MICHELSON. ] INTRODUCTION. 45 
not been rigidly adhered to, for the Foxes themselves are not entirely 
consistent in it, being guided to some extent by the tempo. Final 
vowels are aspirated before initial sibilants. On the other hand, 
it was not possible to normalize the sentence and word accent; so 
that the texts in this volume are “mixed” to this extent. I have 
followed the dictators. It should be noted that the stress-accent at 
times is very weak, and this accounts for the large number of appar- 
ently accentless words in the Indian texts. 
Another point may be brought up here, namely, the treatment 
of final i of one morphological unit before another such unit begin- 
ning with a vowel or diphthong. The elision of this varies among 
different speakers, and is also dependent on the tempo. I have 
followed the usage of the dictator in this respect. The glottal stop 
is merely rhetorical. 
In conclusion it may be noted that aspirations before initial vowels 
and ai are not pronounced, save rhetorically. It may be noted that 
terminally when w is followed by a voiceless vowel, I have heard 
some speakers pronounce a sonant w with a surd glide, and others 
pronounce without (apparently) the sonant w. In such cases I have 
followed the dictator of any given text. 
