GENERAL DISCUSSION. 



101 



The polygon can not move as an undisturbed whole, but will start the 

 formation of a new polygon at the next center of variation. With the 

 increase in the polygon at the new center, the old polygon decreases. Such 

 a phenomenon is generally thought to be necessarily that of a unit-char- 

 acter, but we have seen that no such germinal discontinuity is necessary 

 to account for it. The term ' ' unit-character ' ' should be reserved for those 

 cases of truly distinct units analogous to chemical compounds or to the 

 rungs of a ladder, rather than to the dentation or crenation on a leaf- 

 margin. But, if we grant the existence of such centers of variation as 

 figured, then, in addition to evolution proceeding by even flow and by leaps, 

 it may also proceed by waves or undulations. 



TABLE 22. Hypothetical case of somatic centers of variation. 



-Nmnlior reduced to 



The evolution of varieties and subspecies is readily enough accomplished 

 in these beetles. But the passage of the variety or subspecies into a spe- 

 cies, that is, the acquisition of intersterility, seems much more difficult of 

 accomplishment. The hypothesis of Romanes, that it has arisen by the 

 endowment of certain individuals with mutual fertility but with interster- 

 ility with the rest of the species, seems to me extremely unlikely, as such 

 individuals would have very low fecundity from the slim chance of finding 

 their interfertile mates. Of course there are a few exceptions in cases of 

 particular causes of endogamy. A more probable hypothesis is that of 

 correlation of some genital or developmental feature with the varietal 

 features, so that they become sufficiently unlike to cause intersterility. 

 There is a partial intersterility of Hippodamia convergent var. caseyi with 



