When a gas expands into the atmosphere it may be regarded as “raising the atmosphere” through a certain height, for the atmosphere possesses weight, equal on the average to 1033 grams on each square centimetre of the earth’s surface, or between 15 and 16 lbs. on each square inch. Suppose a quantity of air, weighing 1 gram, to be enclosed in a long cylindrical tube of one square centimetre in section. At the usual pressure of the atmosphere on the earth’s surface, and at 0° Centigrade, the volume of the air would be 773·3 cubic centimetres; and, as the sectional area of the tube is 1 square centimetre, the air would occupy 773·3 centimetres’ length of the tube. If heat be given to this air, so that its temperature is raised from 0° to 1°, it will expand, as Gay-Lussac showed, by 1/273rd of its volume. Now the product of 773·3 and 1/273 is 2·83 centimetres; the level of the surface of the air will rise in the tube through that amount. In doing so it will perform the work of raising 1033 grams through 2·83 centimetres, or 2927 gram-centimetres. Careful measurements have shown that, in order to do this work, heat to the amount of 0·0692 calory must be given to the gas. But it has been found that to heat the air through one degree, without allowing it to expand, requires 0·1683 calory; that is, the same amount of heat which would raise a gram of air through one degree, its volume being kept constant, will raise a gram of water through 0·1683°; or, in other words, the specific heat of air is 0·1683. But if allowed to expand, more heat is required--an additional 0·0692 calory must be given it; consequently its specific heat at constant pressure is greater; it is actually the sum of these two numbers, 0·1683 + 0·0692 = 0·2375. We have thus-- Specific heat at constant pressure 0·2375 " " " volume 0·1683 Ratio between these numbers: 0·2375/0·1683 = 1·41 This ratio is termed the ratio between the specific heats of air, and such a ratio is represented usually by the letter =γ=. But it is not necessary to determine both kinds of specific heat in order to arrive at a knowledge of the value of this ratio. One plan, adopted by Gay-Lussac and Désormes at the suggestion of Laplace,[28] is to actually measure the fall of temperature by allowing a known volume of gas, of which the weight can of course be deduced, to expand from a pressure somewhat higher than that of the atmosphere to atmospheric pressure. It is true that heat will rapidly flow in through the walls of the vessel; but by choosing a sufficiently large vessel, and surrounding its walls with badly-conducting material, the entry of heat will be so slow that it may, for practical purposes, be neglected. The number for this ratio, actually found by Gay-Lussac and Welters for air, was 1·376; but subsequent and more accurate experiments have given as a result 1·405, which is almost identical with that calculated above.