772 
ME. H. TOMLINSON ON THE COEFFICIENT OF VISCOSITY OF AIR. 
readings,* the ten corresponding arcs from rest to rest will be o 1 +6 1 , b l -f-a 2 , ci z -\-b. 2 , 
/> e +« 3 , a 3 + & 3 , & 3 +a 4 , a 4 +& 4 , & 4 +n 5 , a 5 + & 5 , & 5 + a 6 . The means of 0 ^+^, ^+a 6 ; 
6i+a 2 , r(--|~ h- ; « 2 + &o, 6 4 +a 5 ; b. 2 -\-a g , a 4 +& 4 , and of a 3 +6 3 , & 3 H-« 4 were written down, 
and if these agreed well with each other, which was almost invariably the case, the 
logarithmic decrement of the mean of the five means was taken. Now, say that 
n single vibrations have taken place betw r een the end of this and the beginning of 
the next set of consecutive readings, the difference between the logarithms of the 
first and second total means will, when divided by n-\- 10, give the mean logarithmic 
decrement for a single vibration. The logarithmic decrement w r as found to be con- 
stant in each experiment within the limit of probable error; the deviations from 
uniformity were sometimes in one direction and sometimes in the opposite, and there 
was no evidence of any law of increase or diminution of the logarithmic decrement as 
the amplitudes decrease. In the intervals between one set and another of the 
readings, taken in the manner mentioned above, other readings were taken for the 
purpose of determining the vibration-period ; the time of transit of the light across 
the centre of the scale, first in one direction and then in the opposite, was recorded 
for ten successive passages by means of a good watch provided with a seconds-hand, a 
similar series being recorded after every 200 vibrations. These last observations 
enabled the period of vibration to be determined with such exactness that we may 
completely disregard any error arising from want of precision in this respect. From 
time to time, at regular intervals, the readings of all three thermometers and of the 
aneroid barometer were taken, so that the mean pressure of the atmosphere, the 
temperature of the wire, and the temperature of the air in the box B could be calcu¬ 
lated with the necessary accuracy. The greatest care was taken that the cylinders or 
spheres suspended from the horizontal bar TV should hang vertically ; also that there 
should be no appreciable pendulous motion of the wire ; if such motion existed it 
was checked by the hand before any of the readings were taken. Very great care 
was also taken in determining the moments of inertia of the vibrator in the various 
experiments, these being each obtained by several different methods,! which gave 
very concordant results. I shall have occasion in a future memoir to dwell on the 
various sources of error to which determinations of moments of inertia are liable ; so 
it wall suffice, perhaps, here to mention that this part of the work alone occupied my 
entire attention for nearly two weeks. The following five experiments, or rather 
sets of experiments, were made :— 
Experiment I. 
I'lie wire was of well-annealed copper, 97 centims. in length and 0'06272 centim. 
in diameter. Two cylinders, each having a mass of 70T9 grammes, were used. These 
* This number was always taken. 
f The moments of inertia could be calculated with sufficient accuracy from the dimensions and mass 
of the vibrating system ; they were, however, determined also indirectly by the two methods employed 
by Maxwell. 
