The existence of non-diffusive momentum transport can be considered as firmly established from transient transport analysis in several devices [1-4]. In NBI driven JET H-modes, the existence of a pinch has been established both using NBI modulation techniques [2] and using database analysis [3,4], with largely consistent results. Comparisons of the observations with linear gyrokinetic calculations using the GKW code [5,6] have shown similar parameter scalings with R/Ln, q and e = r/R, allowing the pinch to be identified as the theoretically predicted Coriolis pinch [6], but predicted pinch magnitudes are on the whole lower than observed. Several effects may be responsible for this difference: 1) systematic errors on the current profile reconstruction introducing a bias in the gyrokinetic modeling, 2) a difference between Carbon rotation (measured) and bulk ion rotation (predicted), 3) the presence of residual stresses (RS), 4) simplifying assumptions used in the gyrokinetic modeling in [3,4], namely the assumption of electrostatic fluctuations, a circular geometry and the simple quasilinear estimate for the ratio of heat and momentum fluxes. In order to address these points, the present work introduces several refinements into a subset of the data presented in [3,4], which are all from the JET Carbon wall phase (before 2010). It uses EFIT equilibria constrained by polarimetry and allowing for finite edge pressure and current. Torque and heat deposition profiles are calculated using the ion orbit code ASCOT [7] for all samples in the subset, thereby taking into account any classical fast ion transport. All gyrokinetic simulations are repeated with the improved magnetic equilibrium and electromagnetic perturbations. Last, but not least, the difference in neoclassical rotation velocity between the measured impurity (C6+) velocity and that of the main ion species (D+) is calculated using the NEOART code [8].