On Sat, 31 Oct 2009, Kehl Dániel wrote: Use the "sprintf" command -- see below. Just set up a suitable sized matrix and write into it. Revised version: <script> nulldata 10 alepes = 10 bmax = 2 # dummy 'Hozam' series series Hozam = normal() # count the number of rows we need k = 0 loop b=1..bmax loop for (a=-b/2; a<b/2; a+=b/alepes) k++ endloop endloop # initialize statistics matrix matrix stats = zeros(k,2) # do the actual calculations k = 1 loop b=1..bmax j = 1 loop for (a=-b/2; a<b/2; a+=b/alepes) sprintf vname "x_%d_%d", j, b genr @vname = round((Hozam-a)/b)*b+a stats[k,1] = mean(@vname) stats[k,2] = sd(@vname) j++ k++ endloop endloop print stats </script> Allin Cottrell