25 



seems unescapable. It cannot be without significance that so 

 many of our endemics are in genera, or are related to species 

 which, if the phrase be permitted, appear to flutter along the edge 

 of things. This generic and specific waywardness is forcibly im- 

 pressed on us by the dropping out of species as the genus dimin- 

 ishes through the area, or by the number of species related to 

 endemics, which find their outposts here or very near here. 



Excluding the species of the endemic genera, which are as we 

 have seen, nearly all relict endemics, it proves interesting to see 

 what relation our eastern American proportion of species in each 

 genus containing endemics, bears to the distribution of the whole 

 genus. The following table gives these figures, for all the 

 endemics except species of endemic genera and three others 

 which will be considered presently. 



Native Species in 

 E. N. America 



Total New World 

 Species 



Total Old World 

 Species 



Sporobulus . 

 Juncus. . . . 



Salix 



Prunus .... 

 Dentaria . . . 

 Hibiscus . . . 

 Hypericum . 

 Kneiffia. . . . 



Pyrola 



Stachys .... 

 Eupatorium . 

 Senecio . . . . 



19 



47 



39 



8 



6 



5 



21 



8 



40 



60 



47 



31 



6 or 7 



63 



82 



12 



13 



16 



130 



500* 



43 



103 



196 



86 



10 



200 



166 



3 



7 



150 



265 



535 



(?) 



As throwing a broader light on this question of the production 

 of endemics far from the center of distribution of a genus, it is 

 not without significance that in only three cases, in the table 

 above, does the region in eastern North America seem to be 

 anywhere near the generic center. In Kneiffia, which may be 

 all American, as there is some doubt as to the Old World species 

 being correctly credited to the genus; and in Pyrola and Dentaria, 

 which are nearly all of the North Temperate Zone and almost 

 impossible of generic centralization, we have the only exceptions. 

 It would seem as though these three were not enough to upset 

 the main contention of this paper which is that endemism is 

 most likely to occur at or near the periphery of generic distribu- 



* Nearly all tropical. 



