HARRIS: OVULES AND SEEDS IN CERCIS 543 
FE. = .67449 X 0.1/1 60 = .0087. 
This line of argument has been followed out in the preparation 
of Table VIII where for convenience I assumed that 2 = 100 for 
each tree.* In this table, the constants actually found are com- 
pared with the values one should expect them to have if the corre- 
TABLE VIII 
MEAN VALUES OF 7oz FOR THE INDIVIDUAL TREES 
| 60 Trees 40 Trees tro Trees 
Theoretical mean........... -0000 +.0087 .0000+.0107 | .0000+.0064 
Calculated mean............ —.0292+.0106 | —.0175+.0154 —.0232+.0080 
PU CPOOOE, x. . See vas 20'S —.0292+,0137 | —.0175+.0187  —.0232+.0102 
Theoretical standard deviation -1000 +.0062 .1000+.0075 | .1000.0045 
Calculated standard deviation. +1218 +.0075 -1440-.0108 | .1242 +.0056 
Re EIOONE yi dc oe ee eas | -0218 +.0097 0440+.0131 | .0242+.0072_ 
lation between the number of ovules per pod and their capacity 
for maturing their seed were actually o, and the constants as 
found from actual collections were due merely to the errors of 
random sampling. 
For all three series the mean differs from 0 by less than 2.5 
times its probable error. For the standard deviations the differ- 
ence between the observed and the theoretical values is less than 
2.5 in the case of the 60 trees and only about 3.3 times its probable 
€rror in the other two series. 
Notwithstanding these low values of the means and the nearly 
equal areas of plus and minus values on the diagram, one must not 
lose sight of the facts, (a) that there are more negative than 
positive coefficients, the ratio being 57 : 43 for the smaller col- 
lections and 64 : 46 for the series from all the individual trees, 
and (6) that all the means are negative in sign. 
This series like those discussed above seems to indicate that if 
there is any relationship between the number of ovules per ovary 
and the capacity of the ovary for maturing its ovules into seeds 
it is of such a nature that the ovaries with the larger numbers of 
ovules are slightly less capable than those with the smaller 
numbers. 
his is not strictly true; I think that the approximation is quite close enough 
= Present purposes. Sheppard’s correction was used for the empirical distribution. 
