HALL AND CAMPBELL. — MAGNETIC EFFECTS IN SOFT IRON. C33 



The tubes W and W show where the two streams of water enter 

 the apparatus. The course of these streams is first downward past 

 the thermometer-bulbs, which give the temperature readings called 

 Tn and 7'^, through the massive copper blocks JJ and B. Then, if 

 electromagnetic experiments are in order, the course followed is that 

 shown by the arrows in Figure 7 ; but if a temperature-gradient for 

 therniomagnetic work is to be established and maintained, the two 

 streams, one warm and the other cold, pass off at once from B and B 

 to the sink. 



The longitudinal electric current, when one is used, enters by one 

 of the heavy wires C, C, and passes out by the other. 



Plate 1 was studied at intervals from June 5, 1909, to June 29, 

 1910. All the observations we have to publish on Plate 2 were made 

 during December, 1910. 



Calibration of the Thermo-Electric Couples. 



The "thermo-electric height" between copper and our iron was 

 detej'mined some years ago for various temperatures ranging from 27° 

 to 71°, and from this set of data a curve was plotted which gives the 

 thermo-electric force of a copper-iron couple within these limits. As 

 our iron wires were drawn from a piece of the same iron bar that 

 furnished the iron used in determining these thermo-electric heights, 

 and as they were annealed after drawing, it was assumed that the 

 curve just described would serve for couples made by joining these 

 wires with copper ; that is, we assumed the iron wires to be thermo- 

 electrically so like the plate to which we attached them, that we should 

 be justified in ignoring any thermo-electric forces which might arise 

 from differences between the plate and this wire. 



But to our surprise and disappointment we found that the thermo- 

 electric force of the couple Ii-j^late-Is was about one tenth-part as 

 that of the couple Ci-plate-Ca. This discovery, which was not made 

 at the beginning of our work, taken with the fact that the thermo- 

 electric quality of iron is considerably affected by magnetization, led 

 to the rejection of iron wires and the use of constantan wires in their 

 place, when we came to the preparation of Plate 2. 



Experiments to determine whether constantan wire would be ap- 

 preciably effected in its thermo-electric quality by magnetism were 

 conducted as follows : A piece of annealed constantan wire was 

 fastened by means of shellac, applied melted, within a narrow glass 

 tube across, or through, which ran side-tubes carrying streams of 

 water, one at a temperature near 16° degrees, the other near 76°. 



