for that elegant figure, whereupon the precedent of the two Patna statues in the Indian Museum was cited in that they were described by Cunningham as Yakshas on the authority of the inscriptions on them. This made me desirous of examining those inscriptions. Copies of those inscriptions prepared for the Patna Museum chanced to have arrived at the time. These impressions were practically worthless, being badly taken on single sheets, yet they sufficiently showed that the alleged "Yaksha " was in neither inscription, a view in which the Hon'ble Mr. Walsh, Vice-President of the Society, agreed with me the moment I showed the impressions to him. The letters, however, which Cunningham had declared to be later than Asoka, presented to me a wonderful problem. They did not fully tally with characters of any period yet known to Indian Epigraphy. While one letter, n, at first sight appeared to belong to a later age, all others disclosed forms more archaic than the oldest known Brahmi characters. The archaism was so marked that four letters, afterwards identified as bh, dh, s and s, appeared to me to be new forms. To them value could be assigned only on presuming them to be ancestors of such Aśokan letters to which the latter can be carried back on principles of epigraphic evolution. I arrived at a tentative reading and deferred final judgment for a few days until I went to Calcutta, which I had to visit on business, towards the end of the same month. I utilized that opportunity and examined the inscriptions on the statues during my spare time in Calcutta on six different days.

The inscriptions are on the folds of the scarf just below the shoulders on the back of each statue (see photograph D). It seems that the artist thought it profanity in art to cut the letters into the body. After a long scrutiny I came to the conclusion that the letters had been carved before the parallel lines to denote folds on the scarf were chiselled. I consulted Mr. Arun Sen, Lecturer in Indian Art to the University of Calcutta, on the point, and he confirmed my view. The fold-lines have continued in spite of the letters. Over the letters they have been

very delicately handled while the symmetry of the lines is kept on, the forms of the letters have not been interfered with, the original strokes of the letters being scrupulously avoided and kept separate.

I had six impressions of each inscription taken, thanks to the courtesy of Mr. D. R. Bhandarkar, the officer in charge of the archæological section of the Museum. My reading is based on those impressions and verified from the actual letters inscribed on the statues.

The inscription on the statue with the head on (figure A) is as follows:

Bhage ACHO chhoni'dhise.

The aksharas ACHO are larger than others, as if they are put in capitals. The first title bhage is grouped separately to make one word. ACHO again is grouped separately. The first letter is taken to be bh. The upward projection of the top line as it appears in Aśokan bh is not present here. That is a later evolution. The letter in our inscription is written in three strokes, the pen being set and taken off three times, the left-hand and the right-hand strokes having been drawn independently of the top line, while the Asokan bh tends to be done in two strokes (cf. Bühler's Table II, line 31, column VI; and the Bhattiprolu letter). The Asokan bh's have been written in the following way which led to the introduction of the upper projection: the right-hand vertical line was drawn first and then in one flourish commencing over the top of the right-hand line the rest of the letter was completed. The point can be clearly followed by a reference to the bh in the Sohgaura plate (J.R.A.S., 1907) where the two divisions are separate. In the attempt to draw the top line and the left-hand line together, the initial end became pointed upwards. One can verify this by attempting to produce the letter with ease on the principle of the Sohgaura bh. Our three-stroke letter on the statue is thus older in evolution. It is indeed not possible to take it as any other letter than bh.

"

The peculiarity of the second letter, g, consists in that it is composed of two lines, a left-hand line projecting into a hook, and then a right-hand, slightly curved line drawn from top to bottom. The Asokan letter on the other hand is made up of two equal and convenient parts or it begins to be written in one stroke, e. g. at Jaugada, Delhi and Siddapur. The uneven strokes still linger at Bhattiprolu.

The third letter, a, would be recognized at once by epigraphists to be an old form. I may only point out that the two ears which are so widely apart in our letter, tend to coalesce and to lose their curve and become a two-stroke letter in Aśoka's time, e. g. at Kalsi, Jaugada and Siddapur.

The fourth letter, ch, has a special feature in its perpendicular line being produced independently of the lower body. The latter is composed of three strokes. As against this the Asokan ch is made up of only two strokes, the straight line and the base diagram, done without lifting off the pen. The only exception to this in Asoka ch's is the third specimen at Girnar which is the nearest approach to our ch in the whole range of Indian epigraphy. The next letter, ch2, in our record consists of three strokes, while the Aśokan tends to a two-stroke composition. The sixth character, n, is again composed of three strokes as against two of the Aśokan. Its similarity with later n is more apparent than real; for later n's are really two-stroke letters.

The penultimate letter is done in three strokes, two lines drawn down from one point and a base-line joining the two. It is a new form, and assuming a previous history to the Asokan dh our letter can only be ancestor to the latter, the other two possible cases of 9 and t being excluded by their actual occurrence in the inscription. Here again the Asokan letter (dh) is much easier-decayed in form-than this dh, the former being written in only two strokes-a curve and a straight line. The Bhattiprolu dh (Bühler 26, xiv, drawn upside down) is a compromise between our dh and the Aśokan. There the strokes

'It cannot be an e on account of the vowel-mark attached to it.

are still three but the right-hand line begins to curve. The original form still survives in the Andhra group though with a distinct tendency to a two-stroke form.

The last letter is still more original and its identification was a matter of some time. A long perpendicular line is drawn first and then by its sides, about the middle, two hooks are added in two separate strokes. At first sight one would be inclined to take it as a fourth century (A.C.) two-stroke k but the absence of seraph and the lower flourish together with the number of strokes would dislodge that proposal. It is radically different from k. If we follow the method of presuming an earlier form, we can on palæographic considerations trace the ancestry of the Asokan (Bühler, 37, II) and Bhattiprolu s (Bühler, 37, XIV) to this letter. The pivotal line has been contracted in the latter, its upper portion totally disappearing and the lower still remaining longer than the sidal legs. The legs, again, tend to hang down, while they hang on in the Saiśunāka letter.

In all these cases we find the Aśokan figures having reached a stage which costs much less exertion than the Saiśunāka ones. They are much easier, or to adopt an art expression, they are decadent as compared with our letters. The degree of evolutional decay between our letters and the Aśokan is nearly the same as between the Aśokan and Rudradāman's.

Coming to the palæography of the second inscription, the first letter is a new form. I was first inclined to take it as an older form of sh. Dr. Mazumdar, whom I consulted about the letter, discovered on the rock a fine chiselled line from the elbow joint upwards to the fold-line above. This line is so thin that the impressions do not reproduce it sufficiently. It is equally, or more, probable that the letter is a dental s. The lower flourish ends on the level of the base of the letters and does not turn upwards. The right-hand line is separated by a small ridge on the rock; it is therefore part of the next letter. The corre sponding upper flourish is a fork. The whole letter is composed

1 The continuation of the mark beyond the line is produced by a crack in the reeks; it is not chiselled.

of three sections, first the fork commencing with the inner line, then the crescent commencing at the elbow joint and ending by the bottom of the horizontal line of the next letter, and, finally, the upper stroke above the elbow. In Asokan alphabet both dental and cerebral S's are produced in two strokes, and the middle stroke ceases to be straight.

The second letter is made up of three distinct strokes: the right-hand, the base, and the left-hand lines. The right-hand one is drawn from top to bottom and the left-hand one from bottom to top, i.e., it has the composition of p as against l. The left-hand line is a shade shorter than the right-hand one. P's in Asokan groups, except the Delhi letter, are produced in one stroke, the left-hand end becoming short. The older form persists at Delhi and later at Pabhosa, Mathura and Hathigumpha. The lefthand line becomes shorter still as time proceeds. The third letter, kh, again, has an older feature. The body is formed of four lines, which becomes round or tends to disappear in Aśoka's time. The hook in our letter almost touches the quadrilateral and has a nose to the right. In the Aśokan the latter detail is already lost and the letter becomes much easier in shape, lessening the curve.

The fourth letter (t) consists of three parts, two making up' the legs and one, the top vertical line, put on separately. All Asokan and later t's, on the other hand, are only two-stroke forms. Our letter has a faithful descendent in the Kalsi letter (Bühler, 23, III), but that also bears the mark of time in being a two-stroke diagram. The next one, v, is a combination of two side strokes, curvish in form, a straight base, and finally a vertical line above the body. The Aśokan v becomes completely round and with the vertical line a two-stroke character. The

form nearest to our letter is preserved in a Bhattiprolu variety (Bühler, 36, XIV). The older form lingers at Mathura, Bhar

hut and Hathigumpha, but there the curve in each case has long disappeared and a straight line taken its place. The sixth character, f, is like the Asokan letter. The next one, n, is, as in the first inscription, drawn in three strokes. The last letter, d, is