HALL AND CAMPBELL. — MAGNETIC EFFECTS IN SOFT IRON. G59 



but we soon found that the thermo-electric behavior of the iron wires, 

 as well as that of the plate, was atfected in such a way by the magneti- 

 zation that it was impossible to get satisfactory results with this 

 arrangement. 



Accordingly we attached two other iron wires, one to each of the 

 copper blocks B and B of Figure 7, near the ends of the iron plate but 

 well outside the powerful intrapolar region of the magnetic field, within 

 which the iron wires previously used lay. Working with this arrange- 

 ment we got. very considerable effects which showed that transverse 

 magnetization produced in the plate a change of quality or condition 

 such as to make the copper block at the hotter end of the plate elec- 

 trically positive compared with the copper block at the colder end. 

 This difference of potential, Ai-", we could and did measure, but there 

 was, and still is, some uncertainty as to the value which should be 

 assigned to d in this case. It should not be the whole length of the 

 plate, for the whole length w^as not subjected to the full strength, //, 

 of the magnetic field. We have taken d as 80 per cent of the whole 

 length of the plate between the blocks, and have reckoned y,Ze accord- 

 ingly. Each value of this coefficient, as given below, was found from 

 two sets of observations, one made with the heat-current flowing south, 

 the other with it flowing north. As usual, the numbers given in the 

 last three columns are approximate only. 



The relation of /,Zg to H in these two groups of data is shown by 

 means of the two curves of Figure 14. The x points are for the earlier 

 observations, the O points for the later ones. The two curves are very 

 like, and neither suggests a reversal of sign of the effect in question at 

 any stage of magnetization. Why one curve runs notably lower than 

 the other through its whole course we cannot say with full confidence. 

 It is, however, a fact that in the period of the June 14-17 observations 

 the arrangement for getting the sensitiveness of the galvanometer was 

 under suspicion. In the warm, moist atmosphere of a Cambridge 

 summer the problem of insulation in delicate electrical measurements 

 becomes considerable. Leakages occur to a disturbing extent where 



