Determinate Variation 163 



upon 'coincident' or correlative variations which are yet 

 not determinate but fortuitous in the strict sense. 



On the doctrine of natural selection, the only way to get 

 determinate evolution is to secure the survival of a surplus 

 or balance of variations of a particular kind in each single 

 generation considered for itself. So the opponents of 

 determinate evolution have brought the challenge to show 

 that, in each particular case, such a predominance of vari- 

 ations in a particular direction is found. Weismann 

 recognizes the force of this challenge, but does not see 

 how it can be met (especially in the form urged by the 

 paleontologists), with all his machinery, including intra- 

 selection, and so he produces the theory of ' germinal selec- 

 tion ' to account, as he puts it, for 'variations where and 

 when they are wanted.' But the question is one of fact : 

 do we actually find a balance of variations in a particular 

 direction, antecedent to the process of elimination by natural 

 selection ? Recent statistical work points directly in the 

 opposite direction, as is said above. 



Now, the point is that the view suggested under the 

 term Orthoplasy, with organic selection, does not require 

 determinate variations, although it results in determinate 

 evolution. On this view the determination is secured, not 

 by an original balance of variations in one direction, but by 

 a shifting of the mean of variation in a certain direction 

 through the selective results of the creature's accommoda- 

 tions. These not only make their own repetition secure 

 by repeated intra-selection in each generation, as Weismann 

 showed, but they shield and keep alive the set of variations 

 which they in any way involve, so that in the next genera- 

 tion the gamut or range of variations, while subject to the 

 same lazv of indeterminate distribution (called ' chance dis- 



