarce attestation of unequivocal instances of three-position verses in the manuscript poses an insurmountable obstacle to the case for their metrical authenticity.

In an article published in the present issue of Journal of Germanic Linguistics, Seiichi Suzuki has made a case for a three-position interpretation of seven verses from the text of Beowulf and 20 verses from that of Genesis A, on the grounds that they contain no defects of either sense or alliteration. Unlike Weiskott, he seriously engages with the problem of their relatively scarce attestation in the manuscripts, offering a sophisticated solution that deserves careful consideration. Suzuki defines catalexis as the result of an inherent tendency in the verse structure toward the deletion of a weak metrical position, or drop, especially if the drop is verse-final or follows two clashing lifts. According to Suzuki, Old English poets, on account of the normative status of the four-position principle, conscientiously resisted this natural tendency. Occasionally, however, when their vigilance against catalexis faltered, the poets composed three-position verses. From this point of view, the scarcity of catalectic verses is not an indication of their metrical inauthenticity, but a logical consequence of the poets' resistance to the composition of such verses. In support of this view, Suzuki argues that, among the three-position verses that he identifies, the catalectic type D- configuration / / x is the most frequent (nine instances in Genesis A and five in Beowulf), followed by type C-, x / / (five instances in Genesis A).Footnote ⁴ This is exactly what, according to his account of catalexis, should be expected. Both types D (/ / x x) and C (x / / x) contain second drops that are subject to the two conditions favoring their deletion. Because of that, Suzuki reasons, it must have been particularly difficult for poets to avoid the sporadic composition of verses with a three-position type D- or C- configuration.Footnote ⁵

This response subjects Suzuki's theory of catalexis to critical analysis. First, I examine the verses identified by Suzuki as genuinely consisting of only three metrical positions. I demonstrate that many of them are either more susceptible to an alternative scansion or, despite Suzuki's contention to the contrary, present additional problems of sense and grammar that signal textual corruption. Most of these verses must obviously be excluded from the list of genuine instances of catalexis. Nevertheless, they are interpreted by Suzuki as types D- and C-, precisely the three-position verse types that, according to his account of catalexis, ought to have a higher incidence. Therefore, the definition of catalexis upon which Suzuki's theory is predicated appears to be empirically unsatisfactory, since it makes the wrong predictions, as shown by the evidence.

Second, I examine the few unambiguous three-position verses that remain on Suzuki's lists after the abovementioned instances are excluded. Of these, the majority seem to have a type A- structure (/ x /) with impeccable sense and alliteration. Are they to be regarded as genuine manifestations of catalexis or as errors of transmission? The present response contends that the balance of probability supports the longstanding assumption that catalectic verses are invariably the result of scribal interference.

Careful examination of the five verses from Genesis A that Suzuki classifies as instances of type C- may lead one to the conclusion that they are not genuine instances of catalexis. The catalectic interpretation of three of them—1349a, nū ofor seofon niht ‘now after seven nights’, 1449a, Hē þā ymb seofon niht ‘He then after seven nights’, and 2322a, ymb seofon niht ‘after seven nights’—hinges on the assumption that the manuscript sequence seofon niht ‘seven nights’ represents two simplexes, both of which receive primary stress. However, these two words constitute a compound in both Middle and Modern English (seoueniht and sennight, respectively), and so it is only natural to suppose that the reading seofon niht also represents a compound in Old English.Footnote ⁶ Consequently, Genesis A 1349a, 1449a, and 2322a are not catalectic type C- verses, as Suzuki maintains, but type A3 verses with a half-lift instead of a drop occupying the last metrical position of the verse, as in, for example, Beowulf 1995a, þæt ðū þone wælgǣst ‘that you the slaughtering creature’.Footnote ⁷ This scansion is corroborated by their distribution: These and similar verses, like Exodus 63a, Heht þā ymb twā niht ‘He commanded then after two nights’, and Andreas 185a, Nū bið fore þrēo niht ‘Now it will be before three nights’, always appear in the on-verse, while type C is more frequent in the off-verse.Footnote ⁸ Therefore, Suzuki's claim that these three instances from Genesis A are derived from a type C verse through deletion of the final drop is undermined by the evidence.

Suzuki's type C- analysis of l. 2142b, nis woruldfeoh ‘isn't worldly wealth’, is likewise inadvisable, because this verse probably became deficient in the course of textual transmission, as a glance at its context suggests:

1. (1)

   

As can be seen, l. 2143, þe iċ mē āgan wille ‘that I wish to possess’, apparently an unpaired type A half-line preceded by three anacrustic syllables, is anomalous. If nis woruldfeoh ‘isn't worldly wealth’ were taken at face value, as Suzuki's analysis requires, one would have to believe that the co-occurrence in the manuscript of this aberrant verse with two other metrical irregularities (an unpaired verse and trisyllabic anacrusis) is accidental. In order to avoid crediting such an improbable coincidence, it seems better to concur with Eduard Sievers that there is a lacuna after woruldfeoh ‘worldly wealth’ (Sievers Reference Sievers1887:476). The advantage of this hypothesis is that it provides a unitary explanation for the three contiguous metrical aberrations, conceiving of them not as three separate authentic exceptions to metrical norms, but as the defective outcome of the accidental loss of a portion of text. Even if the original reading is now irrecoverable, on account of its suspect text-critical character l. 2142b should not be adduced as genuine evidence for catalexis in Genesis A.

The remaining instance on Suzuki's list of type C- verses, l. 2600a, Hīe dydon swā ‘They did so’, is especially problematic. As Sievers indicates (1885:498; see also Fulk Reference Fulk1992:§355.4), the presence of the West-Saxonism dydon ‘did’ often generates metrical problems both in Genesis A and elsewhere in Old English poetry, as in, for example, Genesis A 142a, drugon and dydon ‘they carried out and did’; Daniel 196b, cūð ġedydon ‘they made known’; Beowulf 1828b, hwīlum dydon ‘formerly they did’. In all of them, the substitution of the Anglian poetical form dēdon systematically restores metricality. So, if one wants to account for this regularity, one cannot avoid the conclusion that Genesis A 2600a is, in actuality, a four-position type B verse, with scribal substitution of dydon for dēdon. This verse does not, therefore, constitute legitimate evidence for the authenticity of catalexis. The exclusion of this instance from his body of evidence leaves Suzuki with no genuine examples of catalectic type C- verses. If the Genesis A poet could not resist the tendency to delete the verse-final drop after two clashing lifts, as Suzuki maintains, it is remarkable that the entire text of the poem does not contain a single unambiguous instance of type C-, such as the hypothetical *ond sinċfæt ‘and a precious object’. It follows that type C- is not a reality of the meter. The inauthenticity of type C- in turn casts serious doubts on the reliability of Suzuki's definition of catalexis as a whole. Indeed, as the analysis of the alleged type D- verses from Genesis A and Beowulf confirms, there seems to be no reason to postulate the natural tendency toward drop deletion upon which Suzuki's theory of catalexis is predicated.

Following A. N. Doane's second edition of the poem, Suzuki reads Genesis A 907a and 2876a as instances of type D- (bearm tredan ‘walk the bosom’ and dæġes þriddan ‘of the third day’, respectively). In fact, however, these are regular verses that Doane failed to interpret correctly. The traditional reading of ll. 906–907, as printed in Krapp Reference Krapp1931, is the following:Footnote ¹⁰

1. (2)

   

According to the traditional interpretation, l. 907a is a regular expanded type D*2 with double alliteration on b. The editorial reading wēriġ ‘weary’ at 906b, which stands for manuscript werg ‘accursed’, is introduced in order to avoid the catalectic verse werg þīnum ‘accursed on your (belly)’. Unwilling to accept either editorial wēriġ or manuscript werg þīnum, Doane decides to place brēostum ‘belly.dat’ at the end of 906b, thereby generating the three-position verse bearm tredan ‘walk the bosom’ at 907a, which Suzuki takes as an instance of type D-.Footnote ¹¹ Nevertheless, as Fulk (Reference Fulk2004) has demonstrated, l. 906b is probably a standard four-position type A verse with scribal substitution of werg for Anglian *werge, a ja-stem meaning ‘accursed’. Thus, the relocation of brēostum, upon which Suzuki's catalectic analysis of l. 907a is predicated, is unnecessary.

In regard to l. 2876a, which has traditionally been printed as dæġes þriddan up ‘(the beginning) of the third day (rose) up’, Doane believes that the adverb up is the first word of the second half-line rather than the last word of the first.Footnote ¹² This lineation causes problems in both halves of the line. L. 2876b as printed by Doane, up ofer dēop wæter ‘up over deep water’, must be analyzed as a type A2b with alliteration on the second lift.Footnote ¹³ This is implausible because type A2b requires double alliteration, which is strictly forbidden in the off-verse (Bliss Reference Bliss1967:§§54, 79). The placement of up in 2876a not only restores metricality to both verses, but also corrects the line's alliteration: Both dæġes þriddan up and ofer dēop wæter are regular four-position verses (type D4 and type C, respectively) with predictable d alliteration on the first lift of the off-verse. Thus, ll. 907a and 2876a as interpreted by Suzuki do not reflect the metrical practice of the Genesis A poet, and hence they should be excluded from his list of instances of catalectic type D-.

There are compelling linguistic reasons to believe that another two verses from Genesis A on Suzuki's list, 2058b, ēað mihte ‘could more easily’, and 2668a, sprecan sīne ‘his counsellors’, are originally four-position verses that have been corrupted in the process of transmission. The manuscript reading at l. 2058b, ēað mihte, is regarded as grammatically deficient by the vast majority of editors. Doane, following Stanley Reference Stanley, Rugg Brown, Ronan Campton and Robinson1986:195, reads the verse as ðe ēað mihte ‘could more easily’, since the comparative expression ðe ēað and the verb magan not infrequently co-occur in Old English, as in, for example, Andreas 368a, þæt hīe þe ēað mihton ‘that they could more easily’. Nevertheless, as Sievers indicates (Reference Sievers1885:513; see also Wells Reference Wells1969:211), comparative ēað is meaningless in this context, in which the poem's narrator, without establishing any comparison, merely reports Abraham saying that God will help the Israelites. Most editors accordingly emend ēað to ēaðe ‘easily’, the positive form of the adverb, on the reasonable belief that the failure of a scribe to copy the final letter of authorial ēaðe resulted in a verse that is defective in terms of both meter and sense.

Concerning l. 2668a, Doane, following the manuscript, prints it as sprecan sīne. As he mentions in the commentary to his edition (Doane Reference Doane2013:388; see also Sievers Reference Sievers1885:453), however, the word sprecan, ‘counselors’, which translates Latin servōs, is a hapax legomenon. Since the Old English idiomatic term for ‘counselor’ is ġespreca, it appears reasonable to concur with the majority of the poem's editors that a scribe accidentally failed to copy the prefix ġe- (as is assumed, for example, in Beowulf 652a, [Ġe]grētte þā ‘Then (one man) addressed’); therefore, l. 2668a is, in actuality, a regular four-position type C verse with resolution of the first lift.

The addition of an extra syllable to another two verses classified by Suzuki as type D-, Genesis A 1098b, wāt ġearwe ‘I know readily’, and 2629a, heht bringan ‘he commanded to bring’, not only improves their meter, but also renders them more idiomatic in Old English. The co-occurrence of the adverb ġearwe and different forms of the verb witan, which is very common both in Genesis A and elsewhere in poetry, frequently involves the presence of a pronominal subject for witan, as in, for example, Genesis A 2626a, þy¯ hē wiste ġearwe ‘because he knew readily’; Beowulf 2656b, Iċ wāt ġeare ‘I know readily’; Azarias 170b, þæt iċ ġeare wiste ‘That I readily knew’. Accordingly, most editors of the poem (Holthausen Reference Holthausen1914, Krapp Reference Krapp1931, Wells Reference Wells1969, Anlezark Reference Anlezark2011), following Friedrich Graz's original suggestion (Graz Reference Graz1896:70–71), emend l. 1098b to Wāt iċ ġearwe, thus assuming that a scribe unintentionally failed to copy the subject of the verb.Footnote ¹⁴ Without this addition, the passage seems defective, since the verb lacks a subject, and wāt is not distinctively first person.

In regard to l. 2629a, heht bringan, Suzuki's analysis is a particularly questionable one, because the passage in which this verse occurs is defective (Wells Reference Wells1969:227, Doane Reference Doane2013:386). One of its several defects is the absence of an object for the verb heht, which is otherwise very common (see Genesis A 2667b, Heht him fetiġean tō ‘He commanded (them) to fetch him’; 943b, hēt hīe from hweorfan ‘he commanded them to depart from’). As a result, many editors emend this verse to either heht hīe bringan (Krapp Reference Krapp1931, Anlezark Reference Anlezark2011) or heht him bringan (Doane Reference Doane2013), thereby improving both its sense and its meter. Regardless of whether these emendations are accepted, the textual problems that attend l. 2629a render Suzuki's catalectic analysis suspect.

As can be seen, there are good nonmetrical reasons to believe that six of Suzuki's nine instances of type D- in Genesis A are more likely to be regular verses; they have either been misinterpreted by Doane or suffered some kind of corruption in the course of transmission. I turn now to the verses from Beowulf that Suzuki identifies as instances of type D-. Three of them (1889a, 1926a, and 1129a) present problems of sense or grammar, which suggests that these verses are not authentic. The first, l. 1889a, hæġstealdra ‘of young men’, is part of a clause that lacks a grammatical subject. This absence is particularly questionable because the genitive plural hæġstealdra and its apposite felamōdiġra ‘of very courageous men’, in the previous verse, require a nominative head (as in, for example, Genesis A 1862b, hæġstealdra wyn ‘pleasure of young men’). It is therefore probable that a word in the nominative case is missing after hæġstealdra, and so l. 1889a as it appears in the manuscript is incomplete. The editors of Klaeber's Beowulf supply hēap ‘band’, which rectifies both the meter of this verse (rendering it a type E with double alliteration on h) and the grammar of the clause.

If l. 1926a, hēa healle ‘lofty halls’, is taken at face value, as Suzuki's analysis demands, then the noun healle must be interpreted as a nominative plural apposite to the singular noun bold ‘building’ in l. 1925a (Fulk et al. Reference Fulk, Bjork and Niles2008:222). This is unidiomatic, however, both because the reference in 1925a is to a single hall, not several, and because nouns in apposition usually agree in number with each other. Indeed, the editors of Klaeber's Beowulf, crediting Ernst A. Kock's hypothesis (Kock Reference Kock1918:116), emend hēa healle to hēah on healle ‘exalted in hall’, a regular four-position type A verse. In this case, hēah is a singular masculine adjective parallel to the singular noun phrase bregorōf cyning ‘very valiant king’ in l. 1925b.Footnote ¹⁵ Because this emendation, in addition to rectifying the meter of the verse, results in a grammatically correct appositive relation, hēah on healle is likely to be the original verse as composed by the author.

Suzuki's acceptance of the manuscript reading at l. 1129a, unhlitme ‘not reluctantly’, on which his catalectic interpretation of this verse is based, is likewise inadvisable. The immediately preceding word in the manuscript, finnel, is nonsensical. Therefore, virtually all the editors of the poem emend it to Finne, the dative of the proper name Finn, which this passage (part of the Finnesburh episode) unambiguously requires. The most logical way to make sense of this notorious corruption is to assume that l. 1129a as it stands in the manuscript is incomplete, and that some linguistic material preceding unhlitme was accidentally attached to finne, in the previous verse, in the form of an l.Footnote ¹⁶ The manuscript reading at l. 1129a, unhlitme, should be taken, therefore, not as the product of a metrical rule, but as the outcome of defective textual transmission.

What is common to all of the verses discussed so far, in addition to their peculiar metrical structures, is that they exhibit problems of a nonmetrical character, which would normally disqualify them as evidence for metrical argumentation (Pascual Reference Pascual2013–2014:65). Once these verses are excluded from Suzuki's lists of catalectic instances in Genesis A and Beowulf, the total number of three-position verses left is 13. Of these, eight are classified as type A- and five as type D-.Footnote ¹⁷ The ratio of type A- to type D- verses undermines Suzuki's conceptualization of catalexis as the result of the poets’ sporadic failure to resist the natural tendency toward drop deletion in the structure of the verse. He argues that the second drop of type D (/ / x x), inasmuch as it follows a drop that has already been subjected to prominence reduction by the two preceding lifts, is most likely to undergo deletion. Therefore, if his conceptualization of catalexis were correct, it is type D-, not type A-, that should be the most substantially represented in the texts. The smaller number of three-position type D- verses indicates that there is no empirical basis for a strong tendency to delete a verse-final drop preceded by two clashing lifts. Consequently, Suzuki's hypothesis that the scarcity of catalexis is a principled consequence of the poets’ resistance to such a tendency appears at odds with the evidence. Thus, even though the 13 three-position verses discussed above are acceptable in terms of sense and alliteration, the incidence of unambiguous instances in the texts is so low, and the probability of scribal error is so high, that they are better regarded as accidents of transmission devoid of metrical significance. Indeed, this is the view shared by the majority of editors of Genesis A and Beowulf, who have proposed reasonable emendations for most of them.Footnote ¹⁸

The traditional notion that the exceptionally scarce attestation of a verse type in the manuscripts is an indication of metrical inauthenticity—and hence of scribal error—is confirmed by poems surviving in more than one copy. For example, while l. 8a of The Capture of the Five Boroughs is transmitted as a three-position verse in the A version of this poem (dēora by¯ ‘Derby’), its three other versions preserve the four-position verse ond dēora by¯ instead; Soul and Body I 81b, wilddēora ‘of wild beasts’, a catalectic verse, corresponds to the regular type A wildra dēora ‘of wild beasts’ in Soul and Body II; at l. 52a, the B text of Solomon and Saturn reads heofonrīċes ‘of the kingdom of heaven’, a three-position verse, whereas the A text reads heofona rīċes ‘of the kingdom of heavens’, a regular type A; and the two-position ond þurh help, at Daniel 292a, appears as the regular type B ond þurh hyldo help ‘and, by your favor, help’ at Azarias 13a (Neidorf Reference Neidorf2016).Footnote ¹⁹

These examples, of which many more could be found, demonstrate that it is the accidental omission of linguistic material from an original four-position verse that is responsible for the abnormal metrical structure of the catalectic equivalent. Note also that the effect of these omissions on grammar and alliteration is either insignificant or nonexistent. Therefore, the 13 remaining instances of catalexis on Suzuki's lists should be given the same treatment, as suggested by the editors of Genesis A and Beowulf: Even though these verses present no grammatical or alliteration problems, they have been generated by similar errors of transmission.