o2 SISSANO. 
43. malau Awalama. 58. keyama Mukawa. 
44. kasu_, Ffaté. 59. Kelama Galavi, Boniki, Mukawa. 
45. gazu Nggao. 60. rogona Awalama. 
46. hasie Wango. 61. rogoma Taupota. 
47. matiu Roro. 62. roi Hamatana. 
48. madiu Mugula. 63. ruai Petat. 
49. maziu Roro. 64. liye Ambrym. 
50. diwai Duke of York. 65. mokomoko Uni. 
51. andivai Gazelle Peninsula. 66. hiwo Tubetube. 
52. a davai Kabakaul. 67. kabakil Misima. 
53. no Ndeni. 68. abei Barriai. 
54. ono Uni. 69. rubwa Tagula. 
55. ena _ Nifilole. 70. daha Bisapu, Punadm. 
56. vengara Kiviri. 71. kumbau Nokon. 
57. vegara Kubiri. 72. saqereu Nada. 
When discussing this stem upon its identification in Subanu, I drew 
attention to the fact that we here assemble three stems, and it is 
by no means within our present power to prove that they derive 
from a common source. With all the material in this list which we 
may justifiably associate with each stem for purposes of examination 
these three stems are ai (1-34), au (35-43), asu (44-49). Before 
taking up the consideration of these major stems we shall make a 
cursory examination of the other words from this area which do not 
come into association with the triad. In citing these extraneous 
forms for the note of such interassociation as may appear, I adduce a 
central form, but wholly without prejudice of eventual determination 
of source. 
diwai (50-52). These three have clearly one source. They are found 
within a narrow geographical range in the eastern Bismarck Archipel- 
ago, the Duke of York being the group of islands within the narrows 
of St. George’s Channel, the Gazelle Peninsula and Kabakaul being 
within eyeshot on the northeastern promontory of New Pomerania. 
There is no grave phonetic difficulty in associating therewith ruai (63) 
and roi (62); the mutation r—-d is very frequent in these languages, 
and duai is not remote from diwai, and from ruai to roi is a simple step. 
There is no geographical obstacle, for Petat is on the eastern shore of 
St. George’s Channel, and Hamatana in the northern Solomons is 
within the range of recent folk-movement. 
ono (54). Itis not unlikely that no and ena may possess a common 
source. Superficially ono is closely affiliated herewith. Ndeni and 
Nifilole are not widely separated in the northern edge of central Melane- 
sia along a track of migration which I have sought to establish as issuing 
through Torres Straits. Uni is remote, at the extreme bight of the 
Gulf of Papua. No matter how the movement of possible migration 
along that coast be explained, whether according to my theory as 
from west eastward along a main channel out of Indonesia, or accord- 
ing to Friederici’s view as a backward movement from east westward 
as mnigrants coasting New Guinea from its north shore, this stem 
should make its reappearance somewhere between the Gulf of Papua 
