The retorts are first heated to a temperature of about 600 deg.  C. or faint redness, then the air pumps, C C, are started.  Air is drawn by them through the purifier, D, where it is freed from carbon dioxide and moisture by the layers of quicklime and caustic soda with which the purifier is charged.  The air is then forced along the pipe, E, into the small air vessel, F, which acts as a sort of cushion to prevent the baryta in the retorts being disturbed by the pulsation of the pumps.  From this vessel the air passes by the pipe, G, and is distributed in the retorts as rapidly as possible at such a pressure that the nitrogen which passes out unabsorbed at the outlet registers about 15 lb. to the square inch.  With the baryta so disposed in the retorts as to present as large a superficies as possible to the action of the air, it is found that in 11/2 to 2 hours—­during which time about 12,000 cub. ft of air have been passed through the retorts—­the gas at the outlet fails to extinguish a glowing chip, indicating that oxygen is no longer being absorbed.  The pumping now ceases, and the temperature of the retorts is raised to about 800 deg.  C. The workman is able to judge the temperature with sufficient accuracy by means of the small inspection holes, H, fitted with panes of mica, through which the color of the heat in the furnace can be distinguished.  The pumps are now reversed and the process of exhaustion begins.  At Westminster the pressure in the retorts is reduced to about 11/2 in. of mercury.  In this partial vacuum the oxygen is given off rapidly, and if forced by the pumps through another pipe and away into an ordinary gas holder, where it is stored for use.  With powerful pumps such as are used in the plant under notice the whole of the oxygen can be drawn off in an hour, and from one charge a yield of about 2,000 cub. ft. is obtained.  With a less perfect vacuum the time is longer—­even as much as four hours.  The whole operation of charging and exhausting the retorts can be completed in from three to four hours.  As soon as the evolution of oxygen is finished, the doors, K, and ventilators, L, may be opened and the retorts cooled for recharging.

The cost of producing oxygen at Westminster, under specially expensive conditions, is high—­about 12s. per 1,000 cub. ft.  When we consider, however, that the cost should only embrace attendance, fuel, wear and tear, and a little lime and soda for the purifiers, that the consumption of fuel is small, the wear and tear light, and that the raw material—­air—­is obtained for nothing, it ought to be possible to produce the gas for a third or fourth of this amount in most of our great manufacturing centers, where the price of fuel is but a third of that demanded in London, and where provision could be made for economizing the waste heat, which is entirely lost in the Westminster installation.  Moreover, in estimating this cost all the charges are thrown on the oxygen; were there any means of utilizing the 4,000 cub. ft. of nitrogen at present blown away as waste for every thousand cubic feet of oxygen produced, the nitrogen would of course bear its share of the cost.