209 



Mackenzie dj (Petitot) Labrador ts, dj WGr. s- 

 midjiapk écheveau, \ 



plur. nudjiat, cf. 7iyapk, > nutset nut'ät, nutsät 



plur. nutçat ] 



tamadja vraiment tamadja (Bourquin g 192) tamas'ci 



madjia ceci mas'a 



nagdjiuk (p. Vll) or | 



, ,v- 7 i nakjuk nac'uk 



nakdjiuk corne ] "^ 



adjipapk (p. XLIX) or | 



, . , . ; aqqak cic-ak 



adgipapk main j 



[^ç] nekpetchidjoapk faire nerisis'o- 



raanger 



The explanation of the distinction which has been drawn 

 between the three kinds of s in Greenland (point-s, mid-s, 

 blade-s, cf. § 9) is probably to be found in the different origins 

 of the sound. But the matter is not altogether clear to me ; 

 among other puzzling points is the fact that the sounds in 

 Greenlandic seem to pass over into each other, and this is 

 the reason why 1 have not as a rule assigned them separate 

 symbols in my spelling. Closer empirical investigations may 

 perhaps throw Ught upon this sound and its conditions in the 

 two dialects. The relation between j and s is the same as 

 that between any voiced sound and an unvoiced sound, as, for 

 instance, between g and ^, or between v and f. Yet in the 

 group tj in the Labrador language, 1 think the j must be un- 

 voiced and accordingly the whole sound-group very nearly related 

 to [s], cf. the description of_;' in Bourquin, Gr. g 8. So where 

 this sound-group (tj) is given, it is not necessary to presume 

 that the transition from j to s has taken place through the 

 medium of a voiced s [z\, which, however, is probable for that 

 series of cases where only j is given. — Now it is not in all 

 cases by far that Greenlandic has got s where the Labrador 

 language has j. Both languages have j in words like : 



qujawoq thanks, ujamik necklace, ujap'oq stretches up over 

 XXXI. 14 



