CHARACTERS IX AN UPLAND-EGYPTIAN COTTON HYBRID. 11 
Lint index. Seed cotton from five bolls of each plant waa weighed and ginned, 
the total weight of the seeds was determined and the weight of fiber was ascertained 
by subtracting the weight of the seeds from the original weight of the seed cotton. 
The total number of seeds was then determined and the lint index was computed as 
weight of fiber -s- number of seeds X 100. A high lint index indicates that the fiber is 
abundant. Holdon gave a higher mean for lint index than Pima. 
. graded. Types were selected representing the range of variation in 
this character of the parental and F._, populations, nine grades having been distin- 
guished. The sample of seeds of each plant obtained by ginning the five- boll sample 
of seed cotton was compared with the types and the nearest equivalent was take 
the grade of seed fuzziness for the plant in question. The grading was based upon 
the relative area of the seed coat covered with short hairs or fuzz, although variation 
in the length and color of the fuzz made the mat.ching somewhat difficult. Of the 
parental populations. Holdon gave the higher mean grade for seed fuzzii 
STATISTICAL CONSTANTS OF THE PARENTAL AND OF THE F, AND F 2 
HYBRID POPULATIONS. 
Table 1 gives the means, standard deviations, and (for the measured 
characters) the coefficients of variation of 39 characters which were 
determined on the parental and F, hybrid progenies in 1919 and on 
the F x hybrid progeny in 1920. As was stated on a preceding page, 
three F 2 Holdon X Pima populations were grown in 1919. each hav- 
ing been the progeny of a single ¥ 1 plant of 1918. The F ? parents 
were not selected as representing the extremes of variation in the ¥ x 
population of 1918 but may be regarded as constituting a random 
sample. Comparison of the frequency distributions and means of the 
separate F 2 progenies for 11 characters in respect to which the parent 
varieties differed very significantly showed no important differences 
and indicated that there had been no pronounced segregation in the F x . 
For 7 of the 11 characters the means of the three F 2 progenies showed 
no significant differences, while for 4 characters the highest and the 
lowest mean differed by from three to five times the probable error 
of the difference. The modes of the three progenies for none of the 
characters differed by more than one grade in a total of seven to nine 
grades. It was decided therefore to treat the F 2 as one array. The 
total number of plants in this F 2 population was 215, but all of the 
characters were measured or graded on only ISO individuals. As 
only the plants on which all characters had been determined were 
used in computing the coefficients of correlation it was thought best 
to compute the means and other statistical constants from the same 
ISO plants. 12 
It is regrettable that the Y 1 progeny of the hybrid grown in 1919 
was not of adequate size for comparison with the F 2 population of 
that year. The F x grown in 1920 was perforce used for this purp 
but the validity of the comparison is lessened not only because of the 
probable effect of different weather in the two seasons but also be- 
cause the Y x progeny of 1920 was situated on better soil than that 
upon which the Pima and Holdon parental progenies and the greater 
part of the F 2 population in 1919 were grown. This makes the com- 
parison unsatisfactory, especially in regard to size characters. 
!2 In order to ascertain whether the 180 nlants on which all character- were measure 1 <ir g 
stituted a fair sample of the whole population, the Frequency distributions and means [or 
which had been dete-mined on all or nearly all F- : in li\ id lals n-213 to 31 '■ . were i omp u • I for the two 
populations. The differences in the frequency distributions were negligible, an i in nocas< 1 i . 1 the means 
of the 180 individuals differ from the means of the larger number byanamounl greater thai 
able error of the difference. 
