If we were to put water into a
glass cylindrical container of circular cross section and transport it to outer space,
the surface of the water would be uniquely determined as a lower hemisphere. Concus
and Finn [1] have shown that if the circular cross section is replaced by a square
section then no such surface can exist as a graph of a function. The question then
arises, for what kind of cross section can we expect the existence of a surface in the
form of a graph?