144 THE AMERICAN NATURALIST [Vol. LV 



the Braconidae. Structurally this relation seems plaus- 

 ible, but as the Alysiids attack almost exclusively the 

 highly specialized Diptera it is very difficult -to regard 

 them as closely related to the ancestors of the Ichneu- 

 monidae, so highly diversified in habits and structure. 

 The latter then are not so easily derived and may go 

 back to Evaniid-like forms. 



One extremely interesting fact in connection with the 

 primitive families of parasitic Hymenoptera is their as- 

 sociation with wood-boring insects. Thus the Oryssidse, 

 the most generalized group of Evaniidae, the more primi- 

 tive Braconidae, many of the structurally primitive Ich- 

 neumonidae, and the Capitoniidae are restricted to hosts 

 having such habits. This shows undoubtedly that such 

 habits have not easily been changed and that similarity 

 of host-habits is an important factor in determining what 

 insects may be attacked. This supports strongly our 

 thesis of the interrelation of taxonomy and habits. 



In connection with the parasitism of certain chalcis- 

 flies, the French entomologist Marchal ( '98) discovered, 

 some years ago, a most anomalous method of precocious 

 multiplication which he designated as polyembryony or 

 gei-minogony. In species exhibiting this phenomenon, 

 the embryo becomes dissociated into a large number of 

 parts, and from the numerous germs thus produced there 

 is formed a veritable swarm of minute parasites, the 

 extent of which is limited only by the available food 

 supply in the host. Marchal's first observations have 

 been much extended since by himself and numerous other 

 workers, and the same condition has been found to exist 

 in many other Chalcidoidea and also in the Serphoidea 

 (Marchal, '03). It has recently been recognized in an- 

 otlr r widely different family, the Dryinidae by Korn- 

 Jiauser ('19) and probably occurs sporadically in several 

 other parasitic families, although I believe no other cases 

 have been absolutely substantiated. From the regular 

 association of numerous individuals in single hosts in 

 the case of Micro g aster, allied genera of the Braconidae, 



