95

truth of this view is offered in the effect ofthe cooling agency of metallic surfaces uponvery small flames. Let the smallest pos-sible flame be made by a single thread ofcotton immersed in oil, and burning im-mediately upon the surface of the oil: itwill be found to be about one-thirtieth ofan inch in diameter. Let a fine iron wireof one-one-hundred-and-eightieth be madeinto a circle of one-tenth of an inch indiameter and brought over the flame.Though at such a distance, it will instantlyextinguish the flame, if it be cold : but if itbe held above the flame, so as to be slightlyheated, the flame may be passed throughit. That the effect depends entirely uponthe power of the metal to abstract theheat of flame, is shown by bringing a glasscapillary ring of the same diameter and sizeover the flame; this being a much worseconductor of heat, will not extinguish it evenwhen cold. If its size, however, be madegreater, and its circumference smaller, itwill act like the metallic wire, and requireto be heated to prevent it from extinguish-ing the flame.*

f Let a small globe of metal of one-twentieth of an