Finding Degas’s Lost Portrait With a Particle Accelerator

For decades, a mysterious black stain has been spreading across the face of an anonymous woman in Australia. She is the subject of a painting by Edgar Degas, the French Impressionist painter, and since the 1920s, the oil paints in her portrait have gradually faded, revealing the hints of another, hidden portrait underneath.

Until recently, attempts to capture the image underlying “Portrait of a Woman” with conventional X-ray and infrared techniques have only yielded the shadowy outline of another woman. In a study published on Thursday, however, a team of researchers reports that they have revealed the hidden layer underneath the painting, which hangs in the National Gallery of Victoria in Australia, at a very high resolution. It seems to be a portrait of Emma Dobigny, a model who was a favored subject of Degas.

“The fact that you get to see the lower image in such incredible detail is really exciting,” said Joris Dik, a professor of materials in art and archaeology at Delft University of Technology, who was not involved in the research.

To get their high-resolution image, the research team used a type of particle accelerator called a synchrotron. Synchrotrons are sources of extremely high-energy light. They work by directing that light, which is a million times brighter than the sun, into an X-ray beam that’s one tenth the diameter of a human hair.

Working at the Australian Synchrotron, the researchers used this intense X-ray beam to scan Degas’s painting with a method called X-ray fluorescence, or XRF. A special detector analyzed the painting point by point, to pick up the unique chemical signatures of different paint elements on a fine scale.

Over the course of about 30 hours, the researchers watched an elemental map of the hidden portrait emerge on a computer monitor.

“You’d never think a group of people would get so excited watching dots appear on a screen, but we were glued to the appearance of a face that nobody had seen for over 100 years,” said David Thurrowgood, who was a senior conservator at the National Gallery of Victoria at the time and an author of the new paper.

After the synchrotron scan, the researchers were left with a terabyte of X-ray data and a black-and-white map of Ms. Dobigny’s portrait. Then came months of data interpretation. The researchers wrote software that was able to convert different metal elements known to be found in different paint colors into a plausible color reconstruction of the painting. Mercury was assigned red, chromium green, cobalt blue, and so forth.

Because previous attempts at capturing the underpainting only produced vague traces, the researchers didn’t know what to expect. “I wouldn’t have in 100 years been able to make the connection that it was Emma,” said Daryl Howard, a scientist at the Australian Synchrotron and an author of the paper.

The underpainting resembled other portraits that Degas painted of Ms. Dobigny, particularly one from 1869 in which the model sports a similar hairstyle. Painting over old canvasses was a common practice for artists at the time, but letters from Degas suggest he had a special fondness for Ms. Dobigny, which might explain why he kept the otherwise unfinished or unsatisfactory painting for a decade or so before painting “Portrait of a Woman.”

Aside from revealing a lost painting, the project also reveals information about Degas’s painting techniques, which could help with identifying forgeries. “We now have a whole new level of information about how Degas actually worked,” Mr. Thurrowgood said.

Before this project, researchers had scanned several other paintings as proof of concept, but this was the first time the technique was used on a major painting with a high-resolution result.

The technique works better on some paintings than others, Dr. Dik said. But in this case he called the results remarkable.

“You get to see the personal characteristics — the painting style of Degas himself,” he said. “It gives you the feeling that you’re looking over his shoulders and watching him make the painting, which is of course long gone for the human eye. But thanks to this XRF work, it’s there again.”