1855.] Ancient Indian Numerals. 563 



sources of knowledge from whence he may have obtained his approv- 

 ed outline. 



The Gupta units vary somewhat from the Sah exemplars. As yet 

 I have only been able to discover three definite and complete 

 forms, the v—four above adverted to, the one which is shaped as an 

 ordinary hyphen, and the curious figure that occurs on No. 57. Pi- 

 ll. J. E. A. S. XII. here its outline follows that of an alphabetical 

 ?T ; but, in treating of Gupta numbers I must fairly warn my 

 readers of a preliminary difficulty that I have experienced in regard 

 to the correct point from whence they should be viewed. The Na- 

 sik inscriptions display the symbol for one hundred as written per- 

 pendicularly, and if that is the correct direction of the cipher, our 

 Gupta dates placed in front of the profile of the King, ought to be 

 read Mongol fashion, like the names of the monarchs of the Gupta 

 race, as usually expressed on the field of their gold currency. On 

 the silver pieces of Kumara Gupta, however, whether the sign for 

 100 may be reversed or not, the arrangement of the tens and units 

 clearly demonstrates that the whole must be read as consecutive 

 rather than as superposed figures, while strange to say, the dates 

 on Skanda Gupta and Buddha Gupta's coins seem to necessitate a 

 contrary mode of decipherment. 



I have entered the outlines of the Gupta numerals both tens and 

 units iti accordance with this somewhat arbitrary arrangement, 

 leaving the point fairly open to correction, when more numerous 

 and more perfect specimens of this coinage may decisively instruct 

 us on the general question. 



I have yet to advert to the outlines last on the unit line, but the 

 single example of this sign to be found on a coin of Kumara is too 

 imperfect to justify my quoting it with any confidence. 



For the assigument of the several symbols representing the 

 numbers from 30 to 80, we have equally but very imperfect data. 

 The 10, 80 and 90 are established by the copper-plate duplicate 

 dates. The 10 in a new form, the 20 and 30 are contributed by 

 the Nasik inscription, and we have some slight aid in determining 

 the relative priority of the various ciphers in the occurrence of two 

 of their uumber on the Valabhi grants* of one, and the same cen- 



* No. 1, Valabhi Grant (Wathen) J. A. S. B. IV. 481 and No 5, PI. XX. Vol. 

 VII. No. 4 ditto (Burn) J. A. S. B. VII. PL XX. No. 4 and VII. p. 966. 



