80 De Vries: ATAVISTIC VARIATION IN OENOTHERA 
this number was sufficient to show the whole range of the vari- 
ability of the race. Nine plants bore typical cruciate flowers with 
linear petals, two specimens had broad obcordate petals like those 
of O. muricata, and the five remaining had an intermediate breadth, 
varying in all degrees between these two extremes. Even on the 
same spike the petals of the different flowers were in this regard 
unlike each other, and sometimes of the four petals of one flower 
some were narrow and others broad. I also found petals large on 
one side of the median nerve and small on the other side. 
I obtained pure seed of some of the narrow-petalled ones and 
of some of the intermediate specimens, separately for each plant, 
and the next year (1899) had four groups flowering. Two of 
them from two typically cruciate mothers repeated this type only, 3 
each group in thirty plants. One group, also from a cruciate 
mother, produced twenty-two typical and eight atavistic indi- 
viduals, meaning thereby those with broad obcordate petals. In- 
termediate types were wanting. So it was with the fourth group, 
the children of an intermediate mother, of which only four were 
purely cruciate, the remaining fifty-eight being atavistic. 
Afterwards I continued these four strains, sowing from the 
two last named in 1900, from the first named in 1901. The ata- 
vistic mothers gave only children which were in this regard like 
them. The purely cruciate strains on the other hand continued to 
produce atavistic and intermediate types, and these last were also 
inconstant as far as I observed them. 
Taking the five generations together, I had 232 children from 
cruciate mothers, 103 from intermediate and 141 from atavistic 
parents. The first group consisted of 85 per cent. cruciate, 5 per 
cent. intermediate and 10 per cent. atavistic individuals. The 
second, of 4 per cent. cruciate, no intermediate and 96 per cent. 
atavistic specimens. The third was nearly constant, all individuals 
bearing the broad obcordate petals. 
I say nearly constant, for among those atavists I observed 
some cases of bud-variation, by which they returned to the cruci- 
ate type. This rare phenomenon occasionally presents itself with- 
out apparent reason, but it may be induced to appear oftener in 
the following manner. Around the main stem the plants produce 
a circle of smaller stems growing upwards from the axils of the 
