Probably goes back to the days of mercury max/min thermometers.
Thats true. I asked once was it Norwegian YR about this and the way averages are calculated goes back to the old days, so new and old values are comparable. Some metoffices use something like (8 values per day + max + min / 10).