Sounds good to me – and it is much easier to understand!
It occurred to me that there may be a simple way to do this.
Pick a calm day.
Tweak your heading until your GS is minimum (if flying into wind) or maximum (if flying with tailwind).
Then the GPS track is your heading.
I tried it yesterday and it seemed to work pretty well. What helps a lot is flying in calm air because then your TAS or GS is much greater than the wind speed, and the error curve is then pretty flat (and small) over a good heading range.
I think this is just a special case of what @davids posted above.
but it is unlikely to be accurate in flight, due to different currents in various circuits. For example my TB20’s compass (vertical card type) swings 5-10 degrees according to how much current the alternator is putting out.
You won’t be able to compensate for this in flight either. In this case you have compass problem, which you should solve first, then an accurate compass swing could be done. This amount of influence is out of tolerance, and one shouldn’t even start the calibration with this much distrubance.
Your Sandel would be better as the flux valve is located in a better location. In your case you should either relocate the compass or get the electrical current flow sorted. Good control of current flow will minimize all EMI effects,
within one degree accuracy.
Accurate under what conditions ? Great for not getting lost on the taxiways but, as Peter suggests, not reliable in the air .
Sure, that’s what everybody does, but it is unlikely to be accurate in flight, due to different currents in various circuits. For example my TB20’s compass (vertical card type) swings 5-10 degrees according to how much current the alternator is putting out.
But nobody can tell if the compass is 5 degrees out when you are flying, because there is not much to check it against 
I am pretty sure my SN3500 EHSI is accurate – because the Sandel fluxgate seemed totally unaffected by anything in the aircraft.
The GPS method would also take far less than 2 hours and would avoid the hassle of driving the aircraft round on the compass swing area which is often grass and thus not available when water-soaked.
While it might be fun to try and prove another methode, I don’t see a good reason. Full calibration will be done in about 2 man hours, and within one degree accuracy.
A year and a half later, Genghis is nowhere to be seen 
so I wonder if anybody else would like to have a go at this?
Ideally it should be an algorithm, probably an iteratively solved one, so it could be implemented as e.g. a spreadsheet which would then run on any modern mobile device, so you could calibrate the compass fully during one flight.
It would be a really useful tool – even more useful than the relatively trivial ASI calibration that uses three GPS groundspeeds on three headings 120 degrees apart.
Lowry's 'Performance of Light Aircraft' - if it hasn't been mentioned before - free download or read online: http://goo.gl/8Vqm6t ?
2+ months later, I suspect Genghis is no longer "around".
Is there anyone else willing/able to have a go at working out a procedure for doing this, which we could try in flight?
