DATA AND NOTES. 431 



The Melanesian identifications are satisfactory, Baki ilo exhibiting the 

 same vowel change as in 362. 



The three Indonesian identifications agree in the accretion of a nasal and 

 are satisfactory. 



There is nothing in form and nothing in signification in this Semitic to 

 afford suspicion of relationship with ula. 



365- 

 uti na, the penis. 



Viti: uti, id. 



New Guinea (Astrolabe Bay); uti, id. 



Another root form is quite similar in the same sense and is adjoined for 

 comparison. 



Samoa, Tonga, Niue, Nuguria, Hawaii: ule, the penis. Maori, 



Tahiti, Mangareva, Paumotu: ure, id. 

 Motu: use, id. New Ireland (Port Praslin) : usik, id. New Ireland 



(Carteret Harbor) : ausidi, id. Pala: «*, id. Lamotrek : ul, id . 

 Arabic: 'uss', id. 



We have here forms in /, s, t. In the table of mutations we find very 

 scanty evidence of l-t mutation, but none at all of l-s. Starting from s we 

 find very scanty evidence of s-t mutation, none at all of s-l (s-r resting on 

 one unsatisfactory instance). Starting from t we find considerable proof 

 of t-s mutation, and, while /-/ does not appear, the similar t-r is found in a 

 few cases (see note 258). 



It is, therefore, just possible if we regard uti as the parent form, to make 

 a plea for the derivation of use and ule therefrom. Apart from the fact 

 that our evidence in support is scanty, we find yet another difficulty. We 

 should be put in the position of finding in the Proto-Samoan a secondary 

 form, yet in all our material we have found that to be the primary form and 

 the Melanesian secondary or loan material. We are not warranted, there- 

 fore, in accepting this identification. 



Not having anything in particular with which to identify it, the mere 

 presence of the Semitic here offered cuts no figure. 



