CLASSIFICATION OF VARIATIONS 93 



ment which Gates introduces to the effect that the difference 

 between such quantitative types cannot be represented in terms 

 of presence and absence. We are quite accustomed to the fact 

 that in the rabbit self-colour segregates from the Dutch-marked 

 type. These two types differ in a manner which we may reason- 

 ably regard as quantitative. It is no doubt possible that the 

 self-coloured type contains an ingredient which enables the colour 

 to spread over the whole body, but it is, I think, perhaps more 

 easy to regard the Dutch type as a form from which a part of 

 the colour is absent. It may be spoken of in terms I have used, 

 as a subtraction-stage in colour. Following a similar method we 

 may regard rubricalyx as an addition-stage in colour- variation. 

 The fact that crosses between rubrinervis, or rubricalyx and 

 Lamarckiana give a mixture of types in Fi, does not I think show, 

 as Gates declares, that there is any system here at work to 

 which a factorial or Mendelian analysis does not apply; but that 

 question may be more fitly discussed in connexion with the other 

 problems raised by the behaviour of Oenothera species in their 

 crosses. 



I do, however, feel that, interesting as this case must be 

 admitted to be, we cannot quite satisfactorily discuss it as an 

 illustration of the de novo origin of a dominant factor. The 

 difference between the novelty and the type is quantitative, and 

 it is not unreasonable to think of such a difference being brought 

 about by some "pathological accident" in a cell-division. 



Recognition of the distinction between dominant and reces- 

 sive characters has, it must be conceded, created a very serious 

 obstacle in the way of any rational and concrete theory of evolu- 

 tion. While variations of all kinds could be regarded as mani- 

 festations of some mysterious instability of organisms this diffi- 

 culty did not occur to the mind of evolutionists. To most of 

 those who have taken part in genetic analysis it has become a 

 permanent and continual obsession. With regard to the origin 

 of recessive variations, there is, as we have seen, no special 

 difficulty. They are negative and are due to absences, but as 

 soon as it is understood that dominants are caused by an addition 

 we are completely at a loss to account for their origin, for we 



