[Illustration: Earnshaw compensated balance wheel for watches.] The Best Clocks in the World. At the Paris Observatory the standard clock, by Winnerl, is in a vault twenty-seven metres underground. At that depth the temperature changes are less than one fifth of a degree during the year, yet the effect of barometric changes on the rate of the clock have proved to be serious. This difficulty is avoided in the Naval Observatory at Washington, by enclosing the standard clock in an air-tight case within which the air is reduced to a pressure lower than that ever shown by a barometer at that level. To avoid risks of air leaking through this case were it to be pierced by a moving axle, this clock is actuated by weights lifted electrically by a small primary battery. The slight electric current required has no perturbing effect on the clock. This time-piece, provided with an escapement of great excellence, was manufactured by Clemens Riefler of Munich. At the Observatory of the Case School of Applied Science, Cleveland, Ohio, another Riefler clock has a mean error of but .015 second per day. This means that in a year the total error is not more than 5.475 seconds, or one part in 5,760,000 of the 365 days. Such errors, minute as they are, give a good deal of trouble when they are irregular, that is, when the clock is sometimes slow, sometimes fast, in a fashion apparently lawless. When the divergences are fairly constant they can usually be traced to their source, making it feasible to apply a remedy. [Illustration: Riefler clock.] Ascertaining the Force of Gravity. A pendulum which swings once in a second at the base of a tall tower will require for the same travel a little more than a second when borne to the top of the tower, because then further from the centre of the earth. Still greater will be the difference in its periods as it swings first at the base of a mountain and next at its summit. A pendulum, therefore, is a means of learning the force of gravity at a given place, and without sacrifice of accuracy it is well that it should be as small as possible. In 1890, Professor T. C. Mendenhall, then superintendent of the United States Coast and Geodetic Survey, designed a pendulum one fourth the length of those previously used, and of admirable precision. Afterward pendulums were built of dimensions further reduced to about two and one half inches in length, with periods of oscillation of one fourth of a second. Such pendulums are easily carried to stations difficult of access, and have been employed on the summits of high mountains, including Pike’s Peak: their indications agree well with those of the larger and somewhat cumbersome apparatus previously used. Heat Measured.