Practical TTiermometric Standard. 523 



account of the supposed greater facility of purification, but 

 according to my experiments, and those of other observers 

 who have tried gold for the purpose, there would be very 

 serious objections to its adoption. Mechanically, the pure 

 metal is far too soft, and too easily strained and broken. 

 Electrically, the conductivity is too high, necessitating the 

 use of a very fine and delicate wire. From a thermometric 

 point of view, it is also much less convenient than platinum, 

 because the scale deviates from the parabolic reduction formula 

 to a very measurable extent at moderate temperatures, and 

 the softness of the material leads to continual changes of 

 zero. Of common metals, copper is the most generally use- 

 ful at low temperatures, in spite of its high conductivity and 

 liability to oxidation. Its scale approximates closely to that 

 of platinum, and it is very easily obtained of uniform purity 

 (fundamental coefficient c = - 00I28) in different parts of the 

 world as manufactured for electrical purposes. It also 

 possesses the advantage that there are no thermoelectric 

 effects produced at the junction of the tbermometric wire 

 with the copper leads. Iron and nickel have also been pro- 

 posed on account of the extremely rapid change of resistance 

 with temperature (fundamental coefficient c = 0*00625 in the 

 case of the purest metal) . The objections to these metals are 

 that their thermom-tric scales differ from the theoretical 

 scale nearly ten times as widely as that of platinum, that they 

 exhibit effects of hysteresis, and that their resistance curves 

 present remarkable singularities within the experimental 

 range. In the great majority of cases, the cost of the 

 thermometric wire is a very small item. From every other 

 point of view, platinum is the most suitable metal for a 

 standard, and it derives additional advantages in this respect 

 from the great care with which it has been studied by inde- 

 pendent observers. 



On the Necessity of Adopting a Practical Standard. 

 When the constant-volume hydrogen-thermometer was 

 adopted in 1887 as the theoretical standard at the International 

 Bureau at Sevres, it was necessary at the same time to select 

 certain mercury thermometers as the practical standards of com- 

 parison on account of the great difficulties of manipulation and 

 calculation attending the direct use of the gas-thermometer. 

 It has recently been shown by the experiments of Chappuis 

 that there are grave objections to the use of hydrogen as the 

 standard material at temperatures above 200° C. The mercury- 

 thermometer is also quite inadmissible as a practical standard 

 beyond this limit. Nevertheless the need of a practical 



