[pspencer]

I always liked this visual representation of the central limit theorem: http://blog.vctr.me/posts/central-limit-theorem.html. There is a faster one here (I think written in R): http://vis.supstat.com/2013/04/bean-machine/

These are computer simulations of Galton boxes: http://en.wikipedia.org/wiki/Bean_machine

[stdbrouw]

As the comments to that page mention, though, it's not a representation of the central limit theorem at all. It shows how a binomial distribution "becomes" a normal distribution in the long run, but it doesn't show why this would apply to the distribution of errors of a series of sample means.

[andy_wrote]

I have fond childhood memories of the "Mathematica" exhibit at the Museum of Science in Boston, where there was a large physical copy of one of these machines.

Also, "quincunx" is a great word.