STRAIN ON THE PHYSICAL PROPERTIES OF MATTER, 
811 
the torsional vibrations, this would not be the case ; but, as a matter of fact, even if 
we could start the wire without any pendulous motion, the axis of the wire never 
exactly passes through the centre of mass, and, as a consequence, when the wire 
vibrates torsionally, so-called “ centrifugal force ” causes torsional motion to be 
partially converted into pendulous motion, or vice versd. 
When, as in Experiment I., tire period of pendulous vibration does not approxi¬ 
mately synchronise with the period of torsional vibration, the phenomenon is not very 
pronounced; but when, as in Experiment II., the two periods approach to synchronism, 
it is very apparent. Indeed, when the two periods approach sufficiently to synchro¬ 
nism, and the damping is not great, the phenomenon may become so marked that, 
after a time, almost the whole of the torsional motion becomes converted into 
pendulous motion, and presently reconverted again, so that the torsional vibrations, 
having first diminished to nearly zero, begin to increase again until nearly the initial 
amplitude has been reached, when once more the torsional motion begins to be con¬ 
verted into pendulous motion, and so on. # An interesting fact in connexion with 
this phenomenon is that, if we allow the wire to rest, with oscillations at intervals, the 
rise and fall becomes much less pronounced, in consequence, probably, of the wire 
gradually allowing its axis to pass more nearly through the centre of mass, but we 
can easily reproduce it in all its intensity by jarring the wire,t or even sometimes by 
change of temperature. 
Unless there is nearly synchronism between the pendulous and torsional vibrations, 
there is little or no appreciable effect on the value of the logarithmic decrement, and, 
if we reckon from maximum to maximum, I have found that the logarithmic decre¬ 
ment is sensibly constant, whatever the amplitude, provided the latter does not exceed 
a certain limit. Similarly there will be little error introduced in our estimation of 
the internal friction of the metal from the mean logarithmic decrement deduced from 
a number of trials, which number was in my experiments very considerable. 
When, however, the synchronism between the two periods is sufficient, there may 
be a very serious interference with the mean logarithmic decrement, and even a 
sensible interference with the torsional vibration-period ; and I hold it as a matter of 
some considerable importance that, in all experiments connected with the damping, or 
even, when great accuracy is required, with the vibration-period of torsionally vibrating 
bodies, this point should be well looked to.J] 
* I liad the pleasure of showing this phenomenon to the Physical Society, June 26, 1886. 
t A very slight shock sometimes will effect this. 
X In the case of a magnet suspended with its axis horizontal, the axis of rotation will of necessity not 
pass through the centre of mass anywhere except at the magnetic equator. 
5 L 2 
