402 De Vries. — On Biastrepsis in its 
within the dense rosette ; and at the beginning of November 
I had rather more than half of the plants removed as being 
undoubtedly normal and atavistic. Towards the end of May, 
1 892, when the shoots were being vigorously pushed up, the 
plants were finally gone over and counted. Including the 
figures for the previous November, I obtained the following 
result for the whole sowing : — 
Twisted stems 37 = 34% 
Stems with } phyllotaxis 12 = 11% 
Normal, (atavistic) 58 = 55% 
Total 107 
It is of importance to remark that the number of twisted 
individuals had increased, with the larger space, not merely 
relatively in percentage but absolutely per square metre. 
In the third generation there were fifty plants to the square 
metre of which 4 per cent. (1-7 per cent.), or about 1-4 plants, 
were twisted : in the fourth generation I had four square 
metres with thirty-seven twisted individuals, that is, about 
nine to the square metre. 
I selected from this bed the seven best plants as seed-bearers, 
all of them having local biastrepsis in some of the branches, 
and I isolated them before flowering. 
The fifth generation, 1893-4, gave less favourable results: 
it yielded only 20 per cent, of twisted main stems. In this 
case I had not, as previously done, sown the seed on the 
beds, but in pans standing in the greenhouse of my labora- 
tory. This method has since proved itself to be the more 
convenient and certain, and it was adopted with both the 
succeeding generations. 
The seed harvested in September, 1892, was sown in the 
middle of March, 1893. About the middle of April the best 
seedlings were transplanted singly into 10 cm. pots, containing 
well-manured loam ; and about the middle of May they were 
planted out in the beds at about the same distance from 
each other as in the previous experiment (twenty-two plants 
to the square metre). In the next year all the shoots 
