26: 



Journal of Agricultural Research 



Vcl. XII, No. s 



one direction only, and therefore has more significance than is indicated 

 by a mere comparison of the averages. 



EFFECT OF SYSTEMATIC DISTRIBUTION OF PI.OTS OVER THE AREA STUDIED 



The importance of distributing plots over the experimental area is 

 more or less obvious, and has been dwelt upon by many writers. Its 

 value arises from the fact that the soil varies over the area, and it is 

 better to have similar-sized plots on both high- and low-yielding areas 



than to have them 

 solely on one or the 

 other kind of soil. 

 The method should be 

 of special value on 

 areas which vary 

 rather uniformly in 

 one direction. 



Increasing the num- 

 ber of trees to the plot 

 k^^" \ \'^'\^ in scattered units of 



^^y- ,s> T«L _ either four or eight 



trees gives a more 

 typical sample of the 

 productivity of the 

 total planting than 

 the same number of 

 adjacent trees. In 

 scattering the plots 

 throughout the area 

 studied, they were sys- 

 tematically repeated. 

 For example, if there 

 were loo plots in all to 

 be grouped in pairs, 

 the first and fifty-first, 

 and the second and fifty-second were united, and so on through the series. 

 If a' quadruple series was desired, the first, twenty-sixth, fifty-first, and 

 seventy-sixth plots were combined. 



Table IV shows the results of scattering 4- and 8-tree plots, respectively, 

 in the plantations studied. Figure 8 illustrates the reduction of the 

 coefficient of variability by increasing the number of trees to the plot in *. 

 both 4- and 8-tree scattered units, compared with the average coefficient 

 of variability for the several fruits by increasing the size of a plot from i 

 to 24 adjacent trees, together with the theoretical curve calculated from 

 the mean coefficient of variability of all the i-tree units. 



A comparison of the curve for adjacent trees and those for scattered 

 units shows at once the marked decrease in favor of the scattered units. 



Fig. 8. 



—Graphs of the reduction of the coefficient of variability by 

 increasing the number of trees to the plot. 



