It is shown that for diolic perfluoropolyethers having short chain lengths the "anomalous" (non monotonic) relationship between isothermal melt viscosity and molecular weight, reported in recent papers, is not anomalous at all but fully explainable when one takes into account the fact that the glass transition of the samples T-g is itself a function of the molecular weight. For low molecular weight polymers the viscosities have to be compared not at the same T but at the "normalised'' temperature T - T-g.