1916] DEV RIES—DIMORPHIC MUTANTS 257 
the percentage figures come so much nearer to equality of the two 
types than in the other cases is probably owing to this striking vigor 
of the race. The means are 25 per cent cana for no. 1 and no. 2, but 
53 per cent cana for no. 3. The proportion of mutants among the 
seedlings of the cana individuals is 18 in 350, or about 5 per cent. 
O. cana from lata nos. 4 and 5.—As previously mentioned, the 
progeny of two further mutants of the same origin have been 
studied. The offspring of one of them embraced only 15 indi- 
viduals, of which 13 have flowered. There were 3 cana, 2 mutants 
(one oblonga and one of the same new type as in no. 2), the 10 
remaining plants being externally like O. Lamarckiana. The 
second original mutant yielded only 11 offspring, among which 
7 were cana, one Lamarckiana, and 3 oblonga. Although these cul- 
tures do not justify the calculation of percentage figures, they 
evidently support the conclusions drawn from the three former 
ones, and argue for the conception that this form of splitting is 
typical for O. cana. 
INFLUENCE OF CULTURE ON PERCENTAGE FIGURES.—I have 
shown’ that the percentage figures for the splitting of O. scintillans 
in the succeeding generations may differ for different families. 
Sometimes it is only 15 per cent, more often it varies between 34 
and 39 per cent, and in rare cases it reaches 69-93 per cent. Sub- 
sequent experiences have suggested the idea that these differences 
are due mainly to outward conditions or to the method of cultivation, 
and that favorable influences must increase the percentage of indi- 
viduals with the type of scintillans and diminish the percentage of 
Lamarckiana-like specimens. 
The self-fertilized seeds of the cana individuals previously 
mentioned have given the following percentages of specimens 
with the cana type: 15 and 16 per cent, 24-34 and 35 per cent, and 
46 and 60 per cent, the two latter being found in a culture which 
excelled the others in vigor. Evidently these figures run parallel 
to those of scintillans and the variability must have the same cause 
in both cases. 
In order to ascertain the nature of this cause I have tried to 
answer two questions, namely: (1) are the percentage figures 
7 The mutation theory. Chicago. 1909. pp. 388-391. 
