HOGG. — VlSCOSIlt OF AIR. 613 



aiul has deduced an expression for /a in terms of the radius and length of 

 the! tube, the mean pressure in the gas, and the excess of tlie pressure at 

 one end over that at the other. 



That there are discrepancies in the results obtained is probably ac- 

 counted for, as Meyer has pointed out, by the failure of the experimental 

 methods to conform to the conditions imposed in the mathematical analy- 

 sis. Whatever method is chosen, then, the prime requisite is to be certain 

 that the mathematical conditions are fulfilled. 



As has been said, Maxwell gave a theoretical discussion of the case of 

 a circular plate performing oscillations between two fixed parallel disks 

 and parallel to them. He used the formula obtained to get the value of 

 /x, but his mean result is much larger than that obtained by more recent 

 investigators using in some cases the transpiration method and in others 

 the oscillation method. The probable cause of this was discussed by 

 Stokes in a note appended to Tomlinson's paper on the determination 

 of IX. Stokes showed that unless the swinging disk is adjusted so that, 

 as it swings, it deviates very little from the horizontal plane, there will 

 be an appreciable loss of energy, on the part of the moving disk, due 

 to the crowding of the air between the fixed and moving disks. 



Tomlinson, in the investigation referred to, by oscillating first a long 

 cylinder and then a sliort one whose time of swing was the same as that 

 of the long one, obtained, by eliminating in this way the end effects, 

 practically a cylinder of infinite length, and was thus able to make use 

 of Stokes's formula for the case of an infinitely long cylinder, suspended 

 vertically and performing oscillations about its axis. There does not 

 seem to have been any objection raised to the use of this formula for the 

 determination of /x. His most consistent results were those obtained in 

 this way from a single cylinder, but during the course of the investigation 

 he made use of a system of two vertical cylinders, and of one of two 

 sjiheres suspended side by side. 



More recently F. G. Reynolds used a single spherical shell, and also 

 a single cylinder, and he obtained results from the two methods which 

 agree well with each other, but are considerably larger than the mean of 

 Tomlinson's results, and larger than the result obtained by the latter 

 with a single cylinder. For /x at 0^ C, Tomlinson got 0.0001715, using 

 the single cylinder, while F. G. Reynolds got 0.0001768. 



The seeming completeness and simplicity of the theoretical treatment 

 of the motion of an infinite fluid in which a solid sphere is oscillating 

 about a vertical diameter commends for use in determining fx a pendulum 

 spherical in form, provided it can be shown that, under the given ex- 



