Bedroom secrets

Much to my delight The bedroom is back ‘home’ in Amsterdam after having been on loan to Japanese museums for a period of several months. The painting was thoroughly restored before it travelled to Japan. We reported on this in the museum and in this blog under the title Bedroom secrets. This culminated the presentation of the restored painting in the gallery. The public’s reactions were overwhelming.

To give visitors the opportunity of following the technical and scholarly decisions underlying such a restoration, we are presenting Bedroom secrets once again.

Reconstructions in 2D and 3D

A special attraction has been added this year: a life-size reconstruction of the bedroom! The bedroom in the Yellow House in Arles has been faithfully recreated on the basis of descriptions in Van Gogh’s letters and naturally the painting itself. While the objects and the furniture in it are not authentic, they reproduce the room’s atmosphere to the extent possible.

Reconstruction of Van Gogh's bedroom in Arles

Also available now is a colour impression of the painting as it probably looked when it had just been completed. Technical analysis of the painting revealed that it had severely discoloured in the course of time. Look and compare for yourself!

Enjoy this ‘multi-dimensional’ presentation, which will be on view until January 15th, 2012.

On the play of complementary colours: cool and warm, primary and secondary

The clash of the primary colours

In its current restored state, Van Gogh’s The bedroom is a clean, crisp, vivid painting. Yet its cool atmosphere and lively colours are only part of the story. The bright blue walls suggest a spaciousness that turns out, on closer scrutiny, to be at odds with the relative proportions of the floor and the bed and with the painter’s description of the room in his letters. Currently, the three primary colours are in conflict, with blue predominating. From a painter’s point of view, the work is fraught with tension.

Colour and tone

‘I had wished to express utter repose with all these very different tones,’ Van Gogh wrote to Gauguin. After staying indoors for two full days to rest his strained eyes, the first thing he had painted was the interior of his small bedroom.

Calm, or ‘repose’, is suggested in a painting by avoiding contrasts. This means using colours that are equally light or dark – in other words, equivalent tones. The quickest way to determine whether you’ve actually created a calm atmosphere is by squinting at the painting through your eyelashes, so that the details fade and only the main forms and contours remain. If we look at The bedroom this way, we immediately see what is wrong with it: namely, the walls and the blanket. The walls are too light in tone, and the blanket too dark. The digital impression restores the lavender colour of the walls (by removing the white) and brightens the red of the blanket, bringing the tonal values into balance and restoring the sense of repose.

Vincent writes that the only white he wanted in the painting was the reflection in the mirror. As he puts it, the fourth pair of complementary colours, white and black, is represented by the mirror and its frame. At present, this effect is cancelled out by the abundance of white in the walls.

Impression of the original colours

Impression of the original colours of The Bedroom.

Cool and warm colours

The reconstruction shows the logic of the original colour scheme, based on a single, dominant primary colour and the interplay of mutually reinforcing, complementary secondary colours, both warm and cool.

The cool colours:

Purple (violet) – walls
Terracotta – floor
Lime green – sheets and pillow
Emerald green – window frame

The warm colours:

Scarlet – blanket
Chrome yellow – bed
Ultramarine – door frames

Van Gogh also uses a range of blues, such as cool Prussian blue, tending towards green, and warm cobalt blue, inclining towards red. Combined with white and the warm but unstable pigment known as geranium lake, which was a favourite of Van Gogh’s, cobalt blue gives a beautiful violet-purple hue.

Leading part for the blanket and the bed

The central colour in the painting is red, which Van Gogh applied quite thickly. Around this primary colour revolve several others, most of them complementary and secondary: cool purple (composed of red, white and blue), warm yellow and various warm and cool hues of green (composed of blue and yellow).

The eye is first drawn to the red blanket – which is currently a cool, discoloured red, comparable to Cadmium Red Dark – embedded in the warm yellow of the bed frame. The floor and the furniture are linked by their yellow and red elements and form the warm part of the scene. The distinct benefit is that, as a group, these elements clearly project themselves into the foreground and form the basis of the composition.

The walls and doors are a cooler variation on the pink of the floor and seem to embrace the foreground. At this stage the function of the green window, ensconced in the violet, becomes apparent. Green is complementary to the primary colour red. Because green is composed of the other two primary colours, red and green reinforce each other and are often used in combination.

Against the faded blue-white of the walls in the painting in its present state, the green window frame is much too heavy and isolated. It seems to float free of the wall, drawn to the red blanket and the green lines at the foot of the bed. The violet of the walls in the reconstruction reveals the warm light entering the room from outdoors, softened by the shutters. This effect is currently overpowered by the bright bluish-white in the walls, which resembles a primary colour. The reconstruction restores Van Gogh’s small, warm room in an old house with thick walls in the sun-drenched south of France.

Thanks to the latest technology and, even more importantly, to collaboration between the various disciplines that study art and colour, we can now see and understand the artist’s work in the way he most probably intended. For me as a painter, knowledge of the original materials is an absolute necessity before I can form an opinion about a painting or offer an art-historical interpretation. Vincent would be greatly relieved to know that his intentions are now visible once again.

Finally, I give you a short summary of the recent months in pictures. It has been a real pleasure to share the restoration process with you. Thank you for all your comments.

.

The bedroom has been restored. Until April 10, 2011, it can be admired in Japan.

From April 2011 the painting will be back home in the Van Gogh Museum in Amsterdam.

Under what conditions can we conclude that the restoration of The bedroom has been successful? My answer is: if its condition and appearance have been improved. By this standard, Ella has done magnificent work. Old restorations have been replaced with better ones that do more justice to what Van Gogh wanted (namely calm, uniform areas of colour), and the painting as a whole has a new intensity that brings out the colours much better. It was already a strong painting, but its ‘wall power’ is now even greater.

But that’s not all. Ella and all the people who supported her did extensive research, unearthing not only a great deal of information but also new questions. I think this result of the project is at least as important. Should the museum redo all the other old restorations? Were the edges of other paintings folded around the corner when they were re-stretched? Should the discoloured, greenish patch under the chair at left in The bedroom (which was not intended as a shadow) be retouched to be more like the rest of the floor?

The most intriguing issue, of course, is discolouration. Now that we know the purplish-peachy colour of the floor and the blue of the walls were originally more like orange-red and violet, respectively, we would love to figure out how the work originally looked. Though I doubt we will ever be able to reconstruct the original colour scheme exactly, in the future we will look at other works for which Van Gogh’s letters mention one colour (violet, for instance) while we now see another (such as blue). To begin with, there are the other versions of The bedroom, but we could also look at Van Gogh’s oil copies of black-and-white prints, which he called ‘translations in colour’. For instance Snow-covered field with a harrow.

Vincent van Gogh, Snow-covered field with a harrow, 1890

What looks almost like white snow in this painting must originally have been orange-pink, because the setting sun was shining on the white snow.

In short, there is still plenty of work ahead. In fact, we’ve already begun: you can follow the conversation at our blog on Van Gogh’s studio practice. One day we’ll discuss The bedroom there, along with some of these new questions. It just goes to prove what they say about great works of art: there’s always more to be discovered.

Compromises 

Despite the perfectionist streak common to most conservators, in fact many of the steps we have to take are based on compromise. This was certainly the case for the decision to re-varnish The bedroom after cleaning, despite the fact that the painting seemed not to have been varnished originally.

A new varnish

A main reason for this choice was the sensitivity to solvents of certain colour areas, which had made it impossible to remove the old varnish everywhere from the painting. For safety reasons, the varnish was left intact in the vermilion bedspread, as well as in smaller areas, such as the Prussian blue mirror frame.

Application of a new matt varnish would help to knock back the disturbing gloss of these areas, whilst slightly improving saturation of others, evening out the surface.

New retouches

In addition, removal of extensive old overpaint had revealed many spots of damage and open cracks scattered across the painting. A new brushed varnish layer was felt desirable to isolate original paint from the fillings and retouches I would apply to reintegrate these paint losses.

A ‘non-varnished’ look

As a final measure, a very thin layer of matt varnish with added wax would be sprayed onto the painting to avoid a varnished look. Unlike the dammar varnish applied by the previous restorer, stable and non-yellowing materials would be used that are anticipated to remain easily soluble in mild solvents if required (Regalrez 1094, a low-molecular weight hydrocarbon resin varnish with Cosmolloid 80H wax added) .

With this choice it is felt that both the safety and satisfactory appearance of the painting have been guaranteed.

A retouching palette

Van Gogh and his contemporaries were keen to apply modern theories of colour science to the art of painting. For me, an innovatory aspect of The bedroom treatment was the opportunity to work together with colour scientist Roy Berns in order to select an ‘ideal’ retouching palette that I could use to touch in old paint losses (so-called ‘inpainting’). Based on the known spectral properties of the Gamblin range of retouching colours, it was possible to determine a recipe to match different spots of colour that had previously been measured on the painting.

Importantly, this match is predicted to persist for different light sources, so that my retouches would not suddenly ‘change’ if the painting is hung in different lighting. I was curious as to how this process would compare to the usual trial-and-error approach of mixing and matching colours by eye.

Trial and error

Whilst waiting for the colour recipes and some missing pots of colour to arrive, I experimented with the Gamblin colours we had in stock. Intuitively, it turns out that I did reach the same choice of blue pigments that, based on calculation, would be recommended for use in the blue walls and outlines of the door panels (cobalt blue and ultramarine respectively) mixed with titanium white.

The recommended set of colours

However, I had problems with matching the elusive chrome yellow colour of the foot end of the bed with its aged greenish-brown translucent skin. Great was my delight when the recommended set of Gamblin colours arrived from America and indeed turned out to provide a very good basis for retouches in the bed frame (roughly equal quantities of  Naples yellow deep and Permanent green light with a little Dioxazine purple and Titanium white added).

Detail of the foot end of the bed after cleaning, before retouching.

The same area after retouching. Some old residues of brown overpaint that could not be safely removed have been touched out with new retouches.

Science and intuition

I am quite certain I would not have reached the same choice alone, since especially the rather lurid  Permanent green light seen in the pot was not an obvious candidate to me. Working with this basic set of colours, the old-fashioned skills of the conservator remained essential to make the subtle adjustments of colour, texture, opacity and layering required at each spot on the painting. As for Van Gogh, it seems that colour science and intuition can work well hand-in-hand.

Retouching materials and palette

 .

During 2002 and 2004, I made spectral reflectance measurements of several areas of Van Goghs painting Daubigny’s Garden. Unusually, it is painted on a tightly woven, red-striped linen cloth known as torchon. Van Gogh prepared the canvas himself with a pink ground layer containing the red pigment geranium lake and lead white. Today that layer has faded to gray. I measured the gray ground layer and an area along the edge that, due to damage, revealed the fantastic pink colour shown below. 

Edge of the canvas of Daubigny's garden

.

This area was quite small and my spectrophotometer was averaging the pink and some of the surrounding gray. As a consequence, when simulating the ground colour, I used spectral information from a paint-out of eosine dye, the chromatic ingredient of geranium lake, and titanium white gouache. 

Because geranium lake was also used in The bedroom, I revisited my measurements of Daubigny’s Garden. Using light absorption and scattering theory and nonlinear optimization, I was successful in estimating the colour properties of Van Gogh’s geranium lake. A concentration series with lead white is shown below. Clearly, this colourant would have been very appealing. 

Possible colour concentrations of the ground layer of Daubigny's garden

.

During 2008, I measured over 50 locations on The bedroom: mainly to evaluate imaging accuracy (to determine the current colours of the painting), but also to help Ella Hendriks choose appropriate pigment combinations for retouching. Here I am measuring the painting. 

Doing colour measurements on The bedroom

.

Here’s a close up of one area I measured and you can see that there is a large mismatch in colour between retouchings from 1931 (pinkish) and the surrounding original paint (purplish). It appears that the floor has continued to fade since 1931. 

Detail of the floor with retouches that do not match with the original paint

.

A question was posed, ‘Could the geranium lake from Daubigny’s Garden be added (computationally) to the colour of the floor of The bedroom and match the 1931 retouching?’ The answer was, ‘yes’, and here is colour series showing the result of adding back to geranium lake, both to match the 1931 retouch and earlier in time, though going back earlier is speculative because we don’t know the original concentrations. 

Although I have already started with some tests, I hope that in the future I can attempt a digital rejuvenation of the entire painting.  For now, here is a preview, of the possible colours of the floor in 1931.

Digital juvenation of the floor

Detail of the floor in current colours

When you write about the use of colour as an art historian, it is very important to realise that you can base your remarks solely on the object itself. For most reproductions of artworks are far from reliable. That is because every camera, every film, every scanner, and every computer screen produces its own colour variations. Just try browsing the internet for The bedroom and you will be astonished at the different variants that appear on your screen!

To illustrate how great these differences may be, I am including with this post four of the search results I obtained myself. Not one of these four reproductions resembles the actual painting.

It was long believed that the dissemination of art reproductions – at first in the form of photographs and today also through the internet – would lower the status of the original work. After all, these days anyone can hang a reproduction of an important work of art like The bedroom in their own home. But the unreliability of these images actually has the reverse effect: the original becomes all the more important.

We have already mentioned that the intensity of the colours of The bedroom have changed, moving away from the intentions of the artist.

The Van Gogh Museum is conscious of the difficulties that surround the reproduction of works of art, and therefore has its own ‘colour management’. A photographer specially hired for the purpose and an image specialist are busy all day producing accurate images of the items in our collection. The images of The bedroom that are shown on this site can therefore be described as reliable. Well, except … have you checked the colour settings on your monitor?

What material is responsible for the discolouration of the floor?  

During the examination of The bedroom, Ella asked me to come by and look at the painting with her. She had noticed something that was probably a result of the materials used by Van Gogh. Specifically, she was curious about why the floor in The bedroom had changed colour. What material could be responsible? Could the paint contain eosin, an organic red pigment that we had found in the ground (the first paint layer) during our earlier examination of the painting Daubigny’s garden? 

Cross-section of a paint sample from Daubigny’s garden. The ground (the pink paint layer at the bottom) consists of a mixture of lead white and eosin.

Geranium lake 

Eosin is a synthetic dye, first produced in 1873. It is a compound of carbon, hydrogen, and bromine. When this dye is precipitated on a substrate (usually alum), it produces a pigment that can be used in oil paint. This pigment was sold under the name of Geranium lake, because it is similar in colour to geraniums. Even though Van Gogh knew that Geranium lake faded when exposed to light, he used it frequently. In eight letters to his brother Theo (from Arles, Saint Rémy, and Auvers-sur-Oise), he asked for Geranium lake from the Paris firm of Tasset et l’Hôte. In total, he ordered no fewer than 44 tubes! 

Searching for evidence 

To take a closer look at the discolouration of the floor in The bedroom, we first asked our colleague Luc Megens to analyse the painting with X-ray fluorescence (XRF) spectrometry. This technique involved direct examination of the painting. In the floor in the foreground, he established the presence of zinc, lead, and mercury, but not of bromine. We were disappointed, because bromine is a telltale sign of eosin. This finding suggested that no eosin, or very little, was present in the pigment that Van Gogh had used for the floor. 

But I was not ready to give up. I obtained a sample from the bottom edge of the painting and took a cross-section. 

Cross-section of a sample from the bottom edge of The bedroom. On top of the light ground, we can see the pink paint layer used in the floor.

This made it possible to study the composition of the paint layers under a microscope, at a magnification of approximately 1,000 times. With the aid of a scanning electron microscope with energy-dispersive X-ray spectroscopy (SEM-EDS), which I use with Kees Mensch at Shell, I managed to identify most of the pigments in the pinkish paint layer. It proved to be a mixture of zinc white and lead white, coloured with small amounts of vermilion, Schweinfurt green, and chrome yellow. But this did not answer my question, because it did not seem possible that these pigments could be responsible for the discolouration in the floor. 

In a final attempt to discover the cause of the discolouration, another colleague of mine, Maarten van Bommel, took a sample of the pinkish paint used for the floor. He analysed this sample using a technique called high-performance liquid chromatography (HPLC), which is particularly well suited for analysing dyes. What he discovered was that, in fact, there was a small amount of eosin in the pinkish paint. And just as we’d suspected, the eosin was almost certainly the cause of the discolouration. Finally, the mystery was solved!

In theory, one could choose to overpaint the floor to return it to how it looked around 1931, using the peachy colour of the preserved strip of original paint along the bottom edge and of Traas’ first retouches as a guide. However, nowadays, such integral overpainting is not an option, since we prefer to show as much as possible of Van Gogh’s own paint. Furthermore, any such attempt would have a too speculative character. Instead we will try to exploit modern technologies to digitally rejuvenate the colour scheme of the painting in a computer image. This might give us an idea of how it could have looked when made. Besides adjusting the colours of the floor, this may include a reconstruction of the ‘pale violet’ walls and ‘lilac’ doors described by Van Gogh, which, again as a result of faded Geranium lake, have now turned to light blue. A blog post by colour scientist, Roy Berns, will describe how this is achieved. 

The disturbing mismatch of retouches in the floor provides a dilemma for the current restoration. Should I try to remove part or all of Traas’ retouches and attempt new ones that provide a better match to the present colours of the floor? And if it is chosen to replace the old retouches, will the original colours of the floor continue to slowly change, despite the stringent museum conditions of light and climate under which the Bedroom is now kept? I will come back to these issues.