CHARACTERS IX AX UPLAXD-EGYPTIAX COTTOX HYBRID. 49 
Egyptian plants are sterile or otherwise malformed, but many are fertile and appar- 
ently normal. 
Specific instances of coherence of characters in Egyptian-upland 
hybrids are mentioned by Cook, although no determinations of the 
coefficient of correlation were made. One of these is pale color of the 
petals with an open cuplike corolla, both being upland characters, 
and the converse, or Egyptian combination (11, p. 17; 15, pp. 42, 43). 
In the Holdon-Pima hybrid, however, there was no significant correla- 
tion between corolla index (width as a percentage of the length) and 
petal color. Coherence of the three upland characters— short and 
wide corolla, pale petals, and long filaments— is mentioned by Cook, 
McLachlan, and Meade (15, p. 43) , but in the Holdon-Pima hybrid 
stamen length was not significantly correlated with corolla index, 
and for the°correlation stamen length with petal color the coefficient 
was positive instead of negative and was only 0.168 ± 0.049. A further 
instance of coherence in the hybrid of two upland characters, pale color 
of both petals and pollen, is cited by Cook, McLachlan, and Meade 
(15, p. 44), but there was no significant correlation of petal color and 
anther (pollen) color in the Holdon-Pima hybrid. 
Cook (11, p. 14) directs attention to— 
A very general correlation between the shape of the boll and the length of the lint 
running through all the varieties of cotton that have been studied with this idea in 
mind, plants with more rounded bolls having shorter lint. 
In a later paper (12, p. 30) the statement appears: 
The general association of longer lint with more pointed bolls in any particular type 
of cotton may be connected with the other general fact that the long-lmted types of 
cotton have more gradually tapering bolls than short-linted types of cotton. 
The Pima parent of the Holdon-Pima hybrid has more pointed 
bolls, as expressed by the lower boll apex index, and longer fiber than 
the Holdon parent f hence, there should be in the hybrid a negative 
correlation between these characters if the coherence pointed out by 
Cook obtains in this case. The coefficient of correlation in F 2 , although 
low, was in fact negative and was apparently significant (r- 0.172 
_0.049). 
It is of interest to consider to what extent the whole series of char- 
acters determined on the Holdon-Pima hybrid is connected by mutual 
correlation (9, p. 34). Coefficients of correlation were determined m 
F 2 for all possible combinations of 38 characters with respect to which 
the parental means differed significantly (difference of three and one- 
half or more times its probable error). Thirty-seven of these char- 
acters are involved in a continuous chain of correlations having a value 
for r of three and one-half or more times the probable error. The 
onlv character which appeared to be entirely independent was seed 
fuzziness, which was significantly correlated in F 2 with no other 
character. 
It seems at first glance remarkable that nearly all of the characters 
determined in this hybrid should be intercorrelated, in view of the 
fact that the number of the chromosomes in Gossypium is large. 
Cannon (7, p. 167) determined the number in the haploid stage in the 
first generation of a sea-island-upland hybrid as 28, while Balls ;'. 
p. 12) working presumably with Egyptian cotton, gives the haploid 
number as 20. It is to be considered, however, that the coefficients 
of correlation in the Holdon-Pima, hybrid are mostly too small to 
