Harris: OVULES AND SEEDS IN CERCIS 545 
and below. What one does find is that the empirical mode falls 
on the class — 3.50 to — 2.50. The area of the polygon which 
is composed of statistically significant constants is shaded in. 
ime 
a a 
CE ie a eae ae 
ICAL MODE 
Er PULT Terre. 
2 
teh alee lt a te 
ake 
+ 
GUARDAR MRE RS el 
j 5 Ea a a ha 
i HO A a 
2 AOS A A LS gS 
MPII 
EEE 
=-3.50 -2.50 —1.50 —.50 
-1.50 - 50) +.50 
- 6.50 —5 50 — ASO 
—5.50 —4.50 ~3.50 —2.50 
Fic. 3. Distribution of ratios of roz coefficients to their probable errors, showing 
large number of significantly negative correlations for the eastern Kansas series. 
C. The Individuals from the Vicinity of Sharpsburg, Ohio 
In this collection, the lumped data gave values of Tos = -455 
+ .009, fo. = — .034+.011. Thus the correlation between 
number of ovules and the capacity of the ovary for maturing its 
ovules into seeds while negative in sign is not only low but is 
only 3.15 times its probable error. 
The calculated correlations for the individuals are set forth 
in Table X. 
- These results differ essentially from those secured for the 
Kansas series in two regards; (a) only 6 out of the 26 may be 
considered statistically significant with regard to their probable 
error as compared with 14 out of 21 in the Kansas series; (b) the 
constants are about evenly distributed between positive and 
negative, there being 14 positive and 12 negative signs. The 
general average is, however, negative. 
Again, the ratios of the constants to their probable errors are 
plotted out in a polygon (FIG. 4) showing the scatter of the con- 
stants on either side of zero. 
