In Language 3. 109-22 (1927) I undertook to show that original bh initial yielded Hittite h. It was already fairly clear that the other IE voiced aspirates (dh, 'h, gh, and gvsh corresponded to Hittite stop-sounds, and that IE bh itself in the medial position appeared as p in Hitt. nepts ' sky' (see op. cit. p. 121). Under such circumstances the thesis that bh initial yielded Hitt. h could be established only on the basis of a large number of evident etymologies, and, while those suggested in my article were numerous enough, few of them were really attractive. Friedrich' was undoubtedly right in thinking that I had not proved my case, and the three additional etymologies suggested in Lang. 4.159 did not help matters much. In fact, a better understanding of Hitt. phonology and morphology shows that several of the proposed etymologies are impossible. Hitt. hurt-' curse ' cannot be connected with Greek crept, since litt. u must correspond to IE u, eu, ou, or au.2 Hitt. halza'call' cannot come from **bhels-, since is remains unchanged in kulganzi 'they observe', etc. The final consonant of the verb stem hatkc'close' is certainly not a suffix, and so the word cannot be connected with the IE root *bhat'strike' (whose existence is besides rather doubtful). There are minor phonetic difficulties with several of the other etymologies suggested; but more serious is the fact that not more than two or three of them are really attractive. I am now convinced that initial bh, like all the other aspirates, yielded in Hittite the corresponding voiceless stop. This thesis is in itself much more plausible, and the following etymologies, which it makes possible, are far more attractive than those which it displaces. Hitt. parkus' 'high' clearly belongs to the IE root *bher'h, 'high' (Sanskrit brhant'high,' etc.3) Hitt. pankus ' all, whole' corresponds sound for sound with Skt.