54 BULLETIN 1164, U. S. DEPARTMENT OF AGRICULTURE. 
Many of the characters of this cotton hybrid fulfill the require- j 
ments laid down by East as necessary to establish the fact of segre- 
gation when the frequency distribution of F 2 is unimodal and no 
definite ratios can be ascertained. These requirements may be para- 
phrased as follows: 
Uniformity of F x comparable to that of its parents. 
Absence of segregation in F r (This point was not adequately tested in the 
material here dealt with.) 
Variability much greater in F 2 than in F^ 
Recovery of grandparental types in F 2 . 
Deviation in F 2 beyond the grandparental extremes. 
Marked differences in the modes and means of F 3 progenies of individuals represent- 
ing different points on the F 2 frequency curve. 
A range of variability in the several F 3 populations, extending from that of the 
populations representing the original parents of the hybrid to that of the F 2 generation. 
Variability in generations succeeding F 2 never greater than that of the preceding 
generation. 
It may, therefore, be concluded that segregation has taken place, 
the apparent nonoccurrence of definite ratios in F, being due to the 
fact that the characters each involve several factors and to the rel- 
atively small size of the population. 
Several of the characters determined by measurement or grading 
showed variation, in the second generation, far beyond the combined 
ranges of the parental populations. Numerous other characters, not 
expressed in either parent variety, appeared in F 2 . Some of the more 
striking of these variations are absent or of very rare occurrence in 
any cultivated variety of cotton known to the writer. Abnormal or 
even pathological characters, such as persistence of tomosis in the 
adult state, partial chlorosis, extrusion of the pistil in the bud, mal- 
formation of the corolla, and abortion of the anthers were exhibited 
by a few of the second-generation plants. 
Further evidence of the wideness of this cross between upland and 
Egyptian cottons is the frequency of sterility. Of the F 2 individuals, 
7 per cent were completely sterile, and many more showed a very low 
degree of fertility. Many of the F 3 progenies, including one of which 
the F 2 parent had been exceptionally fertile, showed a low average 
fertility and comprised numerous sterile individuals. 
The coefficients of correlation were determined for 703 pairs of 
characters in the second generation of the Holdon-Pima hybrid. Of 
the coefficients, 93 were probably significant. Many of the correla- 
tions are undoubtedly physical or physiological or are of a secondary 
nature, resulting from the mutual correlation of several characters. 
Of the 93 pairs of characters which showed a presumably signifi- 
cant correlation in F 2 , the coefficient having been at least three and 
one-half times its probable error, the sign in 66 cases indicated coher- 
ence or expression of the characters in the hybrid in the same relation 
as they were received from the parents. Omitting cases of more or 
less obvious physical or physiological correlation, the number of signifi- 
cant coherences is reduced to 22. In 27 of the significant correlations 
the opposite tendency, disherence, was shown. Few of these disher- 
ent correlations obviously were of a physical or physiological nature. 
The occurrence of disherent correlations, also noted by Collins and 
Kempton in their study of a hybrid between teosinte and maize, 
seems contrary to any generally recognized principle of heredity. 
No adequate explanation of the phenomenon has suggested itself. 
