THE MARQUESAS IN THE FAIRWAY TO HAWAII. 135 



Sea, coals to a Newcastle which needs no heat, for it is the commonest 

 shell on the island beaches. But it has been with me all my life, a hun- 

 dred years ago a great-grandfather brought it back with him, a shell 

 from Owhyhee. That was all the story with which it came to me, just a 

 name which now we spell othergate, a name to recall ; and in its cham- 

 bered recesses at the ear the whisper of booming reefs breaking in marble 

 fleece of foam and of the susurrus of the palm. Just a reminder that 

 Polynesia from my beginning has called me with a voice I have never 

 sought to gainsay. The personal digression may be pardoned if it but 

 serve to set the accent on the regret with which are written these notes 

 of the unrepining passage of a race as attractive as the butterflies and 

 scarcely more thoughtful of the responsibilities of life. 



Had Bishop Dordillon but thought to be more specific in his researches 

 the language of the Marquesas might properly be considered under the 

 heads of two dialects. In a loose sense they are to be regarded as the 

 dialects of the southeastern and northwestern groups which compose the 

 archipelago. Phonetically the critical points are marked in the variant 

 treatment of two palatals. Neither dialect retains in its purity the 

 nasal palatal ng; each may elide it; but in general it undergoes hori- 

 zontal mutation in the southeastern dialect into n and vertical in the 

 northwestern into k. In the matter of the mute palatal k there is less 

 distinctive quality; regarded broadly, we may say that the northwestern 

 dialect inclines to retain it, that the southeastern tends to elide it under 

 the influence which we have already found operative in Hawaii and 

 Samoa. The notes of Monsiguor Dordillon and of Mr. Christian upon 

 the variety of k are to be disregarded ; each has fallen into the error 

 of basing comparisons upon the employment of the k as mutation 

 ng-product with the n similarly produced, which can, of course, have 

 no bearing upon the treatment of the true Polynesian k. 



In the exhaustive collation of the vocabulary I have been led several 

 times to the feeling that in the choice of vocables we might in the Mar- 

 quesas, as elsewhere, find circumstantial evidence of variant migration 

 streams. The proof has always escaped me because the dictionary 

 omits the note of dialect distinction. This point, which would be of 

 interest and probably of value, we are obliged to pass over, for it is now 

 too late to hope to recover the dying speech. 



Bearing in mind this irrecoverable difference of two strongly marked 

 dialects, we present the comparative table of the Marquesan alphabet 

 set upon the Proto-Samoan base. 



a a, e, i, o 

 e e, a 00 



it UK 



\r, - 

 ng ". k, - n n, - mm 



h h, - 

 sh,f,~ 



v v 



if.v.h 

 k k, - 1 1, n, - p p 



