ADAMS AND JOHNSTON! STANDARD TEMPERATURE SCALE 277 



The identity of the readings of the thermoelement and the resistance 

 thermometer at boiling points and melting points. The initial series 

 of measurements gave differences between the freezing point of 

 tin and the naphthalene point on the one hand and between the 

 freezing point of cadmium and the benzophenone point on the 

 other, — using our own apparatus and materials — which were 

 0?2 higher than the corresponding differences obtained by Waid- 

 ner and Burgess at the Bureau of Standards. This lack of agree- 

 ment disappeared 5 when we determined all the points on the 

 identical samples of material used by Waidner and Burgess, which 

 they were kind enough to lend us. 



In Table IT we present our results. In the boiling point experiments, 

 three forms of apparatus were used ; one (B in the table) of glass 6 - — with 

 an aluminium shield for the thermo-couple; the second (A) of brass with 

 an inner tube of thin copper to prevent radiation; and the third (S) 

 similar to the second, but of slightly different dimensions. Four differ- 

 ent elements were used, one (E) of platinum — platin-rhodium, the other 

 three (d, C 2 , C 3 ) of copper-constantan (No. 30 wire = 0.25 mm. diam.) 



The electromotive forces at the freezing points were reduced to degrees 

 with the aid of the table; the boiling points were reduced to normal 

 pressure by means of this table and the formulae: for naphthalene, 

 T m = T - 0.058 (p - 760*) ; for benzophenone, T m = T - 0.063 

 (p — 760). The freezing points have been given to hundredths of a 

 degree, as relative values for purposes of comparison only; their absolute 

 accuracy is of the order of 0?1. 



This table shows that the boiling points are independent of the appa- 

 ratus and sample of material employed, and that the freezing points 

 can be reproduced satisfactorily on different days and with varying 

 set-up; e.g., the results are independent of the size and kind of tube- 

 glass or porcelain — used to protect the thermo-couple from the metal, 

 and of the size of the charge. 



In order to make the data of Waidner and Burgess more truly com- 

 parable with our own, we have reduced them to the scale of temperature 

 on which our own values are based (cf. postea, Table IV), and present 

 these reduced values in the last column of Table II. The differences 

 between the values for the adjacent freezing and boiling-points are com- 

 pared in Table III with the analogous differences derived from our ther- 

 moelectric measurements. 



The differences, therefore, as determined by us with platin- 

 rhodium and copper-constantan thermoelements in various forms 



6 The divergence was due to slight impurities in our tin and to the fact that the 

 benzophenone used was from Merck (cf. footnote, to Table I). 

 6 Lent to us by the Bureau of Standards. 



