220 



MASS MUTATIONS AND TWIN HYBRIDS OF 



are therefore mutated before fecundation, we may deduce from 

 table IV that normal egg cells of the species, after fecundation by 

 0. biennis L., give uniform hybrids of the biennis type, whereas the 

 mutated egg cells reproduce the type of ochracea. This would explain 

 the dimorphous condition, where uniformity would otherwise be 

 expected. 



Table IV 

 Crosses of 0. grandiflora and O. biennis, made in 1914 



0. grandiflora x 0. suaveolens. — I made this cross on two speci- 

 mens in 1915, but got a very small harvest of seeds, yielding only 

 8 and 13 seedlings. In June the same two types were seen as in 

 the cross 0. grandiflora x biennis. I counted 2 and 4 ochracea 

 with broad leaves, a low stature, and weak constitution, and with 

 stems without preparatory rosettes. Among the remaining plants 

 6 and 8 were intermediate between the two parents, with stout, 

 sparely branched stems and dark green leaves of an intermediate 

 form. Besides these there was one lorea. All of these plants flowered 

 in August, and showed in their flowers intermediate characters; 

 but I have not continued the experiment. 



D. First generation of crosses with Lamarckiana. 



Both 0. grandiflora and 0. Lamarckiana produce twin hybrids 

 in a number of crosses. If they are fertilized among themselves, 

 therefore, combinations of these twins may be expected. Moreover, 

 it is known that from crossed seeds and from seeds of hybrids the 

 same mutations may arise as from the parent species. In this way 

 I observed two hybrids with the characters of gigas, some with those 

 of the dwarfs, and a third type of doubtful relations. These mutants 

 are rare, however, whereas the products of the splitting were ob- 

 served in all my experiments. 



For my crosses I have not only used the two species themselves, 



