expansion of all gases decreases with an increase of pressure, as is seen from the researches of Amagat. The difference between the two coefficients of expansion, for a constant pressure and for a constant volume, is explained by these relations. Thus, for example, for air at a pressure of one atmosphere the true coefficient of expansion (the volume varying at constant pressure) = 0·00368 (according to Mendeléeff and Kayander) and the variation of tension (at a constant volume, according to Regnault) = 0·00367. [27] Permanent gases are those which cannot be liquefied by an increase of pressure alone. With a rise of temperature, all gases and vapours become permanent gases. As we shall afterwards learn, carbonic anhydride becomes a permanent gas at temperatures above 31°, and at lower temperatures it has a maximum tension, and may be liquefied by pressure alone. _The liquefaction_ of gases, accomplished by Faraday (_see_ Ammonia, Chapter VI.) and others, in the first half of this century, showed that a number of substances are capable, like water, of taking all three physical states, and that there is no essential difference between vapours and gases, the only distinction being that the boiling points (or the temperature at which the tension = 760 mm.) of liquids lie above the ordinary temperature, and those of liquefied gases below, and consequently a gas is a superheated vapour, or vapour heated above the boiling point, or removed from saturation, rarefied, having a lower tension than that maximum which is proper to a given temperature and substance. We will here cite the _maximum tensions_ of certain liquids and gases _at various temperatures_, because they may be taken advantage of for obtaining constant temperatures by changing the pressure at which boiling or the formation of saturated vapours takes place. (I may remark that the dependence between the tension of the saturated vapours of various substances and the temperature is very complex, and usually requires three or four independent constants, which vary with the nature of the substance, and are found from the dependence of the tension _p_ on the temperature _t_ given by experiment; but in 1892 K. D. Kraevitch showed that this dependence is determined by the properties of a substance, such as its density, specific heat, and latent heat of evaporation.) The temperatures (according to the air thermometer) are placed on the left, and the tension in millimetres of mercury (at 0°) on the right-hand side of the equations. Carbon bisulphide, CS_{2}, 0° = 127·9; 10° = 198·5; 20° = 298·1; 30° = 431·6; 40° = 617·5; 50° = 857·1. Chlorobenzene, C_{6}H_{5}Cl, 70° = 97·9;