GENERATION AT HIGH TEMPERATURES.--At temperatures largely exceeding those of the atmosphere, the reaction between calcium carbide and water tends to become irregular; while at a red heat steam acts very slowly upon carbide, evolving a mixture of acetylene and hydrogen in place of pure acetylene. But since at pressures which do not materially exceed that of the atmosphere, water changes into vapour at 100° C., above that temperature there can be no question of a reaction between carbide and liquid water. Moreover, as has been pointed out, steam or water vapour will continue to exist as such at temperatures even as low as the freezing-point so long as the vapour is suspended among the particles of a permanent gas. Between calcium carbide and water vapour a double decomposition occurs chemically identical with that between carbide and liquid water; but the physical effect of the reaction and its practical bearings are considerably modified. The quantity of heat liberated when 30 parts by weight of steam react with 64 parts of calcium carbide should be essentially unaltered from that evolved when the reagent is in the liquid state; but the temperature likely to be attained when the speed of reaction remains the same as before will be considerably higher for two conspicuous reasons. In the first place, the specific heat of steam in is only 0.48, while that of liquid water is 1.0. Hence, the quantity of heat which is sufficient to raise the temperature of a given weight of liquid water through _n_ thermometric degrees, will raise the temperature of the same weight of water vapour through rather more than 2 _n_ degrees. In the second place, that relatively large quantity of heat which in the case of liquid water merely changes the liquid into a vapour, becoming "latent" or otherwise unrecognisable, and which, as already shown, forms roughly five-sixths of the total heat needed to convert cold water into steam, has no analogue if the water has previously been vaporised by other means; and therefore the whole of the heat supplied to water vapour raises its sensible temperature, as indicated by the thermometer. Thus it appears that, except for the sufficient amount of cooling that can be applied to a large vessel containing carbide by surrounding it with a water jacket, there is no way of governing its temperature satisfactorily if water vapour is allowed to act upon a mass of carbide--assuming, of course, that the reaction proceeds at any moderate speed, _e.g._, at a rate much above that required to supply one or two burners with gas.