118 THE SUBANU. 
fuan was transformed to funa and thence in compensation to funga. 
As an open stem this would tend to permanence, while fuan must in the 
course of Polynesian speech-growth slough off its final consonant. 
23. ia he; Visayan Sia, siya id. 
ia Samoa, Tonga, Futuna, Niué, Uvea, 
Rotuma, Fakaofo, Marquesas, | iya Malay. 
Rapanui, ‘Tahiti, Mangareva, | sia Sulu, Visayan. 
Rarotonga, Manahiki, Maori, | siya Tagalog, Visayan, Bontoc Igorot. 
Hawaii, Aniwa. hia Kayan. 
koya Viti. ya Pampangas. 
Here it suffices to note the substantial identity of these forms. 
This and the other pronouns will better repay study when grouped for 
examination in relation to the theory of evolution from position desig- 
nations which I have advanced in a paper on “Root Reducibility in 
Polynesian”’ (27 American Journal of Philology, 369) and which I shall 
prosecute more exhaustively in writing the comparative grammar of 
this family of isolating languages. 
24. ikan fish; Subanu, sora, seda id. P. W. 350. 
ika Fakaofo, Tonga, Futuna, Niué, | ackan 
Uvea, Moiki, Nuguria, ian 
Sikaiana, Maori, Marque- 
Silong. 
Lariko, Wahai, Gani, Saparua, 
Ahtiago, Matabello,Ceram. 
sas, Rapanui, Mangareva, | iyan Liang, Morella, Nuf6r. 
Tongarewa, Mangaia, Pau- | iani Batumerah, Awaiya, Caima- 
motu, Rarotonga, Mana- rian. 
hiki, Viti. iano Ceram. 
i‘a Samoa, Rotuma. ein Mysot. 
ia Nukuoro, Tahiti, Hawaii. yano Teluti. 
jikan Borneo. 
ikan Malay, Massaratty, Teor, Ilo- | nyan Tidore. 
cano, Wayapo, Gah, Rum- | guihan Chamorro. 
bia, Bontoc Igorot. nik Uap. 
maran—igan Menado. iwa Java. 
itjan Maronene. ka Kar Nicobar. 
ikani Bouton. ga Central Nicobar. 
ikiani Amblaw. isda Sulu, Visayan. 
The concord of the Malayan affiliates is so preponderating that we 
can entertain no doubt that the stem was originally closed with the 
nasal 1. That we can not identify this closed zkan from any of the 
Polynesian uses of zka is susceptible of a simple explanation. When an 
attributive most strongly inclines toward what we know as the noun 
use, it is not susceptible of modification by the suffixes used to particu- 
larize the employment of the more diffuse attributives; it is lacking in 
the protection to the stem afforded by these additional members, and a 
final consonant drops off and leaves no sign. As ordered in this table, 
the Indonesian affiliates fall into a readily comprehensible series of 
devolution forms. ‘Thisistrueofall but thelastform. Ihave included 
isda in the list in order to complete the record, but it is clearly a dis- 
tinct stem. It affiliates readily with the Subanu seda by metathesis of 
the former syllable, and seda is just as distinctly a mutation of sora. 
IT regard either zsda or seda as primal, but which of these two came first 
we can not discover until a further series of the stem is discovered. 
