﻿Harris: Distribution and correlation of sexes 669 



better than is done by a straight line. Such deviations as do 

 occur are probably due for the most part to the unavoidable errors 

 of sampling. At any rate, the discussion of this point cannot be 

 carried farther without more adequate biological information. 



It is interesting to examine the results obtained for each of the 

 sub-habitats from which A. proboscideum was collected.* These 

 are the data combined to form our Table III. 



The results for correlation between number of staminate and 

 pistillate flowers, r^p, are given in the third column of Table V. 

 With but two exceptions the correlations are positive in sign. 

 These two exceptional cases show very substantial negative coef- 



Constants for Individual Correlations of A. p 



oboscideum 





Loca,U, 



o^pt^I 



'•.p 



r^lE. 



r. 



r. U.. 





490 



46s 



685 

 164 



+.331 ±.033 

 7h.090i.031 



+.290 ±.089 



+.276 ±.070 





+T76 



+.981 

 +•952 



+^9^5 



+.921 



-;482±:o23 



+.399±-i57i 



13.68 











Contrada 



1.65 







Lareto . 





Liquerini 



Montefredani 



^2% 



ficients, which are apparently significant in comparison with their 

 probable errors. Since the number of individuals is so small — only 

 13 and 16 — the conventional method of computing the probable 

 error can not be given too much weight. The two exceptions to 

 the rule cannot, therefore, be regarded as a real argument against 

 the general existence of a positive correlation. The average value 

 of these correlations, obtained by weighting with the numbers of 

 individuals upon which the several coefficients are based, is 

 r= +.170. The difference between this value and the r= +.216 

 obtained from the whole (massed) material gives some idea of the 

 influence of heterogeneity on the magnitude of the correlation. 



Consider now the relationship between the total number of 

 flowers produced per inflorescence and the relative numbers of the 



