How to reconcile the Rosenbluth and the polarization transfer methods in the measurement of the proton form factors

The apparent discrepancy between the Rosenbluth and the polarization transfer methods for the ratio of the electric to magnetic proton form factors can be explained by a two-photon exchange correction which does not destroy the linearity of the Rosenbluth plot. Though intrinsically small, of the order of a few percent of the cross section, this correction is accidentally amplified in the case of the Rosenbluth method.