ro.MI'AWSOX OF I'l. . \TI\I M AM) (l.\s TMKKNK >MK'I 'Kl:- 



XX. Determination of tin- Temperature-Coefficient of the Coils. 



Preliminary determinations had shown that tin- tcinjiemturc-coeificieiit of the wire 

 use<l for tin- coils was extremely small, ami had we been able to keep the box-tempera- 

 ture anywhere near constant we would bnitlly have needed to take it into account at 

 all. As, however, considerable variations of the temperature of the room were 

 inimitable, as previously explained, a method bad to be devised to determine the 



ilicii nl u ith eonsidcrable accuracy. It bad been previously found by one of us 

 that the annealing process, to which the wire must l>e subjected to minimise sulwe- 

 i|iicnt time-changes in its resistance, has an appreciable effect on the temperature 

 coefficient of the wire. In nearly all the specimens examined, the point where the 

 characteristic change in sign of the temperature-coefficient takes place was displaced 

 so as to occur at a lower temperature. 



In view of uncertainties in the method of subjecting a sample of the wire to treat- 

 ment exactly similar to what the coils themselves had received, and determining the 

 coefficient of this piece the process usually followed we attempted to measure 

 directly the actual coefficient of the coils themselves in sitfl. 



To do this we first tried a method consisting in the determination of the apjmrent 

 \alue in box-units of a resistance kept at constant temperature, while the box- 

 temperature was varied between that of the tap-water circulated through the outer 

 tank and a maximum of about 35. During these determinations every care was 

 taken that the temperature of the coils as registered really represented their mean 

 temperature at the time. Without going into details as to how this was attained by 

 keeping up a continuous circulation of hot or cold water in the outer tank, and other 

 precautions, \ve may say that the results of the measurements made were somewhat 

 unsatisfactory, and the reason of this was traced to a curious and, we believe, not 

 previously observed behaviour of the alloy in not taking up instantaneously the resist- 

 ance corresponding to a new tempi- /"/<// to u-hich i may be subjected, esix-cially when 



I 2 



