We describe a double-scattering experiment with a novel tagged neutron beam to measure differential cross sections for np back-scattering to better than ±2% absolute precision. The measurement focuses on angles and energies where the cross section magnitude and angle-dependence constrain the charged pion-nucleon coupling constant, but existing data show serious discrepancies among themselves and with energy-dependent partial wave analyses (PWA). The present results are in good accord with the PWA, but deviate systematically from other recent measurements. PACS numbers: 25.40.Dn, 25.10.+s, 28.20.Cz The neutron-proton elastic scattering database at intermediate energies is plagued by experimental inconsistencies and cross section normalization difficulties [1, 2, 3]. These problems have led the most sophisticated partial wave analyses (PWA) of the data [4, 5, 6] to ignore the majority (including the most recent) of measured cross sections, while the literature is filled with heated debates over experimental and theoretical methods [7, 8], including proposed radical "doctoring" (angle-dependent renormalization) to "salvage" allegedly flawed data [9]. Meanwhile, an empirical evaluation of a fundamental parameter of meson-exchange theories of the nuclear force-the charged πNN coupling constant, f 2 c-hangs in the balance [8]. We report here the results of a new experiment, carried out utilizing quite different techniques from earlier measurements in an attempt to resolve the most worrisome experimental discrepancies. The present experiment involves a kinematically complete double-scattering measurement to produce and utilize a "tagged" intermediate-energy neutron beam [10], thus greatly reducing the usual systematic uncertainties associated with the determination of neutron flux. Products from the second scattering were detected over the full angle range of interest simultaneously in a large-acceptance detector array, to eliminate the need for cross-normalization of different regions of the angular distribution. The use of carefully matched solid CH 2 and C targets permitted frequent measurement and accurate subtraction of quasifree scattering background, thereby minimizing reliance on kinematic cuts to isolate the free np-scattering sample. These methods, combined with multiple internal crosschecks built into the data analysis procedures, have allowed us to achieve systematic error levels in the absolute cross section below 2%. In addition to addressing the previous discrepancies, the results provide a useful absolute cross section calibration for intermediate-energy neutron-induced reactions.