HI PUS. J. A. HARKER ANP P. CHAPPUIS OX A 



Several of these values are the ordinary ones adopted for standard resistances, so that 

 \\ ith a suitable arrangement comparisons with a standard could be made from time to 

 time. Coils of these values are also much easier to measure off accurately than the 

 larger multiples of ;01 ohm on the binary scale. We will revert to this question as 

 to the best arrangement of coil-values in the discussion of the method of standardi- 

 zation adopted. 



The resistance box is fitted with three interchangeable bridge-strips of different 

 resistances, on which a change of '01 ohm in the box-coils causes a displacement of 

 the slider, to restore the balance, of 10, 5, and 2 centims. respectively. The slider is 

 fitted with a vernier giving -g^th millim. directly, it being considered unnecessary, 

 perhaps indeed impossible, to determine the position of the knife edge forming the 

 contact to a greater accuracy than this, without taking extraordinary precautions 

 with the scale, the slider, and the bed in which it travels. 



With the three bridge-strips the resistance corresponding to a movement of one 

 millim. is '0001, '0002, '0005 ohm, respectively. 



The strip of medium resistance was the one employed exclusively during this 

 research. 



The slider is of the form employed on the best potentiometers, and can be displaced 

 either by hand or by a fine-adjustment screw with large milled head, projecting beyond 

 the outside of the case. This screw moves longitudinally a rectangular frame carrying 

 the slider ; this frame is also capable of lateral movement in the massive brass casting 

 which surrounds the bridge-wire, and tends to protect it from injury, and to equalise 

 its temperature from end to end. The return contact from the slider to the 

 galvanometer was originally made by means of this movable frame, but from some 

 unexplained cause, apparently not thermoelectric, this led to unsteadiness of the 

 galvanometer zero.* Coupling the various parts of the framework together electri- 

 cally by flexible copper wires did not remove the difficulty, and ultimately it was 

 found best to have a silk-covered return lead attached to the spring contact on the 

 slider, and to cut off the frame from all electrical connection with the apparatus. 



The marble slab forming the top of the resistance-box was supported from the 

 inside of the tank by an iron framework, carrying racks for the coils, the sides being 

 left quite open, and all the coils easily accessible for inspection at any time. The 

 whole was placed in a very heavy double- walled copper trough holding about 50 litres 

 of water, and was covered by a doubly-hinged lid, glazed with thick bevelled plate 



* In these experiments, in which a Griffiths' thermoelectric key (described later) is used, in the normal 

 position of the key the galvanometer circuit remains made. When the platinum thermometer is not 

 changing rapidly in temperature, the stability of the galvanometer zero is a good criterion, from whioh 

 much may be gathered as to the working state of the bridge, and the magnitude of the thermo-currents 

 present. We have reason to believe from our own experience that the use of a well-constructed key of 

 this type considerably facilitates the carrying out of low-resistance measurements, where high accuracy is 

 desired. 



