44 BULLETIN 1489, U. S. DEPABTME5TT OF AGEICULTLEE 
allelomorph dominant to dwarfness. Suck crosses consequently 
would be unusually vigorous and productive and would tend to in- 
dicate a negative relation between the yields of the parents and the 
crosses. 
It is not improbable that lines homozygous for very deleterious 
characters frequently may carry a superior assortment of other fac- 
tors. It has been shown that a stock segregating for dwarf plants 
produced a marked deficiency of dwarfs under field conditions and a 
smaller deficiency in the greenhouse where conditions for emergence 
were more favorable (23, Ifi). This may mean that many of the 
plants surviving in the field were those with the better assortments 
of other factors. That is, dwarfness or other deleterious characters 
may tend to eliminate individuals with the less desirable complements 
of general growth factors by increasing the difficulty of survival. If 
such a condition exists it might even be utilized as a means of selec- 
tion were it not for the practical necessity for producing a good 
quality of seed at reasonably low cost. 
UTILIZATION OF SELFED LINES 
The first suggestion of selection within selfed lines as a method 
of corn breeding resulted from investigations planned primarily to 
study various genetic principles. In spite of the inferiority of the 
selfed lines obtained, it was noted that the crosses between these 
lines were vigorous and that some yielded more than the original 
variety. As a consequence it was suggested that such crosses could 
be utilized commercially (69). The greatest objections to this were 
(1) that the selfed lines were unproductive, making the cost of the 
seed high, and (2) that the seed produced on the selfed plants was 
of poor quality so that it might not germinate well under unfavor- 
able conditions. It then was suggested that crosses between crosses, 
or double crosses, as they were called, be used commercially (39, JfO) . 
For example, letting A, B, C, and D designate four unrelated selfed 
lines, crosses would be made between A and B and between C and D. 
These two single crosses then would be crossed and the first genera- 
tion of the seed from the resulting double cross, (A X B) X (C X 
D). would be utilized commercially. As double-crossed seed would 
be produced on vigorous hybrid plants, its production would be 
cheaper and its quality better. The method of crossing is the same 
as that already considered under varietal hybridization (p. 33). The 
system for maintaining a supply of pure seed of the four selfed 
lines and producing single and double crosses is indicated diagram- 
matically in Figure 8. 
The results of later experiments under more intensive selection 
indicate that the objections which double crosses were suggested to 
meet no longer exist. Some of the selfed lines now being obtained 
are productive enough so that their use in single crosses for com- 
mercial seed production will be entirely practical. The idea long 
has been prevalent, however, that a mixture of seed of two varieties 
of slightly unlike season would be more productive than the average 
of the two varieties as grown separately. The theory is that the 
variation in blossoming and earing would cause the plants of the 
two varieties to make their maximum demands on the soil at different 
