234 J. HORNELL ON 



In considering the suggestion I make that a Polynesian immigration into India 

 occurred prior to or soon after the arrival of the Dravidians, the fact should not be for- 

 gotten that the Malayo-Polynesian settlement of Madagascar antedated the spread of 

 Hindu influence in the Malay Archipelago (in the early centuries of our era), for there 

 are said to be no Sanskrit words in the Malagasy languages. To pass from Sumatra 

 to Madagascar in numbers sufficient to settle a country and this not once but in 

 several successive waves, is difficult to credit, but if we intercalate India with its 

 dependency of the Maldive Islands, the difficulty diminishes to reasonable dimensions. 

 Against this Sir H. H. Johnston' has urged that they have left no traces of their pas- 

 sage, but this argument is, I think, invalidated now that I have shown how extensively 

 Polynesian boat designs have survived on sections of the Indian coast line, taken in 

 conjunction with the round-headedness of certain of the most numerically important 

 castes in the same localities, which I have recently discovered. 



I append tabulated measurements of men belonging to four of the principal castes 

 in Tinnevelly, — Vellalar, Parawar, Shanar and Paraiyar, — from which it will be seen 

 that the three latter which are sub-brachycephalic, agree very closely in the average 

 cranial index, whereas the Vellalas approach much more closely to the index accep- 

 ted by Thurston and others as representing the typical Dravidian. I also give 

 measurements of a number of men from the Tanjore coast. Comparison of the latter 

 with the former shows these representatives of the Tanjore population to contain much 

 less of the round-headed strain. The Vellalas of the two districts evidently have the 

 same ethnic origin ; the sub-brachycephalic Parawas of the Tinnevelly coast, a caste 

 not found in Tanjore, are possibly represented there by Ahambadians and coast 

 Muhammadans or I.,abbais (both occupied with fishing and fish-curing at the places 

 visited) with indices respectively of 79- 9 and 8o-i. Conversely the Sembadavar fisher- 

 men of Adirampatam are distinctly sub-dolicocephalic with an index of 76 03. These 

 men, however, are not usually deep-sea fishermen ; they are the principal users of the 

 casting net and fish chiefly in rivers and backwaters, a fact which, when taken in 

 conjunction with the shape of their heads, points to a Dravidian origin with its con- 

 comitant of fishing customs derived from the Tigris, Euphrates and Nile. 



The Shanars present several very curious divergences from the other castes 

 named. In Tinnevelly district they form a very important section of the popula- 

 tion ; their occupational calling is Palmyra palm cultivation ; in Travancore their 

 kinsmen, the Izhuvans, cultivate the coconut. As will be seen from the tables 

 appended, the cranial index of the 100 individuals measured was 807; the maxi- 

 mum was 90-3, the minimum 71 ; as many as 60 were 80 and over. They are de- 

 cidedly broader headed than the Parawas ; more stoutly built than the latter, fre- 

 quently short and thick-set, with an entirely different caste of features to the typical 

 Parawar type. Among the latter curly hair is not uncommon; among the Shanars 

 the hair of the head is coarse, black, and straight. In features considerable variation 

 is seen ; one well-marked type can, however, be made out as predominant. In this the 



1 Living Races of Mankind, Vol. II, p. 420. 



