(viii.) ou, whether original or from eu, when in one syllable became -ū-, probably about 200 B.C., as in dūcō, Old Lat. doucō, Goth, tiuhan, Eng. tow, Ind.-Eur. *deṷcō.
(ix.) ei became ī (as in dīcō, Old Lat. deico: Gr. δείκ-νυμι, fīdo: Gr. πείθομαι, Ind.-Eur. *bheidhō) just before the time of Lucilius, who prescribes the spellings puerei (nom. plur.) but puerī (gen. sing.), which indicates that the two forms were pronounced alike in his time, but that the traditional distinction in spelling had been more or less preserved. But after his time, since the sound of ei was merely that of ī, ei is continually used merely to denote a long ī, even where, as in faxeis for faxīs, there never had been any diphthongal sound at all.
(x.) In rustic Latin (Volscian and Sabine) au became ō as in the vulgar terms explōdere, plōstrum. Hence arose interesting doublets of meaning;—lautus (the Roman form), “elegant,” but lōtus, “washed”; haustus, “draught,” but hōstus (Cato), “the season’s yield of fruit.”
(xi.) oi became oe and thence ū some time after Plautus, as in ūnus, Old Lat. oenus: Gr. οἰνή “ace.” In Plautus the forms have nearly all been modernized, save in special cases, e.g. in Trin. i. 1, 2, immoene facinus, “a thankless task,” has not been changed to immune because that meaning had died out of the adjective so that immune facinus would have made nonsense; but at the end of the same line utile has replaced oetile. Similarly in a small group of words the old form was preserved through their frequent use in legal or religious documents where tradition was strictly preserved—poena, foedus (neut.), foedus (adj.), “ill-omened.” So the archaic and poetical moenia, “ramparts,” beside the true classical form mūnia, “duties”; the historic Poeni beside the living and frequently used Pūnicum (bellum)—an example which demonstrates conclusively (pace Sommer) that the variation between ū and oe is not due to any difference in the surrounding sounds.
(xii.) ai became ae and this in rustic and later Latin (2nd or 3rd century A.D.) simple ē, though of an open quality—Gr. αἴθος, αἴθω, Lat. aedēs (originally “the place for the fire”); the country forms of haedus, praetor were edus, pretor (Varro, Ling. Lat. v. 97, Lindsay, Lat. Lang. p. 44).
19. Vowels and Diphthongs in unaccented Syllables.—The changes of the short vowels and of the diphthongs in unaccented syllables are too numerous and complex to be set forth here. Some took place under the first-syllable system of accent, some later (§§ 9, 10). Typical examples are pepErci from *péparcai and ónustus from *ónostos (before two consonants); concIno from *cóncano and hospItIs from *hóstipotes, legImus beside Gr. λέγομεν (before one consonant); SicUli from *Siceloi (before a thick l, see § 17, 3); dilIgIt from *dísleget (contrast, however, the preservation of the second e in neglEgIt); occUpat from *opcapat (contrast accipit with i in the following syllable); the varying spelling in monumentum and monimentum, maxumus and maximus, points to an intermediate sound (ü) between u and i (cf. Quint. i. 4. 8, reading optumum and optimum [not opimum] with W. M. Lindsay, Latin Language §§ 14, 16, seq.), which could not be correctly represented in spelling; this difference may, however, be due merely to the effect of differences in the neighbouring sounds, an effect greatly obscured by analogical influences.
Inscriptions of the 4th or 3rd century, B.C. which show original -es and -os in final syllables (e.g. Venerĕs, gen. sing., nāvebos abl. pl.) compared with the usual forms in -is, -us a century later, give us roughly the date of these changes. But final -os, -om, remained after -u- (and v) down to 50 B.C. as in servos.
20. Special mention should be made of the change of -rĭ- and -ro- to -er- (incertus from *encritos; ager, ācer from *agros, *ācris; the feminine ācris was restored in Latin (though not in North Oscan) by the analogy of other adjectives, like tristis, while the masculine ācer was protected by the parallel masculine forms of the -o- declension, like tener, niger [from *teneros, *nigros]).
21. Long vowels generally remained unchanged, as in compāgo, condōno.
22. Of the diphthongs, ai and oi both sank to ei, and with original ei further to ī, in unaccented syllables, as in Achivi from Gr. Ἀχαιϝοί, olīivom, earlier *oleivom (borrowed into Gothic and there becoming alēv) from Gr. ἔλαιϝον. This gives us interesting chronological data, since the el- must have changed to ol- (§ 16. 3) before the change of -ai- to -ei-, and that before the change of the accent from the first syllable to the penultimate (§ 9); and the borrowing took place after -ai- had become -ei-, but before -eivom had become -eum, as it regularly did before the time of Plautus.