I have been testing a Tamaya MS 833 which I recently borrowed. I corrected perpendicularity and side error as indicated in the manual. The residual index error is +1.1’, computed by superposing upper and lower limbs of the Sun and Moon, which give the same result.
The instrument appears to be in very good conditions, and the reported instrumental error is 0.0’ for all reported angles.
I have been testing it by measuring the angles between pairs of stars and comparing the measurement with the prediction from Stellarium software, which incorporates refraction effects. The sextant appears to systematically overestimate the Stellarium value by ~ 1’. Do you have an idea of why?
Here is my guess: when I superpose the direct image of one star and the double-reflected image of the other one, their angular separation depends on the orientation of the sextant, see the figure attached.
I first orient the sextant in such a way that the two stars are far from the line separating the non-coated from the coated half of the horizon mirror (A), which means that they are close to the right edge of the coated part. If I rotate the sextant in such a way that the stars move to the middle of the mirror, the separation increases (B). I think that the solution would be to make the measurement by keeping both stars close to the middle of the horizon mirror?