Hello, Lynn won the race! I do not feel mmCIF needs to perpetuate the sequential numbering of atoms through a complete entry. This is a problem for the PDB format. All that is needed is an _atom_site_model_number or the equivalent and then use the same atom_id value for identical atoms in each model. There is a problem with atom nomenclature and relating the constraint data to a specific model. But, that is another issue. Eldon Ulrich