38 A RESPIRATION CALORIMETER. 



mixture shaken. The remainder of the potassium hydroxide is then 

 added, and after thorough mixing the reagent is allowed to cool to room 

 temperature. The resultant effect of mixing the two solutions and of 

 the heat generated is generally a slight evolution of gas, which usually 

 remains in the form of small bubbles on the surface of the reagent. 

 Before being used for an analysis these bubbles are withdrawn by con- 

 necting a 10 cc. pipette with the rubber tube R, figure 16, and drawing 

 the froth into the pipette. 



It has been found advisable on mixing up fresh reagent to shake it 

 well and, if possible, allow it to stand some time before use. The 

 quantity of reagent in the small pipette, i. <?., 155 cc., suffices for five 

 oxygen analyses, in which 100 cc. of gas are taken for each sample. 

 Thus the absorbing constant of this reagent is not far from 2.5. 



The usual precautions are taken to prevent the deterioration of the 

 reagent, such as keeping both ends of the pipette closed when not in 

 use. The open end of the reservoir on the pipette is closed with a 

 rubber bag. 



Converting percentage by volume to percentage by weight. As the 

 amount of oxygen and nitrogen admitted from each cylinder into the 

 ventilating air current is determined by weight rather than by volume, 

 it is necessary to convert the percentage composition by volume to that 

 by weight. The percentage is calculated in the following manner : 



Weight of i liter of nitrogen 1.25668 grams. 1 

 Weight of i liter of oxygen = 1.42853 grams. 



Example: In cylinder No. 31089 analysis showed 5.13 percent of 

 nitrogen by volume. 



Grams. 



100 liters of gas contains 5.13 liters of N = 6.43 

 100 liters of gas contains 94.87 liters of O == 135.53 



loo liters of gas weighs 141.96 



6.43 H- 1.4196 = 4.541 percent of nitrogen by weight. 



Computation of percentage of nitrogen by weight by using factors. The 

 percentage of nitrogen in oxygen seldom falls below 2 per cent or 

 exceeds 6 per cent, but the fluctuations are too great to rely on the 

 constancy of composition of a lot of cylinders, even if shipped from the 

 factory at the same time ; consequently it becomes necessar)' to analyze 

 each cylinder. After a few determinations it was found that instead of 

 carrying out the somewhat elaborate calculations just referred to, a fac- 

 tor could be used in calculating directly the percentage by weight from 

 the percentage by volume. Thus, for all samples of oxygen containing 



1 See page 82. 



