iQis] DE VRIES-OENOTHERA BIENNIS 185 



be cultivated together on a bed of my experiment garden, which 

 enabled me to inspect them almost every day during their develop- 

 ment and through the whole summer. The characters which 

 distinguish the dwarfs in the stage of young rosettes, with leaves 

 a few centimeters in length, were discovered in the following way. 



The self-pollinated flowers of the dwarf specimen of Stomps 

 in 1913 had set no good seeds, but flowers pollinated from pure 

 biennis had produced some fruits. Now my 0. Lamarckiana mut. 

 nanella, when crossed with O. biennis, yields only, or almost only, 

 dwarfs. Therefore, the expectation was justified that such might 

 also be the result of the cross 0. biennis mut. nanellaX.0. biennis. 

 Seeds from this cross had been sown about the same time; they 

 yielded 108 seedlings, all of which have been planted out and have 

 flowered. They were dwarfs without exception, reached in Sep- 

 tember a height of 40-45 cm. only, were richly branched, and had 

 all the marks of 0. biennis combined with the dwarfish stature 

 and the liability to the same bacterial disease as is shown by the 

 dwarfs of 0. Lamarckiana. The young rosettes of these crossed 

 biennis dwarfs clearly differed from the rosettes of the pure biennis. 

 After the three or four first leaves with long petioles, there followed 

 a group of leaves with smaller stalks and some sessile ones, thereby 

 rendering the whole rosette far more compact than the corre- 

 sponding ones of the pure biennis. With this character as a cri- 

 terion, I isolated from my pure line boxes 8 individuals. One of 

 them proved afterward to be a mistake; it was a pure biennis. 

 Seven were dwarfs and have flowered; they were, in all external 

 respects, like the crossed dwarfs of the control culture. Among 

 the 8500 remaining plants I discovered later, in the field, only one 

 dwarf. This shows that the characters were sufficiently reliable. 

 All in all, I had 8 dwarfs in 8500 plants, making about o . i per cent. 

 They occurred among the progeny of one of the self-pollinated 

 mothers in the second generation (3 dwarfs), and of three of 

 the parents in the third generation (5 dwarfs). Some of them have 

 set good fruits after self-fertilization. 



One of the most interesting and useful features of 0. biennis L. 

 is its propensity to make lateral rosettes from the base of the 

 flowering stem. It is possible to isolate these rosettes and to have 



