Two million specimens and one search box

The Weatherby has been catalogued continuously since 1904, which means it has been catalogued in four different systems by people with four different ideas of what a record is for. The one thing every record has is a determination — the accepted scientific name — and that is exactly the thing a member of the public does not have.

The trap here is obvious in hindsight and was not obvious at the start. Everyone involved, ourselves included, began by scoping a recataloguing project. Two million records, a controlled vocabulary, a five-year programme, a funding application. The keeper was right to be sceptical.

What we built instead sits beside the catalogue and never writes to it. It adds a second index — common names, colloquial names, the county a thing was found in, the century, the collector, and a coarse “what does it look like” facet that curators initially found offensive and now use themselves. A search that fails against the determination falls through to that index rather than returning nothing.

The coverage is honest about itself. The site says how much of the collection a given facet reaches, because a search interface that silently covers a fifth of the holdings while looking like it covers all of them is worse than no search at all.