﻿672 Harris: Distribution and correlation of sexes 



for the whole material. The average (weighted) value is 

 rt, = - .223. This value is not significantly different from the 

 rt2 = — .234 obtained from the combined materials. 



III. Summary and Conclusions 



As a suggestion for more adequate investigation of variation 

 and correlation in the specialized inflorescence of the aroids this 

 note presents the results of a statistical analysis of two rather large 

 series of data collected for the genus Arisarum by Cannarella. 



The following points may be emphasized: 



In both A. vulgare and A. prohoscideum, the coefficient of 

 variation for number of pistillate flowers is far higher than that 

 for number of staminate flowers. The interpretation of this 

 result presents considerable difficulty. 



In A. vulgare the inflorescence not only produces a larger 

 number of both staminate and pistillate flowers than does that of 

 A. prohoscideum, but shows a far higher variability, both absolute 

 and relative. 



Apparently the collections of A. prohoscideum came from a 

 wider range of habitats than those of A. vulgare. If this be true, 

 the variability which they show is in part due to heterogeneity 

 arising through environmental differentiation. This probability 

 renders more significant the observed lower variability in A . pro- 

 hoscideum. Any subsequent investigation should if possible take 

 into account this factor in drawing comparisons between the 

 variability of these or any other forms. 



The inflorescence of A. vulgare is not only more variable, but 

 shows a higher degree of correlation of its parts than does that of 

 A. prohoscideum. In both cases the correlation between number 

 of staminate and pistillate flowers is rather small. The relation- 

 ship between the numbers of the two types of flowers may be fairly 

 well described by a straight line. The fit of the straight line 

 equations is not altogether satisfactory, but it seems idle to carry 

 the statistical analysis farther on the present data. The effort 

 would much better be expended in securing more adequate bio- 

 logical material. 



The correlation between the total number of flowers produced 

 by the inflorescence and the number of staminate flowers is neces- 



