[en] The contribution of gravitationally bound subsystems to the potential energy of systems of galaxies is investigated. A statistical correction factor to the virial mass is derived using Holmberg's observed frequency function of multiple galaxies. It is of the order of ten for groups of galaxies and negligible for rich clusters. However, it is emphasized that an individual treatment is necessary. As an illustration the contribution to the binding energy by observed binary galaxies in the de Vaucouleurs groups 48 and 49, and the Coma Cluster, is calculated. In all three cases a single double galaxy removes the virial mass paradox. It is concluded that corrections for subclustering are substantial in stability analyses of systems of galaxies, and they can remove easily mass discrepancies of the order of ten in groups and rich clusters, and that they should be made individually and not statistically. (Auth.)