108 Psyche [September 
suction pressure from the activated bulb (Eisner, 19571 Eisner and 
Brown, 1958). 
Aside from its obvious advantage in making prolonged crop storage 
possible, the permanent restriction of the proventricular portal poses 
a problem. Solid particles passed into the crop have no place to go, 
since they obviously cannot be pumped through the proventriculus. 
What, then, happens to indigestible solids that are swallowed ? Or are 
they perhaps not swallowed at all? Ants, as well as many other 
Hymenoptera, have a so-called infrabuccal chamber, a ventral infold- 
ing of the hypopharyngeal surface, that could conceivably act as an 
effective solid-withholding device, guarding the opening to the crop. 
Janet (1895a, 1895b, 1905) has described the anatomy of this struc- 
ture, and has shown that in both ants and wasps, debris gathered while 
cleaning themselves or their nestmates, as well as solid residue from 
food ingested, often collects in this pocket and is ejected intermittently 
as small discrete pellets. His observations were excellent, but left some 
basic questions unanswered. The present study deals with an experi- 
mental evaluation of the function of the infrabuccal pocket of a 
formicine ant, Camponotus pennsylvanicus (DeGeer), in which crop 
storage and regurgitative food transmission are known to be well- 
developed social attributes (Plate 6). 
Several laboratory-maintained colonies of Camponotus were avail- 
able, but only nestmates, rather than internidally mixed lots, were 
used for any one series of tests. For experimental purposes they were 
fed honey mixed with various samples of corundum powder ranging 
in particle diameter from iO/x to 300 [i. 
EXPERIMENTS WITH INDIVIDUAL ANTS 
An initial series of tests was designed to determine just how effective 
a filtering device the infrabuccal pocket really is. Individual ants, 
including only medium-sized workers, were confined in Petri dishes in 
which they had access to single drops of honey-corundum mixture. 
Eight corundum samples were tested (10, 20, 30, 80, 100, 150, 200, 
and 300 /i ) , each on 10-15 ants. The ants were starved for one or more 
days before the tests, and, when introduced into the dishes, each would 
promptly commence feeding and remain at the food source uninter- 
ruptedly for up to several minutes. Only in a few exceptional cases 
would an ant pause briefly partway through the meal and, after back- 
ing away slightly and spreading its mandibles, would regurgitate a 
small, typically kidney-shaped infrabuccal pellet, consisting of a densely 
clumped packet of corundum plus small pieces of wood and other 
