388 BOTANICAL GAZETTE [MAY 
Since the occurrence of a mutant gigas gives full right to the 
expectation of mutants of the type semigigas with 21 chromosomes, 
I find no difficulty in the determination of the described specimen, 
but its value is only of a confirmatory nature. 
B. TWIN HYBRIDS OF O. GRANDIFLORA 
One of the most interesting peculiarities of O. grandiflora is the 
production of twin hybrids in certain crosses, analogous to the twins 
of O. Lamarckiana. This splitting was discovered by Davis (3) 
and since confirmed by my own experiments (4). The analogy is 
very close. All those species which split O. Lamarckiana into the 
twins Jaeta and velutina provoke the same phenomenon in O. grandi- 
flora. Moreover, O. biennis Chicago, when used as a female parent 
in. the crosses, splits both of them into Jaxa and densa. In their 
characters the twins of both species resemble each other so closely 
as to be easily identified, although it is evident that they cannot 
agree in all their characters. In those of O. grandiflora the differ- 
entiating marks are not so sharp as in the twins of O. Lamarckiana, 
and it is sometimes difficult to recognize them in the first culture 
which offers them. As soon as a second generation is grown, how- 
ever, all doubts disappear. 
The species which split O. Lamarckiana into laeta and velutina 
are O. biennis, O. syrticola (muricata), and O. suaveolens when used 
as female parents; O. biennis Chicago, when its pollen is used; and 
O. Cockerelli in both reciprocal crosses. O. biennis Chicago fecun- 
dated by O. Lamarckiana produces the twins axa and densa. All 
these instances are duplicated by the analogous crosses of O. gran- 
diflora. Moreover, O. Hookeri produces twins in the reciprocal 
crosses with O. Lamarckiana and also with O. grandiflora, but the 
results of these crosses are of a more complicated nature, and there- 
fore will not be dealt with in this article. Table III gives a list of 
my crosses, together with their main results. 
In O. suaveolens X grandiflora 18 per cent of yellow specimens 
appeared; in the other crosses, however, only the twins mentioned 
appeared. If we sum up the figures for Jaeta and velutina and take 
their mean, we find 52 per cent Jaeta and 46 per cent velutina, show- 
ing that the figures do not deviate essentially from equality for the 
