When light bent

Observation of the 1919 solar eclipse from Brazil and Africa provided the first experimental proof of Albert Einstein’s theory of relativity

Townsfolk gathered at the Patrocínio plaza in Sobral, northeastern Brazil, ahead of the eclipse

National Observatory Archive

No solar eclipse has been as momentous in the history of science as the one that occurred on May 29, 1919. That year, two separate expeditions of British astronomers were sent to photograph the eclipse and take measurements: one voyaged to Sobral, an inland town of northeastern Brazil, and the other to the island of Príncipe, then a Portuguese possession, off the coast of West Africa. Their mission was to determine whether starlight is bent as it traverses a region with a strong gravitational field, in this case the limb of the sun, and the angle of any detected deflection. Apart from possible surprises, they assumed one of three possible results would occur: either the path of light would be uninfluenced by gravitation; or it would deflect as predicted by calculations based on Isaac Newton’s (1643–1727) law of universal gravitation; or it would bend according to Albert Einstein’s (1879–1955) general theory of relativity, by approximately double the amount calculated from Newtonian mechanics. Six months later, photos and calculations published by the British astronomers proved Einstein right.

The expeditions provided the first experimental evidence for the general theory of relativity that Einstein had published four years prior, which suggested that matter and energy caused warps in the spacetime fabric and could also deflect the path of light traveling through it. In lending support to Einstein’s ideas, the results from the eclipse expeditions gave humanity a new understanding of the universe. They also helped to make the German physicist one of the most respected and celebrated scientists in the twentieth century.

Today, one hundred years after the 1919 eclipse, there is now a consensus in the scientific community that general relativity more accurately predicts the trajectory (deflection) of starlight than calculations based on Newton’s theory of gravity. For decades, however, astrophysicists, physicists, and historians of science debated whether the data from the 1919 observations were sufficiently robust to endorse Einstein’s ideas, as indeed they eventually proved to be. Some critics argued that the measurements had not been accurate enough to decide which of the two theories was right; others contended that British astronomer Arthur Stanley Eddington (1882–1944), then director of the University of Cambridge Observatory, who headed up the Príncipe expedition, had deliberately discarded data from the Sobral observations that appeared to support Newton’s theory. “Eddington was not only an enthusiast of Einstein’s ideas, but was keen to experimentally verify his theory as a gesture toward a reconciliation between the United Kingdom (UK) and Germany after World War I [1914–1918],” says physicist Luiz Nunes de Oliveira of the São Carlos Institute of Physics at the University of São Paulo (IFSC-USP). “But there is no evidence that the data was fudged.”

Irish astrophysicist and historian of science Daniel Kennefick, of the University of Arkansas, also dismisses claims that Eddington skewed the data in Einstein’s favor. “Not only was Eddington not in Sobral and therefore not personally involved in taking the measurements, but he also had no hand in analyzing the data from that end of the expedition. Those analyses were done by Frank Dyson [1868–1939] and his assistants at the Greenwich Observatory in London,” argues Kennefick, who is launching a book on the 100th anniversary of the eclipse.

Star fields—the name astronomers give to discrete areas of the sky populated by stars—are continually shifting in space. But the relative position between individual stars is always the same on a small time scale of, say, a few months. “If we take a photo today and another in three months’ time, the stars in a given field line up perfectly,” explains astronomer Augusto Damineli of USP. “But around a solar eclipse, the stars will appear to be slightly offset in relation to a photo of the same star field taken at night. The closer a star is to the sun, the more its light rays are bent during an eclipse.” This was the predicted-but-not-yet-experimentally-observed effect that the British expeditions were able to confirm.

In his book Opticks, first published in 1704, Newton also suggests that light is bent by gravity, but provides no calculations for the angle of deflection. According to Newton, gravity is a force acting between point masses that is proportional to their mass and inversely proportional to the square of their separation distance. In Newton’s time the nature of light was unknown. There were then two competing hypotheses: that light consisted of corpuscles (particles), or that it was a type of wave. Assuming light to be corpuscular, British astronomer John Michell (1724–1793) and French scientist Pierre-Simon Laplace (1749–1827) independently calculated the effects of gravity on light near the end of the eighteenth century. But during the course of the nineteenth century it was established that light was a form of electromagnetic wave. “When light came to be understood as a type of wave, rather than matter, it became completely uncertain whether it would be affected by gravity,” says Daniel Vanzella of the São Carlos Institute of Physics at USP (IFSC-USP). “That remained an open question for more than 100 years.”

Einstein began to make a name in the scientific community when in 1905 he introduced a new conception of space and time. “With the publication of his special theory of relativity, space and time ceased to be understood as absolute,” explains astronomer Reinaldo Ramos de Carvalho of the Brazilian National Institute for Space Research (INPE) in São José dos Campos. Einstein posited that space could deform, shrink, and even collapse, forming black holes, and that time could expand. However, the initial and incomplete version of his theory still yielded the same value for light deflection as Newtonian gravitation: 0.87 arcseconds. It was only after publishing his theory of general relativity, in 1915, that Einstein took his ideas a step further.

He proposed that gravity was not a force exerted between masses, as Newton described it, but rather the effect of a property of energy: that of deforming spacetime and everything that moves across it, even waves, such as light. “Space as described by Newton was flat. But in Einstein’s general relativity, spacetime is curved near bodies possessing significant energy or mass,” explains physicist George Matsas of the Institute for Theoretical Physics at São Paulo State University (IFT-UNESP). After factoring in the assumption of spacetime curvature, Einstein’s figure for light deflection virtually doubled to 1.75 arcseconds.

The world’s eyes on Sobral
When general relativity was unveiled, astronomers from around the world were eager to test the theory through observation of solar eclipses, which would provide the opportunity to photograph stars near the sun’s corona and determine whether their light would be deflected due to proximity to the sun. However, because of bad weather or difficulties stemming from World War I, none succeeded in obtaining data that could substantiate Einstein’s ideas until the eclipse of 1919.

In mid-1918, researchers at the Brazilian National Observatory in Rio de Janeiro, who were anticipating an eclipse the following year, determined that Sobral, a small town roughly 200 kilometers from Fortaleza, would provide optimal geographical conditions for observation. Astronomer Henrique Charles Morize (1860–1930), then director of the institution, prepared a detailed report on the region and sent it to scientific institutions around the world, including the Royal Astronomical Society, in London.

National Observatory Archive The 13-inch telescope used by the Sobral expedition to document the eclipseNational Observatory Archive

Frank Dyson, president of the Society, had been exposed to Einstein’s theories through Arthur Eddington, the institution’s secretary. Eddington was then a rising star in the European astronomical community, says historian Matthew Stanley, a professor in the Department of History of Science at Harvard. “His work in statistical cosmology had established his reputation as a creative and talented scientist, and his later work in stellar structure was a crucial element in the development of theoretical astrophysics as a field,” Stanley wrote in an article in the journal Isis in 2003. “Both Eddington and Dyson knew that the May 1919 eclipse would be special,” says Oliveira. “The sun would pass across a large cluster of stars in the constellation of Taurus, so there would be plenty of bright lights to observe.” The eclipse would provide a window of only a few minutes to photograph stars near the sun’s edge, 150 light-years away from Earth—a light-year equals 9.5 trillion kilometers.

Eyes on the sky
To determine which theory—Newton’s or Einstein’s—was correct, the Royal Astronomical Society organized expeditions to regions providing ideal observation conditions. Eddington led an expedition to the island of Príncipe, 300 kilometers off the coast of Africa. The other team, consisting of two members of the Greenwich Observatory—Charles Davidson and Andrew Crommelin—went to Sobral, with Dyson coordinating the expedition from overseas.

The Greenwich team arrived in Brazil on March 23, 1919. They disembarked at the port of Belém, Pará, where they awaited a few weeks as Henrique Morize, of the Brazilian National Observatory, made arrangements for their arrival in Sobral. By courtesy of the Brazilian government, their gear was waved through customs without inspection, as reported by the British researchers in an article later published in the Philosophical Transactions of the Royal Society.

Davidson and Crommelin brought two astrographic telescopes coupled to mirror systems known as coelostats, which are mounted such that they can track the sun’s movement across the sky and reflect the sun’s image back to the telescope. The main telescope, brought from the Royal Greenwich Observatory, offered a very wide field of vision, in theory allowing them to photograph a larger number of stars around the sun during the eclipse. It had a 13-inch aperture and was mounted to a 16-inch coelostat. A smaller telescope was borrowed from British Jesuit astronomer Aloysius Cortie (1859–1925) as a kind of backup, with a 4-inch aperture and 8-inch coelostat.

The scientists arrived in Sobral on April 30, 1919 and were welcomed by the then mayor, Jácome de Oliveira. “They then met Colonel Vicente Saboya, who offered the foreign visitors one of his houses,” says physicist Emerson Ferreira de Almeida of Vale do Acaraú State University, in Sobral. “The observations would be made at the town’s Jockey Club.” Two other expeditions with more modest equipment, one Brazilian and the other American, joined the English astronomers days later in Sobral, although their measurements were neither intended nor later used to verify the validity of Einstein’s theory of relativity.

Although a source of controversy, Eddington’s and Dyson’s conclusions were proven correct in later decades

Across the Atlantic, Eddington and his team had arrived at the port of Santo António in Príncipe on April 23, 1919. In their baggage they carried a telescope borrowed from the Cambridge Observatory, similar to the larger one sent to Sobral. The day of the eclipse was met with poor weather, and the overcast sky compromised the quality of the images. On some plates, the stars appeared clearer, while on others they disappeared in the cloudy sky. “That day also dawned cloudy in Sobral,” says astronomer Carlos Veiga of the Center for Astronomy and Astrophysics at the National Observatory. “But the clouds gradually began to thin, and the sky cleared.” Shortly before 9:00 a.m., the moon’s disk began to slide over the sun’s, completely obscuring it within minutes. The eclipse lasted exactly 5 minutes and 13 seconds.

The Greenwich team would remain in Sobral until July to photograph the same star field at night without the influence of the Sun’s gravitational pull. Eddington returned from Príncipe to England ahead of the Sobral team and produced images of the same star field in the Oxford sky, although the comparison plates would have been best taken at the site from which the eclipse plates had been captured.

Differing results
The astronomers produced three sets of photographic plates to measure the deflection of starlight near the sun’s limb. At Sobral, the main telescope recorded 12 stars and the backup telescope 7. The telescope used at Príncipe captured five stars. The plates from all three revealed some degree of deflection during the eclipse, confirming both Newton’s and Einstein’s ideas. But each of the three instruments captured different deflection figures, with different error margins. Two agreed with Einstein’s calculations; one was closer to the Newtonian prediction.

The most reliable calculations were derived from the clearest images of the eclipse—ironically, these were obtained with the smaller telescope at Sobral. Back in the UK, the team analyzed the plates and calculated the deflection to be 1.98 arcseconds (with 0.12 arcseconds of error), more than Einstein’s figure. All images produced by the larger telescope at Sobral were blurred or out of focus. “This may have been caused by the effect of the sun’s heat on the mirror array,” suggests USP physicist Ramachrisna Teixeira. The Sobral team was still able to analyze these poorer-quality plates and arrived at a deflection of 0.86, consistent with predictions based on Newton’s law of gravity. However, the poor quality of the images led the British astronomers to discount the larger telescope’s deflection values from their final analysis.

National Observatory Archive Photographic plates produced by the Brazilian team for spectroscopic observations of the sun’s coronaNational Observatory Archive

At Príncipe, due to bad weather, the images of many stars were either lost in the diffuse halo created by the Sun’s light, or covered by the Moon’s disk. Atmospheric turbulence further compromised the quality of the photographic plates. Despite their suboptimal conditions, Eddington was able to analyze the eclipse plates and compare them with check plates he took of the same star field in Oxford. The result was a deflection of 1.61 arcseconds, with a margin of error of 0.30 arcseconds, slightly lower than Einstein’s prediction. “The greatest weight must be attached to those obtained with the 4-inch lens at Sobral. From the superiority of the images and the larger scale of the photographs it was recognized that these [results] would prove to be the most trustworthy,” Dyson, Eddington, and Davidson announced in a written statement during a meeting of the Royal Astronomical Society in London, on November 6, 1919, declaring that Einstein’s prediction had been confirmed.

While their findings became a source of controversy, Dyson’s and Eddington’s conclusions were ultimately proven correct. Several other eclipses were observed over the following decades, and the resulting measurements consistently pointed to a deflection close to Einstein’s. Confirmation of his theory helped open new and wide avenues of research in fields such as physics, astronomy, and cosmology. “The German physicist’s ideas found especially fertile ground in Soviet physicist Alexander Friedmann [1888–1925], who, building on Einstein’s theory, proposed that galaxies were moving away from us because spacetime, that is, the universe, was expanding,” says Carvalho.

General relativity also provided the groundwork for important concepts in astrophysics, including black holes (extremely compact regions in spacetime where gravity is so strong that not even light can escape it) and gravitational waves—disturbances in the curvature of spacetime that propagate as waves. Gravitational waves would only be confirmed in early 2016.