OF ARTS AND SCIENCES. 43 



of results obtained by a similar apparatus to that used by Meyer. 

 From these Maxwell concludes that the viscosity is independent of the 

 pressure upon the gas, aud that it increases as the first power of the 

 absolute temperature. If, however, the results published in that paper 

 be all upon which this law is based, we cannot regard it as very 

 securely established. A third pa{)er was published by Meyer, in Pogg. 

 Ann. cxliii., 14 ; in which the results of seven experiments with oscil- 

 lating plates after Maxwell's pattern, but with bitilar suspension, were 

 given. These, like the others, are iusutiicient to determine the etFect 

 of temperature. In three subsequent papers* by Meyer a large num- 

 ber of experiments are described. These were made by the method 

 of transpiration through capillary tubes, and preliminary experiments 

 were made to prove the validity of the law of Poiseuille. This law 

 may be expressed by the following equation : — 



where V is the volume of gas transpired in the time t, measured at 

 the temperature of the capillary, and under the pressure p ; the pressure 

 at entering tlie tube being jo,, and at leaving it jo,. The length of the 

 capillary is 1, and its radius R ; ij being the coefficient of viscosity of 

 the gas. This law may, I think, be regarded as established for varia- 

 tions of pressure not exceeding two atmospheres, and for tubes in 

 which the length is very large as compared with the diameter. 



]Meyer gives a series of twenty-five experiments, and selects eleven 

 as the most reliable. These all seem to indicate an increase of viscos- 

 ity with rising temperature greater than the \ power, but appear at 

 the same time quite discordant among themselves. Upon the ac- 

 companying figure, I have shown the extremes of these by a graphical 

 representation. The method used to discuss them is one described in the 

 Proceedings of the Academy for 1874, page 222. If we have a line 

 of the general form represented by the equation yz=.jnx^, we may 

 take logarithms of both sides and get the equation, log y=-n \ogx 

 -\- log m, which has the form of the ecjuation to a straight line. Hence, 

 if we have the coordinates of a series of points which we suppose may 

 be connected by a curve of the exponeutitU form, we may determine 

 this fact by plotting logarithms of these cooi'dinates, which should give 

 us points along a straight line whose tangent is the exponent in the 

 primary equation. Thus, if our equation to the variation of ri with the 



* Pogg. Ann. cxlvui., 1, 203, 526. 



