stitute one of the largest linguistic families of the world; and because of the many exciting problems connected with them they also belong to those linguistic groups which will undoubtedly draw the interest and attention of linguistic science of the future.

The disintegration of the Proto-Dravidian parent languages (already in PDr there must have existed distinct dialectal differences) and the separation of South Dravidian seems to have taken place well before the 15th century B.C. The separation of the Telugu (-Tuḷu) and allied dialects (the division into the Āndhra and Drāviḍa groups of ancient grammarians) might have taken place between the 9th-6th Centuries B.C. Before the 5th Century, PSDr was probably still a more or less uniform speech though with at least two strongly marked dialects; they were the Proto-Kannada, developing later into Kannada (with Baḍaga) and Koḍagu, with some relation to Kota and possibly Toda, yet to be established in detail, and Proto-Tamil, developing later into Old Tamil and possibly Toda, and still much later, during the Early Middle Tamil period (roughly 5th-10th Century) into Malayalam.

At the end of this early period, probably sometime in the 3rd-4th Century B.C, the final split of Proto-Tamil and Proto-Kannada probably occurred. As far as Toda is concerned, it seems that the pre-Toda speakers separated from the pre-Tamil speakers at the close of this period.

During the immediately following age, the development began of Proto-Tamil into pre-Tamil and finally Old Tamil, our first recorded stage of any Dravidian language. The final stage of the process of Kannada-Tamil split, and the Old Tamil literature were accompanied by conscious efforts of grammarians (*Agastyam, Tolkappiyam) and a body of poets and connoisseurs to set up a norm of standard literary Tamil language, the so-called centamil, the ceyyuḷ. At the same time, however, the colloquial speech (valakku) developed according to its own lines and from time to time found entrance into literature and inscriptions. It seems that we have to distinguish, for this very early period itself, at least three forms of Tamil: first, the written language of literature, the earliest texts of which seem to have been some poems occurring in the Puranānūru and Kuruntokai collections of the so-called Sangam anthologies, and the brilliant descriptive grammar known as Tolkāppiyam; second, the language of the inscriptions: the earliest data about epigraphic Tamil may be gatĥered from the curious hybrid language of short Brāhmī epigraphs, from the graffiti on Arikamedu ware, and, finally, from the earliest inscriptions in Tamil "proper" (the Tirunātarhunra and Paḷḷankoyil inscriptions from circa A.D. 400 and 550 respectively); finally, the informal day-to-day speech of the people, which