relations of thought which lie at the basis of the syntax of different families of languages, though not the grammatical structures or paradigms themselves, constitute secondary racial characters of the nature of predisposition. There is reason to believe also that accent systems, though originally acquired under persistent climatic, dietic and social conditions, have now come in many cases to be more or less stable, more or less transmittable characters and may have given rise to predispositions of the sort. But besides these primary and secondary characters there is a third sort of racial peculiarities of speech which, no doubt, distinguish one people from another but which are acquired under the influence of the tradition or of the environment, physical and social, and have to be so acquired by the individuals from generation and generation, and which disappear whenever the tradition or environment is changed. This traditional element in a speech constitutes that part of it which is a social tradition and has no ethnic significance in the biological sense of the term. Among these traditional elements of speech, which may be loosely termed tertiary racial characters, I would place the phonetic system (the vowel and consonant system) as well as the grammatical paradigm (including the tense-formatives) of a language or family of languages. But as we have seen, all the racial characters, secondary no less than tertiary, the predispositions no less than the merely traditional elements, are liable to be changed under change of environmental conditions, and replaced by newly acquired or induced characters.
The fact, therefore, stands that different sections of the Bengali people have the capacity of speaking the Bengali speech alike. But I must utter a caveat here lest a wrong anthropological use be made of this philological fact as has been done in so many cases. On the basis of this fact we
