296 The Pedigree Faniilics. 



the leaves of these three types, ohlonga, alhida and sc'in- 

 tillans, can really only be distinguished by their color 

 and substance. 



The various new species (with the single exception 

 of O. Icptocarpa which is not recognizable until it flow- 

 ers) can, however, be recognized with certainty at any 

 stage in their development by their leaves. All the later 

 characters, the form of the flowering branches and of 

 the flowers themselves, in 0. lata the absence of pollen, 

 the size of the fruits and the abundance of seed — can 

 be predicted from the leaves, if the species in question 

 lias once been seen in flower and fruit. But, as a matter 

 of fact, I have planted out every year, for further growth, 

 a greater or smaller number of the plants which had been 

 sorted by their seedling characters; and in no case has 

 my identification proved to be erroneous. 



I have sought for mutants among the crops from 

 Oenothera seeds as often as possible by this method. 

 Whether, and in what quantities, they appear depends in 

 great measure, as I have already said, on the extensive- 

 ness of the sowing. But these things are undoubtedly 

 influenced by other causes operating during germination 

 (see p. 263) or fertilization or even earlier. 



As a rule the new species proved much less mutable 

 than the original O. Lainarckiana from which they orig- 

 inated. It is only the inconstant forms amongst them 

 which exhibit a very high degree of mutability, as for 

 example O. scintillans. 



The power to mutate is maintained after crossing 

 and in cases where the original form O. Lamarckiana 

 reappears out of a cross it does so with its full — I had 

 almost said normal — capacity for mutation. 



