46 
BULLETIN 1164, U. S. DEPARTMENT OF AGRICULTURE. 
The only significant correlations, not of an obviously physical 
nature, between characters that may be regarded as especially impor- 
tant from an agricultural point of view are the positive correlations 
between boll length and fiber length and between boll diameter and 
lint index. In both cases the coefficient of correlation is small. The 
existence, in upland cottons, of a small degree of correlation among 
certain characters of practical importance is indicated bv results 
obtained by Hodson {23). 
Many of the characters here dealt with are doubtless correlated 
physically or physiologically (8. p. 436), the correlation involving 
either a mathematical necessity or a tendency presumably common to 
different parts of the plant, e. g., to greater or less length in several 
organs. It is, however, impossible to determine the category of the 
correlation merely by inspection of the list of character pairs. As 
Collins has pointed out (8, pp. 438, 439), comparison of the coefficients 
of correlation in the F 2 population with those obtained in the parental 
populations (if approximately homozygous) or in F x is the only 
available means of distinguishing genetic from physical or physio- 
logical correlation. 
It is regrettable that this test could not be applied satisfactorily 
in the present case because of the very small size of the parental and 
F x populations grown in 1919 for comparison with the F 2 . An F^ of 
24 individuals was grown in 1920 and the coefficients of correlations 
were computed for most of the character pairs for which they had 
been computed in F 2; but a comparison of correlations based upon 
determinations made in different seasons would be of doubtful value. 
Furthermore, the F x showed almost no variation in respect to several 
of the most interesting characters, so that many of the possible 
correlations could not be determined by the " product-moment fi 
method. For these reasons it was decided not to attempt a detailed 
comparison of the F t and F 2 correlations. 
Of the 93 presumably significant correlations in F 2 , as given in 
Table 12, 66 have signs indicating coherence, i. e., a tendency for the 
characters to be associated in the hybrid in the same manner as they 
had been received from the parents. For the remaining 27 significant 
correlations the sign of the correlation indicates the opposite tendency 
or disherence. These disherent correlations are listed in Table 13. 
Table 13. — Significant and disherent correlations in F 2 of the Iluldon-Pima cotton 
hybrids. 
[The value r-rEis that obtained by dividing the coefficient of correlation by its probable error.] 
Correlation. 
Axis length and leaf extension 
Axis length and bract dentation 
Internode length and bract dentation 
Internode number and callus color 
Internode number and bract dentation ' 
Internode number and boll apex 
Fruiting branch lengtharid bract dentation. 
Fruiting branch length and boll diameter.. 
Leaf length and boll diameter 
3..". 
5. 5 
3.7 
3.7 
5.3 
1.4 
4.2 
3.7 
4.3 
Correlation. 
Callus color and corolla index 4.5 
Bract length and calvx dentation 4.0 
Bract length and boll diameter 3.9 
Bract con nation and calyx glands .5.4 
Bract connation and lint index 3.6 
( alyx dentation and boll length 3.5 
Calyx glands and petal color 4.2 
Calvx glands and boll index 3.9 
Corolla length and stamen length 5.1 
Leaf index and leaf extension 13.2 Ftamen length and pistil length 4.1 
Leaf vein angle and leaf extension 16.7 Boll lock number and midlock furrow 4.0 
Leaf extension and fiber color i 3.9 Boll length and boll diameter 6.3 
Callus color and petiole hairiness 6.5 Fiber length and fiber color 4.s 
Callus color and calyx glands 3.9 
