The Eclipse That Made Einstein Famous

On 12 August, the Moon will cover most of the Sun across Britain. In 1919, another eclipse turned Einstein into a global celebrity and revealed how uncertain evidence becomes public authority.

Solar eclipse 2026: UK viewing guide

On Wednesday 12 August, the UK and Ireland will see their deepest partial solar eclipse since 1999. The Sun will be about 90 per cent covered from London, roughly 95 per cent from Cornwall and almost 97.5 per cent from the south-west of Ireland. Totality will cross parts of Greenland, Iceland and Spain, lasting up to 2 minutes and 18 seconds.

All times are local British Summer Time. The eclipse will be low in the western sky, so find an open view of the horizon. Never look directly at the Sun. Regular sunglasses are not safe. Use undamaged eclipse glasses marked ISO 12312-2, a properly filtered solar telescope or an indirect method such as a pinhole projector. The Royal Observatory Greenwich has full viewing guidance and will stream the event if clouds intervene.

At 7.12pm on 12 August, the Sun above London will have been reduced to a narrow crescent. Cornwall will be closer still to totality. The temperature may dip, shadows will sharpen and the late-afternoon light will begin to feel strangely thin.

It will remain dangerous to look at. Even when nine-tenths of the Sun has disappeared, the uncovered portion is bright enough to damage an unprotected eye. Nor will Britain fall into the sudden darkness experienced inside the path of totality. The spectacle will be incomplete, but unusually close.

For most people watching, the eclipse will be the event. In 1919, darkness was part of the apparatus.

During a few minutes of totality, British astronomers tried to measure whether distant stars appeared to move when their light passed close to the Sun. The shift they sought was far too small for a person to see. It had to be extracted from glass photographic plates, compared with reference images and separated from changes in focus, temperature and scale.

The result was neither as clean nor as final as the legend suggests. Yet when it was announced in London that November, newspapers declared a revolution in science. Albert Einstein, already respected in specialist circles but little known to the wider public, became an international celebrity almost overnight.

The eclipse did not make general relativity true. It made the theory publicly legible.

That distinction explains why some discoveries travel far beyond the laboratory while others remain obscure. Evidence can change scientific knowledge. To change public belief, it must also survive a journey through instruments, experts, institutions and explanation. The 1919 eclipse gave that journey a stage large enough for the world to see.

The theory waiting for darkness

Einstein completed the general theory of relativity in 1915. Its central idea replaced gravity as an invisible force pulling objects across fixed space with something more unsettling: mass and energy curve spacetime itself, and objects move through that curved geometry.

Light should follow it too.

A ray passing close to the Sun would therefore bend slightly. From Earth, the star that emitted it would appear to sit a little farther from the Sun than it ordinarily did. Einstein calculated that light grazing the solar edge should be deflected by about 1.75 seconds of arc. A calculation using a Newtonian-style treatment produced roughly half that value, about 0.87 seconds. No measurable movement would support neither.

This gave astronomers an unusually sharp test. The trouble was seeing it.

The Sun was the massive object needed to create a measurable deflection, but its glare concealed the background stars. At night the stars became visible and the Sun was elsewhere. Only a total eclipse could place both in the same photograph, using the Moon as a temporary shutter.

Earlier attempts had failed. In 1914, the German astronomer Erwin Finlay-Freundlich travelled to Crimea to observe an eclipse, only for the First World War to begin. He was interned by Russian authorities and his equipment was confiscated. Had the expedition succeeded, it would have tested an earlier calculation that Einstein later corrected.

By 1919, the prediction had changed and the sky had supplied an exceptional opportunity. On 29 May, the eclipsed Sun would sit in front of the Hyades, a field containing several relatively bright stars. Totality could last almost seven minutes along the eclipse path, long enough to expose a series of plates.

The theory had already achieved something important. It accounted for a small anomaly in the orbit of Mercury that had resisted a satisfactory explanation within Newtonian gravity.

Mercury's orbit, however, was difficult to turn into a public event. An eclipse was different. It offered a visible contest with a date, a setting and apparently distinct outcomes. Nature would close the curtain on the Sun and reveal which universe lay behind it.

Two expeditions, three answers

The project was organised by Frank Dyson, the Astronomer Royal, with the support of the Royal Society and Royal Astronomical Society. Arthur Eddington, director of the Cambridge Observatory, had become one of relativity's most capable advocates in Britain after the Dutch astronomer Willem de Sitter helped carry Einstein's work across the scientific isolation of wartime Europe.

The politics mattered. Einstein had worked in Berlin during the war. Eddington was a Quaker and conscientious objector who opposed the rupture of scientific relations with Germany. Historian Matthew Stanley has shown that Eddington regarded international science as one way to repair what nationalism had broken. A British test of a German-born physicist's theory, conducted months after the Armistice, could carry a meaning beyond gravity.

That did not make the experiment insincere. Stanley's study of the expedition finds no reason to believe Eddington or his colleagues were anything other than rigorous. It does mean that the test arrived with moral and institutional significance already attached to it.

Two teams left Britain in March. Eddington and the clockmaker Edwin Cottingham travelled to Príncipe, an island off the west coast of Africa. Charles Davidson and Andrew Crommelin of the Royal Observatory went to Sobral in north-eastern Brazil. Dyson remained in London to coordinate the work and later supervised much of the analysis.

On Príncipe, the morning of 29 May brought heavy rain. Clouds persisted into the eclipse. Eddington managed to photograph stars on several plates as the cloud thinned, but only two were judged good enough for the final measurement.

Sobral had clearer skies and two telescopes. The larger astrographic instrument was meant to provide the strongest evidence. During the eclipse, however, its star images became blurred and out of focus, probably because heat affected the mirror feeding light into the telescope. A smaller four-inch backup performed much better and produced seven useful plates.

The expeditions had returned with three data sets, not one decisive picture.

The smaller Sobral telescope produced a deflection of 1.98 seconds of arc. The two usable Príncipe plates gave 1.61 seconds. Allowing for their uncertainties, both were consistent with Einstein's prediction of 1.75 and the Sobral result was strongly inconsistent with the smaller Newtonian-style value.

The troubled Sobral astrograph was another matter. Depending on how the change in its image scale was treated, it implied either about 0.93 seconds, close to the classical value, or 1.52 seconds, closer to Einstein. The team assigned the set little weight because the eclipse plates and comparison plates were visibly different in quality. It was excluded from the final conclusion.

This decision later became the basis of a powerful accusation. Critics argued that Eddington, already convinced by the beauty of relativity and sympathetic to Einstein, had discarded the result that did not fit.

The surviving record tells a more complicated story. The Sobral data were reduced at Greenwich under Dyson, not by Eddington. The focus problem had been noticed when plates were developed at the site, before anyone in Britain knew the numerical answer. The published report discussed the bad images and competing analyses rather than concealing them. A 1979 remeasurement using modern equipment later found 1.55 seconds for the disputed plates and reproduced the strong result from the four-inch telescope.

The evidence was imperfect. It was not casually rigged.

A revolution announced

On 6 November 1919, the results were presented at a joint meeting of the Royal Society and Royal Astronomical Society. The room was full. J. J. Thomson, the Royal Society's president and the physicist credited with discovering the electron, called the evidence the most important result in gravitation since Newton.

The following morning, The Times placed the announcement beneath a headline declaring a revolution in science and the overthrow of Newtonian ideas. In New York, readers were soon told that the lights were “all askew in the heavens”.

The phrasing was more confident than the plates. It was also irresistible.

The public was offered a contest between two recognisable figures. Newton represented the ordered universe that had stood for more than two centuries. Einstein was the obscure continental genius who claimed that space and time were not fixed after all. A rare celestial event had allowed British astronomers to decide between them. A German idea had been validated by former wartime opponents. The new theory was said to be so difficult that only a handful of people could understand it.

Each element enlarged the story. Together they created a modern scientific celebrity.

Einstein was 40. His 1905 work on relativity, Brownian motion and the photoelectric effect had already established him among leading physicists, and general relativity had attracted serious attention before the eclipse. What changed in November 1919 was his scale of recognition. He ceased to be merely a scientist with a formidable theory and became “Einstein”, a public symbol of genius.

Historian Alistair Sponsel has documented how Eddington, Dyson and their colleagues spent months explaining the planned experiment, answering objections and keeping journalists informed. The Times had covered the expedition before the teams departed, while they were abroad and as the plates were being measured. By November, its readers already knew that a verdict was coming.

This was not the invention of a result. It was the construction of an audience capable of recognising one.

How evidence becomes authority

The eclipse shows that scientific evidence and scientific authority are related but different things.

The evidence consisted of tiny differences in the recorded positions of stars, together with judgements about focus, image scale and error. Only a small group of specialists could inspect the plates and reproduce the calculations. The public could not independently verify the result, and neither could most scientists outside the relevant fields.

They had to decide whose judgement to borrow.

That is not a defect peculiar to 1919. Modern knowledge depends on intellectual division of labour. A person can reasonably accept the existence of gravitational waves, the safety profile of a medicine or the age of a fossil without rebuilding the detector, rerunning the trial or dating the rock. The question is whether trust has been placed in a chain designed to expose mistakes.

In 1919, four parts of that chain aligned.

First, general relativity made a risky prediction in advance. It did not merely explain the plates after they appeared. The measured shift could have landed elsewhere.

Second, the eclipse turned an abstract disagreement into a legible test. Readers did not need to understand tensor calculus to grasp that stars should appear in different places.

Third, recognised institutions and practitioners stood behind the result. The Royal Society, Royal Astronomical Society, Royal Observatory, Dyson, Eddington and Thomson supplied accumulated credibility that no photograph possessed by itself.

Finally, the finding could be compressed into a story: Einstein against Newton, tested in the sky. That compression carried the evidence around the world, but it also stripped away much of the uncertainty.

Public authority is not the same as proof. It is a chain of warranted trust.

The quality of that chain depends on whether each link can be questioned. Were the predictions stated beforehand? Were the instruments understood? Were awkward results disclosed? Could other teams repeat the work? Did the institutions earn the confidence being placed in them?

The 1919 story survives these questions better than its most sceptical retellings suggest. It also survives less perfectly than the newspaper headlines implied.

The problem with one decisive experiment

There is a tempting cynical reading of Einstein's rise: the evidence was ambiguous, powerful insiders promoted the outcome and newspapers manufactured a genius.

That goes too far.

The best modern reassessments support the scientific integrity of the original analysis. Daniel Kennefick's examination of the archive found sound reasons for discounting the distorted Sobral images and showed that Eddington did not control that decision. A later statistical reanalysis by Gerry Gilmore and Gudrun Tausch-Pebody concluded that the two accepted data sets supported Einstein's prediction and that including the rejected set would not overturn the result.

Yet the opposite heroic account also goes too far. No single eclipse photograph proved the whole of general relativity or made Newton useless on 6 November. The Príncipe measurement had wide uncertainty. The main Sobral instrument malfunctioned. The observations tested one consequence of a much larger theory, and even a successful measurement could not eliminate every alternative explanation.

Further evidence mattered. Observations during the 1922 eclipse in Australia produced stronger support. Radio astronomers later measured the bending of signals from quasars with far greater precision. The European Space Agency's Hipparcos satellite confirmed the effect across tens of thousands of stars without waiting for darkness. Gravitational lensing eventually passed from being a test of relativity to a tool used to study galaxies, dark matter and the large-scale universe.

Newton survived too. General relativity showed that Newtonian gravity was incomplete, not worthless. For ordinary speeds and gravitational fields, Newton's equations remain an extraordinarily effective approximation. The world did not exchange one entirely false system for one entirely true system overnight.

A result can be scientifically serious and publicly oversimplified at the same time.

The dilemma is difficult to escape. Public understanding requires compression. Few people will follow a finding described only through instrument calibration, confidence intervals and rival models. Remove too much detail, however, and evidence begins to look like revelation: one image, one meeting, one verdict.

Science rarely moves that cleanly. Its authority is strongest when institutions can communicate a conclusion clearly without disguising the uncertainty that still surrounds it.

Where the shadow ends

The 2026 eclipse will resemble the event of 1919 in the sky, but not in scientific purpose.

No astronomer now needs the Moon to establish whether the Sun bends starlight. Radio telescopes and space observatories can measure the effect more precisely under less fragile conditions. The August eclipse is a spectacle, an educational opportunity and perhaps a setting for other observations, not a fresh courtroom for general relativity.

The information environment is different as well. In 1919, a small number of institutions and newspapers could turn a scientific announcement into a shared public event. Today, results move through journals, preprints, press offices, videos, social platforms and partisan communities at once. Authority is less concentrated, scrutiny can be faster and corrections can travel. So can confident errors.

The lesson is not that every discovery needs theatrical proof, or that expertise becomes legitimate only when the public can see an experiment. Much of the best science is cumulative, technical and visually unremarkable. A dramatic test can attract attention, but spectacle is not a substitute for method.

The eclipse reveals something narrower and more useful. When evidence must travel beyond the people who produced it, its reception depends on how well a society connects specialist judgement to public understanding. The strongest chain combines a test that could fail, methods open to scrutiny, institutions with earned credibility and an explanation that clarifies rather than distorts.

What to watch when the next breakthrough arrives