The pair correlation function g(r) is one means of evaluating the atomic spatial arrangements of coals. With coalification expectations are of limited growth in aromatic stacking and a slight growth in the aromatic cluster size over much of the rank range. In this investigation, the pair correlation function was evaluated with molecular models of the aromatic and aliphatic portions for several Argonne Premium coals (Beulah-Zap, Illinois No. 6, Upper Freeport, Pocahontas No. 3, and non-Argonne anthracite Hon Gai) generated directly from high-resolution transmission electron microscope (HRTEM) lattice fringe images. The Fringe 3D approach populates aromatic moieties matching the distributions of fringe: length, layers per stack, interlayer spacing, and orientations to produce a slice model of limited depth but reproducing the lattice fringe micrograph height and width. In this manner, hundreds of fringes can be duplicated rapidly and with greater accuracy than existing model construction approaches. The slice models, devoid of heteroatoms or cross-links, were utilized to predict the pair correlation function and to evaluate the fine detail of the frequency spectra. Agreement between simulated and experimental g(r) functions validated the model construction approach in capturing structural features including aromatic stacking and interlayer spacing. The intensity of simulated g(r) functions increases with increasing rank. The results were consistent with the expected slight increase in the stacking number observed from low- to high-rank coals with a more dramatic transition for anthracite.