154 



• THE AMERICAN NATURALIST [Vol. LV 



develop are to be deposited. Certain Clialsis-flies and 

 Serpkoids have adopted this curious method of trans- 

 portation (Brues, '17) which occurs sporadically in di- 

 verse insects (Banks, '11). The way in which many 

 modifications of this kind appear in similar form makes 

 it impossible to consider them as guides to taxonomic 

 affinity. The elongated ovipositor, the active first-stage 

 larva, and many other adaptations for attaining the host 

 are of course good taxonomic characters, but they reap- 

 pear independently in more than one group, and can be 

 used only in combination with characters of less vital 

 importance to the animal, to characterize completely any 

 extensive groups. Nevertheless, the lengthened oviposi- 

 tor can be used to separate numerous families and 

 smaller groups in the parasitic Hymenoptera and as it 

 bears a certain relation to habits, the latter are thus re- 

 flected in taxonomy on a structural basis. However, the 

 habits of many such insects do not seem to require such 

 a long ovipositor and represent not the primitive habit 

 for the group, but recent modifications which break down 

 the homogeneous correlation of structure and habits. 



Closely connected with the specific association of nat- 

 ural groups of hosts and parasites is the great variation 

 shown by different parasites in the number and diversity 

 of the species that serve as their hosts. Just as we can 

 find among phytophagous insects, omnivorous forms, 

 strictly monophagous ones, and all intergrades between 

 the two, so there exists among parasites an almost equally 

 varied series of associations with one, several or many 

 hosts. 



Although pai-asitic Hymenoptera are so abundant, 

 both in species and individuals, their food habits are not 

 so easily observed as those of plant-eating insects and 

 our knowledge concerning them is far less complete. The 

 large number of secondary parasites also lead to con- 

 fusion, as these may not always be distinguished on a 

 structural basis. 



If parasitism demands a nice physiological adjust- 



