434 Mr. W. Sutherland on the 
ratios k/yMc is added to show proportionality under 
discussion. 
Al. Cu. Ag. Au. Ni. Zn. Cd.” Pb. Sn Pi, Pde 
10Mc... -58 60 61. 64 64 62 63 65 66 63 63 GL 65 
110 111. 112 114 109 108 112 110 112) 120 1205235 
By taking account of the molecular heat we get a decidedly 
more accurate relation between thermal and electrical con- 
ductivity than is given by the Wiedemann-Franz law. Now 
this improvement on that law is in reality contained in the 
theory just given of thermal conduction, for we assumed 
provisionally “for simplicity that a is constant in the equation 
mv?/2=a0. But for the metals we should take «@ to be pro- 
portional to Mc. Then (22) gives at once that k/yMe is to 
be the same for all metals, as we have just found it to be to a 
considerable degree of approximation. Of course the atomic 
heat at 18° and 100° ought rightly to enter into the com- 
parison of &/y at these temperatures, and would perhaps 
improve the results of Drude’s comparison. 
Riecke and Drude make some enterprising and elaborate 
attempts to build up a theory of thermoelectricity on the 
perfect gas analogy for free electrons, treating of the Peltier, 
Thomson, Hall, and Nernst and Ettingshausen effects. I do 
not propose at present to apply the theory of electric gyro- 
stats to these subjects, because Liebenow (Zur T hermodynamik 
der Thermoketten, Wied. Ann. |xviii. 1899) has sketched a 
most promising thermodynamic deduction of the Peltier and 
Thomson effects from the electrical and thermal conductivities 
of metals. So a successful molecular theory of thermoelec- 
tricity presupposes a satisfactory electronic theory of the 
Second Law of Thermodynamics as applied to electricity. 
Moreau has shown ( Compt. Rend. exxx. 1900) that the Nernst 
and Httingshausen effect can be deduced from the Thomson 
and Hall effects. There is reason to believe that the Hall 
effect is connected with the electric and magnetic relations 
of elasticity, and as we have shown that rigidity in its origin 
is electrostatic and its temperature variation is electrokinetic, 
it is evident that a satisfactory electronic account of the Hall 
effect probably involves a complete theory of the effects of 
electric and magnetic fields of external origin on the elastic 
properties of metals. This large subject, first opened up by 
Kelvin half a century ago, has probably to be worked up 
before we get a satisfactory theory of the Hall effect. It is 
interesting to reflect that Kelvin, in applying the laws of 
thermodynamies to electricity, many years ago foreshadowed 
the fundamental similarity between matter and electricity 

