NavList:

   

Reply

Peter,

On Fri, 9 Mar 2007, Peter Fogg wrote:

> I guess this is true so long as
> the errors are truly random,

Yes. Mathematicians say "independent and equally
distributed" which approximately relates to certain
type of errors in reality.

> In practice only a limited number
> of observations of the same body is
> usually possible,

Various manuals recommend 5 to 10.
(If more, you get tired. This is verified by my experience).
My air sextant averager averages 60, and this does not
help. I mean my single individual observation
are better than these mechanical averages of 60.

> In practice the sort of gross error that
> tends to crop up is writing
> down the wrong minute of time,

Sure. This is in complete agreement with my experience.
Wrong reading of the watch is the most frequent blunder.
Wrong reading of the sextant scale is rare:-)
Then come various other blunders. Using the wrong limb.
Even using the wrong date, and even the wrong year
of the almanac:-) But such gross
errors are easily detectable

> Averaging would have simply led to an
> erroneous result, particularly with few observations.

When I just look at the results before the averaging,
the gross outliers are usually visible with the naked eye.
For example, if the alt changes in the wrong direction.

> Comparing the slope is a kind of averaging,

In principle I agree that graphic methods are
better, especially to select the outliers.
But they are very time consuming. And prone to additional
blunders.

Alex.

P.S. From my own experience (several years with SNO-T)
I know that random errors play a lesser role in comparison
with systematic errors. I mean that in series of 5-10
observations I usually get sigma=0.2'.
However, the systematic errors,
whose source is unclear to me,
skew the result by 0.5 and more, sometimes to 1'.
This could be a sextant defect (which I cannot find despite
my efforts), or some fault of the observer which
I also cannot understand.


--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to NavList@fer3.com
To unsubscribe, send email to NavList-unsubscribe@fer3.com
-~----------~----~----~----~------~----~------~--~---

   

Reply