COMPARISON OF PLATINUM AND GAS THF.KMMMKTKRS. 43 



conditions of't-xpci imcnt, such as changes in the laboratory-temperature, the different 

 treatment of the apparatus by different observers, &c. 



Tlie.se trials, which were continued over a considerable period, showed that one of 

 tlic disadvantages of this form of apparatus is the almost inevitable difference of 

 lag" between the mercury thermometer employed to indicate the coil-temperature, 

 and the platinum-silver coils themselves, which in this case hang loosely in air. 

 l-V.-in this cause, especially when the box-temperature is changing rapidly, some 

 uncertainty as to the coil-temperatures is introduced. During the winter, when the 

 imiperature of the laboratory often fell very considerably during the night, and also 

 in summer when it rose to over 20, the temperature of the coil-space changed 

 r.-ijiidly during the daytime, although the regulator nevertheless maintained the water 

 in the tank very near 20 throughout, showing that the protection afforded by the 

 glass cover of the resistance-box was insufficient under the prevailing conditions. The 

 measurements made showed conclusively that in this case the coils followed 

 temperature changes faster than the mercury thermometer selected to indicate the 

 coil-temperature. 



The temperature-coefficient of the alloy of which the coils are constructed is '00026 

 and that of the platinum wire of the thermometers is '00386 ; if then we wish to 

 determine a platinum-temperature to '001 (whatever the resistance of the thermo- 

 meter chosen) we must know the coil-temperature to '015. Therefore, unless great 

 precautions be taken with the mercury thermometer, it is difficult to see how the 

 measurements of coil-temperature can be sufficiently trustworthy. 



GRIFFITHS in his later experience has got rid of the first-mentioned difficulty while 

 retaining platinum-silver as the resistance metal, by immersing the coils in a well- 

 stirred bath of highly insulating oil, into which the mercury thermometer is placed 

 directly, thus rendering the measurement of the coil-temperature much more certain.* 



V. CONSTRUCTION OF THE NEW APPARATUS. 



As it was anticipated that the experiments at Sevres might occupy some time, and 

 it was not thought advisable that the Observatory should be deprived altogether of 

 the use of platinum thermometers for a long period by this apparatus being taken to 

 France, a new resistance-box was ordered specially for this work. The construction 

 of this box was entrusted by the Committee to Messrs. CROMPTON and Co., Limited, 

 and its behaviour has on the whole been very satisfactory. 



In view of the fact that it was not easy to maintain the platinum-silver coils at a 

 sufficiently uniform temperature winter and summer by any simple means, and in 

 view of the difficulty previously mentioned as to the indication of the true coil- 

 temperature with sufficient accuracy by a mercurial thermometer, it was decided in 



* A description of GRIFFITHS' subsequent improvements on the original Kew apparatus, here described, 

 is given by (5. M. CLARK (' Electrician,' vol. 38, p. 747). 



O? 



