406 BOTANICAL GAZETTE [MAY 
mutated condition before the moment of self-fertilization. If we 
apply this to the mutability of O. grandiflora, we may assume that 
its sexual cells are divided into two main groups, about one-half 
remaining typical, whereas the other half belong to the type 
ochracea. ‘Therefore the question arises, which of the triple hybrids 
just described are produced by the typical gametes and which by 
the mutated ones? In order to answer this question I made some 
crosses in which I used O. grandiflora mut. ochracea instead of the 
species itself. The ochracea constitutes a constant and uniform race 
and must obviously give the same hybrids as the mutated sexual 
cells of the parent species. 
O. grandiflora ochracea XO. Lamarckiana.—I made this cross in 
1914 and had two sets of seedlings in 1915, one on the bed and the’ 
other in the box. The first embraced 28 ovata and 2 lutea, the 
second 23 ovata with 3 lutea; together 56 plants. Those on the bed 
were left to flower in August and the counting was then repeated. 
The culture was one of the most beautiful in my garden and no 
doubt was possible concerning the identity of the types. Notwith- 
standing this, no brunnea and no contraria were observed. 
O. Lamarckiana XO. grandiflora ochracea.—The result was 
exactly the same as in the reciprocal cross, but the amount of /utea 
was larger (16 specimens among 69). By the end of August almost 
all the plants had flowered and were carefully compared with the 
adjoining cultures of the first and second generations of the other 
crosses. It was quite evident that no brunnea and no contraria were 
present. Especially the brunnea constitute a type so widely differ- 
ent from the others that no error could be possible. 
O. grandiflora ochracea XO. Lamarckiana nanella.—Cross of 1914; 
first generation in 1915, embracing, as in other instances, two sets, 
one in a box kept until the end of July and the other on the bed; 
observed during the whole period of flowering. There were 25 and 
27 ovata and 5 and 3 lutea, but no brunnea nor contraria. 
A résumé of these facts, confining the observations to those made 
in August at the time of flowering, is shown in table VI. 
The conclusion is evident that the gametes of O. grandiflora 
ochracea produce, in their crosses with O. Lamarckiana, only two 
types, ovata and lutea. These are exactly the same, in all respects, 
