ALTERNATIVE SEXUAL INHERITANCE 



523 



FIG. 7. 



1 9. Limits of deviation. The limits thus assigned will be of course 

 on both sides of the ancestral type. They are not necessarily at equal 

 distances from it, for (owing to natural selection or other causes) there 

 may be more tendency to diverge on one side than on the other : and 

 the limit will be proportionately greater. But it will be seen that we have 

 arrived at an important conception of 



the path followed by a descendant. If 



OP be the ancestral type, then the 



paths of the descendants are confined 



between certain limits, OQ, OR, on 



opposite sides of the ancestral type. 



These limits are not actually defined 



so sharply as we have drawn them ; 



they are more like the edges of a 



shadow ; and an accurate mathematical 



expression for them would express a 



gradual shading off a fall rapid at 



first and very gradual afterwards. But 



if for simplicity we think of actually well defined limits, we shall not be 



far from the truth, and the results arrived at will have corresponding exact 



propositions, differing from them only in detail. 



20. Transmission of sexual characters. The general law of trans- 

 mission above considered has however a class of important exceptions, 

 applying to cases where the differences between parents are not small, 

 and are not continuously graded. The most conspicuous and important 

 of these is presented by the sexual characteristics themselves. The child 

 of male and female parents is not something midway between (with 

 possible extremely rare exceptions), but either the one or the other. We 

 are clearly in the presence of a totally different law of transmission. 



2 1 . We can illustrate the difference graphically by adopting separate 

 planes for the representation of the two sets of characters. Our diagrams 



have hitherto been drawn in 

 one plane, OXY : if we devote 

 this plane to paths such as OP, 

 representing the growth of a 

 female organ, we shall not in 

 general be able to represent 

 the growth of the correspond- 

 ing male organ on the same 

 diagram, as the definitions 

 would not apply. But we 

 could take another plane, OXZ, 

 in which OX still represented 

 the time scale, and OQ the 

 FIG. 8. growth of the male organ. In 



so far as the organ in one sex 



is rudimentary in the other, we could of course represent it by a line 

 near OX in the other. Thus if OP represents the growth of breasts 



