NEW HAMPSHIRE EXPERIMENT STATION 



[Bulletin 271 



In Table II the combined yield of the 10 cuttings is reported, to- 

 gether with the probable error and the difference of the yield of each 

 treatment from the yield of the check plots, with the probable error of 

 the difference. 



Table II — Combined yield all harvests (5 yrs., 10 cuttings), 



Whenal alfalfa 1927-31 



X .6745 



error (P.E.) was determined by Bessel's formula: 



in which E D.^ refers to the sum of the squares of the de- 



*Probable 



n(n-l) 

 viation from the mean and n to the number of plots. 



In interpreting probable error, we have considered that to be significant 

 a difference must be at least three times its probable error. 



See H. H. Love, "The Role of Statistics in Agronomic Experimentation," 

 Sci. Agr. 5: 84-92. 



From a study of these data it is evident that all of the differences 

 from the check-plot yields are significant, except those for lime (L), 

 nitrogen (N), and phosphorus (P). 



Of the substances used alone, potash (K) and manure (M) gave 

 significant and substantial increases. It is interesting to note that an 

 annual application of 150 pounds of muriate of potash (K) has been 

 slightly more effective in increasing the yield of alfalfa than the addi- 

 tional application of 20 tons of manure (M) applied at the outset of 

 the experiment. 



Table III — Increase in yields from potash and 



manure 



