866 . THE AMERICAN NATURALIST. | (Vor. XXXIX. 
specific determination. Walsh has called the maker of each of 
the fifteen willow galls he describes a distinct species. If the 
galls themselves could be taken as a means of determining spe- 
cies his results would in most cases stand. Although I have 
been working with only a few of the cone galls, it is necessary 
to consider some others in settling this question. 
Of Walsh's fifteen willow galls, six are more or less cone- 
shaped bud galls of somewhat similar structure. The larvae 
found and described by Walsh in these six are precisely alike as 
to size, shape, * breast bone," etc. Walsh gives, however, the 
following differences in the cocoons : — 
Cecidomyia salicis brassicoides, cocoon scarcely larger than 
larva. 
Cecidomyia salicis strobiliscus, cocoon 1} to 2 times as long 
as larva. 
Cecidomyia salicis gnaphaloides, cocoon 1} to 2 times as long 
as larva. 
Cecidomyia salicis coryloides, cocoon 2 times as long as larva. 
Cecidomyia salicis strobiloides, cocoon 2} to 3 times as long 
as larva. 
Cecidomyia salicis rhodoides, cocoon 21 to 3 times as long as 
larva. ; ; 
The difference seems to be considerable but it is easily 
explained. I have examined all except the fourth (coryloides), 
ten or more galls of each, and have found the length of the 
cocoons to vary in the different galls in proportion to the length 
of the central cell or cavity (see Fig. 5). The gall S. drassicoides 
has a short cell which permits of only a short cocoon, while the 
gall S. strobtloides has a long central cell giving plenty of space 
for a long cocoon. If the cocoon is exuded from the larva and 
is expanded by a gas, as Walsh supposed, then the length and 
size of the central cell certainly would determine the size of the 
cocoon. 
In these same six species Walsh found the pupæ of only five. 
The only difference I can find in his descriptions is that C. s. 
gnaphaloides (in dried specimens) is slightly shorter than the 
others. I could not find this difference. 
From a careful comparison of ' Walsh's descriptions and from 
