136 



THE AMEBIC AN NATURALIST [VOL.L 



perfect pollen and are little characterized by variability ; 

 while where the species are numerous and coincident in 

 their range both variability and pollen sterility are con- 

 spicuous. 



We may now consider another highly variable group, 

 which has not infrequently been called a hybrid family, 

 namely the Rosacese. The genera Rosa, Rubus and Cra- 

 taegus are notable for the extreme difficulty they have 

 offered from the systematic point of view. Three of my 



graduate students have investigated these genera and the 



results may be conveniently summarized by reference to 

 the genus Rubus. In the case of Rubus, in regions where 

 it has been exhaustively studied, there is almost no end to 

 the species which may be set up. In Europe, in fact, the 

 species have mounted into the thousands. The situation 

 may for the sake of brevity be considered under three 

 heads. First, there are species which range together and 

 have flowering periods which overlap— a condition com- 

 mon to the mass of our ordinary Rubi. In Rubus villosus, 



