Similarly, the data available on the northern Athapaskan languages indicate that the [x]-glide is found only in Chipewyan. Y In Sarci t and k are described as' more strongly aspirated than the voiceless stops of English'8 but are never written [tx] and [kx], 9 in Carrier t and k are written [th] and [kh] respectively,'0 and in Ten'a (an Athapaskan language of Alaska) t and k are defined as' as-pirated surds, sometimes written th and kh'. 1" It should be pointed out that these languages, in all probability, belong to distinct sub-stocks of the northern Athapaskan group.

It would appear from the foregoing that we are justified in deriving the [tx] and [kx] of the Apachean languages from an original PA*[th] and*[kh], respectively. San Carlos probably represents the nearest approximation to the Primi-tive Athapaskan phonemes, in Navaho the tendency towards [tx] and [kx] has only begun to assert itself, and in Chiricahua and Mescalero (mutually intelligible Apachean languages) the change from PA*[th] and*[kh] to [tx] and [kx] is complete. So far, of course, phonetic change alone has taken place. However, a phonemic shift took place in Jicarilla and Lipan. Here, it is to be inferred, the [x]-glide, following a [k] in a majority of occurrences, eventually became strong enough to force a preceding [t] to the palatal position of articula-tion. As soon as this shift occurred, it is evident that the phonemic systems of Jicarilla and Lipan were altered; a pre-Jicarilla-Lipan* t fell together with the hitherto distinct phoneme k.