Re #7, the divisors of a squere-free integer.
What's wrong with the following proof? By induction on k > 0, starting with the obvious case of a singleton, check that the family of all subsets of a finite k-set can be partitioned into subset-related pairs (by keeping the old arrangement + fattening the old arrangement by the new point everywhere).
