TOXIC GASES AS A CONTROL OF THE PEACH-TREE BORER. 11 
season have the added disadvantage of subjecting the tree to a double 
fumigation and a proportionately greater chance of injury, although 
no injury was observed on the trees so treated. 
While very good results have been obtained from applications 
made all the way from the last of August to the last of September, 
very late application has been less satisfactory. In 1916 the 1-ounce 
dose was applied to 48 trees at Springfield on November 1. Exam- 
ination was made the following spring on May 28, about seven months 
later. Large numbers of dead and decomposing larvae were found 
on the trees. An unusual number, however, had escaped. Compared 
to adjacent untreated trees the indicated mortality was 87 per cent. 
A slight amount of the material was still in the soil, and from the 
appearance of the larvae it was suspected that a considerable part of 
the fumigation had taken place in the spring. There was also more 
or less injury about wounds and exposed tissue. The larvae are 
probably much more resistant to fumigation during the dormant 
period than at any other season of the year. 
It has been found that with all gases the trees recover more readily 
from injury which occurs in early spring just at the beginning of 
the growing season than at any other time of the year, and other 
things being equal that would be the logical time to apply fumiga- 
tion. Eecords of soil temperatures made at Springfield in 1916, how- 
ever, indicate that the soil warms up rather slowly in the spring. At 
a depth of 6 inches the temperatures of the period from September 
15 to October 15 were not attained in the spring until about May 15 
to June 15. It is hardly probable that effective application could 
be made in the spring much before May 15. By this time a great 
number of the larvae have reached a considerable degree of maturity 
at the expense of much injury to the tree, and while the gas would 
probably reach a large proportion, the depth and character of the 
burrows of the larger larvae provide more or less chance for their 
escape. 
As already stated, early fall, from two to three weeks before the 
end of the hatching period, proved the most effective time for appli- 
cation. In the latitude of Washington this is about the 10th of Sep- 
tember. At this time larvae that have already entered the trees are 
mostly small, feeding in the outer layers of bark in more or less ex- 
posed locations, while the soil temperature is sufficient to evaporate 
the material before winter. 
For other localities both north and south of the latitude of Wash- 
ington, the time of application undoubtedly can be established ap- 
proximately from what is known of the insects' seasonal history in 
various parts of the country. In Table VI it will be seen that there 
was a wide variation in the dates applications were made in the same 
locality in 1916, 1917, and 1918. As a matter of fact, all of the appli- 
