278 Reversion and [ch. 



segregation. The spine may be taken as a thing added 

 and "present," or in our terms, a dominant character. In 

 each division there is found a dominant half which has the 

 spine, and a recessive half without it. If only we could 

 know what would happen as the result of conjugation 

 between two Paramoecia possessing similar "spines," it is 

 possible that the manner in which segregation occurs in the 

 gametogenesis of the Metazoa would be elucidated. In 

 them, as mentioned above (p. 195), the gametes are almost 

 always formed in sets oi four, and a presumption is thus 

 created that the members of each set of four are not all 

 equivalent to each other. If they were equivalent we could 

 represent the set as AAAA, supposing the zygote which 

 produced them were homozygous. If the organism were 

 heterozygous we should then imagine the series to be 

 AaAa. But if the differentiation is not by pairs but by 

 fours, the series must be represented by AaA'a' (or 

 AA'aa!) for the heterozygote, and as AA'AA' for the 

 homozygote. 



(^ Reversion and Variation, • 



It follows from what has been said of allelomorphism 

 thaT^ variation must now be regarded in the mam as a 

 phenomenon due to the addition or omission of one or more 

 definite elements. When, as in the case of the combs of 

 Ifbwls, the types which have obviously arisen later in the 

 evolution of the species, for example, rose, pea, and walnut, 

 dominate over the primitive type, in this case the single, 

 the variation by which those dominant varieties came into 

 existence must have consisted in the addition of an element 

 (or in the case of the walnut-comb, two elements) to the 

 original stock of the species. JWhen, on the contrary, the 

 new variety is recessive, it is clear that the variety occurs iiTZ 

 consequence of the omission of an element. (The suggestion 

 made by de Vries, that the dominant is always the phylo- 

 genetically older form, has not been substantiated by further 

 investigation.) The problem of the_ causation of variation 

 is thus to some extent narrowed down. The "cause" of a 

 Variation is the event which brings abou t tEe a dditibrT^ f 

 olTTission of a factor. 



