My question is this....

Please correct me if I'm wrong, or how to get around the issue, but how about a real-life scenario that I have:

I have been given a logical definition of a duplicate row to be a composite of 4 columns. However, there are actually 10 columns in the table. This data has duplicates in the logical composite key, but it's not necessarily a duplicate if you included the other data fields in the "group by". If I want to copy the data to the other "temp" table, I have to include all the columns in the "group by" or I will lose the other data, but in my case, I can't do that or I will miss some duplicates.