The airstone drawing from the air inside the tank makes sense, but most air pumps I've seen don't draw air from a hose.

To fill the inverted tank you could put a tube up inside it and vacuum the air out. This will draw water up into it. As long as air isn't allowed in, it shouldn't let the water out. I wonder if the structural requirements of holding the water in that way differs at all from holding the water in the traditional arrangement. It would be a vacuum type stress, rather than pressing outward.

The pipe one would have worked the same way, but much greater vacuum would be required. Plus you would have to have lots of water available to add to the tank as it is sucked up. If the seal were to become breached by the water level dropping below the tube, it would let air in and water out... probably not in a good way!

Michael