J. A. Harris 
493 
Data are available in Tables XLI — XLIII. Ooly the critical comparison 
C — A need be made. Table XXVI. gives the difiference for the mean number 
TABLE XXVI. 
Badial 
Asymmetry 
Mean "Odd" 
Locules, C — A 
•0000 
- •sss 
•4714 
- •loe 
•8165 
-•115 
•9428 
- ^454 
1^2472 
-■157 
of "odd" locules for the first five radial asymmetry classes. Beyond this the 
frequencies are too few to be trustworthy. Table XXVII. gives the difference 
TABLE XXVII. 
Character of Ovary 
Mean Eadial 
Asymmetry, C — A 
3 "even" 
-•017 
2 "even," 1 "odd" 
- -019 
1 " even," 2 " odd " 
- -034 
3 "odd" 
-•077 
between the mean radial asymmetry of C and A for the four classes of ovaries 
with respect to number of " odd " locules per ovary. 
I attach no importance to the numerical value of these differences, for I have 
not calculated their probable errors, but the negative sign throughout seems to 
nie fairly satisfactory evidence that both of these characters are to some extent 
of independent significance in determining whether an ovary shall develop to 
maturity. 
2. The Correlation between the Coefficient of Asjjmmetry and the Number 
of Ovules 2)er Locule. 
Both the actual number of ovules per locule — or per ovary — and the radial 
asymmetry of the fruit with respect to the number of ovules per locule seem to 
have significance in determining the fate of an ovary. Are they really both 
significant, or is one dependent upon the other ? 
This is a fundamental physiological question. I believe a trustworthy answer 
is given by the coefficient of correlation between the asymmetry of the fruit and 
the number of ovules per locule. Series B is the most like the original popula- 
Biometrika vii G3 
