Dec. i, 1920 Permanence of Differences in Experimental Plots 355 



The first purpose of the present study has been to determine whether 

 such differences in fields selected for their apparent uniformity by skilled 

 agronomists are of a purely transitory nature or whether they are of a 

 relatively permanent character. 



This problem can be solved by determining whether in such series of 

 uniformly treated plots the yields of the same plots in different years 

 are correlated. 



The results of the present study show that of the 152 correlations be- 

 tween the yields of the plots in different years, 133 are positive as com- 

 pared with 19 which are negative in sign. The average value of the 

 positive correlations is + 0.335, whereas the average of the negative 

 constants is — 0.148. The general average is + 0.274. With the excep- 

 tion of the 191 1 crop of sugar beets the correlation between the yields 

 of each individual crop and the yields on the same plots in the eight 

 other years of the experiment are on the average positive. 



The data available for half and quarter plots fully substantiate the 

 results for whole plots. 



The results show conclusively, therefore, that plots, even of the small 

 size and apparent uniformity of those at the Huntley Station, are 

 characterized by differences which may persist throughout a period of 

 years. Thus, in general, plots which produce more in one year will 

 produce more in another year. 



This is, of course, a well-recognized principle for large tracts. Its 

 validity for small plots has apparently not been recognized heretofore. 

 It is probably not a principle of universal applicability, because of the 

 fact that meteorological as well as soil conditions play a large part in 

 determining yield. It is quite probable that certain soil characteristics 

 would result in maximum yields with one set of meteorological con- 

 ditions but in minimum yields with another complex of aereal con- 

 ditions. 



The determination of the proximate factors to which these corre- 

 lations are due presents a problem of considerable difficulty. Unfor- 

 tunately (for this phase of the problem only) alfalfa was introduced 

 early in the rotation and occupied the ground for three of the nine years 

 covered by the experiment. It seems quite possible that the correlations 

 between certain of the yields is due in part to the variation in nitrogen 

 content of the soil referable to the variation in thickness of stand and 

 strength of growth of the alfalfa crops. 



The results show that there is but little correlation between the alfalfa 

 yields of 1912 to 1914 and the ear corn yields of 191 5, whereas the corre- 

 lations for ear corn in 1916 are high. Thus the influence of alfalfa 

 upon the yield of a subsequent crop is not fully evident until the second 

 year after it is turned under. 



There is a definitely demonstrable residual influence of the variation 

 of alfalfa yields upon the yields of subsequent crops. The influence of 



