198 



DESCRIPTIVE REMARKS ON 



carved so frequently throughout the monoliths, the caves and 

 the older structural temples here. The outer edge of the veran- 

 dah roof is carved into a continuous bold projecting cornice 

 without a break all round, of single (convex) curvature, and 

 ornamented as usual with the horse-shoe facets containing the 

 human heads or faces. These faces or masks, it should be 

 observed, are or were well wrought and diverse ; but most of 

 them have a profusion of hair often quite a shock, or like an old- 

 fashioned wig. 



The ridge of the roof has the bases or pedestals of 1 8 kala- 

 sanis, and one more is missing from the peak of the south 

 gable, which would make up 1 9, though the two end ones may- 

 have been flat horns or spikes, such as those that surmount 

 the gable ends of all the oblong dormer cells and of the 

 horse-shoe facets, or they may have been triMl finials, such as 

 those that cap the gables of the Ganesa temple or Arjuna's 

 Hatha, No. 24, at the north end of the rocks. This latter mono- 

 lithic Eatha in fact much resembles Bhima's Eatha, No. 42, only 

 it is three stories high instead of two, is very much smaller, and 

 is a mere model of a lofty building, containing but a little shrine 

 or image cell and a very narrow small portico, without any 

 verandah besides, whereas the large waggon-roofed monolithic 

 Eatha of Bhima might well have contained a fair sized chapel ; 

 but from the fact that the cavity of a central cell has been begun 

 in the middle of the west verandah, which would here take the 

 place of the usual portico, it seems probable that perhaps only 

 a shrine cell was intended. 



This very beautiful design is almost unique and would seem 

 to be the prototype of Kamaraja's or the Ganesa temple (or 

 Arj una's Eatha) at the north end of the Mavalivaram rocks, and 

 also of the ordinary waggon roof of all the Gopurams or gate- 

 way towers of the South Indian (Dravidian) temples. 



The only roof like it that I remember to have met is that of 

 one of the large halls (of no great antiquity) in the lower part 

 of the fortress of Ohenj i ( * 1 Gingee' ' ) . The waggon-vaulted roofs 

 of the Nayakkan style in Tan j ore and Madura are seldom if 

 ever between vertical (gable) ends, and the copper roofed halls 



