HALL AND CAMPBELL. — MAGNETIC EFFECTS IN SOFT IRON. 663 



studied has never been tested against lead directly, but from Matthies- 

 sen's value of the thermo-electric height of "galvanoplastic " copper 

 relative to lead, and from the known thermo-electric height of this iron 

 relative to copper wire of the present day, values of 0' have been esti- 

 mated for various temperatures, from '20° to 100°. With these values 

 of 0', and with values of the other factors taken from or reckoned from 

 direct observations on the iron here studied, the Voigt formula has been 

 tested. (See Table VII.) Like the Moreau formula, it seems to be 

 correct at one temperature, between 80° and 90°, and to be untrue at 

 other temperatures. Unlike the Moreau formula, it fails because the 

 left-hand member is too large below this particular temperature. No 

 constant added to the value of©' would make the formula hold true 

 with varying temperature. In this connection it is pointed out that, 

 the ordinary thermo-electric diagram being merely a temperature- 

 entropy diagram (with the temperature-axis horizontal and the entropy- 

 axis vertical, unfortunately contrary to the familiar custom of ordinary 

 thermodynamics), thermo-electric heights are merely entropy-differences. 

 Accordingly, Voigt's 0', the thermo-electric height of lead relative to 

 any metal, cannot be regarded as a fundamental datum for the metal ; 

 for certainly we cannot suppose the entropy of electricity to be zero in 

 lead, as we know there are metals which have a negative thermo- 

 electric height, a negative electric entropy, relative to lead. 



7. A thermo-electric diagram is given for copper and iron in which 

 the temperature-axis is made vertical and the entropy-axis horizontal. 

 This diagram shows graphically, by curvature of the iron-line, the law 

 of change of the Thomson-effect coefficient with change of temperature, 

 which law was in an earlier paper expressed algebraically. (See 

 Figure 12.) 



8. The longitudinal effect which is shown as a change of resistance of 

 the iron plate by magnetization in the direction of its thickness, the 

 coefficient of which effect is called gLe, was observed in two iron plates, 

 one cut with its width parallel to the fibres, or grain, of the iron, the 

 other with its length parallel to these fibres. In the former plate there 

 was a very slight increase of resistance in a field of 10,700 absolute units, 

 in the latter a still slighter decrease of resistance in a field of 5400. 



9. The longitudinal effect which is shown as a change of longi- 

 tudinal potential-gradient, due to magnetization in the direction of the 

 thickness of the plate, in a plate along which a heat-current is flowing, 

 was observed in one of the plates mentioned in ( 8 ), the one first de- 

 scribed there. It was not looked for in the other plate. The coefficient 

 of this effect is called t^Le. (See equation ("25).) According to Zahn, 

 who quotes Houllevigue and Moreau as authorities, the potential- 



