12 THE DEVELOPMENT OF KAW COTTON 



know that if the abnormality had been simply due to 

 accidents of nutrition it would not have reproduced 

 itself. A few remarks on the material from which com- 

 mercial varieties are thus derived may not be out of 

 place. 



We have seen that all cotton-plants can be classified 

 on broad evolutionary grounds into three main species, 



and several equally important, though eco- 

 Subspecies. . ,, . . 



nomically useless, minor primitive species. 



We have abstained from carrying the classification farther 

 on account of the absence of experimental evidence, but 

 the next step would be the grouping of all species of 

 cottons of the Peruvian type, for example, into groups 

 of relations, each group being designated a subspecies. 

 The members of a subspecies would all be alike to ordinary 

 observation, just as all the brambles in a hedgerow are 

 obviously brambles. Closer observation would reveal 

 differences other than those due to accident of situation 

 and nutrition, whose nature could be tested by raising 

 offspring from self -fertilized seed. 



If this test be applied to subspecies of cotton, it is 



found that many separate components which breed true 



at once go to make up the subspecies, in 



Elementary addition to a i arger number of plants which 



Species. 



do not breed true. These latter are neces- 

 sarily of hybrid origin, though the cross which originated 

 them may have taken place even centuries before; the 

 former may or may not have originated from a cross. 

 Where the circumstances are such as to justify the pre- 

 sumption that they did not originate by crossing and 



