I think he means that if the 14 day interval happens to span the day that daylight-savings-time goes 'on' or 'off', then there will be an error (time difference) of one hour. Hypothetically, if you did this calculation around midnight of the day DST kicked in, you might get the wrong result...