Shipwrecked art history: Géricault’s Raft of the Medusa

Théodore Géricault, The Raft of the Medusa (1818-9), oil on canvas, 491 x 716 cm, Musée du Louvre, Paris. WikiArt.

Sometimes nuggets of knowledge and understanding just seem to drift off into increasing fiction. Unfortunately the Internet and Wikipedia do not appear to help this; in many cases they are part of the cause.

If you read my recent review of Julian Barnes’ excellent collection of essays on art, “Keeping an Eye Open”, you may recall that I questioned some of the information he gives about Géricault’s painting The Raft of the Medusa (1818-9). I wrote:

Although commendable, there are some oddities buried within it, most particularly his claim that the “heavy, fast-drying oils he used meant that each section, once begun, had to be completed that day”. A similar account is given in Wikipedia and many online sites which often quote Wikipedia’s words verbatim, but none provides any source for this information.

Whilst not impossible, perhaps by using resin additives, oil paints which dry overnight (alkyds) did not arrive for over a century after this painting was completed.

One possible explanation is Géricault’s extensive use of asphalt/bitumen, which solidifies but does not actually dry at lower temperatures. Barnes appears surprised at the use and effect of asphalt/bitumen in this painting, but its use was extensive even much later in the nineteenth century, and its adverse effects are widespread across the works of many artists of the day.


Barnes does not identify his sources in this new book, as the essays have each been published previously, in this case in his semi-fictional book A History of the World in 10½ Chapters (1989). There he states that his account “relies heavily on Lorenz Eitner’s exemplary Géricault: His Life and Work (Orbis, 1982).”

The passages of Barnes (2015) with which I have difficulties include:
“the heavy, fast-drying oils he used meant that each section, once begun, had to be completed that day.” (p 22.)
“The width of his brushes, by the way, is surprising. From the breadth of his manner, Montfort supposed that Géricault used very thick brushes; yet they were small compared to those of other artists. Small brushes, and heavy, fast-drying oils.” (p 34.)
“To make the shadow as black as possible, Géricault used quantities of bitumen to give him the shimmering gloom he sought. Bitumen, however, is chemically unstable, and from the moment Louis XVIII examined the work, a slow, irreparable decay of the paint surface was inevitable.” (p 39.)


Although he could not have used Wikipedia for research in the late 1980s, as it was not created until 2001, I thought it would be valuable to see what Wikipedia had to offer (as at 3 June 2015):

“Géricault painted with small brushes and viscous oils, which allowed little time for reworking and were dry by the next morning. He kept his colours apart from each other: his palette consisted of vermilion, white, naples yellow, two different yellow ochres, two red ochres, raw sienna, light red, burnt sienna, crimson lake, Prussian blue, peach black, ivory black, cassel earth and bitumen. Bitumen has a velvety, lustrous appearance when first painted, but over a period of time discolours to a black treacle, while contracting and thus creating a wrinkled surface, which cannot be renovated.”

The reference given for that appears to be a preview chapter of another book, The Victorian Visitors (2000) by Rupert Christiansen. Its first chapter starts with a summary of Géricault’s biography, then gives an account of the Raft of the Medusa, including the following:

“In every respect he was orderly and methodical, keeping the colours of his palette (vermilion, white, Naples yellow, two yellow ochres, two red ochres, raw Sienna, light red, burnt Sienna, crimson lake, Prussian blue, peach black, ivory black, Cassel earth, bitumen) carefully separate and the studio clean and tidy. Complete silence was an absolute prerequisite and he told one friend that even the scuffling of a mouse could stop him working. He used small brushes and particularly thick, sticky oils, which dried overnight and left little opportunity for second thoughts.”


Lorenz Eitner (1919-2009) gained his PhD from Princeton University for his thesis on Géricault, and chaired the Department of Art and Architecture at Stanford University from 1955 until his retirement in 1989. He was the authority on Géricault over the last 40 years of the twentieth century, and even today’s harshly critical approach to previous art history has done little to question that reputation or reliability, at least not on matters of fact rather than interpretation and criticism.

Théodore Géricault, The Raft of the Medusa (1818-9), oil on canvas, 491 x 716 cm, Musée du Louvre, Paris. WikiArt.
Théodore Géricault, The Raft of the Medusa (1818-9), oil on canvas, 491 x 716 cm, Musée du Louvre, Paris. WikiArt.

His 1972 monograph Géricault’s Raft of the Medusa provides great detail about the painting and its production in its 176 pages, much of which relies heavily on the biography and catalogue raisonné of Géricault written in the the 1860s by Charles Clément. Eitner provides several long sections translated from Clément, and most of the rest of Eitner’s description and analysis of the final painting process (pages 31-9) is rewritten from Clément’s words.

Eitner (p 32) summarises Géricault’s process in writing that the recollections of those who had come to the studio during the final painting “show that he followed a method which David had introduced among French artists, and which Géricault may have learned from his former teacher, Pierre Guérin.” To Eitner, “the difficulties of this procedure were formidable” because of its “piecemeal execution” (p 33). However it is considerable more in accord with “the psychology of artistic work” and more straightforward in execution than, for instance, the painting of fresco, a task successfully accomplished by large numbers of professional painters over many centuries.

Some of Eitner’s hyperbolic account of these difficulties (pp 32-3) reveals his naïvety. For example, he writes about the difficulty of the task “in the changing light of winter, spring, and summer”. The main reason that painting studios (in the northern hemisphere) are almost invariably lit from the north is to minimise diurnal and seasonal variation in the quality of the natural light entering them. The same issues are common to every painting undertaken over a period of more than a few days.

Although he quotes Montfort’s remarks about the nature of the paint used by Géricault, and the size of his brushes, translations from Clément are edited, and Eitner makes no further comment on those matters. However he does provide possible insight into the matter of brush size when he describes Géricault’s earlier studies thus:
“The heads and limbs in them, about the size of life, are modelled by small, closely blended strokes, while the draperies and backgrounds are very broadly brushed.” (p 36.)
Eitner does not comment on whether his examination of the final work showed similar evidence of differential brushwork according to the object being depicted.


Charles Clément was born in 1821, two years after Géricault completed and showed the painting in Paris, and died in 1887. Eitner praised his biography and catalogue raisonné of the artist as “one of the very best art-historical monographs produced in the nineteenth-century.”

However Clément’s book – Eitner’s main source for his monograph – was first published in 1868, nearly 50 years after the completion of the painting. Clément’s account of the painting of The Raft of the Medusa relies heavily on an oral account by Antoine Alphonse Montfort (1802-1884), who was then aged 17 years and a “young” pupil of Horace Vernet (whose father had been Géricault’s teacher).

According to Clément (p 138), Montfort had visited Géricault many times during the winter of 1818-9, when Géricault was in the early stages of painting the final work. Clément does not record when Montfort gave this oral account of what he saw in Géricault’s studio, but it could hardly have taken place before 1845, when Clément would have been a tender 24, Montfort aged 43 and recollecting what had happened nearly 25 years previously. It is more likely that Montfort gave his oral account to Clément during preparation of the biography, most probably around 1865, when Clément would have been 44, Montfort aged 63 and recollecting what had happened nearly 50 years previously.

Neither Clément nor in his turn Eitner consider how accurate Montfort’s account would have been, nor any errors which might have arisen in Clément’s written notes of their meeting.

Montfort (according to Clément 1879)

The account recorded by Clément appears in largely verbatim translation in Eitner’s monograph, with some significant omissions. I therefore give Clément’s words below, and have emboldened and italicised those sections of significance.

“Il était du reste souvent forcé d’agir ainsi par l’importance du morceau qu’il avait commencé le matin, et qu’il fallait terminer dans la journée. Cette obligation était plus impérieuse pour lui que pour tout autre, car, employant pour peindre une huile grasse des plus siccatives, il n’aurait pu reprendre le lendemain le travail de la veille. Quelques personnes ont pensé que les craquelures survenues dans sa peinture sont dues à l’emploi de cette huile extrasiccative. Il n’en est rien suivant moi, et je pense qu’elles furent causées par le vernis qu’on apposa sur le tableau lorsqu’il était à peine terminé. Très-peu de temps après l’exposition, on le roula pour le transporter en Angleterre et il a certainement beaucoup souffert de cette opération. Quoi qu’il en soit, je fus très-vivement impressionné du soin qu’apportait M. Géricault à son travail.” (Clément p 139.)

“Sa manière de procéder, toute nouvelle pour moi, ne m’étonnait pas moins, d’ailleurs, que sa profonde assiduité. Il peignait au premier coup sur la toile blanche sans aucune ébauche ou préparation quelconque en dehors du trait bien arrêté, et la solidité de l’ouvrage n’en était pas moindre. J’observais aussi avec quelle intensité d’attention il fixait le modèle avant de toucher la toile, paraissant aller lentement quand, par le fait, il exécutait très-vite, posant de suite chaque touche à sa place et n’ayant que rarement besoin de revenir. Nul mouvement, d’ailleurs, soit du corps, soit des bras; il avait l’air parfaitement calme, et une légère coloration du visage indiquait seule la préoccupation de son esprit. Aussi, témoin de ce calme extérieur, était-on d’autant plus surpris de la verve et de l’énergie de son exécution. Quelle saillie! surtout lorsqu’une partie n’était encore que préparée; cela ressemblait à un fragment de sculpture à l’état d’ébauche. A voir cette peinture si large, on pourrait croire que Géricault se servait de trèsgrosses brosses; il n’en était rien pourtant: elles étaient petites, comparées à celles employées par divers artistes que j’avais déjà connus, et il est facile de s’en convaincre à l’aspect de plusieurs figures de son tableau entièrement peintes par hachures.” (Clément pp 140-1.)

Clément also provides us with the following as Footnote 2, on pages 141-2:
“Une note de M. Jamar me permet d’indiquer les couleurs que Géricault employait et la manière dont il les classait. Elles étaient rangées sur sa palette dans l’ordre suivant: vermillon, blanc, jaune de Naples, ocre jaune, terre d’Italie, ocre de Brie, terre de Sienne naturelle, brun rouge, terre de Sienne brûlée, laque ordinaire, bleu de Prusse, noir de pêche, noir d’ivoire, terre de Cassel, bitume. 11 travaillait avec une très-grande propreté, gardant les tons séparés sur la palette, qui le soir paraissait à peine avoir servi.”

Jamar was Géricault’s studio assistant, who was present and assisting throughout the final painting period. The contents of Jamar’s note confirm the source for the lists of pigments given above, but regrettably Clément does not record when Jamar wrote that note. What is significant is that Jamar did not appear to provide any remarks on the size of brushes used by Géricault, the remarkably quick drying of the paint, or other peculiarities apparently recalled so clearly by Montfort.

Montfort’s description of the oils used by Géricault is that they contained “huile grasse des plus siccatives”. In modern English terminology, this means linseed oil medium with added siccatives or drying agents. Indeed the published records of materials purchased by Géricault during the final painting include huile d’oeuillet, huile grasse, and essence: in English, poppyseed oil, regular linseed oil, and solvent (turpentine, or possibly petrol or oil of spike), respectively (Rosenthal 1980).

The use, effects and dangers of siccatives or drying agents in oil paints were well known at the time, and Montfort was correct in reporting that their use could have caused the fine cracking or craquelure in the painting, as would rolling the canvas for transport, as he reports was done for the painting to be transported to England for exhibition after the Paris Salon of 1819.

It is also of interest that Montfort is not quoted as having any concern for the excessive use of bitumen in the painting, nor its potential effect on the longevity or structural integrity of the paint film. This is most probably because, as reported by Eitner, the majority of the bitumen paint was applied late during the painting process, when Montfort was no longer accessing Géricault’s studio.

Montfort reported that Géricault painted (at least the figures, during the winter of 1818-9 when Montfort had access to the studio) au premier coup, and without any ébauche, or underpainting. These terms are still used today, although the first is more usually given in its Italian equivalent, alla prima.

Eitner correctly identifies this method as one used by David for painting the figures and key details in large works on canvas, and there was nothing particularly remarkable about its use by Géricault. Indeed, given the concentration of intertwined bodies in the centre of the painting, Géricault’s alla prima and morceaux approach was probably the most effective and efficient way of tackling the task.

As I have remarked above, Eitner’s examination of the painting comes to different conclusions with respect to the size of brushes used. Whilst he concurs with Montfort that the figures were painted using relatively small brushes (not unusual given the nature of the subject), he states that larger brushes were used for the rest of the painting, something neither observed nor recalled by Montfort, who would only have seen Géricault working on the figures.

Drying of oil paint

I have provided an account of the drying of oil paint and other relevant issues here. I draw attention to these words:
“It is true that thin layers of paint applied with ample solvent can become apparently touch dry quite quickly, particularly in the heat, but the oil paint in those layers requires months or years to undergo proper drying.”

Until the introduction of paints incorporating alkyd resins in the middle of the twentieth century, the sometimes attractive goal of making oil paint dry more quickly had only limited success. Detailed study of the methods employed is reported by Carlyle (2001), and reviewed practically by Gottsegen (2006, pp 73-125). Techniques included any combination of:

  • Addition of solvents such as turpentine, petrol, or oil of spike. These reduce the viscosity of the paint, making it runny, but evaporate over a period of hours. The paint film left after evaporation then usually has a semi-dry surface, although it actually dries no more quickly. This is commonly used for underpainting en plein air, as it enables further paint to be applied quickly. However the underpainting remains reworkable.
  • Addition of a resin such as damar or copal, usually dissolved in a solvent such as oil of spike. When the solvent evaporates the resin makes the remaining paint film very thick and appear almost dry. However the paint film does not actually dry any more quickly. Initially the film is more mechanically robust, thus suitable for overpainting as if wet-on-dry. However resins can leave weaker paint films in the longer term. I cannot find any mention of Géricault purchasing such resins in the records published by Rosenthal (1980).
  • Addition of a wax such as beeswax. Waxes do not normally undergo any form of drying and may inhibit proper drying of the linseed oil in the paint itself. However they increase viscosity of the paint and can make it more mechanically robust, allowing early wet-on-dry painting to succeed.
  • Catalysts, typically certain metal salts (at that time lead was still popular), which do actually accelerate the oxidation which occurs in drying oils such as linseed oil. These result in more rapid and true drying, but even so this will still take several days or weeks before even the surface skin of the paint film has dried. These metal salts can also result in long-term degradation of the paint film, including saponification, forming soap bubbles and further serious problems.
  • Pre-treatment of the drying oil binder to encourage the early stages of polymerisation. The most common such treatment is heating the oil, something which Géricault’s assistant may have done, but is not reported as doing so. This increases the viscosity of the oil but it will still take several days or weeks before the surface skin of the paint film dries. Overheating can cause degradation of the oil and consequent weakness in the paint film.

In the circumstances I consider it most likely that any siccative technique used could have included pre-treating the linseed oil by heating, which would have produced a thicker paint, and the addition of lead (and/or other siccative) metal salts.

Temperatures during the winter in Paris are quite cold, although I can find no evidence of an exceptionally mild or cold winter in 1818-9. Typically December-February average maximum and minimum temperatures remain below 10˚C, and it is not until May that average maximum rises to 20˚C.

In Géricault’s large studio in the Faubourg du Roule (to which he had moved to accommodate the huge canvas), temperature of the canvas would have remained within a very few degrees of that of the outside air. It is interesting that even modern acrylic paints are not recommended for use when temperatures fall below 10˚C because they do not polymerise properly; oil paints are more tolerant of low temperatures but dry very much more slowly.

The reason for Géricault wanting to use siccative or drying agents when painting alla prima (as distinct from using multiple layers) is obscure. The most common reason for taking such risks is when intending to paint in layers, wet-on-dry, to reduce the waiting time required for the next layer to be applied. It is odd that Montfort, Clément, Eitner and all those who have subsequently relied on their accounts have not noticed this puzzle nor tried to solve it.

Even if Géricault’s paint did dry each night so that a surface skin had formed, that would not prevent him from continuing previous work or adjusting it as he went along from day to day. Not only is it perfectly possible to scrape such drying paint back using a knife, but it is also straightforward to overpaint any corrections or adjustments on the forming skin.

Such alterations are frequently noted in paintings, and are widely known as pentimenti (singular pentimento). It is thus misleading to claim that “each section, once begun, had to be completed that day” (Barnes 2015, p 22) or that his technique “left little opportunity for second thoughts” (Christiansen 2000).

Assuming that Géricault painted seven days each week, during the seven months of December to June (Eitner estimates eight months of “painting almost without interruption from daybreak to dusk”, p 32), that provided a maximum of just over 200 days of painting available. His canvas is 491 by 716 mm in size, i.e. 351,556 square mm in area.

This would have required Géricault to have painted an average area of 1758 square mm each day, the equivalent of a square of just 4.2 x 4.2 cm in size. Thus the remarkable technical achievement is more in his ability to sustain this rate of painting over the period of seven to eight months, not in the average coverage which he achieved.

Théodore Géricault, The Raft of the Medusa (detail showing the deteriorating paint) (1818-9), oil on canvas, 491 x 716 cm, Musée du Louvre, Paris. Photo by Dennis Jarvis from Halifax, Canada, via Wikimedia Commons.
Théodore Géricault, The Raft of the Medusa (detail showing the deteriorating paint) (1818-9), oil on canvas, 491 x 716 cm, Musée du Louvre, Paris. Photo by Dennis Jarvis from Halifax, Canada, via Wikimedia Commons.

Use of bitumen

Although hardly ever used now because of its consequent problems, pigments containing asphalt, bitumen, and related substances – which include Cassel Earth, also listed by Jamar as being on Géricault’s palette (Clément 1879, pp 141-2) – were extensively used in paintings over the period 1500-1900 (Bothe 2007). They were also widely discussed in available manuals and handbooks (Carlyle 2001, for example), where reports of their problems were given.

Strangely, Eitner (p 37) states that “in his effort to obtain the most intense black possible, he lavishly used bitumen, which gave him a very satisfactory, glossy, raven black”, an error which has propagated back to Barnes and Wikipedia. All the technical documentation on asphalt, bitumen, Cassel Earth, and their relatives, gives their colour as being a very deep brown. Chromatically the peach and ivory blacks listed by Jamar as being in his palette would have given him the blackest black, particularly when finely ground.

Among others reported to have used these pigments are Velázquez, Rembrandt, Reynolds, Turner, Landseer, David, Delacroix, Courbet, and Ingres. However Bothe gives a caution which we should heed:
“Eyewitness accounts and journal entries, especially, must be treated with special caution, since it cannot always be assumed that what was called asphalt actually was asphalt. It should never be forgotten – especially with regard to the English and continental painters of the nineteenth century – that in addition to the true asphalts there were many substitute and adulterated materials in circulation.” (p 142.)

Objective testing of real asphalt-based pigments has shown that paints so produced may develop a hard surface skin which is prone to wrinkle and crack, while the paint inside remains relatively liquid. They are also quite sensitive to changes in temperature, liquefying in the heat. Furthermore, despite the common belief that they become darker with age, in fact their rich deep brown colour fades over time and becomes paler.

These pigments were also well known for resulting in slowed drying of the paint film, and when not prepared properly, may prevent it from ever drying properly.

Before any degradation in the paint film in Géricault’s work can be ascribed to the effects of bitumen, asphalt, Cassel Earth, siccatives, or any other ingredient, careful physical and careful examination would be necessary. Rolling large oil paintings, as was reportedly done to the canvas when it was moved for display in England, can also result in extensive damage to the paint film. Another widespread cause of liquefaction and loss of integrity is saponification, the formation of bubbles of soap within the paint layer.

Unfortunately I have been unable to find any published record of conservation analysis of The Raft of the Medusa, and given its importance, am surprised that Eitner chose not to include information from a detailed physical examination of the painting in his monograph.

Modern revision

The most recent account that I can find of The Raft of the Medusa and its historical interpretation is that of Athanassoglou-Kallmyer (2010). Sadly she does not give any account of the production of the final painting, but her silence ensures that she avoids the trap of hyperbole passed on from Montfort to Clément to Eitner to Barnes.

She considers an early proposal by Jules Michelet in 1846 that the painting was a metaphor for the French nation, and that Géricault had converted the original story of shipwreck into one of militancy. However her conclusion is that Géricault had adopted “the horror genre as both period style and modernist affirmation.” (p 149.)

I wonder whether Julian Barnes would wish to revise his essay on this painting in the light of any or all of the above.


The motif of this great and vast painting is the tale of a shipwreck. Given that Géricault’s painstaking research resulted in a painting which is not true to any moment in time, it is ironic that subsequent publications about the painting and its production appear not to be true to the painting itself.

Art history has here been shipwrecked on the rocks of multiple fallacies.

Clément, Eitner, and all who have subsequently relied on them should have been more critical of the sources of the information which they uncritically repeated, despite the problems inherent in them.

Anyone who writes a monograph about a painting should at the very least ensure that a skilled conservator examines the work, and should write consistently with those findings.

Art history desperately needs better understanding of the materials and techniques of art.

Just because you can give a reference does not make information or facts true.

Online information sources propagate and amplify errors, and do little or nothing to help a more critical approach.

Accuracy and wisdom grows from critical research, not from mere embellishment of what has been said before.


Athanassoglou-Kallmyer N (2010) Théodore Géricault, Phaidon Press. ISBN 978 0 7148 4400 8.
Barnes J (1989) A History of the World in 10½ Chapters, Jonathan Cape.
Barnes J (2015) Keeping an Eye Open. Essays on Art, Jonathan Cape.
Bothe CI (2007) Asphalt, chapter in Artists’ Pigments, A Handbook of Their History and Characteristics, volume 4, ed BH Berrie, Archetype Publications, pp. 111-149.
Carlyle L (2001) The Artist’s Assistant. Oil Painting Instruction Manuals and Handbooks in Britain 1800-1900 with Reference to Selected Eighteenth Century Sources, Archetype Publications.
Christiansen R (2000) The Victorian Visitors. Culture Shock in Nineteenth-Century Britain, Grove Press.
Clément C (1879) Géricault. Étude Biographique et Critique avec Le Catalogue Raisonné de l’Oeuvre du Maitre, 3rd edn, Didier, Paris.
Eitner L (1972) Géricault’s Raft of the Medusa, Phaidon Press.
Gottsegen MD (2006) The Painter’s Handbook, revised edition, Watson-Guptill Publications.
Rosenthal DA (1980) Géricault’s expenses for The Raft of the Medusa, The Art Bulletin vol 62 no 4, pp 638-640.