Spirally Twisted Races. 545 
as has been observed elsewhere in intermediate races, 
and especially in the tricotyls (see page 439). 
Secondly, I call the attention of the reader to the 
remark which I made in the previous chapter ( 16, p. 
511) with regard to selection. Even as in the case of 
fasciation, a double selection, by the visible characters 
of the seed-parents and by their hereditary value, cannot 
here be carried out under the ordinary conditions of the 
experiments. The most valuable character to select by, 
the hereditary coefficient, fails, and in consequence of this 
the improvement of the race, which in the case of the 
tricotyls so soon resulted in a proportion of from 70- 
90%, obviously cannot be carried to its full height here. 
Nothing less than a lucky chance or the conducting of 
the experiment on a much larger scale can bring this 
about. In the eighth generation I compared the heredi- 
tary coefficients of ten parents. They ranged between 
10 and 55% ; but as only from 100-140 offspring of each 
parent could be compared, the percentage numbers did not 
seem to me to be of much importance from the point of 
view of selection. In the 9th generation I saved the 
seeds from over 100 very finely twisted individuals, in 
the hope of still being able to carry out an experiment in 
selection by hereditary coefficients. 
I shall now give a detailed description of the whole 
experiment. 1 The starting point of my race was formed 
by two individuals with a twisted main stem which flow- 
ered in 1885 in a culture sown in 1884 in my garden. 
All the remaining plants were destroyed before they 
flowered. 
In 1886 I obtained a second generation from their 
seeds. As I have already mentioned, I was not at that 
1 Annals of Botany, Vol. XIII, No. LI, Sept. 1899, P- 401. 
