The Reading Room Grows

The National Library of Luxembourg (BNL) just added 2'700 digitised titles to eluxemburgensia.lu, its portal for digitised collections.[1] The collection includes books, newspapers, posters, and manuscripts published in Luxembourg or connected to the country. It is a significant expansion of what was already one of the more ambitious small-country digitisation programmes in Europe.

There is a catch, though. Most of the titles are still commercially available, which means they are protected by copyright. Under Luxembourgish law, the library can digitise them and make them available, but only within its own walls. You can search the catalogue from anywhere. You can browse the metadata. But to actually read the digitised content, you have to sit in the reading room at the BNL and use a dedicated terminal.[1]

OCR changes the game

The more interesting part of this announcement is the introduction of optical character recognition, or OCR. OCR converts scanned images of printed pages into machine-readable text. This means you can search the full text of a digitised newspaper from 1923 for the word "railway" and find every page where it appears, not just the pages where the cataloguer happened to mention railways in the title.[1]

For researchers, this is the difference between a digital library and a digital archive. A library lets you find what you know you are looking for. An archive lets you find what you did not know was there. Full-text search turns a collection of scanned images into something you can actually mine, cross-reference, and explore. It is the feature that makes digitisation worthwhile rather than just a preservation exercise.

The copyright wall

The restriction to on-site consultation is not a technical limitation. It is a legal one. The BNL worked with the Luxembourg Organization for Reproduction Rights (Luxorr) and the Federation of Luxembourg Editors to make this happen, and the deal is: digitise, index, search, but only in the building.[1]

This is the same compromise that national libraries across Europe have reached. The British Library, the Bibliothèque nationale de France, the Koninklijke Bibliotheek in the Netherlands, all have similar reading-room-only access for in-copyright material. The pattern is consistent: public institutions pay for the digitisation, publishers retain commercial rights, and the public gets access through a keyhole.

It is better than nothing. It is significantly better than nothing. But it also means that the full value of OCR and full-text search is locked behind a physical door. A researcher in Tokyo who wants to search Luxembourgish newspapers from the 19th century cannot do so from their desk. They have to fly to Kirchberg.

What is actually available online

The portal itself, eluxemburgensia.lu, is accessible from anywhere. You can search it, browse document types, and use the advanced search filters. The metadata is open. What you cannot do is read the full text of in-copyright material remotely. Out-of-copyright material is a different story, and the BNL has been making older newspapers freely available online for years through the same portal.[1]

The modernisation of the portal also brought improved search features. Simple keyword search works across the entire collection. Advanced search lets you filter by document type, date range, and other fields. For a small country with a multilingual publishing history, this kind of infrastructure matters. Luxembourgish publications are scattered across three or four languages, multiple alphabets, and centuries of political changes. A unified search portal with OCR is the only way to make sense of it.

Why this matters

Luxembourg has a peculiar documentary heritage. It is small, but it is transnational by nature. Books published here were often written in French, German, or Luxembourgish, sometimes printed abroad, sometimes printed here by foreign publishers. The BNL calls it a "transnational documentary heritage," which is an accurate description.[1]

Digitising 2'700 titles does not sound like much when Google Books has scanned tens of millions. But for a country of 660'000 people with a publishing output measured in the low hundreds per year, 2'700 titles represents a substantial slice of the national literature. And OCR makes each one of those titles individually searchable in a way that a physical archive, however well organised, cannot match.

The reading room is still open. It just grew by 2'700 titles.