[math-fun] Isn't it odd that when you flex a circular wafer,

it fails along a diameter--the widest point? Presumably, sufficient elongation parallel to that diameter will bifurcate the line of probable failure. E.g., is there a shape that half the time breaks 1:2 and half 2:1? Convex? Could it be an ellipse? Is there a shape where the crack might be uniformly anywhere? --rwg