the preparation and publication of collected material. The Wailaki and Tolowa in northwestern California will, when published, present very important linguistic material. The Yukon dialects are practically unknown, with the exception of Ten’a[1]. While, without this at present unavailable Athapascan material, the final word on the subject of a genetic relationship between Tlingit and Athapascan cannot be said, some useful comments and comparisons may be made. These indicate rather clearly what may be anticipated as the final decision on the subject.
In some respects the material to be compared presents unusual opportunities. Both Tlingit and the Athapascan languages have a rather large number of monosyllabic nouns, and the larger number of these are apparently simple and primary. The phonetic changes possible are therefore simplified and reduced in number; for the action of word-accents, both of stress and pitch, are eliminated. Phonetic changes should therefore proceed with unusual regularity. Simple nouns like these present great advantages also in the matter of stable and easily-determined meanings. In the case of Athapascan ca sun, we have a memory association tying a simple phonetic group with a definite single object. In most other instances there is opportunity for varying ranges of application. The word t‘u water may come to be applied to lake and ocean; but, aside from an expansion or contraction of application, a change of meaning in the majority of such simple words, so complete as to make an original identity of form and meaning in the parent language untraceable in the descendants, is not likely to happen. The known history of Indo-European languages shows that certain classes of words — such as numerals, body-parts, and terms of relationship — are particularly stable.
What appears to have happened in the Athapascan languages is that monosyllabic, non-descriptive nouns have been gradually replaced by longer, descriptive terms. A sufficient number, however, of these simple nouns remain in the various languages to furnish a fair basis of comparison. For Tlingit, Professor Boas has furnished upward of three hundred simple nouns. When the Athapascan nouns of identical or closely related meanings are placed beside these Tlingit nouns in parallel columns, only a few words are sufficiently alike to attract attention[2]. With the Tlingit words arranged alphabetically, phonetic sound-shifts between Tlingit and Athapascan, if present, should appear at once. No such shifts are found after careful study.
There are two relationship terms similar in form and of identical or allied meaning. In Tlingit, mother-in-law, is tcan, and in Beaver, tcon. Tlingit ’at‘, father’s sister; and -at in the Athapascan dialects of Northwestern California, where the meaning is elder sister or cousin. This term, in these dialects, is also applied to father’s sister. Were it not for this anomaly in classification, the meanings of the words would not in the least coincide. A connection is possible if it be assumed that in Athapascan the term for father’s sister came to be applied to elder sister. If the change was from elder sister to include father’s sister, the connection in meaning disappears.
A connection might be assumed between Tlingit wan edge, and Ten’a -vwon edge or border, if a parallel of w=vw could be found. The Athapascan sound which appears with a queer alternation in various dialects as b and m, becomes vw, a bilabial spirant, in Ten’a. Without other correspondences, nothing can be made of this single instance.
Perhaps the most striking correspondence
