220 AIR PUMP the piston forcibly to descend for a portion of its motion. To prevent this loss of labor the' double-barrel air pump has been contrived, represented in our second figure. By means of a ratchet wheel moved through about 180 by the double handle H 0', the pistons are alter- nately raised and depressed in the cylinders C C' attached to the base A, which also carries the plate P, with the so-called receiver K, which is the strong glass vessel into which the objects to be experimented upon are placed. An im- portant improvement in such air pumps was the imitation of the manner of working of the steam piston and cylinder, namely, closing the top of the cylinder hermetically, by means of a packing box surrounding the piston rod, and the placing of a valve in this upper cylinder head, so as to give a pump with a single cylin- der three valves, all opening upward one in the piston, and two at the top and bottom of the cylinder. In this way the atmospheric pressure does not act on the piston during the downward stroke, as then the top valve closes, and a partial vacuum is formed over the piston, which is filled by the air under the piston passing through its valve. Such air pumps are represented in figs. 3 and 4. A is the base ; C the FIG. 8. Improved Modern Air Pump. cylinder ; P P the plate ; S the support for the handle II ; E an oil cup, attached to a cap over the top valve, to receive the lubricating fluid which the upward motion of the piston may throw out of the top valve ; and R is the re- ceiver, containing the apparatus prepared to demonstrate that the sound of a bell is not transmitted through a vacuum. In order to ease the labor when such a pump has to be used continuously, a fly wheel may be attached, working by means of one or more cranks one or more pistons, as in fig. 4. Babinet made such pumps, in which the air from one cylin- der was thrown into a second ; and Richard in Paris makes pumps with series of barrels, say eight to each pump, all moved by one axle with cranks. These remarkable pumps pos- sess some other peculiarities, described in Prof. F. A. P. Barnard's report on the Paris expo- sition of 1867. In order to do away with the great friction of a close-fitting piston in the barrel, Deleuil made a pump in which the pid- Fio. 4. Large Improved Air Pump of Ritchie. ton does not touch the barrel at all, but leaves a very narrow space between. In order to guide its motion without contact, it has, be- sides the upper piston rod which moves it, another piston rod with packing box below passing through the under cylinder head. To prevent the air from passing the piston, the latter is a cylinder of great height, nearly half the length of the barrel, and around its cir- cumference are a great number of circular grooves, each of which has to be filled with air before this can pass to the next groove, which takes much more time than each stroke of the piston, so that the pump works as if the piston were tight fitting. But the grooves, filled with air at each stroke, act as so much dead space, and thus as a slight imperfection. Kravogel, of Tyrol, makes his air pump pistons like those of the pump of an hydraulic press, of a simple solid thick iron cylinder, passing through a stuffing box into a barrel which is wider, and in which the space between the two is filled with mercury, thus absolutely an- nulling all dead space. Early attempts to pro- duce a vacuum on the Torricellian principle (see BAEOMETEB) failed, but Geisler succeeded in constructing an air pump on this principle ; it is now one of the most valuable tools in the philosophical cabinet. (See figure 5.) The glass tube C, of which the length is about equal to the height of the barometrical tube (80 inches), contains on the top a glass vessel A, while its lower end is connected by a flexi- ble tube D with the glass vessel B. The glass vessel A is connected with a tube T R, pro- vided with a double-way stopcock P, which
