6 BULLETIN 796, IT. S. DEPARTMENT OE AGRICULTURE. 
veniently carried in an open pail and the dose measured in a gradu- 
ate holding just the desired amount. A little preliminary practice 
with the graduate and scales enables one to gauge the dosage very cor- 
rectly. It is probably best to keep the material from actual contact 
with the bark by perhaps half an inch, although the practical im- 
portance of this precaution is not known. Finally, the material is 
covered to a depth of 2 inches with two or three shovelfuls of soil, 
and this slight mound compacted by a few sharp blows with the 
back of the shovel. The soil is the container for the gas, and the 
success of the fumigation depends upon the vapor being given off 
faster than it is lost. At best there is considerable surface loss of 
gas, and the final compacting of the soil is of considerable impor- 
tance. No lumps or stones are left against the trunk above the 
surface to furnish a harbor behind which newly-hatched larvse 
may begin feeding out of reach of the vapor. No effect, of course, 
can be produced upon larvae feeding in galleries above ground. 
Larvicidal Action. 
To determine the actual larvicidal value of the vapor of p-dichlo- 
robenzene, a series of applications were made to trees in the 
field at Springfield, W. Va., in 1916. The approximate limits of 
effective dosage had been fairly well established by preliminary 
experiments. Each dose was applied to 20 trees and these examined 
in lots of 5 at approximately weekly intervals. 
Table I gives the summarized results of several such tests. It is 
impossible to include all of the experiments made in this connection, 
but the data given bring out the essential facts in regard to dosage 
efficiency. The soil on which these experiments were made is classi- 
fied as Frankstown silt loam. 1 
1 Determination of soil type furnished by the Soil Survey of the Bureau of Soils, United 
States Department of Agriculture. 
