524 APPENDIX 



in a female, the corresponding growth for a male might be a path Op, 

 very close to X, since the growth is slight. And if the organ of 

 one sex were actually non-existent in the other, it could nevertheless 

 be represented by the line OX itself, which represents zero growth 

 throughout. 



22. So that each sex would be represented by two sets of sexual 

 characters, "patent" and "latent" as Dr Archdall Reid calls them. 

 " Patent " female characters would be represented by paths such as OP 

 far from OX, in the special female planes : and " latent " female characters 

 would be represented by paths near OX (or coinciding with it) in the 

 special male planes such as XOZ. 



23. Now we know that our principle (IV.) no longer applies in this 

 case. From wide experience we know that in sexual characters the 

 children represent either one parent or the other : not both (VIII.). 

 Let us see how this principle can be represented diagrammatically. 

 To avoid the complexities of solid geometry, let us make a cross section 

 of Fig. 8 through the lines PTQ, as represented in Fig. 9. The 



point T represents rudimentary or 

 zero characters, P well developed 

 female, and Q well developed male, 

 Now the child has two sets of 

 qualities : we may represent him by 

 a pair of points, AB, connected to- 

 gether, or rather kept separate by a 

 rigid rod. (The notion of keeping 

 Q separate is appropriate, because he 

 has one set patent and the other 

 _ latent : not both patent or both latent, 



A^ B z i.e. A and B must not coincide.) 



FIG. 9. Experience then tells us that there 



are two possible positions of AB, as 



crudely represented in Fig. 9 : either A falling on P and B on T, or 

 A on T and B on Q. 



24. This result would follow at once if we conceive of P, T, and Q 

 as centres of attraction for the ends, A,B, of the rod. And this conception 

 would be quite in keeping with what has gone before: thus in Fig. 7 

 we might regard the type path, OP, as attractive, so that other paths would 

 be attracted near it. If we took a cross section of the path as we did 

 for those of Fig. 8, we should get a single point, P, representing the parent, 

 attracting the single point, R, representing the child. When we come to 

 sexual characters we must represent the child by two points as above : 

 and we then get a new interpretation. 



25. It is scarcely necessary to say that the diagrammatic analogy 

 proves nothing whatever. It is merely a convenient summary of facts 

 which we learn from wide experience. But it may help us to realize 

 the connexion, with these fundamental and well known facts, of others 

 which are not so well known. 



26. For instance, it may help us to realize how Mendelian inheritance 



B, 



