32 CIRCULAR NO. 109, BUREAU OF PLANT INDUSTRY. 



FIELD TESTS IN ENGLAND. 



Wood and Stratton, of Cambridge, England, have published some 

 results relative to the probable error of field experiments. Their in- 

 vestigations covered numerous observations on wheat, barley, oats, 

 mangels, potatoes, and other crops. They observed that there was 

 no significant difference between the probable error of one of these 

 crops and that of any other. This is in a way supported by the close 

 agreement found at Huntley with sugar beets and alfalfa, groups C 

 and D, where the same field was used for loth trials. But Wood and 

 Stratton take 5 per cent as the probable error of a single plat in 

 carefully conducted field experiments. 1 The facts that at Huntley 

 and Scottsbluff widely different probable errors were found for the 

 same crop (oats) and that the probable errors determined at the two 

 farms are considerably greater than 5 per cent indicate that to take 

 5 per cent as " the probable error of field experiments " is not generally 

 justified. 



CONCLUSIONS. 



The material here presented may be used to emphasize the follow- 

 ing points: 



(1) Probable error determinations are valuable as a measure of 

 the reliability of results of field experiments and of the confidence 

 which may properly be placed in conclusions drawn from such results. 



(2) The value of experiments would be greatly increased if the 

 probable error applicable to the field were determined in advance; 

 and where the experiments contain several plats receiving the same 

 treatment it is desirable to determine each year the probable error 

 of the results obtained on such plats, as in this way the effect of 

 seasonal factors on the probable error may be estimated. 



(3) In conducting field experiments all possible precautions should 

 be taken in order to minimize accidental errors. 



(4) It is always desirable to conduct tests in multiplicate, so as to 

 be able to compare averages rather than single results. 



(5) In interpreting and publishing experimental results the prob- 

 able error should be given consideration; otherwise, insignificant 

 differences may be unduly emphasized. 



1 " The probable error of field experiments is investigated by two independent methods, and found to 

 be about 5 per cent of the crop."— Journal of Agricultural Science (Cambridge), v. 3, pt. 4, p. 440. De- 

 cember, 1910. 



[Cir. 109] 



ADDITIONAL COPIES of this publication 

 -ii- may be procured from the Superintend- 

 ent of Documents, Government Printing 

 Office, Washington, P. C. , at 5 cents per copy 



