Cune) 
If the instrument has onee been calibrated, the scale reading gives 
immediately the density of the gas within; while the thermometer 
and the manometer permit the calculation of the density under 
standard conditions, if the compressibility of the gas is known. The 
calibration may be made either with a single gas whose density at 
various pressures is known with sufficient accuracy for any one 
temperature, or by employing in turn several different gases under 
known conditions of pressure, temperature and density, or with a 
rider. Counterpoising the closed sphere with the hemispherical shell 
of equal surface tends to eliminate errors that would be introduced 
if the apparatus contained a vapor which condensed on the glass. 
The instruments should, of course, be protected from changes of 
temperature by proper jacketing or by immersion in a liquid bath. 
A fixed reference mirror (not shown in the figure) is desirable to 
indicate any change in the leveling of the apparatus. 
In order to get an idea of the sensitiveness that could be expected 
from such an instrument, some rough preliminary measurements 
were made. 
The dimensions were as follows: 
Diameter of bulb 1.0 em. 
5 3 capillary beam Oni 
Length ze u A TAO! sop: 
Mass of entire suspended system 0.67 gms. 
Length of quartz fiber 4 em: 
The apparatus was filled with dry air, and the scale readings 
noted for various pressures ranging from 0.3 cm. to nearly 90 em. 
of mercury. With a fiber about 0,005 em. in diameter and the scale 
255 cm. from the mirror, 0.1 mm. change in the deflection was 
gm. 
liter 
density; and this was the same for all densities tried; that is to say, 
a change of 0.1 mm. in the scale reading indicated a change of 
about one part in 6000 in the density of air under ordinary con- 
ditions. The scale might easily have been placed much farther from 
the mirror and the sensitiveness could have been greatly increased 
by using a larger bulb, a longer beam, and a longer and thinner 
fiber. And since the change in deflection is, in the first approximation 
at least, directly proportional to the change in density, an accurate 
knowledge of the deflections for a few densities is sufficient for the 
calibration of the instrument. Certain corrections, as, for instance, 
found to indicate a change of about 0,0002 change in the 
