The new ad hoc "Thin White Line" theory.
Whatever the elevation is that appears to be missing is compressed into a thin optical line at the horizon boundary.
Given this theory, that would mean that on that particular day when MathiasKP made his Turning Torso observations, atmospheric conditions across the channel were such that:
1. when viewed from 29 miles away from an elevation of near sealevel, 371' of the Turning Torso (0.135° of arc) was compressed to a line, leaving the remainder mostly un-distorted and visible.
2. when viewed from 29 miles away from an elevation of 50', 205' of the Turning Torso (0.075° of arc) was compressed to a line, leaving the remainder mostly un-distorted and visible.
At present, the explanation for how the atmosphere does this, optically (other than the ambiguously-used, umbrella term "refraction") is left unexplained. Perhaps a good place to start would be producing a theoretical temperature/height gradient graph showing what kind of density lapse rate would be required to produce, not just extreme light bending of the lower portions of the tower, but also the dramatic demarcation boundary between that compression-producing super-refraction and the staighter, more standard refraction of the upper portions.
Realize, too, this has to be a rather typical and not anomalous, since the "thin white line" theory tries to account for whenever object elevation is apparently missing and atmospheric refraction is given as the reason. Such a temperature gradient has never been recorded. This could be groundbreaking, Tom.