A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Position-Finding
From: Peter Monta
Date: 2019 Jan 7, 13:36 -0800
In addition to the remarks from Brad and Greg, here are a few ideas:
Maybe the AH mirror is not flat, though it would have to be very warped to give errors of 40 arcminutes or more. How is it mounted to the plastic cutting board, and what is the thickness and diameter of the mirror? If it's just set on the plastic and held by gravity and friction, that's one thing, but if there's adhesive involved in fixing the glass to the plastic, that might induce warping. Can you look at the image of the two suns using various parts of the mirror? In other words, set up for a sight as normal, bringing the horizon and index sun-images tangent at the limb, but then physically move the sextant so that the horizon path is reflecting from various parts of the AH mirror, such as the center and then somewhere near each of the four corners. Does the image remain stable or do the two suns start to overlap or separate? That would indicate the mirror is out of flat.
What is the sensitivity of the level? Apparently the Starrett 98 comes in various sizes, the smaller of which has a sensitivity of around 90 arcseconds per 2 mm division. That's probably okay, and certainly it's fine for debugging problems on the order of tens of arcminutes, but ideally you might want a little more sensitivity, say around 20 arcseconds per division. That will put the AH error below the noise floor of the sextant.
What sort of sextant are you using? If it's a metal one, it's unlikely there would be large arc errors. Even a plastic one will be below 10 arcminutes out of the box, and below 5 with even a casual calibration. Does the index mirror look okay? If something catastrophic happened to it, such as being replaced by a plastic toy mirror, then all bets are off. :-)
Greg's guess that you may be using the wrong limb, or dividing by 2 and subtracting the solar SD in the wrong order, sounds plausible. If you're using a star rather than the Sun, then of course this bug can't arise.