I have a table that looks something like this:

groupnum          itemnum          quantity
1                  10                  1
2                  10                  1
1                  17                  5
2                  17                  5
1                  23                  2
1                  26                  1
2                  26                  1
1                  27                  3
2                  27                  4
1                  30                  0
2                  30                  0
2                  33                  1
1                  36                  3
2                  36                  5

The table isn't really sorted, but we order it this way in queries. As you can see, most itemnums are found in groups 1 and 2, and usually have the same quantity. The user wants a query that returns those itemnums found in both groups, but with different quantities. In the above example, such a query would return:

1                  27                  3
2                  27                  4
1                  36                  3
2                  36                  5

There are probably a few ways to do this, involving sequences of queries, but I wonder what would be the most efficient.