354 



THE AMERICAN NATURALIST [Vol. XLIII 



that neither of the series with which we are dealing is 

 large enough to give very trustworthy results for the 

 frequency of individual grades. AVe might avoid to some 

 extent the difficulties introduced by the errors of random 

 sampling by graduating the observed frequencies by some 

 satisfactory fitting formula, but it does not seem profit- 

 able to take up the labor of curve fitting on the available 

 data. We may, however, calculate the means and varia- 

 bilities with their probable errors. The constants and 

 their differences appear in Table II. 



The means show that the "Wood's Holl plants have 

 about three tenths of a seed more per pod than the Danby 

 Dale series. This is to be expected from the diagram 

 where the polygon for the American series is seen to lie 

 a little further to the right than the English. But the dif- 

 ference is only about four times its probable error and so 

 only doubtfully significant even for the samples in hand. 

 I think no one can examine these results and not be con- 

 vinced that so far as is shown by the physical constants 



Table II 









Danby Dale 9. 6425 ±.0691 

 Wood's Holl 1 9.9982±.0502 

 Difference —.3557 :- .0854 



3.5466±.0488 

 3.5236±.0355 

 .0230 -.0603 



36.780 

 35.242 



the two collections, one gathered in England and the other 

 in America and counted and reduced by independent ob- 

 servers, are practically identical when we have regard to 

 the probable errors of random sampling. And this not- 

 withstanding the fact that the age of the plants, the num- 

 ber of plants involved, the ancestry of the individuals, 

 and the climatic and soil conditions under which they 

 grew are not taken into consideration or allowed for in 

 any degree! If the constants were based on some char- 

 acter which the qualitative biologist would guess to be 

 little influenced by the environment one might not be sur- 

 prised at their close agreement; but we are dealing with 



