When integrating record sources (e.g., EDRM systems, web archiving software, ad-hoc submissions etc.) to a digital archiving system, it is common to have to translate from a variety of source schemas into an archival metadata schema. This can lead to some loss of information owing to imprecise mappings. In addition, both the source and archival schema are subject to change over time, meaning that maintaining the necessary transforms is also costly. It is possible to try to reduce the problem by mandating some standards but, given the long timescales for archiving, this can only reduce but not eliminate these problems. Inevitably, the archival system either still needs to cope with heterogeneous metadata schemas or needs to perform even more (potentially lossy) metadata transforms. In this paper we present an alternative approach that has simplified the integration between source systems and the long-term archive by allowing each type of record to use the original descriptive schema used by its source (i.e., use metadata fields appropriate for that type). Appropriate technologies are used to ensure that it is still possible to view, edit (where appropriate) and search within each metadata field. The archive's internal metadata schema thus need only be responsible for maintaining structural, technical and preservation metadata. Hence, the archive allows all the functionality that would be expected if a fixed descriptive schema were used but without the need for complex mappings.