INFLUENCE OF SIZE OR WEIGHT OF SEED KERNEL. 



31 



SO with the other classes of kernels. He tabulates his results as 

 follows : 



The relation between yields of the crops representing different 

 sized kernels is so irregular from year to year that suspicion is 

 aroused regarding the accuracy of the results, due to lack of uni- 

 formity in soil. Sanborn's conclusion is that very little, if any, 

 advantage is to be gained by separating seed wheat and planting 

 the large kernels. 



At the Indiana Experiment Station, Latta" conducted experi- 

 ments in which wheat was separated by means of a fanning mill into 

 heavy and light kernels, but impurities and chaff}^ seed were fanned 

 out of each lot of wheat. The experiments were continued three 

 years, but the separations were made each year from seed that had 

 not been so separated the year before. The average gain from the 

 large seed for three years was 2.5 bushels per acre. 



Georgeson,^^ at the Kansas station, seeded plots of land with (1) 

 light seed weighing 56 pounds per bushel, (2) common seed weighing 

 62.5 pounds, (3) heavy seed weighing 63 pounds, and (4) selected 

 seed, obtained by picking the largest and finest heads in the field just 

 before the crop was cut, weighing 61.5 pounds per bushel. Seed was 

 separated each year from wheat not grown from previously selected 

 seed. The average results for three years were as follows: 



Grade of seed. 



Light 



Common. 



Yield of 

 grain 



per acre 

 (bush- 

 els). 



25.19 

 26.57 



Grade of seed. 



Heavy 



Select (average for 2 years) . 



Yield of 

 grain 



per acre 

 (bush- 

 els). 



27.07 

 25.82 



Desprez'' reports experiments extending through three years in 

 which large kernels were selected from a crop grown from large seed 



« Indiana Experiment Station Bulletin 36, pp. 110-128. 

 & Kansas Experiment Station Bulletin 40, pp. .51-62. 



'^ Abstract, Experiment Station Record, 7, p. 679, from Jour. Agr. Prat., 59 (1895), 2, 

 pp. 694-698. 



