558 Adeney — Dissolved Gases and Fermentative Changes. 



3. The changes of temperature which the solutions undergo during the time of 

 keeping them in bottles, as described, out of contact with air. 



It would be difficult to give a jDrecise value for each of these three sources of 

 error, but their total efPect upon the final results obtained, which are, it must be 

 remembered, the differences between the determinations for each dissolved gas 

 before and after keeping the liquid under the prescribed conditions, will be found 

 approximately indicated in the majority of the experiments which I shall describe 

 in this Paper, by the atmospheric nitrogen determinations. 



Since most of my experiments have been carried on under more or less aerobic 

 conditions, the atmospheric nitrogen has not, except apparently in a few cases, 

 suffered changes in common with the other bodies with which it has been 

 associated. The difference between the volume of this gas before and after 

 fermentation, therefore, affords a very valuable index of the extent to which the 

 above possible sources of error affect the analytical results. 



The total error may also be found practically by blank experiments with the 

 water employed for making up the solutions. 



I have usually found the difference between the determination of the volumes 

 of nitrogen before and after keeping to be under 0'2 c.c. per litre of water. In 

 some cases it has risen nearly to 0"3 c.c, but this is exceptional. 



Blank experiments with tap-water, the particulars of which will be given later 

 on with the details of my other experiments, show that the oxygen determinations 

 before and after keeping do not usually differ by more than 0-1 c.c, and rarely 

 differ by as much as 0'15 c.c. per litre of water. 



The error attending the determination of carbon dioxide varies somewhat with 

 the quantity, and probably also with the composition of the mineral matter in 

 solution in the water. When the quantity is not greater than 40 or 50 ccs. per 

 litre, the error is about the same as that affecting the nitrogen determination, but, 

 as the quantity increases, the error is liable to slowly increase also, and may reach 

 to as much as I'O c.c. per litre for quantities amounting to 200 ccs. per litre. In the 

 case of a hard spring water in which the carbon dioxide reached to nearly 300 ccs. 

 per litre, the determinations of this gas before and after keeping showed differences 

 of as much as 2 ccs., although the atmospheric nitrogen and oxygen determinations 

 were similar, and indicated that no fermentation had gone on therein during the 

 time of keeping it. In consequence of this, I have avoided, when possible, the 

 presence of any large quantity of carbon dioxide, and have employed sodium 

 carbonate only, when it was necessary to add a carbonate to my test solutions. 



The errors of the gas analysis apparatus are very small, and need not be 

 discussed, since I have dealt with them in the Paper already referred to, describing 

 the earlier form of the instrument. 



