A Nezu Thermoelectric Phenoineno)i 123 



none. By thus shifting the flame about I occasionally got very 

 large effects, once reaching -02 volt, at which it was steady for 

 some time. This was by far the largest effect I had yet observed 

 in any metal. I afterwards obtained two wires drawn from 

 standard gold, 92 per cent, gold, 8 per cent, copper, these 

 were about 1 mm. and -5 mm. diameter. Neither gave any 

 effect when heated up to melting point in a naked flame, and I 

 could get no effect at all by heating the coarser one in a clay 

 tube. On heating the finer in a clay tube moderate eff'ects, ^.^i^'-., 

 •0001 volt, were observed. On attempting to draw the hot wire 

 through the tube it parted. On pushing the broken end into the 

 tube again there was an enormously high effect, but before the 

 resistances could be altered it had fallen somewhat, and when the 

 needle had become steady enough to indicate the amount, it was 

 •3 volt, though I think it must have been quite -5, but -3 was 

 the highest I read, and that I can vouch for. On allowing the 

 tube to cool it was found that the wire was stuck. The tube was 

 cracked open, and it was seen that the gold had fused into a lump 

 the thickness of the tube, and about 1 cm. in length. Another 

 wire was heated, and after the flame had been shifted, about -025 

 volt was reached, another shift giving -13 volt. The effect was 

 always much greater after the wire was fused and the ends pushed 

 together. For the most part with gold the effect was temporary, 

 none of those over -1 volt lasting more than fifteen seconds, 

 though on one occasion a steady -18 volt was obtained. The 

 effect was nearly always increased by disturbing the system in 

 any way, such as shifting the flame or pulling the wire along the 

 tube. When the junction of the thick and thin wires was heated 

 the effect was much the same as with the thin wire by itself, but 

 as a rule it was greater and more permanent, and with one 

 exception, which was only for a few seconds, it was always in the 

 same direction — the currents flowing from thin to thick. I 

 alternately heated and cooled such a junction, and at last obtained 

 a temporary effect of 'SS volt, which dropped quickly to '3 and 

 remained steady. After some time I shifted the flame in hopes 

 of raising it, but it fell. On another occasion it increased steadily 

 to "29 volt and remained steady till disturbed. 



In the early part of my work I used the most sensitive galvano- 

 meter obtainable —a low resistance, astatic instrument, with 



