404 Tricotylous Races. 
gether, and planted out under the same average treat- 
ment on the same bed and on the same day. About 140 
plants set seed abundantly. 
On the average, however, this culture was not better 
than those of the previous years, for it only yielded a 
ratio of 4.5%; but the range of variability was much 
greater. Eight plants occurred, the hereditary coefficients 
of which exceeded all previous ones. Of these, six were 
14 to 17%, one 21%, and one 25%. Here the possibility 
of a sudden advance seemed to open up. 
Before I give the whole series of figures, I wish to 
make one further observation. If in the year 1897 I 
had not cultivated 450 plants, but only, let us say, one- 
third, I would have limited myself to the offspring of 
the grandparent with 5.5%, although the value is only 
apparently greater than the other, because the difference 
lies within the limits of observational error. I would 
then have obtained precisely the same result with only 
one-third of the labor. In other words, neither the se- 
lection of tricotyls as seed-parents, nor the attention paid 
to the hereditary values, although this excludes the poor- 
est tricotyls in spite of the latitude of possible errors, 
can make the experiment independent of chance. Noth- 
ing less than carrying out the experiments on a much 
larger scale can effect this. But the results of the two 
following generations will show that even in the present 
very favorable case, no real or permanent advance was 
effected. 
The values obtained, in the spring of 1898, for the 
140 offspring of the best parent of 1895, which itself 
had a value of 5.5%, are distributed as follows: (P 
refers to the figures in percentages and A to the cor re- 
