No. 513] SHORTER ARTICLES AM) ('ORRKSPOXI)EX('E 559 



The constants from Table I are given in Table II. 



Regarding first the means, we note that there is no constant 

 difference between those of the different collections. This is true 

 without considering the probable errors, which at once take away 

 any opportunity for theorizing on these points. Inspection shows 

 the same conclusion to be necessary concerning both absolute and 

 relative variabilities. 



II 



Tree I 



4 .01.-.-2 

 4 -Mi\2 



ii • .u-j:;f. 



mm;:; 



So far as this series of data goes, therefore, we have no reason 

 to think there is any selection in the elimination of the ovaries 

 which do not develop into mature fruits. Most unfortunately 

 the mature fruits are not available for comparison. This ren- 

 ders it necessary to draw the comparisons not between the con- 

 stants for all the ovaries formed and the fruits maturing, but 

 between the constants for the original population of ovaries and 

 those for the fallen flowers. While not so satisfactory as the 

 former method, the latter is quite justified in the present case, 

 since the second collection represented the first of the ovaries to 

 be eliminated from the tree and the third was taken from the 

 main bulk of those which were to fall. During the last two 

 years I have not succeeded in getting suitable data for the num- 

 ber of the ovules in trustworthy samples of ovaries formed and 

 ovaries maturing in either this or other species of Leguminosa?. 

 In our present dearth of quantitative data concerning selective 

 elimination, this seems a good line to follow far enough to get 

 conclusive positive or negative results. Perhaps some other 

 botanist can collect adequate masses of data. 



J. Arthur Harris. 



