Researches tjfon the chemical properties of gases. 2l5 



V. COMPOSITION OF NATURAL GAS. 



The gas used in the following trials was that supplied to Allegheny by the 

 Allegheny Heating Company, and is the product of wells scattered over a consid- 

 erable gas-producing area. It may be said to represent the average composition of 

 an enormous volume of gas. 'No important diiferences have been observed during 

 the period from 1886 to 1892 in the heating or illuminating power of the gas as 

 supplied to the city, except that the odor of petroleum (^. e., of higher paraffins) has 

 been occasionally stronger. 



Tests have also been made of gas from various localities in Pennsylvania, ISTew 

 York and Indiana, and Vancouver, British America, and also at Cleveland, O. In 

 all cases where possible the tests were made at the wells. When this could not be 

 done, it was necessary to use samples brought in glass vessels to the laboratory. 

 In such cases, the samples were examined for oxygen before being subjected to the 

 tests. As a leak in a sample vessel invariably causes an interchange of air and gas, 

 so that air enters in proportion as an escape of gas occurs, much dependence is to 

 be placed on the presence or absence of air in a gas sample as a criterion of its 

 purity. 



IITDROGETST. 



Hydrogen is almost always mentioned in the published analyses of natural gas 

 I have made the following chemical tests : The natural gas, as supplied to Allegheny 

 by the Allegheny Heating Company, was caused to flow through a solution of palla- 

 dium chloride for periods varying from ten days to three months. Five hundred 

 cubic feet have been used in a single experiment. Similar tests have been repeated 

 at various times between January, 18S6, and May, 1892; but in no case was a trace 

 of precipitation observed in the palladium chloride solution. I^atural gas was found 

 likewise to be without action upon solutions of platinum chloride and ammoniacal 

 silver nitrate. A stream of natural gas has been j)assed through dry pure palladium 

 chloride. This extremely delicate test has failed to show the presence of hydrogen 

 even in traces, although tried repeatedly during the period from January, 1886, until 

 May, 1892. As already stated, the results of my study of gas reactions show that 

 palladium chloride produces very different effects according as it is used dry or in 

 solution. Palladium chloride dry is reduced promptly by dry hydrogen when the 

 gas is used in a free state. 



