66 BULLETIN 647, U. S. DEPARTMENT OF AGRICULTURE. 
which a poisoning experiment was being conducted. As the jars 
used would be partly filled with water every time it rained, a shelter 
was made for each of them out of two ends of an orange box nailed 
together in the shape of a gable. Soon after these covers had been 
laid, every one was found to harbor an ant colony which crowded it 
to its fullest capacity. Protection from the rain alone not satisfying 
the ants, they shut out the light and drafts by completely filling in 
the space, between the jar and the top and ends of the cover with 
particles of soil. Each nest was no doubt ideal from the ants' point 
of view. Each was honeycombed with galleries of all shapes and 
sizes, ramifying in every direction. Any desired degree of moisture 
could probably be had in these galleries, those nearest the top being 
driest, and those directly on the surface of the ground most humid. 
The occupants would only need to move to underground quarters 
should prolonged dry weather occur. Each of these " ant castles," 
as they might be called, was perfectly protected from rain by a good 
pine roof. On removing the roof and looking into the galleries, 
thousands of eggs, young, and queens were revealed. Many solid 
masses of young and eggs as large as a hen's egg could be collected 
in these shelters. It was very evident that the ants could be much 
more rapidly destroyed when gathered together in this manner than 
by the tedious and unsatisfactory method of feeding them poison. 
Protection from rain and drafts, good drainage, and darkness be- 
ing the principal nesting requisites of the ants, it appeared that these 
requirements could best be met by a box with a roof. It was also 
found that ants could be induced to mass more thoroughly in numer- 
ous comparatively small colonies than in a few extremely large ones. 
A small covered box-trap was, therefore, given a trial, 15 of them 
being used in the first test. This was so satisfactory that it was fol- 
lowed by a large-scale experiment in which over 400 traps were used. 
Description of Ant Traps and Fumigating Covers. 
The first traps (see PL V) were made of f-inch cypress; but sap 
pine proved to be just as good and was cheaper. Each trap con- 
sisted of the following 9 pieces : Two sides, 12 by 12 inches ; 2 sides, 
10 by 12 inches; 1 bottom, 10 by 10 inches; 2 top pieces, 8 by 12 
inches, and 2 pieces of triangular molding 12 inches long. First the 
smaller sides and bottom are fastened together with rosined nails 
or screws to prevent warping, then the larger sides added. The top 
pieces are fastened together in the form of a gable, with a tight 
joint, this roof being set loosely on top of the box. The pieces of 
molding are nailed across the inner side of the roof where it touches 
the top of the trap to hold it in place. 
Covers to keep the gas in while fumigating are made of 28-gauge 
galvanized iron, each consisting of one piece 38 inches by 13£ inches, 
