12 Tlie Afiiiofip/inr of tJic Snil 



The sif/nijiraitrc of l/ie Jhir/ti/ifians iv compnsilloii Iv llic soil air. 



As alipiuly stated tlio composition of the soil air at any moment 

 is a resultant effect, being the difference between the rate at wliich the 

 carbon dioxide is produced in the soil and that at which it is lost. . At 

 first sight it might appear that the composition must therefore he 

 largely accidental but we have been able to show that it is not, and 

 tiiat the great fluctuations as distinct from the minor ones (p. 33) arc 

 regulated mainly by the rate of production of carbon dioxide in the soil. 

 The method consists in finding some other substance in the soil which 

 is ])rodi(ced in the same manner as the COj, but losi in a different way. 

 If the curve showing the fluctuations of this substance is like the curve 

 for CO, it follows that the fluctuations are largely governed by the rate 

 of production and therefore that the curves given in Figs. 1-5 are 

 essentially production curves. If on the other hand the fluctuations 

 do not resemble those of CO., it follows that the curves are not essentially 

 production curves but that their shape is due to a fortuitous balance 

 of losses and gains. 



The required substance is found in the nitrates of the soil which, 

 like the carbon dioxide, are produced in the decomposition of the soil 

 organic matter by bacteria but which are lost in a wholly different 

 manner. Carbon dioxide is lost by gaseous diffusion, a process wliich 

 proceeds most rapidly in dry conditions when the pores of the soil are 

 most widely open : and least rapidly in wet conditions when the pores 

 are more or less closed. The nitrates, on the other hand, suffer least 

 loss under dry conditions and most loss in wet weather. 



Determinations were therefore made of the amount of nitrate 

 present in each plot on every occasion when samples of gas were drawii 

 for analysis, and the values arc plotted in the curves: unfortunately 

 the necessity for this was not seen when the investigation first began 

 so that no values were obtained during the first four months. 



Inspection of the curves shows that they are all of the same type: 

 there is some displacement in point of time but no difference in character. 

 It follows then that the character of the fluctuations of COj content 

 in the soil air is determined by the rate of biochemical change in the 

 soil. Further proof is afforded by the fact that the curves for bacterial 

 numbers also show a close resemblance to those of COg in the soil air. 



