Oct. 28, i9i8 Experiments in Field Technic in Plot Tests 259 



ARE ALL VARIETIES AFFECTED ALIKE BY SURROUNDING ALLEYS ? 



If, when grown in plots surrounded by alleys, the yields of all varieties 

 or of all cultural or fertilizer treatments are increased approximately alike, 

 then, so far as the comparative results of the different varieties in any 

 single test are concerned, the inclusion of the border rows of plots at har- 

 vest is not objectionable except in so far as has been noted. An opinion 

 regarding this has been expressed by Barber (j, p. 82), as follows: 



All in all, as long as conditions are similar for all varieties of grain in trial in a field of 

 fairly uniform soil, the results of plot tests will show the relative yields of the varieties. 



Referring again to Table IV, it will be noted that in the column headed 

 "No border rows removed" the varieties of oats, wheat, and barley are 

 each listed in order of yield, in bushels per acre, for that method of test. 

 Do the varieties maintain approximately the same relation with regards 

 to yields when one and two border rows, respectively, are removed from 

 either side of each plot? 



An inspection of the rank of the different varieties, when one and two 

 border rows, respectively, have been removed from either side of each plot 

 shows some changes. If these changes in rank, due to the removal of 

 border rows are fairly consistent throughout the 3- or 5-year period in 

 tests conducted by the three methods, what effect will this have on the 

 final interpretation of results ? 



The standard deviation for 20 control oat plots is 3.24^:0.35 bushels. 

 On using this standard deviation in the formula 

 Standard deviation X 0.6745 

 ■yj n 



in which n denotes the number of plots, to derive the probable error in 

 bushels per acre for the yields obtained from plots replicated three times 

 (four plots of each variety), the result is found to be 1.09 bushels {11). 

 A difference of 4.07 between two results is necessary for odds of 40 to i 

 against such a difference in one direction only being due to normal 

 variation {u). Multiplying 4.07 by 1.09 bushels gives 4.44 bushels, 

 which is the least difference between any two varieties, which is signifi- 

 cant. This figure may be used in a broad way in considering the results 

 for three methods of test. In the test with no border rows removed, 

 Victory yielded 12.31 bushels more than any other variety. Since this 

 difference is greater than 4.44 bushels. Victory oats may be considered the 

 highest yielder under that method of test. Considering the other varie- 

 ties in a similar way, Minota is not better than Silvermine, but is a higher 

 yielder than Banner or any other variety in the trial yielding less than 

 Banner. Silvermine is not better than Banner, but is superior to Lincoln 

 or any other variety yielding less than Lincoln. Banner, Lincoln, 

 O. A. C. 72 and Iowa 103 are higher yielders than Swedish Select, ICherson, 

 White Tartar, and O. A. C. 3. 



