60 



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



cooling. We found that if the results of the separate determinations of the value of 

 a constant outside resistance made with a series of steady box-temperatures with 

 temperature rising be plotted, along with those of a series similar in every respect but 

 with the temperature falling, the two do not overlap but form a loop. After a deter- 

 mination commenced at about 15, during which the resistance-box was heated to 31 

 and allowed to cool, the whole temperature change occupying about nine hours, the 

 coils did not return to their original resistance at 15 till they had been at this 

 temperature about three days. We satisfied ourselves that this was due to a real lag 

 in resistance and not in the indications of the box thermometers. The whole 

 hysteresis effect is, however, small, and is quite imperceptible if the temperature 

 changes are very slow, like the variations of laboratory temperature to which the box 

 was ordinarily subjected. We may say that the temperature coefficient of the sample 

 with which we observed the effect is rather abnormally small even for manganine, and 

 that we had not time to see if the same effect could be observed with other specimens. 



Although from the values thus obtained we might have deduced the temperature 

 coefficient, using only the determinations made after a rise of temperature, we con- 

 sidered it advisable to make some fresh experiments, using a modification of the same 

 method. During the first series of observations with thermometer K.8, a consider- 

 able number of zeros had been taken during a period when the box-temperature 

 differed markedly from day to day. The thermometer had meanwhile never been dis- 

 connected from the box ; the contacts remained in the same condition throughout, and 

 we have no reason to believe that any secular change occurred in the leads or ther- 

 mometer wire during the experiments. These experiments, during which the box 

 temperatures ranged from 6 P 60 to 19'65, were accordingly utilised to calculate the 

 temperature-coefficients of the coils, and from them a formula was obtained by least 

 squares for the change of resistance of the box-coils with temperature. 



Choosing as a standard temperature 15, a table was calculated giving the 

 coefficients by which the nominal box-resistances must be multiplied to give the true 

 resistances. This multiplier is alluded to subsequently as the "factor" in the 

 example of the method of calculating an experiment given later. The following 

 numbers extracted from the table, show the magnitude of the coefficient in question : 



