Albert Einstein and Arthur Stanley Eddington are linked by one of the most celebrated experiments in the history of science. In May 1919, Eddington led a British expedition to photograph stars near the Sun during a total solar eclipse, aiming to test whether gravity bends light by the amount Einstein’s general theory of relativity predicted. The results, announced that November to packed audiences in London, made Einstein a global celebrity practically overnight and cemented general relativity as a serious replacement for Newtonian gravity. The story behind those results, though, is more complicated and more interesting than the simplified version most people hear.
What Einstein Predicted and Why It Mattered
By 1915, Einstein had completed his general theory of relativity, which reimagined gravity not as a force pulling objects toward each other but as a warping of space and time caused by mass. One testable consequence was that a massive body like the Sun should bend the path of light passing near it. Einstein calculated that a ray of light grazing the edge of the Sun would be deflected by about 1.75 arcseconds, a tiny angle but one within reach of the photographic techniques of the time.1Revista Brasileira de Ensino de Física. Shadow of the Moon and general relativity: Einstein, Dyson, Eddington and the 1919 light deflection
This was not the only prediction on the table. Newtonian gravity also allows for light bending if you treat a photon as a particle with some effective mass falling through a gravitational field. That Newtonian calculation, which Einstein himself had initially arrived at before completing general relativity, gives exactly half the value: about 0.875 arcseconds.2Revista Brasileira de Ensino de Física. Shadow of the Moon and general relativity: Einstein, Dyson, Eddington and the 1919 light deflection So the experiment was not simply a question of whether light bends near the Sun. It was a question of how much. A deflection near 0.87 arcseconds would favor Newton. A deflection near 1.75 arcseconds would favor Einstein. Anything in between, or anything wildly off, would leave the matter unsettled.
Why a Solar Eclipse
Stars are, of course, invisible during the daytime. The Sun’s glare drowns them out completely. To photograph stars that appear close to the Sun’s edge in the sky, you need the Moon to block the Sun’s light. A total solar eclipse creates those conditions for a few minutes. During totality, the sky darkens enough to photograph the star field surrounding the eclipsed Sun. Those photographs can then be compared to reference plates of the same star field taken months earlier when the Sun is elsewhere, and any shift in the stars’ apparent positions can be measured.
The total solar eclipse of May 29, 1919 was an especially promising opportunity. The Sun would be passing through the Hyades star cluster, a rich grouping of bright stars in Taurus. Having many bright stars close to the Sun’s edge meant more data points and a better chance of detecting the tiny deflections Einstein predicted.
Two Teams, Two Continents
The expedition was organized not by Eddington alone but by the British Astronomer Royal, Frank Watson Dyson, who recognized the 1919 eclipse as an ideal test. Two observation teams were dispatched. Eddington and his assistant Edwin Cottingham traveled to the island of Príncipe, off the west coast of Africa in the Gulf of Guinea. A second team, led by Andrew Crommelin and Charles Davidson, set up at Sobral in northern Brazil. The rationale was straightforward: two widely separated sites doubled the chances that at least one would have clear skies.
The political backdrop added a layer of drama. World War I had ended just six months earlier. Einstein was German. Eddington was British and a Quaker conscientious objector who had nearly been sent to a work camp. A British expedition setting out to confirm the theory of a German physicist was, in that climate, a symbolically loaded act. Some historians have argued that Dyson and Eddington saw the expedition partly as a gesture of scientific internationalism, a way to rebuild bridges across wartime divisions.3arXiv. The Impact Of World War I On Relativity: Part III, The Aftermath
What Happened on Eclipse Day
On May 29, 1919, the Sobral team had reasonably clear skies and exposed plates using two instruments: a larger astrographic telescope and a smaller four-inch lens. The astrographic telescope, however, suffered from a focus problem that degraded the images. The four-inch lens performed well, producing sharp star images on its plates.
Eddington’s team on Príncipe was less fortunate. Heavy rain fell that morning, and thick cloud lingered through much of totality. Most of the photographic plates were useless. Only two plates ultimately proved satisfactory for analysis.4Notes and Records. The 1919 eclipse results that verified general relativity and their later detractors: a story re-told Eddington reportedly said that the clouds parting just enough for those exposures was the greatest moment of his life, though the quote may be embellished.
The Numbers That Changed Physics
When the data were reduced and analyzed over the following months, three sets of results emerged from the two sites. The Sobral four-inch lens gave the most precise measurement: a deflection of 1.98 arcseconds at the Sun’s limb, with a probable error of about 0.12 arcseconds (or roughly 0.18 arcseconds as a standard deviation). This sat comfortably close to Einstein’s predicted 1.75 arcseconds and well above the Newtonian value of 0.87.5Notes and Records. The 1919 eclipse results that verified general relativity and their later detractors: a story re-told
Eddington’s Príncipe plates, based on just two usable images, yielded a deflection of 1.61 arcseconds, with a larger uncertainty of about 0.45 arcseconds (one standard deviation). Less precise, but still consistent with general relativity and inconsistent with the Newtonian prediction.6Notes and Records. The 1919 eclipse results that verified general relativity and their later detractors: a story re-told
Then there was the problematic third dataset: the Sobral astrographic telescope, the one with the focus issues. Its plates gave a deflection of about 0.93 arcseconds, close to the Newtonian prediction. Dyson and Eddington judged these plates unreliable because of the known optical problem and did not include them in the final reported result, describing the value as “discordant by an amount much beyond the limits of its accidental error.”7Notes and Records. The 1919 eclipse results that verified general relativity and their later detractors: a story re-told
On November 6, 1919, the results were presented at a joint meeting of the Royal Society and the Royal Astronomical Society in London. The announcement declared that starlight was deflected by an amount consistent with Einstein’s theory. Newspaper headlines worldwide turned Einstein into a household name practically overnight.
The Accusations of Cherry-Picking
Almost immediately, and periodically ever since, critics have questioned whether Eddington and Dyson handled the data fairly. The core complaint is that the expedition leaders discarded the one dataset (the Sobral astrographic plates) that happened to favor Newton, while keeping the two datasets that favored Einstein. Was this legitimate scientific judgment, or was it confirmation bias dressed up as analysis?
The criticism picked up force in 1930 when the astronomer Charles Lane Poor published a pointed attack, arguing that the non-radial components of the stellar displacements had been thrown out as “accidental errors” and that the remaining radial components were then forced into agreement with Einstein’s theory through assumptions baked into the reduction method.8Journal of the Optical Society of America. The Deflection of Light as Observed at Total Solar Eclipses Poor’s critique was sharp but did not gain lasting traction in the astronomical community. Later reanalyses of the same data, using modern techniques, have generally confirmed that the Sobral astrographic plates were genuinely degraded and that excluding them was defensible. The four-inch result alone was strong enough to favor general relativity over the Newtonian prediction.
That said, the honest assessment is that the 1919 data, taken on their own, were suggestive rather than airtight. The Príncipe result had large error bars. The best Sobral result overshot Einstein’s prediction somewhat. And the handling of the astrographic plates left a permanent opening for accusations of bias, even if those accusations do not hold up well under scrutiny. The experiment’s historical importance outstripped its precision, which is not unusual for pioneering measurements. What mattered in 1919 was that the results clearly favored the general relativistic value over the Newtonian one. The fine-grained confirmation came later.
Confirmation by Later Eclipse Expeditions
Eddington and Dyson’s result did not stand alone for long. The scientific community recognized that more measurements were needed, and subsequent eclipse expeditions set out to repeat the test. One of the most significant early follow-ups was the 1922 expedition to Wallal, in remote Western Australia, which was specifically organized to measure the deflection of starlight during the total solar eclipse of September 21, 1922.9Publications of the Astronomical Society of Australia. Wallal: The 1922 Eclipse Expedition The Wallal results, produced with better equipment and clearer skies than either 1919 site, gave values in strong agreement with Einstein’s prediction and helped quiet many of the remaining skeptics in mainstream astronomy.
Eclipse measurements continued through the mid-twentieth century, each one getting somewhat more precise but still limited by the fundamental difficulty of photographing faint stars near the Sun for just a few minutes. The results consistently pointed toward the general relativistic value, though individual measurements varied. By the 1970s, the technique had largely been superseded by radio astronomy, which could make equivalent measurements without waiting for eclipses.
How Modern Technology Settled the Question
The gravitational deflection of light is no longer a debatable proposition. Radio interferometry, which measures the positions of distant radio sources (typically quasars) with extraordinary precision, has confirmed Einstein’s prediction to a fraction of a percent. A landmark study using very-long-baseline interferometry observations at two radio frequencies across multiple years found that the key parameter of general relativity governing light deflection agreed with Einstein’s prediction to within about two parts in a thousand.10Physical Review Letters. Measurement of the Solar Gravitational Deflection of Radio Waves Using Very-Long-Baseline Interferometry In general relativity, this parameter has a value of exactly one; the radio measurement returned 0.9996, with an uncertainty of just 0.0017. For all practical purposes, Einstein was exactly right.
The European Space Agency’s Hipparcos satellite, launched in 1989, also measured light deflection across the full sky, detecting it even far from the Sun where the effect is vanishingly small. Its successor mission, Gaia, has pushed positional astronomy to yet another level of precision. The accumulated evidence from a century of measurements using optical, radio, and space-based techniques leaves no room for doubt about the reality of gravitational light bending at the level Einstein predicted.
Eddington the Person
Arthur Eddington was not merely a convenient expedition leader. He was one of the few people in the English-speaking world who understood general relativity in its early years. The theory had been published during the war, and the disruption of normal scientific communication between Germany and Britain meant that most British scientists had not even seen Einstein’s papers. Eddington received a copy through the Dutch physicist Willem de Sitter, and he quickly grasped both the mathematics and the physical implications. By 1918, he had written the first English-language account of general relativity, bringing the theory to an audience that would otherwise have been largely unaware of it.
This made Eddington both the theory’s champion and its tester, which is part of what fueled the later accusations of bias. A researcher who has publicly advocated for a theory and then leads the experiment to test it is in an awkward position, even if the science is conducted honestly. Eddington was aware of this tension. By his own account, he expected the expedition to confirm Einstein, and he made no secret of it. Whether that expectation influenced his handling of the data has been argued about for over a century. The charitable reading is that a scientist who understood the theory deeply was in the best position to design and interpret the test. The uncharitable reading is that the outcome was never really in doubt from Eddington’s perspective. Both readings contain some truth.
Einstein Before and After 1919
For Einstein, the 1919 results transformed his career and his public life. Before the eclipse, he was well known within physics but virtually unknown outside it. He held an academic position in Berlin, had published the special and general theories of relativity, and was respected by his peers. After November 1919, he was the most famous scientist in the world, a status that brought opportunities and burdens in roughly equal measure.
It is worth noting that Einstein’s Nobel Prize, awarded in 1921 (though presented in 1922), was not for general relativity. The Nobel committee gave it for his discovery of the law of the photoelectric effect, a contribution to quantum physics.11arXiv. The Impact Of World War I On Relativity: Part III, The Aftermath General relativity was still considered somewhat controversial in some quarters, and the Nobel committee was cautious. The irony is that the photoelectric effect paper, while groundbreaking, was far less revolutionary than general relativity. The prize reflected the conservatism of the committee more than the relative importance of Einstein’s contributions.
Einstein himself had enormous confidence in general relativity well before the 1919 expedition. When asked what he would have done if Eddington’s results had contradicted his theory, he reportedly replied, “Then I would feel sorry for the dear Lord. The theory is correct.” The quip captures something real about the relationship between theory and experiment in fundamental physics. Einstein’s confidence came not from arrogance but from the mathematical beauty and internal consistency of the theory, which by 1919 had already explained a long-standing anomaly in the orbit of Mercury that Newtonian gravity could not account for.
Why the 1919 Story Still Gets Told
The Einstein-Eddington eclipse story has an almost mythological quality that keeps it alive in popular science. It has a clear prediction, a dramatic test with cloud-threatened observations on a remote tropical island, a decisive result, and instant global fame for the protagonist. It also sits at a hinge point in history, a moment when the world was recovering from a catastrophic war and hungry for stories about the power of human intellect to transcend national divisions.
The story’s staying power also owes something to the way it captures a genuine tension in how science works. The 1919 measurements were good enough to distinguish between Einstein’s prediction and Newton’s, but they were not precision instruments by modern standards. The decision to exclude the problematic Sobral astrographic data was almost certainly correct, but it was a judgment call made by people who had a stake in the outcome. Science relies on such judgment calls constantly, and they are usually invisible to the public. The 1919 expedition is one case where the judgment call is visible, debated, and still discussed, making it a useful case study for anyone interested in how scientific knowledge actually gets built.
Gravitational Lensing in Modern Astronomy
The bending of light by gravity, the phenomenon Eddington set out to measure, has become one of the most powerful tools in modern astronomy. Galaxies and galaxy clusters act as gravitational lenses, bending and magnifying the light of more distant objects behind them. This effect, which is just an extreme version of the deflection Einstein predicted near the Sun, has been used to detect dark matter, to study the distribution of mass in galaxy clusters, and to observe some of the most distant galaxies in the universe that would otherwise be too faint to see.
Strong gravitational lensing can produce multiple images of a single background object, or stretch it into dramatic arcs and rings. Weak lensing, where the distortion is too subtle to see in any individual galaxy but shows up statistically across large populations of galaxies, has become a primary method for mapping the large-scale structure of the universe. The tiny effect that Eddington measured with clouded-over photographic plates on a West African island in 1919 turns out to be one of the pillars on which twenty-first-century cosmology rests. The angular deflection near the Sun is measured in arcseconds; the deflections caused by galaxy clusters can be measured in arc-minutes or more, and the science behind them is the same physics Einstein put on paper over a century ago.