658 PROCEEDINGS OF THE AMERICAN ACADEMY. 



The day's work gave a net deflection of 0.03 cm. as the result of mag- 

 netic action, — a quantity smaller than its own probable error. This 

 deflection, if considered reliable, would indicate a temperature-diff"er- 

 ence of 0.0008° C, — about a tenth of the temperature-difference that 

 would be needed to account for the apparent change of resistance which 

 has been described above. 



The coeflicient of the longitudinal effect here looked for but not 

 found may be defined as follows : 



Longitudinal Tkermomagnetic Potential- Difference. — With the 

 magnetic field and the HH current as in Figure 3, we may take note 

 of the difference of potential, AP', set up by magnetic action between 

 two points d cm. apart along the plate. We shall call this effect 

 positive when the potential-gradient thus established is in the same 

 direction as the temperature-gradient along the plate. As the coeffi- 

 cient of this effect we have 



^p\de 

 '^^-'^~^di^^ ^^^^ 



dB 



"where jj means the temperature-gradient along the plate. 



We have not looked for this effect in Plate 2. In Plate 1 we studied 

 it at considerable length, because of a disagreement between our obser- 

 vations and those reported by Zahn on the authority of Houllevigue 

 and of Moreau. He says, " The phenomenon is complicated here [in 

 ferromagnetic metals], for the sign is dependent on the field-strength. 

 With small H the potential-gradient established is in the same direc- 

 tion as the temperature-gradient ; it reaches a maximum at ordinary 

 temperature [bei mittlerer Temperatur] for a field-strength of about 

 3000 gausses, then vanishes, in nickel for // about 4500, in iron and 

 steel at about double this field-strength, and takes, for higher fields, 

 the opposite direction. The reversal of sign occurs at a lower field- 

 strength the higher the mean temperature of the plate is." 



We used at first the iron wires /i and 7s, connected with points (1) 

 and (3) on the plate. The outer ends of these wires were permanently 

 joined to copper wires, in junctions like the one shown in Figure 10, 

 and from these junctions the copper wires led to the galvanometer. 

 After such a balance of thermo-electric forces in this circuit as to keep 

 the scale of the galvanometer in the field of vision liad been effected, 

 the magnetic force was brought into action. An effect was observed ; 



