tion’, Ir. bara: Lat. ferio.—W. credaf ‘I believe’, Ir. cretim (t ≡ d‑d) < *kred d‑: Skr. s̑rad dhā- ‘confide, believe’.
Ar. *k̑red dhē- lit. ‘set (one’s) heart (on)’ was not a fast compound (cf. Skr. s̑rád asmāi dhatta ‘believe in him’); thus the W. credaf is explained by the d‑d coming permanently together in Kelt. (for Ar. d‑dh > W. th § 91 ii), Brugmann² I 670, 691. Lat. crēdo is also irregular, as if *dō ‘give’ had been substituted for *dhē ‘put’, Sommer 251.
When d‑d came together later in Brit., they seem to have been simplified to d giving W. ẟ, as in aẟysg ‘education’ < Lat. addisc‑; so W. aẟef ‘home’ < *ad-dem‑, √demā- § 91 i.
(2) The change of the first t in tt to the affricative tˢ was perhaps due to the tendency in Ar. to avoid double consonants, which in other cases seem to have been simplified. Gemination however was a special characteristic of diminutives and hypocoristic or pet names, and of child-language, which was in a sense a language apart ; and in these even tt remained unchanged. Thus Gk. Νικοττώ (for Νικοτέλεια), Δικκώ, Θεοκκώ, Φίλλιος, Κρίττις, Σθέννις, O. H. G. Sicco (for Sigerīch or Sigbertus), Lat. Varrō (beside Vārus), Brit. Commios (beside Comux, Gaul. Comus), W. Iol-lo (with double l in Ml. W. § 22 ii, for Iorwerth), Gutto (for Gruffuẟ);—Gk. ἄττα, Lat. atta ‘papa’; Skr. akkā ‘mama’, Gk. Ἀκκώ, Lat. Acca Lārentia (: W. y nawfed ach ‘the ninth degree of consanguinity’, lit. ‘the ninth *mother’, cf. “the 4th mother” § 123 v; ach ac edryd ‘descent’, lit. ‘*mat- and pat-ernity’; achoedd, achau ‘lineage’). As the above examples show, the habit of doubling in such forms persisted in new creations, and may account for the qu̯qu̯ in the ogam maqu̯qu̯i, and for the tt in Brit. *genettā > W. geneth r.p. 1359 ‘girl’. So in tribal names: Brittones beside Britannī; Gallī beside Γαλάται. Also in names of animals: Lat. vacca; W. bwch ‘buck’ (ch < kk), Skr. bukkas id.; Gaul. cattos, W. cath; Ml. W. buch ‘cow’ < *boukkā; W. mochyn ‘pig’, Ir. mucc, Germ. dial. mocke ‘sow’; Ir. socc, W. hwch ‘pig, sow’; O. E. dogga ‘dog’; Persson, IP. xxvi 68.
The Spirants.
§ 94. i. Ar. s was of very frequent occurrence. It remained generally in Pr. Kelt. Initially Ar. s before a vowel (Lat. s, Gk. ῾, Germ. s, Lith. s, Skr. s) appears in Ir. as s‑, in W. gene-
