From: Frank Reed
Date: 2020 Oct 15, 11:53 -0700
I would say that your problem is that you were taught (or have read) that you need to eliminate all "side error". That's the apparent left/right gap between the direct and reflected images. This is totally un-necessary. No error arises if there is a gap. The expression "side error" is a misnomer. Adjust your sextant leaving a left/right gap of, let's say, one minute of arc. Then you can place your star images side-by-side without much effort. This will yield good results if you can't try a tabletop approach (as below).
I spent many years testing various methods for testing and reducing index error. The best that I have found requires a small "spotting scope" --a telescope with a magnification of 30-50x. Place your sextant on its side on a table looking at some suitably distant vertical "edge". Remove the standard scope. Mount the spotting scope on a small tripod so that it is aligned with the usual light path that the standard scope would see. You may want to place a temporary paper or cardboard baffle around the horizon glass so that you can guarantee you are only seeing the scene through the normal sextant view. With a magnification around 40x, ever tiny tenth of a minute of arc adjustment is directly visible, and you can repeatably and reliably assess (or better yet, zero out) your index error. With a good metal sextant, this adjustment or assessment will probably last for a whole observing season, possibly even for years.