instead of the spark and explosion, we have a silent stream of fire, much more faint and dilute. The fire with which boiling water is impregnated, escapes very easily in vacuo; and the water, in these circumstances, cools much faster, even in a warm room, than if it had been exposed out of doors in a frosty air, which demonstrates, that fire is confined within the body of the water, by the superficial weight and pressure of the air. The outline by which the flame of a candle is so well defined, is owing to some pressure which acts equally on every side; and this pressure can be no other than that of the air, which L may be discovered by carrying the candle forward; in which case, the side of the flame that meets the air will be bright and well defined, while the side that follows will be ragged and more dilute.

The sight is so familiar, that we pay little regard to it; yet it is a matter of astonishment to reflect upon, that a fluid, so weak as fire might be supposed to be, on account of the infinite subtilty of its parts, should be expanded itself, and expand the air with all that force which experiment demonstrates. The medium near a fire is certainly much more refined than at some distance; and it will

5

will be a véry moderate supposition if we imagine the flame of a candle to be twice as rare at its superficies, as the common air of the atmosphere; whence it follows, that the atmosphere will there press upon it with half its force, that is, with a weight equal to seven pounds on every square inch. If there is a sphere of such flame, whose diameter is one foot, the air will compress its surface with a force equal to three thousand one hundred and sixty-four pounds; yet the fire maintains its dimensions with ease against a compressing power, which seems more than sufficient to drive it all back into its central point. This computation is a gross one, because it supposes a limit which does not really exist: it supposes that the air on the surface hath its whole density, while the air of the flame hath but half its density; however, the computation is the best the nature of the case will admit of.

Secondly, air, by its friction, increases the agitation of fire, and thereby raises the heat of it. If its friction against the hall of a thermometer is sufficient to stir up the included fire, and raise the thermometer, it must have a much greater effect of the same kind when the fire is naked and open, and

the

the air can penetrate its very substance. When these two elements meet, the conflict must needs be very great, The air is naturally impelled in a direction toward the centre of the fire, and being there expanded, a considerable part of it is driven upwards by the pressure of that which succeeds; and during all this, the quantity of the fire continually increases with an expansive motion outwards; so that between these two contrary motions, the agitation of the parts of fire, by the friction of the air, must be very great,

But thirdly, experiment shews us, that the ancients were right in the opinion they universally maintained, that air supports fire as a pabulum; that is, that it actually parts with some of its substance to supply fresh matter, increase the fire by being itself attenuated and refined into the condition of fire, However violent the action of air upon fire may be, it is certain that fire must re-act equally upon air, by the common laws of the Newtonian philosophy. Let us suppose the rays of light to diverge after their manner from a burning point; if they move in right lines, they form an infinite number of angles which meet in the point from whence they

proceed.

proceed. When the air enters and mixes with these diverging rays, the nearer it approaches to the centre, the more it compres ses the fire, and is consequently compressed by it. In this conflict the fire is increased, and the air is diminished. The former of these is so well known, that it is superfluous to prove it; the latter will be evident from experiments.

"I set a lighted tallow candle" (says the ingenious Dr. Hales) "about of an inch "diameter, under the inverted receiver zzaa

66

(plate I. fig. 3) and with a syphon I im"mediately drew the water up to 2%. Then zz. drawing out the syphon, the water would "descend for a quarter of a minute, and "after that ascend, notwithstanding the "candle continued burning, and heating "the air for near three minutes. It was ob"servable, in this experiment, that the sur"face of the water zz did not ascend with "" an equal progression, but would be some"times stationary; and it would sometimes "move with a slow, and sometimes with an "accelerated motion; but the denser the "fumes, the faster it ascended. As soon

as the candle was out, I'marked the height of the water above zz, which difference "was

re

was equal to the quantity of air, whose elasticity was destroyed by the burning "candle. As the air cooled and condensed "in the receiver, the water would continue rising above that mark, not only till all was cool, but for twenty or thirty hours after that; which height is kept, though "it stood many days; which shews that the "air did not recover the elasticity which it "had lost. The event was the same when, "for greater accuracy, I repeated this ex"periment by lighting the candle under the "receiver with a burning glass. The ca

pacity of the vessel above zz, in which "the candle burned in my experiment was "equal to 2024 cubic inches; and the elas

ticity of the part of this air was de"stroyed *." The like experiment had been made before, and is mentioned by other authors, to one of whom Dr. Hales has referred us.

I have made many experiments of this kind myself with glass vessels of different capacities, which sometimes had a mercurial gage adapted to them, and sometimes a water gage to shew the progress more minutely. I usually proceeded, as Dr. Hales did, in the former

*Hales's Veg, Stat. p. 230.