Most complete collection of Michelangelo's drawings, £25.00
Explore / Articles
Temperature compensation devices
By the end of the seventeenth century it was recognized that changes in temperature caused variations in timekeeping. During the eighteenth century various compensation methods were devised. The earliest method was the mercurial pendulum invented by George Graham in 1721. A pendulum rod changes in length as it expands in heat or shrinks in cold. This affects the rate at which the pendulum swings and thus the time the clock keeps. Graham's pendulum consists of a bottom-supported jar containing mercury attached to the pendulum. The top suspended rod expands downwards in heat and the mercury in the jar expands upwards to maintain a constant centre of oscillation for the pendulum. The opposite contraction occurs in cold.
In 1725 John Harrison investigated the expansion of metals caused by the effects of heat and cold. In 1727-28 he developed the gridiron pendulum consisting of four steel rods (in addition to the pendulum rod itself) and four brass rods. The extra rods are arranged in a frame which allows them to move freely in relation to the pendulum rod. However, they are riveted so that the steel rods can only expand downwards and the brass rods upwards. This keeps the pendulum bob at a constant position.
The earliest form of temperature compensation device in watches and chronometers consisted of applied bimetal strips, invented by John Harrison. The strips reacted to temperature change and acted on the balance spring, changing its effective length to make it operate faster or slower. Pierre Le Roy in Paris invented the first self-compensating balance in 1765. Changes in temperature caused the elasticity of the balance spring to change and to compensate, an attached mechanism was fitted to the oscillating balance.
More recently, pendulum rods for accurate regulators have been made from a metal alloy called Invar. Temperature changes cause this material to change in length by tiny amounts or, in some cases, not at all.