PHILLIPS. — COMPOSITION OF NATURAL GAS. 77 



in liquid form in steel cylinders was tried. It was found, however, that the 

 gas leaves a considerable volume of unabsorbed residue when treated witii 

 the solution of a caustic alkali. Sodium carbonate was tried instead of 

 marble. A hot saturated solution of the salt was allowed to crystallize 

 in the generator and the mother liquid poured off. The carbon dioxide 

 produced by action of an acid is purer than that from marble, but the 

 evolution of the gas is tumultuous and uncontrollable. Experiments 

 were tried with marble from various localities, but with little success. A 

 marble from Tate, Georgia, was found to yield very pure carbon dioxide 

 after it had been coarsely pulverised and well boiled in water. Another 

 sample of apparently the same rock, from the same locality, yielded 

 after similar treatment a small residue of gas unabsorbed by jwtassium 

 liydroxide solution. A calcite from Lampasas, Texas, in translucent 

 cleavable crystals, was found to yield satisfactory results. The mineral 

 was crushed coarsely, boiled for six hours in water, and then transferred 

 with a portion of the boiled water to the generator. The carbon dioxide 

 evolved proved to be very pure. Experience has shown that no reliance 

 can be placed upon marble or calcite as a source of carbon dioxide 

 because a specimen of apparently the same mineral known to come from 

 the same locality has proved satisfactory. Every batch must be sepa- 

 rately tested as regards the purity of the carbon dioxide which it 

 evolves. 



To overcome the danger of the action of strong alkali upon stopcocks 

 in such work, it was necessary to use an unsaponifiable lubricant. After 

 numerous trials it was found that a mixture, consisting of 70 parts 

 melted rubber and 30 parts unbleached beeswax, softened with a little 

 vaseline, served the purpose quite well, protecting the stopcocks com- 

 pletely. More than 150 determinations have been made in the same 

 eudiometer without accident to stopcocks. 



Some difficulty was experienced in expelling air from the copper oxide 

 when the apparatus was being prepared for work. In beginning a series 

 of determinations several days were often required for the purpose. The 

 porcelain tube was strongly heated, while a slow stream of carbon dioxide 

 was maintained, and the copper oxide was not considered to be in proper 

 condition for use until the escaping carbon dioxide was found to be ab- 

 sorbed without residue by potassium hydroxide solution. About 300 c. c. 

 of the escaping gas were used for the trial. When the copper oxide was 

 freed from air, determinations of nitrogen in natural gas could follow 

 each other until from experience it was shown to be necessary to reoxi- 

 dize the partially reduced copper oxide. About ten determinations 



