314 
PHYSICS: C. BARUS 
Proc. N. a. S. 
disadvantage, except that the highest exhaustions were not available. 
It is of course possible to keep the pump running and the stopcock open. 
When this is done, however, the needle (quite apart from tremor) is 
always in motion, so that gravitation measurements are out of the 
question. 
2. Data. Logarithmic Decrement, X. — In the work of last year the 
logarithmic decrement as obtained from observations largely made at 
night, seemed to remain fairly constant until the highest exhaustions were 
approached. This result needed the further qualification undertaken in 
the present paper. To compute X log e, the two arcs obtained from the 
first three elongations of the needle y — y' , y' — y" were used. In view 
of the specified small leak in the apparatus, the data for X could be found 
with great accuracy in this way in the course of time, while the vacuum 
slowly degenerated from about .0002 to .271. 
These data and the vacua to which they belong (observations being made 
in the morning (A), afternoon (P), and night (N)), are constructed in re- 
lation to the vacua (mm. of mercury) in figure 1. F denotes that the 
attracting weight M, on the right, is in front, and K to the rear of the case 
and needle, the opposite conditions holding for the M near the left end of 
the needle. The first arc of swing was usually between y = 20 and y 
= 30 cm., depending on the time of day. 
Inspection of figure 1 shows that changes of viscosity due to changes of 
pressure, are not discernible in results of the present kind. In fact the 
logarithmic decrement at the highest exhaustions (July 5) happens to be 
larger than at the lowest exhaustions (July 8, N). The logarithmic 
decrement is therefore, also, primarily under the influence of the radiant 
forces. It is largest in the morning, and least at night. It obviously 
oscillates once per day, though I did not make observations after 11 P.M. 
If dO/dt be the change of atmospheric temperature per second in the 
environment of the apparatus, then this coefficient is the controlling 
factor in the marked variations of X log e. 
Moreover the values of X log e are always larger for the rear positions 
R of M, than for the front positions F, under otherwise like conditions. 
One or the other, or their mean, must therefore be taken in association with 
dO/dt. This may be due to a lack of symmetry of the position of the 
needle to the case; but it is more probably due to a lack of symmetry in 
the environment. 
3. First Semi-Periods. — ^The accurate method of finding T/2 from two 
successive passages of the needle through the position of equilibrium is 
very troublesome; for T/2 is long and the position of equilibrium varies. 
Similarly the variations of T/2 axe large, and precise values of it are here 
apparently of little use. Hence the measurement of T/2 from elongation 
to elongation was accepted as adequate. These values, sp far as taken. 
