>>> You're welcome. Note that Janis' solution and mine will report different
>>> values if one manufacturer uses the "single" enclosure type twice
>>> (Janis' will count it as 1, mine will count it as 2). I'm not sure which
>>> is correct but it should be fairly easy to tweak either solution to do
>>> the opposite thing.

>>>        Ed.

>> Aaah,

>> although I did a quick test with a test sample I didn't notice that Janis'
>> version count them as 1. Now I checked that with the real data and yes, there
>> is the difference between yours and Janis' script.

> I'm still not sure whether the small modification below is what you want...

>    awk '
>      BEGIN {FS=";"}
>      NR==1 {next}
>      !serv[$1]++ {s++}
>      !encl[$2]++ || $2=="single" {e++}
>      !manu[$2,$3]++ || $2=="single" { m[$3]++ }
>      END { print s, e
>            for (i in m) print i, m[i]
>      }
>    '

> If not, it would be helpful if you'd provide some of those real data to see.

> Janis