835 



Even then, however, real deviations remain in the range 200° K.- 

 240° K. showing that the Leiden-temperature-scale lies here somewhat 

 above that of the Phys. Techn. Reichsanstalt. 



A good correspondence at the lower temperatures is obtained with 

 the following formula 



= 0.008.50 1 



0.001515 



The greatest deviation amounts to 0.06 of a degree at 220° K. 

 It would seem to me, that this formula cannot be far wrong, and 

 for the following reasons. Looking at fig. 2 on page 653 in Henning's 

 paper, we see that the curve T — Tc as a function of the tempera- 

 ture, in the temperature region under consideration, allows a small 

 shift upwards, without the agreement with the observations becoming 

 much impaired, which shows, that the deviation from Cai.lendar's 

 formula begins even at a somewhat higher temperature. 



In the range in question we can further utilize the freezing point 

 of mercury. Henning ') has made a very accurate measurement of this 



i) F. Henning. Ann. d. Phys. (43), 282, 1914. 



