26o Journal of Agricultural Research voi.xv, no. 4 



The differences in yield per acre for the various oat varieties brought 

 about by the removal of one border row do not necessitate any material 

 changes in the rank of the varieties as discussed for the test made with- 

 out the removal of border rows. 



By using 4.44 bushels as the least significant difference between any two 

 varieties in the test of the oat varieties with two border rows removed, 

 Victory still maintains the lead, but with less margin than in the test 

 with no border rows removed. Minota can not be considered a higher 

 yielder than Silvermine, Iowa 103, or Banner, but is superior to O. A. C. 

 72 and the other varieties lower than it in yield. 



In the test with two border rows removed the variety Iowa 103 shows 

 up as very promising, while in the test with no border rows removed it is 

 of indifferent value. 



The standard deviation for 20 control wheat plots is 2.45 ±0.26. By 

 using the standard deviation in deriving the probable error in bushels per 

 acre for the yields of the wheat varieties as for the oats it is found to be 

 0.83 bushel. Multiplying 4.07 by 0.S3 bushel gives 3.38 bushels as the 

 least significant difference between any two varieties in the tests. 



By using the 3.38-bushel difference in considering the varieties in the 

 test with no border rows removed, it is clear that Marquis and Preston 

 are not significantly different in yield, but are superior to the other 

 varieties in the test. The removal of one or two border rows does bring 

 about any significent changes in the rank of the wheat varieties. 



The standard deviation for the 20 barley control plots is 2.53 ±0.27. 

 On deriving the probable error for the barley tests in the same way as the 

 oats, the result is found to be 0.86 bushel. Multiplying 4.07 by 0.S6 

 bushel gives 3.50 bushels as the least difference between any two barley 

 varieties in the tests which may be considered significant. 



In the test with no border rows removed the Manchuria cross, listed 

 fourth, is significantly lower in yield than the three other varieties and 

 retains that position in the tests with one and two border rows removed. 

 The rank of Manchuria and Chevalier change in the test with one 

 border row removed, but the difference between their yields is not 

 significant. 



In the test of the barley varieties with two border rows removed, the 

 Chevalier yielded 2.44 bushels more than the Manchuria and 3.52 bushels 

 more than the better of the two Manchuria crosses. For this method of 

 test, the Chevalier is not significantly higher in yield than the Manchuria, 

 but may be considered superior to the ]Manchuria cross. It seems that 

 the barley varieties grown in plots surrounded by alleys are not equally 

 efficient in utilizing the additional space. 



The results indicate that, unless there is considerable fluctuation in the 

 response of varieties to border eft'ect when grown in plots surrounded 

 bv alleys, in a 3- or 5-year triaUsuperior types of oats and barley may 

 not be given their true rank unless at least two 6-inch drill rows within the 

 margins of the plots are removed before har\'est. 



