I9i6] 



DeVRIES— DIMORPHIC MUTANTS 



2 59 



limits of ordinary chance, although all the seeds from the lo cap- 

 sules of each group have been sown. Thus it is clear that the 

 first and the last fruits of a spike and those of a side branch may 

 give the same percentage figures of specimens of the parental 

 type. Moreover, the mean value is not essentially different from 

 the means of the pedigrees, as just given, which was 33 per cent. 

 We may conclude, therefore, that the mean percentage for all my 

 annual cultures is about 30-40. 



In order to compare the influence of biennial culture upon this 

 figure, I chose three healthy and very vigorous rosettes of 1913 and 

 kept them through the winter under glass. They had been reared 

 from seeds of a biennial mutant belonging to the group of cana 

 mutants from lata, from which pedigrees no. 2 and no. 3 were 

 derived; but this special culture stayed in the rosette condition 

 during 1913. In 1914 three plants of the cana type became very 

 vigorous, reaching about double the height of the annual plants 

 and growing up to more than 2.5m.' Their stems also had twice 

 the thickness of the others, the foHage and flower spikes were very 

 dense, and the flowers much stouter. Every evening 4 or 5 flowers 

 opened on the same spike, against i or 2 in ordinary cases. The 

 number of fruits on a spike was 60-80, whereas 40 fruits, as just 

 given, is a high value for an annual plant. All of these fruits were 

 self-polHnated in little bags, and yielded 1-1.3 cc. of seeds from 

 10 fruits, whereas the annuals give ohly 0.5-0.9 cc. of seeds in 

 10 fruits. We may summarize these details by saying that my 

 biennial specimens of 1914 were about twice as vigorous as the 

 very best of all my annual cultures. 



TABLE IV 



All the seeds were sown in boxes in 1915 and the seedlings 

 counted out, without being transplanted, in the stage correspond- 

 ing to fig. I, when the differentiating marks were very sharp. The 

 three plants gave the results shown in table IV. 



