4 



Indiana University Studies 



It cannot be hoped that I have now entirely untangled the con- 

 fusions regarding even the species treated in this paper, but 

 perhaps my method will ultimately prove profitable. 



By the use of trinomials the specific unity of a group is 

 recognized, and means are provided for the introduction of 

 comparisons to discover the character of the variations and 

 possibly some of the factors; to distinguish differences in 

 physiologic characters such as host preference, gall form, date 

 of emergence, and occurrence of heterogeny, in regard to all 

 of which closely related varieties may vary; and to give a 

 basis for the understanding of questions of distribution. The 

 only possible objection to the recognition of varieties will be 

 the less convenient nomenclature necessary, but this is not a 

 great consideration in view of the advantages of the practice. 



In recognizing the limits of varieties these are the general 

 rules I follow: (1) The morphology of the adult insect, rather 

 than the gall characters or other such data, is the prime con- 

 sideration. (2) Any character, no matter how trivial or in- 

 tangible, is of importance if it is constant in any large pro- 

 portion of individuals. (3) No variety is based on trivial 

 morphologic characters alone; physiologic characters as ex- 

 pressed in the structure of the gall, in the choice of host, 

 geographic distribution, or other general biologic characteris- 

 tics always contribute evidence for a similar interpretation. 

 (4) Conversely, no single sort of biologic data, nor set of such 

 data, is the basis for recognizing a variety unless it is paral- 

 leled by morphologic evidence. (5) No variety has been desig- 

 nated unless at least one of the varieties of the species is rep- 

 resented by material from more than one locality. If there is 

 any exception to the above rules, it is because the data of cer- 

 tain sorts are overwhelmingly good for the attitude taken. Sta- 

 tistical methods applied to lesser differences may disclose the 

 existence of still further order in variation. I have not em- 

 ployed this method simply because other, more important 

 problems with Cynipidse demand our present attention, and 

 are likely to do so for some time. 



The question of host relationships of cynipids is consid- 

 erably cleared up by this treatment of varieties. It appears 

 that some species show no constant variation (unless possibly 

 discernible by statistical methods) when occurring upon dif- 

 ferent oaks in the same region, while other species have de- 



