“…(14), (15), (17) and (18) respectively), and replacing r d 1 ϕ (for the curved meridian) by dx (for the cylindrical shell), one can transform the results for the elliptic-toroidal shell to those for a cylindrical shell. However, the cylindrical-shell model will not exactly replicate the behaviour of the toroid, since dx (parallel to the axis of revolution) is only an approximation for the real r d 1 ϕ, an approximation that is very good in the neighbourhood of /2 ϕ π = ± (errors are of the order of /6 2 ψ , where ψ is the angle from the edge), and that becomes better as r 1 becomes larger, and exact as r 1 approaches infinity.…”