the Tooth Fairy's algorithm

[Paul Erlich:]
>I tried n=32 and it's really bad. The "e" is 1584; 33:32 is 70.3203
steps, and 34:33 is 68.2209 steps.

[I wrote:]
> >Actually if n = 32 then e = 1552 and you get a max deviation of
~12� at the 39/32...

[Paul:]
> A max deviation of what from what?

Harmonic series 32-64 from 1552-tET. But if you want to look at just
the superparticular ratios it would be a max deviation of ~4� at the
64/1552 = 33/32.

[Paul:]
>In 1552-tET, 33:32 is 68.8997 steps, and 34:33 is 66.8427 steps.
Doesn't look like it's going to be a consecutive integer sequence to
me.

Your thinking best (LOG(N)-LOG(D))*(e/LOG(2)) approximations, which I
never said it did and just a couple of posts prior to this one said it
of course couldn't once the sum of "e" is too high... so that was
never the "gee-wiz, that's interesting..." part (to me), it was the
shared incremental increase by 1 of the superparticulars and the
fractions of e, and the fact that it worked the way that it did at
all, and the curious circumstance by which I just sort of stumbled
into it (etc.).

Dan