Comment by mannykannot

As the ratio used is a rational approximation of an irrational, I would guess that the ratio-preserving feature breaks down well before you get to atomic sizes, though I have neither proved that to be so nor figured out how to calculate the divergence.