MICHELSON AMERICAN INDIAN LANGUAGES 229 



show only a moderate amount of common lexical and morpho- 

 logical resemblances. In some cases, such as Uto-Aztecan, it 

 must be conceded that the burden of proof is now definitely 

 on those who maintain that the two "stocks" are true distinct 

 stocks, and not differentiations from a common ancestor. It is 

 on the other hand equally certain that the genetic connection 

 has not been established with absolute nicety. In the same man- 

 ner Athapascan, Koluschan, and Skittagetan are almost certainly 

 genetically related. Their morphological resemblances are so nu- 

 merous and so special that it is incredible that they are due solely 

 to borrowings, and not to genetic relationship. However, the 

 amount of lexical material the three have in common is an ex- 

 ceedingly small percentage of their total vocabulary. The 

 extremely large percentage of the unexplained lexical material 

 forces us to admit that this has been derived from outside 

 sources, and with our present definition of "stock," a purely 

 genetic relationship between the three breaks down. After all, 

 our difficulties all hinge on our definition of "stock," and the 

 proofs necessary to show that one or more languages constitute 

 such a "stock." Though the definition of "stock" given above, 

 may be rigorously correct, the actual application of it would 

 practically obliterate the total number of "stocks" in the world, 

 and we should be worse off than ever. For in that sense, there 

 are few, if any, languages which constitute a stock. In the 

 writer's opinion philologists have taken over biologists' concepts 

 without inquiring whether they are suitable to their own science. 

 "Stock" must be redefined in a way that has some real meaning, 

 and some term or terms invented to cover those larger groups 

 which apparently are only remotely related, which may be re- 

 lated in sounds, morphology, syntax, and vocabulary, but not 

 all combined as a unit. It goes without saying that the 

 nature of the proofs then demanded will be in accordance 

 with our definitions. 



To revert to a point brought out above. If the morphology 

 of 'anguages can not be borrowed, with scrupulous nicety we 

 must assume an enormous number of distinct stocks at the very 

 dawn of man, which certainly is not plausible. If on the other 



