Total eclipse — Eddington proves Einstein
During the total eclipse of 29 May 1919, astronomers measured how much starlight bent as it passed the covered Sun, giving the first test of Einstein’s general relativity. The announcement made Einstein famous around the world.
📍 Visible from where you are right now — computed in your browser
Available views: Orbit ✓ · Sky ✓
The story
On 29 May 1919, on Príncipe, a small island in the Gulf of Guinea off West Africa, the British astronomer Arthur Eddington was waiting for a total eclipse. Rain and wind had blown all morning, but at around 2 p.m., when the Moon covered the Sun, he was able to take photographs through gaps in the cloud. At the same time, across the Atlantic at Sobral in Brazil, a second team led by Andrew Crommelin and Charles Davidson photographed the same eclipse.
Both teams were aiming not at the Sun itself but at the stars right beside it. According to Einstein’s general theory of relativity, the heavy Sun bends the space around it, and starlight passing close by is bent as well. That should make stars next to the Sun look slightly pushed away from their usual places. Normally the Sun is far too bright for nearby stars to be seen, so the difference could only be measured during a total eclipse, when the Sun is covered.
The line-up that day suited the experiment well. The Sun stood in front of the Hyades, a star cluster in Taurus, so several fairly bright stars surrounded the covered Sun. The eclipse was also a long one, lasting up to 6 minutes 51 seconds at its longest point, which left time for photographs. The teams compared the positions of the stars on the eclipse plates with the positions of the same stars photographed in the night sky without the Sun.
After months of analysing the plates, the results agreed with Einstein’s prediction. They were announced at a joint meeting of the Royal Society and the Royal Astronomical Society in London on 6 November 1919, and newspapers gave the news prominent coverage. The orbit view reproduces the day’s line-up of Sun, Moon and Earth, and choosing Príncipe on 29 May 1919 in the sky view reproduces the sky with the Hyades beside the covered Sun.
Viewing guide — where, when, how
The path of totality in 1919 ran from northern South America across the Atlantic to Africa. At Sobral in Brazil the Sun was covered for about 5 minutes 13 seconds at around 9 a.m., and at Príncipe totality came at around 2 p.m., seen through gaps in the cloud. Two sites were chosen so that if bad weather spoiled one, the other might still deliver a result.
The observers used photographic plates, glass sheets that record light, and precise instruments for measuring star positions by hand. The same experiment can still be done today: photograph the stars beside the Sun during a total eclipse, photograph the same stars in the night sky some months later, and compare their positions. At the 2017 total eclipse in the United States, an amateur astronomer drew attention by repeating the measurement with a digital camera.
Setting the sky view to Príncipe (São Tomé and Príncipe) at around 14:00 Universal Time on 29 May 1919 reproduces the sky of that day. The Hyades cluster and the bright star Aldebaran can be seen beside the covered Sun. Switching the location to Sobral reproduces the same eclipse happening in the morning sky.
See this for yourself — open that sky at that moment · replay the moment in the orbit view
Why this eclipse is special
The two theories made clearly different predictions. Einstein’s theory said starlight grazing the edge of the Sun would bend by 1.75 arcseconds, while Newton’s theory of gravity gave about half that, 0.87 arcseconds. An arcsecond is a tiny angle: 1.75 arcseconds is about the size of a coin 2 centimetres across seen from roughly 2.4 kilometres away. Measuring it demanded great precision.
The results were 1.98 arcseconds (give or take 0.12) from a small 4-inch (10 cm) telescope at Sobral, and 1.61 arcseconds (give or take 0.30) from Príncipe. Both were closer to Einstein’s value than to Newton’s. Plates from the larger telescope at Sobral came out blurred and were left out of the conclusion, a choice that later became the subject of debate.
That debate was later examined again. In 1979 the original plates were remeasured with modern equipment at the Royal Greenwich Observatory, and the verdict in Einstein’s favour held. Since then radio telescopes and the Gaia space telescope, among others, have measured the Sun’s bending of light far more precisely than in 1919, confirming again and again that it matches general relativity.
Background and earlier eclipses
Einstein’s predicted value was not 1.75 arcseconds from the start. In 1911 he calculated about 0.83 arcseconds, close to what a Newton-style calculation gives. Only after completing general relativity in 1915 did the value roughly double to 1.75 arcseconds. In between, expeditions to Brazil in 1912 and to Crimea in 1914 came back without results because of weather and the outbreak of the First World War. Had they succeeded, the earlier, wrong value would have been tested first.
Eddington was a Quaker who opposed war and sought to refuse military service. Frank Dyson, the Astronomer Royal, argued that leading this eclipse expedition was scientific work of national importance, and secured Eddington’s exemption from conscription. Coming just after the war, a British scientist confirming the theory of a German-born scientist was also seen as a sign that science crosses national borders.
Eclipses repeat in a similar pattern about every 18 years and 11 days, a cycle called the saros. The 1919 eclipse belongs to Saros 136, a series that produced several very long total eclipses in the twentieth century: 8 June 1937 (7 minutes 4 seconds), 20 June 1955 (7 minutes 8 seconds) and 30 June 1973 (7 minutes 4 seconds) all belong to it.
FAQ
What did they measure?
How far the stars beside the covered Sun appeared to shift from their usual positions. Einstein’s theory predicted 1.75 arcseconds; Newton’s predicted about half that, 0.87.
What were the results?
1.98 arcseconds from the small telescope at Sobral and 1.61 at Príncipe, both closer to Einstein’s prediction.
What happened to the claim that data were cherry-picked?
There was debate over leaving out the blurred plates from Sobral’s larger telescope, but a 1979 remeasurement of the original plates with modern equipment upheld the conclusion. Radio telescopes and Gaia later confirmed the effect far more precisely.
Why did the Hyades matter?
Measuring the bending needs several bright stars right beside the Sun. On 29 May 1919 the Sun happened to be in front of the Hyades cluster, giving plenty of stars to measure.
Related
- The 1999 European total eclipse
- The 2027 total eclipse — six minutes of darkness
- The 2009 total eclipse — longest of the century
- The Sun — object index
- Aldebaran — object index
Times and figures are approximate — verify with official sources when planning travel. Simulation uses J2000 mean elements.