“…Moreover, the corresponding curve passes close to the central values of several data points, although the experimental accuracy (which is of the same order as both the three-and four-loop contributions) does not allow one to make a definite choice between four-and three-loop approximations." Second, a certain duality between higher orders and higher-twist contributions to the Bjorken sum rule (74) has been detected 6 in [87]. Indeed, the fitted value of first non-zero higher twist contribution, µ 4 , has proved to be strongly dependent on the order of PT terms kept in (79).…”