HEAT APPARATUS. 137 



cient to explain the" phenomena discovered by Gumming, and that 

 a current of electricity in an unequally heated conductor must, in 

 general, produce absorption or disengagement of heat, according 

 as it passes from hot to cold parts, or from cold to hot. He veri- 

 fied experimentally that this " Electric Convection of Heat " is 

 positive in Copper and negative in Iron. Le Roux has since 

 made valuable measurements of this quantity. 



All the phenomena of Thermo-electric currents can be shown 

 graphically by means of Thomson's Thermo-electric Diagram. 

 When such a diagram is constructed from experiment, it appears 

 that the lines representing (in terms of absolute temperature) the 

 thermo-electric position of a metal are, in general, approximately 

 straight, at least for temperatures within the range of mercurial 

 thermometers. There are, however, at least two marked excep- 

 tions nickel and iron, both highly magnetic metals. These give 

 zig-zag or broken lines : nickel changing its sign about 2ooC, and 

 again about 32oC ; iron at a low red heat, and again at a white 

 heat. With iron, and certain varieties of platinum in which there 

 is no electric convection, a thermo-electric circuit can be formed, 

 giving a current due entirely to electric convection in the iron 

 alone. 



III. MEASUREMENT OF HEAT AND TEMPERATURE. 



The absolute measurement of temperature' depends upon Car- 

 not's Cycle, and is in fact involved in the Second Law of Thermo- 

 dynamics. To make absolute temperature measurement agree as 

 nearly as possible with the indications of the air-thermometer, W. 

 Thomson defined the ratio of the absolute temperatures of the 

 source and refrigerator of a perfect heat-engine as the ratio of the 

 heat taken in to that rejected. The actual determination of the 

 Absolute Zero (i.e. the temperature of a body totally deprived of 

 heat), and of the absolute temperatures corresponding to the indi- 

 cations of any particular thermometer, requires, therefore, experi- 

 ments made under conditions approaching as nearly as possible to 



