Vos-curve entropy

Well, when I use the mediant-to-mediant distance instead of the 1/sqrt(n*d)
weights, the validation exercise still looks fishy, but at least looks a lot
neater . . . I'll post the results of the corresponding harmonic entropy
calculation . . .

Gene, I was surprised to find that in this case, after decreasing the
"fuzz", the shape of the curve stayed qualitatively the same! Something's
fractally about this version, as opposed to the usual normal-curve based
version.

But I don't hear 9:7 and 12:7 as huge minima of dissonance anyway.