EESEARCHES UPON THE CHEMICAL PROPERTIES OF GASES. 223 



5 vols. 2 vols 77.11 per cent. 22.89 per cent. 



3 " 1 " 76.89 " 23.11 " 



7 " 2 " 76.70 " 23.30 " 



4 " 1 " 76.56 " 23.44 " 



5 " 1 " 76.30 " 23.70 " 



6 " 1 " 76.15 " 23.86 " 



9 " 1 " 75.82 " 24.18 " 



From a gravimetric analysis of natural gas, it is easy to determine the relative 

 proportions by weight of carbon and hydrogen in unit volume, and from these the 

 composition may be stated volumetrically in terms of ethane and methane, by the 

 use of the preceding table, and with a fair approximation to the truth. 



It is probable that minute quantities of propane and perhaps higher paraffins 

 occur, but these cannot be identified. The nitrogen and carbon dioxide being deter- 

 mined, the volume of 



CH, + C,H, + C3H3 + 



is obtained as a difference. The error involved in such a method may then be 

 exactly defined as follows : 



The hydrocarbons may consist of methane with traces of propane or of methane 

 with ethane or butane, but the analysis will be stated volumetrically in terms of 

 methane and ethane only. 



As regards the question of fuel value, I have endeavored to show (see Reyort 

 of Oeol. Survey of Penna. for 1886) that the above method will give closely approx- 

 imate results when certain factors relating to available heat of combustion of par- 

 affins are used. 



The gravimetric method affords at the same time a means of control, for it is not 

 only true that in a given volume of a particular paraffin, or of a mixture of paraffins, 

 the hydrogen and carbon will occur in definite quantity, but the ratio ^ is a constant 

 and will be greater as the proportion of higher to lower paraffins is greater. 



These considerations will serve to show the limits of accuracy of the method. 



I^itrogen was determined by passing a measured volume (100 c.c.) over ignited 

 copper oxide contained in a porcelain tube, and then into a eudiometer containing 

 soda solution. By means of a stream of carbon dioxide continued for several hours, 

 the air was expelled from the apparatus previous to the combustion of the gas. In 

 presence of large excess of carbon dioxide, combustion by copper oxide is greatly 

 retarded, and the process must be conducted very slowly to effect complete oxidation. 



Oxygen, as already stated, occurred in too small proportion to allow of a quan- 

 titative determination. 



Carbon dioxide was determined by soda solution in a eudiometer over a mercury 

 trough. 



