156 FAMILY CYNIPIM. 



similar habits, and with which structurally they are in 

 the closest agreement. Thus Foerster's definition for 

 the gall-makers " The second abdominal segment 

 quite half the abdomen (excluding the first) or longer " 

 will not apply to some species of Aulax which have the 

 second segment not much longer than the third, and 

 certainly considerably shorter than half the length of 

 the whole abdomen. Aulax rhoeadis, in fact, agrees 

 with Reinhard's definition of the Figitides, second 

 abdominal segment half or shorter than half the 

 length of the abdomen. I cannot consequently look 

 upon the relative length of the abdominal segments as 

 a character of primary taxonomic value. 



The division into Gymnogastri and Cryptogastri is 

 not altogether satisfactory. Clearly the distinction 

 in the form of the radial cellule given by Ashmead 

 cannot hold good ; for in many Allotrinae the radial 

 cellule is quite as " long and narrow/' if not even more 

 so, than it is in species of Aulax or Synergus. And in 

 certain species of the latter genera the venter is quite 

 as much hidden as it is in Allotria. It is indeed diffi- 

 cult to find any absolute characters which can be used 

 for purposes of classification ; for, whatever structure 

 we may use, we find a gradual gradation from the 

 extreme to the mean when we trace it through the 

 various genera in the family. But although this is 

 the case we can, by employing various characters, 

 divide the group into four more or less clearly defined 

 sub-families. These four groups have a general, but 

 not absolute identity in habits. Thus, while the 

 majority of the Cynipina are attached to galls, either 

 as gall-makers or inquilines, yet one species at least is 

 parasitic, Diastrophus aphidivorus, Cam. 



A. Basal four abdominal segments subequal ; abdomen semi-sessile ; 



a stout process on the second joint of hind tarsi. Radial 

 cellule elongated, the three cubital cellules complete, the 

 second placed opposite the base of the radial cellule ; cubitus 

 complete and issuing from above the middle of the transverse 

 basal nervure. Ibalina. 



B. Basal segments of abdomen not subequal nor subsessile; no 



process on hind tarsi : 



