ure
of the flame any longer than necessary_. This may be accomplished by
doing all heating, shrinking, etc., of the glass in a flame more or less
tinged with yellow, and only raising the temperature to the highest
point when ready to blow the glass. This kind of devitrification is
apparently due to volatilization of the alkalies from the glass in the
flame, and it is said that it can be partly remedied or prevented by
holding a swab of cotton saturated with a strong solution of common salt
in the flame from time to time as the glass is heated.

The toughness of glass, _i.e._, its ability to withstand variations of
temperature, depends on its composition and the care taken in its
annealing. In general, large pieces of glass should be heated very
slowly in the smoky flame, and the larger the diameter of the tube the
greater the length which must be kept warm to prevent cracking. All
large pieces should be carefully heated over their whole circumference
to the point where the soot deposit burns off, before being finally
cooled. After being thus heated they are cooled in a large smoky flame
until well coated with soot, then the flame is gradually reduced in size
and the object finally cooled in the hot air above it until it will not
set fire to cotton. If thought necessary, it may then be well wrapped in
cotton and allowed to cool in the air. If not properly annealed the
place heated may crack spontaneously when cold, and it is quite certain
to crack if it is reheated later.

Next in importance to the glass are the blow-pipe and the bellows. Any
good blast lamp, such as is ordinarily used in a chemical laboratory for
the ignition of precipitates, will be satisfactory; provided it gives a
smooth regular flame of sufficient size for the work in hand, and when
turned down will give a sharp-pointed flame with well-defined parts.
Where gas is not available, an ordinary gasoline blow-torch does very
well for all operations requiring a large flame, and a mouth blow-pipe
arranged to blow through a kerosene flame does well for a small flame.
Several dealers make blow-torches for oil or alcohol which are arranged
to give a small well-defined flame, and they would doubtless be very
satisfactory for glass-work. Any good bellows will be satisfactory if it
does not leak and will give a steady supply of air under sufficient
pressure for the maximum size of flame given by the lamp used. A bellows
with a leaky valve will give a pulsating f