360 STUDIES ON FERMENTATION. 



The primary condition which enables us to rely on the exact- 

 ness of this method is the fact which we have mentioned above, 

 that a liquid if shaken up with air for one minute will become 

 perfectly saturated with oxygen. Substantially this is the case. 

 In estimating the oxygen in different parts of a liquid treated 

 thus, we have invariably obtained the same figures to within 

 about vTjt^- 



It is true that the variable quantity of the oxygen held in 

 solution in the liquid contained in the part of the tube cb, as 

 well as the oxygen absorbed during the treatment of the liquid 

 in contact with air, constitute causes of error. Experience, 

 however, proves that these causes of error are insignificant, as 

 long as we have to deal with a liquid the aeration of which is 

 not very far removed from the point of saturation, and whose 

 solubility-coefficient for oxygen is not widely difierent from 

 that of water for the same gas. Under such conditions we have 

 always found a constant ratio, to within about ^V^^j between the 

 same liquid and air-saturated distilled water, placed under the 

 same circumstances. 



If, on the other hand, we have to deal with a liquid which 

 holds but a minute quantity of oxj'gen in solution, the causes of 

 error mentioned may very seriously aficct our results, and it 

 will be absolutely nccessar}^ to avoid them. The liquid experi- 



