No subject

change in the rod length.  This has actually been measured, and is
accredited to crystal lattice slippage of the material.  This has been
found in metal and recently, crystal rod pendulum.  It is significant in
the latter because experimenters went to crystal rods for their
temperance invariance, only to find the unpredictable spontaneous length
change.

Carl Dreher
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
In : Accurate Clock Pendulums by Robert Matthys (2004) Oxford =20
University Press ISBN 0198529716, Pp264
In Chapter 8 , "The Allen variance and the rms time error", on page 38
he writes:
Figure 8.1 shows another characteristic of pendulum clocks - the =20
clock will run relatively
smoothly at one rate , and then after 3 - 6 months it will suddenly =20
jump to a new rate
as shown if Fig 8.1.

He goes on to say that this is part of a random walk process.

I hope that is of some help.
Neville MIchie