POLYNESIAN RELICS IN MELANESIA. 145 



As soon as we trace the contours connecting the islands whose 

 languages show approximately equal coefficients of Polynesian qual- 

 ity it becomes at once manifest that we are concerned with two 

 great areas. These are so distinct, so widely separated, that it is 

 practically impossible to conceive of them as having the same origin, 

 at least in this Melanesian tract of the Pacific with which we have 

 to do. We shall therefore do well to consider them in detail, each 

 by itself, and to reserve the discussion of their diversity of origin 

 until after that preliminary survey. 



One of these areas embraces southern Melanesia, from the Torres 

 Islands to New Caledonia. In these islands, the New Hebrides form- 

 ing the principal and determining mass, the contours of equal quality 

 extend from northwest in a general direction toward the southeast. 

 Along the axis cutting these contours from northeast toward south- 

 west we find that the coefficient of quality diminishes from the east. 

 In the southern New Hebrides we find two instances of purely 

 Polynesian speech, at Aniwa and at Fotuna, these representing the 

 position of the contour of ioo. It is important to note that Aniwa 

 and Fotuna are the most weatherly of the New Hebrides archi- 

 pelago, that is to say they are the points to be reached by a fleet 

 steering full and bye, the best sailing point of canoes and the only 

 sailing point which gives the helmsman on unknown seas a sense 

 of direction for his course. The three larger islands in this section 

 of the archipelago show scant traces of Polynesian admixture, and 

 the quality coefficient is low; Eromanga 58, Aneityum 46, and Tanna 

 no more than 29. Still farther west the Loyalties make no better 

 showing: Nengone 45, and Lifu 42, while Uea records a Polynesian 

 content entirely of the modern epoch and known to be derived 

 from an involuntary voyage from Uvea in Nuclear Polynesia. 



In the examination of the central and northern New Hebrides, 

 for except in the history of discovery it is not advisable to disso- 

 ciate the Banks Group, we are able to draw contours of 90, 80, 

 and 70 quite plainly. Along the windward face of the archipelago 

 appear spots which might establish the curve of 100, such being the 

 Polynesian settlement in Efate and Mae on the island of Three Hills. 

 At the north the curve of 90 is established on Vanua Lava by Leon, 

 the other languages of that island standing at 40 and the bush 

 language (Alo Teqel) at the lowest mark of 19. Working down the 

 weather aspect of the archipelago, Arag falls but little below this 

 contour and Makura, Nguna and Efate lie above it. Mota in the 

 extreme north shows the same influence as Leon on Vanua Lava. 

 The contours of 80 and 70 are satisfactorily drawn in close paral- 

 lelism with that of 90 and are well established by a sufficiency of 

 points of identification. The high value of Marina 89 in the deep 

 bay of the north coast of Espiritu Santo need not prove as anoma- 



