In a multi branch setting any library might fill an article request, but at the moment only the pick-up library is stored. There is no way to tell which library actually filled the request which is important for statistics.