26 EASTER ISLAND. 



when they reduced the language to writing; it has swept over Samoa 

 since the corresponding period, and is too powerful a force to be stayed 

 by the efforts of the teachers. This as yet evades explanation, it stands 

 as an anomaly. Yet by way of comparison we are able to discover a 

 very few instances where in secular mutation a Latin t has become k 

 in descendant languages, which, it will be seen, is not an exact parallel.* 



A word also remains to say as to the aspiration. Few students of 

 phonetics admit it to a consonant place, yet it is clearly not a vowel. 

 Whitney sets it to one side in his classic table of the alphabet which 

 in other respects we have been following. It is as though a detail of 

 composition, which an artist had found it difficult to dispose of on the 

 canvas, were painted on the frame. Despite this doubt I have had no 

 hesitation in establishing two aspirates and in assigning them to posi- 

 tions within the table of the alphabet; but because I can not identify 

 any part played in the formation of these aspirations by the three 

 consonant-forming organs I have set the two Polynesian aspirates not 

 quite in the lingual and labial series, but proximate thereto. The exist- 

 ence of the duality of these aspirates is readily to be established in this 

 language family. In the lingual series h is the mutation terminus of t 

 and of 5 ; and in the labial series h is the mutation terminus of v, of/, of p. 

 Yet when we find an h carrying on to secondary development a word 

 which at last resumes its former estate, this portative h does not carry a 

 lingual over into the labial column or a labial into the lingual series ; the 

 aspirate delivers properly that which it has received. This could not 

 be the case if the h resultant from lingual mutation and the h resultant 

 from labial mutation were indistinguishable by the people who speak 

 these languages. An exception, a case in which an error in delivery was 

 really found, is so unusual that I discussed it at length in "The Polyne- 

 sian Wanderings," page 287. 



Throughout these languages runs a consistent principle of word 

 mutation quite independent of the mutation by consonant modification. 

 In this principle the word is subjected as word to a mutation which is 

 governed by other than the simple phonetic laws applicable to conso- 

 nant variety. This principle is metathesis, which in Polynesia is far 

 more cogent than apparent. In Rapanui I have noted but thirteen 

 instances of metathesis, involving twelve words, a very minute per- 

 centage of the 3,000 principal entries of this dictionary. In the 

 dictionaries of other languages of this family this more or less com- 

 plete disguise of familiar words is equally rare of record. 



In the fact metathesis is very prevalent ; it is constantly met with in 

 the speech of these islanders. The reasons for the paucity of its dic- 



*One of these instances is the word busk, now obsolescent, which is derivable from the 

 Middle Latin bustutn. The Latin original, itself of uncertain etymology, affords us in 

 forking channels bust in anatomy and busk for the bodice whereby anatomy is tolerated 

 in modest society within the temperate zones; near the equator the distinction is far less 

 requisite. 



