________________
No. 29.]
BANSKHERA PLATE OF HARSHA.
209
prasasti and in the Barada, and consists of a loop below the top-stroke of the ka. The upadhmdniya is represented by a semicircle, open above, with curled ends, just as on Vinayakapila's platel of Sri-Harsha-] Samvat 188 and in later inscriptions. It stands above the pa, but on the level of the top-line of the letters. Similarly the superscribed ra, too, never rises above the top-line of the consonants. Dr. Fleet has noticed this peculiarity as something exceptional in the Aphsad prasasti. But Professor Kielhorn has found it also in the Kudarkot prasasti, and Harsha's two land-grants (that from Madhuban with some exceptions) offer further instances. It is also quite regular in the Såradå ligatures and in those of many Nigari manuscripts of the 10th and 11th centuries A.D. Its causo is, in the cases of the four inscriptions, the desire of the calligraphists to make the tops of all mátrikds without vowel-signs perfectly level in order to gain room for the ornamental medial d, s, i, eto. The superscribed ra of these inscriptions consists regularly of & wedge; but in varnnásrama (1. 3 of the Bapskhêrs plate) it is represented by a full ra, attached to the right of the lower na. Strictly speaking, the group is equivalent to nnra, and we have here another instance, showing that the Indian scribes even of late times did not hesitate to change the natural order of the component parts of a group of consonants in order to form a shapely sign. The fact is of some value for the correct interpretation of the irregular ligatures in the Girnar and Siddapura versions of the Asoka edicts. With Vinayakapala's abovementioned plate agrees also one of the Banskhêra forms of na, e.g. in ograharatvēna (1. 11), where the loop on the left of the sign is connected, not with the vertical, but with the top-stroke. The letter thereby becomes somewhat similar to a ga, for which Dr. Fleet has mistaken it in the word 'nauo (1. 1 of the Vinayakapâla plate), rendering it in bis transcript by gão. The virama in Samvat (1. 16) stands to the right of the final t, hanging down from its top. In the later Madhuban plato we have in the corresponding word the older form of the virama, which consists of a stroke above the final letter.
The characters of the sign-manual in line 18 differ very considerably from those of the body of the grant. They are about three times larger and very elaborately ornamented, in fact of the florid type of the 80-called "shell-characters." The vowel i in the dhi of mahardjadhiraja consists of more than a dozen separate strokes, and the preceding & of seven. If king Harsha really used these characters in signing all legal documents, he must have been a most accomplished penman, and the cares of government and the conquest of India must have left him a great deal of leisure.
Among the numeral signs, those for 20 and 1 agree with the letter-numerals of the period. But the sign for 2 very closely resembles the modern Dovanagart figure of the decimal system. The Devanagari sign for 3 occurs also occasionally in the Bower MS., and it would seem that advanced forms of the decimal numerals were in existence mucb earlier than is usually assumed.
There is only one sign of interpunctuation, the single danda in the shape of & curved stroke. In line 11 this sign is need even between the two names of the donees, though they belong to one and the same dvandua compound. At the beginning of the technical portion of the grant, the neglect of the sandhi in the words Harshah Ahichchhattrao (1. 7) does duty for the sign of interpunctuation.
The language of the Banskhera plate is very good and correct Sanskrit, which is better than that of the Madhuban plate. Even in the technical portion there are only two mistakes, the Prakritic form pramåtdra for pramátri (Il. 8 and 14) and the bad compound sarvvaparihritapariharð (. 9). The orthography is regulated by the pedantic system of the
1 Ind. Ant. Vol. XV. p. 140.
Gupta Inscriptions, p. 208. • See my Indian Studios, No. III. p. 77£.
Kp. Ind. Vol. I. p. 180. • Compare Ind. Ant. Vol. XV. p. 364.
2 B
