________________
168
EPIGRAPHIA INDICA
[VOL. XXXI
employed in these caves was first to engrave the record on the wall and then paint the letters with a kind of red pigment with a view to bringing out the letters clearly. The paint is still intact in a few of the inscriptions. The necessity of painting the letters is emphasized by the fact that the walls of the caves had not been always properly smoothened before the engraving was done. The surface is so rough in some of the caves that it is difficult to read the inscriptions from the rock itself. Probably the inscriptions were first written on the rocks with a similar paint also before the engraving was done. This accounts for Cunningham's finding the Ginja hill inscription1 of Bhimasena in red paint which for some reason unknown to us was never engraved. In these caves I found also a few short records written in red or black paint which are now so badly obliterated that they defied all attempts at decipherment. Besides the inscriptions of historical importance which alone are considered here, some of the caves contain short records incised in them. Many of these are apparently pilgrims' records though the sense conveyed by them is not always clear. In two of the caves I found a few letters in shell characters beautifully carved.
The palaeography of these records shows some peculiar features. Though written within a short space of 50 years, the inscriptions offer several types of writing. The earliest record belonging to the year 51 of the time of Bhimasena shows early features; but in the others the letters show both archaic and more advanced forms. Inscription No. XI of the time of Bhaṭṭadeva and dated in the year 90 is the typical example of the second category. The writing usually shows southern characteristics but in certain cases northern forms are also discernible. The vertical stroke of a and r is straight but also shows an occasional hook at the bottom, e.g. in Nos. I and IX. Initial u is found in Ujhasa (No. X). The vowel e is shown by a triangle with the base slanting upwards; but the base is straight in No. XI. The vertical bar of k is usually straight and the horizontal bar bent; but the latter is straight in No. XI. The letter kh has a large triangle at the base with a short hook above; but sometimes the base takes the form of a loop, e.g. in Rakhitika and pakhe in No. IV. The letter g usually has a rounded top; but it is angular in No. I. The left vertical line of j is bent in the middle; but, in Nos. I and XI, the vertical and horizontal bars are all straight. Dh has narrow points and rounded back; but the back is sometimes angular, e.g. in dhamo (No. XI). P and b show a notch to the left; but the left horizontal limb is straight in Nos. I and XI. M has a triangular base, though in rare cases it is inclined to show a rounded form as in Chhamikasa and negamasa in No. VI, line 2. Y has usually a hook or a loop on the left. The right limb of 1 is usually bent to the left at the top; but it is straight in No. I, e.g., in Balamita (line 4). A second form of this letter is also found in No. VIII, corresponding to that found in the Allahabad pillar inscription of Samudragupta. H has also two forms, the usual one with a hook to the right and also, though rarely, another like the letter found in the inscription of Usavadāta, e.g. in Mähäräja in No. VIII. Of the numerals, symbols for 1, 2, 5, 6, 7, 8, 10, 50, 80 and 90 are found. The advanced form of the numeral 2 is noticed in No. IX, line 4, as against the earlier forms in Nos. VII and XVII. These few details will show that the script, as found in these records, partly shows features of the Kushaņa records but is on the whole more allied to that found in the cave inscriptions of the Kshatrapas of Western India and also of the cave inscriptions of the Western Deccan belonging to the second century A.D.
With the exception of Nos. XIV, XVIII and XIX which refer to Sivamagha and Vaisravana and are written in Sanskrit, the language of the remaining inscriptions is Prakrit. Sanskritisation is, however, noticed in a few instances in the Prakrit records; cf. särthika (No. III, line 4), siddha (Nos. IV-VI, IX, XVII), Chakra (No. IV), and layana (No. XI). The main orthographical features noticed in the inscriptions in Prakrit are as follows: single consonants are used for double as usual in such inscriptions although doubling is noticed in bhaṭṭaraka (Nos. VIII, IX), Bhaṭṭadeva
1 Cunningham, ASI, Vol. XXI, p. 119 and Plate No. XXX.