SUMMARY AND CONCLUSION 459 



the preponderance of one or other parental group in any par- 

 ticular part of the plant depends on the presence of a larger 

 number of homodynamous determinants representing the part in 

 question, and on their possession of a greater controlling force. 



Reversion to grandparents and great-grandparents, or to 

 uncles and aunts, may be accounted for by the fact that, in the 

 first place, the idants and ids are not formed anew in the germ- 

 plasm of the parents, but are derived from the grandparents ; 

 and, secondly, that the combination of ids contained in the 

 individual germ-cells of the parent becomes very diversified in 

 consequence of the ' reducing division.' The usually accepted 

 assumption of breeders that one-fourth of the ' blood ' of the 

 grandchild is derived from each of the four grandparents, and 

 one-eighth from each of the eight great-grandparents, is there- 

 fore inaccurate. The number of ids of any particular ancestor 

 which are contained in the germ-plasm of a ripe germ-cell 

 depends entirely on the manner in which the reducing division 

 occurs ; and, under certain circumstances, a germ-cell might 

 presumably contain half the entire number of ids of one grand- 

 parent, and none of those of the other three. The larger the 

 number of ids derived from an ancestor, the greater is the prob- 

 ability that some of the characters of this ancestor will appear 

 in the descendant ; but this depends on the force of the ids of 

 the other parent, which comes into play when amphimixis takes 

 place, and also on whether the ids derived from this ancestor are 

 the dominant ones which determined his ' type ' (* Bild '). 



Reversion to an ancestor must consequently always occur 

 when, in consequence of the ' reducing division,' the ids deter- 

 mining the type of this ancestor reach a particular germ-cell 

 of the individual in question, — if they are not opposed by a 

 stronger group of ids derived from the other parent in the 

 process of amphimixis. This holds good for each individual 

 part of the offspring, as well as for the entire organism, for the 

 number of homodynamous determinants may be, and generally 

 is, difi"erent in the various parts, — at any rate in the case of the 

 individual differences between human beings. 



From this theory, it could be predicted that hybrid-plants 

 fertilised with their own pollen must produce very variable 

 offspring, and that individuals of these hybrids must, moreover, 

 revert to one or other of the ancestral species : both these state- 

 ments are borne out by fact. 



