EFFECTS OF AMPHIMIXIS ON ONTOGENY 289 



of the dominant group of the mother. It might be considered 

 improbable that both of these groups should come together in 

 one germ-cell, or that this can occur as often as must be the 

 case in reality, considering the frequency with which a combina- 

 tion of the two parental characters exists. 



In reply to this it may be stated that the meeting of two 

 germ-cells in the process of fertilisation, one of which contains 

 the dominant group of idants of the mother and the other that 

 of the father, must take place from time to time, for every pos- 

 sible combination will occur at one time or another. It must, 

 moreover, not be forgotten that it is extremely difficult to dis- 

 tinguish between the pure individual characters of a parent and 

 those of one of the immediate ancestors of this parent ; but the 

 production of these characters in the latter does not necessitate 

 the presence of the whole of the dominant group of parental 

 idants, for a portion of them will undoubtedly suffice if the 

 character in question is represented by a majority of homody- 

 namous determinants. The characters which alternate are very 

 often not the specifically paternal or maternal ones, but those 

 which in general characterise the mother^s or the father's 

 family. These, however, must be represented in most of the 

 ids of the dominant group of idants, and may therefore become 

 apparent even when the reducing division only causes a certain 

 number of idants, instead of the entire dominant parental group, 

 to pass into the germ-cell of the child. 



On the other hand, the frequency with which a child bears a 

 closer resemblance either to its father or mother — and this in 

 many families is actually the rule — can also be easily explained 

 by our theory. In such cases, not only can the w'hole number 

 of dominant idants produce the type of the parent in question 

 with approximate accuracy, but even a majority of them wdll be 

 sufficient to do so, provided that a large number of ids with ho- 

 modynamous determinants are contained in them. Many germ- 

 cells will, therefore, contain a sufficient number of the dominant 

 idants of the parent, while others, although perhaps containing 

 an equally large number of these idants, will consist ot various 

 combinations of ids enclosing relatively few homodynamous de- 

 terminants. If, then, an idant of the former kind, derived from 

 the father, should, in the process of amphimixis, meet with one 

 of the latter kind, derived from the mother, the type of the father 

 would predominate in the child, and vice versa. If two germ- 



