Can anyone please give a definitive answer on how to do this.

I have always calculated a monthly mean by the sum of the mean min=max / 2 and then round up if necessary if the figure = 0.05, 0.015, 0.25, etc - and I believe that this is the COL way.

But I believe that the Met'O 'rules' for when that sum = 0.05, 0.15, etc is thus:

15.05 change 15.1

15.15 leave at 15.1

15.25 change to 15.3

15.35 leave at 15.3

15.45 change to 15.5

15.55 leave at 15.5

15.65 change to 15.7

15.75 leave at 15.7

15.85 change to 15.9

15.95 leave at 15.9

16.05 change to 16.1

Now i like this way, because the thinking of 'one up, one down' must simply be to compensate for the 'average' 6 occassions a year when you would have a mean of 0.05 - so that by average, 3 go up and 3 come down and eh a balanced figure - and so not all data is simply rounded up..

however, it also ignores the rounding up/down of the mean min/max and that will effect if the final mean is bang on 0.1 or a 0.05 - now I see less sense.

But then So how do you work out seasonal and yearly averages then?

A - the sum of the mean temps divide by 12 and then apply those rules re 0.05, 0.15, etc or

B - the sum of the mean min/max divide by 2 and then apply the above rules?

i should add that all means in COL use A

i will give my yearly figures as an example of how this matters.

2009

using A above my yearly mean would = 8.83c (8.8c)

Using B above my yearly mean would = 8.85c - which then becomes 8.9c

2010

using A above my yearly mean would = 7.39c (7.4c)

Using B above my yearly mean would = 7.35c - which then becomes 7.3c

So it does matter and I know which I prefer (A), but I bet you tell me B

After applying this 'one up, one down' system I then checked out the effect that they had on the years.

first I used A - this changed one year

then I used B - and this changed two of the years but not the year that A changed!!!!

It should not be this complicated.

Any offers?