Sorry better to use a diiferent set of data  :)

I would like to find the average temperature combining a  number of
locations where the record overlaps.

for a particular month I could use ...
sum(round(avg(decode(to_char(to_date,'mm'), '01',temp, null)), 1))
"Jan" from table where locatios in (a, b, c)

If I had the following data..

    location 1    location 2      location 3
Jan 1999    25.1        26        40       
Jan 2000    25.1        20
Jan 2001    25.3
Jan 2002    25.3

the above script would result in 26.7 because it averages all the
figures avalable

however I would like to average the temperature across all available
locations for each year first and then add the averaged figures

Jan 1999 = 30.4   
Jan 2000 = 22.5   
Jan 2001 = 25.3
Jan 2002 = 25.3

results in 25.9 and this the figure I mant.

Is there a way to decode by the year, find the average of each year
and then average the resulting figures.

I want only the one figure for each month as an average across all
years of record using many locations with varying overlaps.

This would seem to be a simple task but has me stumped!

I also wish to save the average figure as a new_value variable and use
elsewhere down in the script. I am fine with this part.

thanks Dave