massive, therefore, the Glass is, the more difficult it is for the caloric safely to be given off. Therefore, when a piece of Glass has to be reheated, as in the case of a lamp Glass, it should be of a medium thickness; not too thick to prevent the heat from quickly travelling through its substance; nor too thin, to be subject to break by the friction or tension of cleaning. The same remarks will apply to tubes intended for steam gauges: which should be subjected to special care and time in the process of annealing. Hot water, or heated sand, would be equally serviceable as an annealing medium: heated air is most convenient for general purposes, although the two former are occasionally used.

A piece of unannealed barometer-tube, forty inches long, measured when just drawn, will become about one-fourth of an inch shorter if annealed; whereas, if quickly cooled without annealing, it will only contract about one-eighth of an inch. Should Flint Glass of unequal substance be insufficiently annealed, such as is not broken in cutting may be placed in a sand and water bath nearly cold, there very gradually heated to the boiling point, and kept several hours in a state of ebullition; the heat may then be reduced, and the whole suffered to get gradually cool. This process has scarcely ever been found to fail.

However carefully tube gauges for steam-boilers may be annealed, even after long use, great danger is caused by cleansing the inside: it is well known that a few grains of sand shaken inside a piece (or proof) of unannealed Glass will cause a violent fracture; and the same effect to a less degree may be produced in an annealed tube. Probably the outside, being more exposed to the lear fire, gets better annealed,