ti3 



for maintaining this closed nitrogen atmosphere in contact 

 with the water supply, and for compensating its changes 

 of temperature and pressure. 



It has already been explained that the custom- 

 ary method of controlling the soil atmosphere by passing ■ 

 a stream of the desired gas was impossible because of the 

 very large gas consumption which it would have required. 

 On the other hand, simply to replace the normal soil 

 atmosphere with the desired gas and then seal the pot 

 as, for instance, by closing the inlet and outlet tubes, 

 is impracticable because of the considerable variations 

 of the gas pressure inside the pot which accompany 

 changes of temperature. If such a tightly sealed pot 

 be warmed, as at midday, the gas pressure inside be- 

 comes sufficient to blow out the seal or cause impor- 

 tant leakage through it. On the return of lower temper- 

 atures, as at night, the pressure inside she pot becomes 

 lower than that of the general atmosphere and leakage 

 occurs inward. It is obvious that the repetition of this 

 process will pump atmospheric air- in and out of the pot 

 and will destroy the value of the exp< intent* Theoretical- 

 ly there might be^ constructed a seal capable of with- 

 standing this %9mmiL pressure change but praotically it 

 was found impossible to do so, at least with any si - 

 pie modification of the sealing technique described above. 

 It is necessaty, therefore, to have some means by which 



expansion and contraction of the soil gases may be per- 



