1878.] of Iron and Steel caused by Magnetism, 111 



metres from the iron bar; the bar was also propped by a wooden 

 support at the other end. A preliminary experiment had been made 

 for the purpose of ascertaining whether the heat radiated from the 

 electro-magnet would affect the thermo elements to any appreciable 

 extent, and this was not found to be the case. 



The water in the cube was now kept boiling for about nine hours, 

 and after that time had elapsed it was ascertained that the flow of heat 

 from one part of the bars to another compensated the loss from radia- 

 tion, and the light on the scale remained stationary. (It should be 

 observed that, as it was impossible to cause one thermo-element 

 exactly to neutralise the other, a slight use of the adjusting magnet 

 of the galvanometer was made to bring the light near the zero of the 

 scale.) The current from a single bichromate cell was now sent through 

 the coil of the electro-magnet. In a few moments the spot of light 

 began to move very slowly across the scale in a direction indicating 

 that the flow of heat through the iron bar was being checked, and in 

 about fifteen minutes a deflection of about 25 divisions of the scale was 

 obtained ; the current of the cell was then stopped, when the light 

 began slowly to return, and finally settled in apparently the exact 

 position which it held before sending the current through the coil. 

 This experiment was repeated several times on the same day, and on 

 several other days, but always with the same result as regards direction, 

 namely, showing a diminution of the flow of heat from longitudinal 

 magnetisation. 



As it was thought, however, that the electro-magnet might when in 

 action perhaps produce an apparent diminution of flow of heat by 

 slightly disturbing the bar in its position in the aperture, though every 

 precaution had been used to prevent the chance of this, the bar in the 

 next set of experiments was securely soldered into the cube, and again 

 tested as before. Here, again, some five or six trials gave the same 

 result, and seemed to show, most conclusively, that the thermal con- 

 ductivity of soft iron is diminished by longitudinal magnetisation. 



As the method of observation, however, was tedious and would give 

 no idea of the amount of alteration of conductivity, other expedients 

 were now adopted. 



A piece of soft iron, BC, J inch long, ^ inch broad, and 2 milli- 

 metres in thickness, was soldered, as in the figure, to two pieces of 

 brass, AB, CD, of the same breadth and thickness. 



