As I look at the page (which I believe changes randomly), I especially appreciate the 0.81 correlation between "letters in winning word of Scripps National Spelling Bee" and "number of people killed by venomous spiders".
It's a good one in two ways. First, it's nicely obvious that these things are unrelated. Second, these are two variables that fluctuate (together) over time -- a lot of the spurious correlations turn out to be two things that basically just increase or decrease more or less constantly over the period in question.
