136 The Mecha^shsm of EvoLUTioisr in Leptinotaksa 



there results, in conjunction with conditions of the medium, changes in the 

 association of agents or groups of agents in the gametes, such that in a portion 

 of the population the F^ heterozygous condition in appearance resemhles one of 

 the parent species and continues to breed true until the gametic system is again 

 changed and the agents freed from their position of nonvisible manifestation. 

 In both there is the same separation of the fraternities in Fj of the suspected 

 race into trimodal polygons, showing that the species form-base as measured by 

 the form-index is not affected by this firm association of the several agents in 

 the heterozygous individuals. It is true that the series in both species is quite 

 unusual — call it a non-Mendelian reaction if need be — but there is at no point 

 that I have been able to demonstrate with certainty anything either in action or 

 in result that is in any manner at variance with the principles of factorial com- 

 position and reaction, and further in the series the reactions, as far as they have 

 been analyzed, are most clearly in entire accord with the principles of the 

 Mendelian reaction. When it is possible to carry this further I have no doubt 

 that it will be capable of as complete analysis as was the simple case in the cross- 

 ing of signaticollis and diversa. 



A feature of interest that has come out in these crosses is the slow breeding 

 of the undecimlineata race that appears in F^, giving about one generation in 

 10 or 11 months. As far as I know this slow breeding persists under ordinary 

 conditions indefinitely and remains a permanent character in this race. Some 

 indications of its origin have been obtained that are of interest. In Chapter II 

 it is shown that in signaticollis the normal cycle is two generations in rapid 

 succession, then a period of rest for about 9 months, passed either in the ground 

 in aestivation or in part on the plants in practical inactivity. Under the con- 

 ditions of the laboratory the same rhythm is maintained, but the length of 

 ontogeny spread out and the period of quiet between successive cycles is 

 shortened. In experiment, when the rate of reproduction is shortened by con- 

 ditions in the medium, the stock always shows longer periods of rest and more 

 uniformly enters into hibernation for this period than when the Ac values are 

 larger, at about 60. On the other hand, undecimlineata, while it has the same 

 rhythm of reproduction, is much less sharply delimited and the period of repose 

 between successive cycles is shorter and to a large degree a product of the con- 

 ditions of the medium. 



I may later show something of the nature of this rhythm and its relations to 

 agents within and without the organism and its general methods of inheritance. 

 In general in these forms in pure reacting lines, in which the rhythm is homo- 

 zygous, the sequence is followed with precision, of an overwintering or resting 

 generation that emerges from the period of inactivity, especially on the advent 

 of favorable conditions in the medium, reproduces at once, giving a summer 

 generation that reproduces without rest, giving a second generation in the cycle, 

 which, however, hibernates or rests in one way or another for a varying period 

 of time. There are thus in the cycle two conditions in reproduction that follow 

 in rhythmic manifestation — winter generation, which after resting produces 

 the summer generation, which in turn produces a winter generation without 

 resting, which winter generation rests from reproductive activities in signati- 

 collis as much as 9 months in nature and in the laboratory strains with Ac values 

 of 40 or thereabouts. 



