750 Bulletin American. Museum of Natural History. [Vol. XXIIJ, 



Nest D was unique in having Chs. II and III opening directly into each 

 other. Nests of the simpler racemose type, like G, are more frequent than 

 simple nests with as many as four and five chambers, like D and E. In 

 nests G-I the second gallery sent off a branch terminating in a chamber of 

 its own (Ch. Ila). The terminal chamber of nest H (Ch. Ill), like that of 

 nest E (Ch. IV), was very small and obviously in process of being excavated 

 by the ants. In nest I the insects had completed at least a portion of the 

 gallery (Gal. Ilia) leading from Ch. II and the ants, had they been left 

 undisturbed, Avould probably have widened its end into another chamber 

 (Ch. Ilia). In nest J, the largest and most complicated of the series, not 

 only did Gal. II form two branches, but one of these divided in turn, so 

 that there were three galleries, each terminating in two chambers (Chs. II a, 

 b, c, and Ch. Ill a, b, c) separated by a gallery (Gals. Ill a, b, c). Since 

 in all of the nests the galleries formed an angle with the surface of the sand, 

 their total depth, as given in the last column of the table, does not represent 

 the vertical distance of the floor of the terminal chamber from the surface, 

 but the oblique distance from the entrance. Both simple and racemose 

 nests, moreover, though represented in the figures as lying in a single plane, 

 are often bent, or, like nest I, of the latter type, radiate out from the entrance 

 in three different intersecting planes. 



When establishing their formicaries the ants select only those spots in 

 the woods where the sand is permeated with fine rootlets. They are careful 

 to leave these untouched, while hollowing out their chambers, as supports 

 for their gardens, which in this, as in other species of Trachymyrmex, are 

 always pendent and do not rest on the floor of the chamber like the gardens 

 of Atta s. str., Acromyrviex and Mwllerius. The substratum on which the 

 fungus is grown consists very largely of caterpillar excrement and withered 

 oak-catkins, both picked up under the trees, but often small dead leaves or 

 berries are used, and occasionally as Morris and McCook observed, flowers 

 or green leaves are cut from the small herbaceous plants in the neighborhood. 

 These substances are comminuted and placed on the pendent rootlets where 

 they become knitted together by the rapidly proliferating fungus mycelium. 

 The whole garden then hangs from the roof of the chamber as a cluster of 

 nodular strands or plates separated from the walls and from one another by 

 spaces sufficiently large to admit the ants to all parts of the structure. The 

 first chamber, in which the original worker brood was reared by the queen, 

 is often empty or has lying on its floor particles of exhausted vegetable sub- 

 stances ready to be carried out of the nest, or materials that have just been 

 brought in. This chamber seems to be the work-shop in which the materials 

 are prepared for insertion into the hanging gardens of the lower chambers. 

 The appearance and arrangement of several of these gardens are shown in 



