pl. also; thus Brit. *kárass-o-mmos, *kárass-o-nti giving W. karassom, karassont; from these followed carasoch.
iii. To the 3rd sg. two things happened, (1) It remained unchanged; thus *kárăss > W. *kar, which was extended to karawẟ to distinguish it from the pres.; for ‑awẟ see § 182 iii. The Ir. ro-char implies *karass with short a from the pl.
(2) It was re-formed with the thematic vowel, following the 1st and 2nd sg.; thus *kárāsset; or with ‑a- from the pl. as *kárāssat. Either of these would give W. *karas (caf-as § 175 i (2)). The first gives Ir. carais.
iv. The treatment of ‑ī- stems was precisely similar. The stem-ending in the sg. was *‑īss‑; this survives in the W. 3rd sg. peris. In the 1st and 2nd sg. it was replaced by ‑eis of ‑ā- stems; but in Gwyn. dial. ‑is survives in these persons also. In the pl. the stem-ending was ‑ass‑, as for ‑ā- stems, the ‑a- representing ə, the R-grade of the ‑ē- from which the ‑ī- is derived.
v. (1) Consonant stems formed the aorist with *‑is‑, cf. Lat. vid-is-tis, which developed similarly, and gives W. ‑ỿss- in eisteẟyssant, etc. In the 3rd sg. it appears in W. as ‑es from *‑iss‑at. In the 1st and 2nd sg. it was replaced by ‑eis.
(2) The *‑iss- suffix seems to have intruded into the thematic conjugation; thus Brit. *kóm bere-iss-at > W. kymerwys, kymerws.
vi. The impersonal forms ‑ad, ‑id, ‑ed, ‑wyd seem to have been formed on the analogy of the impf. impersonal, with the vowels of the 3rd sg. aor.
vii. (1) The root-aorist, § 178 iv (3), was treated similarly in Kelt. Thus for the root *qan‑ ‘sing’ the orig. Kelt, root-aor. would be sg. 1. *kan-m̯, 2. *kan‑s, 3. *kan‑t. The 3rd sg. became the stem, and the new tense formed from it was sg. 1. *kantū, 2. *kantīs, 3. *kantet or *kantat. These forms gave W. sg. 1. keint, 2. *keint, 3. kant. To the 1st and 2nd sg. the perfect endings ‑um, ‑ost, § 182 iv (1), were added, § 175 iii (2).—gwant ‘wounded’ from gwanaf < *gwonaf: Ir. gonim, √ꬶu̯hen‑, is probably formed on the analogy of cant. The root *bher‑ has this aor., which survives only in the 3rd sg. in W.; thus W. kymyrth < *kóm bertet or kymerth < *kóm bertat, § 175 iii (1).
(2) Other examples that survived are from roots ending in gutturals: dyrreith b.t. 54 ‘returned’ < *do-(p)ro-rek-t-et, √reg̑‑: W. dyre ‘come!’ § 193 x (8);—maeth b.t. 74 l. 1 ‘nursed’ < *makt‑ < *māk‑t‑, √māk̑‑: magaf ‘I nourish’. The root *u̯ereꬶ‑ ‘work’ had sg. 1. *u̯rek-t-ū, 3. *u̯rek-t-et giving W. gwrith, gwreith; the former occurs in ef gwrith b.t. 26 (? 3rd sg.); the latter seems to occur in gwnaeth [read gw(r)eith] gwynnẏeith gwreith e law b.a. 2 lit. ‘work of vengeance wrought his hand’; but this verb (gwnaf) being in the pres. conjugated like af, this tense was assimilated to the perf. of af, and became sg. 1. gwneuthum. 3. gwnaeth. The quotation shows that scribes changed old gwreith to gwnaeth, the wrong gwreith, viz. the noun, being changed here. In Bret. the old form survived: Ml. Bret. sg. 3. grez.
