6 



(888 per ounce) with Mediterranean wheat, was 2.06 bushels 

 in favor of large grains as compared with the commercial 

 sample, aad 5.18 bushels in favor of large grains over 

 small grains. The difference in yield between the large 

 grains and the commercial sample chiefly occurred the first 

 year; but it is possible, though hardly probable, that the 

 difference was partly due to variation in the soil. The 

 experiment has been carried out in different parts of the 

 field for the last two year4 and the difference in the yield 

 is now only 0.32 in favor of the large grains." 



Cobb (13) gives a summary of the comparisons of the re- 

 sults of some very carefully conducted experiments to deter- 

 mine the relative advantages of large seed and small seed 

 of wheat. He used sieves with rectangular openings of meas- 

 ured width which give a fairly accurate grading on the basis 

 of diameter of kernel. Three grades are cited in the summary: 

 very large, large, and medium which were respectively 3.25, 

 .00 and 2.50 millimeters in diameter. 



If the comparison is made on the basis of yield of grain 

 and straw, the following is a general statement of the results; 



The very large or 3.25 grade excelled in 66.7$ of the ttials. 

 " 3.00 89.7$ n " " 



medium n 2.50 98.1$ " " " 



* 



This statement fails to tell the whole story for the 

 reason that the excess of yield in the majority oases is 

 greater than in the minority cases. Thus 



Grade When excelling did When excelled 



so by per cent was only by $ 



Very large or 3.25 14.7 8.5 



Large or 3.00 30.4 12.3 



Medium or 2.50 44.5 3.5 



In these statements the basis of the percentage cal- 

 culations is the weight of the lower of the two constants. 



