208 TIIE INDIAN CEMETERY OF THE GRUTA DAS MUMIAS. 



Examining the surface of the rock in the interior of the grot- 

 tos it will be seen that the gneiss is suffering a very rapid decom- 

 position, and is scaling off in thin flakes, which are sometimes so 

 soft as to break up readily between the fingers. 



As the gneiss is very compact and had originally but few frac- 

 tures, and as the decomposition progresses from the outside 

 inwards, the rock of course decomposes concentrically, giving rise 

 to more or less regular, concave surfaces. The surface of the 

 rock inside the cavern is constantly damp, but not sufficiently wet 

 to drip. I suppose that this dampness is for the greater part 

 caused by the soaking through the solid rock of water from above, 

 and that the decomposition is caused mainly by the action of car- 

 bonic acid derived from the air. 



Large caverns like that just described are rarely encountered in 

 the gneiss of Brazil, but small ones abound and may be seen in 

 the precipices of the gneiss hills of the vicinity of Rio. 



It is somewhat difficult to determine just how the caverns of 

 the Morro de Diogo Velho at first originated, but it is very likely 

 that they commenced by the decomposition of an isolated mass in 

 the gneiss, that had a somewhat different mineralogical composi- 

 tion than that of the rest of the rock. Ordinarily, cavities of this 

 kind soon disappear from the surface of a cliff, because of the 

 scaling off of the thick, half decomposed sheet, which falls from 

 time to time, leaving a new surface exposed. It is not at all as- 

 tonishing that the decomposition should go on irregularly and that 

 the cavity should enlarge itself in some parts more rapidly than in 

 others, giving rise to the pot-hole-like excavations above described. 

 A very slight difference in the hardness of the rock, or in the 

 amount of moisture, would be sufficient to determine the more 

 rapid decomposition of a certain part of the surface, giving rise 

 to a hollow. On the Rio Tapajos the edges of the beds of coal- 

 measure limestone, exposed to the action of the waters of the 

 [garap6 de Bomjardim, during the rainy season, are not dis- 

 solved away evenly, but are honeycombed with grottos. Witness 

 also the way in which metals and other substances are honey- 

 combed by acids. 



The upper grottos of the Morro de Diogo Velho are like the 

 lower caverns, but smaller. I shall not describe them particu- 

 larly, because archseologically they do not appear to be of interest, 



