498 
PHYSICS: C. BARUS 
Proc. N. a. S. 
In the figure, a line has been drawn through a succession of points 
which seem to suggest it. For this g/m increases per day (owing to the 
decrease of m) at a mean rate of r = 0.00039, which is equivalent to and 
air loss of 5 X 10 ~^ grams per day. Though it was formerly my opinion 
that the phenomenon could be completely explained in terms of the diffu- 
sion of air through water, the cusps (s) and subsequent intervals of re- 
lapse (r) when temperature increases, again do not bear this out. The 
whole is more complicated. The mechanism by which the air loss is brought 
about is largely solutional. Air is dissolved when temperature falls 
and is released from solution when temperature rises again. But for this 
the rate of increase oi g/m throughout many months is in all the results 
remarkably constant and could be allowed for. Only at the beginning of 
the experiments with a new diver, and owing probably to gas adhering 
to the glass walls, are these mean rates irregular. 
But the solutional effect in question under rapidly varying temperatures 
is fatal to the purpose for which the experiments were made. Though 
much of the discrepancy could be removed by placing the whole apparatus 
in an adequately constant thermostat, the outlook is not encouraging. 
^ Advance note from a Report to the Carnegie Institution of Washington, D. C. 
The work was done in deference to a request of President R. S. Woodward, that a variety 
of methods of the kind in question be looked into. 
2 Carnegie Publications, No. 185, Washington, 1913. 
NOTE ON TORSIONAL MEASUREMENT OF VARIATIONS OF 
THE ACCELERATION OF GRAVITY BY INTERFERENCE 
METHODS' 
By Cari, Barus 
Department of Physics, Brown University 
Communicated July 1, 1920 
1. Apparatus. — ^These experiments were undertaken to find in how far 
interferometer methods might contribute to the measurement of changes 
of g, under conditions in which the pendulum is inapplicable. The apparatus 
as a whole is a horizontal torsion balance with the deflection readable in 
terms of the displacement of the achromatic interference fringes. The 
method developed would be applicable in case of any ordinary chemical 
balance. The interferometer described admits of moderately rough hand- 
ling; but it requires a level base, at least during observation. The mir- 
rors are adjusted normally to a vertical plane, as the rays are in this di- 
rection. 
Torsional weighing, by passing the counterpoise from pan to pan, is 
in itself a delicate operation when fringes are used. Moreover, the varia- 
tion of viscosity is particularly large at the beginning. Nevertheless 
