HOGG. — VISCOSITY OF AIR. 619 



justed so that (he mirror should come to rest in the proper place, and 

 time had been given for tlie amplitude of the oscillation to diminish to 

 the desired magnitude,* the points of rest of the consecutive arcs were 

 observed. No observations were made until the arc had diminished to 

 about 15 cm. The scale distance was 2.3 metres, so that the largest angle 

 through which the sphere turned from the position of equilibrium was 

 about 0.015 radians. Every point of rest was observed until the ampli- 

 tude of oscillation had diminished to about 5 cm. A series of twenty 

 consecutive arcs could thus be obtained. By turning the cap, D, a fresh 

 start was given to the vibrator and the same process repeated, care being 

 taken to allow any pendulous motion to subside before any observations 

 were made. Generally, from four such series the mean decrement was 

 obtained. Before removing the sphere from the vessel the time of swing 

 was determined. The pressure was measured at the beginning and end 

 of the experiment, and the temperature observed from time to time dur- 

 ing the course of the work. 



The suspended system was then taken from the vessel, the sphere and 

 disk removed, and the lead disk fixed on the wire at the same place as 

 that previously occupied by the other disk. When the new system had 

 been returned to the vessel and the swinging of the disk had sufficiently 

 subsided, the same process as was carried out with the sphere was re- 

 peated with this new pendulum. The time of swing, the pressure, and 

 temperature were again observed. 



These observations gave all the data necessary for the calculation of fx 

 with the exception of the moment of inertia of the vibrator. To get 

 tliis, three cylinders were made having different radii, and each a length 

 sufficient that its weight should be just that of the lead disk. The 

 weights were made the same to a milligram. Thus the fibre when sup- 

 porting any one of the cylinders was under the same strain as when it 

 was supporting the sphere and disk. The moment of inertia of each 

 cylinder was calculated, and then a knowledge of the time of swing of 

 each and of that of the wire and mirror when allowed to oscillate alone 

 gave at once the value of the moment of inertia of the mirror and wire. 

 This was added to the moment of inertia of each cylinder, and then from 

 each result the moment of inertia of the vibrator was obtained. The 

 moment of inertia of the vibrator could be calculated, but it seemed bet- 

 ter to get it from the time of swing which the vibrator had in the course 

 of the experiment. 



* This process usually took between one and two hours. 



