482 Syncotylons Races. 
of 1890, selected for the syncotylous race, the following- 
percentage numbers of syncotyls : 
19 48 54 56 62 68 69 73 80 84 88 96 
If we compare this series with the pedigree on page 
470 and especially with the figures given there for 1890 
and later years, it will be seen that it is almost only the 
J * 
first two figures 19 and 48% which fall outside the group 
of values of the syncotylous offspring of the selected 
seed-parents. 
A selection of atavists as seed-parents continued 
through several generations, and a selection amongst 
these of specimens which produce the smallest percentage 
of syncotyls, will obviously reduce the mean value of 
such a race, but not to such an extent as to justify the 
expectation that a continuation of the process will lead 
to the origin of a half race. I started such an experiment 
in the summer of 1894, from the seeds of the harvest of 
1892. I selected a specimen from my syncotylous race, 
(p. 470), which had produced 92% syncotyls and only 
8% atavists, and planted out the latter only. Since that 
time I have cultivated the race continuously and on an 
isolated spot and have planted out every year only the 
seedlings from those parents which had produced the 
largest number of atavists amongst their offspring. From 
these I always selected the pure dicotylous seedlings only. 
I determined the hereditary value for each example 
and combined these in small groups in the usual way. At 
the top of the pedigree on page 483 will be found the 
mean values of these groups (5 = 1 9, 15 = 10 19 
etc.), and, vertically below these figures the corresponding 
number of individuals. The figures at the top give the 
percentage numbers of syncotyls. The pedigree can, 
