222 Proceedings of the Royal Society of Victoria. 



As to the calculation it is to be remarked that the numerical 

 value of the index of friction given in my paper is much too low. 

 This arises in great measure from my having corrected for the 

 residual air in Baily's swings at reduced pressure (about one inch 

 of mercury) on the supposition (which seemed to be conformable 

 to the single experiment that Sabine had made on the subject) that 

 the coefficient of viscosity, the //, of my paper, varies as the 

 density. We know that Maxwell's law, according to which it is 

 independent of the density, is very accurately true in experiment. 

 The true coefficient is now well known for air. I have not got 

 here books of reference, but towards the end of a paper of 

 Tomlinson's in the Phil. Trans., in which he treats of the viscosity 

 of air, you will find collected the numerical results of various 

 observers, himself included. The effect of reducing, in my paper, 

 by a law as to the relation between viscosity and density now 

 known not to be the law of nature was to exaggerate the effect of 

 reduction of pressure, in other words to under-estimate the effect 

 of the residual air, and therefore, in equating the theoretical 

 expression for the difference between thirty inches pressure and 

 one inch in the observed result, to bring out a coefficient which 

 was decidedly too small. The adoption however of the true law, 

 though it raises considerably the coefficient of viscosity as got 

 from Baily's experiments, leaves it still too small. I do not see 

 how to account for this except on the supposition that the motion 

 of the pendulums was not small enough to allow of a strict appli- 

 cation of the formulas of my paper. I have remarked in my 

 paper (at least with reference to a suspending wire, and the same 

 would of course be true generally) that the effect of the formation 

 of eddies would be to tend to throw the effect of the resistance 

 from off' the time on to the arc. Whether any sensible part of 

 the resistance is due to the formation of eddies, may be tested by 

 seeing whether the arc of vibration decreases strictly in geometric 

 progression as the time increases in arithmetic. I examined in 

 this way some of Sabine's experiments in the Phil. Trans., and 

 some of Bessel's experiments with the long and short pendulums. 

 Plotting a curve with the time and log-arc for co-odinates, it came 

 a straight line for the long pendulum, but the curve, though 

 very nearly a straight line for the shorter pendulums, had a 

 sensible though slight curvature. It appears therefore that with 



