167 



long n just as well to kn^ fii as to nu, etc.*) As a rule it 

 must be decided in each case separately which is the most 

 probable stem-form. The reconstructed intermediate forms are 

 mainly scientific working formulas, in which the result of the 

 analysis of the word is expressed, but they do not necessarily 

 mean that the unassimilated forms ever were really pronounced. 

 Perhaps the second /t-sound in *kamiktd:q has never been 

 pronounced as к directly before ^, but has merely been thought, 

 merely approximately aimed at. The 'result may have been a 

 marked elevation of the back of the tongue before the ^-sound, 

 somewhat like a loosely articulated unvoiced j, in which the 

 ^-position has been anticipated. 



It would be exaggeration to assert that all the geminated 

 consonants in the language are eqvivalents for original groups 

 of two different consonants, but such geminations are undoubt- 

 edly more numerous than those which can now be dissolved. 

 The assimilating activity has no doubt been of no less import- 

 ance in older stages of the language than is the case now. 

 j\Iy notes and the tales etc. which I have written down contain 

 enough of examples to show that assimilation is constantly 

 taking place in the spoken language, in all essentials similarly 

 for all who speak the language because they all have the same 

 habits of pronunciation, and yet in many little details varying 

 according to different individuals on account of individual 

 arbitrariness. As contrasted with the assimilations handed down 

 by tradition, which belong to the history of the language, these 

 cases last referred to appear as vigorous demonstrations of the 

 present life of the language. 



*) Just as we should not be able to know anytliiiig certain about tiie 

 original elements of the long /' in words like arnäl'o, iniiÅ'o (and the 

 women, and the human being), if we could not take into consideration 

 such suffixiiigs as nnalo (and lie) etc. and conclude from the latter 

 that in the former cases too the suffix is merely -/o, which in the flrsl 

 word is added to arnät (woman), in the second, to iniilc (a human 

 being); k^ has accordingly in the one case originated from tl . in the 

 other, from Id. 



