KATURAL SELECTION COMPLETES GERMINAL SELECTION 135 



understood it, but, as he said, to the " laws of growth, and of the 

 reciprocal influence of the different parts," or, as Wilhelm Roux 

 termed it, to " histonal selection." ^ It is histonal selection which 

 determines the mutual position of the parts, each part being 

 determined, in its constitution, by the degree of specific attrac- 

 tion exercised by its component cells. And we must also re- 

 member that a correlation exists between germinal and natural 

 selection. A variation which is originated by germinal selection, 

 and, having attained biological value, is favoured by natural 

 selection, may produce a correlated variation in another group 

 of determinants ; the determinate of the latter may in itself 

 possess no appreciable biological value, and yet it will be main- 

 tained as long as the primary variation is maintained. For 

 correlation exists, not only in the fully formed organism, but 

 at every stage of the life-history, from the germ-cell until death. 

 Thus, secondary characters cannot determine the constitution 

 of a species. They are due, it is true, to purely intragerminal 

 action, but not as Nageh understood it, nor as De Vries seems 

 to understand it ; for they, too, are a result of selection — 

 germinal selection — ^both in their origin and in their disposition 

 in the fully formed organism. AU biologically important 

 characters, however, are the result of natural selection com- 

 fleting the work of germinal selection, and acting on those 

 variations which the latter presents to it. The difEerence 

 between Darwin and Weismann lies in this, that for Weismann 

 the element of pure chance is considerably reduced ; each time 

 an adaptation to new conditions becomes necessary in the 

 interests of the species, that variation wiU be presented by 

 germinal selection, and natural selection has simply to act on 

 it. When we consider that there are only two possible modes 

 of variation in the germ — ^progressive and regressive — it is 

 evident that the chances are 100 to 1 that one or other of these 

 will set in ; and as the number of individuals in a species is 

 1 Vide pp. 35, 36. 



