COXTIXUOUS SELECTION FOR EAR TYPE IX CORN 9 
— 0.78 ^vas due to the negative correlation between the number of 
rows in a cross and the difference between the numbers in its parents. 
There is little reason to suppose that selection for number of 
kernel rows should result in decreased productiveness in every case 
unless all of the selected strains possessed something in common. 
The wide range of row numbers used in these experiments, extend- 
ing Avell to each side of the mean of the parent variety, seems to pre- 
clude number of rows, as such, from being the determining factor. 
The reduction in yield accompanying close breeding in corn is well 
recognized. There is little reason to suppose, however, that close 
breeding in the ordinary sense has taken place in any of the strains 
with the possible exception of Xo. 6. This strain may have origi- 
nated from the four 8-roAved ears recorded in strains Xos. -i and 5 
in the crop of 1918. Xot less and usualty more than 50 ears were 
used to represent each of the other strains each j^ear. It seems prob- 
able that this number of open-pollinated ears taken at random 
should include enough germinal heterogeneity to satisfy the re- 
quirements even of corn. If so. conditions similar to those follow- 
ing actual close breeding have been due to the selection. 
Much of the detailed evidence in favor of the present vievN' that 
hybrid vigor is due to the complementary action of dominant favor- 
able growth factors was obtained from inbreeding experiments with 
corn. In those experiments a homozygous condition for many 
factors was obtained automatically by self-fertilization. A decrease 
in vigor and productiveness accompanied the approach to a homo- 
zygous condition, and crossing unrelated inbred lines restored a 
heterozygous condition and yield. 
A standard variety of long standing was used in the present ex- 
periments. Strains differing markedly in certain characteristics 
were obtained from this variety by mass selection without inbreed- 
ing. Presumably each of these strains became more or less homo- 
zygous for the combination of genetic factors determining its charac- 
teristic number of kernel rows. Crossing unlike strains would pro- 
duce a heterozygous condition among the factor pairs concerned, 
and there would be more factor pairs concerned in crosses between 
strains differing more widely. Under these conditions crosses be- 
tween the more unlike strains would be heterozygous for more factor 
pairs and under the Mendelian interpretation of hybrid vigor would 
be more productive. The observed facts are in reasonable accord 
with this hypothesis, and the experiments afford further evidence in 
favor of this interpretation of hybrid vigor. 
In their practical application the experiments indicate that a de- 
crease in vigor and productiveness similar to that following inbreed- 
ing may result from too close selection for a particular kind of ear. 
Careful experiments have failed to demonstrate a marked consistent 
superiority for any specific kind of ear. Other experiments have 
shown that the yields of crosses between varieties of corn frequently 
are more productive than the average of the parents, thus indicating 
that the parent varieties are too homozygous to permit maximum 
yields. Just what constitutes too close selection is not Iniown. In 
view of the lack of evidence in favor of any particular kind of ear 
and the abundant e^ddence of the decreased yields that follow close 
breeding, however, it seems best to stay on the safe side by avoiding 
