________________
$ 24, A.]
INDIAN PALEOGRAPHY.
(2) In consequence of the elongation of the ends of the wedges and of the use of long straight top-strokes, the heads of A, A, gha, pa, pha, ma, ya, sa, and sa are gradually closed, both in the soute-angled and the Nāgari scripts.
(3) The lower portion of the left half of A and I almost invariably consists of a curve, open towards the left, which first appears occasionally in the Kuşana inscriptions (see above, $ 19, B, 1) and later regularly on the Ucos kalpa plates (plate IV, 1, IX). It is preserved in the Balbodh of the Marāthās and is common in the Bombay editions of Sanskrit works. In other late specimens of the Nāgari; it is replaced by two slanting strokes (plate V, 1, 2, XVF), to which a third, & remnant of an earlier wedge at the foot of the vertical, is added lower down. This form is the parent of the 4, A, ased in the Benares and. Calcutta prints. Up to the 8th century, the long A is invariably, differentiated by the addition of a curve to the right end of A. Later, its mark is a downward stroke, which is attached either to the right of the top (e. y., plate IV, 2, XXI) or to the middle (plate IV, 2, XXII) and thus reoccupies the same positions which the corresponding horizontal bar has in the Asoka odicts. In the MSS., the downstroke at the top is found even earlier (plate VI, 2, VI).
(4) The sign for I is mostly derived from the Gupta form of Iudor (plate IV, 3, VII) by the substitution of a curve for the third dot (plate IV, 3, XI-XXIII ; V, 3, II-IV, &c.; VI, 8; V-IX). But in addition there is (plate V, 3, V, XII, XIII, &c.; VI, 3, XI-XV) a derivative from the T of the Uccakalpa plates (IV, 3, IX), in which the upper dot is replaced by a straigbt line; and this I is the parent of the modern Devanagari I, in which the two lower dots have been changed into carves and finally have been conneeted. ha Jaina M3S., the I with two dots above and a curve below occurs occasionally as late as the 15th and 16th centuries. The unique early forms of the long I (plate VI, 4, V, VII), as well as their later development (plate VI, 4, XV), which has followed the analogy of I, deserve attention.
(5) U and O invariably show at the lower end a tail; drawn towards the left, which in course of time is developed more and more fally..
(6) The curve of R, attached to the right of the ra, becomes very shallow and long in the Horiazi palm-leaves (plate VI. 7,V), and this shallow carve is the precursor of the vertical line of the later palm-leaf MSS. of Western India (plate VI, 7, XV-XVII). In the Cambridge M8. No. 1049 (plate VI, 7, VII) and in No. 1691, ther-curve is attached to the lower end of the ra.
(7) Among the signs for , Land , which are first traceable in the MSS. of this period (plate VI, 8-10, V, VII, X), the long R is clearly formed by the addition of a second -curve to the short R. In the Cambridge MSS. Nos. 1049 and 1691, Ę is represented by a cursive southern la (see plate VII, 34, VI-IX), jast as the oldest medial ! in le! (VII, 42, XIV) is identical with another form of la; and the long L is derived from the short vowel by the addition of a second la, turned in the opposite direction. In the ļ and / of the Horiazi palmleaves (plato VI, 9, 10, V), the In has been turned round towards the left, and respectively one and two r-curves have been attached to the foot. And the combination l(a)-r remains (64) also in the Nägari both of the palm-leaves from Western India (plate VI, 9, 10, XV) and of our days. the reason being no doubt the pronunciation lị, which is customary both there and in other parts of India. These paleographical facts agree with the tradition of the Chinese Buddhists. who, as 8. Lévi has discovered,' ascribe the invention of the sigos for the liquid vowels to
South-Indian, either to Sarvavarman, the minister of the Andhra king Satavābana, or to the great Buddhist teacher Nagarjuna.
1 Soe above, $ 16, D, 1, 2; and plate II, 2, II-X
1 Seo above, $ 23, page 80. . Communiontion by letter.