1907.] Wheeler, Fiingus-groxving Ants of North America. 781 



compact mass of very much branched and contorted hyphpe. The formation 

 of the oidia, or process whereby, according to Hohermann, the hyphie in 

 the interior of the spherules breaks up ahnost completely into very short 

 oval cells, is by no means so complete in our species. To be sure, the 

 hyphre are constricted in the interior so that they appear as rows of short 

 oval cells, completely filled with protoplasm, but these cells even in the larg- 

 est spherules, which have reached their full development, remain attached 

 to one another so that when a thin section is pressed under the cover glass, 

 only a few of the cells escape. In the spherules described by Holtermann, 

 on the contrary, slight pressure on the cover-glass sets free thousands of 

 oidia." 



The mounds of T. vulgaris {— affinis Trag&rdh) are as large as those of 

 natalensis (1.4 m. high and 5.5 m. in diameter at the base), but the structure 

 and arrangement of the chambers is very different. They are separated 

 by thick walls and communicate with one another by very tenuous galleries. 

 Each chamber has a flat floor with a peripheral groove and an arched roof. 

 The gardens, which are shaped like inverted dishes and are not confined to 

 the smaller peripheral chambers, are often concave beneath, with a ridge 

 around their border fitting into the circular groove in the floor of the chamber. 

 The substratum consists of the same materials as in natalensis and is per- 

 forated with numerous transverse galleries. Concerning the fungus Tra- 

 g&rdh says: "The spherules are much smaller than in natalensis, are like 

 these nonpedunculate, and occur in great numbers on the walls and espe- 

 cially on the roofs of the cavities and galleries in the peripheral portions of 

 the gardens. These portions are also stuffed with larvie and nymphs. The 

 spherules are unlike those of T. natalensis in structure, since as shown in 

 Figs. 2 & 3 PI. Ill [reproduced in the present paper as Figs. 58 & 59, PI. 

 LIII], the cells in the outer layer of the spherules are larger than those in 

 the interior. Both the inner rows of cells, M-^hich ramify dichotomously, 

 and the outer ones, are in part empty, in part filled with finely granular proto- 

 plasm." Although Tragt\rdh found fungus-gardens in the nests of T. 

 trcegardhi {— incertus Tr'Agh.) which seems to live as an inquiline in the nests 

 of T. hellicosus, natalensis and vulgaris, he believes that these had been 

 stolen from the host termites and that trcegardhi does not itself grow fungi. 



Doflein (1905, 1906) has contributed more recent observations on the 

 gardens of termites. He studied colonies of T. obscuriceps in Ceylon. 

 The mounds of this species are about 2 m. high and terminate above in one 

 or more huge tubular, chimney-like orifices Avhich open into the galleries 

 and chambers in the interior of the nest. The chambers arc about as large 

 as a cocoa-nut or smaller, with smooth walls and excavated to a depth of 

 1^ m. below the surface. The gardens, which consist of comminuted wood 



