528 Prof. H. L. Callendar on a 



and should be raised to 1055° or 1060°, which would agree 

 very closely with the value given by the platinum resistance- 

 thermometer according to the experiments of Heycock and 

 Neville (Journ. Ohem. Soc. 1895). The latter observers 

 have also shown that the F.P. of copper is at least 20° above 

 that of gold; so that Violle's value 1054 for the Cu F.P. 

 should be still further raised. 



Quite recently (Wied. Ann. Aug. 1899) Holborn and Day 

 have used a zinc-boiling apparatus for obtaining steady tem- 

 peratures in thermometric comparisons. The apparatus was 

 not intended, and is not suited, for absolute determinations 

 of this point, as it permits considerable superheating of the 

 vapour as the zinc boils away, but the limits of their observa- 

 tions, 910° to 930°, allowing for superheating, would appear 

 to indicate that the Zn B.P. is lower than 920°. 



Thus by several independent observations we are led to 

 the conclusion that all the numbers of Violle's calorimetric 

 scale require to be raised, but that his value for the Zn B.P. 

 should be lowered. I have recently observed this point with 

 a platinum resistance-thermometer, which gave the value 

 916° 0. ; and although I could not feel quite certain of 

 avoiding superheating on the one hand, or excessive cooling 

 by radiation on the other, I believe that this result will prove 

 to be within one or two degrees of the true boiling-point. 



The experiments of Holborn and Day afford a good illustra- 

 tion of another point of some importance. Their observations 

 with a platinum-iridium bulb gave results on the average 

 about 8°*5 lower than their porcelain bulb at 1150° 0. They 

 attribute this difference to the rapid increase in the expansion 

 of platinum-iridium at the higher points. Their results were all 

 calculated assuming constant mean values for the coefficients of 

 expansion, namely, 0*0000132 for porcelain, and 0'000025 

 for platinum-iridium, which they assumed provisionally to be 

 the same as platinum. If we adopt instead of this for the 

 coefficient of expansion of platinum at t° C. the formula 

 (dv/d0A' = 0-00002564 + 0-00000000962*, given by my own 

 experiments in conjunction with Mr. Eumorfopoulos, we 

 should find the expansion correction 29 o, 0at 1000° C, instead 

 of 22 0- 5, i. e. the results would be raised about 6°*5. It must 

 be noted, however, that the individual observations of Holborn 

 and Day at this point vary about 6° or 7° on either side of the 

 mean with the porcelain-thermometer, and about 3° or 4° 

 with the platinum-iridium. The uncertainty of the expansion 

 correction is therefore of much the same order as the 

 fortuitous errors of observation in comparing the gas-thermo- 

 meter with the thermocouple. 



