266 



Professor Deioar 



[Jan. 10, 



made in the numbers representing the deviations of the Thomson 

 galvanometer : — 



Table I. 



Tablk II. 



If tlic (liflcrcuces in the galvanometer readings for every ten 

 degrees in the first table be tabulated, it will be observed the second 

 difference may bo regarded as constant, considering the errors of this 

 kind of observation. A parabolic formula can therefore represent the 

 results with sufficient accuracy. These second differences are far more 

 constant than similar numbers deduced from Rosetti's observations, 

 and his more complete formula in terms of the absolute temperature 

 is too extensive, considering the range of the experiments where tem- 

 perature was accurately known. Tho results of Table II. extend to 

 the boiling points of mercury and sulphur, and the numbers are in near 

 accord with the simple square of the temperature. The alteration in 

 the condition of the radiating surface at high tem2)eratures causes 

 great complications, and until this difficulty is overcome, experiments 

 at high temperatures must remain uncertain. All the experiments 

 show that for an approach to a knowledge of temperatures beyond the 

 range of our actual thcrmometric scale, the law given in 1873 is a 

 sufficiently correct reproduction of the facts, considering the limited 

 data at our disposal. 



The intensity of the radiation of the positive pole of the Siemens' 

 arc, as compared with the same surface heated with a large oxy- 

 hydrogen blowpipe, was determined by emi)loying a hollow negative 

 carbon which allowed the intensely heated surface to radiate directly 

 on to the pile, as shown in Fig. 1. A largo number of observations 



