NavList:

One other issue with a whole horizon mirror: how do you do an accurate index error test? If you try the usual sea horizon test, you'll find that the views from the two directions (reflected from the index mirror and direct through the horizon glass) merge "too easily". It's visually difficcult to detect an offset between the two sides. An index error of +/-5 minutes of arc? That's visible with a whole horizon usually. But +/-1 minute? Not really. Maybe that's just an acceptable cost with that type of horizon... Or maybe there are better ways of testing index error with sextants that have whole horizon mirrors. What's the best practical approach?

Frank Reed