534 Mr. W. Sutherland on a 



between zero and 320° C, we can get the true latent heats 

 we require by subtracting ic «T 2 from Person's values. 

 Thus we can compare our theoretical values of X with the 

 experimental in the case of six metals. The following Table 

 contains the values of pi and p 2 , Person's value of the latent 

 heat X 1? the true latent heat X 2 , and the value X 3 calculated 

 on the supposition that the molecules are invariable : — 



Ag. Zn. Cd. Hg. Sn. Pb. 



o x 9-92 6-85 8-366 14-19 7*183 11-00 



P , 9-51 6-48 7-99 13-69 6-99 10-64 



\ 21-1 28-1 13-6 2-8 14-2 5*4 



X, 16-2 24-5 12-5 2-8 12-8 4-8 



X 3 81-0 117-0 53-3 9-7 46-0 23-6 



Eatio X 3 /X 2 . 5-0 4-8 4'3 3'2 3-6 4-9 



In the case of silver it was not safe to extrapolate for 

 specific heat up to the melting-point of 954° G. from experi- 

 ments up to 320°, so I made a determination of its mean 

 specific heat between 20° and 860° in comparison with that 

 of gold, which is known ; the result was "06 63, which, over a 

 range of 934°, gives 61'9 calories in place of Person's 57. 

 In the case of tin I used Tomlinson^s determination of the 

 specific heat, as Naccari omitted it. The ratio X 3 /X 3 comes 

 out nearly constant, especially if we omit mercury as uncer- 

 tain on account of the smallness of X, and tin as uncertain on 

 account of its tendency to crystallize ; the mean value of the 

 ratio for the other four metals is 4*75. In each case the cal- 

 culated value of X is about 4*75 times the experimental, and 

 here again, while the actual equation and the assumption of 

 inalterability in the molecules are not verified, the form of the 

 equation is confirmed in a striking manner if we remember 

 that the latent heat ranges from 2*8 for nercury to 28 for 

 zinc. In other words equation (10) asserts that 



2c(p 1 —p 2 )/3bp i 'k 



is constant on purely theoretical grounds, and the tabulated 

 comparison shows it to be constant, only that the value of the 

 constant is, on the average, 4*75 instead of 1 ; still it is pro- 

 mising confirmation for the kinetic theory that it should be 

 constant at all. 



We may state, then, that collections of unaltering mole- 

 cules, having the coefficients of expansion of the different 

 metals, would, on the kinetic theory, have bulk-moduluses 



