38 



THE BOOK OF WHEAT 



selected parent sort yielded only 22.5 bushels, an increase dur- 

 ing four years of 5.8 bushels per acre. In ten years nearly 25 

 per cent in yield was gained. 1 



Ninety-six tests of selected wheat seed during the years 1900 

 to 1902 at the Canada experiment farms gave an average gain 

 of about 3.6 per cent in favor of selection. 2 Principles differ- 

 ing somewhat from those usually followed in selection were 



CROSSING AS A CAUSE OP VARIATION I 



Yield in grain of 100 plants, showing greater variation in yield of hybrid than 

 of parents. Yield of hybrid shown oy x line. (After Hays.) 



utilized by Lyon. 3 His selections were for quality rather than 

 quantity. He experimented with the smallest and lightest ker- 

 nels on account of their high nitrogen content. Heavy seed 

 planted at the rate of 1.5 bushels per acre gave a greater yield 

 of wheat the first year than light seed sowed at the same rate. 

 Selecting heavy seed grown from the heavy wheat and light 

 from the light wheat, the difference in yield in 3 or 4 years was 

 small. After the first year of the separation, the light seed 

 gave much the greater amount of proteids per acre. Lyon 

 points out, however, that proteid nitrogen is no index to the 

 amount of gluten, which is the better basis for improvement. 

 It is not yet decided whether selection should be for plants with 

 large heads or for plants with a large number of medium-sized 

 heads. In general, the results of many experiments seem to 

 favor the selection of large seed. 4 



1 Hays, Plant Breeding, p. 10. 



3 Evidence of Wm. Saunders, 1903, p. 48. 



U. S. Dept. Agr., Bu. of Plant Indus., Bui. 78 (1905). 



Hunt, Cereals in Amer. (1904), pp. 87-89. 



