Depends on the wire-insulation class, I suppose... If you use class-A wire you'd be in trouble at 250 deg. F:

http://www.anotherpower.com/gallery/dinges/Leeson_5

I use class-H insulated wire but if one uses scrounged wire one can only guess at the temperature the insulation is rated for.

Using solvents to dissolve the glue wouldn't seem right to me either. How can you be sure the solvent doesn't damage the wire insulation ?

To the original poster:

If those were my coils I would cut my losses and wind new ones. Yes, copper is expensive. But you could at least sell the old coils for scrap value and recuperate some of the expenses. Reducing the number of the turns after they're bonded with glue (or even cast?) is an exercise in futility. (unless you used lousy glue to begin with).

 Sorry to hear, that you've got some negative comments. Just come's with the territory I guess.

 One other thing, say you actually went thru the trouble, of inspecting the wire. Then found a few bad-spots, where the insulation/varnish is comprimised. I would think a spray-can of varnish, would be able to re-coat the wire with a puff or two. Then it would still be, ok to rewind(after drying). By the way, I really like Hikers comment to boil the coils, before un-winding them with less trouble, Hiker, does it make a difference if they are un-wound hot, or after they have cooled?. This should work with the super-glue, types of glue.

   I have discussed isolating 'phases' with individual diode-bridges with Flux before, he pointed out to me, the pole-to-coil ratio determines how many possible phases there may be. Personally Im going to try a single phase machine with what I have. I think it wont be the most super-duper alt, but am willing to bet it will work. Besides no matter what, I will learn something about all the variables involved.