268 



Journal of Agricultural Research 



Vol. XII, No. s 



Tabl^ VII. — Comparison of the reliability of a plot of 32 adjacent trees with that of a J2- 

 tree plot of four scattered units of 8 trees each. Navel oranges {Arlington) 



On turning now to the navel oranges (Antelope Heights), which the 

 calculations show to be the most uniform planting of any observed, the 



question might arise 

 as to what degree of 

 reliability may be ex- 

 pected in comparing 

 two i6-tree plots. 

 Table VIII shows a 

 comparison between 

 such plots made up of 

 adjacent trees and 

 plots composed of 

 scattered units of four 

 trees each together 

 with the theoretical. 

 A much greater range 

 is found with the plot 

 of adjacent trees, while 

 the necessary differ- 

 ence between two 

 plots for a 10- to- 1 as- 

 surance is 62.94 per 

 cent. With plots 

 made up of scattered 

 units the necessary 

 difference is 32.05 per 

 cent. The hetero- 



!0 ^ 10 

 f\j W^ K 



\ 





Fig. 10.— Curve of yields of individual trees, navel orange (Arlington). 



geneity of this soil is clearly shown by both the increase in range 

 and the increased probable error when plots are composed of adja- 

 cent trees. This block of trees, however, appeals to the observer as 

 unusually uniform and would be considered desirable for plot experimen- 

 tation, the fluctuation in the productivity of the trees approaching closely 



