1855.] Ancient Indian Numerals. 559 



supposition I can scarcely bring myself to accede to. The conclu- 

 sion is chiefly based upon the similarity traced in certain forms of 

 the figures, to the original letters of the Arian writing ; in order 

 to carry out the comparison however, very great liberties have to be 

 taken with the normal forms of the characters themselves, and even 

 these rather forced identifications are confined to a very limited pro- 

 portion of the entire suite of the numbers ; while on the other hand 

 many of the figures are clearly and indubitably composed of letters 

 of the alphabet in which the inscriptions at large are expressed. 

 That these latter in their original constitution actually were indige- 

 nous letter symbols seems to be fully established by other more 

 recent inscriptions, where such forms are seen ordinarily to follow 

 the progressive modification of the cognate alphabet. 



I now proceed to examine the figures in detail. 



To commence with the units, I conclude that no one will wish to 

 gainsay the simple and obvious determination of — = 1 = = 2 or 

 EE = 3, as suggested by me iu 1848 ; the first mark occurs on Sah 

 coins,* on the Huriswamini inscription at Sanchi,t and on the Peacock 

 coins of the Guptas, the second is found on Sah coins, J and on the 

 Udayagiri cave inscription,§ and the third is met with amid the early 

 Pali of the Bhilsa Topes, || on the Sah coins,^[ and in the Chandra Gup- 



ing resemblance between the character denoting a thousand (Sahasra) and the Bac- 

 trian S reversed. This induced me to examine the rest of them, and I think it 

 exceedingly probable that they are all derived from that source. The Bactrian Tx. 

 pronounced in Sanscrit J or Dsch, will represent well the figure, which is first in 

 5 or 10 (Dasha.) The sign for 5 {Pancha) is the P, or the old Indian ^ inverted. 

 The Bactrian double T, also approaches very nearly to the 8 of our inscriptions, 

 as if to denote ^3". It would appear, then, that the Bactrian letters had been in- 

 troduced into the Satrap Indian inscriptions as numerical cyphers. The system, 

 also, is the ancient Roman and Greek one, that in which there are different sigus 

 for the 1 in tens, hundred and thousands ; our present decimal notation being, as 

 1 have noticed elsewhere, a comparatively modern invention of the Scindian mer- 

 chants of the middle ages, (J. R. A. S. Bombay IV.) Future research will pro- 

 bably show, as Mr. Prinsep has done with a few of them already, that the old 

 Indain numerals are also ancient letters," (J. R. A. S. Bombay V. 39.) 

 * J. R. A. S. XII. 62. f J. A. S. B. VI. 458. Bhilsa Topes, PI. XXI. No. 198. 

 X J. R. A. S. XII. 39. § Bhilsa Topes, No. 200. || Bhilsa Topes 252, No. 121. 

 % J. R. A. S. XII.— 38. 



4 E 



