506 Conclusion. 



ample for the breadth of the leaves, the length of the 

 fruits, the size of the flowers, etc. But exactly the same 

 can be done for those forms which have been recognized 

 as species by the best systematists and exhibit such ex- 

 cellent points of difference as, for example, the size of 

 the flowers in Oenothera biennis L. and 0. ninricata L. 



The boundaries between my new species are no more 

 obliterated by transgressive variability than are those 

 between recognized types. The phenomenon is a very 

 general one in the whole animal and vegetable kingdom ; 

 and is very apt to land any one who confines his attention 

 to one character at a time, into difficulties. In settling the 

 identity of a doubtful case situated in the borderland be- 

 tween two types the other characters of the doubtful in- 

 dividual must receive attention ; for, in spite of the phe- 

 nomenon of correlation, the other characters will almost 

 always show to which side the doubtful form really, be- 

 longs. And if this comparison does not suflice, we must 

 resort to experimental sowings ; the direction in which 

 the offspring regress will leave no doubt as to the specific 

 type to which the doubtful individual belonged. 



In this respect as in all others, my new forms of 

 Oenothera are species and not varieties ; with the single 

 exception of O. nanella (see § 18, pp. 360-363).^ The 

 difference between a species and a variety is that whilst 

 a variety is marked by the possession of a single distinc- 

 tive character a species or subspecies differs from its 

 nearest allies in almost all its parts. Lack of pigment, hairs 

 or spines ; branching of leaves or stems ; or an abnormal 

 development of these characters; laciniate leaves and 

 petals, etc. — such are the characters of true varieties. 

 There are two further points about varieties. First the 



* On this point see, especially, the second volume. 



