408 BOTANICAL GAZETTE [MAY 
laeta and velutina. These are found, on the average, in about equal 
numbers. Our formula now becomes: 
Xlaeta =25 per cent ovata 
o per cent pure ‘ 
cee F Xvelutina=25 per cent brunnea 
O. grandiflora x La- 
marckiana= laeta) =25 per cent ovata 
x 
$e per cent ochracea Xvelutina = 25 per cent lutea 
It is easily seen that this formula opens a deeper insight into the 
whole phenomenon of twin and triple hybrids. This point will be 
discussed further at the end of this paper. 
O. Lamarckiana lata XO. grandiflora.—I made this cross twice 
in 1914 and rors and cultivated the first generation in 1915 and 
1916, respectively. In the boxes it was clear that besides the 
hybrids described for the parent species, specimens with the type 
of O. ata were present. They had the broad leaves with rounded 
tops which are so characteristic of this mutant. They were planted 
separately and developed their typical marks during the summer. 
Their stems remained low and flexible, the foliage was dense, the 
petioles short, the blades full of bubbles and paler green than in the 
mutant from Lamarckiana. The flowers and fruits were almost 
like those of this mutant, but there was plenty of pollen, and the 
artificial self-fertilization gave a good supply of seeds. I counted 
(in 1915) 18 data among 30 plants, and in the next year 24 among 
76; together 42 among 106, or about 4o per cent, a figure which 
does not differ essentially from the hereditary percentages of O. 
Lamarckiana mut. lata. The remaining plants were mostly (41) 
ovata, with some lutea and some brunnea, some dwarfs, and some 
other mutants of different types. Thus we see that this cross gave 
exactly the results that might be expected. 
In 1916 I sowed the seeds of three self-fertilized specimens of the 
lata type. The cultures showed the same splitting and the same 
types as in the first generation after the cross. I counted the /ata 
in May and found 13, 15, and rg per cent, and the ovata in May and 
August; they amounted to 35~51 per cent among 234 individuals. 
The remainder were partly /utea and brunnea and partly mutants 
of different types. Self-fertilized specimens of /ata from crosses with 
O. Hookeri, O. Cockerelli, and O. biennis Chicago have given analo- 
