SHAPE AND SIZE OF PILOTS IN TESTS OF VARIETIES OF GRAIN. 75 



error is sufficient to lead one to conclude that even one shaking 

 in settling the grain manifests itself in a higher mean and a 

 smaller variability. 



In all of these methods it is evident that the standard devia- 

 tion and the coefficient of variation are absolutely small. While 

 shaking five times lowers the variability of the mean weight 

 determinations as is shown in a reduction of 38.5 percent in 

 the standard deviation and 44.9 percent in ithe coefficient of 

 variation, the mean weight per bushel increases 11.6 percent. 

 On account of its very low standard deviation and coefficient of 

 variation Method IV 'has been adopted by this Station as the 

 one giving the greatest accuracy in determinations of the bushel 

 weight 'of oats. In our tests of varieties of oats the practice 

 is to take for each variety the mean oi three weighings as the 

 weight per measured bushel. This gives us according to the 

 results derived from data presented in this paper a measure- 

 ment possessing a very high degree of accuracy. Method I we 

 understand is the one ordinarily practised by grain dealers, and 

 the data shown herein warrant the conclusion that determina- 

 tions derived thereby also possess a high degree of accuracy. 

 However, as already shown, this degree of accuracy is very 

 much less than that attainable in Method IV. 



The use of the. standard bushel measure in getting the weight 

 of seeds heavier than oats would undoubtedly show less varia- 

 tion in the determinations. 



SUMMARY. 



Data presented in this paper indicaite the necessity of follow- 

 ing a particular scheme of handling the standard grain tester in 

 determining the weight per bushel of grain. 



Of the four metliods tried the one wherein the grain is poured 

 into the bucket and settled by shaking five times gives the most 

 accurate results. This method in comparison with that involv- 

 ing no settling of the grain lowers the standard deviation 38.5 

 percent and the coefficient of variation 44.9 percent. 



Settling the grain not only decreases the range of variability, 

 thus giving more accurate results, but also increases the mean 

 weight per bushel. 



In conclusion it can be said that the standard grain tester as 

 a means of determining the weight per measured bushel of 

 grain gives results possessing a high degree oi accuracy. 



