526 Prof. H. L. Callendar on a 



appreciable error in his results owing to the control afforded by 

 the thermocouple and by the number of separate determinations. 

 He also repeated his observations with a porcelain bulb, and 

 obtained the still lower results, Zn B.P. 891° C., Ag F.P. 

 916 C, thus increasing the discrepancy to 150°. In spite 

 of this, the results of Deville and Troost were generally 

 accepted, and were in fact adopted by Benoit, the present 

 director of the International Bureau at Sevres, as the basis 

 of his work on the variation of the electric resistance of 

 metals in 1873. The higher value of the B.P. of zinc was 

 also confirmed about the same time by an experiment of 

 Weinhold's with a porcelain air-thermometer which gave the 

 result 1035° C. 



Violle (1877-1879) made a skilful application of the 

 calorimetric method to the determination of high temperatures. 

 By observations with an air-thermometer in the neighbour- 

 hood of 1000° C, be determined the mean specific heat of 

 platinum between 0° and 1000° C. to be 0*0377. Combining 

 this value with the mean value 0*0323 between 0° and 100° C, 

 and assuming the specific heat to be a linear function of the 

 temperature, he obtained the formula, 



*' = 0-0317 + 0-000006 t, 



for the mean specific heat between 0°and t° C. Assuming that 

 this formula held up to the fusing-point of platinum, he 

 measured the fusing-points of several metals by the calori- 

 metric method, with the results given in line (2) of Table II. 

 These results are very often quoted, and constitute what may 

 be called " the Violle scale/' They have been of great value 

 in high-temperature research as a practical standard of 

 reference. 



Admitting that the boiling-point of zinc must be below the 

 F.P. of silver, and that the latter could not be much higher 

 than the value 954° C. found by Violle, Deville and Troost in 

 1880 were compelled to revise their determination of the B.P. 

 of zinc. Using a greatly improved form of gas-thermometer, 

 they found the values 940° w T ith air, and 920° with hydrogen, 

 which agreed fairly well with the original determination of 

 Becquerel. They also showed that the higher values obtained 

 with iodine and C0 2 were to be explained by partial dissoci- 

 ation. This raises another interesting question, as to whether 

 diatomic gases like nitrogen may not be similarly subject to 

 partial dissociation at the higher points of the scale, and 

 whether it would not be better to employ argon, or helium. 



In 1882 Violle, who appears to have been unacquainted 

 with these results, made a direct experiment with a porcelain 



