268 University of California Publications in Agricultural Sciences [Vol.3 



of this paper. It is beyond question that the soil is an extremely 

 variable quantity and that its variability must be measured and taken 

 into account before any determinations made upon it become of value. 



Summary 



A study of variability as regards nitrate production of eighty-one 

 samples of soil taken from a limited area of an apparently uniform 

 soil, is reported. Both surface and subsoil samples were taken. The 

 nitrate present at the time of sampling, the nitrate produced from the 

 soil nitrogen, from ammonium sulphate, and from dried blood were 

 determined. Statistical methods were applied for the interpretations 

 of the results. The following conclusions seem justified from the data 

 presented : 



1. The variability of the field samples of soil, even from an appar- 

 ently uniform area of limited extent, is high and is a factor of extreme 

 importance in an estimation of the reliability of any series of results. 



2. The variability of the samples treated as in the tumbler method 

 for nitrification studies is increased over that found for the nitrate 

 produced in the field. 



3. Subsoil samples vary more in the field, and when treated with fer- 

 tilizers in the laboratory, than surface samples taken from the same 

 area. No explanation of this fact can be offered at the present time. 



4. A single sample of any soil is of little value as regards deter- 

 minations which may be made upon it. 



5. A limited number of samples as ten or sixteen, are subject to 

 wide variations and can only be used when the results are to be inter- 

 preted as having a low degree of accuracy. 



6. A composite sample may be considered as of value only after the 

 probable error to which it is subject is known and this can only be 

 determined by the use of a large number of individual samples. 



7. The distance apart samples are taken is of little importance as 

 long as the samples are uniformly distributed over an area which is 

 apparently uniform. 



8. In the light of these results, the conclusion seems inevitable that 

 much of the past work done, as regards nitrification at least, must be 

 critically examined to determine the degree of reliability, if any, it 

 may have. 



