314 THE GERM-PL.4SM 



occur ; and only those instances are possible in which the rever- 

 sion concerns more or less extensive groups of characters which 

 already existed in an individual of a former generation. 



We have now discussed the question of transmission from the 

 parents, grandparents, and great-grandparents, to the child ; and 

 it will be as well, before going further, to summarise the various 

 possible cases, and to see whether observation and theory con- 

 tradict one another in any respect. 



The most frequent case seems to be that in which the child is 

 a mixture of the types of the two parents, the characters of both 

 of whicli are either completely united, or else they alternate with 

 one another in the different cells, parts, and organs, and even in 

 the systems of organs themselves. In all such instances, the 

 group of idants which predominated in the ontogeny of the 

 parent, or rather of both parents, must have been contained in 

 the germ-cell from which the child arose. 



The question might here be asked as to how it can so fre- 

 quently happen that only the dominant group of idants is 

 present in the germ-cell destined to undergo amphimixis. We 

 must, however, remember that a perfectly equal blending of the 

 types of the two parents in the child occurs much more rarely 

 than is generally supposed ; and that it is difficult, and frequently 

 in fact impossible, to say whether the maternal portion of a 

 character is really derived from the type of the mother, and not 

 from that of her father or mother. General characters only, of 

 the mother's family are most usually combined with others 

 derived from the family of the father. But those characters 

 which have been peculiar to many members of a family for 

 several generations are the most likely to occur in numerous ids 

 and idants, and they therefore reach the germ-cells in larger 

 numbers in many of the modes of ' reducing division.' Theo- 

 retically the dominant group of idants of the parent would not 

 be necessary for such a transmission of a general family resem- 

 blance, but only a majority of the idants of this group. 



The case, however, is different when the resemblance refers 

 mainly to the type of one parent ; and w^e must then assume that 

 the dominant group of idants of this parent is present, and is 

 opposed to another weaker group derived from the other parent, 

 and provided with a smaller number of homodynamous deter- 

 minants. 



