Bayeux Tapestry science: dye analysis, hydro-mechanics and hyperspectral imaging reveal its secrets

When the loan of the Bayeux Tapestry to the British Museum was announced in July 2025, there was opposition in France. Some said it was too fragile to make the journey across the channel for the first time since the Norman conquest. A 2020 conservation report had noted 24,204 stains, 16,445 creases, 9646 holes or gaps and around 30 rips. A legal challenge from French conservators even warned that the risk of tears during the journey was particularly high, given the number of potholes in UK roads.

But the loan was agreed – in fact, plans were already underway to move the tapestry from its usual museum home in Normandy, so it could be conserved and then returned to a new purpose-built museum in 2027. The French Ministry of Culture collected a group of conservators and scientists to safely manage the journey to London, including several ‘dry runs’ using a dummy tapestry.

Mechanical engineer Cecilia Gauvin, who runs the consultancy Smach, was part of that effort. Her team built a zig-zag folding frame to hold the tapestry inside a custom temperature-regulated crate for the journey, designed to keep vibrations below two millimetres per second. She also supervised the Tapestry’s removal from and placement in each museum’s display case, which took more than 80 people moving in concert.

Why all the fuss? ‘It is much more than a work of art. It is simultaneously a historical document, a masterpiece of embroidery, a political narrative and a symbol of medieval Europe,’ says Clarisse Chavanne, whose PhD at the Sorbonne University in Paris analysed dyes used in the embroidery. It illustrates the events leading up to and including the Norman conquest of England in 1066, but also ‘serves as a unique record of daily life in 11th-century Western Europe’, says Clémentine Paquier-Berthelot, head of exhibitions at the Bayeux Tapestry Museum. ‘Consequently, it is studied by experts across a wide range of disciplines: history, art history, archaeology … but also chemistry.’

Given its current condition and importance to scholars, it has seen several scientific studies examining its condition, but also its materials and how these have been altered over the years. The knowledge being gained could help answer some of the Tapestry’s mysteries, including where it was made, and what exactly is depicted.

The 70 metres of storytelling is made of wool embroidery threads on linen, a strong flax-fibre textile which has lasted better than the wool threads themselves. Where it was produced is unclear, but most experts think it was created in England in the 1070s, possibly in Canterbury.

Over the years the dyed embroidery threads have significantly faded from light exposure. Paquier-Berthelot says this is seen most acutely in the parts that were displayed facing windows and the effect is most pronounced in areas of past restorations (518 restoration patches have been identified) carried out with less durable synthetic dyes; ‘the blues have turned green’, she says, and ‘some wool threads have turned white’.

The effect of heat and humidity 

Following the 2020 condition report, Gauvin was tasked with getting a better picture of the hydro-mechanical properties of the Bayeux Tapestry; that is, the strains the fragile material experiences with changes in temperature and humidity. She has studied these properties for multiple works of art, including the Mona Lisa.

Linen is a hygroscopic material: its cellulose fibres are rich with hydroxyl groups so it tends to pick up water molecules and swell as humidity rises. Cycles of moisture absorption and drying become particularly destructive and can cause permanent damage, because hydrogen bonds shift are broken and remade across cellulose chains as the fibres dry, but in new stress-shifted positions. ‘If it’s very small and repeated over the years, you will have what we call fatigue damage, micro-cracks that become visible only after so many cycles of movement,’ Gauvin explains.

She has developed a way to monitor these movements in artworks using digital image correlation (DIC). The method images and then mathematically tracks how thousands of pixels move relative to each other. Measuring the hanging tapestry from the back of an additional linen layer that had been secured to the Tapestry in the 19th century, she took a series of one minute exposures every six hours over several months.

The data showed ‘the tapestry handles slow climate change very well’, says Gauvin, with little change. ‘Although if we see a very quick climate fluctuation, then the tapestry reacts very quickly,’ she adds. This means as long as short term fluctuations are avoided it should be possible to seasonally vary display conditions in the future museum to reduce energy consumption – a huge concern for museums today. Over that time, the relative humidity varied from 50% to 59%.

The other important issue she investigated was the optimum backing material for the Tapestry to rest on. To test this, Gauvin was able to experiment with several contemporary copies of smaller sections, one even produced by the mother of a fellow museum scientist. She tested the behaviour of brushed cotton and polyester microfibre, checking for cycles of ‘stick–slip’. This occurs when the static friction between two materials – the resistance to starting motion – is much higher than the friction occurring during movement, leading to sudden and damaging jerks before the surfaces re-grip.

She found significant stick–slip between the brushed cotton and linen, attributable to the high static friction from hydrogen bonding between the abundant cellulose surface hydroxyl groups on cotton and the cellulose and hemicellulose on the linen surface. The events were undetectable by eye, Gauvin says, but worrying because they were unpredictable which she says is the ‘worst-case scenario’. By contrast, the chemistry of polyester microfibres affords no surface hydroxyl groups to hydrogen bond with the linen and provided more consistent behaviour.

What dyes can tell us about Harold’s death

Conservators today ‘seek to stabilise areas of active deterioration while keeping interventions as minimal as possible’, says Paquier-Berthelot. That’s very different to previous restorations of the Bayeux Tapestry, particularly in the 19th century ‘At that time, the aim was often to restore an object so that it looked complete and visually coherent,’ says Chavanne, which included using the synthetic dyes starting to appear at that time.

In the 1980s, destructive testing of small samples of wool from the Tapestry were carried out using thin layer chromatography and identified three types of natural plant dyes used in the 10 original colours: madders, a group of red dyes from anthraquinone compounds; welds, giving yellows from flavonoid compounds; and indigos or woads, from the blue bis-indole molecule indigotin (two joined indole rings).

But that work was unable to pinpoint the exact synthetic dyes used in the Tapestry’s historic restorations – though they are easy to spot now. ‘The restoration threads look very different from the original colours,’ says Chavanne, ‘because early synthetic dyes were far less light-fast than natural dyes’.

Understanding the chronology of the major 19th century restoration when these synthetic dyes were used became one focus of Chavanne’s PhD, supervised by Philippe Walter, a chemist and director of the Laboratory of Molecular and Structural Archaeology at Sorbonne University. They hoped this might help provide some important historic context for some historical anomalies by dating some of the dyes used.

For example, the oft-repeated story that Harold II was killed by an arrow through his eye is now thought to have gained ground due to incorrect restitching during these restorations. Conservators in 1982 looked at detailed marks on the reverse of the tapestry which suggest Harold was actually holding a spear – also backed up by the appearance, in the earliest drawings from 1720, of holes in the fabric where threads had been lost. So the arrow may have been erroneously added by the restorers. Could the dyes used for these restorations help confirm this theory and provide a more precise date for the restorations?

Chavanne’s study, published in 2023, used non-destructive diffuse reflectance spectroscopy, measuring reflected light for each coloured thread, to be matched to reference spectra of the possible dyes. For the synthetic dyes she found an original 1876 book from French chemist Adolph Wurst, and a similar publication by English chemist Frederick Crace Calvert from 1877, containing synthetic protocols for trade-named dyes, along with fabric samples. ‘The textile samples had been protected from light inside the volumes for almost 150 years. They therefore preserved the colours and chemical characteristics of the dye,’ she explains.

One of the red coloured threads matched the synthetic dye aurin, the oldest of the triphenylmethane dye family, discovered through distilling coal tar in 1834 by German chemist Friedrich Ferdinand Runge. Chavanne also uncovered methylaniline violet (also known as methyl violet or ‘Violet de Paris’), a mixture of three triphenylmethane compounds with varying degrees of methylation, first synthesised by the French chemist Charles Lauth and commercially produced from 1866.

Pinning down the exact synthetic dyes used means the restoration must have occurred after 1866 when they were both available. ‘Since the tapestry was already restored in photographs taken in 1872, this allows us to date the restoration much more precisely, to between 1866 and 1872,’ says Chavanne.

But the 19th century embroiderers did not exclusively use synthetically dyed wool for their repairs. For example, she found both natural indigo and the synthetic rosaniline blue dye, ‘suggesting that the restorers simply selected whichever dyed threads best matched the original colours’, says Chavanne. The question of the specific dye used for Harold’s arrow remains unknown. While the yarn seems to be synthetic, it is so faded that Chavanne thinks it’s unlikely to be identified by spectroscopic analysis. But it does add weight to the idea that the arrow was not an original tapestry feature.

There are still other puzzles, including an identified dark yellow brown dye, which Paquier-Berthelot says is distinct from walnut ink or oak bark, both of which were used in the medieval period. ‘Yellow natural dyes are difficult to identify using reflectance spectroscopy,’ says Chavanne, as their signature spectra tend to be close to the low-wavelength limit of portable instruments, where measurement quality is poor.

Reconstructing the Tapestry’s original colours

Today the Tapestry looks faded and has certainly changed in appearance since it was made. Dyes used in the tapestry have degraded over time, due to light and UV radiation in particular, which Walter says has led to ‘a drop in chromatic intensity across the visible spectrum without altering the hue itself’. But there are complications, as not all dyes break down at the same rate: for example, flavonoids age quicker than indigo – so greens composed of yellow weld and blue indigo start to appear more blue. There is an additional effect from the yellowing of the wool fibres caused by the chemical alteration of some amino acids in keratin, the fibrous protein found in wool, which in reverse tends to give blues a greener tint.

Despite these changes, Chavanne and Walter realised that spectroscopy might help them recreate a complete facsimile of the original colours, restoring it to its former glory, without resorting to guesswork or approximations. Instead of measuring one spectrum at one point, hyperspectral imaging captures a full spectrum for multiple pixels across a 2D surface. While the colours are faded, their spectra recorded for each pixel represent the dyes present and can be translated into a representation of the original colour.

To do this took a few stages. First they recorded reference spectra for the 28 colours now found in the tapestry, from 10 original and 18 restored threads. They then used digitised photographs of these colours from the unfaded reverse side of the Tapestry, taken in 1982 during conservation work, to create digital RGB values corresponding to each of the 10 original colours. These would stand in for as close to the original colours as they could get.

Finally, they assigned the corresponding RGB colour to each reference spectra, with spectra from the 18 colours added during historic restorations each assigned an RGB value of one of the original 10 colours they would have been chosen to replace.

Hyperspectral support for the Canterbury theory

The subsequent hyperspectral imaging at a resolution small enough to differentiate each individual thread created a colour-corrected image of the tapestry, as it would have appeared 900 years ago. ‘This calculated image takes us back to the 11th century, reviving the vibrant, colourful world of the era,’ says Walter. He says it bears a striking resemblance to a hand-colored engraving from 1820, likely meaning most of the fading occurred after that date. Before this time it was only displayed rarely in Bayeux Cathedral, where Walter thinks the stained-glass windows would have effectively shielded it from UV radiation.

So far they have imaged the first nine metres of the Tapestry but Walter says it’s already giving hints to historians. ‘The virtual image reconstructing the original colors reveals intense shades, very similar to those seen in certain manuscripts, such as the ones from Canterbury housed at the British Library. It reflects the color palette and aesthetic tastes of that period and culture,’ he says. This does further support the theory that the Tapestry was made in Canterbury, known to be the home of a celebrated school of embroidery at the time. ‘Still, this finding is not a proof,’ says Walter, ‘just one possibility among others.’

The imaging method, initially developed for remote sensing, also provides a new non-destructive in situ method for studying cultural heritage, using technology not available to museum scientist in the past. ‘It represents an advance because we can now obtain chemical information without taking samples from the object. These techniques were simply not available in the 1980s,’ says Chavanne.

Lessons for a new museum in Bayeux

If it was made in Canterbury in the 1070s, its current sojourn at the British Museum will be its first return to the UK since then. Viewing all 70 metres of the iconic stitched history for the first time will be exciting for many Brits, but Gauvin is already planning its return journey and working out how to install it safely in its new Bayeux museum home.

In London the Tapestry will be displayed completely horizontally to eliminate any stress on the fabric, but Gauvin has also tested displaying it at either a 45° or 75° angle. ‘It is no surprise that 45° gives better results in terms of the creep and the load applied on the tapestry,’ she says. Although ultimately the new museum has opted for 75°, a compromise according to Gauvin between ‘the visitor experience and the conservation of the tapestry’.

Paquier-Berthelot says as well as a controlled climate, the new museum will keep monitoring the Tapestry with regular photography of sensitive areas to track any changes. ‘In the future, advances in imaging technology will allow for further analysis, incorporating 3D techniques (photogrammetry) to gain a better understanding of the embroidery’s textile layers,’ she adds.

Chavanne, who is now an outreach coordinator at the Musée Curie in Paris, says ‘working so closely with the Bayeux Tapestry was a real privilege. It is an extraordinary object, not only because of the story it tells, but also because of its remarkable survival.’ It is one of very few embroidered work from the 11th century to have survived in such a complete state, escaping wars, fires, the French Revolution and centuries of handling, ‘yet it remains largely intact’, she says.

Walter thinks the Tapestry still has more to tell us and with the right scientific tools we might one day definitively know where it was made. Although this would require physical sampling, he says, ‘one option could be genetic testing [to identify] the sheep that supplied the wool’. Advances in the field of paleo-genetics might make it possible to tell the differences between sheep flocks in Normandy and those in Southern England, for example. ‘There is still so much research left to be done,’ says Walter.

Rachel Brazil is a science writer based in London, UK