Expanding on this coil business...
As I see it now, the whole idea of "Test Coils" takes on a new meaning. If I was building alternators and not following a specific plan I might try this to calculate coils. First build a test bed of a steel disc, say 12 inches diameter. Affix it to a motor that can be set to specific rpm's. One way would be to use a DC motor with a Pulse width modulator to control it.
(side note, for those of us that don't have anything to measure rpm's do they still make those discs with lines like bar codes on them that when spun under flourescent lights indicate the rpm's? They were used to calibrate or check phonograph turntables.)
Anyway now you have a base disc where you can draw circles at 8 inches, 10 inches, etc. Also you could draw out lines indicating the 360 degrees of a cicle at 5 degree intervals. Now you can arrange your magnets on the test bed for 6 pole, 8 inch diameter or 12 pole, 10 inch, 16 pole 12 inch whatever. Then get a $5 package of magnet wire from Radio Shack and you wind test coils for the TYPE and SHAPE magnet you will be using. I think there's enough wire in that 3 pack to make a 50 turn, 60 turn, 70, 80, 90, ect coils, and it doesn't matter what gauge you use. You find which coil gives you the volts you want at a given rpm* using given magnets at a given speed/ diameter disc and then make your real coils using larger wire at that number of turns. To determine what size wire you can get into the area alloted it you could use the info Flux gave me in this thread,
http://www.fieldlines.com/story/2008/1/19/32356/2251 I know that there are variables, physical dimensions, losses due to resistance, eddy currents etc. that will skew results somewhat, but on the whole, is this sound thinking?
* for 9 coil, 3 phase I think that would be 1 coils volts times 3 times 1.7