80 Inheritance of Melanism in Tephrosia (Ectropis) 



cases whole chromosomes are lost in the cytoplasm, others pass undivided 

 to one pole, others divide regularly or irregularly, and so on. Thus the 

 gametes finally developed are endowed with varying chromosome com- 

 plements. Clearly enough the gametes produced by the fusion of such 

 gametes will also possess abnormal chromosome combinations, so that 

 on the completion of development the insects bred ought, as just urged, 

 to be of novel guise. 



How does- that help to explain the present series of facts ? If in the 

 gametogenesis of the F^ bistortata-crepuscularia hybrids the chromo- 

 some carrying the gene for melanism finds its homologue, and this pair 

 of chromosomes follow the usual course of division with genes uncon- 

 taminated, or should it in default of finding a mate pass undivided to 

 one pole, then in either case ordinary Mendelian ratios should obtain 

 amongst the F^ insects. Such ratios did not manifest themselves, so we 

 are driven to consider further explanations. 



Giving due weight to the cytological facts, the following possibilities 

 lie open : 



(1) that the chromosome in which the factor for melanism lies is 

 amongst the number of those which, during the first maturation division, 

 split equationally, which in turn demands the probability that the chro- 

 mosome homologous to it should do likewise ; 



(2) that it divides into unequal portions ; 



(3) that the homologous chromosomes pair as in the gametogenesis 

 of a pure species, but that whilst they are united a levelling up or 

 averaging of the values of their component determiners ensues ; this 

 would necessitate, on separation, the production of hosts of gametes 

 with varying powers so far as the induction of melanism is concerned, 

 their variation in this respect depending on the degree of completeness 

 of the blend ; 



(4) that the chromosome carrying the gene for melanism is lost in 

 the cytoplasm during the later stages of division. 



Distinct as they would seem to the first glance, possibilities (1) and 

 (2) are not different in effect unless one is prepared to grant that a 

 mere mechanical division is able to fractionise the determiner for a given 

 Mendelian character. If it is not possible so to break it into smaller 

 fractions then, since it must proceed during cell division one of two ways 

 unaltered, no departure from Mendelian expectation should result. For 

 these reasons this possibility must be dismissed. In any case, even if it 

 were possible to split a gene, then the resulting broods should have 



