An experimental study of the early stages of crystallization in iPP has shown a qualitative difference between the behavior at low supercooling with that observed with a deep quench. To address previous misgivings in the limits of resolution of crystallites by wide-angle scattering, a new detector has been used that has many orders of magnitude improvements in count rate. At low degrees of undercooling there is a substantial gap between the appearance of a peak in the small-angle scattering, associated with electron density modulations, and the resolution of crystallites. This early growth in electron density has been analyzed in terms of a spinodal decomposition process and the stability limit of isotactic polypropylene determined for three different samples of varying molecular weight. The underlying physics of the early stages of crystallization are discussed and a number of scenarios eliminated; at high temperature Avrami kinetics are not observed whereas at low temperatures the structure in both the small-angle and wide-angle regimes grow contemporaneously following secondary nucleation.