262 



qernertoq. se'rnartoq. orsoq. Whether the stems of these words 

 have in tlie distant past been without an uvular or had nr. sr 

 instead of rn, rs, the uvuhirization of their first syllable must 

 have taken place at a time when the Mackenzie R. Eskimo 

 and the East Eskimo had not yet separated from each other. 

 It is highly characteristic that in the following form in the IVl. 

 dialect, where the retrogressive uvularization is distinct enough, 

 present Greenlandic has , not uvularization , but a geminated 

 consonant: 



M. amepk, skin, plur. aptngit (or amit) 

 Gr. am'eq, skin, plur. ^am'Ht (or anVit]. 

 It lies near to assume that the Greenlandic long m in the 

 plural is in reality an assimilation of an original nn , which is 

 preserved in the West Eskimo form. If this is the case, this 

 form has had the same fate as Gr. irse (Egede) > «.s-e, an eye, 

 except that the latter assimilation belongs to one of the newest 

 strata in the language, the former to a very old stratum. 



The retrogressive uvularization, then, has perhaps already 

 begun in a few words before the Mackenzie R. Eskimo separ- 

 ated from the East Eskimo, but it has not struck root in this 

 dialect, which undoubtedly belongs together with the western 

 dialects. The chief demarcation for this sound-change, indeed 

 for the dialects on the whole, is no doubt the one that I have 

 already suggested. 



In the far distant past, then, the original Eskimo horde 

 must have separated into two flocks, of which the flock toward 

 the east, before it was again subdivided, carried out the meta- 

 thesis. Since this is common to both the inhabitants of La- 

 brador and of Greenland , it may be assumed that it was car- 

 ried out before the invasion of Greenland, in the districts west 

 of Davis Strait. 



But here we are again confronted with a surprising form in 

 Frobisher. Of the language in Baffin Land, which geograph- 

 icallv lies between the two dialects mentioned last and bars 



