The impregnation took place on February 10th, and on March 15th the liquid was nearly saturated.  The bubbles then began to lodge in the bent part of the exit-tube, at the top of the flask.  A glass measuring-tube containing mercury was now placed with its open end over the point of the exit-tube under the mercury in the trough, so that no bubble might escape.  A steady evolution of gas went on from the 17th to the 18th, 17.4 cc. (1.06 cubic inches) having been collected.  This was proved to be nearly absolutely pure carbonic acid, as indeed might have been suspected from the fact that the evolution did not begin before a distinct saturation of the liquid was observed. [Footnote:  Carbonic add being considerably more soluble than other gases possible under the circumstances.—­Ed.]

The liquid, which was turbid on the day after its impregnation, had, in spite of the liberation of gas, again become so transparent that we could read our handwriting through the body of the flask.  Notwithstanding this, there was still a very active operation going on in the deposit, but it was confined to that spot.  Indeed, the swarming vibrios were bound to remain there, the tartrate of lime being still more insoluble in water saturated with carbonate of lime than it is in pure water.  A supply of carbonaceous food, at all events, was absolutely wanting in the bulk of the liquid.  Every day we continued to collect and analyze the total amount of gas disengaged.  To the very last it was composed of pure carbonic acid gas.  Only during the first few days did the absorption by the concentrated potash leave a very minute residue.  By April 26th all liberation of gas had ceased, the last bubbles having risen in the course of April 23rd.  The flask had been all the time in the oven, at a temperature between 25 degrees C. and 28 degrees C. (77 degrees F. and 83 degrees F.).  The total volume of gas collected was 2.135 litres (130.2 cubic inches).  To obtain the whole volume of gas formed we had to add to this what was held in the liquid in the state of acid carbonate of lime.  To determine this we poured a portion of the liquid from the flask into another flask of similar shape, but smaller, up to the gaugemark on the neck. [Footnote:  We had to avoid filling the small flask completely, for fear of causing some of the liquid to pass on to the surface of the mercury in the measuring tube.  The liquid condensed by boiling forms pure water, the solvent affinity of which for carbonic acid, at the temperature we employ, is well known.  This smaller flask had been previously filled with carbonic acid.  The carbonic acid of the fermented liquid was then expelled by means of heat, and collected over mercury.  In this way we found a volume of 8.322 litres (508 cubic inches) of gas in solution, which, added to the 2.135 litres, gave a total of 10.457 litres (638.2 cubic inches) at 20 degrees and 760 mm., which, calculated to 0 degrees, C. and 760 mm. atmospheric pressure (32 degrees F. and 30 inches) gave a weight of 19.700 grammes (302.2 grains) of carbonic acid.