THE PHENOMENA OF REVERSION 31I 



aggregates of the characters of the grandparents undoubtedly 

 reappear in the child, and these may be more easily explained 

 theoretically, without the necessity of supposing that chance 

 plays so great a part. For in order that perfect reversion may 

 be possible, numerous most infrequent occurrences must take 

 place together. To use the same example, in each individual of 

 three consecutive generations, only one of the two groups of 

 parental idants could determine the type of the child, and the 

 reducing division, moreover, must affect one of the mother- 

 germ-cells of each of four individuals — i.e., the two grand- 

 parents and the two parents — in such a manner as to separate 

 the two groups of parental ids. All these possibilities may 

 sometimes concur, but we can only imagine this to be so if 

 it is proved that an exact resemblance exists between the child 

 and grandparent. 



A partial resemblance is theoretically far more likely to occur. 

 Let us suppose, for instance, that the germ-plasm consists of 

 sixteen idants, eight of which are derived from the mother and 

 eight from the father, and that the reducing division causes six 

 idants of one parent and two of the other to be present in one 

 germ-cell : the former might possibly contain many of those 

 determinants to which the type of this particular parent was due. 

 This need not necessarily be the case, for all the sixteen idants 

 may have had a farely equal share in the production of this type, 

 and six out of the sixteen idants could not then possibly give 

 rise to the same type even approximately. 



It might, on the other hand, happen that the type of the 

 parent was essentially determined by the group of idants derived 

 from the father or mother of this parent only, and a resemblance 

 of the child to its father could then only occur when the whole 

 or a great part of that group which controlled the development 

 of the father was also present in the germ-cell from which the 

 child was developed. Even then, the production of the type of 

 the father would not absolutely be ensured, for in the process of 

 fertilisation by the other parent a group of idants might be 

 added, in which the controlling force preponderates over those 

 already present, either entirely, or at any rate as regards many 

 or most of the determinants. 



We may, however, conclude from all these considerations, 

 which are unfortunately still very vague, that a large number of 

 idants of the grandfather, for example, may be present in the 



