42 PROCEEDINGS OF THE AMERICAN ACADEMY 



tance between the centres of two molecules at impact. The value of t] is 

 expressed in units of length, mass and time, since it is a tatigentifJ 

 force. This formula, if true, shows that the viscosity of any gas sliould 

 be independent of its density at a constant temperature, and should in- 

 crease proportionally to the value of u. But u'^ is proportional to the 

 absolute temperature, whence we see that the viscosity should increase 

 proportionally to the square root of the absolute temperature (which 

 we may reckon from — 273*0.). Maxwell has also pointed out * that 

 in this expression we should obtain the same result with regard to the 

 pressure, whatever ast^umption we adopt of the mutual action at impact 

 of the molecule ; but tiiat it is necessary to make some special assump- 

 tion upon the nature of this action to determine the variation with the 

 temperature. 



Previous to this deduction by Maxwell, there had been but little 

 work done upon the viscosity of gases, and almost nothing as to its 

 variation with temperature. Subsequently, experiments have beeu 

 made by Meyer, Maxwell, Puluj, and v(m Oliermayer. The forms of 

 apparatus used have depended upon two fundamental methods : I*', the 

 retardation of pendtdums by the surrounding ga-es ; 2°, the transpira- 

 tion of gases through capillary tubes. In the present pai)er, I propose 

 to discuss somewhat the value of these experiments in determining the 

 variation of the viscosity with the temperature, and to describe some 

 recent experiments made with a modification of the second of the 

 above methods. 



In a paper published in Poggendorff's Annalen, exxv., 177, 1865, 

 O. E. Meyer describes a series of experiments upon the internal friction 

 of air made by measuring the retardation of three circular glass plates 

 oscillating around a vertical axis in a closed receiver containing the gas, 

 whose temperature and pressure could be varied. From the results 

 of these measurements, Meyer concludes that the coefficient of viscosity 

 is independent of the pressure. It will, however, be evident, upon an 

 inspection of the published results, — especially by application of the 

 graphical method, — that no reliance can be placed ui)on them for de- 

 termining variation with the tempeiature. IVIeyer's second paper 

 (Pogg. Ann. cxxvii., 199, 353) is devoted to a discussion of Graham's 

 transpiration experiments,t from which we may derive quite a satis- 

 factory proof of the law of Poiseuille as a|)plied to gases. In the 

 Philosophical Transactions, London, 186G, Maxwell published a series 



* Pliil. Mag. XXXV., 211. 



t riiil. Trans. Roy. Soc. Lond. 1846-49. 



