46 BULLETIN 1489, U. S. DEPARTMENT OF AGRICULTURE 
so that a degree of hybridity will be maintained by random pollina- 
tion sufficient to insure productiveness. The ordinary commercial 
variety of corn in reality is a mixture of crosses or hybrids. Many 
of the better component elements of these crosses are isolated during 
the period of selection and self-fertilization. After the value of the 
selfed lines has been determined in hybrid combination, those lines 
can be selected which give productive crosses in all combinations. A 
synthetic variety consisting of a mixture of crosses among these lines 
grown under conditions of random pollination, as in ordinary corn 
culture, will be maintained in a high degree of hybridity and at the 
same time should yield more than the ordinary variety. Theoreti- 
cally it should be possible to obtain material increases in yield from 
such synthetic varieties. Theoretically also, such varieties should 
not be as productive as the Fi generation of the best component cross. 
Very little actual evidence is available as yet as to the possibilities of 
this method. The synthetic variety possesses such practical advan- 
tages in the way of seed production that it might well have an 
important place in corn culture, even though the yields obtained were 
somewhat less than those from single or double crosses. 
The elimination or the fixation of a single characteristic or condi- 
tion sometimes may be of much importance. Susceptibility to a dis- 
ease, the development of husks which afford insufficient protection 
against insect attack, or a tendency to tiller profusely may be a seri- 
ous disadvantage from the standpoint of the quantity or quality of 
the crop or of its handling. If the undesirable condition is rela- 
tively simple genetically, and particularly if it is due to recessive 
factors, application of the principles involved in the production of 
synthetic varieties offers much of value. It was, in fact, in connection 
with such a problem, the production of a high-oil corn, that the use 
of this principle first was suggested {21). Thus, a large number of 
unrelated lines may be selected, all of which breed true for the desir- 
able condition of the one character to be modified. These lines then 
are intercrossed and the mixture of crosses propagated as an ordinary 
variety under conditions of random pollination. By using enough 
lines and selecting only for the one character, the recombined lines 
will differ little, if any, from the original variety except with respect 
to that one character. The yield or quality will be improved, how- 
ever, in so far as the character modified previously had been 
unfavorable. 
The operation of this method and the reason that it is particularly 
applicable in eliminating recessive characters (or in fixing dominant 
characters) is well illustrated in the hypothetical example of selection 
for purple and nonpurple aleurone already given (p. 42). Thus, if 
one selfed ear with all kernels purple were obtained from each of a 
considerable number of lines, these ears could be shelled together. 
The mixed seed propagated as a variety would produce only purple 
kernels. If enough unrelated all-purple ears were not obtained the 
first season, pollination of plants from purple seeds could be prac- 
ticed another year, or until enough different lines were represented. 
The resulting purple variety should be essentially like the parent 
variety, except for kernel color, but would have any advantages that 
came from purple kernels. 
