vacuum produced in the glass vessel, burnton the surface exposed to the glass vesselfor some seconds, prpdu<3ng a murmuringnoise.

A circular canal one-twenty-fifth of aninch in diameter, an inch and a half in cir-cumference, and one and seven-tenths of aninch deep communicated explosion, but fourconcentric canals, of the same depth anddiameter, and of which the smallest was twoinches in circumference, and separated fromeach other only by their sides, which wereof brass, and about one-fortieth of an inch inthickness, did not suffer the explosion to actthrough them.

It would appear then, that the smallerthe circumference of the canal, that is thenearer it approaches to a tube, the greatermust be its depth, or the less its diameter torender it safe.

I did not perceive any difference in theseexperiments, when the metals of the aper-tures were warmed by repeated explosions;it is probable, however, that considerableelevation of temperature would increase thepower of the aperture to pass the explosion;but the difference between the temperature