16 THE POISON OF ANTS. 



is not thus fired at the enemy ft-om a distance, there 

 are two cases in which the sting might be allowed to 

 fall into disuse. Firstly, those species which fight 

 with their mandibles might find it on the whole most 

 convenient to inject the poison (as they do) into the 

 wounds thus created. Secondly, if the poison itself is 

 30 intensified in virulence as to act through the skin, 

 a piercing instrument would be of comparatively small 

 advantage. I was amused one day by watching some 

 specimens of the little Gremastogaster sordidula and 

 the much larger Formica cinerea. The former were 

 feeding on some drops of honey, which the Formicas 

 were anxious to share, but the moment one approached, 

 the little Cremastogasters simply threatened them 

 with the tip of their abdomen, and the Formicas 

 immediately beat a hasty retreat. In this case the 

 comparatively large Formica could certainly have had 

 nothing to fear from physical violence on the part of 

 the little Gremastogaster. Mere contact with the 

 poison, however, appeared to cause them considerable 

 pain, and generally the threat alone was sufficient to 

 cause a retreat. 



However this may be, in their modes of fighting, 

 difi*erent species of ants have their several peculiarities. 

 Some also are much less mihtary than others. Myr- 

 mecina LatreiUii, for instance, never attack, and 

 scarcely even defend themselves. Their skin is very 

 hard, and they roll themselves into a ball, not defend- 

 ing themselves even if their nest is invaded ; to pre- 



