inevitably contain a slight portion of impurity, the causes of the deficiency can easily be understood.
In other experiments made in the same way, but with smaller quantities, more accurate results were obtained: 1 cubic inch of olefiant gas with 12.25 cubic inches of chlorine, produced by the action of light 3.67 cubic inches of muriatic acid gas, 4.963 of the chlorine having been used. 1.4 cubic inch of olefiant gas with 12.5 cubic inches of chlorine produced 5.06 cubic inches of muriatic acid gas, 6.7 cubic inches of chlorine having been used. Other experiments gave very nearly the same results; and I have deduced from them, that one volume of olefiant gas requires five volumes of chlorine for its conversion into muriatic acid and chloride of carbon; that four volumes of muriatic acid gas are formed; that three volumes of chlorine combine with the two volumes of carbon in the olefiant gas to form the solid crystalline chloride; and that, when chlorine acts on the fluid compound of chlorine and olefiant gas, for every volume of chlorine that combines, an equal volume of hydrogen is separated.
I have endeavoured to verify these proportions by analytical experiments. The mode I adopted was, to send the substance in vapour over metals and metallic oxides at high temperatures. Considerable care is requisite in such experiments; for if the process be carried on quickly, a portion of fluid chloride of carbon is formed, and escapes decomposition. The following are two results from a number of experiments agreeing well with each other.
Five grains were passed over peroxide of copper in an iron tube, and the gas collected over mercury; it amounted to 3.9 cubic inches; barometer 29.85; thermometer 54° Fahr. Of these nearly 3.8 cubic inches were carbonic acid, and rather more than .1 of a cubic inch was carbonic oxide. These are nearly equal to .5004 of a grain of carbon. Hence 100 of the chloride would give 10 of carbon nearly, but by calculation 100 should give 10.19. The difference is so small as to come within the limits of errors in experiment.
Five grains were passed over peroxide of copper in a tube made of green phial glass, and the chlorine estimated in the same manner as before. 17.7 grains of chloride of silver were obtained, equal to 4.36 grains of chlorine. This result ap-