1910] HARRIS—FERTILITY IN CERCIS 12I 
shown in the form of fractions, in which the number of ovules is 
given as the denominator and the number of seeds developing as 
the numerator. For each of the twenty-four seed-ovule classes found 
in our material, the frequencies of the different pod lengths are tabu- 
lated out. 
The physical constants describing these characters are: 
ORAS OE ee SE 4.6947+0.0111 
Standard deviation of ovules.................-- ©.904I1+0.0079 
Coefficient of variation of ovules................ 19.258 
Prveraie O6igs os 4.0613+0.0134 
Standard deviation of seeds.................--: 1.0849+0.0094 
Coefficient of variation of seeds............-..-. 26.712 
Avetage Gg Hodes oy ee he 76.1810 .095 
Standard deviation of length of pod®............ 7.6814+0.0669 
Coefficient of variation of length of pod.......... 10.496 
Average index, seeds/ovules .2.... 6:26 46:6 .%s 0. 8650+0.0020 
Standard deviation of index..............-....- ©.1587+0.0014 
These constants require no discussion here; they enable us to 
pass to the determination of the degree of interdependence of the 
fertility characters and the length of the fruit. ; 
Consider first the correlation between number of ovules per pod 
and the length of the pod. We find r,;=0.4278+0.o101. Remem- 
bering that correlation is measured on a scale of o to +1, we see (a) 
that the sign of the relationship is positive, that is, that as the number 
of ovules per pod increases the length also increases, and (b) that the 
relationship is a moderately close one. 
The degree of interdependence between the two characters may 
be made clearer by expressing it in terms of regression instead of 
correlation. The equation to the regression straight line is 
y= 56.119 +3.634%. 
In this case y=length of- pod and x=number of ovules per pod. 
From this equation we see that a pod having an ovule more than 
the average of the population would be 3.6 mm. longer than the 
average length. 
° SHEPPARD’s correction was applied to the second moment in the calculation of 
the constants for length of pod. It was not used for the fertility characters, where we 
are dealing with integral variates, nor in the seed-ovule indices. 
