THE INFLUENCE OF THE SEED UPON THE SIZE OF 
THE FRUIT IN STAPHYLEA. II 
J. ARTHUR HARRIS 
(WITH ONE FIGURE) 
IV. On the nature of the correlation between the number of 
seeds and pod length . 
Heretofore we have contented ourselves with such analysis of 
our data as is necessary to the establishment on a sound quantita- 
tive basis of the fact of a correlation between the number of seeds 
developing and the length of the pod. This correlation is that of 
the statistician, not of the physiologist. It shows the existence of 
a relationship between two characters and measures its intensity, 
It does not prove that this relationship is due to a direct physiologi- 
cal interdependence between the two characters. To demonstrate 
such physiological interdependence one must remove the influence 
of other possible factors. All of the factors which seem possible 
sources of the correlation between the number of seeds developing 
and pod length and which can be investigated on the available 
data are discussed below. 
I. THE HYPOTHESIS OF THE INFLUENCE OF THE RELATIONSHIP FOR 
OVULES AND LENGTH 
As emphasized in the paper on Cercis, one of the sources of con- 
fusion in interpreting an observed 7, is the fact that both / and s 
may be correlated with 0, and so differences in the fruit length (/) 
which appear to be due directly to the number of seeds developing 
may be merely resultants of the relationship 7, and 7,;, and so 
indicate no physiological relationship between s and /. But it has 
been conclusively shown that n,>7 jo, thus demonstrating that the 
interdependence for seeds and length is not solely dependent upon 
the relationship for ovules and length. Indeed, it is only when 
both 7, and 7,, are high that they will greatly affect 7. It seems 
desirable, however, not to leave the question without showing just 
how much influence may be attributed to this factor. I use the 
two methods suggested in the paper on Cercis, namely, the corre- 
Botanical Gazette, vol. 53] [396 
