Tricot y Ions Races Do Xot Arise by Selection. 401 
set as much seed as those grown in groups. The seeds 
do not fall out and as soon as the desired quantity is ripe 
the whole raceme is cut off and rubbed between the hands. 
One cubic centimeter per plant, and often less, is suffi- 
cient for next year's seed ; as a rule I obtained about 
4 to 5 cubic centimeters from each individual. 
After the transplantation of the seedlings the length 
of the cotyledons, and of their stalks especially, increases 
considerably. In this way it may happen that specimens 
which, at their first examination, appeared to be pure 
tricotyls are found to have two of their cotyledons united 
at the base, and therefore to be deeply cleft hemi-tricotyls. 
I have as a rule removed such specimens when I discov- 
ered them, and not cultivated them further, although this 
is not actually necessary. 
In the summer of 1892 I had 11 tricotylous speci- 
mens which flowered in a group as far removed as pos- 
sible from the hemi-tricotylous culture already men- 
tioned. The harvest gave a maximum of 11.9% tricotyls, 
but on the average only 1 to 6.5% with a mean of 3.5%. 
In 1893 I planted out the purely tricotylous seedlings only 
of a seed-parent with a hereditary value of 6.3%. I saved 
the seeds of 15 plants, examined between 500 and 1000 
seedlings per seed-parent, and obtained values which fluc- 
tuated between 2.6% and 7.4%; mean 47% (1894). 
I selected the parent with 7.4% for the continuation of 
the race and only planted out the best tricotylous speci- 
mens from amongst its offspring. I had 44 tricotyls 
amongst 1000 seedlings to choose from; besides these 
there were 31 hemi-tricotyls and one tetracotyl. I re- 
peated the evaluation of the ratio with 4000 seedlings and 
found 7.2%, that is to say, about the same value. 
At the end of the summer of 1894 my culture con- 
