THE PHENOMENA OF REVERSION 315 



A third case may also occur in which the type of an uncle or 

 aunt is more or less accurately reproduced in the child, or in 

 which this type is blended with the characters of the parent on 

 the other side. I know a man who closely resembles a maternal 

 aunt, but nevertheless possesses many general characters of his 

 father's family. This fact may in all probability be explained in 

 terms of the idioplasm as follows : — the egg-cell from which 

 this person was developed contained the group of idants which 

 was dominant in the ontogeny of the mother's sister, and not 

 that which predominated in the case of the mother. Theo- 

 retically this might very well be the case. Let us suppose that 

 the primary germ-cells of the maternal grandfather (m p) 

 contained the idants a, b, c, d, e, f, g, h, and those of the grand- 

 mother the idants i, k, 1, m, n, o. p, q ; and also that the fertilised 

 egg-cell from which the mother was developed contained the 

 idants a, b, c, d x i, k, 1, m, and that from which the aunt was 

 developed the idants a, b^ c, i x /, n, o, p. We will further 

 assume that the group of idants which was dominant in the 

 ontogeny of the aunt were those indicated by the letters a, b, c, 

 and /, printed in italics. It will then be seen that the same 

 combination a, b, c, I can also be formed from the germ-plasm 

 of the mother by means of the reducing division, for all four of 

 them are present in this germ-plasm («, b, c, dxi, k, /, m). It 

 is doubtful whether such a case ever occurs so accurately, and 

 I know of no instance w^hich renders this hypothesis necessary : 

 the resemblance is always an imperfect one. 



We must also consider a fourth case, in which the child 

 neither specially resembles its father or mother, presents a 

 recognisable combination of the characters of both, nor bears a 

 striking resemblance to one of the four grandparents, but dis- 

 plays an entirely new combination of characters. Such a child 

 would probably always bear some resemblance to both, or at 

 any rate to one, of the parental families, but it would not exhibit 

 certain marked characters of the respective fore-fathers. 



These cases do not contradict our theory, for by means of the 

 reducing division it may possibly happen that none, or only 

 certain of those very idants which were dominant in both 

 parents are present in the germ-cells of the parents which 

 undergo amphimixis. 



If, for instance, the germ-plasm in the ontogeny of the father 

 had the composition a, b, c, d, e,f, g. h, of which the idants a, b, 



