I guess the worst problem with the DATABASE_PDB_MATRIX category is that it fails to respect the very elegant category structure of mmCIF, which is a powerful tool for searches. The experimenter's PDB submission coordinate frame (ORIGX), the PDB reported transform from orthogonal to fractional (SCALE), the non-crystallogtaphic symmetries (MTRIX), the translations linking domains of a polysacchatides (TVECT) are not the sorts of things upon which one might search a database for the fact that it was reported once upon a time in the PDB, but because each of them helps in understanding the reported structure, and more usefully belong linked to other structural information using existing categories, with, perhaps, appropriate subcategories. The comment about consistency with "older" PDB reported entries is irrelevant. If an experimenter feels such information is useful in explaining a structure, it should be a matter for his scientific judgment to report it. If he does not feel it is useful, he should not report it. What he needs in the dictionary is what he can find in the rest of the dictionary: a scientifically sound, non-judgmental, clear exposition of what information is conveyed by items in the category. -- H. J. Bernstein P.S. The submitted coordinate frame may seem like a minor issue, and often it, but sometimes can provide useful information about what happened to make some coordinate directions behave differently than others.