34 THE POLYNESIAN WANDERINGS. 



varying languages of Indonesia. As soon as he attempts to link 

 Indonesia in the backward bight of the chain, his figures fall to 

 ioo, and when he links in Melanesia he finds but 28 available words. 

 Accepted even at his own valuation, philo logically a most faulty 

 one, these are but small foundations for so lofty an erection of a 

 Babel building which is to confound Max Miiller and other names 

 by no means inconsiderable in systematic philology. 



In these studies my position is eastward. From Polynesia I am 

 looking through all available linguistic material toward the west, 

 hoping to find some data that will establish the partition of the 

 Polynesian and the Melanesian elements in Viti, welcoming any- 

 thing that promises to enable us to comprehend the Melanesian in 

 areas where its proportion in speech is the more considerable, 

 rejoicing at the accession of new knowledge which will equip us 

 to give better study to the possibility of some manner of proof that 

 kinship does or does not exist among Polynesian, Melanesian, and 

 Indonesian, convinced that only as we establish these three links 

 of the chain in positive knowledge can we approach the yet earlier 

 history of these tongues with clear sight. 



In Dr. Macdonald's material I see three classes of data. 



1. A central area, Melanesian material for which we have not yet 

 established connections with either of the known areas, Polynesian 

 and Indonesian, respectively. We therefore want the means to 

 subject this to comparative study, and on this account it remains 

 unavailable. 



2. An eastward extension, through so much of Melanesia as is 

 known, to the well-established knowledge of Viti and Polynesian. 



3. A westward extension, through Melanesia to the equally well- 

 established knowledge of the Indonesian; and, yet more remote, 

 Dr. Macdonald's confident projection to the Semitic. 



The second class is that which shall provide the instruments for 

 our study of these data upon which we are now to engage at much 

 length and with the utmost attention to the minuteness of detail. 

 The Indonesian languages, and in yet greater measure the Semitic 

 tongues, have their own enthusiastic students, and to such we may 

 confidently leave the prosecution of similar research from the 

 vantage-ground of their own knowledge, in case this ambitious 

 theory should seem at its western extremity as proper a subject of 

 debate as it does to those of us who are engaged at the eastern 

 extremity. 



In the following detailed study I have purposely omitted many 

 examples, for simple inspection will show our author patently in 

 error. At the same time I have endeavored to afford room for all 

 such as seemed at least debatable. The process of elimination will 

 be continued in these studies. 



