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

The apparent discrepancy between the Rosenbluth and the polarization transfer method 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 kinematically enhanced in the Rosenbluth method while it is small for the polarization transfer method, at least in the range of (Q^2) where it has been used until now.
Comment: 4 pages, 4 figures. Version accepted for publication in Phys. Rev. Lett