Anyone dealing with arrays containing a billion elements or more really ought to be using 64 bit arithmetic to avoid problems like this. Certainly better to do this the right way though.