CHARACTERS IX AX UPLAND-EGYPTIAN COTTON HYBRID. 13 
MEANS OF THE PARENTAL AND HYBRID POPULATIONS. 
Comparing the means of the progenies representing the Holdon and 
Pima parents of the hybrid as to the significance of the difference. 
(Table 1) it is seen that in respect to all but 2 of the 39 characters, 
pedicel length and bract number of teeth, 13 the difference amounted 
to three and one-half or more times its probable error. The parental 
means of 1919 for 28 characters differed by 10 or more times the 
probable error. Pima gave a higher mean than Holdon for all size 
characters excepting stamen length, boll diameter, and boll lock, 
number. 
It is of interest to compare the means of the F x population grown 
in 1920 with those of the parental and F 2 populations grown in 1919, 
although such comparison, as previously noted, is not free from 
objection. Of the characters which showed differences between the 
parental means amounting to three and one-half or more times the 
probable error of the difference, 38 of which were determined in F lr 
17 are size characters, and in respect to 14 of these the Pj mean 
approached or exceeded that of the parent which gave the higher mean. 
For 9 of the remaining characters the F : mean was approximately 
intermediate, these characters having been: Leaf vein angle, leaf 
callus color, petiole hairiness, bract tooth spread, calyx dentation, 
calyx gland abundance, petal color, boll index, and fiber color. Dom~ 
inance in greater or less degree was indicated by the Fj means for the 
following 12 characters: Leaf index, leaf lobe index, leaf extension 
index, bract connation index, bract dentation index, petal spot, anther 
color, stigma index, boll apex index, midlock furrow index, lint 
index, and seed fuzziness. Of the characters showing a marked tend- 
ency to dominance, seed fuzziness is the only one for which the F, 
mean approximated that of Holdon. In all the others F t resembled 
or exceeded Pima. The greater seed fuzziness of Holdon appears to 
be almost completely dominant while the midlock furrow, a charac- 
ter appearing in the Holdon but not in the Pima parent, was com- 
pletely absent in F x . 
The F 2 mean for all characters fell between the parental means or 
did not differ significantly from the mean of one or the other parent. 
STANDARD DEVIATIONS AND COEFFICIENTS OF VARIATION. 
In comparing variation in the different populations with respect to 
measured characters, including indices derived from measurements, 
the coefficient of variation affords the most satisfactory criterion. 
The use of the coefficient does not seem to be permissible in the case 
of characters which were determined by grading; for, with standard 
deviations of equal size, a population having a high mean grade 
would give a disproportionately low coefficient of variation in com- 
parisou with a population having a low mean grade. 14 The standard 
deviation, therefore, has been used as the criterion of the variability 
of the graded characters. The standard deviations for all characters 
I3 These characters are included because the means of the parental progenies in 1917 and 1918 diiTered 
significantly, the Pima mean having been the higher for both characters in both years. For pedicel 
length the difference was 13.4 times E (probable error; in 1917 and 18.6 times E in 1918. For bract num- 
ber of teeth, the difference was 3.9 times E in 1917 and 5.S times E in 191S. The Pima population of 1919 
gave an abnormally low mean for pedicel length. 
m For example, the mean grade of seed fuzziness was 1.5 in Pima and 7.1 in Holdon, a having been 
respectively, 0.6 and 0.5. The coefficients of variation in this case would have been 40 and 
assuredly the Pima progeny was not six times more variable than the Holdon progeny in respect to 
this character. 
