A COMPARISON OF MAIZE-BREEDING METHODS 17 
character into expression was the last of a series. The stock in 
which the abnormality appeared already was homozygous recessive 
for one or more factors that produced no visible effect by themselves. 
If most recognizable characters are multiple-factor characters is 
it not probable that many of the less conspicuous characters that 
are assumed to make up the phenomenon of heterosis also are 
equally complex? Observed multiple-factor characters represent 
varying degrees of departure from normal, small departures being 
more frequent than large, and it seems a logical assumption that 
expressed variations too small to be recognized individually also are 
of a compound nature. 
Multiple-factor characters will come into expression in large num- 
bers in generations later than F,, and if they are of a nature to be 
recognized continued selfing would assist in their elimination. But 
afjthe characters must be treated en masse and the choice of,strains 
1924 1923 1922 1921 1920 I9I9 1918 
oF 
22/ 
SFXSE en Fr 
CLE. 
6S 
'FIG. 6.—Representative pedigree of progenies in the Fs seed stock of maize tested in 1924. Numbers 
indicate individual plants. Selfed generations are shown by double lines. F=27.1 
be governed by behavior in crosses the value of continued selfing 
may be questioned. If selfing were continued to complete homo- 
zygosity, most of the characters would still be represented by homo- 
zygous recessive factors in some line, and the characters would re- 
appear when the lines were combined. 
Our problem here is to free a commercial variety or a cross from 
multiple-factor characters that come into expression from time to 
time through recombination. 
Confining the discussion to dihybrid characters and with selfing 
continued until complete homozygosity is obtained, one-quarter of 
the characters will be brought into expression, and of these more 
would appear following the second generation of selfing than in any 
other generation. A corresponding quarter would be represented 
entirely by dominant allelomorphs. The remaining half would again 
become heterozygous when the lines were peeensbined. Assuming 
ce ne 3 ee eee 
