Running mean of Q_b in the Hubble sequence, with and without spiral arms and dark halo corrections. Left panel: mean bar torque parameter obtained using the two disk thickness estimates applied in this study (see Sect. 3.1.2): de Grijs (1998) and Speltincx et al. (2008) (blue and red solid lines, respectively.). The lower and upper limits for the disk thickness are calculated using the de Grijs’ relation, from which the corresponding mean bar torque parameter delimits the uncertainty in the estimate of Q_b (blue shaded area). Central panel: mean torque after the halo correction, following the different approaches described in Sect. 3.6.2: (i) using the universal rotation curve; (ii) following the methodology of B2004; and (iii) adding V_{3.6 μm}+URC halo, and correcting the URC halo amplitude to match the observed velocities. The unrealistic extreme case (iv) of assuming to be constant at all radii is also shown. Corrections (i), (ii), (iii), and (iv) correspond to the light green, orange, red, and dark green solid lines, respectively. The shaded area is computed from all the reliable bar strength measurements presented in this paper, while the remaining moving averages (including that one of the raw Q_b) are computed using only the galaxies with available halo correction. Right panel: moving average of Q_b after the spiral arm correction. Galaxies with prominent bulges are studied separately, showing the role they play in diluting the bar forces. We obtain setting the m> 0 Fourier components to zero (1) beyond the bar length and (2) beyond 1.2 times the bar length in the potential calculation.