NavList:

Paul Hirose, you wrote:
"Lunar 4.4 should be accurate to about 1 mas in the Almanac and Wallace examples."

But that's not real. While there are some things in astronomy that can be computed to that level of precision (and higher) including certain positions of certain "safe" star cases in certain defined coordinate systems, these are not meaningful for lunars (or any other form of traditional celestial navigation). For lunars, if we do not include lunar limb profile data in some fashion (no does that I am aware of, and the use of limb data would have to be defined based on the nature of the observation) then the limit on the lunar distance numbers is no better than a couple of arcseconds. That's it. Doesn't get any better than that. Anything below that level of precision is simply meaningless. Even 2 or 3 arcseconds is better than we really need by at least a factor of two. Of course, the nominal limiting precision of any modern sextant is six seconds of arc (a tenth of a minute) and realistic accuracy under excellent conditions is about the same but frequently much less. Chasing milliarcseconds could have some merit for some astronomical task unrelated to celestial navigation, but it's not significant to standard celestial altitudes by a long shot and even for lunars that level is meaningless.

Is there any risk in chasing a level of precision that is roughly 2000 times higher than necessary for lunars? Yes, because it muddles the whole process and creates imaginary requirements. For example, you include humidity as an option for refraction calculations in your "Lunars" tool, but this is entirely insignificant for lunars. It doesn't belong in there at all (disclaimer: I have never looked at your software itself --I don't like being influenced by other people's software development work-- but I did see the screen captures in Modris Ferster's recent posts).

Frank Reed