xiii VARIATION AND EVOLUTION 151 



in appearance from those of the wild form. Such a plant 

 would breed perfectly true, and a botanist to whom it 

 was presented, if ignorant of its origin, might easily rele- 

 gate it to a different genus. Nevertheless, though so 

 widely divergent in structure, such a plant must yet be 

 regarded as belonging to the species Lathyrus odoratus. 

 For it still remains fertile with the many different varie- 

 ties of sweet pea. It is not visible attributes that con- 

 stitute the essential difference between one species and 

 another. The essential difference, whatever it may be, is 

 that underlying the phenomenon of sterility. The visi- 

 ble attributes are those made use of by the systematist in 

 cataloguing the different forms of animal and plant life, for 

 he has no other choice. But it must not be forgotten that 

 they are often misleading. Until they were bred together 

 Eur alia wahlbergi and E. mima were regarded as perfectly 

 valid species, and there is little doubt that numbers of 

 recognised species will eventually fall to the ground in the 

 same way as soon as we are in a position to apply the 

 test of breeding. Mendelism has helped us to realise that 

 specific characters may be but incidental to a species — 

 that the true criterion of what constitutes a species is 

 sterility, and that particular form of sterility which pre- 

 vents two healthy gametes on uniting from producing a 

 zygote with normal powers of growth and reproduction. 

 For there are forms of sterility which are purely mechan- 

 ical. The pollen of Mirabilis jalapa cannot fertilise M. 



