C. Berger passed along a link to me today from a talk she went to. Check out colorbrewer.org. It looks useful for picking out color schemes for displaying data, or more fun things like presentation templates, webpages, etc.
You input the “number of classes” of data (read, number of colors in your scheme), what type of data you have — sequential, diverging, or qualitative (not sure exactly what these man) — and then choose from one of 9 different color schemes. The colors you’ve selected are displayed on a map so you get an example, and of course, they give you the rgb, cmyk, hex, etc. values of the colors. It also tells you some other important qualities of the color scheme chosen such as whether it is friendly to color blind people or will translate well if printed as black and white (useful for all of us publishing color online and black and white print figures).
Lunchtime! 
That’s a fantastic idea for a website. Bookmark’d!
Very nice, but from the title of the post I thought the website would help me dress myself!
I remember attending a talk about how it had been proven that you need five distinct colors to color countries on a map so that the same color didn’t touch anywhere. On a torus, you only needed four. It was a pretty cool talk, in part because they made a big deal about how there had been an early proof that was found to be wrong after 12 years. The speaker claimed that it was extremely unusual for an incorrect proof to last that long.
Are you saying that on a flat map I need 5 colors and on a Torus I only need 4? I think this must be backwards. Since there is a mapping from the flat map to the Torus that does not destroy any borders, but that’s not necessarily true the other way around. So, if I’m given a flat map, I can roll it into a Torus, color it, and unroll it and since the unrolling doesn’t create any new borders, I will have the flat map with no same colors touching. I can imagine the Torus requiring more colors than a flat map, but not less.
You’re correct Brad… according to Wikipedia you only need four for a plane and seven (!) for a torus. Either the speaker or, much more likely, my memory was off.
The Four Color Theorem