Chai". VI] 



ANIMALS 



M9 



Df the more or less thin partition-walls: the chambers themselves contain 

 \\r and are inhabited bj- ants. Fairlj' numerous, but ver\- small, openings 

 illow for communication with the exterior. From them the ants rush 

 3Ut ready to attack, as soon as the tubers are touched. I have observed 

 VI)'rmecodia echinata and Hydnophytum montanum growing wild in 

 lifferent parts of Java and ha\-e always found the tubers inhabited. The 

 argest of the tubers, that I have observed, is shown reduced to one-third 

 if its size in Fig. S6, having been drawn from a specimen preserved in 

 Icohol. 



Fig. 85. Myrmecodia echinat.i. Tuber cut longitudinally. Below is an epiphytic fern. 

 West Java. Natural size. 



The structure and development of the tubers of Myrmecodia and 

 lydnophytum have been admirably described by Treub. He proved 

 lat the structures, which were considered by Rumphius and other later 

 bservers as a kind of ant-gall, are, with all their peculiarities, brought 

 bout without any assistance from ants. As regards their function, Treub 

 xpressed himself very cautiously ; yet he did not think himself warranted 

 1 considering the tubers as adaptations to ants, but was rather inclined 

 ) connect the utility of the chambers with aeration. IMost botanists who 



