159 
The sexes are usually recognized, as in all other Hemiptera, by the 
structure of the venter beyond the sixth segment. In the male there is only 
a single plate or capsule, open posteriorly, rarely transversely divided ; 
in the female, the basal part is divided medially into two plates, while the 
apical part shows several small plates. The median division in the female 
is the most conspicuous difference. Secondary sexual characters are 
rarely seen. In Amnestus the distinction is very difficult, but in this 
genus there are interesting secondary sexual characters consisting of 
peculiar spines on the fore and hind femora, and differences in thoracic 
sculpture, which are indicative of the sexes. 
[The general appearance of sonie of the commoner species is well 
represented in Plate XXI, Figures 78 to 83 inclusive. ] 
CLASSIFICATION 
Kirkaldy believed that the pentatomids were the most primitive 
Heteroptera, and I am inclined to agree with him. Probably the Ochter- 
idae represent a generalized stem, as Reuter suggests, but I believe that 
the five-segmented antennae of the pentatomids is older than the four- 
segmented of other families, and is not a case of specialization, even 
though the antennae of pentatomid nymphs is uniformly four-segmented. 
The venation of the Coptosomidae is by far the most generalized in 
the Pentatomoidea studied by me; yet, singularly enough, the scutellum 
would seem a climax of specialization. The Scutellerinae come next in 
wing venation, leading to the Pentatominae, along with a reduction in the 
size of the scutellum. In the Pentatomidae, the development of ventral 
spine and sternal ridge would seem to begin in the subfamily Penta- 
tominae in such forms as Banasa, Nezara, and Arvelius, with the Acantho- 
sominae and Asopinae as offshoots, and lead up naturally through the 
Edessinae to the Tessaratominae. Yet the venational evidence shows that 
this order must be reversed. 
The most primitive type of heteropterous caecal structure I should 
consider to be the one with the largest number of rows and pockets, 
specialization proceeding by numerical reduction and differentiation, fre- 
quently reaching complete elimination, as in Asopinae. The data are as 
yet very incomplete, but it is evident that the four-rowed type, found in 
the scutellerids and pentatomids, should precede the two-rowed type, rep- 
resented by the Cydnidae, sens. lat., and succeeding families. In the four- 
rowed series Brochymena, with its large number of caecal pockets and 
its well-developed wing hamus, is clearly primitive. 
THE IMAGINES 
The Caecal Appendages 
Much light has been thrown on pentatomid relationships by the 
interesting studies of Dr. H. Glasgow * on the caecal pockets of the 
Heteroptera. His paper appeared in a general biological publication. It 
was primarily a study of the symbiotic bacteria always inhabiting these 
* Biol. Bul., Vol. 36, p. 101, pl. I-VI1T. 
