Last modified: 2011-09-10
Abstract
In 2009, the Muséum National d'Histoire Naturelle initiated an extensive renovation of the Paris Herbarium, a collection of ten million specimens amassed through botanical collecting since the 1600’s. For such a huge collection, this kind of move happens once a century. It includes remodelling the building and collection storage facilities as well as sorting out and reconditioning specimens. Thanks to the two-dimensional nature of those specimens, digitization can be integrated in the process without extra manipulation, and achieved at a marginal cost.
Each specimen is labelled with a barcoded inventory number, allowing software recognition on digital pictures with a link to taxonomic and geographic data. It is also possible to take advantage of the physical ordering of the specimens to alleviate data entry: species name is entered at the folder level, genus entered once for a set of folders, while family and geographic areas are entered for large sets of folders.
The average processing rate is over 10 000 specimens/day for a project duration estimated at less than two years. A dedicated site was necessary, of which setup and operation have been outsourced to an industrial company. As academics and industrials have a different perception of priorities, quality control is a key point to ensure that scientific excellence governs the industrial throughput. Hence, the information workflow requires specific software to provide traceability and failure detection, in addition to the data acquisition itself.
In the middle of the project and after several months under constant time pressure, the operations are under tight supervision and still need frequent adjustments. But its completion is only a first step. Indeed as very rich scientific information is available on the photos (collection date, localization, ecology notes, outdated determinations, collector name, etc.), making it discoverable will be another project. This should be done without manipulating the specimen sheet. A wide range of initiatives should be federated (specialists of taxonomic groups, citizen scientists, countries or regions of origin, historians, optical character recognition) to feed a common repository.