The study of Sanscrit has shown us that two thousand years ago it occurred to Hindoo grammarians to investigate the origin of the words of their language, when they discovered that all words could be reduced to roots, and that these roots all expressed various forms of activity; that they were therefore verbs, and that the number of these roots was very restricted. Our present philologists have continued this work and are not only able to acknowledge the accuracy of the Brahmanic discovery, but also to certify that the grammatical analysis of the Hindoos, put forth 500 years before our era, has never been surpassed. It is important to remember that roots are the fundamental elements which permeate the whole organism of the language. Hebrew has been reduced by Renan and other Hebraists to about 500 roots; the work has still to be done for the whole Semitic family. The same process has been carried out with regard to the Aryan languages; we find the number of roots in Sanscrit reduced to about 800; of Gothic about 600; rather more than 400 in the Teutonic family, and 600 in the Slavic. The Ural-Altaic languages have also undergone a partial analysis of the same kind, and the result at present corresponds to that obtained by the examination of the other families. After eliminating the tertiary and secondary roots from the Sanscrit the residuum is 600 or 500, and we arrive at the fact that this entire language, and, in a great measure, all the Aryan languages, can be traced back to an extremely small number of roots.