¹ Apparently shortened in the late Alemannic of Notker, the only source using these longer optatives that consistently marks long vowels.

² Rudolf Kögel, PBB 9.507–8; Josef Schatz, Althochdeutsche Grammatik §517; Braune-Mitzka, Althochdeutsche Grammatik ⁸ §310.

³ In the participle and gerund, where *ōj was followed by a syllable with secondary stress, the regular phonetic development was to /j/ after a short accented syllable and to zero elsewhere, as illustrated by Old English dialectal dobgendi and tācnendi (Corpus glosses). West Saxon la þianne and la þiende have syllabic i from the other ōja-forms (Sievers-Brunner, Altenglische Grammatik ² §412 Anm. 10). To judge by the material in C. Günther, Die Verba im Altostfriesischen 67, Old East Frisian generalized the alternant -and in the participle and in the gerund took over /i/ from the infinitive. I have not examined the development in Old West Frisian.

⁴ Hence a total of 47, higher by five than the figure given by E. H. Sehrt, Lg. 16.100. Sehrt does not include bimurnie C 1869, morniat C 4728, friehan C 1451, or rumeat M 1554; and I find two long forms of frāgon (CM 2417, C 5410) and 20 of tholon, for which his figures are one and 19; on the other hand, he counts scadoian C 279 ‘shade’ as belonging to an ō-verb *skadon, but I believe it is rather to be interpreted as a spelling for skad(o)wian (M has scadouuan), a ja-verb precisely equivalent to Go. skadwjan.

⁵ Studien über die ō-Verba im Altsächsischen 71–84.

⁶ Sehrt, loc.cit,, objects that oppraiu ‘is a loan word from ... Vulgar Latin oprare, and therefore a later analogical formation after verbs of the first weak conjugation.‘ But I know of no other examples of a Germanic -ajan-inflection being invented to accommodate Latin e-verbs, and I see no reason not to accept oppraiu as a genuine second-conjugation form, mutatis mutandis exactly equivalent to OHG opfarōm.

⁷ Of course, likiu by itself could be a Frisian form, examples of which are frequent in the Lublin Psalter; but oppraiu, with its a (parallel to the a of uundraian, M 2261), can hardly be Anglo-Frisian. The only certain case of ō-inflection in the Psalter is to bedant 28.2 ‘adorate’.

⁸ Northumbrian forms like 3d sg. ind. somnigaþ are of course quite recent and belong to the general disruption of inflectional patterns in late Northumbrian.

⁹ Streitberg suggested, UG 313, ‘Der Grund ist ... wohl der Gegensatz, in dem die 2. 3. Sing. 2. Pl. zu der 1. Sing. Plur. 3. Plur. stehn. * salbojō [sic] : salbōþ = *sōkijō : *sōhīþ.‘ Although I do not believe that this proportion offers an adequate basis for reducing two present-stem types, *ō and *ōji/ōja, to a single hybrid ō/ōja, Streitberg's comparison of the Ingveonic ō/ōja-inflection with that of the first weak conjugation is essentially sound and is what had led me to the explanation of the Ingveonic inflection given in §12.

¹⁰ Mahlow says that 'jedes j vor ... i verloren gehen muss, ausgenommen vielleicht im Anlaut'—so that pre-Germanic *ōji from PIE *āye would become *ōi—and then 'der Diphthong ōi ... wird zu ō contrahiert'. The only disagreement I have with this is that I would expect the immediate result, when the *j of *ōji was lost, to be dissyllabic *ōï, which may conceivably not have passed through a diphthong *ōi on its way to ō. The discrepancy between the development of PIE *āye to Gmc. ō and that of PIE *ēye to Gmc. ai which seems to have occurred in the prehistory of the third weak conjugation is not a serious counterargument, and in any case it is the development of *ēye to ai, with change of timbre from to , that is phonetically hard to account for.

¹¹ This formulation expresses the end result. Quite possibly *j was at first lost after all unaccented vowels, including *i, and then a new [ḽ] arose as a glide between *i and a following low or back vowel. The explanation of Ingveonic ōja given in §12 requires that the *j after *i be phonemic in prehistoric West Germanic.

¹² Paul actually writes the etymon of Go. bairau ‘feram’ as *bharajm; I have modernized the spelling. The laryngeal in the ending does not bring Paul's theory into conflict with the hypothesis that the Germanic Verschärfung of *j and *w was caused by adjacent laryngeals (Lg. 34.203 ftn. 1), since the Verschärfung occurs only immediately after the first vowel of a word.

¹³ OS dōmi and OHG tuomi are apparently analogic to the short-stem ja-verbs.

¹⁴ Evidently after the short -stem ja-verbs, type OHG hevis, hevit, with PGmc. *i < *ji < PIE *ye by Mahlow's rule. (Go. hafjis, hafjiþ have been analogically remodeled.)

¹⁵ The imperative *-i would be phonologically regular if PIE *-iye and *-eye became *-ī, not *-ĩ, in Proto-Germanic, or if PGmc. *-ī and *-ĩ both became *-i in Proto-West-Germanic. I know of no decisive evidence for or against either of these hypotheses; relevant forms seem all exposed to analogic remodeling.

¹⁶ I doubt that the short-stem ja-presents contributed much to the transformation of the ō-presents, since the West Germanic consonant gemination had introduced an additional complication in their stem alternation. To a 3d singular like *framiþ ‘brings forward’ the o-grade alternant, *frammija-, would be formed by adding ja and doubling the stem-final consonant. (Or, if Sievers' Law was no longer operating, the o-grade form would be *frammja-, and the influence on the ō-presents even less.)

¹⁷ The relative chronology presented here is not crucial to the argument. I put the loss of *j after the change of *e to *i because the outcome of PIE *eye in Germanic is ī, not *ē OIc. &thorn;rír, Go. gasteis, frawardeiþ), and after Grimm's and Verner's Laws because the differing treatment of *j after vowels of initial and noninitial syllables is most naturally explained as a result of the Germanic initial accent, which presumably developed later than the operation of Verner's Law. Any historical connection between the loss of intervocalic *j in Germanic and the very similar developments in Italic and Celtic must have involved interaction between already well differentiated dialects.

¹⁸ It is important to remember that what I write ō was phonemically and phonetically a:/ [a:] until well after the Proto-Germanic epoch (for the phonetics witness Go. Rumo- neis* from Lat. Rōmānī, a loan that can hardly be older than the expedition of the Cimbri and Teutoni at the end of the 2d century b.c., and Caesar's Silva Bacenis ‘Beechwood‘). Hence there is no difficulty at all in the development of [a:a] to [a: :]; and that of *ōai to õ categorically rejected by Wilmanns, 3.86) is no more difficult than that of *ōï to õ: [a:aḽ] would have become [a:ḽ] (hardly [a::ḽ]), and then the long diphthong would have lost its off-glide. Cf. the Attic sequence Athēnáai > Athēnâi > (Hellenistic) Athēnâ.

^18a Indeed, the Germanic optative is based entirely on the Proto-Indo-European optative (cf. §10 for the 1st singular active of Gothic and North Germanic) ; the often asserted falling together of the Indo-European subjunctive and optative in Germanic is a myth. The only Germanic verb form so far identified with any plausibility as an old subjunctive is the imperative Go. ogs ‘fear’ (used without ni R 11.20, 13.4). But cf. now Karl Hoffmann, Münchener Studien zur Sprachwissenschaft 7.91 (1955) : Go. qimiþ : Skt. gámat.

¹⁹ Replacement of the 2d plural indicative and imperative by the optative (OIc. laþeþ in place of *laþaþ) simply follows the same development in the simple thematic and short-item ja-presents (cf. Prokosch, CGG 212), where the consonant clusters resulting from syncopation of both vowels of PGmc. *-idi led to evidently intolerable morphophonemic complications.

²⁰ Wissmann, 199, would add the influence of the nō-presents. But I doubt very much that Proto-Germanic—let alone West Germanic—still had athematic continuations of the Indo-European *nā/nₔ-formation. Where WGmc. dō- ‘do’ comes from is a question which fortunately does not have to be answered here.

²¹ But the agreement of OS 1st sg. lađon with OHG ladōm is no evidence for this second possibility, since we do not know whether the form replaced by OE laþiu etc. ended in *-ō or *-ōm(i). The ending of the pronominal dative singular masculine and neuter in Old Saxon shows a similar mixture of Ingveonic and non-Ingveonic, -m agreeing with Anglo-Frisian and -mu with High German.

²² In the following I base myself principally on Hermann Flasdieck's exhaustive study of the third weak class, Anglia 59.1–192 (1935).

²³ Or rather, since Germanic *ai and *ē had probably already fallen together in medial syllables, read *libēþ, Hibēda, *libējan.

²⁴ Note also that the inherited ai/ja-alternation already fitted in pretty well with the alternation of e- and o-grade in the simple thematic and ja-presents, whereas the uniform ō inherited in the second weak conjugation did not.

²⁵ If prehistoric ā-presents contributed at all to the Germanic second weak conjugation, their inflection was completely assimilated to that of the āye/āyo-presents. Germanic forms cannot be used as evidence for a Proto-Indo-European ā-inflection.