THEORY OF ELECTROMAGNETISM. 
751 
— 877 [{<£') UV] a + 6 = (ff — lfG. 2 \Jv ' a ; 
or 
- [{*'} W] n + 4 = - 277 [I7WI .(37). 
Hence for both paramagnetic and diamagnetic isotropic bodies surrounded by non¬ 
magnetic media, Maxwell’s stress leads to a surface traction which is always a 
r tension (except as in the anchor-ring -when it is zero). 
78. Consider now the well-known ordinary case of a soft-iron ellipsoid (ff a constant 
scalar) brought into a uniform field. Inside the ellipsoid B , H , and I' are all constant, 
and therefore [ff] A' = 0, so that there is no bodily force. Also since [equation (27), 
§67] 
= VCB' - V , 1 SI , H' 1 + W'vnr/a ..... (38), 
there is no bodily force in the surrounding medium. Hence, in the present case, the 
only force is the tension at the surface. The ellipsoid will therefore be expanded by 
Maxwell’s stress. 
D. Thermoelectric, Thermomagnetic, and Hall Effects. 
79. It will be found convenient to discuss these various effects together. 
The natures of the thermoelectric and Hall effects are well known and need 
no description here. The thermomagnetic effects are perhaps not so well-known. 
The original papers of von Ettingshausen and Nernst (the discoverers of these 
effects) are in ‘ Wied. Ann.,’ xxxi. (1887), 737 and 760, xxxiii. (1888), 126, 129,. 
474. The effects are briefly described in Professor J. J. Thomson’s 'Applications 
of Dynamics to Physics and Chemistry,’ 1st ed., § 57. The principal features of 
these effects are that the electromotive forces due to differences of temperature 
are modified in two ways by the presence of magnetic force. First, parallel to © 
there is an electromotive force that varies approximately as H 2 T© (the “ longitudinal ” 
thermomagnetic effect); and, secondly, at right angles to both © and H there is an 
electromotive force BV0K, where B is a scalar dependent on the temperature, but 
approximately independent of T© and TH (the “ transversal ” thermomagnetic effect). 
The latter effect is especially large in bismuth. There is evidence that these effects 
are closely connected with the Hall effect. 
80. The natural way to discuss these results would appear to be to attempt to 
explain them by suitable terms in l. But on the present theory it is possible that 
they may be explained by terms in x. According to the first explanation they would 
be reversible phenomena, and according to the second irreversible phenomena involving- 
dissipation of energy. Thermoelectric effects are certainly at present looked upon by 
physicists as reversible phenomena. 
The two explanations—which will for the future be referred to as the theory of 
reversibility and the theory of irreversibility respectively—will be found in many 
respects very analogous, though, of course, we must expect some striking difference of 
