COMPAUISON (iF ![,. \TINTM AND c\^ TIIKKNH >\!KTI.I 111 



The coils used for this exjieriment were those lettered C and E, 10 and 3 ohms 

 respectively, which may he taken as fairly representative. They enter into nearly all 

 tin- oompariaoofl with K.8. 



It may be pointed out here that the influence of the variations of box-temperature 

 on the results is largely eliminated in the experiments, as the fundamental points of 

 the platinum thermometers were determined before, during, and after each series of 

 r\l>eriment8, and if a wrong value were taken for the coefficient to reduce all these to 

 standard temperature, the errors committed would practically compensate one 

 another. 



We may mention here that the coefficients deduced by the method described above 

 show a satisfactory agreement with those found from the ascending series of observa- 

 tions made by the first method, although the coils used in the two cases were not 

 exactly the same. 



XXI. FIXED POINTS OF THE PLATINUM THERMOMETERS. 



Before the commencement of each series of platinum thermometer comparisons s. 

 set of determinations of the zero and steam-points, generally about six in number, 

 was always made ; frequently check determinations of these points, especially of the 

 zero, were interspersed between the comparisons themselves, thus giving an indication 

 of the exact time when changes, if any, really took place. The zeros were taken in 

 an apparatus similar in all respects to that described later in treating of the gas 

 thermometer. 



A few of the first steam-points were taken in an early form of the boiling-point 

 ajijwimtus usually employed at the Breteuil Laboratory, originally designed for 

 mercury thermometers. During a long series of preliminary control comparisons 

 between the platinum and gas thermometer at 100 we found, however, a very 

 small but systematic discrepancy in the results, which disappeared when the steam- 

 points of both thermometers were taken in the same apparatus. We therefore 

 arranged that the same steam-point apparatus should be used by both of us in all 

 the subsequent experiments. 



The apparatus for the determination of the boiling-point of sulphur, and the special 

 experiments made with it, are described later on p. 97. 



XXII. HEATING OF THE THERMOMETER WIRE BY THE CURRENT. 



It is manifest that however small the current employed in the thermometers may 

 be, it must needs heat them to some extent. Although the amount of this heating 

 would I*- ditlicult to calculate, yet we thought it advisable to make a few experiments 

 with a view to determine it, and at the same time to get some data from which we 



