Chris:

I'd agree that mathematicians probably peak somewhere around 40. I can see two things contributing to the subsequent decline: mental degeneration, and an increasingly irrelevant skill set; if your specialty is narrow enough, you may have solved all the easy problems in your microfield. I expect life extension technologies to fix the former problem. For the latter problem, Feynman recommends changing fields every 7 years. Of course, he's Feynman; maybe 10 years is better for us ordinary folks.