Odeuropa: Tracing the History of Smell Through Texts, Paintings and Museums

“Scent is part of our cultural heritage.”
— Inger Leemans, Odeuropa project leader

What smells surrounded people in the past? A small scented container held by the sitter in a portrait. A record complaining about the stench of a canal. Hints of smoke and decay conjured by a sermon. Even when a smell itself has disappeared, texts and paintings retain traces of the people who perceived or imagined it.

Odeuropa was a European research project devoted to finding and interpreting those traces and communicating them as cultural heritage. It connected the use of AI to examine old books and paintings with perfumers’ work on scents for exhibitions. Historians, art historians, heritage scientists and museum curators provided the links between them. The aim was not to decorate the past with fragrance, but to use another sense to consider what people felt about smells and what meanings they gave them.

Contents

Odeuropa at a Glance
From Sensory History to Research Across Borders and Disciplines
1. An Encounter with an Art Historian Opens Up the History of Smell
2. Extracting Information, Interpreting History and Reaching Visitors
Where Does AI Find Smell in Texts and Paintings?
1. Reading Acts of Smelling, Not Just Flower Names
2. Finding Objects and Nose-Pinching Gestures in Paintings
At Museum Ulm, a Portrait’s Accessory Acquires a Scent
Creating a Canal Smell and Analyzing a Surviving Object
Tools and Outcomes Available After the Project
1. Three Entry Points: Searching, Reading and Exhibition-Making
2. A Scent Collection and Conversations in Osaka
Using Smell as a Clue to History
References

Odeuropa at a Glance

Full name: Odeuropa: Negotiating Olfactory and Sensory Experiences in Cultural Heritage Practice and Research
Funded research period: January 1, 2021–December 31, 2023
Project leader: Inger Leemans
Coordinator: Humanities Cluster of the Royal Netherlands Academy of Arts and Sciences (KNAW)
Funding: The EU’s Horizon 2020 program, grant agreement 101004469
Official website: Odeuropa

Thirty-five researchers participated from six countries: the Netherlands, France, Germany, Italy, Slovenia and the United Kingdom. The research focused on European olfactory culture from approximately 1600 to 1920. The funded project has ended; its search tools, encyclopedia and museum guides remain as publicly accessible outcomes.

From Sensory History to Research Across Borders and Disciplines

An Encounter with an Art Historian Opens Up the History of Smell

Leemans became a professor at VU Amsterdam in 2010 and established ACCESS with colleagues to study emotions and the senses across disciplines.

Art historian Caro Verbeek approached her with an interest in studying the relationship between scent and art. Verbeek’s research into the Futurists’ use of scent helped broaden interest in the history of smell.

Odeuropa was publicly launched in 2020, with its formal, EU-funded three-year research period beginning in 2021. Libraries, archives and museums across Europe hold vast quantities of digitized texts and images. Yet search systems organized around titles and authors make it difficult to trace the olfactory experiences hidden across them. Odeuropa developed ways to search accumulated digital collections through smell.

Extracting Information, Interpreting History and Reaching Visitors

The research involved three main tasks. First, computer vision and natural language processing were used to identify smell-related information in images and texts. Next, that information was connected in a common format and interpreted in its historical context. Finally, the findings were brought into exhibitions and walks, and their reception by visitors was studied.

This division of work matters. Detecting a scent-related object in a picture does not explain why it was painted. Even when a text mentions a stench, its interpretation depends on who used the term and for what purpose. Historical research asks questions of the clues found by technology, while museums consider how to communicate them to people.

Where Does AI Find Smell in Texts and Paintings?

Reading Acts of Smelling, Not Just Flower Names

Text processing focused on explicit references to smell. Instead of gathering every occurrence of the word “rose,” it extracted sources and qualities of smells from the context around words such as “smell,” “sniff” and “stink.” This distinguishes a rose listed in a vocabulary from a rose whose sweetness someone actually smelled.

Human-annotated historical examples provided training material for identifying who smelled what, where, how it was described and what emotions were associated with it. The multilingual research benchmark included medical, legal and literary texts in seven languages: English, Italian, French, German, Dutch, Latin and Slovenian.

By contrast, the public description of Smell Explorer specifies material in six languages, excluding Latin. The scope of a research dataset is not identical to that of the material included in a public search service. Likewise, the benchmark’s dates of 1620–1925 should not be conflated with the 1600–1920 period described for the public outputs.

Finding Objects and Nose-Pinching Gestures in Paintings

In images, the system detects smell-related objects such as flowers and perfume bottles, places associated with smells, and gestures such as sniffing or pinching the nose. People mark relevant areas of images to create training examples, which machine learning then uses to find similar elements.

Object recognition in historical paintings presents particular difficulties, however. Depiction changes with period and style, and objects of the same kind can look very different. The researchers’ ODOR image-recognition challenge also highlighted the problem of objects for which few training examples exist. Correctly distinguishing details within paintings is itself a research task.

The information extracted from texts and images is organized into the European Olfactory Knowledge Graph. A knowledge graph is a database that connects things and events with their relationships. Smell Explorer provides an entry point for searching it, allowing users to follow historical sources by the origin of a smell, location or period. The information concerns smells recorded in images and words. The AI is not measuring scent molecules from a screen or directly producing perfume formulas.

At Museum Ulm, a Portrait’s Accessory Acquires a Scent

A Pomander Connects Protection Against Illness with Prayer

In April 2022, Museum Ulm in Germany launched scented guided tours developed with Odeuropa and fragrance company IFF. The aim was to connect clues to smells depicted in artworks with scents visitors could actually experience.

In Martin Schaffner’s 1516 portrait of the councillor Eitel Besserer, a small openwork container hangs from a rosary. This is a pomander, an ornament containing aromatic materials. At the time, bad smells were thought to harm health, while good smells were believed to protect the body. The accessory in the portrait brings together precautions against illness and habits of prayer.

IFF created a scent based on typical sixteenth- and seventeenth-century pomander formulas. It combined spices, herbs and rose with elements of ambergris and civet. The contents of the painted container cannot simply be analyzed. Nevertheless, smelling a formula recorded in historical sources shifts attention from the ornament’s richness to the significance of carrying scent close to the body.

The Smell of Hell Interprets a Historical Imagination

The same tour accompanied Schaffner’s painting of Christ’s descent into the underworld with a scent of hell. Its starting points were the smoke depicted in the painting and religious associations between hell, smoke, sulfur and decay. IFF combined a smoky impression with elements suggesting decomposition and feces through indole and skatole.

This was not an attempt to collect air from a hell that once existed. It was an interpretation in scent of fears imagined by people of the time, drawing on paintings and historical descriptions. Heritage scientist Cecilia Bembibre described the work as follows.

“We developed a smell to evoke a conceptual smellscape …” Bembibre explains that researchers and perfumers created it together.
— Cecilia Bembibre; short excerpt followed by a paraphrase

Perfumers received a historically informed brief, and prototypes were evaluated by historians, art historians and specialists familiar with fragrant materials. Deciding how strongly to convey an unpleasant smell is also an exhibition decision. A balance is needed between allowing visitors to think about fear and discomfort and making an experience too difficult to continue.

Refining the Stench of Leather and the Scent of Gloves

Curator and researcher Lizzie Marx identified a mismatch between the language of historical research and that of perfumery.

“the scent of leather in perfumery is not akin to historical tanned leather.”
— Lizzie Marx; excerpt

For the gloves depicted in the portrait of Helena Schermar, the aim was to convey the relationship between the unpleasant smell of tanned leather and the perfumes used to mask it. IFF refined prototypes in response to the museum and researchers; three versions were needed before they were satisfied. The tour used handheld dry diffusers so that visitors could control how much scent they experienced. Adding smell to an exhibition means designing the entire process, from creation to presentation.

Creating a Canal Smell and Analyzing a Surviving Object

For the canals of seventeenth-century Amsterdam, complaints preserved in notarial records and paintings showing public privies helped establish a brief that perfumers could work with.

Leemans emphasizes that this does not prove the canals smelled exactly that way in the seventeenth century. A recreated scent provides a starting point for considering what smells existed and how they affected people.

“the starting point for thinking about what that smell might have been”
— Inger Leemans; excerpt

For an object that still exists, by contrast, methods such as headspace analysis can collect volatile compounds released into the surrounding air. Odeuropa cites the interior smell of Royal Car P5B, a historic vehicle used by Queen Elizabeth II, as an example of reconstruction based on analytical data.

The Heritage Smell Library distinguishes this kind of analytical reconstruction from recreations based on old formulas or historical research. Alongside samples, it records how a scent was made and what it means culturally. Physical samples are stored at the Osmothèque. The aim is not only to preserve scents but to allow later users to trace the evidence behind their interpretation.

Tools and Outcomes Available After the Project

Three Entry Points: Searching, Reading and Exhibition-Making

Smell Explorer provides an entry point for searching smells in historical sources. Its public description reports processing approximately 43,000 images and 167,000 texts. A 2025 account of the project’s outcomes reported around 2.5 million smell references in the knowledge graph. That does not mean 2.5 million different smells were recreated; it indicates the scale of descriptions and depictions gathered from texts and images.

The Encyclopedia of Smell History and Heritage draws on those sources and specialist research to present readable accounts of olfactory history. Andrew Kettler’s entry on sulfur, for example, connects associated smells with ideas of hell, volcanoes, hot springs and industry. The odor of rotten eggs relates to compounds such as hydrogen sulfide, but the meanings people attach to it change with religious, health-related and industrial contexts. The entry invites readers to consider chemical properties alongside cultural meaning.

The Olfactory Storytelling Toolkit is a practical guide for museums and heritage institutions using scent. It addresses finding smell stories within collections, creating historical scents with perfumers, selecting presentation methods and assessing risks, with eleven practical resources including worksheets. This work by Sofia Collette Ehrich and her collaborators turns experience gained from individual exhibitions into methods other institutions can reconsider for themselves.

A Scent Collection and Conversations in Osaka

The Historical Scent Collection created with IFF contains twelve scents, including a pomander, canals, hell and perfumed gloves. It is a limited collection of samples for sharing research-based smells. Two inclusions, Beuning Room and Waterloo, were developed in 2015, before Odeuropa’s funded research period, by IFF and the Rijksmuseum as part of Verbeek’s research. Treating every scent as a new Odeuropa creation would obscure the research that preceded it.

In April 2025, Leemans, Bembibre and Raphaël Troncy visited Expo 2025 Osaka, Kansai, and presented the findings at the EU Pavilion. Crowded conditions prompted a change from a smell-walk experience to sessions in which they showed, explained and offered scents to visitors. Conversations brought up smells rooted in Japanese everyday life: parks, flowers, the seaside, incense-burning rituals and tatami. Presenting historical scents also became an opportunity to listen to olfactory knowledge from another place.

That year, Odeuropa received a European Heritage Awards / Europa Nostra Awards honor in the Research category. Although its funded research concluded in 2023, the public data, explanations and exhibition methods continue to inform later research and education.

Using Smell as a Clue to History

Odeuropa’s work suggests that approaching the smells of the past requires questions as well as aromatic materials. What smelled? Who perceived it that way? Why was a smell considered protective, frightening or indicative of status?

These questions change how we read the pomander in a portrait, a complaint about a canal or smoke around a devil. Scent enriches visual exhibitions while also giving us a reason to return to historical evidence. A single inhalation does not settle what the past was like. Rather, the sensation prompts us to reconsider the lives and values of people who lived before us.