Using inferences from his Large Numbers hypothesis (L. N. h.) Dirac (1979) derived a cosmological model, described by an ‘atomic’ metric, that is not in agreement with the Einstein theory of gravitation. He then showed that the same model described by a different metric appears to be the Einstein–de Sitter (E. S.) universe, and thus to agree with the Einstein theory. According to the first description, relative to electromagnetic forces, gravitation becomes weaker with advancing epoch. So long as the L. N. h. is retained, we point out how this physical feature persists no matter what metric is used. Therefore Dirac’s E. S. model is observably different from that of standard relativistic cosmology. We make this explicit by writing into the calculations a ‘Coulomb’ constant to match the gravitation constant. We show how the way we do so is internally self-consistent. In the course of the work it emerges that one of the two main inferences from the L. N. h. is automatically ensured by the E. S. model, which may call for some re-examination of the status of the hypothesis. We discuss the different cosmological redshifts predicted by Dirac’s E. S. model and by the standard model.