UPON THE ARCHITECTURE OF WESTERN INDIA. 175 a hard jasper, and therefore as unfavourable to dehcacy of work as can well be imagined. This circumstance seems to have caused the architect to despair of effecting his exact pro- portions, and to leave this to be accomplished by a coat of stucco, traces of which are to be found everywhere in these excavations. Hence it is not possible in some cases to ascertain what the mouldings have been, as A, fig. 3. This figure represents a portion of the screen which sepa- rates the vestibule from the apsidal chamber. It exhibits all the architectural characters of the screen, which is covered with sculptures and inscriptions. The screen which separates the vestibule from the outer court is quite plain on its inner side, Avith the exception of some niches, which may have been subsequently fashioned in the naked wall. The outer side of this screen is ornamented with sculptures of Buddhs and lotus flowers. A low screen at the entrance of the outer court is of a similar character, and of this a portion with its mouldings is given, fig. 4. This screen is strictly Buddhist, the lotus flower being their favourite emblem. The parabolic moulding then has not been made use of in the ordinary ornaments of this temple. In the small cells, however, it frequently occurs, and its position is remarkable. In the excavations elsewhere, both of the Deccan and of the Concan, the rock being trap, the cells are openings in preci- pitous escarpments. But at Kenneri, the surface having a gradual slope, it has been necessary to form a little court in clearing away a sufficient height for the front of the cell ; and into this court the water, during the rainy season, must pom- down from the slope above. Just at the edge of the slope, where the water would thus run over, there is cut a parabolic moulding, remarkable for its magnitude and boldness. The cells have very frequently, carved on a pillar or some smoothed surface, the emblem, fig. 5; sometimes it is the only ornament to be seen. The curve is evidently that of the para- bolic moulding. In one instance the stone ap])eared so smooth and the lines so fine that it seemed possible to ascertain whether the cm-ve is truly parabolic. Accordingly the mea- sures AB = BC = CD = DE=1:F = 3 in. being laid off, and GBL, HCM, IDN, JEO, KEF, being drawn perpendicular to AF, the folio win Qj measurements were made :
