kthnolooy of the !NBO-pacipic islakds* 9 



it woulii Hpp^ar to have been possessive ( and oblique), aa it ianow 

 foil ml in ali tlie Southern Jialects, save Tclugu, ia those cases, or 

 as the agentive postfix to verlg, which ig radically possessive. In 

 Telugii, by a diaiL»ctic variation, it oce ill's only in the nominative, 

 tlie oblique cases taking tlie primary a. Tlie Northern diutects, 

 Uraon, Male, in their preference for e, follow Teltigu, or more 

 probably the Southern Tiiluva, which has other special affinities 

 with the Kortliern dialects including the Kol. It is probable 

 from this that a (sometimes varied to o) was the proper nomina- 

 tive vowel, and that the substituiion of the possessive e for it was 

 a dialectic variation which spread from Telugn or Tulnva lo most 

 of the Northern dialects, or was internally produced by the loss of 

 the ideologic distinction between ihe two forms. It is clear that 

 the use of e in the po^cssi\re like that of m in the plural bctongi 

 to a very archaic condition of the foriiiation or some of its brdiuches. 

 It ia not probable that in any single branch there were originaSIy 

 two modes of indicating the pliinds atid possessives, and it is still 

 le^ss probable that both admilttid of being combined. When we 

 now find such combinations it is to be inferred that one of t!ie 

 particlus is primuiry and other secondury, the combin;itio;i>j 

 having been produced by the blemlijiig ot" a foreign system of 

 postfixes with the Dravirian or of two Draviriun systems previous- 

 ly characteristic of dilferent bmnches of the formation. The anti- 

 quity and wide prevalence of the ordinary jdurul particles in /, r 

 ^e are proved by their occurrence not only in South Dravirian, 

 Kol and Gungelico-Ultraindian languages but in Asonestu. lint 

 one branch may have originally possessed labial plurais. The 

 possessive in e whether poslfixaal or flexional must hava |»receded 

 the use of the superadded possessive postfixes. The most probable 

 explanation afforded by ihe Dravirian particle system by itself is 

 that the pronominal root na took the archaic possessive in i (ii], ia 

 &c South Dravirian, Kol) and that tliis bccamo e by the coales- 

 cence of the root vowel a with the definitive vowel i (na'T7t=ne«). 

 But even the current possessive bas sometimes e. Thus in Tamil 

 we find ei, in Malayalum ye, in Dhimal eug Sec. 



The u ol the *2ud pronoun can hardly be explained as a merely 

 phonetic variation of the radical L In the Anc. Tamil it occurs 

 in the full form nu- ui tlio possessive plural only nu-wa-r/tf, the 



