128 Proceedings of the Royal Society of Victoria. 



found the behaviour much the same in each case, "IS volt being 

 observed frequently, and "2 occasionally. It had nearly always 

 been necessary to shift the flame and thus set up irregular heating 

 before the effect could be observed in any considerable degree. 

 To see if the effect could not be obtained from a perfectly 

 symmetrically heated system, I took a fresh tube, and passing a 

 wire along it applied a flame to the middle, prepared to watch it, 

 if necessary, half-an-hour. There was no effect for a minute, the 

 e.m.f. then became manifest and rose quickly to -035 volt, and 

 fell again to "03. Then after rising and falling irregularly for ;i 

 time it soon reached -13 volt. Forty-five minutes after lighting 

 the readings changed sign but did not get higher than "05 on the 

 other side. After an hour I began to heat it irregularly, but did 

 not get more than -15 volt. With lead a very slight cooling of 

 the tube caused the effect to disappear. Merely cutting off the 

 supply of air from the bunsen flame was always followed by a 

 very great decrease, if not a complete disappearance, of the effect, 

 although the luminous flame kept the tube at a moderate red 

 heat. An estimation of the temperature reached inside the tube, 

 made by means of a copper platinum junction with specimens in 

 which I had not been able to observe this other effect, showed 

 that with the bunsen flame 900° C. was reached, and with a 

 blowpipe 1050°. I only used the blowpipe occasionally, generally 

 using an ordinary bunsen flame. 



Filling a tube with tin as the first had been filled with lead 

 no efiect higher than -0001 volt was observed for half-an-hour 

 either by steady or irregular heating. At last there was a large 

 and sudden swing of the needle and on its coming to rest it 

 indicated -15 volt, remaining between -12 and -15 for several 

 minutes. Various parts of the tube were heated, and as in the 

 case of the first lead tube there was one point in particular 

 which gave great effects. After leaving it for a couple of days 

 and again applying the flame to this point -3 volt was indicated 

 almost at once, and for half-an-hour from -28 to -31 volt was 

 maintained, only once did it fall to -21 but instantly rose again. 

 While at its height the gas was turned off. There was a steady 

 and very slow fall, -01 volt still remaining after ten minutes. 

 The gas was lighted again and -28 was soon indicated again. 



