OENOTHERA GRANDIFLORA AIT. 239 



stoutness of all its organs. These two types are constant from seed, 

 but the gigas keeps on mutating into lorea and ochracea. 



3. The crosses among 0. grandiflora, 0. ochracea, and 0. lorea 

 show that these forms are isogamic, the pollen carrying the same 

 hereditary qualities as the egg cells. 



4. 0. grandiflora yields twin hybrids with the same species which 

 produce twins in their combinations with 0. Lamarckiana. The 

 female organs of 0. biennis, 0. syrticola (muricata), 0. suaveolens, 

 the pollen of 0. biennis Chicago, and both sexes of 0. Cockerelli split 

 0. grandiflora into laeta and velutino, whereas the cross 0. biennis 

 Chicago x grandiflora yields the twins densa and laxa. The twins 

 appear, on the average, in about equal numbers. This splitting 

 fails when the crosses are made with mut. ochracea instead of 0. 

 grandiflora. Their progeny is uniform and corresponds, so far 

 as investigated, to the laeta among the twins. 



5. In other crosses of 0. grandiflora the hybrids also resemble 

 the corresponding ones of 0. Lamarckiana. 0. grandiflora x syrti- 

 cola produces the type gracilis, 0. grandiflora x biennis hybrids of 

 the types of biennis and ochracea, among which the first is constant 

 in its progeny, whereas the second repeats the splitting. 



6. In crosses with 0. Lamarckiana, 0. grandiflora produces com- 

 binations of the two groups of twins. I found three such types. 

 One of them embraces about one-half of the offspring and corre- 

 sponds to the laeta; it is called ovata. The two others appear each 

 in one- fourth of the whole culture and are called lutea and brunnea. 

 The first corresponds to the combination ochracea x velutina, the 

 second to grandiflora x velutina. These triple hybiids are constant 

 in their progeny, inasmuch as they do not produce individuals of 

 the other types, but split off some forms which constitute different 

 combinations of the parental characters and partly also of those of 

 the mutants. One of them, lacking the initial rosette of radical 

 leaves, appears in percentage figures which seem to correspond to 

 the formula of Mendel for monohybrids. 



7. If the crosses are made with the mutant lorea, this character 

 is latent in the first generation and reappears in the second in about 

 one-quarter of the individuals; but this rule shows some exceptions. 



8. From these facts, in combination with the occurrence of about 

 25 per cent of barren grains among the seeds, we arrived at the con- 

 clusion that the yearly production of large numbers of ochracea is 

 a phenomenon of mass mutation, analogous to the instances descri- 

 bed by Bartlett and due to an initial mutation of the ordinary 



