One Author, Several Names
Panizzi and Cutter made the catalogue answer a reader's questions, including when different names refer to the same person.
A person changes their surname, publishes under a pen name, or appears in an old record with an initial instead of a first name. Their work does not cease to belong to them. Yet a catalogue can scatter it into several apparently unrelated careers.
The opposite error is just as easy. Two people share a name, and a system combines their publications, qualifications, or employment records. Anyone who has tried to untangle that mistake knows that a matching string of letters is a poor substitute for an identity.
Librarians had to solve this problem long before customer databases and knowledge graphs. Their work on names offers a particularly clear connection between historical cataloguing and professional ontology practice.
Rules for a reader, not just a clerk
Antonio Panizzi and his colleagues at the British Museum developed the ninety-one cataloguing rules published in 1841. The rules dealt with the practical problem of making entries consistently enough for a large catalogue to be usable. Decisions about names, titles, and the treatment of works could not remain private habits in the head of whichever employee happened to make the entry. Panizzi’s Rules for the Compilation of the Catalogue, 1841 edition record
Charles Ammi Cutter later stated the catalogue’s purposes in unusually accessible terms. His rules asked whether a person could find a book through a known author, title, or subject; discover what the library held by an author or on a subject; and choose a suitable book, including an appropriate edition. The rules first appeared in 1876 and were revised. The 1904 fourth edition preserves those purposes clearly. Cutter’s 1876 publication; Cutter, Rules for a Dictionary Catalog, 1904
Notice where Cutter begins. A person wants to do something. The catalogue is judged by whether that person can do it.
A modern team might call these user requirements or competency questions. The useful connection does not depend on the label. Before designing the data model, ask what someone needs to find, distinguish, or choose.
For a professional register, “Show me everything this engineer has contributed” is one question. “Show me everyone called J. Patel” is another. A model that treats those questions as equivalent will eventually assign someone’s work to the wrong person.
Keeping the name without confusing the person
Libraries use authority control to keep names and other access points consistent. An authority record can connect a preferred form of a name with other forms, supporting information, and references. “Authority” here concerns the controlled record used by the catalogue. It does not mean the person named is an expert on every subject, or that the record is immune from correction. Library of Congress, understanding MARC authority records
Consider a hypothetical author who published as A. M. Rivera and later as Ana Rivera. A catalogue can preserve both forms while linking them to the same identified person. A reader who knows the earlier name still has a route to the work.
That is different from deleting the earlier name and replacing every occurrence with the latest one. A transcription of a historical title page may need to preserve exactly what appeared there. The system can record the printed name and separately identify the person it refers to.
This distinction matters in ordinary organizational work. An invoice might carry the supplier’s trading name at the time it was issued. The supplier register might carry the current legal name. An ontology can connect those records without pretending the invoice used words that it did not use.
There is a further question. Did the company merely change its name, or did a new legal entity acquire its business? The answer cannot be settled by how similar the two names look.
What would justify a match?
An identity decision needs evidence appropriate to the setting. For an author, that might include publication history, dates, affiliations, or an established identifier. For a supplier, registration details may matter. A similarity score can identify candidates, but it cannot decide which distinctions the organization is entitled to discard.
Imagine two records for “A. Rivera.” One belongs to a civil engineer in Madrid. The other belongs to a botanist in Bogotá. Joining them creates a very accomplished person who does not exist.
A more careful workflow keeps the records separate until there is enough evidence to merge them. It records the basis of a confirmed match and leaves a way to reverse an error. The uncertainty should be visible to the person using the result, especially if a later application would use that identity to grant access or attribute responsibility.
This is not fussy housekeeping. Identity determines which facts travel together. A mistaken merge can make every individual fact appear well sourced while assembling them into a false account of one person.
Cutter’s reader-centred purpose remains useful here. A successful system helps someone identify the intended person and choose the relevant work. A long list of near-matches may be necessary, but the system should explain why the candidates differ.
The international cataloguing principles continued this concern. IFLA’s 2009 statement broadened the earlier Paris Principles beyond entry choices for textual resources to bibliographic and authority data more generally. Cataloguing principles were being revised for changing materials and systems, while retaining the need for consistent, usable descriptions. IFLA, International Cataloguing Principles
The name is now attached to an identified person. The book is distinguishable from another edition. A reader can find either without knowing the cataloguer’s preferred spelling.
But finding a known author is only one entrance. What if the reader knows the subject and has no title in mind? Then the catalogue has to make a more contentious decision: which subjects belong together, and who gets to decide.
