Savery quaintly relates the action of his machine in "The Miner's Friend," and so exactly, that a better description could scarcely be asked: "The steam acts upon the surface of the water in the receiver, which surface only being heated by the steam, it does not condense, but the steam gravitates or presses with an elastic quality like air, and still increasing its elasticity or spring, until it counterpoises, or rather exceeds, the weight of the column of water in the force-pipe, which then it will necessarily drive up that pipe; the steam then takes some time to recover its power, but it will at last discharge the water out at the top of the pipe. You may see on the outside of the receiver how the water goes out, as well as if it were transparent; for, so far as the steam is contained within the vessel, it is dry without, and so hot as scarcely to endure the least touch of the hand; but so far as the water is inside the vessel, it will be cold and wet on the outside, where any water has fallen on it; which cold and moisture vanish as fast as the steam takes the place of the water in its descent." After Savery's death, in 1716, several of these engines were erected in which some improvements were introduced. Dr. Desaguliers, in 1718, built a Savery engine, in which he avoided some defects which he, with Dr. Gravesande, had noted two years earlier. They had then proposed to adopt the arrangement of a single receiver which had been used by Savery himself, as already described, finding, by experiment on a model which they had made for the purpose, that one could be discharged three times, while the same boiler would empty two receivers but once each. In their arrangement, the steam was shut back in the boiler while the receiver was filling with water, and a high pressure thus accumulated, instead of being turned into the second receiver, and the pressure thus kept comparatively low. [Illustration: FIG. 14.--Papin's Two-Way Cock.] In the engine built in 1718, Desaguliers used a spherical boiler, which he provided with the lever safety-valve already applied by Papin, and adopted a comparatively small receiver--one-fifth the capacity of the boiler--of slender cylindrical form, and attached a pipe leading the water for condensation into the vessel, and effected its distribution by means of the "rose," or a "sprinkling-plate," such as is still frequently used in modern engines having jet-condensers. This substitution of jet for surface-condensation was of very great advantage, securing great promptness in the formation of a vacuum and a rapid filling of the receiver. A "two-way cock" admitted steam to the receiver, or, being turned the other way, admitted the cold condensing water. The dispersion of the water in minute streams or drops was a very important detail, not only as securing great rapidity of condensation, but enabling the designer to employ a comparatively small receiver or condenser.