METHODS OF MEASURING RESPIRATORY EXCHANGE 31 



Usually the oxygen for feeding a respiration apparatus of large 

 dimensions is taken from a cylinder. This oxygen is never pure and 

 generally contains as much as 2 per cent, of nitrogen. In an experi- 

 ment of long duration the nitrogen percentage in an apparatus may 

 therefore rise appreciably. Oxygen from oxylith generators is pure 

 but expensive, and the generators are inconvenient and wasteful. By 

 special arrangement with the manufacturers of oxygen it is often pos- 

 sible to obtain cylinders with less than -5 per cent, of nitrogen, a per- 

 centage which will be small enough for almost all purposes. 



It is impossible on account especially of the variations in tempera- 

 ture and barometric pressure to maintain the internal atmosphere in a 

 large closed-space respiration apparatus constant with regard to quantity 

 and composition. The oxygen admitted is not therefore an accurate 

 index of the oxygen absorbed by the animal, nor is the carbon dioxide 

 absorbed in the purifiers an accurate index of the quantity produced. 

 The larger the apparatus compared with the respiratory exchange of 

 the animal and the shorter the duration of the experiment, the greater 

 is usually the influence which variations in the internal atmosphere are 

 likely to have. Sampling and analysis of the air at the beginning and end 

 of each experimental period are therefore practically always necessary. 

 Most investigators take small samples of a 100 c.c. or less and use gas- 

 analytic methods, and in the opinion of the writer this procedure is un- 

 doubtedly the safest as well as the simplest. Benedict and Carpenter 

 [1910] have devised a method by which gas analysis is avoided. 

 Before and after each experimental period they allow air to go 

 through a bypass from the piping in front of the large absorbers. 

 They absorb and determine by weight the moisture and CO 2 in this air 

 current, measure the volume by means of a gas meter, and let the air 

 enter the chamber again. When the whole apparatus is absolutely 

 airtight they are justified in assuming that the quantity of free nitrogen 

 in it remains constant except for the known volume of nitrogen which 

 is admitted as impurity with the oxygen. When therefore the total 

 enclosed quantity of air at the beginning and end of each period is 

 calculated, and the corresponding quantities of CO 2 , water vapour and 

 nitrogen subtracted, the rest represents the oxygen, the variations of 

 which can thus be determined without actual analysis. Except for the 

 rigorous tests of tightness to which every part of their apparatus is 

 subjected before and after each experiment this ingenious method of 

 calculation would be very unsafe. 



The results of the analysis, however made, must be multiplied with 



