J. A. Harris 
315 
locules separately. The constants were worked out largely to furnish an extra 
safeguard against arithmetical blunders, but the results seem worth publishing 
as this may indicate to the non-statistical reader something of the amount of 
divergence between constants which may arise purely as a result of chance. 
The correlation coefficients and their probable errors are given in Table 4. In 
two cases I have also compared the correlation ratio, 77. See Table 5. In these 
two cases Blakemau's formula cited above gives : — 
For length and ovules, ^/ PJ^ = 4<-l2o, 
For length and seeds, ^/A'j = 4-131. 
Here again there is some question concerning the linearity of our regression. The 
equations to the regression lines are : — 
For ovules per placenta, ?/ = 5-863 + -532.^, 
For seeds per placenta, y — 2-987 + 'oOOx. 
Number of Ovules and Number of Seeds. 
0 
' '"V 
9' \ 
> \ 
Diagram II. Kegression of number of ovules per placenta on peduncle length and of number 
of seeds per placenta on peduncle length. Solid dots and line = ovules ; circles and dotted 
line = seeds. (Means depending on a few isolated observations not joined up.) 
