524 Prof. H. L. Callendar on a 



standard is more severely felt at higher temperatures, o wing- 

 to the increased difficulties and uncertainties attending the 

 use of any kind of gas-thermometer. The backward state of 

 the science of heat at the present time, our comparative 

 ignorance of the fundamental laws of radiation and dissocia- 

 tion, and of many other phenomena as dependent upon 

 temperature, are principally due to discrepancies of gas-ther- 

 mometry at high temperatures. It would be a great step in 

 advance to establish a practical standard of comparison, even 

 if it did not coincide perfectly with the theoretical scale. 

 The apathy which has characterized the attitude of the 

 scientific world with regard to thermal units and thermo- 

 metric standards can only be attributed to ignorance of the 

 facts. It is impossible for those who have never worked with a 

 gas-thermometer to realize the extent of its shortcomings. 

 Such is the general ignorance of these defects that it has 

 often appeared sufficient merely to mention the words " air- 

 thermometer " in connexion with some difficult measurement, 

 without giving any details of observations or apparatus, in 

 order to secure the widest and most unhesitating acceptance 

 of the results. On a previous occasion I have already given 

 (Proc. Roy. Inst., Mar. 1899) some illustrations of this point, 

 drawn from the recent researches on the law of radiation. 

 Another very instructive illustration of the difficulties and 

 uncertainties of gas- thermometry at high temperatures is 

 provided by the history of attempts to determine the 

 boiling-point of zinc, a summary of which is contained in 

 Table I., together with some results for the P.P. of silver for 

 comparison. 



The first high-temperature determinations of any import- 

 ance were those of Pouillet in 1836, with a porcelain 

 air-thermometer. He gave the F.P. of silver as 1000° C, 

 and that of gold as 1200° C, in round numbers ; results which 

 were still generally current in text-books as late as 1880. 

 The first determinations of the B.P. of zinc were those of 

 Deville & Troost in I860, with a porcelain weight-thermometer 

 containing the vapour of iodine, which was chosen on account 

 of its density. It will be observed that their result, 1040° C, 

 was above the F.P. of silver as found by Pouillet. Becquerel 

 in 1863 determined the F.P. of silver by observing the 

 temperature in a muffle with an air-thermometer of platinum 

 at the moment of fusion of a silver wire. His individual 

 measurements ranged from 917° to 960°, but he selected the 

 highest value as being the most probable. He determined 

 the B.P. of zinc (932° C.) and various other points indirectly 

 by means of a Pt-Pd thermocouple, employing a reflecting- 



