The line of sight to the moon, or stars, or whatever does have more atmosphere in the way than the line of sight to the horizon in most instances, as thesublime has already pointed out. From Wikipedia (
http://en.wikipedia.org/wiki/Horizon), you can calculate the distance to the horizon depending on how high you are. At 1.70 m, the distance to the horizon is approx 4.7 km. So that's 4.7 km worth of atmosphere you are looking through just to see to the horizon. However, to look at the moon, you are looking through at least 6 km of atmosphere and at most 20 km of atmosphere (since the Troposphere (which contains around 3/4 of the earth's atmosphere) stretches from 6-20 km, the higher numbers being around the equator). So even if you were looking at the moon straight on you would still be looking through around 20 km of troposphere at the equator. The closer to the horizon the moon gets, the more atmosphere you are looking through. Even if you couldn't see stars near the horizon (which in a a really dark area would not be a problem) Tom, you're still looking through more atmosphere when they're 5 degrees above the horizon!
Furthermore, the semi-opaque nature of the atmosphere on a clear day cannot account for the dissapearence of ships at the horizon, since if you rise higher, say to 100 meters (where the horizon line is now at a distance of 36 km) you can once again bring the ship into view, despite the fact that you are not looking through more atmosphere than before. So I'm sorry Dogplatter, but when you're looking at the horizon from a boat or the seashore, the semi-opaque nature of the atmosphere cannot account for the dissapearence of the ship, for if it was responsible, then it should remain obscured even if the observer's altitude rises. Furthermore, the ship does not 'fade' until you can't see it anymore, it 'sinks' very distinctly. I have not seen the same effect in very choppy seas so I cannot give you an honest answer to that. First of all, you'd have to show me evidence that the sinking effect is more pronounced. Second of all, if it is, I would suggest that very choppy seas are typically associated with bad weather, in which case the horizon itself may be poorly visible. Or depending on just how choppy the seas are, waves could directly obscure the line of sight to the horizon or the ship itself (although this is where we're talking about those tsunami-like waves). In any case, having not observed the phenomenon in both set of circumstances myself, I will refrain from speculating further.
I would also say that the diagrams do attempt to compensate for your 'compounded wall of water.' thesublime didn't correct for it appropriatelly as I pointed out, but I did mention in the post right after what it would look like (i.e. the pale blue line obscuring the view). You will agree I'm sure that any compound wall of water to the horizon would only obscure the true horizon by a set distantance, and no more. The ship would therefore shrink before it became obscured by this wall. It would not 'sink'. But if this phenomenon isn't clear enough with the ship, then let's discuss the same effect with the sun or the moon, it makes no difference to me. When they set, neither shrink until they are obscured. They both sink beyond the horizon very definitivelly and within a couple of minutes or so, without shrinking at all. Please do not counter by saying they actually look bigger and thus some effect of the atmosphere is covering the shrinking that happens in reality. I can assure you, they only 'look' bigger upon initial observation. If you take a ruler even and compare their sizes, you will see that they are not at all bigger near the horizon than at their zenith.