204 Mr. E. B. Rosa on the Specific 



longer inversely proportional to /e 2 , but decreases more rapidly 

 than k 2 increases. 



Thus the conditions which the function a x must fulfil are 

 not only what we should not expect, but are actually incon- 

 sistent. Turning now to the other alternative, namely, that 

 tf is zero (and therefore a and a x are zero), we have the force 

 expressed by F = AKV 2 , an expression which applies equally 

 to dielectrics and electrolytes. In both cases K is independent 

 of A and Y, as we have seen. It changes little if any with a 

 variable conductivity, but changes notably with the temperature. 

 The only other determinations of temperature-coefficients 

 which I have seen are those of W. Cassie *, made at the 

 Cavendish Laboratory. He worked with a number of solid 

 and liquid dielectrics, for the most part between about 15° 



and 50° or 60° C. Here are his mean values of t^-7m : — 



Solids. 



Mica . . . . + -0003 



Ebonite . . . +'0004 



Glass . . . +-0012 



... +'0020 



Liquids. 



Turpentine . . —'0011 



Carbon bisulphide — *0040 



Glycerine . . . —'0056 



Benzoline . . . -'0011 



Benzine . . . -'0013 



Olive oil . . . --0024 



Paraffin oil . . 4- -0023 



We notice that the solids and paraffin oil have positive 

 coefficients, while all the other liquids have negative co- 

 efficients, like water and alcohol. Further, the values of the 

 temperature-coefficients for water and alcohol are not only of 

 the same order of magnitude as those for the given liquid 

 dielectrics, but both ( — '0038 and — '0046) fall within the limits 

 of the series, water having a value less than carbon bisulphide 

 and glycerine and greater than turpentine, benzoline, benzine, 

 and olive oil. 



VII. It is very convenient to think of the force acting on 

 the needle as due to a given charge tending to flow down a 



dV 



given slope of potential. Letting — represent the resultant 



slope of potential, the displacement over the surface of the 



dV 

 needle is AjK-^-, A 1 being a constant depending on the 



needle, and K a constant depending on the medium. One 

 can just as readily imagine this displacement occurring in a 

 conducting as in a non-conducting medium. Indeed, Maxwell 



* Proc. Roy. Soc. vol. xlvi. (1889), p. 357, and Phil. Trans. 1889. 



