You Have Only Seen the Postcard

Géricault's Raft of the Medusa is 4.91 metres tall and 7.16 metres wide. You have almost certainly seen it. You have almost certainly not seen it.

The distinction matters more for this painting than for most, because its subject is a distance. A rescue ship appears on the horizon and does not come. Géricault chose that moment specifically, having considered and rejected the mutiny, the cannibalism and the actual rescue. What the picture is about is the gap between a straining arm in the foreground and a two-centimetre sail at the back.

On a screen, your eye crosses that gap instantly and for free. In the Salle Mollien you have to walk.

Théodore Géricault, Le Radeau de la Méduse, 1818 to 1819. Oil on canvas, 491 × 716 cm. Musée du Louvre, INV 4884. Public domain, via Wikimedia Commons.

Four works that reproduction cannot carry

Table 1. Four works that shrink badly and one that grows badly.

Las Meninas: the painting assigns you a place

Diego Velázquez, Las Meninas, 1656. Museo Nacional del Prado, P001174. Public domain, via Wikimedia Commons.

The Prado's own text calls it one of the most credible spaces depicted in western art, and a consideration on the act of seeing that makes the viewer reflect on the limits between painting and reality. The mirror on the back wall shows Philip IV and Mariana of Austria watching the scene.

The reading that the viewer therefore stands where the monarchs stand is scholarly rather than institutional, and it has a serious counter-argument, which is that the mirror may be reflecting the canvas rather than the couple. But whichever reading you take, both depend on the room being a room: figures near life size, several metres of depth receding behind them, and your own body at a particular distance. On a page, all of that collapses into "outside, looking at a rectangle."

The Orangerie: a device for removing the horizon

Monet gave the Nymphéas to the French state the day after the armistice of 11 November 1918, as a peace offering. They were installed in 1927, a few months after he died. Eight compositions, ninety-one metres of canvas in total across two oval rooms, oriented west to east along the path of the sun, lit from skylights.

The museum states his intention directly: the illusion of an endless whole, of a wave without horizon and without shore, so that the visitor could immerse themselves totally in the painting and forget the outside world.

There is no way to photograph that. A single panel cropped to a page is a landscape painting, which is exactly the category Monet built the rooms to escape. Reproduction converts an environment into a picture. The environment was the work.

Bayeux: a story you walk

Nearly seventy metres long and fifty centimetres high. The Bayeux Museum is precise about what it is, and it is not a tapestry: it is an embroidery in wool thread on linen cloth, in ten hues.

The aspect ratio is roughly 140 to 1. No page has that shape and no screen does either. Every reproduction takes one of two options: cut it into panels, which destroys continuity, or shrink it to a ribbon a few millimetres high, which destroys legibility. There is no third option. The object is a walk, and the sequence of events is metered by the viewer's own pace.

The best digitisation that exists is two files of sixty-four thousand pixels each. They are the closest thing to the real experience anyone has made, and on a page they still render as a thread.

The Lacemaker: the failure running the other way

Vermeer's Lacemaker is 24 by 21 centimetres. With its frame it is 68 by 62, so the object is mostly frame. It is smaller than the hands of the person looking at it, and the red and white threads spilling from the sewing cushion are close to the true size of real thread. It is, in a literal sense, a one-to-one object, and looking at it is an act of leaning in.

The best free reproduction of it runs to about 2,800 pixels across a painting 21 centimetres wide. That is roughly 340 pixels per centimetre, which is more resolution than the eye can use at reading distance. You can see more of those threads on a screen than any of Vermeer's contemporaries ever could.

Reproduction does not always lose information. It loses scale, which is a different thing, and here it loses it by inflating.

A jeweller's object becomes a mural. The intimacy is not reduced. It is inverted.

The raft, and the thing scale is carrying

Back to Géricault, because the history is the argument.

The frigate Méduse ran aground on the Arguin Bank off the west African coast on 2 July 1816. Her captain, Hugues Duroy de Chaumareys, had the command because he was a returned émigré whose royalist loyalty the Bourbon Restoration was rewarding, and he had hardly sailed in twenty years. A raft was built from the ship's own timbers and towed behind the lifeboats. The ropes were cut. The boats reached shore.

The numbers are worth stating carefully, because the received version is tidier than the record. The Louvre's own object entry says a hundred and fifty people were put on the raft, that it drifted for twelve days, and that fifteen were alive at the rescue. The scholarly literature says at least a hundred and forty-six men and one woman, thirteen days, fifteen alive, of whom four or five died shortly afterwards aboard the Argus. So the figure everybody repeats, fifteen survivors, is not the number of people who lived.

Two of them, the engineer Alexandre Corréard and the naval surgeon Henri Savigny, published their account in 1817. Géricault met them both. He also worked with the raft's own carpenter, Lavillette, to build an accurate scale model of it, which is why the planking in the painting shows the gaps between some of the boards.

His preparation has accumulated a lot of romantic narration, so here is what is documented rather than atmospheric. He sketched bodies in the morgue at the Hôpital Beaujon, studied the faces of dying patients, brought severed limbs to his studio to watch them decay, and kept a severed head borrowed from an asylum on the studio roof for a fortnight and drew it. Those studies exist as objects. Head of a Guillotined Man, 41 by 38 centimetres, is in the Art Institute of Chicago.

Théodore Géricault, Head of a Guillotined Man, 1818 to 1819. Oil on panel, 41 × 38 cm. Art Institute of Chicago, 1992.628. Public domain, via Wikimedia Commons.

He shaved his head and worked in near isolation from November 1818 to July 1819. His friend Montfort recorded that he painted straight onto the white canvas with no underdrawing, moving very little.

He also made a watercolour study of cannibalism, a figure gnawing the arm of a headless corpse. It is in the Louvre. It is not in the painting. He took it out.

At the 1819 Salon the classicists called it a pile of corpses. Louis XVIII, to the artist's face: your shipwreck is certainly no disaster for you. Jules Michelet read it as a national diagnosis, that our whole society is aboard the raft of the Medusa. Despite the director of museums pushing for it, no royal purchase came in Géricault's lifetime. The state bought it at the estate sale on 2 November 1824, for 6,005 francs, ten days before he had been dead a year.

Video. "Géricault, Raft of the Medusa," Smarthistory, recorded in front of the painting.

One more thing the reproduction hides

Géricault used bitumen in the shadows. It looks velvety and lustrous when fresh and then, over decades, darkens to something like black treacle while contracting into a wrinkled surface that cannot be renovated. Large areas of the painting are now difficult to read at all.

This was obvious early. In 1859 and 1860 the Louvre commissioned Guillemet and Ronjat to paint a full-size copy for loan exhibitions, because the original had already deteriorated too far to travel. Forty years after it was painted, the museum's answer to the condition problem was to make a second one.

So the photograph at the top of this article is, in one specific way, better than the object: it is colour-corrected and evenly lit, and it shows you things the canvas no longer shows. Reproduction does not simply lose. It trades, and it rarely declares the trade.

What I actually do with this

• State the intended size, always. Every image in a portfolio, a case study or a deck is being viewed at a size nobody specified. If the work was made for a 6-metre wall or a 24-centimetre panel, say so in the caption. It costs a line and it is the single most-deleted piece of information about any visual work.

• Put a human in the frame when scale is the point. I could not find a single freely licensed photograph of the Raft or of an Orangerie oval room with a person in shot. For works whose meaning is physical, the installation photograph with a body in it is the only honest reproduction, and it is the one nobody makes.

• Know which direction your medium lies in. Screens shrink the enormous and enlarge the tiny. Both are distortions and only the first one gets discussed. If you design at 200 percent zoom on a large monitor, you are doing to your own work what a postcard does to Géricault, in reverse.

A detail-by-detail walk across the raft itself, face by face, is on my Korean blog: 베이니월드.

Sources

Dimensions, inventory number, acquisition and the 1859 to 1860 copy: Musée du Louvre, INV 4884. The Lacemaker, including its framed dimensions: Musée du Louvre, MI 1448. Las Meninas: Museo Nacional del Prado, P001174. The Nymphéas, their length and Monet's stated intention: Musée de l'Orangerie. The Bayeux embroidery's dimensions and materials: Bayeux Museum. Corréard and Savigny's 1817 narrative is digitised in full by the BnF at Gallica. Head of a Guillotined Man: Art Institute of Chicago, 1992.628.

A note on where these images come from. The Louvre's own downloads are restricted to a listed set of non-commercial uses and route commercial licensing through RMN-GP; the Prado prohibits commercial use of its files outright; the Bayeux Museum asks that its digitisation not be used commercially and that extraction from its viewer not happen at all. The images here are the public-domain reproductions on Wikimedia Commons, which are not governed by those terms. Where a museum asks for a credit line even so, it seems right to carry it.

Not verified and therefore not asserted: that the figures in the Raft are life size, which is repeated everywhere and stated by no museum; the dimensions of the Orangerie's oval rooms; and the widely printed count of figures in the Bayeux embroidery, which the museum itself does not publish.