CAPACITY FOR HEAT OF METALS AT LOW TEMPERATURES. 
349 
Much weight should not be attached to the fact that the calculated values are 
slightly higher than the observed ones in the case of lead at about 200° C. abs., since 
there is an uncertainty in the value of C p — C y , due to lack of knowledge of the 
variation in the coefficient of compressibility with temperature. At 80° C. abs., the 
calculated value for this metal is also too low. 
Only for the metals iron and sodium does the difference between the observed and 
the calculated values increase considerably with the temperature. 
The divergence in the case of iron must, to some extent, be associated with its ferro¬ 
magnetic properties. 
From the work of Weiss and Beck* it appears that about 5 per cent, of the specific 
heat of iron at room temperature is due to the potential energy of demagnetization of 
an internal field which, according to the theory advanced by Weiss, exists in ferro¬ 
magnetic bodies. The magnitude of this term increases rapidly in the neighbourhood 
of the temperature at which spontaneous magnetization disappears (about 750° C.), 
amounting there to nearly 40 per cent, of the total. 
In the case of sodium, as the melting point is approached, molecular changest take 
place and the theories of Einstein and Debye are not sufficient. 
(3) Comparison of the Experimental Results with the Formulce of Einstein, Nernst 
and Lindemann and Debye, when the Frequencies Inserted give Coincidence 
at a Temperature of about 125° C. abs. 
If, instead of requiring the formulas to fit the experimental curves at the lowest 
temperatures, we take as basis of comparison the range of temperature through which 
our investigation extends, we find that there is a closer approximation of the theoretical 
values to the experimental results, and it is notable that the various formulas give 
results in fair agreement among themselves over this range. 
In the comparison below, values of the frequencies have been chosen to bring the 
calculated values into coincidence with the experimental values at about 125° C. abs., 
which, on the average, was the lowest point to which we carried our experiments. 
It will be observed that, generally, near the boiling-point of liquid hydrogen, 
Einstein’s formula gives values which are too low; from Nernst and Lindemann’s 
the values are too high ; while Debye’s formula gives values which are in fair agree¬ 
ment for Al, Ag, Zn, and Pb, and, in the other metals, agrees with the experimental 
values better than either Einstein’s or Nernst and Lindemann’s. At liquid air 
temperatures all three formulae give values which are too high. 
* ‘ Journ. de Rhys.,’ col. 7, p. 249, 1908. 
t A. E. Oxley (‘Proc. Cam. Phil. Soc.,’ vol. XVII., p. 450, 1914), has suggested that a magnetic 
energy term similar to that of Weiss and Beck could explain abnormally high values of the specific heat 
near the melting point. 
