J. A. Harris 
485 
the results graphically. The three comparisons are made on the basis of the 
percentage frequencies in Table XX. 
TABLE XX. 
Formula 
B— A 
C— B 
C— A 
3 "even" 
2 "even," 1 " odd" 
1 "even," 2 "odd" 
3 "odd" 
+ 1 -660 
+ -357 
-1-671 
- -347 
+ 16-722 
- 3-650 
-10-222 
- 2-849 
+ 18-382 
- 3-293 
-11-893 
- 3-196 
3E 2E+0 E + 20 30 
Series A. Eliminated. 
3E 2E + 0 E+20 30 
Series B. Developing. 
Series C. Matured. 
Diagram 6. Percentage frequencies of four types of locular composition of ovaries in 
the three series. 
In both of the critical comparisons it appears that elimination brings about an 
increase in the per cent, of ovaries with all three locules " even " and a decrease in 
the per cent, of those which contain one or more locules with an odd number of 
ovules. The differences are also conspicuously large, amounting to 18-4 per cent, 
in the case of the " all even " class. 
In view of the results secured for the individual locules, it has not seemed 
necessary to calculate the probable errors of their differences. A better test would 
seem to be that for the probability against a given system of deviations from 
theory, in the case of a correlated system of variables having arisen through the 
Biometrika vii 62 
