212 



THE AMERICAN NATURALIST 



[Vol. XLIV 



vegetative splitting, or reversion to the parental types, 

 in such forms as the reputed graft-hybrid, Cytisus 

 Adami, differ in any essential respect from Mendelian 

 segregation in germ cells. In angiosperms, the great 

 majority of which are hermaphrodite, there is a separa- 

 tion of tendencies, which may be thought of as a "segre- 

 gation," in the Anlage of every flower, the stamens pro- 

 ducing microspores and male gametophytes, while the 

 ovules bear the megaspores, which develop the female 

 gametophyte. These two tendencies— namely, for the 

 stamens to produce microspores and the ovules to pro- 

 duce the female gametophyte — are almost never inter- 

 changed (although Xemec, '{18, has described a case in 

 Hyacinthus, in which the microspores may produce a 

 structure resembling more or less the female gameto- 

 phyte). Whether there is here a separation of substances 

 in the cytoplasm or chromoplasm during the division of 

 certain cells in the Anlage of the flower, is not known, 

 but it is not impossible that such is the case. 



Viewed from this standpoint, all differentiation during 

 ontogeny may be considered a "segregation." A con- 

 sideration of what this really consists in would involve 

 the whole great problem of individual development, 

 which I shall not touch upon here. But it may be pointed 

 out that maturation, particularly in Mendelian hybrids, 

 may be looked upon as a period of active germ cell dif- 

 ferentiation. 



The factors involved in such differentiation may be, 

 in some cases, quite as complex as those involved in 

 development itself, but as I have shown from the evi- 

 dence of these Oenothera hybrids, and as appears from 

 color inheritance in mammals and from other evidence, 

 in many cases at least the difference between Mendelian 

 germ cells must be of a simple quantitative sort, involv- 

 ing either a difference in the amount of certain material 

 substances or a difference in the energy-content of cer- 

 tain constituents. 



From this point of view, many instances of Mendelian 

 behavior are seen to be cases of quantitative inheritance. 



