There is an interesting article (open public access!) in the current Proceedings of the National Academy of Sciences USA studying perfect pitch (aka "absolute pitch") - E.A. Athos et al., Proc Natl Acad Sci USA (2007) 104:14795-800.
Among the interesting findings:
- People with perfect pitch scored in a group completely separate from everyone else, rather than just at one end of a bell curve, on tests of pitch perception;
- Perfect pitch declines with age: people in their 50s with perfect pitch weren't quite as perfect (and they consistently identified notes as sharper than they were);
- G# was the most difficult note to identify -- substantially more than any other note;
- "A" (alone) could be substantially sharp or flat, and still be recognized as "A";
- The results suggested to the authors that perfect pitch is probably due to just one or only a few genes;
- Some people with perfect pitch were consistently a half step sharp! (I guess they're only "semi-perfect"... )