We point out the close relationship between Weyl geometry and the manifestly gauge-independent but path-dependent formalism of quantum electrodynamics of Mandelstam. This comparison shows how the usual Weyl geometry and even the recent modifications of this geometry by Dirac cannot be correct.