450 Mr. J. D. Hamilton Dickson on 



platinum-temperatures, instead of converting them at once 

 into normal air-temperatures, an equally simple S-formula 

 like (4) exists, (19). There only remains one point ; does the 

 proposed formula represent the results of experiment as well 

 as the others? To answer this question, I have carefully 

 calculated formulae, by the method of least squares, from 

 the best experimental results obtainable. I give the details of 

 such calculations for Prof. Calendar's observations, for those 

 of Professors Dewar and Fleming, of Mr. E. H. Griffiths, 

 and of Messrs. Holborn and Wien at low temperatures. I 

 have also calculated formulae for Messrs. Holborn and Wien's 

 high-temperature results*, but it is unfortunately impossible 

 to base much upon them after the break-down of the appa- 

 ratus. Prof. Witkowskif gives a rough set of results below 

 0° C, but without details. I have calculated a formula with 

 equal care for them, but it leads so far astray that I presume 

 he only meant his results to be taken as showing the trend of 

 a possible series of experiments. 



Professor Callendar' 's Observations. — Professor Callendar is 

 at first less concerned with finding an absolute connexion 

 between platinum degrees and the readings of an air-thermo- 

 meter, than with determining the prior question whether 

 there is a strict connexion between the two. The painstaking 

 and repeated efforts made to overcome, as he has done success- 

 fully, the difficulties that arose during the investigation fully 

 entitle him to claim the demonstration of his proposition, 

 and at the same time make his observations of great value. 

 For my present purpose I deal only with those observa- 

 tions contained in tables A.-v. on pp. 213-216 of (Call.), 

 which he considers the most free from error. I have selected 

 the observations chosen, for the following reasons : — Prof. 

 Callendar found that his air-thermometer readings at constant 

 volume gave higher values of t — ot than those at constant 

 pressure, the difference in one case being as much as 1° C. at 

 540° C. (p. 221) : the gauge used was specially designed for 

 readings at constant pressure (p. 209) : readings at constant 

 pressure have less chance of error through dispensing with the 

 manometer (pp. 126, 129 of Call, and Gr.) : the temperature 

 employed is defined to be " the temperature by normal air- 

 thermometer at constant pressure, and is nearly equal to the 

 ahsolute temperature on the thermodynamic scale " (p. 218 

 Call.). Hence I have used only the observations recorded at 

 constant pressure. Of these Professor Callendar considers 



* Wied. Ann. vol. lvi. p. 360 (1895). 

 t Phil. Mag. vol. xli. p. 314 (1896). 



