95 



TWO-ROWED BARLEY. 



The importance of allotting each sort three or more plots is clearly shown 

 in this table, as is also the necessity of giving the actual weight of grain per 

 acre instead of the number of bushels when making a comparison of sorts of 

 different weights. 



An essential feature in breeding work is the correct interpretation of 

 results obtained from comparative trials. Where many sorts must be under Importance 

 test for a series of years and where new sorts are gradually being brought in f corre(: ' i 

 and inferior sorts dropped, a reliable basis of comparison is imperative. *! rpr f 

 Thus the average yield of sorts under trial from 1890 to 1906 cannot fairly 

 be compared with the average of others which came in, say, three years 

 later. The manner in which this difficulty is sought to be overcome at 

 Svalof is shown in the following table of autumn wheat yields (39, p. 231) : 



