92 



VARIATION IN COMPOSITION OF 



The reducmg valve of the compression cyHnder was con- 

 nected with a gas-meter, which in turn was attached to a bunsen 

 burner. Samples of the gas were obtained for analysis from a 

 three-way tube placed between the cj'h.nder and the gas-meter. 

 The gas was allowed to flow from the cylinder at a constant rate 

 of 120 liters per hour. Since each division on the scale of the 

 main pressure gauge of the cylinder corresponded to 62.5 pounds 

 per square inch, samples of gas were taken for analysis after 

 each reduct'.on in pressure of this amount. Determinations of 

 carbon dioxide, illuminants, oxygen and carbon monoxide were 

 made in the usual way. Owing to lack of apparatus for the 

 separation and estimation of hydrogen, methane and ethane, 

 these components could not be separately determined. In 

 Table 1 are recorded the amounts of carbon d'oxide, illuminants, 

 oxygen, carbon monoxide and the mixed lighter components 

 (H2 + CH4 + C2H2 + NP) found in samples of gas taken at var- 

 ious cylinder pressures. 'These results are shown graphically 

 in Figure 1 . 



Table 1 



Component 



Percentage Composition 



At a 



pressure (in lbs. per sq. in.) of : 



Average 



312.5 



250 



187.5 



125 



62.5 1 15 



CO, 



3.8 



4.5 



5.2 



5.8 



6.4 



6.9 



5.4 



111. 



8.2 



9.8 



lt.2 



12.9 



14.8 



16.6 



12.2 



Oj 



1.0 



0.8 



0.9 



0.7 



0.8 



0.7 



0.8 



CO 



21.2 



23.2 



25.3 



28.1 



30.5 



32.0 



26.7 



H. + CH4 -f 



1 















C.H, +N., 



■ 65.8 



61.7 



57.4 



52.5 



47.5 



43.8 



55.0 



It will be observed from the results contained in Table 1, 

 that the averages of carbon dioxide, oxygen and carbon monox- 

 ide agree closely with the amounts of these components con- 

 tained in the gas before compression. The low average for 

 illuminants and the consequent high average for the residue of 

 hydrogen, methane, ethane and nitrogen are quite possibly due 



