1918] DEVRIES—MASS MUTATIONS 397 
O. grandiflora normally produces stems without this preliminary 
step. In the first generation the differences had been the same in 
this respect, the biennis plants having preparatory rosettes and 
stout stems, but the ochracea lacking these characters. Thus we 
see that this mark returned in part of the specimens of the second 
generation. I shall simply call the plants growing up without 
rosettes ochracea, and retain for the others the term “biennis,’’s 
but both types had the broad, yellowish leaves of the ochracea of 
the first generation, and its loose spikes. I counted in August 33 
per cent of the diennis and 67 per cent of the ochracea type. 
Resuming these descriptions we see that the second generation 
of the first ochracea hybrids was constant and like the parent in all 
respects except the rosette character. This was absent in two- 
thirds of the specimens, which thereby were just like their parents, 
but present in one-third, which returned to the mode of growth of 
_the other grandparent. This disposition may be expressed by the 
following pedigree: 
IQI4 grandiflora X biennis 
First generation 191 5 ochracea biennis 
| | 
Second generation 1916 ochracea ochracea biennis biennis 
This pedigree shows the splitting of the biennis character in the 
first and second generations and the constancy of the ochracea 
marks in the second. For the majority of the marks of ochracea 
it runs parallel to the ordinary scheme for the splitting into laeta 
and velutina, but for the character of the rosettes it is parallel to 
that of the Jaeta and velutina produced by the crosses between 
O. Hookeri and O. Lamarckiana, where the /aeta is known to split 
off velutina in the succeeding generations. The explanation which 
offers itself is that the annual growth is here dominant over the 
5 I would have preferred to call them annual and biennial, since one see assumes 
the annual habit of O. grandiflora and the other the biennial growt . biennis L.; 
but as I cultivated all of them as annuals, it does not seem siaabla to use these 
terms here. The chosen terms relate obviously to one prominent character} Peapred 
should not convey the conception that all other characters are the same a: 
prototypes of the names. 
