|
|
|
|
|||||
|
||||||
From: Alexandre Eremenko (no email)
Date: Wed Apr 20 2005 - 01:55:03 EDT
Fred,
> I wonder if you could clarify how the standard deviation,
> average error
> and maximal span were computed.
I measure some distance several times, say 7 times in 10 minutes.
A star-to star distance can be considered constant for this period.
The average is the arithmetic average.
(x_1+x_2+x_3+x_4+x_5+x_6+x_7)/7 in this case.
Here x_1...x_7 are my sextant readings.
The standard deviation.
I subtract the average from each x_j, square the difference
add the results, divide by 6 and extract the
square root. (Actually the calculator does it all for me).
Maximal span. I take the maximal value of x_j and subtract
the minimum value.
For the Lunars (where the distance changes more quickly,
I feed all 7 distances to Fred's program,
and look at the errors this program produces.
Then I do the above manupulations on these errors.
> I would prefer that the standard deviations be given
> to an extra digit.
The purpose was to detect and estimate the systematic error.
It is clearly much larger than the standard deviation.
> the development of
> proficiency is an important component in minimizing error.
Sure. I was shooting distances since October 2004
almost every evening when weather permitted.
Trying to develop this proficiency.
But when I made almost simultaneous shots of the same distance
with two sextants (The other was Bill's new Astra) and obtained
a correct result with the Astra, and +0.4 error with my sextant,
I finally concluded that there is arc error.
> I was very accurate and precise in altitude shots (with a
> Cassens + Plath, not the Husun),
My Sun shots with art horizon were always much better...
until I realized that I compare the final result, not the sextant
reading,
so I divided my error by 2:-) With actual sextant reading the
systematic error is about the same.
> The curves can take
> surprising twists. I regret I don't have one to share
You can see many on the web. Sometimes the pictures
are good enough to read the certificate.
> described calibrating an Astra IIIB using equipment to which he had
> access. He mentioned that he fit the "observations" using Fourier
I know a very clever method of calibrating your sextant
"at home" that is without stars, but you need a theodolite
or another instrument more precise than the sextant,
and with precise level. Unfortunately I don't have a theodolite.
Can describe the method to the list if anyone is interested.
(I recently read about it in one XIX century paper on
the Internet).
Alex.
|