28 BULLETIN L204, U. s. DEPARTMENT OF AGMCTJLTtrHE. 
cotton with a dust cloud so shaped, it was accessary to take only 
narrow strips. This results in much overlapping of poisoning near 
tln> ends of the rows if the strips are spaced close enough to give a 
thorough application at the middle The waste of poison is enormous. 
Theoretically this overlapping could be compensated by varying 
the rate of cranking as the speed of the plane changes. This was 
attempted but proved impracticable, because it depends on the per- 
sonal equation and the judgment of the human. When moving at 
the rate of from 75 to 90 miles an hour, as is the case in this dust nig 
work, ground is being covered so rapidly that the slightest error in 
calculation on the part of the operator of the crank is tremendously 
magnified. Although attempted by several individuals it proved im- 
possible to compensate accurately for the change of speed. 
The need for a feeder which would perform this function auto- 
matically resulted in the construction of the air-suction hopper. In 
this device the dust is carried from the hopper by means of air suc- 
tion, created by the stream of air flowing down through the hopper. 
Since this air is collected by a funnel pointing forward over the 
plane wing, its velocity is determined by the plane's speed, and the 
amount of suction created is thus in turn proportional to the speed 
of the plane. 
To test this principle, a series of straightaway flights were made 
over a stretch of fields, with as much variation as possible in the 
speed of the plane. The same amount of dust (120 pounds) was 
loaded into the hopper for each flight and the flights were 1 stalled 
at the same point. The time requirement for emptying the hopper 
was varied from 45 seconds to 2 minutes and 22 seconds. In every 
instance the hopper was emptied in the same distance, namely 5.470 
feet. The ground speed therefore ranged from 31.2 to 90 miles an 
hour. The fact that the hopper discharge lasted exactly the same 
distance indicates that, in straightaway flights at least, the use of 
air suction for dust delivery provides an automatic compensation 
for variation in plane speed. These figures were of course made 
with the feeder valve wide open and the dust delivery at the heaviest 
rate provided in any of the experimental work. More erratic figures 
probably would be obtained on reduced deliveries, but this point was 
not tested. The records obtained indicate that to treat cotton at the 
rate of 6 pounds per acre it is necessary for the plane to cover a 
swath 160 feet wide, or if this figure is reduced to 2 pounds per acre 
it will be necessary to treat 480 feet at each trip. 
These figures are very satisfactory for straightaway flights, but 
whenshort flights were made, with the plane operating hack and forth 
over the field, somewhat different results were obtained. In fact, 
variation in the feed was noted to some extent in the straightaway 
flights. When first opened, with a full load of dust in the hopper, 
the feed was not very heavy, but as this dust quantity became 
reduced somewhat the feed increased until it quickly reached a 
maximum delivery, which was maintained throughout practically the 
remainder of the flight. In other words, the dust delivery was to a 
slight extent dependent on the amount in the hopper, even in 
straightaway flying. In flying back and forth over a field where 
the st rjps were rarely more than 000 to 700 feet long, this undesirable 
feature was more accentuated. Not only was the dust feed lighter 
