You got to love it when correlations can be made, and a thank you to the ICECUBE Blog

If the histograms and data are exactly right, the paper quotes a one-in-ten-thousand (0.0001) chance that this bump is a fluke. That’s pretty small; although bear in mind that lots of distributions like this get plotted. If you plot 100 different distributions, the chances become about one in a hundred (0.01) that you’ll see something odd in one of them. The Tevatron goes bump

http://imgs.xkcd.com/comics/significant.png

Dialogos of Eide