﻿Harris: Distribution and correlation of sexes 671 



/, = s -\- p, or total flowers per inflorescence, and z be positive it 

 indicates that as the total number of flowers per inflorescence 

 becomes larger the staminate flowers are relatively more riumerous. 

 Using the formula suggested to me* by Professor Karl Pearson 

 I find: 



For A. vulgar e:] 



rtm = + .964 ± -002, 

 ru = — .230 ± .026. 



For A . prohoscideum: 



rtm = + .931 ± -002, 

 rt, = - .234 ± .012. 



In both cases the correlation between total flowers and number 

 of staminate flowers is high. This is as one would expect it to be. 

 The interesting feature of the result is that in both species there 

 is a significant negative correlation between the total number of 

 flowers and the deviation of the number of staminate flowers from 

 their probable value. In short, the male flowers while absolutely 

 more abundant in the larger inflorescences are there relatively 

 distinctly less abundant. Or in other words the larger inflores- 

 cences tend to produce a larger proportion of pistillate flowers. 



Consider these relationships on the basis of the individual 

 collections. 



Table V, in which the constants for the individual collections 

 of A. prohoscideum are given, shows that with a single exception, 

 in which iV is very small, the correlations for total flowers and 

 staminate flowers are very high. 



The critical values derived from these are given in the sixth 

 column. Of these 10 coefificients, 7 are negative in sign, and 5 

 may be considered unquestionably significant in comparison with 

 their probable errors. Neither of the 3 positive values may be 

 looked upon as certainly trustworthy. Thus while the constants 

 for the individual sub-series are very irregular in magnitude they 

 fully substantiate the conclusions to be drawn from the constant 



* Harris. J. Arthur. The Correlation between a Variable and the Deviation 

 of a Dependent Variable from its Probable Value. Biometrika 6: 4.18-443- iQOQ- 



