No. 420.] AN EXTRAORDINARY ANT-GUEST. lOII 
and, when overtaken, escape death by rolling itself up into a 
hard ball, which the ants tried in vain to perforate with their 
mandibles. As soon as they desisted the Armadillidium would 
unroll itself, and the chase would begin again. Finally I was 
compelled to remove the crustaceans, as the larvze seemed to 
be in danger of being injured in the excitement of the chase. 
On the sth of November I obtained satisfactory evidence 
that the dipteron larva is not obliged to remain always with the 
same Pachycondyla larva. During the night one of the large 
larvae had moved and attached itself to the first abdominal seg- 
ment of an ant larva which already bore a commensal around 
its metathoracic segment. The two larva were oriented in 
opposite directions, z.¢., with their heads reaching around oppo- 
site sides of their host. Subsequently one of these commensals 
moved to an unoccupied Pachycondyla larva. I was not present 
when the change occurred, nor was I able to determine whether 
it was the originally stationary or the adventitious larva that 
moved. Although this observation makes it certain that the 
dipteron larvae can shift their position from one host to another, 
I am convinced, nevertheless, that they must do this with great 
reluctance and only under urgent circumstances, such as extreme 
hunger, the death of the larva to which they are attached, or, 
perhaps, when fully mature and about to pupate. 
As the days passed, the mature ant-larvee spun their brown 
cocoons one by one, and one by one the mature commensals 
disappeared. Did they also pupate and for this purpose conceal 
themselves in the soil of the nest? or had the ants at last 
detected the villains and converted them into food for the 
larvae which had not yet pupated? or did they stick to their 
hosts and pupate within the cocoons? In order to ascertain, 
if possible, the true state of affairs, I transferred the whole 
colony to a fresh nest and examined the soil of the old nest 
with great care. There were no traces of the missing commen- 
sals. The only remaining resource was to open the cocoons. 
Several of the cocoons which had been taken with the nest 
October 27 had hatched, but by November 10 there were still 
thirteen cocoons in the nest, as several of the larvae had pupated 
in the mean time. Five of these were opened, and in two, 
