FAMILY CYNIPIM1. 155 



that joint 7 is often excised below, and shows under- 

 neath it a small portion of the tip of the last ventral. 



" Habits, as far as known, insectivorous." 



Practically the same classification is adopted by 

 W. H. Ashmead (Trans. Am. Ent. Soc., xiii, p. 59). 

 He divides the family into two sections : 



i. Gynmogastri. Venter visible, or visible for 

 more than half its length ; radial area long and narrow. 



ii. Cryptogastri. Venter not visible, or with the 

 tip alone occasionally exposed; radial area an equi- 

 lateral triangle. 



The G-ymnogastri contain three groups : 



Ibalini&. Radial area closed ; abdomen cultriform ; 

 segments about equal ; venter visible. 



OynipindB. Radial area seldom dosed ; second abdo- 

 minal segment occupying half or but little more than 

 half the whole surface of abdomen ; venter visible. 



InquilincB. Radial area seldom open; second abdo- 

 minal segment occupying the whole or nearly the 

 whole surface ; venter not visible its whole length. 



The Cryptogastri are divided into two " sub- 

 families :" 



i. AllotrindQ. Abdomen short, globose or semi- 

 globose ; second abdominal segment longer than the 

 other ; scutellum round, convex. 



ii. Figitinx. Abdomen elongate, ovate, compressed, 

 with the apex more or less pointed ; third abdominal 

 segment the largest ; scutellum quadrate, cupuliform, 

 or spined. 



From the preceding sketch of the systems of classi- 

 fication proposed by the chief writers on the subject, it 

 will be observed that great stress is laid on the rela- 

 tive lengths of the abdominal segments for the primary 

 divisions. As used by Hartig it can only be regarded 

 as an artificial method of separating the species, for 

 it separates the Eucoelinse from the Figitides, with 

 which they are much more nearly related than they are 

 to the gall-making species. Even as used by Foerster 

 it will separate true gall-makers from others of precisely 



