If no further contrivance was used, the mercury would fall into the ball A in large drops; which would make the instrument less accurate. For the thermometer’s beginning to rise immediately after a drop is fallen, or just as it is going to fall (in which case it will return back into the tube), will make a difference of such part of a degree nearly as that drop answers to. To prevent this inconvenience, the top of the shorter leg, close to the ball, is contracted, by being held in the flame of a lamp; and the passage is further streightened by a solid thread of glass placed within the tube, and extending from the bottom of the shorter leg to the part near the ball A, where it is most contracted. By this means, as soon as any small portion of mercury is got beyond the end of the thread of glass, it breaks off, and falls into the ball in very small drops. This thread of glass is fastened by the heat given to the tube in making the bend next to the ball. In order to fill the shorter leg with mercury, to fit the instrument for a new experiment, it must be inclined till the mercury in the ball covers the orifice of the tube _n_. The cylinder being then heated, the mercury will be forced into the shorter leg, and will run down the thread of glass in drops, which will soon unite. By this means, such a quantity of mercury must be got into the shorter leg, as, upon the cooling of the instrument, will be sufficient to drive all the spirit of wine into the ball with a less degree of cold than what the thermometer is likely to be exposed to. The ball A must always have some mercury in it, but never enough to fill it up to the orifice of the tube _n_. It must therefore be made of such a size, as to contain all the mercury, which can come into it from the tube without being too full. If it should happen to be made too small, so as to be too full in cold weather, any part of the mercury may easily be driven into the cylinder, and got back again into the ball when wanted in warmer weather.