The Art of Noise: investigating the impact of the hidden threat of sound on artworks

 Checking for vibration in the canvas (Catherine Higgitt and Tomasz Galikowski). @ The National Gallery, London/Bickerdike Allen Partners LLP

By main authors Catherine Higgitt, Tomasz Galikowski, David Trew, and contributing authors Diogo Pereira and Jorge Garcia Garcia 


Like many cultural institutions, as part of its public programme and income generation, the National Gallery, London organises a variety of events and activities involving music and sound and has been doing so for many years. Fundraising concerts were organised as far back as 1922, and the Myra Hess concerts took place throughout the Second World War. 

However, the impact of repeatedly exposing artworks to the vibrations that music and other sound sources can induce is unclear. Although this issue is not new, it has perhaps become more urgent with the more frequent use of amplification and electronically generated music near artworks. Does exposure to sound pose an immediate or cumulative risk? And are there ways to mitigate or reduce any risk, for example, by avoiding certain types of music, setting limits on the sound or vibration levels experienced by artworks or introducing measures to reduce transfer from source to object? 

To begin to address these questions, it is vital to understand how sound-induced vibration reaches artworks. In the case of a sound source in the same room as exhibits, there are both airborne and structure-borne sound-to-object transmission paths, and their relative importance needs to be understood. Sound arriving at the object via each path will contribute varying levels of vibration, depending on a number of factors, such as the distance between the source and object or the properties of the materials through which the sound is transmitted. 

To investigate these factors, we recently undertook a pilot study using an anechoic chamber at London South Bank University. Anechoic chambers are specially designed rooms isolated from outside noise and vibration; the walls, floor and ceiling are covered with materials that absorb any sound that hits them, significantly limiting any sound from travelling through the structure of the room or bouncing back into the room. By placing our test paintings and a noise source within the anechoic chamber, it was possible to eliminate the structure-borne transmission routes (shown in blue and pink in diagram below). This made it possible to assess the importance of the direct, airborne route and explore whether airborne sound alone can cause artworks to vibrate or even induce damage. 



For the pilot study, several different canvas paintings were used. These included two non-accessioned framed paintings of unknown attribution and date (but thought to be from the 19th or possibly 18th century) that were assessed to be in fair condition by a National Gallery conservator. The other two test paintings were unframed canvases prepared specifically for the experiments and designed to be inherently fragile, with one having underbound paint layers and the other showing poor adhesion between paint layers.

The painting under investigation was suspended in the centre of the chamber (to ensure separation from the structure of the room) with a 175 W Line6 bass amplifier and a 15-inch speaker placed on the mesh floor of the chamber, directed towards the painting. The condition of the painting was carefully assessed before and after each round of testing to check for any visual change.  

The controlled laboratory conditions allowed tests to be carried out with a variety of acoustic source signals. Experience from previous acoustic testing informed the range of sounds used in these experiments. Music as a source is not ideal for objective testing or comparative studies, as there is substantial variation in the temporal and frequency content. For example, the low frequency or bass content of music will vary substantially from one track to another. Sounds below a frequency of 200 Hz are normally referred to as low frequency noise (LFN). We were particularly interested in investigating the impact of LFN as it has been known to induce vibration or rattle in objects (e.g. LFN-induced rattle in windows due to airborne aircraft noise). So called “pink noise” has equal sound energy across the frequency range and can be used as a consistent and repeatable source across laboratory and in-situ settings. Sine sweeps (where the frequency is varied with time, from low to high frequency or vice-versa) and individual tones can also be used to test how a sample responds to individual frequencies. This is particularly useful when testing whether a painting is resonating in sympathy with sound energy at a specific low frequency. 

To investigate exactly how the canvases responded to the sound sources, we used a laser Doppler vibrometer. This is an entirely non-contact optical method based on a red laser that allows vibration measurements to be taken at any point across the surface of a painting. The laser vibrometer employed an eye-safe Class 2 red laser operating at 633 nm and was equipped with a shutter to minimise the light exposure of the painting. Alongside the laser vibrometer, we also used standard accelerometers that were attached at various positions on the stretchers or frames of the test paintings to measure vibration. An accelerometer was also mounted on the vibrometer to correct for any vibration within the equipment itself during the experiments. An IEC 61672 Class 1 data logging sound level meter was used to measure sound levels immediately adjacent to the painting.



Our pilot study provided clear evidence that airborne sound can have a direct impact on canvas paintings, causing not only the canvas but also the stretcher and frame (if present) to vibrate; as such, airborne transmission routes must be considered alongside structure-borne routes when assessing the impact of sound on artworks. As has been observed before, the vibration levels measured within the canvases were higher than those measured in the stretchers or frames. These observations are important, both in trying to define criteria for sound levels in the vicinity of artworks, along with its character (time and/or frequency specific), and in thinking about how to approach sound-induced vibration mitigation. 

By analysing the vibration measurements from the laser vibrometer and accelerometers during the experiments, it was also possible to establish the specific frequencies of the various vibrational modes of the test canvases. These are the frequencies at which the paintings will respond most and at which the highest amplitude displacement will occur. With our experimental setup, we were able to expose the various test paintings to sounds (or tones) at specific frequencies corresponding to these vibrational modes. The frequencies involved were in the LFN range which is not surprising as the canvas support of a painting tensioned around a stretcher is similar to a stretched drumskin, which as noted below tend to have frequencies in the LFN range. Even at relatively modest levels of LFN, resonant vibration was clearly measurable during the experiments. 

This observation is relevant as music instruments are capable of generating sounds in the LFN range. For modern music, a low E note played on a bass guitar has a frequency of 41 Hz, and bass drums have a frequency of 50-100 Hz. These low frequencies are more likely to induce vibration within paintings, a finding clearly observed in the experiment.

The experiments also provided what seems to be the first clear evidence that airborne noise alone is capable of inducing damage in our test paintings. We observed that, in some cases, even quite brief exposure of the test paintings (both those specifically prepared for the tests as well as the non-accessioned 19th-century paintings) to such LFN tones caused damage, with paint loss or widening of cracks noted. These effects were observed with LFN at levels of around 95-110 dB at individual octave bands of 31.5Hz, 63 Hz and 125 Hz. From the full series of experiments undertaken, it was also clear that it was not the overall level of the sound source which influenced the impact on the test paintings but the specific frequency content of the source. 

The overall level of a sound source is typically reported as dB(A), where “A” stands for A-weighting. This parameter indicates a measurement of sound pressure adjusted to the response of the human ear and measured over the frequency range 20–20,000 Hz. Due to the characteristic of human hearing, the A-weighted sound is stripped of a significant amount of energy in the LFN range, and as a result, the overall sound levels may be of similar magnitude even if the frequency content varies significantly. As such, the dB(A) parameter is not suitable for assessing the impact of sound on physical objects.



After three very intensive days in the anechoic chamber, we are still working through the data acquired and hope to share the results in a journal article in the near future. We are also contributing to the development of a good practice guide for musical events—one of the anticipated outcomes from our work with the international research group which recently published the results of a questionnaire focused on the impact of vibration from music and transportation on museum collections. We intend to regularly update this guide as new findings—such as those from our pilot study and other research—become available.

As with all good pilot studies, we have raised as many new questions as we have answered, but we have been able to develop a robust experimental approach and explore the potential of laser vibrometry for heritage applications. The impact of exposure to sound is thought to be a cumulative process, but this is yet to be researched in detail. Unfortunately, it was not possible to expose the test paintings to noise for extended periods of time as part of this pilot study. However, we hope to extend the study to address some of the new research avenues we have identified with the aim of providing practical guidance to improve our understanding of the impact on collections of repeated exposure to sound-induced vibration and how to minimise the associated risks.

The National Gallery has, for many years, researched, monitored and managed levels of vibration from various sources including construction projects and events involving music. Those studies relied on the use of accelerometers which are often too large to safely fix to the canvas and heavy enough to affect the response of the canvas. While laser vibrometry has generally been limited to use in laboratory settings, after the anechoic chamber tests, the equipment was brought to the National Gallery. Further tests demonstrated that this method would be suitable for on-site measurements to monitor vibration levels on sensitive objects in heritage organisations or historic buildings where traditional methods of surveying or monitoring using attached sensors are not possible. The laser vibrometer has a range of 30 m, allowing monitoring of not only sensitive but also previously inaccessible artworks (e.g. wall paintings, painted ceilings, etc). We also found that a laser vibrometer could be easily integrated into existing accelerometer-based monitoring systems if required.

Although laser vibrometry has been used in previous studies to assess vibration experienced by artworks,  it is still a rarely used technique within heritage applications, and this study is believed to be the first example of its use to assess artwork in an anechoic chamber.



The authors gratefully acknowledge the help and support of Professor Stephen Dance and the EPRSC-funded UK Acoustic Network Plus (UKAN+) for the grant [EPSRC UKAN EP/V007866/1] that made it possible to access the anechoic chamber in the School of the Built Environment and Architecture at London South Bank University. They would also like to thank colleagues at the National Gallery, particularly Lynne Harrison, for her assistance with preparation of the test paintings. 



Catherine Higgitt is principal scientist at the National Gallery in London. She was previously head of science at the British Museum where her involvement in the World Conservation and Exhibitions Centre project started her interest in understanding the impact of vibration from a range of sources on heritage collections and approaches to mitigation of the associated risks.

Tomasz Galikowski is an acoustic consultant and associate at Bickerdike Allen Partners LLP with over 14 years of relevant experience and a research interest in the effect of structural and air-borne vibration on cultural heritage artefacts and historic buildings. Examples include the British Museum WCEC and National Gallery NG200 projects and work at the V&A and Science Museum Group.

David Trew is a partner at Bickerdike Allen Partners LLP. He has an engineering degree in acoustics and vibration from the Institute of Sound and Vibration Research. David has over 25 years of experience as an acoustic and vibration consultant. He is a visiting lecturer at the UCL Institute for Environmental Design and Engineering, part of the Bartlett School of Architecture. 


Read the article in the June-July 2024 "News in Conservation" Issue 102, p. 12-18