OBSERVATIONS ON SOME SOUTH AFRICAN TERMITES, 41 



latter. If by chance an ants' nest is somewhere near to a 

 trinervius mound and the mound is broken the ants will 

 swarm around ; they do not, however, display that delirious 

 blood-lust exhibited, under like circumstances, against bilo- 

 batus, but behave with the greatest circumspection, never 

 facing the soldier termite, but awaiting an opportunity to 

 seize it from behind, in the region of the nape. 



As the viscid fluid ejected by the soldier of this species, 

 considered as a missile of defence, can only be most 

 ineffectual against a large marauder of the nest, it must 

 be regarded as their chief form of defence against true 

 ants. 



In the foregoing account it is assumed that each triner- 

 vius colony inhabiting a mound is a separate colony, despite 

 the fact that its mound may be connected by undergTound 

 galleries with another mound or series of mounds. No experi- 

 ments have been made to ascertain how far the insects of 

 adjoining mounds are antagonistic or friendly, but Mr. Bag- 

 shawe-Smith assui-es me that whenever he introduces a 

 foreign queen to a mound the soldiers attack her, sliming her 

 over with their viscid secretions. 



V. CLASSIFICATION. 



1. Introduction. 



This contribution is designed to fix the identity of the 

 species whose habits have formed the subject of the preceding- 

 discussion and that of several allied species. That some species 

 are presented as new is incidental. 



For the purpose in view the soldier caste has received chief 

 considei'ation, but worker characters, essential for their deter- 

 mination or for the qualification of those of the soldiers, are 

 also given. The images have not been described. Particular 

 attention has been given to the full characters of the soldier 

 mandibles and more especially to the basal regions which are 

 oi'dinarily hidden by the labrum ; these charactei's ai-e as 



