The bottle-o-phone project never really panned out. A sad story, really. The idea came to me while taking Doug West's wonderful graph theory course at UIUC.
The basic idea was to have a set of twelve two-liter Mountain Dew bottles filled with various amounts of water tuned to the notes of the scale. Obviously it would be difficult to play anything with such a setup, since it would be difficult to move from bottle to bottle, blowing across them. Playing multiple bottles at the same time would be out of the question. So a set of flexible tubes would be attached to each, aimed across the top of the bottles. Early experimentation showed that this could produce a respectable sound.
The tubes would be collected into a hexagonal grid, which the musician would blow into. So the key to the whole bit is figuring out how to arrange the hexagonal lattice of tubes to produce a set of chords. I refined this into figuring out the smallest regular hexagon that could produce all of the major and minor chords in every key, with every chord being a triangle of tubes. I never analyzed the problem in any great detail, but i did find an arrangement of tubes in a hexagon that almost met the requirements. It was missing B minor (ouch) and F# major (meh), but had only three tubes on a side and only two notes (bottles) that needed to get fed by multiple tubes.
So the nice thing about this arrangement is that you can just make a "mouthpiece" of three short, rigid tubes fixed together in a triangle. By walking it around the hexagon, you can move from chord to chord. I was happy enough with this configuration to start construction, and got as far as building a frame, mounting and tuning the bottles, and building the hexagon of tubes.
It was at this point that I realized that it was hard enough to blow on one bottle for any significant length of time, let alone three at the same time. Thus the whole point of building essentially a thing for playing sustained chords was defeated. The only way to get it to work would be to have a pump to blow on the tubes, which was more than I cared to implement.
But all was not lost! The next step was bophosim, the bottle-o-phone simulator...