Electrons concerned in Metallic Conduction. 259 



to 2*44, and writing K = ft 2 , this should involve a dielectric 

 capacity of 5'9, of: like magnitude with all the values in the 

 table. But no element is known with a dielectric capacity, 

 as deduced from the refractive index, greater than about this 

 magnitude. We note that K is the dielectric capacity for 

 steady currents, if the moving ions are all electrons. 



A mode in which negative values of K might arise has 

 been suggested in (14). If this be the cause in the case of 

 tin, magnesium, and cadmium, the effect concerned is not large, 

 for the negative values are small. We may therefore, appa- 

 rently, still rely upon the estimates of p given by the first equa- 

 tion, the absorption of selective type being very small. But it 

 must be admitted at this stage that the Maxwellian hypothesis 

 has given an excellent account of Drude's experiments in so 

 far as wo have considered them. The agreement is very 

 exact, for a small divergence from this law can, as we shall 

 see, produce a great change in the necessary value of K. 

 It would seem that the value of K is capable of deciding 

 between the hypotheses, and of determining the exact extent 

 to which any hypothesis is correct, and if this be granted, we 

 may claim to have shown that the formulae developed in this 

 paper are mathematically sufficient, within the limits of ex- 

 perimental error, to account for the optical behaviour of all 

 these metals in sodium light. 



Before proceeding to examine the results of the other 

 hypothesis, let us consider what the limits of experimental 

 error probably are. For an analysis of Drude's results may 

 be pushed further, and yields very profitable conclusions. 

 In several cases, for example, silver, he has given the results 

 of several experiments performed on the same metal, prepared 

 in different ways. Thus for silver we have the following 

 particulars : — 



(1) For silver made solid by pressure, 



n % = 3-65, n=-169, % = 21'7. 



(2) For silver deposited hj electrolysis from the double 

 cyanide of silver and potassium, 



w% = 3'62, n = -170, % = 21-2. 



(3) For silver chemically prepared from the pure nitrate, 



nX=3'69, n = -206, x=18'0. 



(4) For the chemically prepared silver, after heating, 



n % = 371, n = -180, %=20-6. 

 The mean of all the experiments is 



h X = 3-67, n=-181, x = 20 * 2 - 



S2 



