810 
MR, H. TOMLINSON ON THE INFLUENCE OF STRESS AND 
Remarks on Experiments I. and II. 
It is worth while to pause here and to consider what we can glean by a careful 
examination of these two experiments, because they are typical of most of the others, 
whose number now can be reckoned by hundreds. I had in my mind rather a strong 
conviction that there was an intimate relationship between the loss of energy from 
internal molecular friction and the electrical resistance of a metal, and have already 
observed to that effect.* Accordingly, regarding this apparent relationship as a 
matter of considerable importance, I spared no pains in attempting to get accurate 
values for the coefficients of viscosity of metals, such as have been obtained by 
Maxwell and others in the case of fluids. All the wires mentioned in this paper 
have been examined over and over again ; nearly all have been set up a second time, 
and most of them a third time, and, whilst in no case has a wire been under examina¬ 
tion for less than a week, in some instances the experiments have extended over 
several weeks. As the work progressed, the conviction became more and more forced 
upon me that I was dealing with something very distinctly different from what is met 
with in fluid friction. In Experiment I. we see apparently that one of the laws of 
fluid friction is exactly carried out, for according to the seventh column, in which is 
given the logarithmic decrement for one vibration, the diminution of range per equal 
numbers of oscillations bears a constant ratio to the diminishing range, the departures 
from the mean value of the logarithmic decrements being in no case greater than '3 
per cent. A closer scrutiny, however, shows us that this is only the case because, in 
calculating each of the logarithmic decrements, the mean of a great number of 
oscillations is taken.t If we look into each set of six arcs, we may notice that 
in some instances two consecutive arcs will show no diminution whatever, whilst 
in others the proportionate diminution is much more than the average. These 
are not the results of errors of observation, because the differences are appreciably 
greater than any which could be accounted for in this way, and when we turn to 
Experiment II. the phenomenon is more marked ; we have most indubitable evidence 
of a continued rise and fall of the logarithmic decrement. This rise and fall would 
probably escape notice, were we able to take the mean of as many oscillations as were 
taken with copper ; moreover, if we commence a fresh experiment altogether, we find 
a total mean logarithmic decrement agreeing, within the limits of errors of observation, 
with the previous one. 
[Note added Sept. 18, 1886.—I have since discovered the cause of this rise and 
fall: it is due to the fact that it is impossible to ensure that the axis of rotation 
always passes exactly through the centre of mass. Of course, if the wire were 
perfectly flexible, and we could entirely avoid pendulous vibrations when we started 
* £ Phil. Trans.,’ Part 1, 1883, p. 168. 
f It will he observed that in Experiment 1. the mean logarithmic decrement for 70 oscillations is 
taken in all the series except the last, where we have the mean for 210 oscillations given. 
