¹ Corneli Taciti de Origine et Situ Germanorum, ed. J. G. C. Anderson (Oxford, 1938), p. 26Google Scholar, §40.1.

² I have used the editions by R. W. Chambers (Cambridge, 1912) and Kemp Malone (London, 1936, and Copenhagen, 1962).

³ Widsith, ed. Chambers, p. 197.

⁴ Widsith, ed. Malone (1936), p. 144, and (1962), p. 149.

⁵ In what follows a number of Old English grammars are cited by very abbreviated references: Bülbring, Karl D., Altenglisches Elementarbuch (Heidelberg, 1902)Google Scholar, as ‘B’; Brunner, Karl, Altenglische Grammatik, Dritte, neubearbeitete Auflage (Tübingen, 1965)CrossRefGoogle Scholar, as ‘Br’; Campbell, A., Old English Grammar (Oxford, 1959)Google Scholar, as ‘C’; Girvan, R., Angelsaksisch Handboek (Haarlem, 1931)Google Scholar, as ‘G’; Luick, Karl, Historische Grammatik der englischen Sprache (Leipzig, 1914–1940)Google Scholar, as ‘L’; and Sievers, Eduard, An Old English Grammar, trans. and ed. Cook, Albert S., 3rd ed. (Boston, Mass., 1903)Google Scholar, as ‘S–C’.

⁶ C § 120 (2) and § 753 (6), and Br § 173 Anm. 2 and § 408 (3) Anm. 13.

⁷ G § 26, § 120 (2) and § 753 (6), and Br § 173 Anm. 2 and § 408 (3) Anm. 12.

⁸ Malone, , Widsith (1936), p. 147Google Scholar; (1962), p. 153.

⁹ C § 273; Br § 126 Anm. 2.

¹⁰ For a convenient list of attested forms, see Br § 408 (3) Anm. 13, 14 and 15.

¹¹ Often printed ēowian; that the diphthong is short will emerge from the discussion below.

¹² C § 764; Br § 417 (2) Anm. 7; Flasdieck, H. M., ‘Untersuchungen über die germanischen schwachen Verben III. Klasse’, Anglia 59 (1935), 1–192CrossRefGoogle Scholar, esp. 64–9. Flasdieck postulates two different etymologies for different forms of the verb, and in general his explanations seem unnecessarily complicated; however, he provides a very useful list of forms. The form bœwenn, which occurs several times in the Ormulum, suggests that bȳwan may also have had a Class III infinitive *bēawan; it is not discussed by Flasdieck.

¹³ I have borrowed this useful term from Campbell, who uses it (C § 753 and elsewhere) to denote forms derived by regular sound-change from their historical antecedents; it provides a useful contrast to ‘analogical’.

¹⁴ Gothic awēþi represents a different formation from the same stem.

¹⁵ Often printed mēowle, the long diphthong (difficult though it would be to explain) being postulated to explain the syncope of the medial vowel. On syncopation after a short stem, see G §§ 172 (2), 173, and C § 388.

¹⁶ C § 237 (1) (b); Br § 129 (1) Anm. 2. In C § 753 (7), n. 6, Campbell seems to envisage contraction of -awi- to -ai- and then to -ē-; but -ai- would surely give -ā-.

¹⁷ S–C § 173 (2) seems to envisage loss after long syllables only; but cf. § 408 (3), n. 13, which presupposes loss after short syllables. G § 250 (3) and Aanm. 2 suggest that in verbal forms the stem-vowel was analogically lengthened before loss of w; but cf. § 26 and Aanm. Br §§ 129 (1) Anm. 2 and 173 Anm. 1 (e) envisages loss after short syllables in Northumbrian only. C § 406 is the most dogmatic statement of loss in all circumstances; but § 154 (1), (2), cites without comment a number of forms in which w remains after a short syllable.

¹⁸ The Holy Gospels in Anglo-Saxon, Northumbrian, and Old Mercian Versions, ed. W. W. Skeat, (Cambridge, 1871–87). The occurrences are at Matthew xxvi.31, Luke 11.8 and Luke xii.32. The forms in Lindisfarne are respectively edes, edo and édœ; the forms in Rushworth are edœs, ede and eode (on this last form, see below, n. 19). Ross, A. S. C., ‘Lindisfarne and Rushworth One’, N&Q 224 (1979), 194–8Google Scholar, has shown that from Matthew xxvi onwards Rushworth is dependent on Lindisfarne, so that the Rushworth forms are not independent evidence for the sound change.

¹⁹ The Oldest English Texts, ed. Henry Sweet, EETS o.s. 83 (London, 1885)Google Scholar. The occurrences are at vi.7, XL.4, Lxii.7 and cxxxi.3, and the form is strene in each case. The so-called ‘Junius Psalter’, known to be based on the Vespasian Psalter, reproduces these forms, except that at XL.4 it has the form streone; see Der altenglische Junius-Psalter, ed. Eduard Brenner, Anglistische Forschungen 23 (Heidelberg, 1908)Google Scholar. The form streon also occurs in Maxims I 67b, ‘hord [sceal] in streonum bidan’, and these two forms need to be discussed in conjunction with eode in Rushworth (see above, n. 18), since in both cases -ewi- appears to have developed to -ēo-; such a development would be incomprehensible, and it seems certain that the forms must be due to some kind of error. In Rushworth the second scribe copying Lindisfarne ede may have been influenced by his own eowede; similarly the Junius scribe copying Vespasian strene may have been influenced by his own streowen; in Maxims I the form may be influenced by (ge)strēon, since the context is concerned with treasure.

²⁰ B § 217; G § 250 (3) Aanm. 2; C § 237 (1) (b); Br § 129 Anm. 2, 5.

²¹ G § 250 (2). Br § 130 (1) Anm. 3 has a different explanation not involving the loss of w.

²² Br § 129 Anm. 2, 5.

²³ B § 257; S–C § 258 (1), n. 2; G § 116 Aanm. 1; C §§ 211, 593 (2); Br § 258 Anm. 4. Br § 96 (1) inconsistently has ‘aus *awi, vgl. lat. ovis’.

²⁴ Streitberg, Wilhelm, Gotisches Elementarbuch (Heidelberg, 1910), § 150Google Scholar; Wright, Joseph, Grammar of the Gothic Language (Oxford, 1910), $$$§§ 192–3Google Scholar; and Braune, Wilhelm, Gotische Grammatik, neubearbeitet von Ernst A. Ebbinghaus (Tübingen, 1961), § 97Google Scholar Anm. 1.

²⁵ Prokosch, E., A Comparative Germanic Grammar (Philadelphia, Pa, 1939), p. 245Google Scholar, § 81 (b); cf. the cognate OE mǣgþ, ‘maiden’, and þegn, ‘retainer’.

²⁶ C § 406, n. 4.

²⁷ C § 579 (4)

²⁸ The Old English Orosius, ed. Janet Bately, EETS s.s. 6 (London, 1980), 16Google Scholar, line 28.

²⁹ E.g., B. § 257. According to Wyatt, A. J., Old English Grammar (Cambridge, 1897)Google Scholar, § 147 (c), ew became first euw then ēow. See also S–C § 73 (1), n. 1.

³⁰ L § 230 and Anm. 1. This doctrine was rejected by G § 116 Aanm. 1 but accepted by C § 211 and Br § 110 Anm. 2; there are some reservations, however, in Br § 87 Anm.

³¹ See, however, Br §§ 378 (2) and Anm. 1, 391 (2) Anm. 8. It is perhaps just possible that gesewen might have e from a by i-mutation, but there is grave danger of arguing in a circle. The argument seems to run as follows: original e is always broken before w; e from a by i-mutation is never broken before w; e in gesewen is never broken before w; therefore gesewen must have e from a by i-mutation.

³² So L § 230 and Br § 110 Anm. 2.

³³ So explained by B § 216 and S–C § 174 (4), n. 3.

³⁴ For lists of forms, see C § 753 (6) and (7), and Br § 408 (3) Anm. 12 and 13.

³⁵ The attested infinitive spīwan is usually taken to belong to the cognate strong verb of Class 1. However, some at least of the instances could be developed from WS *spīewan, with the late West Saxon change of īew to īw. On this sound-change, see Br § 107 Anm. 5; but cf. also C § 300, n. 2.

³⁶ For lists of forms, see C § 753 (6), and Br § 408 (3) Anm. 15.

³⁷ C §§ 211, 752; Br § 400 (2) Anm. 2.

³⁸ C § 610 (7).

³⁹ Procopius, , De Bello Cothico iv. 20Google Scholar; see Procopius with an English Translation, ed. H. B. Dewing, (London, 1914–1941) v, 252–5.Google Scholar

⁴⁰ Chadwick, H. M., The Origin of the English Nation (Cambridge, 1907), p. 106.Google Scholar

⁴¹ Prokosch, , Germanic, pp. 92–3Google Scholar, § 33; according to Lehmann, W. P., Proto-Indo-European Phonology (Austin, Tex., 1955), pp. 47–72Google Scholar, the operation of Holtzmann's Law depends on the presence of a laryngeal consonant.

⁴² Widsith, ed. Chambers, p. 197.

⁴³ Widsith, ed. Malone, (1936), p. 144, and (1962), p. 149.

⁴⁴ For references, see Widsith, ed. Chambers, p. 197.