18 
Psyche 
[Vol. 89 
which usually had gathered near the cotton plug. In the following 30 
seconds I observed the reaction of the ants, and between each test I 
waited at least 10 minutes before another sample was inserted into 
the nest tube. These pilot tests (3 repetitions with each organ) clearly 
indicated that only crushed poison glands and pygidial glands eli- 
cited increased locomotory activity and attraction in Cerapachvs 
workers. The ants did not exhibit any particular behavioral reaction 
when sternites, hindgut or crushed Dufour’s glands were intro- 
duced.* For the next series of experiments I first initiated colony 
emigrations either by following the procedure described above, or 
by shaking the colony out of the nest tube onto the arena floor. 
Before each experiment the arena was provided with a new paper 
floor. A new nest tube was offered 15-20cm away from the old nest 
tube or the displaced colony. 
Once the colony emigration to the new nest tube had commenced, 
I covered the floor area between the colony and the new nest site 
with a cardboard, onto which I had drawn two artificial trails, one 
with a crushed glandular organ to be tested, and a second one with a 
drop of water (control). The trails were made to originate either 
from the entrance of the nest tube or from the periphery of the 
clustered colony. Each trail (test and control) diverged through an 
angle of 45° to either side from a possible natural trail (which was of 
course covered by a piece of cardboard). In addition the whole 
paper floor was rotated for 90°, in order to control for possible 
visual orientation (Fig. 10). During the following 2 minutes I 
counted the ants following the trails (10cm long) to the end. Only 
trails drawn with crushed poison glands elicited a precise trail fol- 
lowing behavior in Cerapachvs workers. There was some initial 
following response to trails drawn with crushed pygidial glands, but 
the ants followed only through the first 1-3 cm, then usually turned 
or meandered off the trail. Only once was it possible to conduct a 
similar test during raiding behavior of Cerapachvs. In this instance 
the ants followed only an artificial trail drawn with a crushed poison 
gland. 
Although pygidial gland secretions did not release trail following 
behavior in Cerapachvs, it clearly elicited increased locomotory 
*Cerapachvs has also a very well developed sting sheath gland. It was not possible to 
test whether or not secretions of the gland play a role in communication. 
