NavList:

Frank writes:

And if chasing that accuracy [0.25 arcmin] is the goal, then a leveled mirror is probably hopeless.

For portable mirror setups, the mechanical mount and earth connection is going to be a problem: even a surveyor's tripod might drift 10 arcseconds or more over the course of a 20-minute observation session.

But for a permanent installation, I really don't see a limit to how good a mirror can be.  Fix it to a rock or concrete pier, supply some means of trimming its orientation while keeping rigid, and spirit-level away.  Even the plate level on a construction-grade theodolite is 20 arcseconds per division, so levelling up to 5 arcseconds should be (and is) routine.  A more sensitive level might be 5 arcseconds per 2 mm division; below this they become slow to use, so electronic ones are preferable.  A bit below the sextant noise floor is enough.

Leveling gets quicker with practice.  Just keep any dust particles away from the level's feet or the mirror's surface (they can be "felt" by very gently sliding the level).  And of course levels are self-calibrating by reversal, and they're used here as a null instrument so their linearity is not a concern.

Cheers,

Peter