^{page 253 note 1} For the root determinative -s- cf. Brug. Grdr. ii. § 8 Anm. 2 and -s-lo- ib. p. 186.

^{page 254 note 1} Velar g is printed g in italic type.

^{page 254 note 2} Mr. Darbishire remarks that Lat. pēdo shows an I. E. p- and that there is no law that would voice a preceding breathed consonant; but even assuming an I. E. p- in such forms as vowel + -bzd + vowel, z can be pronounced as a consonant and would then voice the preceding breathed consonant, hence b- would spread by analogy to forms where z was of necessity pronounced as a sonant.

^{page 254 note 3} Mr. Conway points out to me that the old Oscan form of dat. plur. luisarifs (Büch. Rh. Mus. 114 (1889) p. 328), where -fs = original ^*-bh**s, gives an additional proof of pro-ethnic Ital. ^*-bh**s as dat.- abl. instr. plur. suffix.

^{page 254 note 4} Cf. also Umbr. fratrus (? fratfs) karnus (? karfs). Lindsay, Classical Review, ii. p. 277.

^{page 254 note 5} The nom. sg. -és appears to have arisen in proethnic Italic from adjectives in -ēs by the proportion e.g. ^*pub--bhos: pub-ēs = ^*sed--bhos: sedēs. Such an explanation would at any rate avoid the assumption of an I.E. fem, noun in -ēs (Brug. Grdr. ii. p. 397 Anm.) for which there is no evidence outside Italic.