148 



THE AMERICAN NATURALIST [Vol. LV 



the already greatly strained principle of the fixity of in- 

 stinct in the imaginal insect. 



It also aids greatly in understanding the relation pre- 

 viously referred to, where extensive groups of parasites 

 attack discrete groups of host. Adaptation to one host 

 means ordinarily greater physiological suitability for 

 another closely related host than for a widely different 

 one. This, no doubt, applies to cases like the Alysiid 

 parasites, for here the series of hosts, while quite uni- 

 form, is so extensive that it can not be explained on the 

 slowly acting basis of concomitant differentiation of the 

 hosts and parasites. 



Instances, like one cited by Pierce ('08) where several 

 species of parasites suddenly became abundant enemies 

 of the boll-weevil due to the scarcity of their more 

 favored hosts, must depend upon selection, as suggested 

 above, leading to the rapid improvement of partial adap- 

 tations. 



We have already referred to the fact that the parasitic 

 H^Tnenoptera, and quite generally also most parasitic 

 insects, attack other insects, and pointed to this as a 

 characteristic more or less peculiar to insect parasitism 

 or at least to its most prevalent types. 



The attachment to closely related animals as hosts is 

 sho^vTi still more clearly in Hymenoptera that are sec- 

 ondary parasites on parasitic species of the same order, 

 of the same family, or even of related genera. This phe- 

 nomenon is not restricted to Hymenoptera, but is most 

 extensively exhibited by them. Thus certain genera of 

 Ichneumonida?, Braconida?, and Chalcidoidea develop 

 regularly in the larvae of primary parasites which become 

 established in a free living host. 



Secondary parasites are not absolutely distinct from 

 primary ones in some individual cases, for this relation 

 is known to bo facultative in a few species of Hymen- 

 optera which drv.'lo]) in either way. In 1903, Fiske ('03) 

 showed fi-oni careful l)reeding experiments that certain 

 Ichiu'unioiiida} of the genus Itoplectis may be either pri- 



