Serviss — Internal Temperature Gradient of Metals. 463 



covers them (see fig. 1), and may be relied upon to protect 

 them from so small an effect as that due to the passing of the 

 small quantity of air transferred by convection. 



In the early part of this paper (p. 453), I called attention to the 

 fundamental change I made by placing all my junctions be- 

 tween the cylinders in order to expose both rows equally. I 

 wish now to explain why this change was made. 



Let us compute the gradient in a sphere of radius c, in 

 which b calories are generated per gram per second. Take 

 p = density, k = conductivity, h = surface conductivity. Then 

 when a steady state has been set up, 



4 s * 



— tit pb = 



A 27 Cljt 



dr 



r=- 



3k dt 

 pb dr 



r 2 



2 



-~rt + C 



pb 



Integrating, 



To determine the constant of integration, 



4 

 when r ==■ c, — -jrr 3 pb = ^irc 2 hkt c • 



_c P b 



c ~ 3hk 



2 A 



and the temperature at any point is given by the equation 



r 2 3k c 2 c 



Y~~p~b t+ Y + T 



from which it appears that the gradient is a parabola. 



Strutt* in a determination of the ionizing power of common 

 materials, states "radium is 100,000 times more active than 

 uranium, and uranium 3000 times more active than the most 

 active material I have experimented with. So that one part of 

 uranium in 300,000,000 would suffice to account for the 

 observed effects/' His results show as wide a range for differ- 

 ent specimens of the same metal as for different metals. So 

 although he did not try iron, we may reasonably suppose that 

 it would give an effect similar to that set for the maximum ; 

 the order of magnitude is all that can be shown by such a com- 

 putation as this. Curie and Labordef found that radium 



*Phil. Mag. (5). v, pp. 680-685. 1903. 

 fComptes Eendus, cxxxvi, p. 673, 1904. 



