I9I3] 



/. Arthur Harris and Ross Aiken G ortner 



TABLE I 



Weight of root in grams 



525 



coefficient can be determined.^ From their summary, reproduced in 

 slightly modified form in the above table, we deduce, using the con- 

 venient formula 



''«.= 



'2{w's')/N—ws 



<r CT 



where 2(w'/) denotes the summations of the (mid-ordinate) values 

 of the weight and sugar content of the individual beets, the bars 

 denote the means and the sigmas the Standard deviations of the 

 two variables, 



r„. = .116 ± .025. 



The correlation Is over four times its probable error and so pos- 

 sibly statistically significant, but certainly is very low. Expressing 

 the relationship in terms of regression, using the straight line equa- 

 tion employed in the preceding paper 



+ r- 



we find 



5^ = (5 — r-^w] 



\ '^ra J 



s= 18.722927 + 0.000796 w, 



3 Throughout their work these authors speak of their first series as con- 

 taining 699 beets. But the entries in their fundamental table, " Tabelle II a," 

 add up to 701, so we have thought it best to foUow this. In two other series 

 we cannot make our additions agree with theirs. 



