Thanks for the kind words Luke. Unfortunately, I didn't bother to take any pictures of the motor when I had it apart and it's not glued back together again in my electric boat pod.
I found a fair bit of information on motor winding from some really good links that Ron Sommeren posted on one of the RC groups, although I forgot to bookmark the post. I'm sure a search on one or other of the RC forums would find his list of links on motor winding, it's really very good indeed for basic stuff, like how to do the wind.
I did take care to do a sketch showing exactly how my motor was originally wound before I took it apart. I also took careful note of where the wires came out from each coil, the number of strands of wire used etc. The motor was wound as Distributed LRK (often abbreviated to dLRK), which seems to be the most common method for a 12 slot stator.
I searched high and low for a way to work out how many turns I could get on the motor at any particular gauge of wire, including resorting to drawing a slot up in CAD and modelling it. What I found was that it was difficult to predict with accuracy, as it's very hard to keep windings absolutely neat, especially if they are wound as they are originally, with multi-strand wire. I think I would my motor three or perhaps four times in total, until I got a good feel for the best compromise between low winding resistance (for high efficiency) and high number of turns (for low Kv). My aim was to directly drive a boat prop of 12" in diameter at a maximum rpm of around 800 at the highest possible efficiency from a 63mm diameter motor though, so a very different set of requirements from those needed to drive a bike.
Knowing what I know now, and applying it to a bike motor, I'd suggest going for the fastest motor speed that you think you can live with, probably set by the practical reduction ratio you can accommodate without too great a power loss. Spinning motors faster gives more power for little or no increase in heating loss in the windings. The downside of this approach is noise and complexity, plus an increase in mechanical loses, things I didn't want in the boat unit, but which might well be acceptable on a bike.
Jeremy