We have at least three levels of alternate configurations of models with which to deal: atomic level alernate sites, microheterogeneity at the sequence level, and full alternate models at a structural entity level. There may be more levels, but this provides enough meat to make some of the problems clear. A single structure may have all these levels. The current dictionary handles the lowest level very nicely with flags for the alternate sites and declarations of ensembles to show correlated behavior. I would suggest that there is a certain simplicity in simply extending the same approach to the sequence and entity level. The problem that arises is then simply a matter of how to apply these flags to the lists of sites or monomers or entities. Again, we may look to the current dictionary for guidance. The alternate position flags are placed within the rows of atomic site lists, and certainly we should permit more columns of alternate sequence and alternate entity flags to be placed there, which is what would actually happen in the innards of an efficient database, but there is also a representational efficiency to using the indirect approach used in declaration of ensembles and, say, sheets, of having an external declaration which creates groups by sets or ranges of identifiers within the site list or monomer list or entity list. This may require a little fiddle in a database load to properly untangle and distribute the information to the proper rows, but it can be a great courtesy to human readers of a data set, who need a more confined context for comprehension than does a computer. -- H. J. Bernstein