A PRIMEVAL MOTHER TONGUE 



point of fact the case is very different from this. 

 We know that French and Italian are differ- 

 ently modified forms of Latin, because we can 

 trace the modern words directly back to their 

 ancient prototypes, and verify by the aid of 

 written documents their various changes of form 

 and meaning. After carrying on for a while this 

 process of comparison, we find that the modern 

 words vary from the ancient according to cer- 

 tain well-defined rules, which are different for 

 French and Italian, but are singularly uniform 

 for each language. So unmistakable is the reg- 

 ularity of the system of changes, that if all 

 record of Latin were to be swept away we might 

 still reconstruct the language from a compara- 

 tive study of its modern descendants. Mois and 

 mese, for example, the French and Italian words 

 for " month," would give us the Latin mentis, 

 and nothing else; and so on throughout. In 

 similar wise, although the Old Aryan language 

 has left no written documents to tell us of its 

 grammar and vocabulary, we have nevertheless 

 detected such a regular system of phonetic 

 changes among the languages which have de- 

 scended from it that we have been already ena- 

 bled to go some way toward reconstructing this 

 extinct tongue. Month and mentis, for example, 

 carry us back, with little less than absolute cer- 

 tainty, to an Old Aryan mans a ; and so on as 

 before, though here the inquiry is an abstruse 

 137 



