The Boundaries Between Related Species. 429 



is only when we compare other characters as well that 

 we can distinguish the individual species. 



The object of exact inquiry should be first to collect 

 as many of these continuous series as possible ; but second 

 to anah^ze them into their component units. 



This analysis can be effected both by the statistical 

 and by the experimental method. Let us examine the 

 former first. 



A transgression of the limits is only exhibited by 

 isolated and relatively few individuals; the vast majority 

 belong to the mean type of the species. Therefore if we 

 take care not to be too much on the lookout for transi- 

 tional forms, or even, if we try to make our measure- 

 ments as numerous as possible, the separate curves will 

 become discernible. The result will be exactly that which 

 we got from an investigation of the Oenothera flowers, 

 namely curves with numerous apices, such as those with 

 which the work of Bateson^ Ludwig and others has 

 made us familiar. Each apex indicates a group of indi- 

 viduals which belong together, to a type, or even to an 

 elementary species. 



Transitional forms can then be recognized imme- 

 diately by their rarity. It becomes obvious that the 

 transitions are only apparent and that no real continuity 

 between the different centers of variation exists. These 

 curves show no more than that the edges of neighboring 

 curves on the same abscissa may overlap. 



The simplest way of making the experimental method 

 clear is to take the case of the Oenothera flowers again. 

 We collect the seeds from the fruits of two flowers of 

 equal size of which one is one of the largest flowers of 

 Ocn. biennis, the other one of the smallest of Ocn. La- 

 marckiana. There can be no doubt what the plants raised 



