While the experiment itself is interesting, the countermeasures (thermally isolated gripping surfaces) are so inexpensive and simple to implement that it seems a bit reckless not to. Many of the best precision levels use wood or bakelite like materials, seems like questioning orthodoxy on this point doesn't really save a lot of cost relative to the casting itself. The point made by
@MCritchley about the deflection, however small, potentially changing or inverting the curvature of the reference surface is critical. Castings are often beautiful, but this is a case where I think that form should come first. Using different thicknesses and profiles, sweeping curves and variable cross sections, all
could make the tool more susceptible to heat non-linearities. Getting everything thermally stable, even in an environmentally controlled workspace, must still be a challenge.
I know this is the
wrong thread, but in a non-conditioned environment having a (grippable) support with a much higher surface area to cross section than the rest of the tool may be asking for trouble. My instinct is you want the instrument to react as monolithically to the environment as possible. I get that is dogmatic thinking, and this
real world experiment is trying to add some data points to validate if this is really something to worry about, but I suspect we would need to make a more complex experiment (or simulation) if we are trying to validate
that design.
