84 THE ROYAL SOCIETY OF CANADA 



and its bearing on the very important question of the constancy of the 

 MendeHan characters. One school of geneticists explains blending 

 inheritance as the result of the action of multiple determiners, the 

 intermediate individuals representing combinations of determiners 

 greater in number than that of the lower parent but smaller than that 

 of the higher. The other school holds that a real blend of a single 

 pair of determiners occurs. The latter consequently believe that the 

 Mendelian determiner is not unchangeable, while the former believe 

 that the occurrence of blending does not affect the question of unit 

 character constancy. A great body of evidence has been brought 

 forward on both sides. 



In the present case according to the multiple determiner hypothe- 

 sis a difference of sixteen days for example is due to several genes, 

 each being responsible for a fraction of the total difference. In the 

 second hybrid generation combinations of different numbers of these 

 genes will segregate out. If there are four genes concerned each 

 responsible for a four days' difference then segregates will be formed 

 ripening (1) with the earliest, (2) four days later, (3) eight days later, 

 (4) twelve days later, (5) with the latest. If the homozygous condi- 

 tion for a given gene has double the effect of the heterozygous condi- 

 tion the classes will be still further increased. And most of the segre- 

 gates will be in the middle grades. Environmental variations will 

 smooth out the curve. 



The chief argument in favor of the hypothesis that multiple 

 determiners are involved is the increased variability in the second 

 hybrid generation, this being taken to indicate Mendelian segregation. 

 In all the crosses reported in this paper this great variability is very 

 evident. The F2 generation is much more variable than either 

 parent and in fact often varies from the lowest extreme of the lower 

 parent to the highest extreme of the higher parent. Unfortunately 

 comparison with the Fi generation is impossible because of the small 

 numbers of the latter. Moreover when the parents are very unlike 

 the F2 variation does not reach the parental extremes. Under these 

 conditions the parents supposedly differ by a larger number of 

 hereditary factors than when they are more alike, and consequently 

 the extreme combinations in F2 will occur much more rarely. 



The multiple determiner hypothesis may thus explain satis- 

 factorily the results of each individual cross considered alone. But 

 we have yet to determine whether it will explain all the results of 

 the various crosses considered as a whole. We must apply the 

 hypothesis to each successive pair of parents and determine their 

 hereditary differences; then we must sum up these differences and 



