588 Species Accordijig to flic Theory of Mutation. 



its sim])lest terms, if we call those qualities whicli in cross- 

 ing conform to Mendel's law, varietal characters, and 

 those which give uni-sexual unions, specific characters. 

 This would conduce to a more general form of the thesis 

 based on the experiments with Lychnis (p. 582), viz., 

 that two related forms can differ from one another simul- 

 taneously in varietal and in specific characters. 



Are they therefore to be regarded as varieties or as 

 species ? Here we arrive at the boundary which separates 

 facts from conventional terminology. Here lies the point 

 to which GoETiiE refers in his well-known lines : 



''Dich im Unendlichen zn finden, 

 Musst unterscheiden und dann verbinden." 



The process of distinguishing is an objective one, 

 that of combining subjective. The former is the imme- 

 diate result of inquiry; in our case, of systematic studies 

 on the one hand, and of experiments in hybridization on 

 the other. The combination is partly a question of taste ; 

 it has to serve special aims ; and above all it must facilitate 

 general conceptions and mutual understanding. 



It is not my task to go more deeply into the question 

 of systematic subdivisions or to make any definite pro- 

 posals.^ My sole object is to place the actual facts in as 

 clear a light as possible. This attempt, however, again 

 leads to the conclusion that here also this insight can 

 only be obtained on the basis of the theory of mutation. 

 It is only by attempting to analyze the species into its 

 component factors, the elementary characters, that we 



* If two forms were found to differ from one another exclusively 

 in varietal characters, but the number of these were very large, they 

 would probably have to be separated as species. Here also the dis- 

 tinction between species and variety is an arbitrary one, and as a 

 matter of fact, many of the larger groups and sometimes even whole 

 families have among their distinguishing marks some which do not 

 reallv differ from "varietal character^." 



