studies on oat breeding. 115 



Work in 1914. ' 



In 19 14, 31 of these pure lines were tested again. Three 

 lines, viz., 139, 243 and 249, were discarded because they pos- 

 sessed certain undesirable characters. The data from the 1913 

 test were not sufficient to warrant discarding any of the other 

 lines. By this time practically all of the lines possessing weak 

 straw, poor head character, or poor grain had been discarded. 

 The further selection in these pure lines would have to be 

 largely upon the question of yield. Consequently two or three 

 years records are necessary. 



Owing to a large amount of other work in 1914, it was not 

 possible to plant more than two plots of each pure line. Two 

 lines were represented by only a single plot. These pure line 

 plots alternated in the field with the plots of standard commer- 

 cial varieties. They were therefore tested under exactly the 

 same conditions so far as that was possible. 



In connection with these tests a method has been devised to 

 correct the yields of individual plots for differences in the soil. 

 Before discussing the results of the 1914 tests it will be neces- 

 sary to consider this method very briefly. 



The method used by us in testing varieties is to grow several 

 (usually 4) systematically repeated plots of the same variety. 

 These plots are each 33 feet square, or 1-40 of an acre. Where 

 four of these are grown it makes 1-10 acre devoted to each vari- 

 ety. It has been clearly proven that much more satisfactory 

 results are obtained by growing several systematically repeated 

 small plots of one variety than by growing a single large plot. 

 With single plots one of these may fall on better ground and 

 give a much better yield for that reason, although the intrinsic 

 yielding ability of the variety may not be as good as some of 

 the others which were on poorer soil. Where several plots are 

 grown and scattered over the field there is much less chance of 

 all of them falling in very good or very poor soil. 



Even where several plots are grown there is some chance 

 that one or two of these will fall in exceptionally good or ex- 

 ceptionally poor soil and these may unduly affect the average 

 yield. To take account of this, a method has been devised by 

 which the yield of each plot can be corrected for differences in 



