460 Serviss — Internal Temperature H radh-nt of Mc-tals. 



When the cylinders were first placed in the thermostat, the 

 outside cooled down rapidly and the gradient was large. It 

 was more convenient to follow this down by sending the cur- 

 rent directly through the galvanometer. But after some hours 

 the difference of temperature was small/ and it was then meas- 

 ured on the potentiometer. 



a 



TT 



<L 



Three copper slide wires were used, as follows : 



Size. 



Diam. 



Olnns/meter. 



Cm./0° -000001 



18 



l-024 mm . 



0-01979 



0-22 



12 



2-053 



0-004922 



0-86 



3 



5-827 



0-0006106 



7"1 



Summary. — A brief enumeration of the changes made in 

 Dr. Thwing's apparatus may now facilitate comparison. These 

 changes were made, not so much because they were necessary 

 as because they seemed to add to the certainty of the results. 



(1) The junctions were all placed inside the cylinders, in 

 order to expose them in the same manner. 



(2) In the final observations, the circuit was all of copper to 

 avoid disturbing thermo-electric forces. 



(3) An armored astatic galvanometer of unusual qualities 

 was available, instead of a D' Arson val, which usually has 

 internal electromotive forces. 



(4) A potentiometer was used to measure the e.m.f. of the 

 thermal couples instead of sending the current produced by 

 them through the galvanometer. 



(5) Constant conditions were maintained by slowly stirring 

 the ice to drain it dry and using conical bottoms on the ther- 

 mostat and ice vat to prevent the ice from melting away from 

 the thermostat. To this same end the external parts were pro- 

 tected from large or sudden changes of temperature. 



