Eeactions and Peoducts in Intekspecific Ckosses 113 



the formation of the determiners for striping and punetation into a closely 

 bound association, which thenceforward acts homogeneously to produce no trace 

 of stripes in subsequent generations, owing to the lack of interaction of V with 

 Elp. What it is that changes this arrangement in the system I do not know, 

 except that it occurs only when Ac has the relative values of 60 and 40, and 

 imder conditions of the medium as given. It would at present be useless to 

 speculate upon what the actual change is in the material constitution of the 

 gamete. The entire complex series of crosses between these two species thus 

 becomes intelligible and in accord with the principles of factorial constitution 

 and operation, capable of elucidation without making assumptions of agents 

 either internal or external, or relations, associations, or results that can not be 

 put to test in one way or another. The entire series shown in plate 9 and all 

 of its consequences works out as is shown in the following analysis of the opera- 

 tions as outlined above. 



In plate 10 is shown the parental conditions and the events in the F^ and Fj 

 of an array like that shown in plate 9. The complex under which this cross has 

 in general been made is that of the breeding quarters, as follows : 



Table 8a. 



Wind movement continuous and causing a complete change in the medium 

 at least once in 30 seconds, giving an evaporation-rate higher than that in the 

 habitat of diversa in the growing-season, lower than that of signaticollis. 



The two parent stocks with the Ac values of diversa {Ac^°) and signaticollis 

 (Ac*°), when crossed under the conditions given, produce uniformly the stated 

 results in about equal ratio of 1:1. This ratio, however, is only a product of 

 the special complex, both within and without. 



The Fi types both produce the same array of gametes, except that in the 

 mashed heterozygote, the V determiner in the diversa gametes has in this series 

 formed in the interactions in F^ the new association shown in the graphic rep- 

 resentation in plate 10. From this point onwards, Pu-V exists as a group, 

 acting as a single agent in the gametic operations. Each of the F^ heterozygotes 

 forms two classes of gametes, giving in Fj the typical array of three classes in 

 the ratio of 1 : 2 : 1, in one set readily visible, in the other only detectable by 

 biometric measurement and the test reaction. In both the extracted types 

 breed true. In plate 11 is shown the continuation of these true-breeding lines, 

 from F„ onward through F^, Fj, and Fg, the mid-type or heterozygous mode not 

 being continued, as it duplicates repeatedly the Fj behavior. In Fj is shown, 

 in graphic form, the application of the test reaction to the two lines. When 

 crossed with the line derived from the F, higher or signaticollis mode, nothing 

 happens, and the reason is here shown in the gametic compositions that enter 

 into this combination, no elements but pure signaticollis entering; consequently 

 none other is brought out. In the other line, the cross of an F^ extractive from 



