FIELD PLOT TECHNIC 55 



i.e.) a progressive difference of 25 Ib. for each plot in either direc- 

 tion from the central one. The corrected yields are found in 

 the last column of the table. Note that the probable error is 

 3 per cent, less in the corrected than in the unconnected yields. 



The method outlined above may be used only where there are 

 a comparatively large number of similarly treated plots and 

 where the increase or decrease in yield across a field is fairly 

 consistent. If check plots are grown every third to fifth plot 

 as they frequently are, a direct correction for yield is sometimes 

 made as follows : 



DIAGRAM ILLUSTRATING DISTRIBUTION OF CHECKS 



Suppose every fourth plot is a check. The productivity of 

 each intervening plot is estimated on the basis of the yields of 

 the two nearest checks. For instance, the true productivity 

 for plot one equals %C + Y^\\ for plot two equals ^C + %C\] 

 for plot three equals YC + %Ci; etc. For example, by this 

 method the yielding value of plot six could be obtained. The 

 corrected yield could then be obtained by the following pro- 

 portion: Average yield of all check plots: yielding value of plot 

 six = the actual yield obtained from plot six: X. In a similar 

 way corrected yields could then be obtained for all plots in the 

 test. 



Use of Checks as a Probable Error of the Experiment. Other 

 methods of using the checks as direct corrections for yield have 

 been employed, but the tendency in present-day field investi- 

 gations is away from the use of checks for this purpose (especially 

 where yield is being studied). They are, however, very valuable 

 indices of soil variation, thus giving an approximate measurement 

 of reliability for the particular experiment. To illustrate the 

 use of checks in this way, suppose in a certain experiment there 

 were 50 systematically distributed checks grown and that the 

 computed probable error of a single check plot (standard devia- 

 tion X 0.6745) was 4 bu. Suppose each variety or strain being 

 investigated for yield is replicated three times, making four 

 plots in all. The probable error of the average yield of these 

 four plots would be equal to the probable error of a single check, 

 4 bu. divided by the square root of the number of plots, or 4. 

 This gives 2 bu. as the probable error of the average yield of 



