2."i4 University of California Publication.-: in Agricultural Sciences [Vol.3 



The extremes recorded in column two are simply the extreme deter- 

 minations found in any one series. The extreme range may be deter- 

 mined by taking the difference between these two figures. The greatest 

 extremes are shown by the samples to which dried blood was added, 

 with the ammonium sulfate samples a close second. The results for 

 any one series emphasize the very large difference found between a 

 large number of samples treated as uniformly as possible. 



It will be noted that the coefficients of variability as regards the 

 surface samples with their various treatments differ but little among 

 themselves, this difference amounting to only 5.3 per cent. The differ- 

 ence is greater with the four series in which the subsoil was used, being 

 15.7 per cent or nearly three times that of the surface samples. The 

 point is again emphasized here that we are dealing with the summations 

 of all the errors to which the samples are subject, both field and 

 laboratory. One source of error has been allowed for, which is of a 

 purely mechanical nature, namely, that of making the readings with 

 the colorimeter. This error will be considered in the following section, 

 and the coefficients of variability discussed at greater length there. 



The percentage ratios given as the probable error of the mean place 

 all the results on a comparable basis as regards the error to which the 

 various means are subject. Attention is called to the fact that these 

 figures are of similar magnitudes in the four series of surface samples 

 while showing much larger differences in the subsoil samples with their 

 various treatments. 



It will be noted from tables 1, 2, and 3 that in a number of samples 

 there was less actual nitrate nitrogen found in the subsoil samples 

 after incubation without the addition of anything but water, or with 

 the addition of 0.2 gram of ammonium sulphate, than was present in 

 the soil of corresponding samples as they came from the field. This 

 result was not anticipated and no explanation to account for this loss 

 of nitrate nitrogen is offered at the present time. It is, however, 

 regarded as of biochemical interest largely and not of importance as 

 regards the variation between samples at the time the determinations 

 were actually made. 



Error of the Determinations Due to the Colorimeter 



Reference has already been made to the absolute accuracy of the 

 nitrate determination and its bearing upon the results reported in the 

 present paper. Aside from the absolute accuracy of the determina- 



