_ the finished container. It is
convenient to introduce the substance at the stage indicated by _g_. The
important points to observe in making this contrivance are that the
glass rod must be kept hot by working while it is actually in the flame,
and that the skeleton must be made as thin as possible with the
avoidance of heavy masses of glass at any place. If these details are
neglected it will be almost certain to crack.

_Stirrers._--These are usually made from glass rod, and no special
instructions are necessary for their construction, except that the glass
should be in a thoroughly fused condition before making any joins and
the finished join should be annealed slightly by covering with a deposit
of soot, as explained on page 16. The flat ends shown in _a_, Fig. 8,
are made by squeezing the soft glass rod between two pieces of carbon,
and should be re-heated to dull redness after shaping. Fig. 8 also shows
various forms of stirrer.

In order to carry out stirring operations in the presence of a gas or
mixture of gases other than air, some form of gland or seal may be
necessary where the stirrer passes through the bearing in which it runs.
A flask to which is fitted a stirrer and gas seal is shown in section by
_b_, Fig. 8. The liquid used in this seal may be mercury, petroleum, or
any other that the experimental conditions indicate.

[Illustration: Fig. 8]

If the bearing for a stirrer is made of glass tube, it is desirable to
lubricate rather freely; otherwise heat will be produced by the
friction of the stirrer and the tube will probably crack. Such
lubrication may be supplied by turning out the top of the bearing tube
and filling the turned-out portion with petroleum jelly mixed with a
small quantity of finely ground or, better, colloidal graphite, and the
bearing should also be lubricated with the same composition. Care
should be taken not to employ so soft a lubricant or so large an excess
as to cause it to run down the stirrer into the liquid which is being
stirred.

_Leading a Crack._--It sometimes happens that a large bulb or specially
thin-walled tube has to be divided. In such a case it is scarcely
practicable to use the method recommended for small tubes on page 12,
but it is quite easy to lead a crack in any desired direction. A
convenient starting point is a file cut; this is touched with hot glass
until a crack is initiated. A small flame or a bead of hot glass is now
used to heat the article at