log Harris. 



stances, it seems desirable to point out the apparent significance of 

 the data which I have secured for the mature inflorescence. 



The first fundamental requisite to the tenability of the theory 

 that the differences in the eliminated and matured ovaries is due 

 merely to positional differentiation, is to show that such a differen- 

 tiation within the inflorescence really does exist. It would be desirable 

 to do this with young inflorescences but the objection that it is idle 

 to make such determination on mature inflorescences from which 

 many ovaries have fallen may be set aside at once, since it is equi- 

 valent to an admission that the fallen ovaries are different from those 

 which reach maturity. 



The second fundamental requisite is to show that elimination is 

 proportionally larger from the more distal portions of the inflorescence. 

 This has not demonstrated to be true in Staphylea, but I admit its 

 probability in order that the test of the theory of the relative inferio- 

 rity of certain types of ovaries may be as severe as possible. 



We now turn to our data. 



(a) Position on the Inflorescence and Number of Ovules. 



The disadvantage of working with a "general population" of 

 vegetable organs is that the heterogeneity due to the mixing of 

 numbers of parts taken from a small series of individuals may some- 

 times introduce a spurious correlation which may somewhat obscure 

 real biological relationships. This is especially true in work on fertility 

 where one is trying to dissect out delicate physiological interrelations- 

 hips by statistical means. 



For the 1906 series I have, therefore, drawn up correlation tables 

 for Tpoi) in the usual form for the twenty individuals. There are 

 altogether twenty of these tables. To publish them in the con- 

 ventional way would be too great tax on space, and they are con- 

 sequently condensed into a single line each by placing the characters 

 of both variables at the head of the table. The data for ;> appear 

 in Table I. 



1) For the assistance of those who may not yet be familiar with biometric 

 terminology, we may state that the coefficient of correlation between two characters 

 is represented by the letter r. The two characters between which the correlation is 

 calculated are indicated by their subscript letters. Thus, Cpo is to be read 'the 

 correlation between the position of the fruits on the inflorescence and the number 

 of ovules formed per locule". 



