20

municate with the external air; for ex-plosion will be communicated by any aper-ture, however small, provided it be suffi-ciently heated. This circumstance is shewnin a very elegant manner in burning con-centrated mixtures of oxygen and hydrogenat the end of a long tube of one-sixtieth orone-seventieth of an inch in diameter, whenthe experiment begins, (the tube being cool)there is no danger; gradually, however, asit becomes heated, the combustion steals asit were down the tube, and at last reachesthe reservoir of the gases.

Where one set of air feeders only are at-tempted in a lamp, they should present an.uniform surface, so that the radiating powersof the metal, and the cooling powers of theexternal gases, may immediately balancethe heating powers of the internal gases;and where the radiating tissue is connectedwith other parts of the lamp, these partsshould be so massy as to be slightly heatedonly, and present no means for a gradualaccumulation of heat. Wire gauze, as itoffers a greater extent of radiating surfacethan perforated metallic plates, is the bestmaterial for the guard of lamps; and by