There's some speculation that fermat made the same mistake as lamé, who thought he had a proof. However lamé incorrectly assumed unique factorisation of a general number field, which was an easy mistake to make at the time.