It's easy enough to find the number of turns required for a particular
inductance on a known toroid core, but how do you know what size wire
to use so that the turns will all fit on a single layer? That's this
morning's geometry problem. I suppose you can find formulas, but a
quick Google search didn't turn up anything useful for me this
morning. What I worked out, that may be of some use to others, is:

Given D = inside diameter of the toroid core, and d = wire diameter,
same units, and N = number of turns:

N = integer( pi/arcsin(d/(D-d)) (arcsine in radians...)

d(max) = D*sin(pi/N)/(1+sin(pi/N))

If you want to calculate in degrees, replace pi by 180. To allow for
the inevitable little gaps and the wire not hugging the core ID
closely, pick a wire with a diameter at least 10% less than d(max).

Also, if you use a smaller wire so the turns can be spread out or
bunched together, you'll find that you can significantly adjust the
coil's inductance that way, especially with low-mu toroids such as -2
and -6 powdered iron.

Cheers,
Tom