HARRIS: RELATIONSHIP OF OVULES TO SEEDS 255 
ated in number of ovules and number of seeds per pod, and prob- 
ably in the variability and correlation of these two characters. 
Since the general samples are made from a relatively small number 
of these individual trees some differences between collections 
might arise through the errors of sampling in the selection of 
individuals. 
TABLE VIII 
COMPARISON OF THREE SERIES OF Cercis 
| Missouri Series Kansas Series Ohio Series 
Aotel individuals. .......:... eee ger, 123 22 | 26 
Sab petag een eee 8,554 2200 3900 
Ov 
BS Meta, cg archer gee) 4.702 =.004 4.916 +.013 5.493 =.011 
readied deviation. ....... 1.007 +.003 .925 =.009 .983 =.008 
Coefficient of variation. .... 21.43 18.82 
NCA ae aaa lee ger a 3.840 +.005 4.116 +.015 3.944 +.016 
Standard deviation........ 1.204 +.003 1.030 =.010 1.453 +.011 
Cocticient of he ne hits 31.35 25.03 36.83 
ES AND SEEDS 
Correlation, og... 0s esse .648 +.002 .603 +.009 -455 =.009 
WOPrelation; fers. @ ssc o9s —.046 =.004 —,183 +.014 —.034 *.011 
a epeticient of fecundity..... .8166 -8373 -7180 
Again, no account whatever can be taken of environmental 
conditions, either edaphic or meterological. 
With two such factors, which may to some extent tend to 
bring about differences in the constants of the series dealt with, 
it has seemed to me rather surprising that the physical constants 
for ovules and seeds do not differ more widely than they do. 
The coefficients of correlation, 7, differ considerably; but two 
factors influencing this constant must not be forgotten. First, 
heterogeneity, due to the mixing of the pods from a large number 
of individuals, would tend to raise the value for the Missouri 
series. This appears very clearly in a comparison of the mean 
for the 60 individual constants from trees with 100 pods each and 
the constant for the 28,000 and more pods in the lumped sample. 
The former is .599, the latter .648. Second, the coefficients of 
fecundity show that the three series differ very materially in the 
percentage of ovules developing into seeds. The lowest value of 
the coefficient of correlation for ovules formed and seeds maturing 
(in the Ohio series) is associated with the lowest value of the coef- 
ficient of fecundity. 
Finally, perhaps the most important point to be gathered from 
