In the experiment described in the last paragraph, it will be remembered that the ferment-liquid and the germs employed in its impregnation came in contact with air, although only for a very brief time. Now, notwithstanding that we possess exact observations which prove that the diffusion of oxygen and nitrogen in a liquid absolutely deprived of air, so far from taking place rapidly, is, on the contrary, a very slow process indeed; yet we were anxious to guard the experiment that we are about to describe from the slightest possible trace of oxygen at the moment of impregnation. We employed a liquid prepared as follows: Into from 9 to 10 litres (somewhat over 2 gallons) of pure water the following salts[140] were introduced successively, viz:— Pure lactate of lime 225 grammes Phosphate of ammonia 0·75 grammes Phosphate of potassium 0·4 grammes Sulphate of magnesium 0·4 grammes Sulphate of ammonia 0·2 grammes [1 gramme=15·43 grains.] [Illustration: Fig. 69.] On March 23rd, 1875, we filled a 6 litre (about 11 pints) flask, of the shape represented in Fig. 69, and placed it over a heater. Another flame was placed below a vessel containing the same liquid, into which the curved tube of the flask was plunged. The liquids in the flask and in the basin were raised to boiling together, and kept in this condition for more than half-an-hour, so as to expel all the air held in solution. The liquid was several times forced out of the flask by the steam, and sucked back again; but the portion which re-entered the flask was always boiling. On the following day, when the flask had cooled, we transferred the end of the delivery tube to a vessel full of mercury and placed the whole apparatus in an oven at a temperature varying between 25° C. and 30° C. (77° F. and 86° F.); then, after having refilled the small cylindrical tap-funnel with carbonic acid, we passed into it with all necessary precautions 10 c.c. (0·35 fl. oz.) of a liquid similar to that described, which had been already in active fermentation for several days out of contact with air and now swarmed with vibrios. We then turned the tap of the funnel, until only a small quantity of liquid was left, just enough to prevent the access of air. In this way the impregnation was accomplished without either the ferment-liquid or the ferment-germs having been brought in contact, even for the shortest space, with the external air. The fermentation, the occurrence of which at an earlier or later period depends for the most part on the condition of the impregnating germs, and the number introduced in the act, in this case began to manifest itself by the appearance of minute bubbles from March 29th. But not till April 9th did we observe bubbles of larger size rise to the surface. From that date onward they continued to come in