While you're right that Perkins' discrepancy is somewhat larger than one might expect, the stats that you choose to back up that assertion are misplaced I think. We're talking about the difference in...