It is shown that there is nothing paradoxical in Kovacs’ well-known tau(eff) data. At small deviations from equilibrium (delta < a few times 10(-4)), the tau(eff) values are inaccurate, should be rejected, and do not allow any conclusion about the behaviour of tau(eff) for delta --> 0. Thus, there has never been any physical evidence for a ’paradox’ or an ’expansion gap’ at equilibrium. The reliable part of the data (delta > a few times 10(-4)) can be described, within experimental error, by the phenomenological volume-recovery theory. A dependence of tau(eff) on the initial temperature (at constant delta) is a normal feature of linear and nonlinear systems with wide distributions of relaxation times. The dependence may even persist up to equilibrium; however, tau(eff) if then necessarily continues to increase (to infinity) with decreasing delta instead of approaching a finite limit as suggested by Kovacs’ data.