other cases
of ignition; the action and reaction, which necessarily attends the
separation of the constituent principles, exciting probably a vibratory
motion in them.

Since inflammable, air, by agitation in water, first comes to lose its
inflammability, so as to be fit for respiration, and even to admit a
candle to burn in it, and then comes to extinguish a candle; it seems
probable that water imbibes a great part of this extraordinary charge of
phlogiston. And that water _can_ be impregnated with phlogiston, is
evident from many of my experiments, especially those in which metals
were calcined over it.

Water having this affinity with phlogiston, it is probable that it
always contains a considerable portion of it; which phlogiston having a
stronger affinity with the acid air, which is perhaps the basis of
common air, may by long agitation be communicated to it, so as to leave
it over saturated, in consequence of which it will extinguish a candle.

It is possible, however, that inflammable air and air which extinguishes
a candle may differ from one another in the _mode_ of the combination of
these two constituent principles, as well as in the proportional
quantity of each; and by agitation in water, or long standing, that mode
of combination may change. This we know to be the case with other
substances, as with _milk_, from which, by standing only, _cream_ is
separated; which by agitation becomes _butter_. Also many substances,
being at rest, putrefy, and thereby become quite different from what
they were before. If this be the case with inflammable air, the water
may imbibe either of the constituent parts, whenever any proportion of
it is spontaneously separated from the rest; and should this ever be
that phlogiston, with which air is but slightly overcharged, as by the
burning of a candle, it will be recovered to a state in which a candle
may burn in it again.

It will be observed, however, that it was only in one instance that I
found that strong inflammable air, in its transition to a state in which
it extinguishes a candle, would admit a candle to burn in it, and that
was very faintly; that then the air was far from being pure, as appeared
by the test of nitrous air; and that it was only from a particular
quantity of inflammable air which I got from oak, and which had stood a
long time in water, that I ever got air which was as pure as common air.
Indeed, it is much more easy to account for the passing of inf