J. A. Harris 
331 
Since the number of ovules formed and the number of seeds developing on a 
placenta are correlated, the correlation between the number of ovules per placenta 
must bring about a correlation between the number of seeds developing on the 
two placentae. Evidence is adduced to show that the correlation for seeds is not 
entirely due to the correlation for ovules, but that the larger part of the inter- 
relationship for the seeds developing must be due to ecological or physiological 
factors. 
Loudon, December 7, 1909. 
V. Appendix of Correlation Tables of Data. 
The tables for the relationships between the two placentae are not rendered 
symmetrical except for the cross correlation for ovules of " first " placenta and 
seeds of " second " placenta where it is obviously necessary to use each placenta 
once as a " first " and once as a " second " placenta or publish two tables. 
It is unnecessary to publish tables for the following four relationships : — length 
of peduncle and total ovules, length of peduncle and total seeds, length of fruit 
and total ovules, length of fruit and total seeds. The product moments for these 
can be easily derived from those for ovules and seeds per placenta by taking the 
origin at 0 in calculating the first moments of the arrays of number of ovules or 
number of seeds per placenta. Since both placentae of a fruit were entered under 
the same grade of peduncle length or fruit length the first moment for the array 
gives at once the total number of ovules or seeds formed by that grade of peduncle 
or fruit. This is the same for the number per placenta or the total number per 
fruit. The origin of the other character — peduncle length or fruit length in the 
present case — may then be taken wherever we please, and the product moment 
calculated at once by multiplying up the first moments of the arrays by the 
deviations of the classes of the other character from their origin. The means of 
arrays for tests of linearity of regression may be easily obtained. The standard 
deviations for total ovules and total seeds per fruit are available from other tables. 
The correlation tables for number of ovules failing per placenta and per fruit 
are also omitted, for if their standard deviations be known the more important 
correlations can be derived from the tables for ovules formed and seeds developing. 
