Let me add:
The 38º rule probably applies, but it may not be the optimum. I don't know the particulars of your engine, but variations in compression ratio, fuel octane, altitude, piston dome shape, combustion chamber shape, carbon buildup, inlet air temperature, etc. could change this maximum figure. The best way to maximize the advance is to put it on a chassis dynomometer and check for the max. power at various rpm's under load while somehow monitoring threshold detonation. After testing through the rpm range, you can plot a custom distributor advance curve that stays safely below the detonation threshold. As you might imagine, this will take time (& money).
Sherwood Lee
http://members.rennlist.org/911pcars