irregular increase in mean length and a marked and pro- 
gressive increase in average width. Since kernel width 
has increased proportionately more than length, it fur- 
nishes the most useful dimension for detailed study. 
Kven more striking than the increase in average kernel 
width is the progressive increase in the maximum width 
(width of widest kernel) from stratum to stratum. This 
increase is virtually linear, but the linearity is probably 
spurious, since it is scarcely to be believed that the sev- 
eral strata represent identical periods of time, or that ge- 
netic changes of a given order always produce the same 
change in absolute dimensions whether acting upon small 
or large kernels. Furthermore, part of the increase in 
maximum size is nothing more than a consequence of 
increasing the size of the sample. This factor, however, 
does not account for all of the increase in maximum size, 
since there is a substantial increase from Stratum IV to 
Stratum VI, although the samples from the two strata 
are approximately equal in size. Furthermore, the cobs, 
even though they lack kernels, furnish indirect evidence 
of a progressive increase from stratum to stratum in the 
mean and maximum. width of the kernel. 
Although there is a progressive increase in kernel 
width from stratum to stratum, the smaller-kerneled 
types with which the series began did not disappear from 
the scene when the new types with larger kernels came 
into existence. What has actually occurred is that there 
has been an appreciable increase in the range of variation 
with respect to kernel size. This is illustrated by Plate 
XXIV, Fig. A. 
The marked increase in variation in kernel size in a 
relatively short period of time provides some indication 
of the way in which variability accumulates in a man- 
made environment in which the pressure of natural selec- 
tion has been reduced or released. Here, perhaps, is the 
[ 230 ] 
