218 METHODS AND KESULTS OF ETHNOLOGY iv 



upon the " avidity with which the inhabitants of 

 the polyglot islands of Melanesia, from New 

 Caledonia to the .Solomon Islands, adopt the 

 improvements of a more perfect language than 

 their own, which different causes and accidental 

 communication still continue to bring to them ; " 

 and he adds that " among the Melanesian islands 

 scarcely one was found by us which did not 

 possess, in some cases still imperfectly, the decimal 

 system of numeration in addition to their own, in 

 which they reckon only to five." 



Yet how much philological reasoning in favour 

 1 of the affinity or diversity of two distinct peoples 

 'has been based on the mere comparison of 

 numerals ! 



But the most instructive example of the fallacy 

 which may attach to merely philological reason- 

 ings, is that afforded by the Feejeans, who are, 

 physically, so intimately connected with the ad- 

 jacent Negritos of New Caledonia, &c., that no 

 one can doubt to what stock they belong, and 

 who yet, in the form and substance of their 

 language, are Polynesian. The case is as remark- 

 able as if the Canary Islands should have been 

 found to be inhabited by negroes speaking Arabic, 

 or some other clearly Semitic dialect, as their 

 mother tongue. As it happens, the physical 

 peculiarities of the Feejeans are so striking, and 

 the conditions under which they live are so 

 similar to those of the Polynesians, that no one 



