1918] Waynick: A Statistical Study of Nitrification in Soil 267 



As regards the effect of seasonal variation upon the accuracy of 

 results which may be obtained at one season of the year upon those 

 secured at another period, no data are as yet available which are 

 extensive enough to warrant any conclusion. It is evident that if a 

 "representative" sample must be found from time to time whenever 

 new work is undertaken, the procedure would become very laborious. 

 The prediction is made that the probable error of the mean as found 

 in any one sampling will be very nearly the same for other samplings 

 taken at other seasons of the year, so that it will only be necessary to 

 make a composite sample of a previously determined number of 

 individual samples and apply the probable error, as previously found, 

 to the results obtained. 



The segregation of the causes of biologic variation in the deter- 

 minations as made, is of importance. Some evidence has already been 

 presented showing the increased variability due to laboratory treat- 

 ment. Variations in moisture present under incubator conditions, the 

 change in the medium due to the addition of various fertilizers, and 

 the mixing of the fertilizer with the soil are all causal factors in the 

 variations between samples as regards the amount of ammonia, nitrate, 

 or similar compounds finally measured. We are attempting, in this 

 laboratory, to determine the extent, and to limit as much as possible, 

 the variations due to manipulation. In the field the only possible 

 cause contributing to increased variability is the method of sampling, 

 and from unpublished results to be reported later the differences be- 

 tween different methods of sampling are of but little moment in a 

 study of field variability. Variations between field samples may only 

 he allowed for as already outlined and are no longer serious sources of 

 error when once recognized. 



There can be no doubt that much of the work done in the past, as 

 regards nitrification at least, is open to very serious question in the 

 light of the results herein reported. Just how much of it can be 

 retained can only be determined by a careful checking of the actual 

 determinations made, taking into account the error due to sampling 

 and to the laboratory manipulations. If the data reported in table 4. 

 for instance, be arranged in any proper order, it will be found that 

 nearly any desired series of results may be secured, agreeing closely 

 in their relative range with any series of results obtained from plots 

 or tumbler treatments which have been reported by various investi- 

 gators. Individual investigations are not referred to here in detail, 

 since their large number precludes their discussion in the limited space 



