Originally posted by Gimbal I know that more advanced pendulum clocks have the pendel arm built in such a way that temperature variations doesn’t change the length of the pendel, and hence doesn’t affect the clock speed. Is that it?
Yes. The period (=swing time) of the pendulum is critical to the time-keeping accuracy of a clock that is driven by a swinging pendulum. The period is determined by the length of the pendulum only, not by its mass (~weight). and the length is measured from the pivot to the center of mass, not the tip end. The length of the pendulum will change with temperature, longer when it is warmer (causing the clock to run too slow), shorter when colder (causing the clock to run too fast). "Regulator" clocks solve this by having a mass at the end of the pendulum that expands UPWARD when the temperature rises, thus compensating for the increased length of the main pendulum. Typically, on small Regulator clocks such as the one pictured, common in RR stations, schools, business offices and the like, the regulator mechanism is a series of rods behind the primary pendulum mass (the brass disk at the end) with some bit of mass that the rods "push" upward when the temperature rises. Needless to say, it takes some careful calculation and precise cutting and weighing of materials to make a really accurate regulator clock, but once a manufacturer has determined the values for one particular model/size of clock, they can produce them in thousands. I once saw a massive tall-case ("grandfather") clock that had four glass vials of mercury mounted at the end of the pendulum in addition to a decorative brass piece. I suspect that the mercury expanded as the temperature increased, rising up a tiny bit in the glass vials, and thereby "regulated" the period of the pendulum to compensate for small changes in temperature.
On the clock pictured by Sailor, between the "L" and "A" of "Regulator" there appears to be a thin rod that parallels the much thicker primary pendulum rod. That
might be part of the regulating mechanism, but I cannot tell for sure.