The experimental metastable zone width (MSZW) was newly assumed to correspond to a point at which the number density of accumulated crystals (grown nuclei) had reached a fixed (but unknown) value and a new mathematical model was proposed. The effect of cooling rate on the experimental MSZW reported in the literature was explained reasonably by the new model. The nucleation order n in the rate equation for primary nucleation was deduced from the slope of a linear plot of log(MSZW) vs. log(cooling rate). This linear relation is similar to the Nyvlt's one but the physical basis is completely different. The induction time was also explained in the same way and related to the MSZW. (C) 2007 Elsevier B.V. All rights reserved.