184 
cover the surface of the piston until it ultimately will rise 
through the tubular openings aa into the lower part of GG', 
and so raise the valve ring UV. 
At this stage cease to pour in the mercury and make a 
downward stroke with the piston, and again raise it by an 
upward rnovement. 
If the level of UV remain the same as before, or even if it 
should float on mercury, there will be good reason to conclude 
that sufficient mercury has been poured into it; but if the 
mercury does not rise sufficiently to float the ring, more must 
be added in the same manner, moving the piston up and 
down from time to time, so as to expel bubbles or adherent 
air. 
When satisfied that sufficient has been supplied, place the 
glass receiver (which I will assume to contain eight ounces 
of air, and the capacity of the cylinder to be equal to it) on 
PP', and make a downward stroke of the piston. At this 
moment the tension of air in Z, the receiver, and in AB, the 
cylinder, ought to be equal, and hence one-half the quantity 
in Z will pass through LV probably during the stroke, but 
certainly as soon as the steel ball has been left by the sinking 
of the mercury to fall by its own weight. 
Granting that the quantities of air in each are equal, 
though having only a density equal to one-half of the original, 
we commence the upward stroke, the steel ball LV is floated 
against TT', and reflux is prevented. Soon after the piston 
has passed half w'ay, its contents are condensed so as to 
equalise tension with the external air, and thus in the piston’s 
continued course from this point, the ring valve UV is raised 
until the whole air has passed, and even some small surplus of 
mercury. 
The concavity of the lid, the surface of the piston and the 
piston rod, are now full of mercury to the exclusion of air, 
and the next downward movement of the piston produces 
a vacuity which compels the half residuum in Z again to 
