I've also added some t-square stiffeners to limit any tendencies of the square to squash to a trapezoidal shape.
Added a torpedo level and a plumb bob.
Added caps to the tubes so that I didn't have to keep emptying to move and refilling to use. (Caps removed during leveling and sighting of course.)
The transverse sighting line actually tends to be more of a hindrance than an aid. Good for lining up and extending the sightline left-to-right from the two water level tubes, but then it obscures the horizon. So, I've made it so that I can adjust it a little. Use it to level, but then slide it up or down until flush with a cage othogonal wire. It'll still cross the field of view, but well above or below any possible horizon line I may observe at elevations I've chosen.
Or, I might get rid of it altogether. It's just that the best way to line up water levels is to move off center and line up the tubes. But I want shoot from the centroid, which is probably just an aesthetic choice. I want those perspective lines to be centered, if only to confirm that the whole rig is truly square and has been set up horizontally, with no introduced pitch or roll.
After over-engineering this thing, it was getting to unwieldy. I'm on v1.3.
1. I'm down to just the two large sighting tubes, connected by a single tube. I've ditched the cross-connecting tubing in favor a torpedo level and plumb bob to ensure the rig is square and level.
2. I can now break it down and build it up on site so it's more portable when hiking up to the highest planned survey point.
3. I am mounting it on angle irons, set on a cross plank that will balance nicely on a good, tripod (which also has leveling bubbles).
4. I moved the sighting tubes to the outside of the cube so that they don't obstruct the perspective lines of the cube.
5. I've got a sight line strung across the midpoint transverse that, along with the plumb bob line gives me an excellent centroid index.
I've tested it and I can consistently get correlation between the converging perspective lines and the horizontal sight line of the water level tubes. Now, all I need is a horizon to "shoot." Waiting on the weather.
in the meantime, I did some calculating. If we are on a flat surface, I should find the horizon at level no matter what my elevation. But if we are on a convex surface of the size ascribed by globe advocates, then the following should be observed:
At 100' elevation, the visual horizon calculates to around 64,684' without refraction. (69,696' with standard atmospheric refraction; 74,691' on days with exceptional refraction index.)
With a camera set level, 30 inches back from my sighting guides, I should get a 0.089° drop of the horizon below level. Not much. My guide lines are 1/25th of an inch wide (0.039" = 1mm).
From 30" away, that 0.089° drop would measure 0.043-0.047", or just about 1 width of my guideline.
Doing the same thing for other elevations:
At 400', the measured drop would be 0.086-0.93" or ~2x the width of my sighting line.
At 700', the measured drop would be 0.114-0.123" or ~3x the width of my sighting line.
At 1160', the measured drop would be 0.147-0.158" or ~4x the width of my sighting line.
At 1380', the measured drop would be 0.16-0.172" or <4x the width of my sighting line.
I know all I'm doing really is detecting whether or not any drop below my sighting line occurs, but I wanted to see what to expect.
Though the main leveling tool is the water level, the added cube framework should provide lines of perspective that converge on the sight line rather than the horizon, as long as there is no pitch to the apparatus. The water levels, the torpedo level, the plumb bob and the tripod's level itself should all agree and confirm that there is no pitch tilt.
I feel this is much more precise (and duplicate-able) than the vague description given by Rowbotham in Experiment 15, that has supported the "horizon is always at eye-level" claim for years. Plus, I intend to attain elevations much higher than the top floor of the Grand Brighton.
If not--if there's something I'm missing or that's inherently flawed in this approach that Rowbotham avoided--speak up so I can address them before gathering measurements.