Ice cream sale and crime

Whenever I am lecturing on or discussing Waltz's chapter on 'Laws and Theories' (in Theory of International Politics) in an IR theory class or my part of the methodology course, I am fond of repeating an example a Professor of mine mentioned (a long time ago, in a galaxy far, far way) of the correlation between ice cream and violent crime.  The punch line is of course that correlation is not causation and that correlation requires an explanation before pure correlations make any sense.  I didn't realize that this particular example was very popular, nor that it had any basis in any real study.  I used it assuming it was just a social science equivalent of an urban myth.  Apparently, the example is quite popular and even has some empirical support.  Nice story in Slate.