CRUSTACEA AND MOLLUSCS 259 



being destroyed on the ground of these excessive 

 variations ; it does not allow too great an advance of 

 the variations, because it crosses these animals with 

 others that have not the variation in question, or have 

 one in the opposite direction, so that the normal 

 standard is reached again in their offspring. Thus 

 excessive variations tend to disappear in the general 

 crossing, as amphimixis has an equalising effect. 



We have already seen that the indifferent characters 

 of animals are preserved in the general crossing, because 

 the plus and minus variations neutralise each other. 

 Without amphimixis the variations would diverge in all 

 possible directions, and each animal would separate 

 further and further from the others. It would be 

 impossible to comprise a definite group together as a 

 species. Definite species are maintained entirely by 

 amphimixis. As the law of heredity explains the 

 resemblances of animals, so amphimixis explains that 

 its action marks off definite specific types from each 

 other in the world. 



A further result of this levelling tendency of am- 

 phimixis is that isolated changes, even if they are 

 useful, cannot modify the species, because they are lost 

 in the general crossing. It is only when the majority 

 of a species that live together vary that the character 

 in question will be impressed on the whole species by 

 natural selection ; in other words, only plural variations 

 are taken into account by natural selection. The 

 majority have, of course, only to influence the survivors. 

 If, say, a third of a species is extinguished every year, 

 it is sufficient if a little over a third of the species has 



