212 PROCEEDINGS OF THE AMERICAN ACADEMY 



spot was always found to have changed by about 3 cm. The direction 

 of the change was such as to show that the junction on the magnetized 

 arm was becoming warmer. It was at first thought that this confirmed 

 Maggi's results. 



The apparatus was next arranged with one arm of the iron parallel 

 to the axis of the magnet. The arm was passed through the hollow 

 core of an electro-magnet somewhat stronger than the preceding. 

 The same battery was used. The details of the experiment were 

 exactly the same as when the bar was perpendicular to the lines of 

 force. The results of several observations here also showed that the 

 junction on the magnetized arm became hotter under magnetization. 



We had previously assumed that the heat developed in the mag- 

 netic coil would be too slight to affect the iron bar. This assumption 

 was now proved to be incorrect, by placing the unheated bar, 

 arranged exactly as in the preceding experiments, in the magnetic 

 field. When the magnet was made the galvanometer needle began 

 slowly to move, always in the direction showing a heating of the 

 junction on the magnetized arm. This deflection was slightly larger, 

 after the same space of time, than the deflections observed in the pre- 

 vious experiments, and consequently rendered the results of these 

 experiments useless. 



In order to avoid complications arising from the heat generated by 

 the electric current, the following arrangement was adopted. About 

 17 cm. from one end, the iron bar was bent upon itself. At the end 

 of the bent part, and at the point of the bar opposite this end, were 

 soldered two German silver wires. These two thermo-electric junc- 

 tions were about 0.4 cm. apart, and were separated from each other 

 by densely packed asbestos. The arm was placed in a glass tube, 

 arranged as before. The bar was heated about 19 cm. from the 

 thermo-electric junctions. By this arrangement the heating of the 

 mao-netic coils had an equal effect upon the two junctions, while 

 any change, due to altered conductivity, of the flow of heat along the 

 bar would affect the relative temperatures of the junctions. When 

 the German silver wires were connected as before with the gal- 

 vanometer, and the unheated bar was placed either parallel or perpen- 

 dicular to the axis of the magnet, there was, after forty-five minutes 

 observation, absolutely no deflection in the galvanometer. This 

 showed that the arrangement obviated all difficulties arising from 

 heating the coil ; and, moreover, that the magnetic field did not per- 

 ceptibly alter the thermo-electric relation of iron and German silver. 

 When the bar was placed perpendicular to the axis of the magnet, 



