SPECIES BY INFINITESIMAL VARIATIONS. 19 



seven, of which one, being very commonly the parent, will 

 often have the advantage from the start. Now between seven 

 the competition may be very strenuous, but it is evident that 

 the range of variability can on the average be only a very 

 small one, and consequently that progress can only be much 

 slower than is often (consciously or unconsciously) supposed. 

 In the majority of cases there cannot be any competition 

 between two plants separated by a considerable distance, 

 unless, later in life, for the services of insects to pollinate their 

 flowers. Wind, of course, might affect those on one side of 

 a crowd more than others, but any such agent cannot as a rule 

 have any very great effect. 



I find that I was wrong in stating Coleus spicatus to be 

 endemic in South India, for Bentham says that it also 

 occurs in Africa. On the other hand, the argument about 

 Coleus barbatus gains greatly in force from the fact that this 

 species also occurs in tropical Africa. 



I have received many letters deaUng with my former paper, 

 and the usual tenor of them is that a good case has now been 

 made out for mutation, and that if we can now prove that 

 one species can mutate into another, all will be finished. But 

 no one has seriously tried to pick a hole in my argument, 

 and I take this opportunity of repeating that it is impossible 

 to get this species by continuous variation, and that from 

 analogy all the rest follows. 



There are several other arguments that might be brought 

 forward, but want of time compels me to leave them for 

 another paper. 



