44 



DRS. J. A. BARKER AND P. CHAPPUIS ON A 



this second apparatus to obviate the necessity of very accurate measurement of the 

 coil-temperature by using one of the new alloys of very small temperature-coefficient, 

 manganine being the one chosen. The expediency of this change was subsequently 

 emphasised by the fact that we found it was inadvisable to artificially heat the room 

 at Breteuil in which the comparisons were made, on account of the uncertainties 



Fig. 2. 



O 



attending the measurement of the temperatures of the various mercury columns 

 of the gas thermometer. During the year and a-half the experiments lasted, the 

 room temperature varied from about 4 to 23, which would have rendered accurate 

 artificial control of the box -temperature extremely difficult. 



Since in the first resistance-box the thermoelectric effects between the various 

 wires and terminals in the circuit (in which several different metals are used) were 



