Heegaard Floer theory and property P

This paper has been withdrawn due to an error in the proof. The paper implicitly assumed that every homotopically trivial knot is related to the unknot by negative crossings, which is false (as was pointed out to the author by Peter Ozsvath).