Just my two cents, I haven't done a calculation or tests (I have done calculations for mirror warping when you pull on one point at the center with the edges supported, and when you apply pressure evenly over the back, to "figure" a spherical mirror):
The author of that article in Sky and Telescope did not seem to choose the best-fit paraboloid to the distorted surface, and if so he got unnecessarily pessimistic results. OTOH, Texereau is certainly wrong about getting a radially symmetric figure when you support at three points at the edges. Maybe it's *closer* to radially symmetric than if you support at 0.707r.
Now what happens if you use a soft support at the edges so the weight is more evenly distributed? The figure will change, I would think, so maybe it's not a good idea.
Clay Spence cspence@sarnoff.com