The archive class is used to preserve the edit transactions that occur on the enabled dataset or object class. Therefore, the schema of the archive class must be consistent with the dataset or object class.
Schema changes made to the dataset or object class are automatically pushed to the archive class. For example, if you add an attribute column to the feature class, that column is automatically added to the archive class. If you delete an attribute column, that column will also be removed from the archive class, removing all the archived information for this column.
Geodatabase archiving does not create metadata regarding schema changes. The objective of geodatabase archiving is to preserve the edit transactions of enabled classes.
Changes to geodatabase behavior will be available to the archive class, but the actual change to the ArcSDE repository is not archived. Documenting geodatabase behavior change is the responsibility of your workflow procedures.
Do not alter the schema of the archive class directly. Should you directly alter the archive class, you will potentially corrupt this class and lose your archived transactions. ArcGIS limits the access to the archive class to read-only, preventing direct editing of and schema changes to the archive class.