662 PROCEEDINGS OF IHE AMERICAN ACADEMY. 



tested by means of the data for iron. (See Table II.) The formula is 

 true for one temperature, near 80" C, according to our data, but de- 

 parts progressively from the truth with descent from this temperature 

 and probably with ascent above it. Below the temperature mentioned 

 the first member of the Moreau equation, as here given, is too small ; 

 above this temperature it is, probably, too large. Hence Moreau's 

 theory, which would account for the Hall effect by a deformation of 

 the plate under magnetic action (this deformation producing a real 

 rotation of the equipotential lines of the longitudinal electric current), 

 and would account for the Nernst effect as a like and equal rotation of 

 the Thomson-eft'ect equipotential lines of a longitudinal temperature- 

 flow, appears to be ill-founded and, if taken literally, unsafe, though it 

 is likely to be useful in a suggestive way. 



4. Following the suggestion of the Moreau formula, we tested the 

 equation JT^ -i- p = ;, 7^ for various temperatures, to see whether the 

 isothermal lines of the longitudinal temperature-flow are, in a field of 

 given strength, rotated in the same direction and to the same extent as 

 the equipotential lines of a longitudinal electric flow. (See Table III.) 

 The direction is the same in the two cases, but the equipotential lines 

 are, according to this test, rotated about one and a quarter times as 

 far as the isothermal lines, this ratio remaining nearly constant 

 through the whole range of temperature considered, 20° to 100"^. (See 

 Table IV.) 



5. " Rotation " cannot account for the Ettingshausen effect, since 

 there is no corresponding longitudinal condition to be rotated, none, 

 that is, which can be regarded as adequate to produce the observed 

 transverse effect. On the other hand, the ratio of the Hall transverse 

 potential-gradient to the Ettingshausen transverse temperature-gradient 

 appears to be almost strictly proportional, through a considerable range 

 of temperature, to the Thomson effect s. Thus we have, approximately, 

 eTe-^eTh = 200 s, from 20^" to 80° or higher. (See Table V.) It 

 should be said, however, that the temperature-coefficient of ^ T^ cannot 

 be regarded as accurately known. The suggestion is made that the 

 transverse potential-gradient, which is like the Thomson-effect potential- 

 gradient in being static (not attended by flow down the gradient), may 

 be the cause of the transverse temperature-gradient, whereas in the 

 Thomson effect the temperature-gradient causes the potential-gradient. 

 It seems likely that the electron theory must here be used, but the 

 attempt is not made in this paper. 



6. The Voigt formula ^T; -i- p = ;, 7^ -r- ©', in which 0', the " thermo- 

 electric height " of lead relative to the metal, takes the place of the 

 Thomson effect s of the Moreau formula, is considered. The iron here 



