42 
Research Hull* tin Yo. 2 
Disregarding the progeny of ear (60-3x54) 5-2CO because there 
were only 29 individuals, there is a range of variability from 
12.156 ± .t;s per cent to 19.61 ± .98 per cent. There is no question 
then but that the progenies of different F 2 ears show great devia- 
tions in variability of ear length, as should be expected if segre- 
gation of multiple factors occurs. The most interesting fact, 
however, is the differences in mean length of ear shown by the 
various frequency distributions of F 3 . The mean length of the 
progeny of ear (60-3x54) 6-5EO is only 9.7 ± .11 cm. They com- 
pare very favorably with Tom Thumb pop, the smaller parent, 
tho they are sweet segregates breeding true. Perhaps here again 
extremely small segregates are less common than they should be 
theoretically on account of the slight increase in size due to 
heterozygosis. Perhaps also the combination of genes represent- 
ing other character complexes that would be most favorable to 
small size was not obtained. 
On the other hand, the progeny of ear (60-3x54) 1-7ES average 
15.9 ± .15 cm. They have nearly the same length as the long- 
eared parent, Black Mexican. They average about twelve rows 
per ear, and are starchy segregates, however, so perhaps they are 
not as long as they would be with the same zygotic structure for 
ear size, if they were typically 8-rowed and wrinkled like the 
Klack Mexican. At least there is good reason for suspecting a 
physiological correlation between long ears and few rows per ear. 
Cross (60-5x54), Table 14, does not show such uniform in- 
crease in variability in F 2 over that shown by F x as does the 
cross just described. There is an increased range of variability 
and an increase in the calculated coefficient of variability in each 
of the five F 2 families grown, but in three cases out of five this 
increase is not significant. If one adds together the entries in 
all of the five families, he obtains a frequency distribution with 
a coefficient of variability of 15.67 ± .30 per cent. This co- 
efficient is considerably larger than that of F w 12.48 ± .72 per 
cent, but the procedure is questionable. 
The nine F 8 families grown show great range in variability. 
The smallest coefficient is 10.46 ± .60 per cent and the largest 
coefficient 21.68 ± .88 per cent. It is unfortunate that no ex- 
tremely short ears and but one extremely long ear were selfed. 
but this should not be regarded as the fault of the experimenter. 
When the ear is bagged for selfing, one is unable to tell whether 
it will be large or small. Over 300 ears were selfed in the F 2 
generation and of these about 275 matured, but among them were 
found but one normal ear of 7 cm. and but one ear of 18 cm. 
Mice destroyed the short ear, and recourse to ears 9 cm. and 
10 cm. in length was necessary. The average of the progeny of 
