OF GASES AT HIGH EXHAUSTIONS. 
399 
than to air. Before this discovery many months were wasted in taking useless 
observations, owing to the presence of hydrogen as impurity, and several pieces of 
viscosity apparatus and pumps were taken to pieces to find out a supposed leakage. 
The best way to purify mercury is to violently agitate it with a solution of per- 
chloride of iron. The mercury is thus rapidly converted into a pasty mass of metallic 
globules in so fine a state of division as to be invisible to the naked eye. The globules 
are washed, and then dried by squeezing first through a cloth, then through chamois 
leather. In these operations the minute globules coalesce, and the metal reassumes 
its fluidity. 
651. When the high exhaustions are first approached in the apparatus another pre¬ 
caution is necessary. The pump is kept going till an exhaustion of about 0 - 5 M is 
attained, and the whole apparatus, including the connecting-tubes and those belonging 
to the pump, are heated to between 300° and 400°, by passing over them the flame of 
a Bunsen burner. The apparatus itself is then kept at a temperature of about 300° 
for half an hour or more, and the pump is worked the whole time. Glass condenses 
on its surface a certain amount of permanent gas which is let out very slowly and 
incompletely in a vacuum; but the gas is driven off, in quantity sufficient to depress 
the gauge, when the temperature of apparatus and pump is quickly raised to 300° 
or 400°. 
The relation between the viscosity and pressure in the apparatus after this heating 
is not quite the same as before, showing that the gas liberated has not quite the 
composition of atmospheric air. It is therefore necessary to exhaust again to a very 
high point, and then to add a little air (some should always be retained in the air-traps 
for this purpose) and again exhaust. After this treatment it is practicable to attain 
much higher exhaustions. 
Another very necessary precaution is to allow a sufficient time to elapse, after the 
pump has stopped, before the observations are taken. At low exhaustions the pressure 
in the apparatus equalises in a few minutes ; but when the tension is reduced to a few 
millionths of an atmosphere, twenty minutes or half an hour must pass before the 
pressures in the McLeod globe and the viscosity bulb are uniform. The normal pres¬ 
sure is always considered to be 760 millims. When the apparatus is full of gas the 
interior pressure is easily brought to 760 millims., whatever be the height of the 
barometer, by raising or lowering the mercury reservoir, and so varying the height of 
the mercury in the tube supplying the fall tubes. 
652. A noteworthy point in connexion with the elasticity of glass is observed on 
the curves in diagram C. They are not continued beyond the 0‘02 M exhaustion, 
but the general form of the curves indicates that, if they were produced beyond the 
limits of the observations, they would cut the line representing the absolute vacuum. 
The curves representing the repulsion accompanying radiation evidently go up to 
the zero point, showing that at an absolute vacuum there would be no repulsion. 
The curves of viscosity cannot, however, be supposed to end at the zero point 
