Why does a sample size greater than 30 approximate the normal distribution?

Why does a sample size greater than 30 approximate the normal distribution? Why is n>30 in the central limit theorem? What's the math or reasoning behind it?
Update: what blahb31 wrote is good, but I need more details about just what 'empircal evidence' means. What did these statisticians do to decide on the number 30 that I can write about in an essay for a Stats I class?
1 answer 1