A Catalogue Larger Than One Library

In 1895, Paul Otlet and Henri La Fontaine began an international bibliographic enterprise in Brussels. They wanted records that could connect readers with material beyond the holdings of any single library. The practical equipment was familiar: cards, cabinets, indexing rules, and people doing painstaking descriptive work.

The ambition was much less local. A researcher should be able to discover relevant material without already knowing which institution held it or which publication contained it.

The Mundaneum’s account of the Universal Bibliographic Repertory records an initial base of 400,000 cards prepared by Otlet, Henri La Fontaine, and his sister Léonie. By 1897, the repertory contained 1.5 million cards. Its thematic arrangement used the Universal Decimal Classification developed by Otlet and La Fontaine. Mundaneum, the Universal Bibliographic Repertory

Those numbers are worth keeping because they make the labour visible. This was not merely a prediction about a future information network. People were creating and maintaining descriptions at scale.

Agreement across collections

Dewey’s system had helped arrange a library. Universal Decimal Classification adapted the decimal approach for a broader documentation purpose, with ways to express combinations and relationships between subjects. The UDC Consortium describes a system intended to organize recorded knowledge across disciplines, languages, and media. Its history begins with Otlet and La Fontaine’s work in the 1890s. UDC Consortium, history

A shared arrangement lets records from different sources be considered together. It also creates a new obligation. A category has to remain intelligible outside the room where it was assigned.

That is a familiar professional problem now. Two engineering offices may both classify reports as “ground investigation.” One includes desk studies and fieldwork. The other uses the term only for intrusive site investigation. Joining the databases without examining those definitions creates a collection that appears consistent while answering questions inconsistently.

An ontology practitioner needs to establish whether the categories match, overlap, or differ in scope. A useful connection may say that one category is narrower than another. It need not claim that they are identical.

The old bibliographic ambition is still recognizable. Make material discoverable across institutions, but keep enough meaning for the connection to be useful. A shared code helps only when people understand what it includes.

Léonie La Fontaine belongs in this account as a contributor to the bibliographic work, not as a relative mentioned in passing. The Mundaneum now presents its history through Paul Otlet, Henri, and Léonie La Fontaine. Naming the people who prepared and maintained the records gives a more accurate account than describing a great idea that somehow filled its own drawers. Mundaneum, institutional history

Briet asks what counts as a document

Several decades later, French librarian and documentalist Suzanne Briet widened the question further. Her 1951 Qu’est-ce que la documentation? considered evidence beyond printed texts.

Her famous example concerns an antelope. In her account, an animal encountered in the wild is not automatically a document. Once captured, studied, catalogued, and placed within an organized documentary setting, it can serve as one. Reports, photographs, recordings, and other descriptions then develop around it. Briet is examining the organized activity that makes something available as evidence, not claiming that an animal has become a book. Briet’s 1951 text, republished with editorial notes

The example is memorable because it refuses the easy assumption that a document must be a sheet of writing. It also lets us separate an object from the accounts produced about it.

Think of a material sample taken from a building. There is the sample itself, the act of taking it, the laboratory result, the photograph, the report interpreting the result, and the recommendation someone makes afterward. Calling all six “documents” may be convenient for a storage menu. It is not enough for a system that must explain how a conclusion was reached.

Following one sample

Consider an illustrative concrete sample, C-104. A laboratory tests it and reports a strength value. An engineer later cites that result in an assessment.

The ontology should allow the assessment to refer to the test result, the result to refer to the test, and the test to refer to the particular sample. The sample, in turn, needs its collection location and date. A photograph of C-104 should not be mistaken for another sample merely because it has its own file identifier.

Now suppose the laboratory corrects a transcription error. Which assessments used the earlier value? A folder of PDFs may contain all the evidence, but it may take someone hours to establish the chain manually. Explicit relationships make the question answerable without requiring every later reader to reconstruct the work.

The distinction also prevents a quiet inflation of evidence. Five reports that repeat the same laboratory result are not five independent tests. A model that records their derivation can show that they depend on one observation.

Modern provenance standards provide ways to represent the entities, activities, and people involved in producing a result. They help explain how a record came to exist. They do not certify that the result is true. W3C, PROV overview

Otlet’s shared bibliography and Briet’s documentary evidence address different problems, but together they make the modern task easier to see. Connect what is scattered. Describe what each record actually represents. Preserve the relationship between an observation and the later statements made from it.

There is still a stubborn difficulty inside the catalogue itself. A subject may involve a material, a process, a place, and a period at once. Asking a cataloguer to choose a single branch can hide most of what a reader needs. Ranganathan approached that problem by taking the compound subject apart.