NavList:

We can categorize errors as random or systematic. Random errors have equal probability of being positive or negative. If the standard deviation of the random error distribution exceeds the precision (size of smallest division on the scale) of the measurement then by averaging a series of measurements on a single body when can minimize the random error down to the level of the precision of the instrument. By plotting altitude versus time we can spot potential mistakes ie outliers not part of a random distribution. Such measurements are mistakes and should not be included in averaging.

Systematic errors will still remain.

For systematic errors we may or may not know what they depend on. If we are worried about sun heating the sextant causing a shift in our readings, absent measuring the sextant temperature and having a model for the error as a function of temperature we will be unable to deal with such errors. In the case of heating it is not connected azimuth so it tells us nothing about being inside or outside of the cocked hat.

For something like index error(which we should have eliminated down to the precision of our measurement) but if we made a mistake in that factor then that error affects all measurements the same. Likewise, anomalous refraction ( in Dip) will affect all measurements the same and the placing of a fix at points inside or outside the cocked hat based on azimuths is a sensible procedure.

Sextant calibration measurements as was common with older instruments are systematic but must be applied to each measurement. They vary with altitude rather than being constant for all azimuths.

Dave Fleming
----------------------------------------------------------------
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
----------------------------------------------------------------