472 Syncotylous Races. 
used the plants with a mean value (say 50-55%) for the 
continuation of the culture. But then I would have ob- 
tained a non-selected strain, and the figures for the first 
two generations following on the original plant, clearly 
indicate that this mean would have been about 50-55%, 
which is also, as we know, the mean value for tricotylous 
intermediate races. 
The figures given in the pedigree do not present a 
complete picture of the whole improved race, for in each 
year the hereditary value of the best offspring alone was 
determined. Dicotylous seedlings and those with a low 
degree of fusion in the seed-leaves were excluded from 
further cultivation, as were also weak plants. If I had 
not applied this selection, the mean values would ob- 
viously have turned out somewhat lower; but the differ- 
ence would not have been a very considerable one, as the 
next two sections will show. 
The chance of obtaining a pure syncotylous progeny, 
i. e., a crop without dicotylous seedlings, may appear 
to be very great in this experiment. In 13 individual 
crops 96% and over was reached. But appearances 
are deceptive. Only once did I have a perfectly pure 
crop (100%) and that even at the outset of my experi- 
ment, in 1890; but this plant had produced only 105 fer- 
tile seeds ; and of course we must acknowledge the possi- 
bility of some stray atavistic seedlings occurring amongst 
them if the harvest had been larger. In other words, 
selection leads the race as close as possible to the highest 
degree of purity, without, however, enabling it to reach 
it. Moreover, the table shows at a glance, that the pro- 
geny of the plant with 100% would probably have fallen 
back from this high value, in the case of many of its 
offspring at any rate. 
