904 eepoet — 1884. 



feature approximate to the Mongoloid type of North Asia ; but when it comes to 

 cranial measurement, the Esquimaux with their narrower skulls, whose proportion 

 of breadth to length is only 75 to 80, are far from conforming to the broad-skulled 

 type of North Asiatic Mongoloids, whose average index is toward 85. Of this diver- 

 gence I have no explanation to offer ; it illustrates the difficulties which have to be 

 met by a young and imperfect science. 



To clear the obscurity of race-problems, as viewed from the anatomical standing 

 point, we naturally seek the help of language. Of late years the anthropology of 

 the Old World has had ever-increasing help from comparative philology. In such 

 investigations, when the philologist seeks a connection between the languages of 

 distant regions, he endeavours to establish both a common stock of words and a 

 common grammatical structure. For instance, this most perfect proof of connec- 

 tion has been lately adduced by Mr. R. H. Codrington in support of the view that 

 the Melanesians and Polynesians, much as they differ in skin and hair, speak lan- 

 guages which belong to a common stock. A more adventurous theory is that of 

 Lenormant and Sayce, that the old Chaldean language is connected with the 

 Tatar group ; yet even here there is an a priori case based at once on analogies of 

 dictionary and grammar. The comparative method becomes much weaker when 

 few or no words can be claimed as similar, and the whole burden of proof has to be 

 borne by similar modes of word-formation and syntax, as, for example, in the re- 

 searches of Aymonier and Keane tending to trace the Malay group of languages into 

 connection with the Khmer or Cambodian. Within America the philologist uses 

 with success the strong method of combined dictionary and grammar in order to 

 define his great language-groups, such as the Algonquin extending from Hudson's 

 Bay to Virginia, the Athapascan from Hudson's Bay to New Mexico, both crossing 

 Canada in their vast range. But attempts to trace analogies between lists of 

 words in Asiatic and American languages, though they may have shown some 

 similarities deserving further inquiry, have hardly proved an amount of corre- 

 spondence beyond what chance coincidence would be capable of producing. Thus 

 when it comes to judging of affinities between the great American language- families, 

 or of any of them with the Asiatic, there is only the weaker method of structure 

 to fall back on. Here the Esquimaux analogy seems to be with North Asiatic 

 languages. It would be defined as agglutinative-suffixing, or, to put the definition 

 practically, an Esquimaux word of however portentous length is treated by looking 

 out in the dictionary the first syllable or two, which will be the root, the rest 

 being a string of modifying suffixes. The Esquimaux thus presents in an exaggerated 

 form the characteristic structure of the vast Ural-Altaic or Turanian group of 

 Asiatic languages. In studying American languages as a whole, the first step is to 

 discard the generalisation of Duponceau as to the American languages from Green- 

 land to Cape Horn being united together, and distinguished from those of other 

 parts of the world, by a common character of polysynthetism, or combining whole 

 sentences into words. The real divergences of structure in American language- 

 families are brought clearly into view in the two dissertations of Mons. Lucien 

 Adam, which are the most valuable papers of the Congres International des 

 Americanistes. Making special examination of sixteen languages of North and 

 South America, Adam considers these to belong to a number of independent or 

 irreducible families, as they would have been, he says, ' had there been primitively 

 several human couples.' It may be -worth suggesting, however, that the task of 

 the philologer is to exhaust every possibility of discovering connections between 

 languages before falling back on the extreme hypothesis of independent origins. 

 These American language-families have grammatical tendencies in common, which 

 suggest original relationship, and in some of these even correspond with languages 

 of other regions in a way which may indicate connection rather than chance. 

 For instance, the distinction of gender, not by sex as male and female, but by 

 life as animate and inanimate, is familiar in the Algonquin group ; in Cree 

 muskesin = shoe (mocassin) makes its plural muskesina, while e«fawyM=woman 

 (squaw) makes its plural eskwayivuk. Now, this kind of gender is not peculiar to 

 America, but appears in South-East Asia, as for instance in the Kol languages of 

 Bengal. In that Asiatic district also appears the habit of infixing, that is, of 



