1898] THE STUDY OF VARIATIONS 237 



sary to show something further tliaii that variations are definite, or 

 even that direct adaptive modification may occur which can subse- 

 quently become hereditary. Natural Selection must, as far as 

 possible, be totally excluded, reversion and influence of climate on 

 reproductive organs also more or less satisfactorily eliminated, 

 before anything like a test case can be adduced. This case must 

 still further demonstrate, if it is to disprove anything more than the 

 Weismannian position, the uselessness of Natural Selection by separate 

 control tests. 



The same arguments will apply to Cope and other Neo- 

 Lamarckians, although less forcibly, on account of Natural Selection 

 being considered a subsidiary factor. 



But the position is if anything even more difficult for the Neo- 

 Darwinians, as it would be necessary to produce definite hereditary 

 variations by Natural Selection without any change in environment, 

 and the elimination of such a wide series of factors would be 

 exceedingly difficult to accomplish. The extreme difficulty of this 

 position has been, to my thinking, so conclusively proved by 

 Romanes and Spencer, that the only means of obtaining definite 

 results will be by a complete revision of the methods used. 



My object has been to show that equally plausible contrary 

 positions may be taken up from the facts adduced by either side. 



If this is so, it would seem probable that some fundamental 

 mistake must be responsible for this curious position, and the cause, 

 I believe, exists in the extremely uncertain meaning given to such 

 terms as adaptive, definite, and indefinite variations. A more rigid 

 and exclusive use of these terms might, by rendering the difficulties 

 more apparent, help them to be more easily overcome. 



Some such classification as the following might be resorted to : — 



Unadaptive Variations. — Having no selective value on account 

 of all the variations being wholly unfitted for their 

 environment. 



Indefinite Variations. — Capable of selection, but varying in all 

 directions round a central position. 



Definite Variations. — All variations being of selective value, 

 but that value varying in degree. 



Adaptive Variations. — All equally modified for special condi- 

 tions, and therefore incapable of being selected except from 

 general strength or vitality. 



These divisions can only be approximately correct, but would 

 serve to limit the meaning of these terms and aid the formation of 

 clearer ideas on this subject. 



Lastly, some classification of possible origin of variations which 

 might be applied equally to both animal and vegetable life, by 



