There are lots of probable reasons, including that it is approximately as big as it can be for a double integer while limiting the risk of rounding errors.