The model which he had repaired, had a cylinder of only two inches diameter, and six inches stroke. After he had put it in complete order, he found, that although the boiler was much larger in proportion to the cylinder than those of real engines, yet, that it was incapable of supplying the cylinder with steam in sufficient quantity to keep it at work. To enable it to continue to move, he found it necessary to lessen the quantity of water raised by its pump, so as to [Pg085] reduce the load on its piston very much below the proper standard according to the common rules for large engines. He ascribed the great inferiority in the performance of the model, compared with the performance of the large engines, to the small size of the cylinder, and to its material. The cylinder of the model was brass, while those of large engines were of cast iron; and brass being a better conductor of heat than iron, he concluded that more heat in proportion was lost from this cause in the model, than in the larger engines. He observed that the small cylinder was so heated when the steam was admitted into it, that it could not be touched by the hand; but, nevertheless, that this heat contributed nothing to its performance, inasmuch as before the piston descended, the cylinder required to be cooled. (47.) His first attempt to improve the engine, was by using a wooden cylinder instead of an iron one. He accordingly made a model with a cylinder of wood, soaked in linseed oil, and baked to dryness. With this he made numerous experiments, and found that it required a less quantity of water to be thrown into the cylinder to condense the steam, and that it was worked with a less supply of steam from the boiler than was necessary with the metallic cylinder. Still he found that the force with which the piston descended was considerably less than that which the atmospheric pressure ought to supply, supposing a tolerably perfect vacuum to be produced under the piston. This led him to suspect that the water injected into the cylinder was not perfectly effectual in condensing the steam. The experiments which he had previously made on the increased temperature at which water boils under pressures greater than that of the atmosphere, led him by analogy to the conclusion that it would boil at lower temperatures if it were submitted to a pressure less than the atmosphere, and he was aware that Dr. Cullen and others had then recently discovered that in vacuo, water would boil at so low a temperature as 100°. These notions suggested the probability that the water injected into the cylinder being heated by the condensed steam, might produce vapour of a low temperature [Pg086] and reduced pressure under the piston, which would account for the deficiency he observed in the power of the engine.