on some Carnivorous Insects. 333 



" drivers." It was at once covered with the ants, but a 

 hornbill at once strode up, picked it out and, after shaking 

 off only a few of the ants, swallowed it with all the rest 

 that were clinging to it. 



Remarks. — The ants were unconfined, carrying on their 

 ordinary daily avocation. The experiments cannot there- 

 fore be criticised as having been carried out on animals 

 under " highly unnatural " conditions. Yet the ants showed 

 strong " preferences," readily taking some animals when 

 they would not take others at all, and when failing in their 

 attacks on yet others. It is true that some of the winged 

 insects offered were, because winged, not such as the ants 

 would normally have an opportunity of seizing except 

 when hunting at night — but they do hunt at night greatly. 

 Most of these were nevertheless taken by them, and, even 

 if we should exclude these as not forming a part of the 

 ants' normal food, we should find that a number of species 

 were offered that the ants must meet with continually, 

 and that very strong " preferences " weie shown even as 

 between these. 



The butterflies tested, the moth Rhodogasfria bubo, the 

 larva of Papilio demodocus, a hive-bee, the larvae of a bee 

 and of a wood-boring Hymenopteron, a cockroach, the 

 beetles Himaiismus, Systates, and an Elaterid, and a Zono- 

 cerus elegans were probably all less protected from Dorylvs 

 than even the weakened larva of Aletis monteiroae, and 

 certainly than the larvae of Acraea acara, than adult 

 Mylabris amplectens, Epilachna polymorpha and E. hirta, 

 and Antestia variegata, and all these, apparently, than the 

 fly, the larva and pupa of Amauris albimaculaia and the 

 beetle Psammodes (protected by hardness). But the ejection 

 from the tunnel of a large number of Epilachna a day or 

 two after they had been taken in may indicate that these 

 were found to be as bad as any of the objectionable species. 

 Certainly the ants found the fly and the Danaine larva 

 and pupa much more obviously unpleasant than the 

 Epilachnas, Mylabris, Antestia, and Acraea larvae, and 

 it is clear from an experiment to be described below 

 that the latter when well-grown are acceptable enough 

 to them if they give the ants time, and the latter are 

 persistent enough, to raze the juice-deahng hairs. 



The butterflies used were Danaida chrysippus, L., 

 Amauris albimacidata, Butl., Mycalesis campina, Auriv., 

 Acraea neobule, Dbl. and Hew., A. egina. Cram. var. areca, 



