closed, for 6 weeks, or until it no longer glowed, I found that 1⁄3 of the air had been lost.
19. Third Experiment.—I placed 3 teaspoonfuls of iron filings in a bottle capable of holding 2 ounces of water; to this I added an ounce of water, and
gradually mixed with them half an ounce of oil of vitriol. A violent heating and fermentation took place. When the froth had somewhat subsided, I fixed into the bottle an accurately fitting cork, through which I had previously fixed a glass tube A (Fig. 1). I placed this bottle in a vessel filled with hot water, B B (cold water would greatly retard the solution). I then approached a burning candle to the orifice of the tube, whereupon the inflammable air took fire and burned with a small yellowish-green flame. As soon as this had taken place, I took a small flask C, which was capable of holding 20 ounces of water, and held it so deep in the water that the little flame stood in the middle of the flask. The water at once began to rise gradually into the flask, and when the level had reached the point D the flame went out. Immediately afterwards the water began to sink again, and was entirely driven out of the flask. The space in the flask up to D contained 4 ounces, therefore the fifth part of the air had been lost. I poured a few ounces of lime water into the flask in order to see whether any aerial acid had also been produced during the combustion, but I did not find any. I made the same experiment with zinc filings, and it proceeded in every way similarly to that just mentioned. I shall demonstrate the constituents of this inflammable air further on; for, although it seems to follow from these experiments that it is only phlogiston, still other experiments are contrary to this.
We shall now see the behaviour of air towards that kind of fire which gives off, during the combustion, a fluid resembling air.
20. Fourth Experiment.—It is well known that the flame of a candle absorbs air; but as it is very difficult, and, indeed, scarcely possible, to light a candle in a closed flask, the following experiment was made in the first place:—I set a burning candle in a dish full water; I then placed an inverted flask over this candle; at once there arose from the water large air bubbles, which were caused by the expansion, by heat, of the air in the flask. When the flame became somewhat smaller, the water began to rise in the flask; after it had gone out and the flask had become cold, I found the fourth part filled with water. This experiment was very undecisive to me, because I was not assured whether this fourth part of the air had not been driven out by the heat of the flame; since necessarily in that case the external air resting upon the water seeks equilibrium again after the flask has become cold, and presses the same measure of water into the flask as of air had been previously driven out by the heat. Accordingly, I made the following experiment:
21. Fifth Experiment.—(a.) I pressed upon the bottom of the dish A (Fig.
2) a tough mass, of the thickness of two fingers, made of wax, resin, and turpentine metal together; in the middle I fastened a thick iron wire which reached to the middle of the flask B; upon the point of this wire C, I stuck a small wax candle, whose wick I had twisted together out of three slender threads. I then lighted the candle, and at the same time placed over it the inverted flask B, which I then pressed very deep into the mass. As soon as this was done, I filled the dish with water. After the flame was extinguished and everything had become quite cold, I opened the flask in the same position under the water, when 2 ounces of water entered; the flask held 160 ounces of water. Accordingly, there is wanting here so much air as occupies