OBSERVATIONS ON SGML] SOUTH AFRICAN TERMITES. 873 



best in the case of the Pretoria nest (PI. XXVIIT, fig. 2) 

 wherein the cavity was extended right into the dome, and the 

 arching of the shelves quite exaggerated.^ 



The royal chamber is externally a flattened, rounded mass 

 of clay, the exterior form being occluded by the shelves 

 bracketed to it and by the grooves which lead to the tunnels 

 through its walls (PI. XXYIII, figs. 5 to 8). It is suspended, 

 as it were, at a central point in the lower hemisphere ; but 

 not infrequently it partly rests upon some intrusive stone or 

 root, so excavated that it projects into the hive. When not 

 upon such foundation it seems to be in a very precarious posi- 

 tion, but the exact support to such royal cells has not yet been 

 sufficiently traced. 



The pillars which support or suspend each shelf are in 

 reality stairways, excepting those which connect the top shelf 

 with the dome of the cavity ; these latter can only be regarded 

 as suspenders. 



Most of the stairways are canted slightly and twisted a little 

 spirally. In section they are flattened, oval, and slightly 

 grooved on the surface traversed by the termites. This groove 

 causes the pillar to appear to be more spiral than it really is. 

 The canting of the pillar and the pathway upon it are both 

 factors in rendering the ascent easier. 



Upon the upper shelves the individual fungus-gardens are 

 arranged, placed like so many loaves of bread upon a larder 

 shelf. The lower and more restricted shelves at the basement 

 of the hive have so far always been found empty ; and, if 

 there is differential feeding of the young, it takes place in 

 these compartments. 



The gardens present a granular formation throughout, the 

 matrix and the entire surface showing coarse granulations. 

 They are of varying sizes and heights, being of greater height 

 in accordance with the available perpendicular space above 

 them. Irregular as they are in shape, differing in height and 



' This nest was not examined by nie, but the series of insects in the 

 collection bearing the same accession number are typical T. natal- 

 ■ensis. 



