248 REPORT 4, UNITED STATES ENTOMOLOGICAL COMMISSION. 
while others emit the spray in the direction of the axis of the pipe. The 
pipe, ij should be preferably so extended as to carry its jet, «, beneath the 
plants. The pipe, 7, may be inserted in the wall of the can at any point 
above the desired level of the liquid which closes the lower end of the 
pipe so that air cannot ascend to the large, tight reservoir. Hence, the 
liquid can rise no higher than to close its lower orifice. As soon as the 
liquid is lowered so much that air can enter and pass to the upper res- 
ervoir liquid enough will descend to close the entrance again. With 
this arrangement the feeding of the blast goes on, and the small can 
does not have to be filled. This supply -tube may have a stopper, or be 
connected with an automatic supply from a larger, higher reservoir, as 
previously explained. It maybe seen that whether the angle, x, or the 
side, a, or the side, n, be downward the passages at the corner, y, will 
always be above the liquid, so that w hen the ins trument is in operation 
it can be directed downward or upward, as well as horizontally. By 
this machine the blast is fed with certainty and strongly. Since there 
is no immediate communication of the surrounding atmosphere with 
the reservoir, the liquid is not raised in the pipe, x-y, by the ordinary 
process of atmospheric pressure, but the blast-pressure, which is more 
powerful, is applied. Here it should be noticed that the blast also 
bears down upon the liquid in the tube, x-y, as well as on that in the 
chamber, and were its force equal in both these places the water would 
not rise, for the elevation must be due to the reservoir pressure exceed- 
ing that in the feed-tube. To secure this condition the main blast be- 
tween the incurrent and excurrent orifices, at y, should be partially ob- 
structed, and this can be accomplished in various ways, here by one 
side of the tube, x-y, projecting into the main passage. It can also be 
done by a crook, or by having the main passage made larger beyond the 
pressure-hole. It should be noticed that with this device, even when no 
valve is used between it and the bellows, the liquid will not flow back 
into the bellows past the pressure-inlet, for there it naturally enters 
again into the chamber. 
An important feature of this device depends on the fact that it may 
be turned upside down, and so works with less effort, while gravitation 
causes the liquid to feed into the blast uninterruptedly by way of what 
was in the other case the distal pressure-hole, and the tube, x-y, or a 
hole preceding, it now serves to pass the blast-pressure above the 
liquid. 
An apparatus having these characteristics, also adapted to poisoning 
under surfaces and others, is shown by a plan view in Plate XXXII, 
Fig. (>. The bellows, v, has a very long blast-pipe, i, with an angular 
dischai^ing-uozzle, 8, which can be turned on the pipe, i. At the base 
of the pipe, i, is a crook, recess, valve, or partial septum, to prevent the 
poison from entering the bellows. The reservoir, p, has a supply en- 
trance, 7, which may be closed, or can have a pipe connection as an 
automatic supply from a higher reservoir, carried by the operator or a 
