The Catalogue Raisonné as Intervention

A newly surfaced painting lies on the table. It clearly belongs to a series that was catalogued and numbered long ago. The known works carry inventory numbers 0184 through 0192. By now, however, the catalogue raisonné has reached number 0671.

What number should the new picture receive?

The temptation is to insert it among the other works. But that would mean changing existing numbers. Image files, loan forms, condition reports, and earlier correspondence would no longer match the records. An art-historically coherent order would create a practical mess.

The problem does not lie in the later-found painting. It lies in the expectation that an inventory number should already express when a work was made and where it belongs.

A catalogue raisonné combines two tasks that are easily confused with one another. It must identify every work unambiguously. At the same time, it should show how the works relate to one another in time, material, and content. The first task needs stability. The second must remain changeable.

The number fixes the object. The remaining information establishes its relationships.

How much must a catalogue raisonné know?

The basis of a catalogue raisonné is an inventory, yet the two are not identical. The inventory records what exists or is known. The catalogue raisonné, beyond that, decides what belongs to the scholarly-treated œuvre, and under what conditions.

Even a work that is not included, or still unresolved, therefore needs an internal record. Otherwise not only does an object disappear from the documentation, but so does the history of its examination.

There is no globally binding standard specifically for catalogues raisonnés. Even the Catalogue Raisonné Scholars Association points out that it has no universally valid guide for new catalogue raisonné projects. That is understandable: a catalogue of paintings requires different information than one of sculptures, photographs, prints, or site-related works. Catalogue Raisonné Scholars Association

Still, established standards exist that a practicable structure can draw on.

The Object ID standard, supported by the International Council of Museums, names nine basic categories for identifying a cultural object: object type, materials and technique, measurements, inscriptions and markings, distinguishing features, title, subject, date, and maker. Photographs and a short description are added to these. International Council of Museums

The Getty Vocabulary Program goes considerably further with the Categories for the Description of Works of Art, or CDWA. CDWA describes around 540 possible categories and subcategories. But it explicitly marks only a small part of these as the indispensable core needed for unambiguous identification and description of a work. Getty: CDWA General Guidelines, Getty: CDWA Core Categories

This distinction is especially helpful for estates. A good catalogue raisonné need not capture every conceivable piece of information from the outset. It must first reliably establish which work is meant, what is known about it, and how certain that knowledge is.

Many databases have been developed for numerous collection types and exceptions. Long input forms therefore give the impression that every record must be filled out completely. The result is empty fields, inconsistent entries, and later restructuring.

With a large holding, the complexity of the system has immediate consequences for the work. If entering even a simple work already demands numerous fields and links, the backlog grows faster than the catalogue. A lean system, by contrast, allows new works to be recorded quickly and consistently. Deeper information can be added later.

Complexity is not the same as accuracy. And the completeness of a single record is not the same as the workability of the entire catalogue.

What must be settled before the first record

Even before the first inventory number, it must be determined what counts as a unit of record within the given œuvre.

Is a diptych one work, or does it consist of two objects? Does a sheet worked on both sides receive one record, or two linked entries? How are casts, editions, variants, collaborative works, and multi-part installations handled? Are sketches recorded individually or as a bundle?

These questions cannot be answered in general terms. But they should be put in writing and applied consistently to comparable works. Otherwise the structure shifts during the process of recording, and records already processed have to be recreated or split apart.

A short data manual can also determine:

• which fields are mandatory,

• how types of title are distinguished,

• how certain, approximate, and presumed dates are written,

• which units of measurement and terms are used,

• how series, parts of works, and variants are linked to one another,

• which information stays internal,

• and who may review or approve a record.

The data manual need not be an extensive set of rules. It should prevent the same information from being recorded differently by different people.

A sensible core

For each work, a core record should be created first. It might contain the following information:

• Immutable internal identification number It links the work, record, images, and documents.

• Processing status For example: created, in progress, or reviewed.

• Catalogue raisonné status For example: candidate, under review, included, deferred, or not included.

• Maker and attribution status Especially important for collaborative works, workshop contexts, or uncertain authorship.

• Work type and, where applicable, work group For example painting, drawing, or sculpture, as well as the relation to a series.

• Title and title status A title given by the artist is different from a transmitted title or one added later as a description.

• Date and degree of certainty Exact year, period, circa date, presumed date, or undated.

• Material, technique, and dimensions

• Signatures, inscriptions, labels, and distinguishing features

• Images, along with image credit and rights status

• Location and, where known, owner This information may be subject to internal access restrictions.

• Sources and evidence Where do the title, date, provenance, or attribution come from?

• Notes, peculiarities, and assumptions

• Processing note Who created, reviewed, or changed the record, and when?

Provenance, exhibitions, literature, restorations, and loans can be added as expandable sections. They should be repeatable, because a work can have multiple owners, exhibitions, and publications. A long list in a single text field is difficult to search or analyze later on.

Equally important is the distinction between "unknown," "not present," "not applicable," and "not yet reviewed." An empty field does not reveal which of these cases is meant.

A catalogue raisonné need not contain an answer to every question. It must make clear which questions have already been addressed.

The notes field plays an important role here. Peculiarities, contradictory information, and provisional assumptions can be recorded there:

Date presumably 1968 to 1970. Based on three comparable, signed works from the same series and an oral statement from March 12, 2025. Not yet conclusively verified.

Observation and interpretation should remain distinguishable here. A visible inscription on the back is an observation. The assumption that it came from the artist is an interpretation. Both may be documented, but they should not receive the same status.

The free-text notes field must also not become a dumping ground for anything hard to classify. Whatever needs to be searched, sorted, or filtered later belongs additionally in a structured field. Getty, too, recommends capturing important information from free notes in the corresponding controlled fields, so that it remains findable. Getty: CDWA General Guidelines

Why numbers should explain nothing

An inventory number should be unambiguous, permanent, and as meaning-poor as possible. It may identify a work, but it should encode neither date nor medium, location, or art-historical classification.

A simple sequence such as

WORK-000184

is more resilient in the long term than a number like

1972-PTG-SERIES-A-04.

Because dates can change. Works change location. A piece is later assigned to a different series. When such information is part of the number, every new finding becomes a technical problem.

Internationally used documentation guides likewise recommend unambiguous numbers as a stable link between object and information. Simple systems are considered more sustainable than complex numbers that encode various pieces of information. Collections Trust: Numbering, Canadian Conservation Institute

The later-found painting therefore does not receive the number 0187a, and it does not force the following works into new numbers either. It gets the next free internal identification number, for instance WORK-000672.

In the "series" field, it is linked to works 0184 through 0192. Through the date and a separate sequence within the series, it can be displayed at the art-historically appropriate position.

Inventory number, series, and chronology remain separate.

This holds true even when a printed catalogue raisonné number is assigned later. The internal identification number should remain in place. A published number can be added as an additional field. Numbers already in use should not be reassigned or silently changed. Earlier designations and numbers remain in the record as cross-references.

A new work may change the order of the œuvre without changing the identity of all works already recorded.

So the work does not have to be done again

The data should not be usable exclusively within one particular piece of software. A complete export in an open, widely used format must be possible at any time. Images, too, should remain linked to the work records through stable file names or media numbers.

Long-term workability includes:

• regular complete exports,

• separate backups of data and images,

• a directory of the fields and terms used,

• documentation of earlier numbers,

• and a traceable change history.

The database is a tool. The catalogue raisonné should be able to outlive it.

Where to begin?

With the largest works? With the earliest? With the supposedly most important work group?

For the starting point, art-historical significance is usually not the best criterion. Anyone who begins exclusively with the major works may develop a system that does not work for sketches, versos, variants, or uncertain pieces. Anyone who wants to begin strictly chronologically stalls as soon as early datings are unresolved.

A small, clearly bounded, and as varied as possible holding makes more sense. This can be a room, a shelf, a folder, or a manageable work group. The first works recorded should include different formats and degrees of difficulty: a well-documented work, an undated piece, a work with a divergent title, a work from a series, and an object of unresolved attribution.

This group makes it possible to test whether the fields are understandable, the numbers work, and uncertainties can be recorded cleanly. Only after that is the system carried over to the entire holding.

Then the rule is: breadth first, depth later. Every work found first receives a short, reliable core record. Provenance, exhibition history, literature, and scholarly commentary can then be added according to priority.

The recording itself can proceed spatially and continuously: shelf by shelf, folder by folder, room by room. This reduces the risk of overlooking works or entering them more than once.

The later-discovered painting still lies on the table. Its new inventory number sits far apart from the numbers of the rest of the series. In the catalogue raisonné, though, it appears directly beside them, because date, work group, and relationships were recorded separately.

So where to begin?

Not with the largest or earliest work. But with a manageable group on which a lean system can be developed — one that enables swift work, and still holds up when, years later, the next unknown picture surfaces.