NavList:

On 9/28/2010 7:03 PM, Frank Reed wrote:

Ok, but it's very easy to over-state this. As I have noted, I can routinely get lunar distances accurate to nearly a tenth of a minute of arc (standard deviation) when sets of four are averaged. A tenth of a minute of arc error in the angle is generally equivalent to twelve seconds error in the resulting Greenwich Time. To navigators spoiled by modern clocks that might sound pretty bad, but stop for a moment and recall the implications for the vessel's position. A twelve second error in time yields an error of only three nautical miles at the equator (or 2.1 n.m. at 45 degrees latitude). Two or three miles is NOTHING for traditional marine navigation. Even if you triple or quadruple these errors, you're still going to get you very close to your destination after months at sea. Additionally, lunars have another important property: they never get worse. A chronometer can change its rate and develop small random errors that accumulate over time. A chronometer (corrected with its assumed rate) may be in error by only three seconds, let's say, after two weeks at sea, but it might be in error by sixty seconds after three months. But you could always take lunars, and, as long as you had an almanac for the year in question, you could get the absolute time to within some fixed accuracy even after years away from a known longitude.

This opened my eyes. I have always said that celestial is a "position finding" system as are GPS, LORAN, VOR, DECCA, RDF and OMEGA while  DR and INS (and DOPPLER) are "position keeping" systems. But it actually turns out that, as to longitude, that celestial, when done with a chronometer that is not checked with radio signals, is actually a "position keeping" system, albeit with a much greater accuracy than the other "keeping" systems.

gl