Such were the circumstances which forced themselves upon the attention of Watt, in the course of repairing, and subsequently trying, the model of the atmospheric engine, at Glasgow. Being informed generally of the uses of the engine in the drainage of mines, and of the vast expense attending its operation, by reason of the quantity of fuel which it consumed, he saw how important any improvement would be by which the extensive sources of waste which had thus presented themselves could be removed. He saw also, that all that portion of steam which was expended, not in filling the cylinder under the piston, but in heating the great mass of metal composing the cylinder and piston, from a low temperature to that of boiling water, upon each stroke of the piston, was so much heat lost, and that the proportion of the fuel expended in evaporating the steam thus wasted would be saved, if by any [Pg090] expedient _he could make the piston descend without cooling the cylinder_. But in order to estimate the full amount of this waste, and to discover the most effectual means of preventing it, it was necessary to investigate the quantity of heat necessary for the evaporation of a given quantity of water; also, the quantity of steam which a given quantity of water would produce, as well as other circumstances connected with the temperature and pressure of steam. He, therefore, applied himself to make experiments with a view to elucidate these questions; and succeeded in obtaining results which led to the discovery of some of the most important of those physical phenomena, on the due application of which, the efficacy of the steam engine, which he afterwards invented, depended, and which also form striking facts in the general physics of heat. (49.) The first question to which he directed his experiments, was the determination of the extent to which water enlarged its volume, or magnitude, when it passed into steam. To ascertain this, he filled a thin Florence flask with steam, of a pressure equal to the atmosphere, and weighed it accurately. The same flask was then filled with water, and weighed again. Finally, the weight of the flask itself was ascertained. It is evident, that by such means, the exact weight of the steam which filled the flask, and of the same bulk of water, would be obtained. He found that the water weighed about eighteen hundred times more than the steam; from whence he inferred that the steam which filled the flask contained about eighteen hundred times less water than the flask would contain.[17]