50 HEYL— THE CONSERVATION OF [April 22, 



Having no milli-voltmeter at my disposal it was necessary to 

 use a milliammeter, which of course is equivalent to the other if 

 the resistance of the circuit is known. Connections were made 

 to the ammeter on the supposition that the current would leave the 

 cell by the wrought iron electrode and return by the carbon. 

 Care was taken, however, to insert a reversing key in the circuit. 



About ninety pounds of a low carbon steel (o.io per cent.) were 

 placed in a crucible and heated for about four hours. Perfect 

 fluidity could not be obtained, but the mass finally became pasty and 

 viscous, allowing a rod of iron to be plunged into it without great 

 difficulty. A small amount of aluminium was then added to free 

 the mass from gas bubbles, and the crucible was then withdrawn 

 from the furnace, placed on the floor and banked about with coal, 

 which immediately took fire. As quickly as possible, for the metal 

 rapidly hardened, the electrodes were plunged in while a watch 

 was kept at the ammeter. 



Almost immediately the carbon electrode snapped off at the 

 surface of the metal, and so rapid was the cooling that by the time 

 the spare carbon electrode could be brought into action the metal 

 was too hard to allow it to be plunged in ; but the small fraction of 

 a second available to the watcher at the ammeter was sufficient. 



Immediately on plunging in the carbon electrode the pointer 

 of the ammeter started to move in the expected direction. When 

 the carbon cracked the reading was 0.015 ampere. The pointer 

 then fell back and exhibited a tendency to a reverse deflection. 

 Turning the reversing key, the pointer again moved forward and 

 remained stationary at about one third of its previous reading. 

 On looking at the crucible it was found that the upper portion of 

 the electrode was lying loosely upon the surface of the now solid, 

 though still red-hot metal. On lifting the electrode the deflection 

 disappeared, but reappeared as often as the carbon was touched to 

 the hot metal or even to the (now hot) iron electrode. This de- 

 flection was plainly of a thermo-electric origin, while the earlier 

 deflection in the opposite direction is to be explained only as the 

 result of galvanic action. 



How far the pointer would have moved had the carbon not 



