Oops. actually, it is not. You mention that the strange shape where the tangent meets the sphere meets is because of the different lengths of the tangencies, well, as my boss pointed out, this is the same case on the rounded corners of the square section as well. Or should be.