1939] 
Cannibalism Among Larvae 
31 
conditions (Dethier, 1937) may be explained by the fact 
that chance meetings are more frequent. When similar 
conditions were reproduced in a cage twelve by sixteen 
inches, both animals eventually died of starvation. 
While experiments with mutilated animals indicate that 
thirst may be one factor in inducing cannibalism, the follow- 
ing experiments demonstrate that hunger by itself is an 
important factor. 
One E. acrea was kept in a moist atmosphere, given its fill 
of water, and presented with every opportunity to continue 
drinking. When given a mutilated larva of the same species 
the experimental animal began feeding almost immediately. 
Repetitions of this experiment prove that hunger as distinct 
from thirst is one factor in inducing a carnivorous diet. 
The experimental animals never chose a diet of meat in 
preference to plants. Plant food was always accepted even 
after the larvse had gorged themselves with meat. In three 
hours a single E. acrea consumed one E. sordidus, another 
consumed one entire Gryllus assimilis Fab., and an I. isabella 
consumed one full-grown larva of Vanessa virginiensis 
Drury. These three arctiids pupated and produced normal 
adults. 
With regard, therefore, to the role played by hunger and 
crowding in causing cannibalism the following conclusions 
seem justified : First, the degree of hunger is of considerable 
importance. Larvse in the initial stages of hunger are not 
readily induced to eat flesh unless stimulated probably by the 
odor of the body fluids and more certainly by direct contact 
with them. Larvse in the final stages of starvation yet still 
active enough to crawl about do not require such an intense 
stimulation. Since animals in this condition habitually 
nibble at near-by objects, they eventually bite through the 
integument of an intact carcass or a quiescent animal. At 
this juncture they too are stimulated by the flesh within. 
An exceedingly active victim is not actually eaten till it has 
been rendered more or less quiescent although it may still 
be capable of considerable movement. Second, crowding 
facilitates the initiation of the events already mentioned as 
caused by hunger. In addition crowding induces attacks 
not prompted by hunger (Balduf, 1931 ; Dethier, 1937) . 
