HALL. — THERMAL AND ELECTRICAL EFFECTS IN SOFT IRON. 27 



lathe, and was of such shape that, if the layer of varnish could have heen 

 removed from the rod and laid out flat, the groove would have looked like 

 Figure 5, the scale of which is 4 to 5. The two dots shown in this 

 figure mark the positions of the junctions K and A'. It will he seen 

 that each of these points lies in a part of the groove which is not of the 

 spiral, a part which is in a plane at right angles with the axis of the bar. 



These two plane half circles in which the junctions rest are very nearly 

 1 cm. apart, perhaps as much as l.OOo cm. In three trial cases the dis- 

 tances between such grooves, after baking, were 1.0021 cm., 1.0018 cm., 

 1.0032 cm., according to the measurements made by Mr. Churchill by 

 means of a dividing engine. The small lumps of solder at the junctions 

 are probably not quite symmetrical, though they were filed down as 

 much as was considered safe, and therefore the junctions probably do not 

 lie exactly in or over the middle of the grooves. But the conditions of 

 the experiment are so varied, by reversing the current and by revers- 

 ing the temperature gradient, that the result must be very nearly the 

 same as if each junction were placed accurately in the groove made for it. 



The couple L L' has its left junction 9.5 cm. distant from the left 

 shoulder of a, and its right junction 9.5 cm. from the right shoulder. 

 The couple MM' has M distant 2.5 cm. and M' distant 1.5 cm. from 

 the right-hand shoulder of a. The couples NN', O 0', and P P' have 

 corresponding positions on /3. 



Figure 4 shows a vertical cross-section through the bars near their 

 middle. The small copper wires, w and w, leading up from a and /3, pass 

 through glass tubes, t and t, set in the board B, and above are soldered 

 to somewhat larger copper wires, which are gripped near the end by the 

 binding posts/* and p, but extend beyond these posts and out of the main 

 apparatus to make connections with the sensitive galvanometer used to 

 measure the thermo-electric currents. Before the copper wires were 

 soldered together they were hooked to each other, and during the soldering 

 the hooks were kept taut, so that after the soldering the two copper sur- 

 faces were almost, if not quite, in direct contact. An actual interruption 

 of the copper connection by as much as half a millimeter of solder at 

 this point would probably make serious trouble. The board B was 

 baked after receiving a coating of asphaltum varnish. 



Z' is a thermo-electric junction of copper and german-silver laid on the 

 outer side of the bar G 3 , with a layer of thin sheet rubber between. 

 Wires leading from this junction pass around the whole guard-ring in a 

 vertical plane, and the german-silver wire joins another copper wire in a 

 second junction, which can be placed in ice-water or in steam. The 



