The authors propose an approach to optimising schema evolution operation sequences in object databases. The approach separates operations that add structures from those that remove structures so that all additions are performed before any removals. This separation ensures that there is always a state during schema evolution where data can be evolved from structures that are to be deleted to structures that are new or already exist. Their approach also reduces and groups the schema operations to simplify implementation of data evolution functions by developers. They present a case study used as a first evaluation of the approach.