532 



APPENDIX 



FIG. 22. 



51. S<?#w<2/ ^^5 combined with natural selection. We now turn to 



the modifications introduced by mating. Let us suppose that the destruc- 

 tion of the unfit has produced a 

 rise from the ancestral type OP to 

 Op. Then the reasoning of 15 

 shows how mating sets certain limits 

 OQ, OR for deviation from the type 

 Op : but suggests no general tendency 

 to return towards OP. 



52. But there is a particular 

 tendency of this kind arising from 

 the sexual effect, in connexion with 

 the paring down of irregularities. If 

 OP be the type, then we have re- 

 marked that such irregular paths as 



OrKR tend to disappear because in effect there are equally likely paths, 



OqKQ, with which they may be 



mated. But under natural selection 



the paths are not equally likely. 



The path OrKR climbs over the 



V and indicates survival : the paths 



OqKQ does not surmount V, and 



means destruction. Thus paths such 



as OrKR get the advantage and 



survive. This is the first cause for 



the depression of path noticed in 



50- 



53. Ancestral inheritance and 

 natural selection. Let us now consider ancestral inheritance. As before, 



let OP be the ancestral type, DV 

 the destructive limit, and suppose 

 the type has risen, under the 

 influence of natural selection, to 

 OQ. Then by virtue of ancestral 

 inheritance there is a continual 

 tendency for every point of OQ 

 to return to OP; and this acts 

 in the contrary direction to 

 natural selection, and, at any 

 rate, checks indefinite departure 

 from the original type. 



54. But is it sufficient to arrest 

 it ? In whatever form we think of 

 ancestral influence, we cannot avoid contemplating it as becoming weaker 

 in each succeeding generation. Suppose for a moment that for a particular 

 generation the tendency of natural selection to raise the type were just 

 counterbalanced by the tendency of ancestral influence to depress it. 

 For the next generation the depressing influence would be weaker, and 



FIG. 



M 



23- 



FIG. 24. 



