The arms carrying the balls, or the balls themselves, are pinned to rods, _M M´_, which are connected to a piece, _N N´_, sliding loosely on the spindle. A score, _T_, cut in this piece engages a lever, _V_, and, as the balls rise and fall, a rod, _W_, is moved, closing and opening the throttle-valve, and thus adjusting the supply of steam in such a way as to preserve a nearly fixed speed of engine. The connection with the throttle-valve and with the cut-off valve-gear is seen not only in the engraving of the double-acting Watt engine, but also in those of the Greene and the Corliss engines. This contrivance had previously been used in regulating water-wheels and windmills. Watt's invention consisted in its application to the regulation of the steam-engine. Still another useful invention of Watt's was his "mercury steam-gauge"--a barometer in which the height of the mercury was determined by the pressure of the steam instead of that of the atmosphere. This simple instrument consisted merely of a bent tube containing a portion of mercury. One leg, _B D_, of this U-tube was connected with the steam-pipe, or with the boiler by a small steam-pipe; the other end, _C_, was open to the atmosphere. The pressure of the steam on the mercury in _B D_ caused it to rise in the other "leg" to a height exactly proportioned to the pressure, and causing very nearly two inches difference of level to the pound, or one inch to the pound actual rise in the outer leg. The rude sketch from Farey, here given (Fig. 30), indicates sufficiently well the form of this gauge. It is still considered by engineers the most reliable of all forms of steam-gauge. Unfortunately, it is not conveniently applicable at high pressure. The scale, _A_, is marked with numbers indicating the pressure, which numbers are indicated by the head of a rod floating up with the mercury. A similar gauge was used to determine the degree of perfection of vacuum attained in the condenser, the mercury falling in the outer leg as the vacuum became more complete. A perfect vacuum would cause a depression of level in that leg to 30 inches below the level of the mercury in the leg connected with the condenser. In a more usual form, it consisted of a simple glass tube having its lower end immersed in a cistern of mercury, as in the ordinary barometer, the top of the tube being connected with a pipe leading to the condenser. With a perfect vacuum in the condenser, the mercury would rise in the tube very nearly 30 inches. Ordinarily, the vacuum is not nearly perfect, and, a back pressure remaining in the condenser of one or two pounds per square inch, the atmospheric pressure remaining unbalanced is only sufficient to raise the mercury 26 or 28 inches above the level of the liquid metal in the cistern. [Illustration: FIG. 30. Mercury Steam Gauge. Glass Water Gauge.]