New precision measurements of cross sections and polarization asymmetries in the p͑ g, gp͒ reaction at 90 ± c.m. are presented for incident energies between 213 and 333 MeV. A long-standing problem with earlier experiments that appeared to violate unitarity at the peak of the D is resolved. Data are compared to theories based on baryon resonance structure and to dispersion relations. Recent calculations using the proton polarizabilities are closest to the data, although inconsistencies are observed near the D resonance. PACS numbers: 13.60.Fz, 13.88.+e, 14.20.Gk, 25.20.Dc Elastic (Compton) scattering of intermediate energy photons from the proton is a potentially rich source of structure information. It is sensitive to the proton's electric (a) and magnetic (b) polarizabilities [1], to the deformation of the nucleon through the electric gND coupling [2], and even to the sign of the p 0 decay constant, F p [3,4]. There are many published calculations for Compton scattering. These can be grouped into two general categories, those based on the baryon resonance spectrum or its underlying quark structure [5][6][7], and those relying upon unitarity and dispersion relations to phenomenologically describe elastic scattering in terms of photopion production [3,4,8-10]. These calculations give significantly different predictions, particularly in the region of the D resonance.A number of measurements of proton Compton scattering have been reported [1,[11][12][13][14][15][16], and several authors of dispersion calculations have pointed out a significant inconsistency between many of these experiments and p photoproduction data near the peak of the D [3,8-10]. Compton scattering can be described by six independent amplitudes. Their imaginary parts can be calculated from (g, p) multipoles using s-and u-channel unitarity, and dispersion integrals can be written for their real parts. Four of these integrals converge rapidly with energy. However, the remaining two, those involving a photon helicity flip, do not converge rapidly, making subtractions essential. One of these is dominated by t-channel p 0 exchange, the Low amplitude [17], and can be readily evaluated in terms of the p 0 lifetime. However, the other contains contributions from multiple meson exchange in the t channel that are quite poorly determined. In principle, the beam-polarization asymmetry constrains this other amplitude [4]. But prior to our new measurements, only a single datum with large errors had been published for this observable [18]. Alternatively, sum rules can be used to write the subtraction function for this spin-flip amplitude in terms of the difference of the proton polarizabilities, a 2 b, which can then be fixed by fitting a perturbative expansion of the cross section to data below the p threshold [3,4]. Although this provides a good description of scattering below the D [1,19], the peak cross sections appear to be overestimated [3,4].A lower unitarity bound on the Compton cross sections, which avoids the uncertainties of the dispersion calcul...
