OK, I have a data set that has (length $3) char vars x1 to x15; Each obs has a varying population of these vars. That is some have 1 value and the rest nulls , some have 15 , etc.

obs x1 x2 x3 x4 x5 x6 1 a b c d e (want to get a b c d e) 2 d d c (want to get d c ) 3 a b a b (want to get a b)

Each obs is unique and needs to be retained (there is a keyed var called ' ID'). I need to drop all duplicates in the observation array x1-x15 . I also want to sort the file so that the obs will provide a meaningful count (that is all vars in the array will be in the same position so that I don't get something like this: x1 x2 a b b a

because it will give me counts of two unique combinations when there really is only a count of 2 of 'a' and 'b'. I need to concatenate these vars to then get the freqs needed.

Any ideas? Filesize 90K obs.

Thanks Much, Thought I saw something like this back awhile ago. Doug Conrad