42 BULLETIN NO. 3, DIVISION OF ENTOMOLOGY. 
The quantity of liquid and poison used also depends on the various con- 
ditions stated as determining the rapidity. The amount of liquid to the 
acre, aS near as could be estimated under the circumstances, ranged 
from 10 to 40 gallons, according to the size of the spray-discharge and 
of the cotton. The quantity of poison is in direct proportion thereto, 
being one-eighth to one-half a pound of London purple, or one-fourth 
to four-fourths of a pound of Paris green to the acre. 
The stirrer-pump device is a most perfect thing for the purpose and 
gave the greatest satisfaction. This contrivance is described and illus- 
trated in the annual report for 1881~’82 (pl. ix, pp. 159-161). The pump 
has heavy metallic valves, and its piston-head has no softepacking, so 
there is nothing about it that can get out of order. Yarn packing is 
used in the stuffing-box at its top. Being double acting it throws a 
strong and constant stream. Only one trunnion-eye need be moved, 
and a single iron wedge, instead of two, is sufficient for setting it. The 
wedge has on its head a catch whereby it is easily pryed out, and an eye 
by which it is chained fast to prevent losing it. In place of the wooden 
stirrer bar formerly employed I have made an iron one having a spring 
at the middle to clamp snugly in the eye at the bottom of the pump. 
This will not weaken with age or break. The cord or chain for pulling 
out the bar is not essential, as by having the end bung-hole of two 
inches diameter a pair of tongs or pinchers can be inserted to take hold 
of the end of the stirrer and extract it. The main hose or pipe has a 
screw union, by which it is easily separable from the pump. 
The descending pipes between the rows should have flexion and tor- 
sion in their joints or segments or hangings. The ground beneath 
the cotton-rows is highly ridged, and the mid-furrow between each 
pair of rows is deep. The ground thus formed operates upon the 
lower parts of each descending pipe or its appurtenances, when suita- 
bly shaped, so as automatically to guide the pipe and its nozzles 
between the rows, and to follow any crooks therein even when the 
conveyance is not driven in conformity with such irregularities. This 
automatical adjusting is allowed to a large extent even when the 
top of the descending pipe is firmly or non-adjustably attached, pro- 
vided that the descending pipe be flexile in some part of its course. 
This was shown in the earlier flexile forked machines which were 
tested for this Department near the Atlanta exposition. For the ob- 
Jject in question the descending pipe may be flexile throughout, but it 
is more commonly preferable to construct this pipe of stiff segments 
having one or two flexile joints, or very short hose segments; one at 
its top, and another at about two-thirds or one-half of the way down- 
ward therefrom. It is also generally best to make these flexile seg- 
ments or joints of three-ply or two-ply hose, and only of such length 
as to allow them to bend like knee-joints, and to suffer a semi-rotation 
or semi-torsion. This construction prevents the trailing-fork or other 
end part from getting turned upside down, or from remaining in wrong 
