334 



THE AMERICAN NATURALIST [Vol. XL VI 

 TABLE III 



FSHHC 

 FSDC 

 FSDDC 



1 1.221s ± ."_!()->*> 



12.9562 * .1856 

 12.2431 * .2068 

 14.1505 * .2115 



7.4134 ± .1200 

 ^7959 ± !ll04 



6.3729 * .1463 



from the tables of constants. As already emphasized 

 the comparisons between the ancestral series are of in- 

 terest for our present purposes only in so far as they 

 furnish proof that the parents of the comparison series 

 were conspicuously differentiated in type and variability 

 because of the environmental conditions to which they 

 were subjected. The reader must always keep in the 

 foreground the fact that our problem is not to determine 

 in detail what the causes of this differentiation are, but 

 merely to show that a conspicuous differentiation exists 

 and to ascertain whether it has any weight in determining 

 the characteristics of the offspring. 



The differences between the starved and well-fed an- 

 cestral series are so well marked that constants are best 

 represented by graphs for all the series. In diagrams 6 

 and 7, which embody data for all possible comparisons 

 for A and o-, roughly made, the key number of the va- 

 riety is given along the left-hand margin. The value of 

 the constant for the ancestral series is indicated by the 

 position of a solid dot when the series is a starved one, 

 and by the position of a circle when it is a well-fed one. 

 The value of the constant for the offspring of each of 

 these ancestral series grown upon the comparison field 

 is shown by the position of a solid square under a sep- 

 arate scale. Thus the key to the comparison series is 

 given by adding C to the formula for the ancestral series. 



The graphs for the mean number of pods per plant and 

 for the standard deviation of number of pods per plant 



