THE HAMILTON ASSOCIATION. 95 



serviceable for speech, and, as children's words, not likely to be in- 

 fluenced by the analogy of shifted forms. 



2. Again we find the tenues regularly unshifted in certain com- 

 binations : — 



ft phnptos . fifta fifth^'^ funfto fUnfte 



ht okto eahta eight ahto acht 



The permanence of the tenues in these groups is evidently due to 

 the presence of the surd spirants .r, f h. The variation in the sk 

 group shown in many instances in modern English {sh) and in al- 

 most all in modern German (sch) does not belong to the primitive 

 Germanic : the early documents of both languages preserve the k. 

 Similarly mfremde the High German, because of the influence of the 

 sonant m, has retained the d of the primitive Germanic seen in 

 K.-^.fremde. The d of the M. H. G. preterite forms nande, rumde, 

 solde is to be explained in the same way. 



3. A number of seeming exceptions will readily be understood 

 if we bear in mind the laws of so.und change in the several langua- 

 ges. Take for example the cognates Skt. duhitr, Gr. thugater, Goth. 

 dauhtar, A.-S. dohtor, E. datighter, G. tochter, which point to a 

 primitive ^dhughatar. The change in Sanskrit from dh to d was 

 evidently to avoid the use of an aspirate so near the gh ; a similar 

 dissimilation regularly occurs in Greek, as in pephileka for *phephi- 

 leka. Again, primitive ' mediae aspiratae ' between vowels were often 

 represented in Sanskrit by h. The g in Greek is an instance of a 

 change often found, as in eg07i (Skt. aham), megas (Skt. mah) ; com- 

 pare also brhno (Skt. bhramati^ Lat. fremo). The retention of t in 

 Germanic forms has been explained already : the h, gh, ch require 

 further consideration. 



4. When the necessity was shown for postulating the existence 

 of a language which stood in the same relation to the known langua- 



(i) For the th see below ; page 97. 



