Re: ATM Foucault test accuracy

["Bill T." <twentiethwave@hotmail.com>]

If a barely gray shadow is a "1" and an almost black shadow is a "10", I get
my best *repeatability* with a bright shadow of about 3.  My worst
repeatability is for shadows darker than about 6.  By repeatability I mean a
small numerical spread over a series of readings.  I get almost identical
surface results using grays between 2 and 5.  The inner window readings seem
to break down first for the darker shadows.  You can easily make these
experiments for yourself.

I've tried various techniques ranging from "fastest possible approximation"
to "nitpicking exactitude."  I see no big difference in the results.  In
many ways I trust fastest approximation more, gives me less time to fudge
the readings.  I like to record three readings for each window:

1. Left window barely brighter.
2. Right window barely brighter.
3. Windows balanced.

I have confidence in the numbers when the spread is reasonable.  Sorry, no
equations!

FWIW, I found that when I started placing my mirrors lower than my tester,
so the mirrors were tilted back in their cells, my Foucault readings went
from the under-corrected side to the slightly over-corrected side.  I
attribute this to less mirror slump, but it could also be a change in
systematic error.

Bill T.