NavList:

I wouldn't agree with your description of plastic sextants as training aids, suggesting that they are merely training aids. They're not just training aids. Davis sextants are probably the most widely used sextants on yachts and smaller sailing vessels at sea right now. Real ocean sailors navigate with plastic sextants on a daily basis.

And plastic sextants are not that bad. Myself, I routinely get errors with a standard deviation of about 2.0' from well-adjusted Davis sextants. Here I'm referring to pure sextant error as determined by lunars or star-star angles or other dependable angles. Similar observations with an average metal sextant with an ordinary scope have a standard deviation of about 0.5' and about 0.25' with an above-average metal sextant with a better scope. Since system error from other sources in standard celestial navigation (excluding sights like lunars) amounts to about 0.8', most of the additional accuracy in metal sextants is un-necessary and irrelevant. Errors in lines of position with metal sextants amount to about 0.9' while errors in lines of position with plastic sextants amount to about 2.2' (standard deviation sense, in both cases). For nearly all aspects of ocean celestial navigation, this is considered excellent. In fact, most folks navigating on the ocean have so many bad habits, so many general errors in procedures both in the sight-taking process and in the analysis of sights, that few get anywhere near these levels of accuracy in their sights. But that's not the fault of the sextant, whether plastic or metal.

You mentioned the drunken wander of the index error on a plastic sextant. And this is definitely a major problem because the textbooks and manuals on celestial navigation provide no guidance on this. The solution, however, is quite simple: zero it out. When using a plastic sextant, the navigator should zero out index error before each celestial sight if possible. This adds little work to the whole process, but many navigators skip this step because it's not described in the textbooks and because they're accustomed to the idea that index error should be measured and recorded rather than zeroed out. On (Davis) plastic sextants, it's easy to zero out the error since the sextants are built with easily-manipulated thumb screws. By the way, this is one issue where the split horizon mirror on the Davis Mk. 15 beats the whole horizon mirror on the Davis Mk. 25. It's harded to judge index error, let alone zero it, on the latter model.

So what about prismatic error in the shades of a plastic sextant? I've seen some, yes. Then again, I have also seen prismatic error on metal sextants, in particular a couple of Tamaya-likes manufactured in the 1970s. I think it's fair to say that you can discount prismatic error in shades on any metal sextant manufactured in the past 25 years, primarily because there are no "cheap" manufacturers left, but before then it's not guaranteed.

How do you test for shade error? On a plastic Davis sextant, I usually roll in all of the direct shades for a Sun sight. How can I know their effect? In the modern world, it's not difficult. First, let's assume that we're primarily interested in shade error for a particular set of shades that is standard for reducing the brightness of the Sun. One possibility is to shoot Moon-star lunars, which don't require shades and compare with Moon-Sun lunars, which will require full strength shades. Another option: shoot some Sun sights at regular angular intervals. Then compare those altitudes with expected values. And how do you get those expected altitudes? Well, you get your exact observing location from GPS, you get the exact time from GPS, then you do a reverse calculation from proper ephemeris data and other details, like your exact height of eye, to calculate what the sextant altitude should be. Sounds like a lot of work, right? If only we had an app for that... And we do. This, of course, is exactly what my GPS Anti Spoof app does (here's the Ocean Navigator article about it).

I have tried this with two Davis plastic sextants using lunars. On one, there was (if I remember correctly --this was some years ago) a consistent +1' error at all angles (probably shade error?). On another, there was a variable pattern: 0' at 30°, -0.5' at 45°, -1' at 60°, -0.5' at 75°. The results were repeatable. In both cases, I always zeroed out index error before each sight.

By the way, I'm not trying to insult the old Ebbco sextants by leaving them out here. I've taken sights with Ebbco sextants, but I just don't have any hard data on them to form an opinion.

Frank Reed