We tested equilibrium by first starting with sea-water of a known lower tension than the gas mixture, and making a second deter mination on sea-water of a known higher tension than the gas mixture. If the final P„ was the same in both cases we concluded that equilibrium had been sufficiently approximated, but in all later experiments we about doubled this time.
In making the electrometric reading the electrode was simply inverted so that the sea-water ran down into the large tube, the ungreased stopcock immersed in the KC1 bath, and a platinum-tipped wire hooked into the platinum loop to connect to the potentiometer. If the reading was taken with the disk out of the water, the Hy and partial pressures were the same as in the original gas mixture, but if (11.5\ the disk was immersed, 11.5 was added to the Hy pressure and k bar 1 was added to the pressure in making the calculations. The pressures were always corrected for vapor tension of water before the final calculations.
The gas mixture was made in the apparatus shown in figure 17, holding about 25 pounds of mercury. This consisted of a liter separa tory funnel with a narrow tube and 3-way cock attached at the top, and wrapped with wire passed through hot sealing-wax to increase its strength. By graduating the narrow tube in tenths of a cubic centi meter and making a mark, M, near the lower end of the reparatory funnel to denote a total volume of a liter, it was possible to measure the in parts per 10,000. The was washed with solu tion and the Hy with solution and again in in wash bottles shaped like pyknometers, and passed through traps (0, fig. 17) to insure atmospheric pressure. The apparatus was filled with mercury by raising the leveling bulb by means of a rope passing through a pulley in the ceiling. The CO2 apparatus was attached to the 3-way cock and the air was washed out of the connections. The 3-way cock was now turned as in figure 17, so that entered the narrow tube when the leveling bulb was lowered and the lower stopcock of the separatory funnel was opened cautiously. When the required amount of was admitted, the lower cock was closed, the 3-way cock reversed, and the apparatus substituted. The was washed out of the connections and the 3-way stopcock turned as in figure 17 and H2 admitted by opening the lower stopcock. The operator kept his hand on this stopcock until
the filling was complete, because the H2 came faster at first than later, and there was danger of the mercury leaving the separatory funnel faster than the H2 was supplied, in which case water would be sucked in from the trap O. It was necessary to allow a little H2 to escape continuously through the trap 0, as a matter of safety. When the mercury meniscus reached the lower mark, the cocks were closed and the apparatus shaken, the remaining mercury stirring and mixing the gas. The tonometer electrode was now attached, the leveling bulb was raised, and the gas mixture was passed through the tall column of sea-water in fine bubbles, which was accomplished only after a careful adjustment of all the stopcocks.
For the determination of the of sea-water taken directly from the sea, the electrode shown in figure 18 was used. It was first filled with H2 and then sea-water was admitted through a to the mark m, after washing out the air at b. The H2 was shaken with 10 c.c. of sea-water in the large compartment; then by tapping the apparatus it was caused to enter the smaller compartment and again shaken. In this way the loss of CO2 by the second portion of sea-water was minimized. The ungreased stopcock was immersed in the KC1 and the reading was taken in the usual manner.
The calomel electrodes were made from mercury redistilled in Hulett's still (and plated on platinum so as to be portable), KC1 was recrystallized many times in quartz beakers, and calomel was made by the electrolytic method of Lipscomb and Hulett. The 0.1 n KCl calomel electrodes were made at 20° and hence have a very slight error at 10° and 30° due to volume change of the KCl solution in them. The KC1 was weighed with standard weights and the weighings reduced to vacuo. A saturated KC1 calomel electrode was often used as an intermediate on account of its lower resistance, but each time was compared against the 0.1 n electrode. The saturated electrode often varied about 1 millivolt, although the readings were never taken in less than 3 hours after the room was brought to constant temperature, after a maximum fluctuation of about 2 degrees.