Annealing of single crystals and of single‐crystal cakes of polyethylene leads to an increase in density because the growth of crystal thickness with time reduces the area of the density deficient surface regions. When the same process is applied to the isothermal crystallization and to the annealing of bulk samples, one has to consider the long period distribution of the crystals due to their varying age and thermal history; then one is able to reproduce sufficiently well the experimental data. In particular, starting with inhomogeneous athermal nucleation and uniform three‐dimensional growth (n= 3), one obtains a higher initial value of the Avrami exponent (n′ = 3.7), i.e., a faster density increase as a consequence of the relatively rapid decrease of the surface mass defect contribution to the density of the crystals in the first stage of their thickness growth. The extremely slow reduction of this contribution with growing age of the crystals also explains the long tail of secondary crystallization and annealing curves where the crystallinity, although slowly approaching, never reaches unity. The early drop of crystallinity below the calculated values and particularly the observed transition from primary to secondary crystallization, however, being much gentler than predicted by the theory, have to be explained by additional effects, i.e., mainly by impurity rejections during crystallization by negative pressure relaxation developing in areas completely blocked by growing spherulites and by tied molecules partly included in two different crystals so that the intervening loops from pure geometrical reasons are prevented from further crystallization.