"The construction is as follows: Let there be a sphere of a capacity of about six cotyles (about 2¾ pints) made of some metal tough enough to withstand the pressure of the air that is to be produced. Let us place this sphere, _ΑΒ_, upon any base whatever, _Γ_. Through an aperture in its upper part we introduce a tube which runs down to that part of the sphere which is diametrically opposite the aperture, but which leaves sufficient space there for the water to pass. This tube projects slightly above the sphere, to whose aperture it is soldered, and divides into two branches, _Η_ and _Ζ_, to which are affixed two bent tubes, _ΖΜΝΞ_ and _ΗΘΚΛ_, that communicate internally with _Η_ and _Ζ_. Finally, in these tubes, _ΗΘΚΛ_ and _ΖΜΝΞ_, and in communication with them, there is adapted another tube, _ΠΟ_, from which issues at right angles a small tube, _ΡΣ_, that communicates with it and terminates at _Σ_ in a fine orifice. If, taking the tube, _ΡΣ_, in hand, we revolve the tube, _ΠΟ_, the two apertures that face each other can no longer establish a communication, and the liquid that rises will no longer find an outlet. Then, through another aperture in the sphere, we insert another tube, _ΤτΦ_, whose lower orifice, _Φ_, is closed, but which has upon the side, toward the bottom, at _Χ_, a round hole to which is adapted a small valve of the sort called by the Romans _assarium_. Into the tube, _τΦΤ_, we insert another and closely fitting tube, _ΨΩ_. Let us now remove the tube, _ΨΩ_, and pour liquid into the tube, _τΦΤ_. This liquid will enter the cavity of the sphere, through the aperture, _Χ_. The valve will open in the interior, and the air will escape through the apertures in the tube, _ΟΠ_, of which we have already spoken, and which have been so arranged as to communicate with the tubes, _ΗΘΚΛ_ and _ΖΜΝΞ_. When once the sphere is half full of liquid, we incline the small tube, _ΡΣ_, so as to shut off all communication between the corresponding apertures, and then push down the tube, _ΨΩ_, and drive into the interior of the sphere the air contained in _ΤτΦ_. This requires some force, as the sphere itself is full of liquid and air, but the introduction is rendered possible through the compression of the air, which shrinks into the empty spaces that it contains within itself. Let us now take out the tube, _ΨΩ_, again so as to fill the tube, _ΤτΦ_, with air, and let us push down the tube, _ΨΦ_, again and force this air into the sphere. On repeating this operation several times in succession we shall finally have in the sphere a large quantity of compressed air. It is clear, in fact, that the air introduced by force cannot escape when the piston-rod is raised, since the valve, pressed by the internal air, remains closed. If then, replacing the tube, _ΡΣ_, in a vertical position, we set up a communication again between the corresponding apertures, the liquid will be driven to the exterior through the