¹ D'Ovidio and Meyer-Lübke, in Gröbers Grundriss 1.693 (2nd ed., Strassburg 1904–6; hereinafter abbreviated as GGr²); Meyer-Lübke, Grammatica storica della lingua italiana 116 (tr. Bartoli and Braun, Torino, 1931; hereinafter referred to as ML-Bartoli); Grandgent, from Latin to Italian 76 (Cambridge, 1927; hereinafter referred to as Grandgent).

² D'Ovidio and Meyer-Lübke, GGr² 1.674; ML-Bartoli 80; Grandgent 48, 72.

³ ML-Bartoli 113; Grandgent 14.

⁴ ML-Bartoli and GGr²., loc. cit.

⁵ ZRPh 24.159 (1901).

⁶ Grandgent 76.

⁷ REW³ §§3234, 5469, 7521.

⁸ Grandgent 158.

⁹ Caix, GFR 1.229 (1878), Origini della lingua poetica italiana 228 (Firenze, 1880); Gaspary, ZRPh 3.623 (1879). Caix seems to have been the first to oppose the theory of the ‘vocale d'appoggio’ in this connection, favoured by Diez and Delius.

¹⁰ MLN 8.97–99 (1893).

¹¹ ML-Bartoli 160.

¹² The theory of -s > -i (Grandgent 77; ML-Bartoli 116) does not seem at all tenable, and Puçcariu's demonstration of analogical origin of -i in all such cases in Italian and Roumanian (Studii si notite filologice, Convorbiri literare 25.211) is preferable. Grandgent's comparison of Andalusian -h is not to the point; in Andalusian and American Spanish, when h̄ (< s) disappears, it is either assimilated to the following consonant (mimmo < mismo) or disappears without a trace (lo biyete < los billetes).

¹³ Grandgent 51 f.—from which, however, we should of course omit consideration of words with -i allegedly from -s as part of this theory.

¹⁴ Cf. discussion by Norman, Italica 14.59 (1937).

¹⁵ ML-Bartoli 113.

¹⁶ Grandgent 14.

¹⁷ Meyer-Lübke and d'Ovidio, GGr² 1.684 (where the theory of ‘vocalischer Nachklang’ is carried to its logical reductio ad absurdum, with the admission that we should normally expect -a in the first conjugation and -e in the second, as in Sardinian ‘Nachklang’; rightly criticised by Förster, ZRPh 22.521 ff. [1898]).

¹⁸ ML-Bartoli 118; Grandgent 79, 160 without recourse to ‘Murmelvokal’.

¹⁹ ZRPh 22.521–525 (1898). The Italian school have not accepted this theory; cf. d'Ovidio's answer, ZRPh 23.213–220 (1899); Merlo, StR 6.69–83 (1909). D'Ovidio's answer, although generally prevalent, is not acceptable; it is based simply on a dislike for considering OLat. forms in connection with Romance philology (cf. such terms as 'rancido', etc., applied to OLat.) and on an objection to -nn- < -n- after short tonic vowel (which, however, is normal in Tuscan, cf. ML-Bartoli 186). D'Ovidio's suggestion of '-ono (whence, by analogy, '-ano), with -o < -unt, taken over from the strong perfect, runs contrary to the normal lines of analogical force. It is from more common tenses such as the present to less common tenses such as the perfect that analogical transference usually takes place; cf. 1st sing, -o from the present to the imperfect, etc. It was for this reason that Baist refused (ZRPh 16.532 [1892]) to consider the final vowel of Sp. soy, OSp. soe as by analogy from the perfect fuí.

²⁰ Pieri's attempt (Riv. d'istruzione e di filologia classica 33.495–497 [1905]) to explain away danunt as showing merely alteration of the root da- with -n- from donare, and the other verbs in -nunt as based on the analogy of danunt, seems unjustified in view of the OLat. attestations of the -nunt forms (e.g. in Plautus) and related phenomena in other IE languages; cf. the discussion by R. M. Bechtel, Language 13.181 (1937).

²¹ Cf. Rydberg, Geschichte des französischen .

²² Cf. the criticisms made of Bovet's hypothesis of a special [Δ]-phoneme to explain andare, aller, etc. < ambulare (Scritti in onore di Ernesto Monaci 243–262 [Rome, 1901]), by Paris, Rom. 31.606–607 (1902); Gröber, ZRPh 26.639 (1902); Salvioni, Arch. Glott. It. 16.209–210.