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



General Discussion 



It has already been stated that the area under discussion was of 

 very limited extent and as free from apparent variations as any area 

 is likely to be. Also, that the treatment and climatic influences had 

 all tended toward a state of biologic equilibrium as regards nitrate 

 production in this particular soil. With this viewpoint in mind, the 

 importance of applying a statistical interpretation to results obtained 

 from working with a given soil type obtained under conditions much 

 less favorable, becomes doubly important. It is not at all improbable 

 that variations several times greater than those reported will be found 

 in many instances and the number of samples deemed sufficient in this 

 case must be increased to obtain any considerable degree of accuracy. 

 It is possible that areas more uniform than the one used will be found, 

 but it is hardly probable that such will be the case. 



While the present paper deals only with the production of nitrates, 

 not only the variability of the products of microorganic activities, but 

 the chemical constituents of any soil should be studied in a similar 

 manner to determine just how reliable past results in soil investigations 

 may be. It is beyond question that before we can have faith in future 

 results, we must apply the principles outlined. A study of the prob- 

 lems of nitrogen fixation in the soil, which is to be as complete and 

 as carefully controlled as possible, is now under way at this station. 

 Statistical methods are to be applied to the variations in the total 

 nitrogen, total carbon, and nitrates, at least, in two representative 

 soils under observation before a treatment of any kind is applied. In 

 other words, the experimental areas to be used are being standardized 

 as carefully as possible, making due allowance for field variations in 

 the constituents noted above. Further, the field errors due to varia- 

 tions between the samples must not be greater than the experimental 

 error due to the laboratory manipulations, since the increases (or 

 decreases) to be measured are themselves of small magnitude. To be 

 sure, this method of procedure takes much extra time and energy, but 

 it must be employed before the results of any treatments which arc 

 made may be correctly interpreted or recommendations based upon 

 the experimental results safely made. It is hoped that some of the 

 results of these preliminary studies may be ready for publication in 

 the not far distant future. 



