IO2 W. J. CROZIER AND L. H. SNYDER. 



gammarid had difficulty in carrying a large female, and could not 

 clasp her successfully unless his appendages were able to reach 

 around the body of the female. It is not so clear, at first sight, 

 why small females are not found carried by large males. But 

 closer inspection shows that the dactyls of the male pereiopods are 

 neatly inserted under the edges of the coxal plates of the captive 

 female. We incline, therefore, to the view that the graded corre- 

 lation between the sizes of members of pairs is determined by 

 mechanical features of the clasping process. The magnitude of 

 the correlation indices found in these cases is in general agreement 

 with those of the indices reported for Paramecium and for Chro- 

 modoris, in which also there is involved a purely structural adjust- 

 ment. 



The possible result of the selective coupling, in relation to num- 

 ber and size of offspring, remains to be studied. It is known that 

 the number of eggs carried by a female gammarid varies directly 

 with her size, hence it is not impossible that a phenomenon like that 

 suggested in the case of Chronwdoris (Crozier, 1918) 'may be in- 

 volved here also. If the economical utilization of gametes be the 

 chief consequence of selective coupling by sizes, we deal with an 

 adaptive mechanism not necessarily involving factorial inheritance. 

 The possibility of the latter complication can be tested experi- 

 mentally. It is not out of place to call attention to another phe- 

 nomenon in which selective pairing may well play an evolutionary 

 role. The imaginal size of certain parasitic insects is known to be 

 determined by the size and the rate of development of the particu- 

 lar species of host insect in which their larvse grow and pupate 

 (Keilin, 1915; Haviland, 1922). This is due to a developmental 

 correlation between the pupation of the larval host and of its con- 

 tained larval parasites. If adult insects, of the same species, dif- 

 fering in size through this means, practice selective mating and are 

 by some agency compelled to oviposit in a host of the type from 

 which they were themselves reared (cf. Wheeler, 1922), a basis 

 is clearly afforded for the foundation of divergent types. This 

 possibility may be examined experimentally ; it is profitable, in the 

 meanwhile, to study the conditions of assortative mating in a 

 variety of types. 



