Serviss — Internal Temperature Gradient of Metals. 459 



fiber. A small wooden dowelling rod passed up through the 

 cover to reverse the key. ' The thermal disturbances in the 

 commutator and galvanometer could be found by disconnecting 

 the thermopile and short-circuiting the lead wires beyond the 

 commutator. The effects due to the galvanometer could be 

 measured by lifting the reversing key and closing the galvano- 

 meter terminals at their mercury cups by a copper strap, oper- 

 ated through the side of the protecting box. These deflections 

 were always small, never exceeding a few millimeters, and 

 rarely amounting to l mm , although the entire resistance of the 

 circuit was about 11 ohms. 



The galvanometer was a three-cased " Panzer b> of the 

 duBois and Rubens design.* This was an extremely satisfac- 

 tory instrument because of its high and nearly constant sensi- 

 tiveness, its freedom from external magnetic disturbances, and 

 the steadiness of the zero. The resistances of the coils were 

 5*5 and 5*3 ohms at 20° C, the heavier suspension (300 mgm.) 

 was used, hung by a very fine quartz liber. The sensitiveness 

 was tested twice a day or oftener with a potentiometer, using 

 a cadmium cell for a standard of electromotive force and sev- 

 eral standard coils. This circuit was entirely independent of 

 the rest of the apparatus and was disconnected from the gal- 

 vanometer when not in use. The telescope, made by Alvan 

 Clark of Cambridgeport, gave a large, clear image and tenths 

 of millimeters could easily be- read. A current of 4'lXlO" 11 

 amperes gave l mm deflection from zero at a scale distance of 

 4'2 meters. When the thermopile was connected directly 

 through the galvanometer, 60 ohms were in circuit and a dif- 

 ference of 0° "000001 between the inner and the outer row of 

 junctions would give a deflection of l-7 mm from zero, so that a 

 mean gradient of o- 0000001 per cm. could certainly have 

 been detected. , 



The Potentiometer. — The electromotive force of the thermo- 

 pile was usually measured by a potentiometer, kindly placed 

 at my disposal by Professor B. 0. Peirce. The slide wire, 

 which could be easily changed, rested upon a glass scale. The 

 keys were provided with springs, insuring uniform pressure on 

 the wire. The contacts and the cable were copper, and like 

 all the other connections were covered with cotton. A stand- 

 ard cadmium cell was used, and the resistance (R) in the main 

 circuit was adjusted so that the fall of potential was 1 volt 

 over 10,000 ohms in r. A D'Arsonval galvanometer (D), sen- 

 sitive to a change of 1 ohm in the box R, was used for this 

 adjustment. Unless the glass scale of the potentiometer was 

 kept very clean, troublesome leakages and short circuits oc- 

 curred. Washing the glass with alcohol and with ether was 

 found to be effective. 



* Zeitschr. fiir Instr., xx, pp. 65-78, 1900. 



