ON A CRITERION OF SUBSTRATUM HOMOGE- 

 NEITY (OR HETEROGENEITY) IN FIELD 

 EXPERIMENTS 



Dr. J. ARTHUR HARRIS 



I. Introductory Remarks • 



Every one who has had practical experience in variety 

 or fertilizer tests or in any other experiments involving 

 the comparison of field plots must have been impressed 

 by the great difficulty of securing tracts with uniform soil 

 for their cultures. 



A careful examination of the agricultural literature 

 bearing on the question of variety tests will reveal many 

 cases in which the experimenters have noted the difficulty 

 of securing a uniform substratum, or in which there is 

 internal evidence for the influence of substratum hetero- 

 geneity upon the result. 



For example, in 1894-1895 tests of varieties of wheat 

 were made on 77 plots at the University of Illinois. 1 As 

 a check on the other strains, the variety known as Valley 

 was sown on nine different plots * ' well distributed over 

 the area sown." 



. . . the yields of this variety varied from 11.7 bushels to 24.1 

 bushels, an average of 19 bushels which is remarkably close to the 

 average of all the varieties. It is again remarkable that but eight yields 

 were above the highest of the Valley, and but three below the lowest 



The only reasonable explanations that can be given for 

 such results are either (a) that the plots were so small 

 that the results are due purely and simply to the errors of 

 random sampling, or (b) that the wide divergences in the 



i Butt. Univ. III. Agr. Exp. Sta., 41, 1896. 



430 



