The unit‐cell dimensions of a given polyethylene have previously been shown to vary nearly linearly with the reciprocal of lamella thickness. Data obtained at 153.2 and 296.2°K are presented to show that the slope of this dependence is different for crystals of orthorhombicn‐paraffins, melt‐crystallized polyethylene, and solution‐crystallized polyethylene. Within the limits of error, all extrapolate to the same basal area at infinite lamella thickness, and this agrees with the measured value for a sample crystallized from the melt under high pressure to yield a long period of about 3500 Å. Since the effect is a surface one, it is proposed that these differences result from the differences between methyl interactions, fold interactions, different fold planes, domains, etc. It is shown that the variation of cell dimension with lamella thickness leads to a quadratic term in the variation of macroscopic density. This term permits the separation of the thickness and density of a lower‐density surface layer in a two‐phase model.