A collisionality scaling experiment associating a confinement analysis, turbulence measurements across the whole plasma and gyrokinetic modelling is reported. In Tore Supra L-mode plasmas, mid-radius dimensionless collisionality ν* has been varied performing a four-points scan from ≃0.1 to ≃0.7. The normalized confinement time exhibits a dependence with respect to collisionality: BτE ∝ ν*−0.3±0.3 which is strongly modified when accounting for the confinement dependence on the normalized Larmor radius, ρ*, and normalized pressure, β, since one obtains BτE ∝ ν*0.0±0.7. This weak dependence is consistent with ITER L mode scaling laws and dedicated experiments elsewhere (Luce 2008 Plasma Phys. Control. Fusion50 043001). The global analysis is confirmed by normalized effective heat transport coefficients which do not vary outside their error bars in a limited radial range of reliability. The analysis is completed by density fluctuation δne/ne measurements across the whole plasmas. For normalized radius r/a < 0.7, δne/ne does not depart from its error bars and the radial wave-vector spectra are not modified. These observations are well reproduced by non-linear gyrokinetic simulations, where, despite high ν* values, no zonal flow damping mechanism is at play. At the plasma edge (r/a > 0.7), the lowest poloidal wave-vector measured by the Doppler reflectometer exhibits a decrease in δne/ne with increasing ν*, while the other turbulence measurements remain unaffected.