624 PROCEEDINGS OF THE AMERICAN ACADEMY. 



Considered from the point of view of the Kinetic Theory of Gases, 

 the viscosity of a gas is measured by the excess of the quantity of mo- 

 mentum transferred in one direction across tlie mutual boundary of two 

 portions of tlie gas by the to and fro motions of the particles over the 

 quantity of momentum transferred in the opposite direction. The ions in 

 a gas no doubt play their part in transferring momentum from one layer 

 of the gas to another, but their number is so exceedingly small in compar- 

 ison with the nuuiber of unaffected particles that it is scarcely probable 

 that the transfer of momentum is appreciably affected by the ionization 

 of the gas. 



An experiment was performed with the object of ascertaining whether 

 there is any appreciable change in the viscosity of air when in the ion- 

 ized state. No change in the rate of damping of the oscillation could be 

 observed. 



The investigations of Tomlinson and Reynolds, already referred to, 

 seem to be the only ones in which the oscillating body was spherical. A 

 closer examination of these investigations may therefore be permitted. 

 In Tomlinson's experiment two spheres, each 6.3 cm. in diameter, were 

 fixed side by side with their centres 20.78 cm. apart. The system 

 oscillated about an axis, which passed along a vertical line midway be- 

 tween the spheres. The time of one complete swing was 5.76 seconds. 

 The mathematical formula for this case, which was given by Stokes, re- 

 quired a correction for the rotation of the spheres about their respective 

 vertical diameters. When adajited to the case of a single sphere the 

 formula for this correction is that made use of in the present inves- 

 tigation. From the experiment with the spheres Tomlinson found (jq 

 = 0.0001716. 



The results obtained by the same investigator by means of the two- 

 cylinder method, and by means of the single-cylinder method, showed 

 good agreement with each other and with that given for the spheres. A 

 vibrator in the form of a single cylinder was finally adopted and the two 

 results obtained from it were /xq = 0.0001716 and /iq = 0.0001714. 



Where cylinders are used it would appear that there is not the same 

 limitations for the velocity that there are when spheres are used, and 

 consequently, from considerations of what the velocity of a point on 

 either sphere must have been, the result given for the spheres is perhaps 

 hardly wliat one might expect. With an amplitude of oscillation, such 

 that the vibrator would move from tlie point of equilibrium through an 

 angle of 0.02 radians, the maximum velocity of the centre of each sphere 

 must have been about 0.23 cm. per second. The value of Va p/ fx. would 



