46 SCIENCE PROGRESS 



It is not the purpose of this article, however, to derive the 

 most accurate corrections or to discuss the relative merits of 

 the various methods by which such corrections have been 

 obtained, but to indicate the most approved modern lines along 

 which such investigations proceed. One very striking fact is 

 the extreme accuracy of such measurements, even at tempera- 

 tures such as + 500 C. or — 250° C. Tenths of degrees are 

 capable of exact determination, and at the latter temperatures 

 even hundredths of degrees are determinable with certainty, with 

 carefully prepared and calibrated instruments. This accuracy 

 is really necessary at low temperatures, on account of the much 

 higher proportion of the temperature which one-hundredth of 

 a degree has at, say, 50 K. than at 300 K. It is clear that such 

 an accuracy is only obtainable when every possible precaution 

 is taken, and, in particular, when the temperature of the gas 

 which is being measured is kept constant to about one- 

 hundredth of a degree. For all isothermal work at low tempera- 

 tures the reservoir of gas is immersed in a liquid which is caused 

 to boil at the required temperature by adjusting the pressure 

 on it. The vapour pressure of a pure liquid diminishes with the 

 temperature according to the relation expressed by the border- 

 curve between liquid and vapour, which can be deduced by 

 corresponding states from one accurate series of measurements, 

 or, better, by direct measurement in each case. 



As, however, in practice it is impossible to keep gases quite 

 pure, the liquefied gas will be more or less a mixture, and the 

 temperature at which it boils under a given pressure will 

 change as the more volatile component boils away. Such a 

 condition is very well exhibited by the boiling of liquid air. 

 Here the normal boiling points of oxygen, freshly condensed air, 

 and nitrogen are respectively 90 K., 82 K., and 79 K., hence that of 

 air is very nearly obtained by the sum of the proportions of 

 liquid oxygen and nitrogen contained in it. When the air is 

 boiled, the more volatile nitrogen boils away, so that the liquid 

 becomes continually richer in oxygen and the temperature rises 

 until a steady state is reached at which the mixture boils as 

 a simple substance. Hence it is clearly not possible to keep a 

 temperature constant by boiling liquid air at constant pressure, 

 and this is true of all gases used for such work, although, where 

 the amount of impurity is small, the total change of temperature 

 may be small also. It is necessary to have an elaborate system 



