1865.] and Viscosity of Metals. 291 



and the other to a fixed rigid body, from which the wire hangs, bearing 

 the vibrator at its lower end. I arranged sets of observations to be made 

 for the separate comparisons of the following classes : 



(a) The same wire with different vibrators of equal weights (to give 

 equal stretching-tractions), but different moments of inertia (to test the 

 relation between viscous resistances against motions with different veloci- 

 ties through the same range and under the same stress). 



(b) The same wire with different vibrators of equal moments of inertia 

 but unequal weights (to test the effect of different longitudinal tractions 

 on the viscous resistance to torsion under circumstances similar in all 

 other respects). 



(c) The same wire and the same vibrator, but different initial ranges in 

 successive experiments (to test an effect unexpectedly discovered, by which 

 the subsidence of vibrations from any amplitude takes place at very dif- 

 ferent rates according to the immediately previous molecular condition, 

 whether of quiescence or of recurring change of shape through a wider 

 range). 



(d) Two equal and similar wires, with equal and similar vibrators, one 

 of them kept as continually as possible in a state of vibration, from day to 

 day ; the other kept at rest, except when vibrated in an experiment once 

 a day (to test the effect of continued vibration on the viscosity of a metal). 



Results. 



(a) It was found that the loss of energy in a vibration through one 

 range was greater the greater the velocity (within the limits of the experi- 

 ments) ; but the difference between the losses at low and high speeds was 

 much less than it would have been had the resistance been, as Stokes has 

 proved it to be in fluid friction, approximately as the rapidity of the 

 change of shape. The irregularities in the results of the experiments 

 which up to this time I have made, seem to prove that much smaller 

 vibrations (producing less absolute amounts of distortion in the parts of the 

 wires most stressed) must be observed before any simple law of relation 

 between molecular friction and velocity can be discovered. 



(6) When the weight was increased, the viscosity was always at first 

 much increased ; but then day after day it gradually diminished and be- 

 came as small in amount as it had been with the lighter weight. It has 

 not yet been practicable to continue the experiments long enough in any 

 case to find the limit to this variation. 



(c) The vibration subsided in aluminium wires much more rapidly from 

 amplitude 20 to amplitude 10, when the initial amplitude was 40, than 

 when it was 20. Thus, with a certain aluminium wire, and vibrator No. 1 

 (time of vibration one way 1*757 second), in three trials the numbers of 

 vibrations counted were Vibrations. Vibrations. Vibrations. 



Subsidence from 40 initial am- ) K r fii R/t 



plitude to 20 J 



And from 20 (in course of the 1 9g Q8 96 



same experiments) to 10 . . J 



