NavList:

If you draw a line on a wall to serve as a surrogate horizon, the sextant must be at exactly the same height above the ground for every sight. This is feasible for a single individual standing on a solid surface, but if there's a second observer or a change in the ground level or even just a change in posture, a nearby horizon line will be inaccurate. It all depends on the specific distances and required accuracy, of course. For a minimum accuracy of one minute of arc, the horizon line would need to be marked at a height relative to the sextant's height above the ground with an accuracy of about one inch in 300 feet (that's a ratio of 1:3600 while the ideal ratio would be closer to 1 to 3438 so this actually yields an accuracy limit of about 0.95' rather than 1.0' ...close enough!). So if you paint a white horizontal line on a structure 100 yards away, measure a celestial altitude relative to it, and then lower your sextant by just one inch, the observed altitude will decrease by one minute of arc. And the change is proportional: if two observers use the same line at the same distance and one is five inches taller than the other, then their observed altitudes would differ by 5'. And if the painted line is 100 feet away instead of 300, then that difference would be tripled. 

Frank Reed