ON STANDARDS FOR USE IN ELECTRICAL MEASUREMENTS. 



63 



At these low temperatures the observations are not nearly so con- 

 cordant as those already considered. The tei^minals of the coils are stout 

 rods of copper, and whenever the temperature of the room is different 

 from that of the bath in which the coils are placed heat is conducted to 

 them through the copper rods and the temperature becomes uncertain ; 

 besides this it is difficult to prevent the deposition of moisture on the 

 paraffin with which the cases are filled, and this again becomes a source 

 of error. Table VII. gives a series of the differences observed between 

 the various coils and Flat. The coils were in a north room of which the 

 windows were open, and the temperature in the room was on the average 

 about 2° C. The differences are given in bridge wire divisions. 



Table VII. 



Now from Tables III. or IV. we can easily calculate what these 

 differences ought to be if we suppose that the temperature curves are 

 straight lines. In making a comparison of the results of this calculation 

 with the observed values given in Table VII. some allowance must be 

 made for the fact that the bridge wire referred to in IV. was at a mean 

 temperature of about 15°, while in Table VII. the temperature was about 

 2°. Now the temperature coefficient of the bridge wire — platinum 

 iridium — is about •00143 ; thus the change in resistance for 13° of tem- 

 perature will be '0185 of the resistance at 2°, and we shall have to reduce 

 each of our observed values by this fraction of itself. 



We thus get the following Table VIII. of values of the difference at 

 0° between Flat and the various coils. 



On examining these it is at once clear that the supposition that the 

 temperature curves for A and B are straight lines is false. 



