THE SAWAIORI BEGINNING RESTS UNKNOWN. 177 



as fabrications which will stand under no law of evidence, assuming 

 that these obliquities can be made straight, we are to encounter a new 

 jungle of difficulties in the examination of phonetic mutation. 



We have been able to develop for the Polynesian a very simple 

 system. In the primary and even in the secondary Melanesian 

 borrowings of Sawaiori loan material we have developed a system 

 that seems to hold good as far as it can be put to the test, and this 

 despite the confusions of the attempt of the Melanesians to catch 

 in unaccustomed ears and to reproduce with untrained organs of 

 speech, above all the lips, sounds unfamiliar. We have subjected 

 the diffusion of these elements in various Indonesian languages to 

 the same analysis. 



What principle or principles of mutation have we been able to 

 discover in this trine examination as a factor common to the three 

 families which use this common stock of vocables and which possess 

 them in fee or as bailee? They are but few, these principles; corre- 

 spondingly they are simple : 



i. The nasals tend toward a mutation, if any, backward in the direc- 

 tion of the glottis. Less frequent is variety in the series to which 

 each belongs. The mutation is horizontal rather than vertical. 



2. The consonants in each of the series tend normally to mutation 

 downward in the series. 



3. At the foot of each series, palatal, lingual, labial, the mutes 

 tend to mutation upward in the series, surd mute to sonant mute, 

 mutes to spirants. 



4. Mutation extra seriem, horizontal mutation, is rare outside of 

 the nasals, most such cases being explicable as mutation to the aspi- 

 ration in the first instance and then secondarily from the floating 

 aspiration to some adjacent series. 



How far do the proposed Semitic identifications conform to these 

 broad principles which, in the intricate detail of the study put upon 

 them in the foregoing chapters, have been established as perspicu- 

 ously as simply? 



It would be idle to attempt to list all the concordances and equally 

 the discrepancies of the Semitic offered in identification when meas- 

 ured by these established principles. We note from the Arabic, 

 from which Dr. Macdonald has drawn most largely, the proposed 

 mutations in but two of the representative Polynesian consonants, 

 t as being central in the diagram and f as representing the labial 

 series and with it the maximum mutability. For each mutation we 

 note but a single instance in reference; it seems that there will be 

 no lively desire to seek out more. 



Polynesian t may become in Arabic: / (107), t (44), d (38), s (35), 

 f (160), z (267), h (36), n (37), gr (247), rd (306), '(356). 



