458 On Platinum Temperatures. 



Gr.) values with 414 0, 53 for this temperature. The various 

 observations for aniline agree in giving 184° for its boiling- 

 point, but they vary among themselves by 0°'3 ; for naphtha- 

 lene, similarly, the boiling-point is 218° with a variation of 

 0°*2 ; for the boiling-point of methyl salicylate we have 223° 

 with a variation of 0°"3 ; and for benzophenone 306° with a 

 variation of o, 3 ; triphenyl methane is about 356° or 357°, 

 but this is at the unusual pressure of 770 millim. ; and the 

 boiling-point of mercury lies between 356^° and 358°, the 

 variations shown ranging over 1°'6. The results given by 

 the formulas (15), (16), and (17) are therefore fairly com- 

 parable with these in accuracy. 



From these calculations we are therefore justified in taking 

 the proposed formula (9) as competently satisfying all the 

 requirements of an empirical formula, within the greatest 

 limits of temperature in which experiments may be made, 

 where platinum is employed to determine temperature. The 

 variation of the constants in the different numerical formulas 

 given above corresponds with the variation of 8 in the 

 S-formula. 



If it be still thought desirable to preserve " platinum- 

 temperature," a formula similar to the 8-formula may be 

 easily devised. Equation (9) can obviously be written in the 

 form 



(^ + A) 2 =P(* + £), (18) 



giving 



(100 + A) 2 =P(100 + Z>), andA 2 =P6, 

 whence 



100 + 2A = P. 



Thus (18) becomes, writing — by an obvious extension of 

 arithmetical notation, — "O-cj for «r/100, 



t-^=h'{'0^- Qzt), .... (19) 

 where 



PS^ 10,000, 

 or, on reduction, 



^8'=^(b + 100)-\/b. . , . . (20) 



For a first approximation, since b is about 1000, we have 



46S'= 10,000. 



Although it has been demonstrated that platinum is a suit- 

 able substance for determining temperatures over a very wide 

 range, not much different probably from 2000° C, neverthe- 

 less seeing that each platinum thermometer needs at least to 

 have its constant specially and carefully determined, not by 



