314 Prof. W. F. Barrett on some 



give an E.M.F. of -^ of a volt when heated over any flame, 

 the cooler junctions being kept in ice-cold water. A con- 

 venient standard of E.M.F. might thus be made if the means 

 of the readings between 300° and 1000° C. were taken. 

 Whether repeated heating and cooling of the alloy will affect 

 its E.M.F. I cannot say, but I have not yet observed any 

 injury resulting from this cause. 



We know so little of the whole subject of thermo-electricity, 

 that the explanation of the remarkable behaviour of this 

 alloy can only be a matter for conjecture. Some light may 

 be thrown on it by the results obtained from the other alloys 

 of iron, when their thermo-electric behaviour is examined. 

 So far I have found that reducing the nickel in the alloy 

 from 25 to 19 per cent., the other constituents remaining the 

 same, does not destroy the sudden arrest of E.M.F. at about 

 300° C. ; but the range of temperature where the E.M.F. is 

 nearly constant is less, extending from about 400° to 750° C. 



It has been suggested that the peculiar thermo-electric 

 property of this alloy may be connected with the effect 

 observed by Lord Kelvin, the so-called Thomson effect, 

 whereby a kind o£ electric convection of heat occurs. It is 

 very possible that it may have some connexion with this, and 

 hence with the neutral point, which occurs in the thermo-electric 

 behaviour of certain pairs of metals, such as copper and iron. 

 It is well-known that at a certain critical difference of tempe- 

 rature between the hot and cold junctions of, say, a copper-iron 

 couple, the potential-difference due to heat disappears ; as the 

 temperature rises, inversion of the current occurs ; and a 

 second inversion may occur at a still higher temperature. In 

 the phenomenon described in this paper, the E.M.F., it is 

 true, does not fall to zero. If, however, the cooler junction 

 were kept at a temperature of 310° C, this would be the 

 case ; and we should then have a series of three successive 

 small inversions of E.M.F., occurring at 540° ; 810°, and 

 1030° C. In a copper-iron couple the neutral point is 275° C. 



I have noticed that the temperature of the neutral point in 

 a copper-iron, or copper-steel couple, is not the same during 

 the heating as during the cooling of the couple. Moreover, 

 in a couple formed of copper and mild carbon steel, the 

 neutral point becomes lower in successive heatings. Thus, 

 at the first heating the temperature of inversion was approxi- 

 mately 640° C, and in cooling, 500° C; in the second 

 heating, 550° C, and in cooling, 465° C. ; in the third heat- 

 ing, 520° C, and in cooling, 485° C: the difference between 

 the temperature of inversion in heating and cooling becoming 

 less in each successive heating and cooling. The cool junc- 



