346 



SCIENCE 



[N. S. Vol. XXXVIII. No. 975 



icant, being nearly eighteen times its probable 

 error. 



The comparison of the GG series with the 

 other Burpee's Stringless cultures is forcibly 

 brought out by the diagram. This shows the 

 linear graduations (from the equations given 

 in the table) of the number of pods per plant 

 for various seed weight classes. Only the GG 

 series, for which the empirical means are also 



were planted in numerical order across the 

 field, just as is usually done. In the OG 

 series the weight groups were planted in order 

 across a garden plot which was selected for its 

 apparent uniformity of soil. This was again 

 in accordance with ordinary experimental 

 practise. 



But many unknowable factors are involved 

 in the productive capacity of the soil and it 



I 



shown," indicates a decrease in the number of 

 pods associated with an increase in the weight 

 of the seeds planted.* 



The explanation of this result is simple. In 

 each of these 29 experiments with the excep- 

 tion of the LL, LG and GG series, the seeds 

 were individually labelled, thoroughly shuffled 

 and planted at random over the field to 

 counteract the possible heterogeneity of the 

 soil." In the case of the LL and LG series I 

 suspected that the soil conditions were not 

 strictly uniform, but the various " pure lines " 



' The inclusion of all the empirical means would 

 have rendered the graph too confusing. Graphic 

 tests made for each case affords no evidence that 

 a curve of a higher order would be better than a 

 straight line. 



*For the GGD, GGD, and GGDD series the 

 slope of the line is very slight. This is due simply 

 to the fact that these cultures were grown under 

 much more adverse conditions than the others, and 

 such wide variation in number of pods per plant 

 is not possible. 



" The importance of this procedure has been 

 emphasized in Amer. Nat., Vol. 45, pp. 697-698, 

 1911; Vol. 46, p. 325, 1912. 



appears that the particular parcel of ground 

 selected, although only large enough to grow 

 750 plants, changed in productiveness from 

 one side to the other. By chance the seeds 

 were so planted that the smaller ones were 

 given the best conditions. So great was the 

 heterogeneity that it not only neutralized the 

 influence of seed weight which is always 

 demonstrated when experiments are made with 

 proper refinements," but actually brought about 

 a negative correlation between weight of seed 

 planted and number of pods produced which 

 is numerically the highest found in twenty- 

 nine cultures! Had the order of planting been 

 reversed, both soil fertility and seed weight 

 would have been active in the same direction, 

 and an abnormally high positive correlation 

 would almost certainly have been the result. 



J. Abthur Harris 

 Cold Spring Hakbor, L. I., 



June 10, 1913 

 " The seeds used in the GG series were the an- 

 cestors of those employed in all the other experi- 

 ments with Burpee's Stringless. Thus there can 

 be no criticism because of "differences between 

 the pure lines used." 



