664 PROCEEDINGS OF THE AMERICAN ACADEMY. 



gradient established in iron by magnetic action is in the same direction 

 as the temperature-gradient in comparatively weak fields, up to perhaps 

 9000, but is in the opposite direction in stronger fields. In the iron 

 here studied the potential-gradient produced by the magnetization is 

 in the same direction as the temperature-gradient all the way from a 

 field- strength of 1900, the lowest tried, to 12000, the highest tried, 

 though the coefficient f^Le is much larger for weak fields than for strong 

 fields (see Figure 14). 



10. The other two longitudinal effects, the coefficients of which may 

 be called eLh and ^Z;,, were looked for diligently in one plate, but their 

 existence there was not proven.^i 



The Jefferson Physical Laboratort, 

 Cambridge, Mass. 



Postscript added March 11, 1911. — The objection which we have 

 made to the use of ©'„, the "thermo-electric height" of lead with 

 respect to any metal m, as a fundamental property of the metal m, 

 seems to me so plain that I have been at great pains to make sure that 

 we have not fallen into some gross blunder of interpretation of Voigt's 

 theory. There can be no question, however, that Moreau takes Voigt's 

 0'to in this way, and I have seen no protest against his doing so. There 

 is possibly room for doubt as to whether Voigt had lead in mind or 

 some other, unnamed, material of zero Thomson effect, with reference 

 to which his ©'„ is to be reckoned, but this question is of little 

 importance. 



Apparently Moreau does not criticise Voigt's theoretical discussion. 

 He seems merely to point out that it is different from his own, and to 

 test each empirically by means of data which he has taken from various 

 sources. He says, " I consider a circuit \_cha'ine^ formed by a metal M 

 and lead, the junctions being at different temperatures. At a point 

 (.r, y, z) of M, if X, Y, Z are the components of the thermo-electric 

 electromotive force per unit of length, one has, according to ]\I. Voigt, 



X=e'f, Y=&% ^=e1-. where ©'=^. (1) 

 dx by dz dl 



is a function of the absolute temperature T, and is characteristic of 

 the metal M." 



"Let us suppose that a thin slice of the metal Jf is placed in a mag- 



^^ It now appears that magnetization parallel to the thickness decreases 

 notably the longitudinal temperature-gradient in Plate 2. March 11, 1911. 



