496 



javana is very widespread in the Suiida Islands, and makes its nests in 

 other ways besides. Emery is, therefore, of the opinion, which is 

 doubtless correct, that Pheidole merely robs the nests of Irldomyrmex 

 occasionally and is not adapted to the plant. On the other band, 

 Emery regards Cremastog aster deformis as the constant guest of the 

 Hydnophytum. In all the Myrmecodia and Hydnophytuni bulbs that I 

 received in alcohol through Dr. Frick there was a colony of the smaller, 

 darker Javanese variety of Iridomyrmex cordatus Smith ( Var. Myrmc- 

 codke Emery), including males, a few pregnant females, and numerous 

 larvjx3 and pupa*. All the specimens of the plant had the same hollow 

 labyrinth, looking like a nest built by ants. It must be added that the 

 genus Iridomyrmex is very closely related to Azteca and Liometopnm, 

 but comprises many species which build nests of earth. 



Now, Treub has ascertained (Annals of the Botanical Garden of 

 Buitenzorg, Vol. VII, 18SS, p. IDl) that Myrmecodia raised from seeds 

 in hothouses and in the botanical garden of Buitenzorg develop the 

 whole hollow labyrinth in their bulbs, to complete maturity, without 

 the presence of ants, just as well as those which, in a state of freedom, 

 possess ants. At the same time the jilants thrive admirably. This 

 proves that the labyrinth is produced by the i)lant, and not by the ants, 

 though Beccari repeatedly found severed bundles of vessels in the com- 

 partments. Treub has, consequently, resorted to other attempts at an 

 explanation, and has regarded these singular cavities as the breathing- 

 organs of the plant, and connected them with the interior irrigation of 

 their web (the compartments have a very watery web), which, in view 

 of the epiphytic situation of the plant upon trees with little foliage, is 

 I)lausible. My own opinion, however, is that Beccari's observations 

 are correct, and that the ants lend their aid by connecting some of the 

 cavities by bored passages, as the natural cavities do not all appear to 

 me to have a natural communication with each other, such as the ants 

 require. Furthernu)re, the founders of the colony, the mother females, 

 must first bore in. At all events, only the adaptation on the part of 

 the ant is well ascertained — that is to say, in the case at least of 

 Iridomyrmex eordatns^ which finds its exclusive, beautiful, and secure 

 dwelling in the plants of the genera Myrmecodia, Hydnophytum (and 

 Bischidia'i). Botanists say that these plants furnish no special food to 

 the ants, at least nothing has been found corresponding to the Midler's 

 corpuscles of Cecropin. Since, however, most of the Dolhlioderi keep 

 no plant lice, but lick up the secretions of plants, or devour insects, a 

 closer investigation of the mode of feeding of Iridomyrmex cordatus 

 would probably bring some interesting facts to light. Besides, Treub's 

 observations do not prove by any means that the plant does not form 

 the labyrinth for the purpose of serving as a dwelling for the ants. 

 The fierce inmates certainly afford it protection against its enemies. 

 We must investigate, as S(>]iim])er did for Cecropia, whether there are 

 kindred species of plants, not inhabited by ants, with or without similar 



i 



