In this erratum we want to point out that our interpretation and conclusion concerning the closing of our magnetisation curves is not correct. The fact that in our measurements the hysteresis curves close is due to the specific measuring method used in our Quantum Design MPMS SQUID magnetometer, combined with the small length of uniformity of the superconducting magnet. It is, therefore, not related to the fast relaxation of magnetisation with respect to the long measuring time of the SQUID, as we have argued in our paper. [...]