338                       ALBERUNPS INDIA.

called sankha, in Persian sped-muhra.    I have seen this
in the town of PurshiXr.    Pious people have bequeathed
for these clepsydree, and for their administration, lega¬
cies and fixed incomes.
Muhurta.          Furtlicr, the  day  is  divided into  thirty  muhurta,

but this division is not free from a certain obscurity ;
for sometimes you think that the muhitrtas have
always the same length, since they compare them either
with the ghati, and say that two cghati are one muhurta,
" or with the watches, and say that one watch is three and
three-quarters muhHrta. Here the muhurtas are treated
as if they were horce cequinoctiales (i.e. so and so many
equctl parts of the nychthemeron). However, the num¬
ber of such hours of a day or of a night differs on every
degree of latitude, and this makes us think that the
length of a muhurta during the day is different from
its length during the night (for if four watches or fifteen
muhurta represent a day or a night, the muhilrtas
cannot be of the same length in the day and in the
night, except at the times of the equinoxes).

On the other hand, the way in which the Hindus
count the dominants of the muhilrtas makes us more
inclined to the opposite opinion, that, in fact, the
muhurtas are of different length, for in the case of day
and night they simply attribute to each of them fifteen
dominants. Here the muhilrtas are treated like the
horce obliquce tenvporales (i.e. twelve equal parts of the
day and twelve equal parts of the night, which differ
as day and night differ).

The latter opinion is confirmed by a calculation of
the Hindus which enables them to find the number of
the muhurtas (which have elapsed of the day) by
means of the digits which the shadow of a person
at the time measures. From the latter number you
subtract the digits of the shadow of the person at
noon, and the remaining number you look out in the