52 BULLETIN 1164, U. S. DEPARTMENT OF AGRICULTURE. 
pollination with upland cotton merely because definite upland char- 
acters are not observed among the Egyptian plants. The occurrence 
of any peculiarity or abnormality must be regarded as ground for 
the suspicion that such cross-pollination has occurred and all such 
variants should be removed when seed-increase fields are rogued. 
SUMMARY. 
Upland and Egyptian cottons, although both types are possibly of 
hybrid origin, are comparable to very distinct botanical species in 
the nature and magnitude of the differences between them. 
A genetic study of the kind and degree of segregation in an upland- 
Egyptian hybrid was undertaken for the purpose of obtaining 
statistical data as to the nature and behavior of the variants to be 
expected in fields of either type when exposed to natural cross- 
pollination by the other. Such information should be useful in 
determining whether cross-pollination has occurred and as a guide in 
the work of roguing seed-increase fields. 
The Holdon variety of upland cotton, a representative of the 
Texas big-boiled group, was crossed with the Pima variety of 
Egyptian cotton, and the resulting hybrid was studied in the first, 
second, and third generations. 
Holdon and Pima differ in respect to a great number of characters, 
many of which distinguish the two types of cotton in general. By 
measurement or grading 39 characters were determined in the 
parental, F x and F 2 populations, and showed significant differences 
between the parental populations. 
The first generation of the hybrid showed heterosis or intensification 
in respect to most of the size characters, the Y 1 mean having 
approximated or surpassed that of the parent which gave the higher 
value. Of the remaining 21 characters determined in this population, 
the ¥ t mean was approximately intermediate in 9, while a tendency 
to dominance was shown in 12 characters, the F x mean approaching 
or in some cases surpassing that of one or the other parental 
population. 
The mean of F 2 for all characters in respect to which the parental 
populations differed significantly either fell between the parental 
means or did not differ significantly from that of one or the other 
parental population. 
Comparison of the coefficients of variation for measured characters 
and of the standard deviations for graded characters showed Holdon 
to be decidedly the more variable of the parental populations. 
In a large majority of the characters F x was less variable than the 
parental population which gave the higher coefficient of variation or 
standard deviation for the character in question. F 2 was more 
variable than the respective more variable parental population in 
most of the characters and was significant!}' more variable than F x 
in all but one character. 
The relative diversity in F 2 , as compared with that of the parental 
and first-generation populations, is not fully expressed by comparison 
of the coefficients of variation or standard deviations for the charac- 
ters determined. Numerous extraparental characters or expressions 
of characters appeared in F 2 . 
The frequency distributions of F 2 extend to or beyond the 
extremes of both parental populations in 33 of the 39 characters. 
