33s 



THE AMEBIC 



IT [Vol. XLVI 



between individuals whose ancestors had been subjected 

 to distinctly unlike treatment is made in Tables V-VIII. 



Consider first the means. Altogether there are 28 

 inter-ramal comparisons, direct and cross. The number 

 of pods is smaller in the plants whose ancestors had been 

 starved in 16 out of 28 cases. If there were no relation- 

 ship between the conditions to which the ancestors were 

 subjected and the number of pods which their offspring 

 produced, one would expect 14 to be negative, providing 

 the errors of random sampling had not to be allowed for. 

 But the probable error is 



.6745V28 X .5 X .5 = 1.79. 

 Clearly a difference of 2 ± 1.79 has no significance. 



If now we restrict the comparison to differences signifi- 

 cant with regard to their probable errors, and consider 

 DifT./iJdiff. > 3 to be significant, we note that only 11 out 

 of the 28 differences may be regarded as statistically 

 trustworthy. Of these, 9 have the negative and only 2 

 the positive sign. Certainly this looks as though there 

 were a very slight effect of the starvation of the ances- 

 tors, but nevertheless an effect quite detectible by the 

 statistical methods. 



This point may be tested further by taking the aver- 

 ages, regarding sign, of the pertinent differences for the 

 series of the three varieties. To make sure that slight 

 racial differences between ND and NH do not obscure the 

 results we recognize two classes of comparisons, within 

 the strain and between strains. The results are : 



- A =— .515 



General Average, A = — .515 

 Ne Plus Ultra, A = — 1.315 



White Flageolet, .1 = + .585 



In all cases except the White Flageolet series 17 the 

 number of pods is slightly lower when the ancestors have 

 been starved. 



in the White Flageolet variety. 



