NavList:

For index error you need something far away. How far? The parallax between the two lines of sight in the sextant should be smaller than your acceptable error in determination of the index error. For a Davis Mk III, I would say +/- 1' is good enough for index error. The distance between the two lines of sight is roughly 8 cm. So we need a distance such that 8 cm subtends an angle as seen from "out there" that is smaller than 1'. An angle of 1' is a ratio of 1/3437, so we need a distance that is greater than 3437 multiplied by 8 cm which implies a distance greater than about 275 meters. Find a building wall or an antenna or any other terrestrial feature that is further away than that and has a sharp, clear edge. Use that as a substitute for a horizon. If the sharp edge is vertical, hold the sextant horizontal. Of course, if you prefer accuracy five times better in the index error, then you would look for an edge that is five times further away. 

Other options:

• Use Venus or Jupiter (or a bright star, but planets are better since their images usually have better definition in our eyes). Hold the sextant vertical and place the direct and reflected images side-by-side. If you cannot see them both at once, sweeping quickly left to right and back will reveal the vertical offset.
• Look for a freshly-formed contrail from a jet flying high overhead. If it's producing a contrail that's nearly guaranteed to be far enough away even for a very fine index error test. Or if you can track steadily, use the aircraft's wings or fuselage directly as a horizon substitute. Normal cruising altitude for airliners is about 10 km which is farther away than the sea horizon as seen from most small boats.
• Use the Sun test, which you are apparently familiar with, on the Moon. Just be sure to use the edges from cusp to cusp (along the north-south axis of the Moon) if it's less than half full. Better yet, wait for full moon.
• If you have access to a long hallway and a decent carpenter's laser, place the sextant on a table and aim the laser through the eyepiece end of the sextant so that part of the beam travels direct through the horizon side and part bounces to the index mirror. You'll get two beams of light coming out. Measure the spacing between them a few meters from the sextant and then again some tens of meters away. If it's the same, the beams are parallel and at that point you can read off the index error.

Frank Reed
ReedNavigation.com
Conanicut Island USA