Improvement of a Hemi-Syncotylous Race. 479 
tion and the planting out of hemi-syncotylous seedlings 
suddenly increased the proportion of the latter and cor- 
respondingly diminished that of the syncotyls ; but only 
for a time. In the harvest of 1891 both have become 
low, and from that time onwards selection has gradually 
increased, with but slight deviations, both the number 
of hemi-syncotyls and that of syncotyls. In the last 
harvest (1896) the numbers are given for the plant with 
the largest proportion of hemi-syncotyls. The average 
number of the hemi-syncotyls of the whole crop was, 
however, 29%, and that of the syncotyls 31%. 
The result of this experiment of seven years was, 
therefore, that by the double selection of seed-parents, 
which produced the largest number of hemi-syncotyls, 
and of hemi-syncotylous individuals for seed-parents, 
we do not even approach a pure hemi-syncotylous race. 
In spite of the inevitable fluctuation of the numbers, the 
syncotyls remain in about the same proportion as the 
half type. If we think how many degrees of symphysis 
this latter embraces as opposed to the uniform group of 
the true syncotyls, the number of the latter actually found 
J ^ 
assumes even greater importance. 
In the pedigree on page 480 I have given a summary 
of the whole course of this culture. In this I have en- 
tered only the totals of hemi-syncotyls and svncotvls ; but 
~ J ~ V 
as half of these values belonged to hemi-syncotyls, the 
pedigree would not have been different if these alone 
had been entered, except of course that the percentage 
figures at the top of the table would have to be halved. 
As usual, these figures relate to means of small groups, 
which in this case embrace values between 1 and 9, 10 
and 19, 30 and 39, etc. The figures which are given for 
the several generations indicate the number of individuals 
