G. W. Robinson and W. E. Lloyd 145 



for nitrogen, phosphoric acid, coarse sand and so on. By a well-known 

 formula 



provided that, as may be assumed to be the case, the errors are 

 independent. 



The object of this investigation is to obtain values of p.y, the probable 

 field error, for various soil constituents. 



There is practically no hterature on the subject. The only account 

 of previous work accessible to the writers is in a paper by J. W. Leather^. 

 Leather took duplicate samples from several plots in different localities. 

 Each sample consisted of a mixture of a dozen borings. He found that 

 the differences between determinations on the duplicates of nitrogen, 

 available phosphoric acid and potash respectively, varied from nothing 

 up to more than 20 per cent. Unfortimately the data do not give 

 any basis for determination of probable errors due to field variation''^. 



There are various papers which treat of the probable error of 

 experimental plot yields, but since the composition of the soil is only 

 one factor in determining crop yields, it cannot be expected that these 

 probable errors are any measure of the probable error of samphng. 



Two fields were therefore investigated, one on a drift and the other 

 on a sedentary soil. The following notes describe the two fields. 



Field A, at the College Farm, Aber, near Bangor. This field has 

 been for many years in grass. The soil is glacial drift of local origin. 

 It is not particularly uniform in texture and appearance and for 

 ordinary survey purposes would be reckoned too variable. 



Field B, at Cellar, Aberffraw, Anglesey. This field is now in arable 

 cultivation. The soil is derived from the Pre-Cambrian schists. There 

 is a certain admixture of wind blown sand since the field is not far from 

 the sea. The soil is quite uniform in texture and appearance. 



In Field A, 25 samples were taken. The diagram, Fig. 1, shows 

 the order in which the borings were made. 



Each sample was kept separately in a bag. On arrival at the 

 laboratory all the samples were analysed according to the usual methods. 

 The following analyses were performed: Mechanical analysis, deter- 

 mination of hygroscopic moisture and organic matter, and the 



1 Trans. Ckem. Soc. 1902, p. 883. 



^ Since sending in this paper, tlio authors have seen a paper by Pfeiffer and Blanck, 

 Landtc. Versuchs.-Stat. Lxxvrn. Working on the nitrogen content of an experimental field, 

 they obtained a field error of 2-5 per cent. This was, however, over a small area and 

 the result i.5 scarcely of service for survey purposes. 



