Rev. R. Baron — Rock Cavities in Granite. 19 



contains three or four basin-shaped depressions. Other cavities 

 exist further south. 



On the mountain of Ambohitsileo there are several other cavities, 

 one of which is probably larger than any of those mentioned above. 



On the north side of the mountain of Angavokely there is also 

 a large cavity. 



It stands to reason that if in the course of two or three com- 

 paratively short rambles so many rock cavities should be found, 

 a further acquaintance with the granitic area would probably reveal 

 the existence of others in large numbers. 



How, then, are these rock-holes to be accounted for ? I have 

 known of their existence for many years, but they had practically 

 passed out of my mind until I came across the articles in the 

 Geological Magazine for 1904. The explanation of their formation 

 which always occurred to me when I thought about them was that 

 they represented what were originally nests of felspar, which had 

 since decayed and disappeared. I have, however, always felt the 

 explanation to be unsatisfactory, and for the following reasons : — 

 (1) Though segregations of the more basic minerals (often taking 

 the form of true diorite) are quite common in the granite, segregations 

 of felspar (except here and there in veins) 1 have never come across. 

 I do not, of course, say that there are none such. (2) Felspar nests 

 can hardly be supposed to have had in all cases the remarkably 

 rounded outlines and the rounded protuberances on the surface 

 which the cavities, such as I have described, seem to require. 

 (3) If these holes really represent nests of felspar which have rotted 

 out and disappeared, it is difficult to account for the felspar crystals 

 on the present walls of the cavities remaining comparatively fresh ; 

 and (4) it is difficult to see by what process, on this supposition, 

 the decayed felspar has been so cleanly scooped out of the holes. 



The theory that these numerous cavities have been eroded by 

 the action of wind-carried sand cannot possibly account for them, 

 at any rate in the area I am speaking of, for the simple reason that 

 there is no sand anywhere near worth speaking of, and sand-laden 

 winds are unknown in the interior of the island. Moreover, would 

 not sand-abrasion produce striee round the walls ? The theory on 

 the face of it seems far-fetched. Abrasion of plain surfaces by 

 blown sand in the open country is quite explicable, but that the 

 sand-laden wind (supposing sand to be present) should gyrate 

 and bore at particular points sufficiently long or frequently to 

 produce these remarkable cavities seems to me entirely incredible. 



" Noi'," as Professor Bonney says, " does the action of dew or 

 of moisture in any form seem a promising explanation, for," etc. 



If, then, these rock cavities are due neither to the decay and 

 disappearance of knots or nests of felspar, nor to erosion by blown 

 sand, nor to the action of " dew or moisture in any other form," 

 how are they to be accounted for ? 



I make bold to suggest an explanation, to which at present I can 

 see no objection, and which, at any rate, would account for the 

 phenomena ; it is tliat these cavities represent sjjaces occupied hy 



