CEREAL IMPROVEMENT SELECTION 185 



In this series of experiments, the selections were made 

 each year from the general crop, being therefore a repe- 

 tition but not continuous. Actually the same number 

 of kernels were sown to each unit of area. The seed 

 was hand-picked, and even small seed was sound and not 

 shriveled. Out of 40 separate tests 37 gave best results 

 from large seed. 

 Besides a larger 

 yield of grain, the 

 large seed also 

 gave better aver- 

 age weights a 

 bushel in all cases. 

 The yield of straw 

 from the large seed 

 was usually better 

 and never less 

 than from small 

 seed. Similarly 

 good results were 

 obtained in 12 

 years of continu- 

 ous selection of 

 light and heavy seed from the corresponding previous 

 selected crop of Joanette oats, the difference increasing 

 constantly (Fig. 61). 



184. Experiments in New South Wales. Investiga- 

 tions of the relative value of large and small seed wheat 

 were made for several years by Cobb (1903) in New South 

 Wales, in which uniformly better results were obtained 

 with large seed. Four different trials were made and in 

 the third trial 29 varieties were employed. Specially 

 made millimeter sieves were used in the grading, and the 



FIG. 61. Pot experiment with Manchuria bar- 

 ley, showing beneficial effect of selecting large 

 seed : pot on left sown with large seed, on 

 the right with small seed. 



