[ATM] (no subject)

[Vladimir Sacek <vladis.2@juno.com>]

Nils,

>...but what if you don't get a tremedously accurate estimate of a figure
that is middling to lousy anyway?<

The problem is that it may be not even accurate, let alone "tremendously
accurate". In making a mirror, no one starts out with perfect or
near-perfect parabola to Foucault test. The test is there to guide you to
as good parabola as it can be, and to have it confirmed in te end. If you
don't know for sure where you are with the mirror at a given intermediate
moment of transforming it from a sphere into parabola - or have wrong
idea about it - how do you figure out the right way to proceed? Doesn't
look like win-win situation. And such scenarios are exactly the source of
frustrating inability to come to - or to confirm - even a decent
parabola. 

I'm not suggesting that this is caused by test accuracy limitations
alone. I merely point out that they are present. Errors resulting from
setup/application/external factors are likely to be more of a problem.
The test is simple in its principle, but the application can be tricky.
Relatively small deviations/omissions from the proper setup/procedure, or
unrecognized external factors, can easily make it a nightmare. Also, the
Foucault is 
"designed" for nearly perfect conic figures; the rougher/wavier figure,
the lower its accuracy. I think it's fair to say that really effective
use of the test requires not only its proper setup/application, but also
near-standardized parabolizing technic that keeps mirror's conical figure
in a decent (or better) shape throughout the process.

Vlad
_______________________________________________
ATM mailing list http://www.atmlist.net/