12 A. H. Edgren, 



trace of its r, the «-vowel has entirely sacrificed its nasal in 

 Sanskrit and Greek. Even here English has a lesson to 

 teach. Dialectically the r-element of such words as father, 

 mother, arbor, etc., is often omitted in both America and 

 England (cf. Whitney, Orient, and Ling. Stud., II., p. 236). 

 But that means virtually that those words often, instead of 

 being pronounced regularly as fathr, inothr, arbr, are pro- 

 nounced as fathy, niotJu, arbs {p being here used to denote 

 the indefinite ?/-sound in English). The ;/-vowel is not then 

 alone in losing its consonantal element. As, however, the 

 theory of a nasal vowel is nowise needed for the acceptance 

 of the theory that tan is weakened to ta in ta-noini, I will not 

 insist any longer on its merits. 



The last of the roots classified by the Hindu grammarians 

 with the Eighth class is kar {k{) 'make.' It is well known 

 that this verb was regularly conjugated according to the Fifth 

 class in the older language {kr-m-ti), and also in Zend {kere- 

 nao-ti). Its later conjugation, however, is entirely anomalous, 

 and has not yet been satisfactorily explained. The objections 

 brought against Brugman's theory by Van den Gheyn in his 

 second article seem to me, on the whole, justified. Unfor- 

 tunately, he offers nothing satisfactory in its place. Possibly 

 the whole problem might find a solution in supposing that 

 the primitive language possessed, at the side of kar, also a 

 radical karii^, from which the latter (classical) present-system 

 was made. In fact, a root krv (Jcarv) is given in the Dhatu- 

 patha, and analogous radicals in -v (-?i) are not uncommon in 

 Sanskrit (cf. carv, bJiarv, tnrv, dliurv, rajiv, dhanv, and yet 

 others). Probably these are all denominatives from stems 

 in -11. Denominatives, to be sure, form their present-stems 

 in -a or -aya, but karn, existing at the side of krnu (which 

 was the earlier and once prevailing stem), may easily have 

 come to be conjugated in analogy with it, especially as their 

 suffixal vowels coincided. But why, then, the change to 

 kur{ii)- in the weak forms .'' As is well known, the syllable 

 ar often alternates with iir, especially in combination with a 



28 



