A Neiv TJieruioelectric Phenonietwn. 121 



It was in repeating Trouton's experiment that I found that 

 great effects were produced by heating tlie iron wire steadily. 

 This effect however was soon found to be very arbitrary and 

 irregular, and steps were taken to obtain regular and systematic 

 observations. A couple of yai-ds of very hard iron wire were put 

 in series with a sensitive galvanometer, the junctions of iron and 

 copper being immersed in the same vessel of oil to insure their being 

 at the same temperature and that no thermoelectric effects were 

 generated in them. The iron was stretched into a loop and 

 heated in various ways, such as warmed with the fingers, parts 

 immersed in boiling water, in hot oil, in melted tin both bare 

 jxnd protected with asbestos, heated in a bunsen and in a blow- 

 pipe flame, long portions heated in a tube furnace, and a small 

 part was cooled by evaporating ether. In each case some effect 

 was observed, though below 300° C. it was small, about the same 

 order of magnitude as an ordinary thermal junction of silver and 

 copper. Consistent results however were never obtained : if a 

 certain effect were observed by heating part of the wire in a 

 certain way, then on repeating the conditions a different effect 

 would be observed, perhaps greater, perhaps less, and as likely as 

 not of opposite sign. When kept heated steadily the effect was not 

 constant. It increased, decreased, kept steady, changed sign, 

 ■vanished and reappeared in the most arbiti^ary way imaginable, 

 and showed no sign of becoming steadier even after being left 

 alone for half-an-hour. Below a dull red heat these changes were 

 slow, but fast enough to keep the galvanometer needle moving 

 perceptibly, but above a red heat the changes were too fast for 

 the needle to follow dead beat, and it was kept continually 

 oscillating. These oscillations were sometimes small, perhaps ten 

 per cent, of the total deflection, while at others the needle was 

 jerked about so widely that one could not even form a mental 

 estimate of the changes of electromotive force, much less make a 

 note of them. I tried various samples of iron wire of difterent 

 hardness and thickness, from -2 mm. to a bar 1 cm. diameter. 

 The effect was observed in each case, it being as a general rule 

 more marked in the finer wires than the coarser. The highest 

 effect I observed in iron was about -002 volt. On passing the 

 wires through tubes, glass or clay, and heating them, very little 

 •effect could be obtained. 



