310 Prof. W. F. Barrett on some 



the thermo-electric power of the various junctions up to a red 

 heat. When an iron wire was used as the second metal the 

 thermo-electric force quickly rose till a certain temperature 

 much below redness was reached, and then remained almost 

 stationary, notwithstanding that the temperature of the junction 

 rose from a black up to a bright white heat ; the cool junctions 

 being throughout kept in water at the temperature of the 

 room, or in ice. Here, then, we have the change in the 

 thermo-electric force arrested when a certain temperature is 

 reached, and the force remaining nearly the same in spite of 

 the increasing difference of temperature between the cool and 

 hot junctions. We know that so long as the difference of 

 temperature between the cool and hot junctions in most 

 thermo-couples remains constant, and the circuit is unchanged, 

 the potential-difference also remains constant ; but if the 

 difference of temperature between the cool and hot junctions 

 alters, then the potential-difference, as a rule, also changes ; 

 and for small changes of temperature the electromotive force 

 thus set up is, in most cases, proportional to the change of 

 temperature. This, of course, is the principle upon which the 

 thermo-electric pile is used as a delicate thermometer ; and 

 with certain alloys this proportionality holds good through a 

 wide range of temperature. 



In order to measure the exact temperature of the hot junc- 

 tion, a thermo-electric couple, formed of a platinum wire 

 twisted with a wire of an alloy of platinum containing 10 per 

 cent, of rhodium, was employed. The vapour of boiling- 

 water (100° C.) and of boiling sulphur (445° C.) are the most 

 convenient and reliable fixed points for plotting the lower part 

 of the scale*. For the higher parts I have used the freezing- 

 point of pure silver (961° C), and of potassium sulphate 

 (1066° C.)f. 



A reflecting galvanometer was employed, a dead beat high- 

 resistance instrument, of the D'Arsonval type, made by 

 Ducretet, of Paris. Owing to its high resistance the altera- 

 tions in the resistance of the circuit during the heating and 

 cooling of the couple introduced no sensible error, the 

 deflexions being proportional to the E.M.F. and not to the 

 current. 



The value of the scale-reading of the galvanometer in 

 * The thermo-couple must, of course, be protected from the sulphur 

 by being enclosed in a hard-glass tube. To obtain the boiling-point of 

 sulphur it is best to employ a hard-glass flask with a long neck, all except 

 the lower portion of the flask being jacketed with asbestos. A convenient 

 arrangement is supplied by the Cambridge Instrument Company ; this 

 I used, 

 t Heycock and Neville, Trans. Chem. Soc. 1895. 



