96 MISSOURI BOTANICAL GARDEN. 
the degree of individuality of the fruits with respect to 
their capacity for laying down seed-forming substance. 
The 1906 series of weighings were made for this purpose. 
The labor involved in the preparation and verification of 
correlation tables with every seed used once as a first and 
once as a second member of an associated pair is very 
greatly increased by the use of varying numbers from the 
fruits. It is desirable, therefore, to have the same and as 
large a number as possible from each of the 200 fruits. 
Ten can easily be secured from the larger earlier maturing 
fruits. The fruits, therefore, are not a random sample of 
the whole crop, but are intentionally a selection, though 
only in so far as is necessary to secure those containing 
at least ten seeds. The seeds to be weighed were chosen 
quite at random by selecting those falling farthest to the 
right when the fruit was opened above a table. 
The symmetrical intra-fruit correlation surface appears 
as Table X.?2 Within each fruit there are n(n-1) com- 
binations, or a total of 10 « 9 x 200 18,000. The corre- 
lation coefficient is 
r = .6759 + .0082 
and the equation to the regression straight line, shown 
graphically in Diagram III is 
» = 1.4225 + .6759 W,, 
_ where the constants are in the half-gram units of weighing. 
The close agreement of the empirical and theoretical means 
leaves no question concerning the linearity of regression. 
The results indicate that some complex of morphogenetic 
and physiological factors tends very strongly to make the 
seeds of a fruit alike in weight. The nature of these factors 
and the part played by each must be determined by the 
application of proper biometric formule to more detailed 
records. 
22 The seed weights are for convenience expressed in the class units 
given in Table VI. 
