Here's what I think, before reading e-mail this morning.
I think that G & O absorbed a factor of 3986 into their definition of "B,"
in order to express their results in cents. If I multiply the Helsinki
guys' "B" at threshold in the middle of the compass by this factor, I get B
= .002, give or take a factor of two here and there, which is about a
factor of ten below the value of .015 given by G & O for a Thoiry
Blanchet, and below the .03 used for one of the sound files that PP sent
me. The effect is easily heard in the latter, which is not inconsistent
with the threshold being a factor of ten lower.
I think this sort of blows away the idea that the old guys' designs were
sitting close to the threshold, but as I have noted, there
are inconsistencies in how B is defined in the literature, and I could
have made mistakes in the arithmetic - I am notorious for that. Also,
there are really two questions here - where is the threshold, and how
rapidly does the timbre change after the threshold is crossed? I do think
it interesting that B is quite consistent over the upper half of the
compass for the Blanchet.
I still intend to explore this experimentally myself. I think my FFT
software has good enough resolution to measure the i.h., but I have to dig
deeper into the details of how the algorithm is realized before I stake my
life or reputation.
This is fun, albeit not everyone's cup of tea. You have to be a dork.