THE ANTIQUITIES OF MUKHALINGAM. 75 
1062j expressed by the numerical words "nayana, (2) rasa, 
viyat (3), chandra (1)." 
One on tlie second pillar, containing a Sanskrit sloka, and 
one or two on the opposite pillars, but worn out. 
9. The dates and the reigning years of Anantavarma- 
deva given in the inscriptions on the two above-mentioned 
temples do not all tally with one another. But I did not 
examine them critically when I was at Mukhalingavi, and 
my notes therefore require to be verified. 
10. I may here remark in passing that in deciphering old 
inscriptions, one has very little trouble in the case of San- 
skrit ones. When the language is a vernacular it requires 
much more patience and ingenuity to work at them. I 
cannot say that I could understand every word of the 
Telugu inscriptions. They offer some curiosities to be 
noted by a Telugu philologist. Some of the Telugu words 
and forms seem to have now become obsolete. AoKg'eeoeS 
was formerly used to mean the bank of the Ganges. JT'elaS) 
was perhaps some unit of land-measurement. ^otS and 
"^o<Jk) (=to-day) were probably in popular use 800 years ago, 
as -a'oi^. s^oTv now are in sr>me parts. The nasal in the 
middle of the words is now dropped in pronunciation, but 
can be recognised in the semi-anitsvdru substituted by the 
learned in writing. Orthodox Telugu pandits do not allow 
that the semi-aiiusvd7-a after a long vowel represents the 
older full anusvdra sounds. I have met with such words as 
"2s^c2.g&-"o^eaj, which are now written by the learned thus . 
53-«(C)Sr>c©oftoo but which are in colloquial usage »r»5§5n>£^:bo. 
I have not seen Mr. Seshagiri Sastri's recent publication 
on the Ardhanusvara, but I expect such instances are 
noticed and commented upon by him. 
11. (III.) The Somesvarn, temple. This is situated at a 
distance of more than two furlongs south of the village. 
One peculiarity of this temple is that it opens to the west, 
