26 
BULLETIN 1164, TJ. S. DEPARTMENT OF AGRICULTURE. 
or more of the plants were graded 0. It has been pointed out, how- 
ever, that this character varies on the individual plant, and it is also 
more or less influenced by soil variations. It is not unlikely that if 
a larger number of flowers had been examined on each of the F 3 plants 
graded the "zero' 1 individuals in some of these progenies would 
have been found to show a faint development of the spot in some of 
the flowers. 19 
The fact that notwithstanding this uncertainty in regard to the 
grading of some individuals a close approximation to a 3 to 1 ratio 
for presence and complete absence of the petal spot was obtained in 
F 2 makes it likely that such errors as may have occurred in the grad- 
ing offset each other, the population having been a fairly large one 
(n = 215). That the grading was, on the whole, satisfactory is indi- 
cated by the high parent-offspring correlation between the grades for 
petal spot of 22 F, individuals and the mean grades of their F 3 prog- 
enies, the coefficient of correlation having been 0.86 ±0.037. 
Taking as one population the F 3 progenies from F 2 individuals 
which had shown some degree of development of the petal spot, but 
• ^ j. + s- * r * * excluding progenies 44 and 29, 
which were homozygous for pres- 
ence of the spot, and progenies 19 
and 171, which were graded as 
entirely spotless, the total numbers 
of spotless and of spotted indi- 
viduals were, respectively, 88 and 
180. This is far from being a 1 to 
3 ratio, the deviation from the ex- 
pected percentage of spotless hav- 
ing been 7.8 ±1.9 per cent; but in 
view of the likelihood that the 
number of spotless individuals 
Fig. 41.— Seed fuzziness: Frequency distribution of wmild hflvp hppn fnnnrl +n \\& 
Holdon X Pima F s cotton plants, by grades, as %V OU1Q fj ave Deen J 0UnC \ t0 De 
used in computing the correlations. Theordinate smaller II larger numbers 01 110 Wei's 
^f S £Si t ipffi^ plMfaa8aperoen1 ' had been examined the supposition 
that complete absence of the spot, 
as compared with its presence, is a simple recessive character can 
not be regarded as disproved. 
That the degree of development of the petal spot, when present, is 
conditioned by modifying factors seems to be satisfactorily established 
by the evidence from both F 2 and F 3 . 
S^j 
/ 
y 
f> 
•/J 
ANTHER COLOR. 
The frequency distribution for anther color in F 2 (fig. 29) is inter- 
esting because of the failure to recover the very pale anther color, 
or rather pollen color, of the Holdon parent. The golden yellow 
color of the Pima parent was partly dominant, as is shown by the 
position on the F 2 frequency curve of the ¥ t mean and by the fact 
that the mode in F 2 was only one grade lower than the Pima mean. 
It is possible but not probable that the apparent bimodality of this 
!9 Balls (2, p. 367) met with the same difficulty in classifying plants in regard to this character and 
attributes an excess of " spotless " in F2 to the fact that " it was common to record a plant once or twice 
as spotless and then find a flower appearing which bore a faint spot on one or more petals." 
