36 THE POLYNESIAN WANDERINGS. 



had to deal with this matter in the consideration of the phonetics 

 of the more strictly Polynesian languages. The same holds true, 

 mutatis mutandis, of these two Melanesian tongues. I cite the 

 earlier discussion of the subject : 



A man with a quick ear and an obedient tongue may, as the result of 

 long discipline, acquire almost perfect use of the Samoan consonants, but 

 it is most probable that no Caucasian has really mastered the art of the 

 Samoan vowels. It is as in their music : the intervals, the supertones, and 

 the fractions of the tone are developed on a system which we find it 

 impossible to acquire. It establishes a new group of units of vibration 

 of the vocal cords, for which the fundamental diapason of our own speech 

 is not set in unison. 



With this in mind, we shall find a plain explanation of the central triangle 

 of the vowel changes if we regard the short a, e, o as merely so many 

 approximations to a primal obscure short vowel which lies centrally situ- 

 ated in respect of these three apical points. One congeries of the Polyne- 

 sian tongues may have had a vibration series and period which inclined 

 its use of the primal obscure vowel somewhat in the a direction ; to another 

 congeries the e component was the more grateful; to yet another the 

 tendency was in the o or labial grade. * * * Thus we have no hesitation 

 in taking this central triangle of d-e-o out of the group of vowel changes 

 in Samoan, of regarding it as no more than a doubly muffled rendering of 

 a single central sound, and of removing it entirely from consideration 

 among the criteria of vowel changes as dialectic indicia. 



When we diagram upon the common alphabetic scheme the vowel 

 changes not in this class, we find some interesting developments. 

 Along the palatal strut, that which rests upon i and peaks in a, 

 in the data thus compared we find but a single instance of the 

 dialectic vowel change a-i in bila (12) to pick up Viti vili, and of 

 i-e in seri (22) to loose a tabu Viti sereka to untie; and the weight 

 of these instances is considerably lessened by the fact that they lie 

 in an unaccented syllable, and a terminal one at that, unaccented 

 terminal vowels in Viti being evanescent. 



On the labial strut, from a to u, most of these vowel changes are 

 seen to lie. The maximum frequency, four instances, is found 

 between o and u, the change u-o is seen in bure (14) to wash Viti 

 mborea; lume a (19) to dip Viti lomotha; and suba (33) to break 

 Viti sovetaka; the change o-u in mako (41) offspring Viti makumbu 

 grandchild. A change over a slightly longer interval, o-a, is found 

 in trbm (38) turmeric Viti ndamu, red. A still longer gap is found 

 in a-u in bera (27) to crumble Viti vuruvuru; and in the reverse direc- 

 tion, u-a, in mutrei (5) breadfruit cake Viti mandrai. 



There remain three changes which vary from the foregoing simple 

 system of vocalic mutation along one or the other of these struts, 

 for they cut across diagonally. Of the change e-u we have two 

 examples, in tefa ki (24) to range Viti tuva, and in bera (27) to 

 crumble Viti vuruvuru. The change i-o is found in bo ri (28) to 

 break Viti vorota; and u-i in lubwa (30) to pour out Viti livia. 



