196 THE FUNCTION OF SEX 



tion in a patent or latent condition, in which case there is no 

 blending in that generation. In subsequent generations there is 

 blending if descendants of the mutants interbreed, for they will 

 have the character in a patent (dominant) or latent (recessive) 

 condition. It is not to be supposed, however, that mutations are 

 reproduced unaltered in offspring and descendants. Presumably, 

 like other characters, they are subject to fluctuating variations, of 

 which the general trend is retrogressive. They disappear, there- 

 fore, in time by sudden mutation, or probably more usually by 

 gradual retrogression, unless preserved by selection. We have no 

 reason to suppose that they are less liable to retrogression when 

 latent than when patent. No doubt, numbers of latent characters 

 have persisted through many generations in domestic species ; but, 

 speaking comparatively, all domestic species are of recent origin, 

 and the retrogression of some useless patent characters (e.g. limbs 

 when they have become useless) is very slow. Nature is parsi- 

 monious of her materials ; it is improbable, therefore, that the 

 germ-plasm is burdened for ever with useless dormant characters. 

 At any rate, there is not an iota of evidence that it is so burdened. 

 An interesting point in connection with latent traits is the reap- 

 pearance in offspring of numerous narrow, rather faint stripes when 

 the horse, which has no patent stripes, is crossed with the Burchell 

 zebra, which has a few broad ones. This is the most ancient 

 latent character known to me. It is commonly believed that the 

 narrow stripes are due to reversion to a common ancestor. This 

 may or may not be true ; but it does not follow that the dormant 

 trait is carried by both species. Dormant traits are practically 

 unknown in natural varieties such as the zebra ; and horses, even 

 when pure bred, sometimes display traces of stripes. It is pro- 

 bable, therefore, that the horse is the carrier of them, that they 

 were rendered dormant in the ancestry by the artificial selection of 

 mutations, and that they have grown faint through retrogression. 



323. Offspring vary ' spontaneously ' from their parents in two 

 ways. First, they blend parental characters, and so differ from 

 both parents ; or, if there is exclusive inheritance, they differ from 

 one parent. Second, they differ from both parents otherwise 

 than by blending. With this latter sort of spontaneous variations 

 we have already dealt. 1 We have to consider the variations 

 which result from blending. Suppose, now, to choose a homely 

 example, a man with an exceptionally broad nose mates with a 

 woman with an exceptionally high nose. Then, if the result be 



1 See chapter'v. 



