Tg12] HARRIS—STAPHYLEA 217 
As constants describing the degree of interdependence between 
number of seeds developing and length of pod, there seems little to 
choose between these. To satisfy ourselves more fully concerning 
Length of fruit 
‘ 6 7 8 9 sie) II 
oy* ° - 
. s 
I e ig 
‘ 
\ 
* : \ 
2 oe Ae 
| 
7 
* 
3 \jeo 
Ay 
4 AN oe 
x 
x 
5 © 
\ N 
ep ‘ 
g 6 e 
Oo ‘ 
— x 
: 7 of a 
3 \ 
i, ‘ 
3 8 oi ° 
oO Xi 
8 ‘ 
"a 9 °o . , 
i) ‘ 
a 
ay 
ae 
eal x 
\ ae 
7 
‘ 
a7 
13 ‘ 
‘ 
4 
* 
14 
‘ 
: 
\ 
15 oO ¥ 
. 
‘ 
. 
16 %. 
—Regression of pod length on total number of seeds developing per fruit; 
solid WE and line =1906; circles and broken line=1 
the rate of change in length of pod, whether it is a simple arithmeti- 
cal relationship or whether it is some more complex curve, we may 
calculate & and apply BLAKEMAN’s* test for linearity of regression. 
LAKEMAN, J., On tests for linearity of regression in frequency distributions. 
Biometrika 42332-3650. 1905. 
