The scent of the past: Classifying works of art by their smell

Original article:

Do you know that unmistakable smell of new (or old) books? Is the aroma of coffee an essential part of your everyday wake-up routine? Have you ever had food poisoning so bad that the very smell of that food will still make you sick again, even years later?

Most likely, your answer to at least some of these questions is yes, and that’s hardly surprising. The sense of olfaction is powerful, it can trigger deep and instinctual memories, and it is inextricably related to the very way we experience the world.

As it turns out, odors can do more than make pleasant (or bad) memories: they can teach us something about the past. In particular, they can teach us about the past of works of art.

In a 2018 communication to Angewandte Chemie International Edition, Curran et al. demonstrated that volatile organic compounds (VOCs) emitted by artefacts can provide valuable information about chemical degradation phenomena and help inform conservation interventions.

VOCs are a broad class of chemical compounds that in fact do not have much in common. Except, you guessed it, their  volatility (i.e. they evaporate easily) and their organic nature (i.e. they contain mostly carbon atoms and few other elements, such as hydrogen, oxygen, and nitrogen.)

Though many of us know them as air pollutants, not all VOCs are bad guys. For example, VOCs are responsible for the delicious smell of frying bacon and for the fragrance of many flowers (and for the less than pleasant stench of wet dogs, alas.)

The researchers that conducted this project were able to assess the degree of degradation of polymeric artefacts in modern art collections based on their emitted VOCs. They prepared specimens of 6 kinds of polymeric materials, representative of materials found in museum collections and sampled from objects of common use (such as a comb, a doll, and a cigarette case). They then “cooked” the samples in a chamber at 80°C (176 °F) and 65% relative humidity for 2, 4, 6, 8 and 10 weeks, in order to simulate a longer aging process in a relatively short length of time. During the induced degradation, the samples emitted VOCs, which the researchers analyzed using solid-phase microextraction gas chromatography-mass spectrometry (SPME-GC/MS). In this technique, gas chromatography is used to separate all the VOCs into individual ones, which are then identified using mass spectrometry. Mass spectrometry is a superstar technique used very often in analytical chemistry, as it allows us to detect the mass and structure of chemical compounds by breaking them down into smaller fragments and ionizing them (usually by removing an electron and turning the molecular fragments into positively charged ions.) Ionized fragments are then separated based on their mass, and the relative abundances of different fragments are used to identify the initial molecule (Fig. 1).

Figure 1. Mass spectrometry infographic. Reproduced with permission from Compound Interest. Original link:

The reported mass spectrometry results confirmed that different polymers emit different VOCs, and the researchers sought to clearly define their relationship with the degree of degradation. Thus, since changes in a single VOC were insufficient to effectively assess the degree of degradation, they used combinations of several VOCs as degradation markers for their statistical algorithm.

For each material, they divided the artificially aged samples into two classes:

  1. Class 1. These samples were  moderately aged with  0-4 weeks of artificial degradation
  2. Class 2. These samples were severely aged with  6-10 weeks of artificial degradation.

Then, they used data about combinations of VOCs to build a statistical model based on linear discriminant analysis (LDA) and trained it to spot differences between Class 1 and Class 2.

For example, for their 30 samples of polyurethane, they used the VOC results for 26 of them to “teach” the algorithm about differences between Class 1 and Class 2 in terms of types of VOCs and how much their levels changed at different times throughout the aging process, and then they tested the accuracy of the trained algorithm using the remaining 4 samples.

Their conclusions were very interesting. Using their coupled mass spectrometry/LDA methodology, they were able to classify 4 of the 6 investigated materials into Class 1 and Class 2, with accuracies varying between 62% and 83%. Polyethylene (PE) and polyvinyl chloride (PVC), however, were poorly classified, possibly because the degradation conditions weren’t harsh enough to induce significant aging.

The last step was testing their method with three actual artefacts from the Tate Modern in London, shown in Fig. 2.

Figure 2. a) Naum Gabo, Model for the statue of Aphrodite in the ballet “La chatte”, 1927. b) Antoine Pevsner, Head, 1923-24. c) Naum Gabo, Model for spheric theme, 1937. Adapted from Curran et al., Angewandte Chemie International Edition 57 (2018) 7336. Original link:

Two objects (shown in Fig. 2a and Fig. 2c) were made of cellulose acetate (CA). The third  (Fig. 2b) was made of cellulose nitrate (CN). The emitted VOCs were collected by placing a solid fiber close to the artefacts for one week at room temperature (a procedure called solid-phase microextraction, SPME.)

Using the artificially aged samples as references, the researchers found that the CA objects were at an early degradation stage (i.e. Class 1), while the CN Head was highly degraded (i.e. Class 2) and therefore needed to be prioritized for conservation and further analysis.

This work, the first of its kind in heritage science, showed that “sniffing” artefacts with mass spectrometry is an entirely non-invasive method that shows tremendous potential in the assessment of modern collections and in making decisions about conservation, storage, and monitoring. For these works of art, the smell of the past can shape the future.

All figures reproduced/adapted with permission from: Compound interest; John Wiley and Sons, Inc.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s