Improvement of a hemi-Syncotylous Rat\. t/ / 
adapted to such an experiment in isolation. Our expe- 
rience with the tricotylous races teaches us what to ex- 
pect; for, obviously, either a half race or an intermediate 
race must arise, if hemi-syncotyly is at all capable of 
separate existence. 
But if such an isolation cannot be effected, i. e., if the 
hemi-syncotyls are only minus variants of the syncotylous 
intermediate race, the selection of hemi-syncotyls will 
obviously do no more than maintain this latter race, and 
only modify it slightly in the mi Hits direction. We should 
then expect to obtain a strain which should not differ 
essentially from a true syncotylous intermediate race, 
except by a slight shifting of its mean value. In this 
case the number of the hemi-syncotyls will be somewhat 
increased by selection, but not, however, to the exclusion 
of the syncotyls. 
From this discussion it is clear that we can furnish 
the experimental proof that the hemi-syncotyls are minus 
variants of the syncotvls by an appropriate experiment 
in selection. For this purpose I started in 1890 with a 
lateral branch of the pedigree on page 470, by selecting 
every year only hemi-syncotylous plants as seed-parents. 
I have continued this experiment in the same way for 
seven generations, and the result was, as we shall see, 
a confirmation of the above conclusion. 
This experiment was conducted in another garden 
from that in which I cultivated the syncotylous race, but 
otherwise carried out in exactly the same way. In the 
crops the hemi-syncotyls and the true syncotyls were 
always recorded separately, so that for each seed-parent 
two values were obtained. The individuals to be planted 
out were at first chosen from two parents, but later only 
from one ; the selection being made according to the pro- 
