274 # Inscription in the old character on the [March, 



bet : and we see the reason why this was departed from in the Nagari 

 form, q>, by turning the stroke outward, lest by turning inwards it 

 should be confounded with the tj or sh, a letter unknown in our old 

 alphabet. With reference to my former remark on the duplication 

 of alphabetic forms to produce the aspirates, it may be adduced as an 

 additional argument for such an assumption that in the oldest of three 

 plates from Kaira with copies of which I have been lately favored by 

 Dr. A. Burn, the ph of the word phala is twice written qj oy pp in 

 lieu of XS> ; vvhich is the augmented or aspirated form used in the other 

 plates, and which is more consistent with the original type now disclos- 

 ed to our knowledge. 



Of the bh I would merely take this opportunity of noticing that I 

 have discovered the period and cause of the two very opposite forms of 

 this letter which are found in later alphabets, as for instance the Mah- 

 ratta ^f and the Tibetan ?n (which agrees with the Devanagari or Kutila 

 of the 10th century ^) and have proved them both to descend from 

 the original ^ ; the Mahratta may be said to follow naturally from 

 the Sainhadi*i form ; the other I have traced on the Saurashtra coins 

 of Skanda and Kumara Gupta, where sometimes the one and some- 

 times the other form is employed, the latter being the natural course 

 followed by the pen in imitating the sculptured letter rf , beginning at 

 the top, viz : ^ , whence would gradually follow ^, and M with the 

 headstroke, common to all the modern characters. 



The Pali contains but one ,?. We cannot therefore expect to find in 

 our ancient alphabet the prototype of either the Sanskrit "JT or "%. Of 

 these letters I only notice the early forms because I have inserted them in 

 the accompanying lithographed plate. The modern form of ^r would seem 

 to be derived from the *f of the Samudragupta or No. 2 alphabet, 

 where again it might be presumed that it was introduced as a trifling 

 modification of the letter JSJ, or s, — in fact, by closing the outer stroke 

 or doing the same thing to this as was done to the p, to have the effect 

 of duplication or aspiration. Or, it may be more proper to consider it a 

 written modification of the more ancient form y found on the copper-plate 

 grants of the third century dug up in the Gujerat peninsula, whence the 

 transition is more evident and palpable to the various Pali and Sinhalese 

 forms, the Cashmere form and even the modern Nagari and Bengali. 



It is not so easy to trace the origin of the tdliba sha,^, in the old alpha- 

 bet but there is plausible reason to suppose that this was originally merely 

 the murdina or cerebral s ft, turned in an opposite direction, invented to 

 denote another modification of the sibilant required in the refinement 

 of the Sanskrit alphabet. In the oldest Guierati plates, these arc writ- 



