n which it again
extinguishes candles, and that in which it finally becomes fit for
respiration, depends upon some difference in the _mode of the
combination_ of its acid with phlogiston, or on the _proportion_ between
these two ingredients in its composition; and it is not improbable but
that, by a little more attention to these experiments, the whole mystery
of this proportion and combination may be explained.

I must not omit to observe that there was something peculiar in the
result of the first experiment which I made with nitrous air exposed to
iron; which was that, without any agitation in water, it was diminished
by fresh nitrous air, and that a candle burned in it quite naturally. To
what this difference was owing I cannot tell. This air, indeed, had been
exposed to the iron a week or two longer than in any of the other
cases, but I do not imagine that this circumstance could have produced
that difference.

When the process is in water with iron, the time in which the diminution
is accomplished is exceedingly various; being sometimes completed in a
few days, whereas at other times it has required a week or a fortnight.
Some kinds of iron also produced this effect much sooner than others,
but on what circumstances this difference depends I do not know. What
are the varieties in the result of this experiment when it is made in
quicksilver I cannot tell, because, on account of its requiring more
time, I have not repeated it so often; but I once found that nitrous air
was not sensibly changed by having been exposed to iron in quicksilver
nine days; whereas in water a very considerable alteration was always
made in much less than half that time.

It may just deserve to be mentioned, that nitrous air extremely rarified
in an air-pump dissolves iron, and is diminished by it as much as when
it is in its native state of condensation.

It is something remarkable, though I never attended to it particularly
before I made these last experiments, and it may tend to throw some
light upon them, that when a candle is extinguished, as it never fails
to be, in nitrous air, the flame seems to be a little enlarged at its
edges, by another bluish flame added to it, just before its extinction.

It is proper to observe in this place, that the electric spark taken in
nitrous air diminishes it to one fourth of its original quantity, which
is about the quantity of its diminution by iron filings and brimstone,
and also by liver of su