1019] 



field mors. 



145 



A comparison of the Federal v. Minnesota grading system, T. SANDERSON 

 ( Worth Dakota 8ta, Spec. Bui., 5 (1918), Xo. 6. pp. 107-160, tins. 5). A compari- 

 son of the two systems for grading wheat is presented, based upon an analyst* 

 of the Information contained In the two publications noted above and upon 

 the results of numerous milling and baking tests made at the North Dakota 

 College (B. B. R., .'58, p. GG3). 



Variations in seed tests resulting from errors in sampling, O. A. STEVENS 

 (Jour. Ann r. Soo, Agron., 10 (1918). Xo. 1. pp. l-i'.K figs. 3). — Germination and 

 purity tests, made during 1914 and 1915 by the department of botany of the 

 North Dakota Agricultural College, are described in connection with studies of 

 the fundamental factors which occasion variations in seed tests. These varia- 

 tions are said to be largely of two sources, mathematical and personal or 

 economic. The direct causes of variations In purity tests are attributed to im- 

 perfect mixing, random sampling, errors in weighing, personal selection and 

 errors of identification, while in germination tests variations are attributed to 

 imperfect mixing, random sampling, errors In counting, personal selection, im- 

 proper conditions for germination, and special conditions of the seed. These 

 causes are briefly discussed. 



In the experimental work described, 50 germination and purity tests were 

 made from each lot of seed by one individual. Conclusions and recommenda- 

 tions based on the results obtained may be summarized as follows ; 



The probable error of a single germination test of 100 to 400 seeds varied 

 according to the percentage of germination as follows : 



Approximate probable error for germination tests. 



The above figures were increased about one-fifth in the lower percentages of 

 germination for legumes containing the so-called " bard " seeds, the sources of 

 variation being reduced as far as it is possible to do. Xo attempt was made to 

 determine the range of variation where factors other than that of mathematical 

 probability entered to any extent. These values may also be used for other 

 experiments involving similar conditions, e. g., counting 500 seeds to determine 

 the percentage of mixture of two kinds. Samples not containing mixtures of 

 materials which tend to separate readily (such as sand, fine trash, or coarse 

 material), require only a small amount of mixing, while samples which contain 

 such mixtures should receive, when practical, a supplementary test of larger 

 quantity to show the approximate amount of such materials. 



In purity tests the quantities used should receive careful investigation in 

 order to determine whether those in current use could be changed to advantage. 

 The use of the second decimal place is of no value in most work, hut if such 

 accuracy is desired, the test should be based upon a sample of sufficient size. 

 In an instance of such change in some work carried on at this laboratory the 

 following schedule was adopted and the second decimal used: For cereals a 

 measured quantity (about 8 oz. for wheat) was used unless the seed was ob- 

 viously impure, and then the regular quantity (30 gui.) was taken, while for flax 



