NavList:

This was discussed in 2008 and I reproduce a posting of mine, edited to remove one or two errors and irrelevancies:

"In a previous posting about sextant calibration, I gave the re-setting
error of my ex-USSR SNO-T sextant as about 4 seconds, its micrometer
error as about 2 seconds and its backlash error as up to 12 seconds.

On the occasions when I have directed a sextant at the sky rather than
at an autocollimator instrument, I have noticed that my determination
of index error is seldom the same twice running. At first I put all of
it down to my inexperience and shaky right arm, but I have quite a
lot of experience with making and using optical instruments and am a
careful observer. George’s requests for “Error budgets” led me to try
to find out what sort of spread of results I(and possibly others)
might expect when finding the index error of their sextants.

I made three sets of thirty careful observations for each method:
using a sharply defined land horizon about 6 km away; using the sun’s
limbs; and using a 2nd magnitude star. To make things easier and to
avoid fatigue, I clamped the sextant atop a theodolite tripod so I
could make each observation a leisurely one. I also glued a simple but
effective paper vernier over the micrometer index to reduce to some
extent a tendency to bias the results in a more or less favourable
direction when estimating tenths. Here are my results for the standard
deviations:

Method S.D. 95% range
Horizon, reflected image up to direct image 0.142 0.56
Horizon, direct image up to reflected image 0.157 0.62
Sun’s limbs 0.155 0.61
Star 0.174 0.68

For those who might think that statistics is a new form of contact
adhesive, I should point out that the standard deviation is a measure
of the dispersion of the results about the mean value; and 1.96
standard deviations each way will “capture” 19 out of 20 or 95% of
results. 1 S.D. each way will include about 64% of results. So, my
sextant-eye-brain system will give an index error of more than 0.3
minutes away from the best estimate, the mean, one time in twenty.

Physicists are supposed to be good at error estimation and my
education was in the biological sciences, so I leave the rest to
others...

Bill Morris
Pukenui
New Zealand
----------------------------------------------------------------
NavList message boards and member settings: www.fer3.com/NavList
Members may optionally receive posts by email.
To cancel email delivery, send a message to NoMail[at]fer3.com
----------------------------------------------------------------