NavList:

   

Reply

I've been off the air for several days and have lost a number of
messages in the "celestial accuracy" thread, but I have a few items to
toss in.  Apologies if any of this is redundant.


The best celestial nav I've ever seen was performed by a robot, in the
form of an astro-inertial navigation system.  Looking through a flat
porthole in the top of the aircraft, this device located and tracked
several stars per minute from a catalog of 61 stars.  (I never did see
a list of the stars.)  This was not an experimental device; it's in
service today.  It could see the stars even in bright sunlight!
Performance is still classified as far as I know, but I will say that
while flying at jet aircraft speed it could beat an expert human
standing on solid ground with a sharp horizon and a fine German
sextant.

Dan Allen - the .108' error from your mean of 5 observations was
probably good luck.  The standard deviation of your intercepts is
2.40'.  My understanding is that taking the mean of n observations
will reduce random error by a factor of square root of n.  So I'd
expect the standard error of your mean to be 2.40'/2.24, or 1.07'.  (I
realize I'm walking on thin ice, analyzing only 5 observations.)

Did you notice that correcting your refraction to actual conditions
made the result worse?


George Huxtable - a statistical study of dip error has been done.
From the 1984 Bowditch Vol 1, chapter on sextant altitude corrections:

"An investigation by the Carnegie Institution of Washington showed
that of 5,000 measurements of dip at sea, no value differed from the
tabulated value by more than 2.5', except for one difference of 10.6'.
Extreme values of more than 30' have been reported and even values of
several DEGREES have been encountered in polar regions."

Unfortunately, I have no other details.  I wonder how one goes about
measuring dip at sea.  The only method that occurs to me is to set up
a theodolite on shore, oil rig or other steady platform and measure
the sea horizon with respect to the spirit bubble.

   

Reply