28 



DE. E. E. GATES— CONTEIBUTION TO A 



conception that some fundamental germinal cliange has occurred in the origin of 



O. brevisfyUs, presumably from O. La^narcUana, that change exhibiting itself externally 



in a different response as regards leaf-shape, but also in the very short style and badly 



formed stigma, the short sepal tips, and the almost entire failure of the ovaries to 

 mature. 



DeVries was able to show that 0. hrevlstylis will breed true when seeds were obtained 

 by self-pollinating many flowers, but O. hrevidylis has usually been propagated by using 

 its pollen to cross with O. Lamar cMana, the offspring being about 50 per cent. O. h^evi- 

 stylis and 50 per cent. O. Lamarchiana. My cultures of O. brevistylis were from seeds 

 of DeVries which had been grown in that way for five generations. The sixth 

 generation in my garden yielded 92 plants, of which 55 were O. Lamar cUana, 

 34 O. brevistylis and 3 O. lata mutants. The seventh generation, grown last year, 

 gave a simibir result, showing that there is continued sharp alternative inheritance 

 when this mutant is crossed with its putative parent; and there is no tendency 

 for any of the leaf characters or flower characters to separate and be independently 

 inherited in the Fg or later generations. The logical conclusion is, therefore, that all 



these character-changes represent the external manifestations of one fundamental 



germinal change. The sterility of the 



emphasize the depths to which 



the organism has been stirred. It appears that the changes have been so serious that 

 the development of the ovaries is rendered almost impossible. It is possible, however, 

 that the failure to produce seeds results, not from abortion of the ovules, but from the 

 fact that pollen will not germinate on the imperfectly formed stigma. 



O. riibrinervis and 0. rubricalyx. 



The mutant O. rubrinervis probably bears a somewhat different relation to its parent 

 (O. Lamarckiana) from some of the other derivatives. DeVries (1907) has shown 

 that when O. Lamarckiana or one of its mutants is used to pollinate O. biennis, Linn., two 

 types are produced in the F^, which he calls respectively 0. laeta and O. velutina. This 

 type of behaviour will be considered further in the section on hybridization, but it might 

 be pointed out here that, as Honing (1911) has shown, O. Lamarckiana and O. rubri- 

 nervis difl'er from each other in characters which are analogous to the differences 

 between O. laeta and O. velutina. In both cases the former has broader, the latter 

 narrower leaves, the former has darker green leaves and the latter more long hairs, etc. 

 I shall show later that 0, mnricata, Linn., from the wild condition also often°produces a 

 broad-leaved and a narrow-leaved type, so that this " double " condition is of funda- 

 mental interest in the hereditary behaviour of the genus, though its full meaning is not 

 yet clear. 



In my cultures of O. rubrinervis in 1907 a new form (O. rubricalyx) originated 

 suddenly in a single individual. Its origin and subsequent behaviour when bred pure 

 and in crossing have been described in a previous paper (Gates, 1911 b). It need only 

 be said that this form is morphologically identical with O. rubrinervis, but differs 

 strikin gly in a physiological character— pigment-production ; there being a large increase 



