508 Conclusion. 



great many subspecies and varieties are not, or at any 

 rate not yet, constant and that they revert to the parent 

 species from time to time is commonly held. It is this 

 that is meant by the statement that varieties are incipient 

 species. This conception is however entirely erroneous. 



My new species are either absolutely constant from 

 the beginning, without the slightest trace of reversion; 

 or, when they are not, exhibit no increase in constancy, 

 in response to selection. 



In order to test the constancy of the new forms di- 

 rectly after their arising from O. Lamarckiana or other 

 families one has to artificially fertilize the mutants them- 

 selves with their own pollen. Large quantities of the seed 

 are then sown ; and as the new species is easily and cer- 

 taii^ly recognized during the first months of its life, be- 

 fore it develops a stem, some hundreds or even thousands 

 of seedlings can be recorded. Of these only as many as 

 can be conveniently accomodated are chosen, at random ; 

 and cultivated until they flower and ripen their fruits. 

 The constancy of a new form can be determined by grow- 

 ing it in this way for a number of generations. From 

 the very beginning O. gigas, O. ruhrinervis , O. oblonga, 

 O. alhida, O. Icptocarpa, O. semilata and O. nanella were 

 absolutely constant. 



O. schitillans, 0. clliptica and 0. sitblinearis, on the 

 other hand, proved to be inconstant. Only a small pro- 

 portion of the seeds of these forms repeat the parental 

 type. The others revert to O. Lamarckiana or produce 

 other mutations. 0. scintillans usually did not repeat it- 

 self in more than 30% of its offspring and could not be 

 got to increase the proportion by continued selection ; the 

 others did not furnish sufficient seeds for conclusive ex- 

 periments. 



