________________
48
INDIAN ANTIQUARY, VOL. XXXIII, 1904; APPENDIX.
[ $ 22, O.
(8) The palatulis (10, I, II; 4%, I, VI, VII, XI) is frequently made cursive aud round, and is occasionally laid on the side in order to save space; compare ulso jauh, plate III, 40, XIV. But older, angular, forms likewise occur (42, V).
(9) The ça (17, I–III, IX) is often flattened down at the top.
(10) The ņa of 21, I, II, shows a little stroke at the right end, caused by an inexact formation of the hook on the right, and in the secoud sign a cursive loop on the left; in 21, III, the letter has been laid on the side and somewhat resembles the Magari na.
(11) The tha (23, 1, V-IX) is mostly elliptical or fluttened on the right, and a cross-bar often replaces the dot in the centre; but the old form likewise survives (23, 11, 111).
(12) The va (82, 1-IX) is mostly tripartite, but sometimes, particolarly in ye, yai and yo, transitional forms with the loop, like the Inter ones in 32, XIII, XVI, appear, which lend op to the bipartitite yq. The oldest instance of the independent looped yu is found in Fleet's No. 59 of A. D. 371, but the Kuşana inscriptions show the looped subscript ya even earlier (see above, $ 19, B, 12). . .
(13) The left linb of sa (38, I-III, V, VI, VIII) often becomes a loop, as happens already in some Kaşana inscriptions ($ 19, B, 16). A sabstitute for the loop is the triangle (probably giving the outlines of & wedge), which occurs in the three moat ancient inscriptions from Nepāl; compare the later sa of 38, XIL. But the older hook is equally common.
(14) The rare !a (40, I–III) is found also in Fleet's No. 67, live 1.
(15) The signs for the medial vowels agree in many particulars with those of the Kaşana period. But the open semicircle for å in tà (17, 11), which is found also in rā, is an innovation. Further, the medial i, for instance, of khi (8, III, VI, IX), is drawn farther to the left than in the earlier inscriptions. In some inscriptious like Mathura, new series, Nos. 38, 39, the medial i consists merely of Turve, going to the right, though the form with two horns (as in di, 24, I), and a looped one (as in bht, 30, IV), are more common. Medial u is mostly represented by the still used curve, which in ru (33, III, VI) appears abnormally at the end of ra; bat in gu (8,II, 'VI), tu, bhu (30, 1) and ku (36, III) the vowel rises upwards. For medial u there are, besides an old form in gu (9, IV), other combinations in bhi (30, II, VI) and !!ů (42, II) and a later very common, cursive form in dhů (25, II, VI). One of the Mātrās of ai and o is often placed vertically, as in gai, 32, III ; in go, 9, III; and in no, 21, III.
(16) The desire to save space causes the carsive ña, ta (see sta, 45, IX) and tha (see stha, 45, V ; stho, 46, IX) to be laid on the side, in case they form the second elements of ligatures. From the 5th century, rya (45, VII) is expressed by a full ra with a subscript ya.
(17) The first certain Virāma (see ddham, 43, VII), consisting of a horizontal stroke above the small final, dates likewise from the 5th century; the northern Jihvămūliya (bka, 46, II) and the Upudhin niya (bpā, 46, III) occur already in the 4th century.
c.-The Gupta alphabet in manuscripts. Among the types of the Bower MS., which belongs, according to HORRNLE's and my own opinion, to the 5th century, I have given [49] in plate VI, cols. I-IV, only the alphabet of the portion whiclı HUERNLE marks A, since the published parts of his B. and C. are not softciently extensive for a paleographic enquiry. Its characters differ very little from those of the epigraphic documents of the Gupta period, especially from the copper-plates. The Serifs at the tops of the vertical stroker, however, are made moro carefully and nearly throughout worked 1 Compare facsimile in Flert's Gupta Inscriptions (CII. 8), No. 61.
1J.ASB. GO, 83 ff. • J.ASB, 60, 92 1.; WZKM, 5, 104 f. The discovery of an inscription of the 7th century with mostly tripartite ya, EI. 4, 29, makes a modification of HOERNLE's argumentation Decembary, but does not invalidate his final result.
