On Fermat's Numbers