Vol. VIII, No. 9.) The Bakhshali Manuscript. 355 
[N.S.] 
(k) Medial o is also expressed in three ways, of which the 
wavy detatched top stroke appears to be the most 
| (l) Initial e in the earlier inscriptions is almost equilateral 
in shape. The later forms are more right-angled. 
Of these points (c), (e), (2), (2), (j), (k) seem to be the most 
important. The differentiations aaa (a), (b), (d), (f) and (1) 
may possibly be due to the material used. The evidence o 
the whole appears to point to the Bakhshali manuscript com- 
ing between the Baijnath and the (Kullu inscriptions, or say 
between the eleventh and sixteenth centuries a.p. 
Wi. 
There occur in the Bakhshali manuscript at vin two ex- 
amples of word-numerals, viz. répa for ‘one’ an a (the 
the me for ‘ six.’ Now itis possible that rapa may ite e been 
sed for ‘one’ before the word-numeral system was ira nei 
sito India (the system was aa probably not indigenous), but 
the term rasa would not have been used for ‘six’ before that 
time. Although el-Biruni tells us that Brahmagupta ‘ invented 
this system, the earliest epigraphical instance is, according to 
. Fleet, a.p. 945, while according to Li iders the earliest 
instance is dated Vikrama Samvat 898.2 Biihler quotes the 
Cicacole insoription (a.D. 641), but this has since been proved 
to be spurious® and the Kadab inseription (A.D. 813) which 
is deemed doubtful by Liiders and Fleet.* Of epigraphical 
instances of these symbolical words I have come across two 
only of the ot century, three of the tenth, a few of the 
eleventh and numbers of later date. 
Again, Acsaeaines to Biihler, ‘‘the decimal figures of the 
Bakhshali manuscript show the ancient letter numerals for 4 
and 9; but these so-called ‘ancient’ letter numerals are com- 
paratively modern. The tendency to fashion numerical sym- 
bols like letters does not appear before the ninth century. 
However, Biihler’s statement is not really borne out by the 
manuscript itself. There is hese resemblance between the 4 
and ka, the 5 and pa, the 6 and ma, the 8 and ha, and the 9 
and the Om; but the Cecdtiacen is not sufficient to form a pre- 
miss for argument. It would, however, not be difficult to io 
out from other sources letters almost identical in form with 
figures in the Bakhshali manuscript, and therein lies a srt 
from the fact that the place-value notation (which 
possibly came into India about the tenth century a.D.) is used, it 
is somewhat difficult to fix a date from the form of the bear: 


Brahmagupta lived in the sone hedibady ADs, ‘iis too tatis 
for Hoorn’ s theories. 
Ep. Ind., iv, 335. 8 Ind. Ant., xxx, 211. * Ep. Ind., iv, 335. 
