>It is possible. You have to divide the balls into 3 groups of four balls each. And then you should label each group A, B, and C. That gets you started.
>

I thought in this direction, but didn't come to a solution in case, the A and B don't weigh the same.

>I've never heard of Great Ferma theorem. What is it?

http://www.mbay.net/~cgd/flt/fltmain.htm