104 THE SUBANU. 
Visayan form. Inthe former group are tabulated the various dialectic 
forms in Polynesian with a statement of their provenience; in the latter 
group are the Malayan forms so ordered as most conveniently to 
exhibit their alterations from the more simple to the highly complex 
type of variety. It has not appeared advisable herein to exhibit the 
mutations of Polynesian stems as found in Melanesia; that problem 
is quite distinct from that with which we are here engaged. For the 
convenience of such as care to examine this theme I have appended, 
wherever I have collated this material, a reference to the page of The 
Polynesian Wanderings where such data have been discussed. I have, 
however, included this third and intervening element in the discussion 
of every such vocable as was not included in the scope of the former work. 
I. afi fire: Subanu gapoy id. 
afi Samoa, Tonga, Futuna, Uvea, Niué, { ai Siwa, Brissi. 
Aniwa, Sikaiana. hai Vaiqueno, Rotti. 
ahi ‘Tahiti,Mangareva,Marquesas, Rapa- | hahi Timor. 
nui, Hawaii. api Malay, Kolon, ‘Tomohon, Solor, 
ai Rarotonga. Battak, Bugis. 
hapi Java. 
afi Bima, Ceram. yap Mysol. 
efi Muna, Matabello. apoi Silong, Champa, Formosa, Matu. 
afo Malagasy. apui Kayan, Madura, Dayak, Tagalog, 
ngafi Guam. Ilocano, Sideia, Bontoc Igorot. 
quafi Chamorro. wha _ Bouton. 
goifi Guam. pepi Macassar. 
aif Gah. puro Bolanghitam. 
yaf Ahtiago, Teor. 
The last of these forms is highly problematical; it is here included 
for the completion of the record, but it has no suggestion of association 
with aft save through the presence of p, which occurs so commonly in 
the secondary Malayan stem. The Macassar epi is in slightly better 
case, for we may regard the prosthetic p as due to attraction of the stem 
consonant, a precisely similar instance being the prosthetic h in hah. 
Bouton wha, if associable with this stem, is a mutilated fragment. 
Three forms, az/, yaf, and yap, exhibit different phases of inversion, 
a structural method which we have already discovered in the discus- 
sion of the Subanu. The remaining forms fall into accord through 
the operation of well-established laws of mutation. ‘These mutations 
are triple in their incidence. The stem consonant f varies in one direc- 
tion to h, in the other to p, even undergoes extinction, variations of 
frequent occurrence in the phonetics of these languages. Prosthesis 
operates through four agents, h, ng, g, k; these, it will be observed, 
run the whole scale of palatal consonants. ‘The final vowel undergoes 
a modification to what may prove a diphthong, a mutation which we 
shall observe again in the study of this material (cf. 4); a change such 
as this is wholly foreign to the spirit of the Polynesian languages, where 
the vowels are of the stoutest constancy, but we note with interest the 
occurrence of the converse in Niué, where weencounter several instances 
in which ae of the stem, essentially not diphthongal, becomes e. In 
