We ran into Pelican Pete (literally) in Gove Australia.  He was 
ready to depart across the Indian Ocean, but had to wait until the 
local ship's chandlery received shipment of the latest 
                Nautical Almanac, 
so this seemed like a perfect time to ask a question.Pelican Pete!  What type of Ephemeris do you use?”

My favorite is Reed's Nautical Almanac, but the government Nautical
Almanac and Air Almanacs are good too.  There are also Long Term 
Almanacs for the Sun and stars.  These are very limited because the 
best celestial bodies for navigation are the Moon and planets, 
but it's a good thing to have one in your Ditch Bag, since it all fits 
on 1 page.

In the latest versions of NavPak, I use formulas and data written
 mostly by The USNO and JPL.  These appear to be better than 1 minute for 
 the Moon and 0.5 minutes for the Sun, stars, and planets, and they are 
 supposed to work until 2150!  This level of accuracy, speed, and 
 compactness was not possible before about 2010.  Previous to this, 
 the coefficients and code to calculate the planetary positions were 
 16MB and up, depending on which data set you use.

Some may think of celestial navigation as an archaic method, but modern 
tech with these compact and  accurate long term almanacs is pushing it 
over the top.  It is now viable to have one on a hand held computer, 
where previous solutions were clumsy for one reason or another.

I started using “The Almanac for Computers”, published by the US Naval 
Observatory in the early 1980s. This was very accurate, but required that 
a large number of coefficients be updated every year.  The publication 
was free, but not always easy to get in far-flung ports.

In the later half of the 80s' I used “Compact Data for Astro Navigation
by the UK Nautical Almanac office.  This was similar to theThe Almanac for Computers, except it was more compact and the
 coefficients were good for 5 years.  This was a great improvement, but 
still not easy to get in the middle of no where, and not cheap either. 

Also in the early 1980s' there was another method in the technical paper: Low-Precision Formulae for Planetary Positions.  This was compact enough 
for a handheld computer and did not require entering coefficients every 
1 to 5 years, but was not accurate enough for navigation.

   To be continued...