504 



THE AMERICAN NATURALIST [Vol. XLIX 



have seen to be true for almost every characteristic sufficiently studied — 

 for the comb form, the nostril form, cerebral hernia, crest, muff, tail 

 length, vulture hock, foot feathering, foot color, earlobe and both gen- 

 eral and special plumage color. Everywhere unit characters are changed 

 by hybridizing. 



In crosses between (Enothera rubricalyx and (E. grand- 

 iflora I have studied with care the modifications which 

 take place in the expression of the various character- 

 differences in Fj, F 2 and later generations. Many of the 

 results have been recorded in detail elsewhere (Grates, 

 1914, 1915a, pp. 250-282). It need only be said that the 

 foliage characters in F 2 form an absolutely continuous 

 series so that it is impossible to apply to them usefully 

 the unit-character conception. In F 3 a large number of 

 races were obtained differing in many ways as regards 

 their foliage, many of them breeding true and others 

 varying within wide or narrow limits. Occasionally in 

 back-crosses an apparently complete reversion takes 

 place to one or other of the parents, but blending and 

 fractionation of the characters is the rule. 



It is, however, difficult to obtain critical evidence from 

 the foliage because, while the original differences are 

 sharply marked, yet it is always possible to assume that 

 the continuous F 2 series and the numerous F„ races result 

 from the presence of many independent units. 2 I will 

 therefore confine my attention to the sharp pigmentation 

 character (R) of rubricalyx, for in the inheritance of this 

 character crucial evidence may be obtained. The origin 

 of this dominant unit-character through a single muta- 

 tion, and the subsequent attainment of the duplicate con- 

 dition (RR') for this character in some of the offspring 

 of later generations (1915&), have been pointed out else- 

 where. Here we will examine the modifications of R 

 which take place when rubricalyx is crossed with (E. 

 grandiflora. 



The main facts regarding the variability of R in these 



