CORK BREEDING 47 
The production of a synthetic variety with nonpurple aleurone, 
the recessive condition, would be more difficult. Before combining 
the lines having nonpurple kernels it would be necessary to determine 
experimentally that they did not produce purple kernels in any com- 
bination. Once determined, however, the proper lines could be com- 
bined into a synthetic variety which would breed true for colorless 
aleurone. If enough independent lines were combined the resulting 
variety would be approximately the same as the parent variety except 
for the absence of kernels with purple aleurone. 
A relatively simple and unimportant character has been chosen 
for an example. The basic principles apply equally to any char- 
acter, no matter how complex, provided its mode of inheritance is 
known. It is this certainty of selection within selfed lines that 
makes it so important in corn breeding. Even in the absence of any 
direct evidence of increased yields following selection within selfed 
lines, the certainty with which characters can be controlled by this 
method is sufficient warrant for its use. Fortunately, however, more 
direct evidence of the success of selection within selfed lines is accu- 
mulating rapidly. Some of this evidence will be considered in the 
paragraphs that follow. 
RESULTS OF SELECTION WITHIN" SELFED LINES 
Most of the data so far available on the results of selection within 
selfed lines are on the yields of the selfed lines themselves or on 
crosses or double crosses between lines. The yields obtained from 
the selfed lines are of little interest. It is enough to say that practi- 
cally all have been markedly inferior to the parent variety and that 
no selfed line has been reported which represented an improvement 
over the parent variety. The proportion of superior lines would 
be expected to be very small, however, so that the failure to obtain 
any up to the present is of little importance. The later experiments 
are encouraging in that much better selfed lines are being obtained 
under the more extensive selection now being practiced. 
It is recognized that not all crosses will be superior, and that the 
problem of the corn breeder using this method is to find the few 
which are best. From this point of view the yields of inferior 
crosses are of as little interest as those of selfed lines and will receive 
scant attention here. It seems desirable to point out, nevertheless, 
that some cross has been more productive than the parent variety in 
every reported experiment in which a number of crosses between 
selfed lines have been compared. 
The first reported yields of crosses between selfed lines in compari- 
son with a comparable noninbred stock appear to be those presented 
by Shull in 1908 (68). The acre yields of reciprocal crosses were 
74.4 and 78.6 bushels in comparison with 75 bushels from the parent 
f varieties. In 1909, Shull obtained acre yields of 98.4 and 96.1 bushels 
from the reciprocal crosses of the best hybrid combination and only 
88.1 bushels from the best noninbred strain (69). In 1910, 7 crosses 
averaged 68.07 bushels per acre as compared with an average yield 
of 61.52 bushels from 10 noninbred strains (70). It was on the basis 
of the experiments in connection' with which these yields were ob- 
