5 2o APPENDIX 



general, follow either a path Oc, which is a return towards the 

 ancestral path OA, and may even go on the other side of it, or 

 a path OC, which is a further deviation from the ancestral. We are 

 not yet in a position to say anything of the relative chances of these 

 two possibilities, but experience of a wide kind tells us that both are 

 possibilities ; and similarly it is possible for the children OD, Od of Ob 

 either to return towards the ancestral type (as OD) or to diverge further 

 from it as Od. Hence there is a continual tendency to produce descen- 

 dants differing more and more from the original type in both directions, 

 and if this tendency were unchecked we should get animals of all sorts 

 of miscellaneous shapes. There are, however, three important checks to 

 this tendency which may not be entirely independent, but which we will 

 consider separately in the first instance; they are (a] the sexual, (b) 

 the ancestral, (c) the retrogressive. The first two we can deal with at 

 once : the third is more conveniently deferred until we come to natural 

 selection ( 55). 



10. Influence of Two Sexes. Most living forms are the children not 

 of one individual but of two. Let the path of one parent be OQ, and 



of the other OS. What will be the path 

 of the child? We know from wide ex- 

 perience that as a rule children resemble 

 both parents (IV.), and thus the child's 

 path will, as a rule, be intermediate between 

 the paths OQ and OS ; not necessarily 

 half-way between them, but somewhere 

 intermediate. The child's path may be 

 such as OC nearer to OQ, or OD nearer 

 r^ to OS. There might, for instance, be a 

 tendency for the child to resemble the 

 father rather than the mother ; and if OQ 



be the path of the father, the child's path would be like OC. But if 

 for every pair of 'parents ; OQ OS, in which OQ represents the father there 

 exists another pair in which OQ represents the mother (V.), then for every 

 child OC above the half-way path we shall have another child OD equally 

 below the half-way path, and on the average the children will follow the 

 half-way path. 



11. The condition italicized is roughly in accordance with experience. 

 There are tall men with short wives and short men with tall wives. 

 Whether there are more of the former than the latter we cannot say 

 without statistical inquiry : but it is easily seen that the condition 

 must be roughly fulfilled. For take any ten married couples : divide 

 the men into tall and short five of each : the women also into tall and 

 short five of each. Suppose three tall men marry tall women : then two 

 tall men marry short women. There are thus left five short men married 

 to three short and two tall women. 



12. The argument is quite general. If we take n tall men and n 

 short men : n tall women and n short women : and if x of the tall men 

 have tall wives : then n - x have short wives. And since x tall women are 





