A ‘REVOLUTION IN SCIENCE’ 100 YEARS LATER
ROBYN ARIANRHOD looks back on the events that made Albert Einstein a global superstar.
IT IS NOW 100 years since Albert Einstein burst onto the global stage like a supernova. Strictly speaking, the anniversary of his spectacular public debut is this November, but the occasion that sparked it was an extraordinary British scientific expedition centred on the solar eclipse of 29 May 1919.
BY NOVEMBER, THE DATA had been analysed and the results were in: newspapers around the world announced a “revolution in science”, thanks to the British astronomers’ confirmation of Einstein’s prediction that light is bent by gravity.
This was a radical new idea in the early 20th century. It was known that light is electromagnetic radiation, not a material particle, and in Newton’s physics no mass means no gravitational interaction; ergo light should be unaffected by gravity. Not surprisingly, it was Einstein himself who suggested using a solar eclipse to test his strange prediction, which he’d made with his new general theory of relativity.
The idea is that when the moon blocks out the sun and darkens the sky, “the fixed stars in the parts of the sky near the sun are visible … and may be compared with experience”, as he put it.
He meant that to see whether or not the light from those stars bends as it passes through the sun’s strong gravitational field and then reaches us on Earth, astronomers could take photographs of the apparent positions of the stars during the eclipse. These could then be compared with photographs taken several months
earlier or later, when the sun would be in a different part of the sky and so the light from the photographed stars would reach us in a direct line of sight.
By all accounts the 1919 eclipse expedition returned a stunning result. In one fell swoop it appeared to confirm Einstein’s prediction and overturn Newton’s venerable theory of gravity.
It is not often that a new theory topples another so dramatically – especially a theory as successful as Newton’s, which had passed virtually every test for more than 200 years. In this case, though, the timing was especially poignant, because in that first year after the Great War, a British team had experimentally confirmed a German theory. Arthur Stanley Eddington, a leader in the eclipse expedition, and a Quaker, hoped that this would speed up reconciliation in a world weary of war but still seething with resentment between the former enemy nations.
But there’s a catch to this famous and thrilling story. The excited newspaper announcements of November 1919 made Einstein into the first scientific superstar, launching him on a trajectory that would so capture the public imagination that 90 years later he was named Time magazine’s Person of the Century.
It was a good choice to make in 1999, when Einstein’s theories had been tested many times to considerable precision, and his personal warmth and community engagement were legendary.
In 1919, however, things were not so clear, and the publicity surrounding the eclipse results inadvertently propagated a myth about science that has dogged it – and the expedition itself – ever since.
So here I want to do two things: to celebrate general relativity and the recent 29 May centenary of its first experimental hurdle, and to highlight what the 1919 eclipse experiment and its aftermath tell us about science – about how it is done, and what it really means to “successfully” test a theory.
THE GENERAL THEORY OF RELATIVITY – general relativity (or GR) for short – was published at the end of 1915. The First World War had interrupted communication between German and British scientists, but the Dutch physicist Willem de Sitter managed to send his friend Eddington a copy of Einstein’s new paper. Eddington was entranced by it.
GR is perhaps the most beautiful scientific theory of all time, and one of the most extraordinary creations of the human mind.
The special theory, which had come 10 years earlier, is also brilliant. It gave us E=mc2, and a precise formulation of the concept that time and space have no universal existence; rather, they are artefacts of our consciousness, and of the way we choose to measure them. The special theory shows how different observers locate a physical event differently in time and space if each is moving relative to the other with a constant speed.
General relativity extends this to a theory about measuring the influence of gravity. This was no mean feat in 1915, before space travel had become possible, because gravity is always with us here on earth.
With other forces, such as electromagnetism, you can build up a theory based on experiments that show how one charged particle moves in the field of another, and so on. But you cannot readily isolate the gravitational effect of one body on another, because the force between everyday objects is negligible compared with the Earth’s gravitational field, which has the remarkable property that all material bodies are drawn downwards at the same rate. This of course is Galileo’s famous result.
Newton’s ingenious approach to the problem had been to look to the skies for inspiration. He had derived his “inverse square” law of gravity by using Kepler’s laws applied to the (roughly) circular orbits of moons. Then, by making his bold hypothesis that our moon, like every other earthly body, is falling towards (or around) the Earth, he was able to equate the circular acceleration of the moon with the known rate of gravitational acceleration on Earth.
His calculations showed “pretty nearly” that the Earth’s gravitational pull on the moon does indeed
depend on the masses of the moon and Earth and the inverse square of the distance between them.
He then showed that the planets are likewise pulled around in their orbits by the sun’s gravity.
The words “roughly” and “pretty nearly” are important in the context of the 1919 tests. New theories must quantify the intuitive leaps that underlie them, but often there are technical limitations in the equipment used to test them. For instance, the difference between Newton’s moon calculation and his theoretical prediction was due largely to the inaccurate value then available for the lunar distance – and as it turns out, the inverse-square law of gravity is remarkably accurate when applied to most things in the solar system.
Indeed, in 1915 the only known discrepancy between Newton’s law and actual planetary motion was the slow precession of the perihelion of Mercury. Newton’s theory was out by a tiny 43 angular seconds a century. (An angular second, or arc second, is 1/3600th of a degree – a miniscule amount that will prove crucial in analysing the 1919 eclipse experiments.)
For this reason, as well as to fit the theory of gravity into the kind of relativistic framework that had worked so well in his special theory, Einstein approached the problem in a radically new way. His great insight was to recognise that we on Earth can, in fact, imaginatively “transform away” the effect of the Earth’s gravity by changing our everyday frame of reference. Our usual point of view is from a frame that is fixed firmly to the ground, but what if we are in free fall – in an elevator falling freely down an empty shaft, for example? Imagine that you are in this vertiginous situation, and then suppose you let go of a ball you were holding. A person standing on the outside and watching will see that both you and the ball (along with the lift) continue to fall towards the Earth at the same rate. That is how falling objects behave when we view them from terra firma. Relative to you in your falling elevator, however, the dropped ball stays motionless – just as if there were no gravity acting on it at all.
In other words, in the reference frame of the falling lift there appears to be zero gravity – just as we see nowadays in videos of astronauts floating freely inside their capsules.
What this “thought experiment” showed Einstein was that he could indeed devise a relativistic theory of gravity, since we are able to make – or unmake – a gravitational field simply by changing reference frames. And what this means is that the mathematical key to both relativity theories lies in knowing how to transform time and space coordinates from the point of view of one relatively moving observer to another.
It took Einstein 10 long years to find the right mathematical language for this task, and to extend special relativity to a general theory where the relative motion is not constant but accelerated (as it is when one is falling under the influence of gravity).
What he found was that while special relativity played out in “flat” space-time, the addition of gravity meant that space-time itself would have to curve. We are used to this idea now, through the famous metaphor of the way the fabric of a trampoline curves around a heavy bowling ball placed on it, but in 1915 the idea that gravity curves space-time was incredible. So incredible, in fact, that many physicists showed little interest in it.
This was despite the fact that GR accounted for those missing 43 arc seconds in the precession of Mercury’s perihelion – a result that had made Einstein dizzy with joy. Nevertheless, of those who did show interest in the new theory, some hoped to disprove rather than confirm it. Giving up Newton seemed a step too far.
One of these doubters was Sir Frank Dyson, Astronomer Royal. Einstein had first made a prediction of light bending in 1911, before he had finalised his theory. A number of eclipse expeditions that had subsequently been planned were abandoned because of the outbreak of war. Ironically, this was lucky for Einstein, because in 1916 he had improved his prediction using his new, finished theory.
His final estimate for the angle of light bending (or “deflection”) by the sun was tiny: 1.7 arc seconds
(or 1.75 with modern values of the relevant physical constants). But Dyson realised that the solar eclipse of May 1919 would be the perfect occasion to test it.
Two suitable locations were found where the eclipse would be total. Eddington and Edward Cottingham’s team went to a cocoa plantation at Roça Sundy on Principe (in West Africa, then under Portuguese colonial rule, now the tiny independent island nation São Tomé e Principe); Andrew Crommelin and Charles Davidson went to Sobral, in north-eastern Brazil.
Eddington, a distinguished Cambridge mathematician and astronomer, was one of the few people at the time who really understood GR – and critics would later say that he was so captivated by it that he was not an objective observer. As he pointed out in a 1919 article, though, his colleagues in the team that went to Brazil were “very sceptical as to Einstein’s theory, and were inclined to expect a result opposed to it”. Actually, Newton himself had speculated – in a throwaway remark at the end of his book on optics – that gravity would have a tiny bending effect on light, because he thought of light as being made of tiny material particles. You can see the way gravity curves the path of material objects by throwing a ball horizontally, and watching it arc down to the ground.
In GR, however, it was not Newtonian gravitational force between the sun and the light particles that pulled light towards the sun, but the curving space-time around the sun. In 1919, this seemed too outlandish to take seriously – and the formidable mathematics that underlies it didn’t help win converts to Einstein’s new theory.
So, the task the British eclipse teams had set for themselves was monumental: to collect the data that would help decide which theory was correct, Einstein’s or Newton’s.
ON 29 MAY 1919, the dense cluster of stars known as the Hyades was right near the sun – perfect for photographing during a total eclipse. However, the technical challenges involved in this task comprise a story in itself. The main equipment consisted of telescopes and astrographs, devices comprised of telescopes and photographic plates, with lenses especially designed to give a wide field of view.
The total eclipse was to last barely six minutes, during which time the astronomers had to be able to focus sufficiently to distinguish different stars, and to record good enough images to discern on the photographic plates position changes of around 1/50th millimetre. Six minutes may sound like a reasonable window of opportunity, but during that time the Earth itself will have moved noticeably because it is spinning on its axis at about 1600 kilometres per hour. You can see this effect at sunset as you watch the disc of the sun on the horizon disappear below the Earth in a couple of minutes.
The teams’ equipment needed to be able keep track with this terrestrial rotation so that the tiny distant stars remained in focus. To make this a little more concrete, the sun’s diameter is about half a degree or 1800 arc seconds, compared with Einstein’s tiny 1.7 arc seconds of light deflection. So it was very important to keep track with the moving Earth so as
not to blur the miniscule deviations Eddington was hoping to record – and that others were hoping could be ruled out.
And then there was the weather. As Eddington reported, it had been beautifully fine every day in Principe except for the day of the eclipse. Luckily, during the actual eclipse the clouds began to thin, and the astronomers managed to see and photograph a few stars.
At Sobral, the weather was better, but the team there had another problem. Crommelin set up and focused the main telescope the night before, but the excessive daytime heat caused the photographic plates, mirrors and telescope tubes to expand, which played havoc with his delicate calibrations.
For example, expansion would change the telescope’s focal lengths and the scale on the plates, so any recorded change in the stars’ positions during the eclipse would represent a different angular displacement from those calibrated on the check plates made the night before, and on the comparison photos taken when the sun was far away from the Hyades. This made it impossible to get an accurate value of the actual amount of light bending.
Moreover, weather changes can introduce changes in atmospheric refraction. Refraction is the bending of the path of a light ray when it passes through a denser
medium, and it is easy to notice its effects when you immerse something in water. It is more difficult to discern atmospheric refraction, but some physicists believed that if there were any light bending to be seen in the eclipse photos, it would be due to refraction, not to gravity.
Over the next few months, the astronomers sifted through this very limited data, painstakingly applying analysis techniques such as least squares to standardise errors. Eddington was jubilant about his team’s early results, which seemed clearly to favour Einstein, and by September rumours had filtered through to Einstein himself, via the Dutch physicist Hendrik Lorentz.
Einstein quickly telegraphed his reply: “Heartfelt thanks to you and Eddington.” Then he wrote to his mother, telling her the “joyous news” that “the British expeditions have actually proved the light deflection near the sun”. By November, the data analysis was complete, and very soon the world knew that Einstein’s theory had trumped Newton’s. The rest, they thought, would be history.
Back in early November 1919, however, when Dyson, Eddington and Crommelin first announced their results in London, the mixed response from scientists was at odds with the triumphalist announcements that soon followed in the press. In fact, doubts about the accuracy of the results, and
accusations of bias, continued to surface – not only at the time, but until recently.
These doubts and accusations hinge on the lack of data, and on the fact that the astronomers had obtained three different sets of results, only two of which were used in the final conclusion.
For Principe, a deflection of 1.61 ± 0.40 arc seconds was found; the ± term indicates the amount of uncertainty in the results, due to the kinds of errors mentioned above. For Sobral, photographs taken with a smaller back-up lens yielded a deflection of 1.98 ± 0.18 arc seconds. These results clearly confirm that the light rays from the stars were deflected or bent during the eclipse. But they show only a rough agreement with Einstein’s prediction of 1.75 arc seconds.
On the other hand, the images taken with the large astrograph at Sobral had given a value of only 0.93 arc seconds, far too low to give credible support to Einstein. Dyson, however, had decided that this third result should be discarded. Although the temperature had fallen somewhat during the eclipse, causing the plates to shrink again, the resulting images were blurred, and there was great uncertainty as to how the expansion and contraction had changed the scale.
By discarding the blurred Sobral data, Dyson had unwittingly left himself and Eddington open to accusations of bias – of selective interpretation of the results in order to prove Einstein right. In 1979, however, astronomers at the Royal Greenwich Observatory re-examined the discarded plates. Using new equipment and new data analysis software to tease out the true data hidden in the blurred images, the Greenwich team showed that when all the scale and other technical errors were sifted out, the data definitely favoured Einstein.
It was a vindication of Dyson’s decision that was brought to wider attention only in 2009, in a paper by Daniel Kennefick.
Such uncertainty in experimental data can fuel the kind of anti-scientific bias we’ve seen in the climate change debates. What laypeople also often misunderstand is that disagreements within the scientific community about the worth of an experimental test do not necessarily mean that science and scientists can’t be trusted. On the contrary, robust scientific scrutiny and debate are vital when new results are announced: back in 1919, most scientists knew that despite what the newspapers had trumpeted to the world, a new theory cannot overturn an existing one on the basis of a single test. Dyson himself planned a new expedition in time for the solar eclipse of 21 September 1922, but it was hampered by bad weather.
However, a remote corner of northwest Australia had a special part to play in re-testing Einstein’s prediction during the 1922 eclipse. A US team from the Lick Observatory had made measurements during a 1918 eclipse, but Lick director William Campbell was so concerned about accuracy that the results were not released.
However, his interim data – analysed with the help of a female assistant, Adelaide Hobe – did not appear to favour Einstein’s prediction. In September 1922, Campbell and some US colleagues joined a 20-person international team at Wallal Downs, a cattle station on the coast between Port Hedland and Broome. Alexander David Ross led the Australian contingent. (There were two other Australian expeditions, one in South Australia and one in Queensland.)
The teams had to deal with the same kinds of technical challenges as in 1919, as well as “flies and dust [that] brought their own crop of discomforts”, according to Campbell. (Fortunately, he also found that the hospitality and general interest shown the astronomers was better than he had experienced “in any other part of the world on any other occasion”.)
Not surprisingly, the 1922 measurements were no more accurate than those of 1919, but they did consist of far more data, and the best results gave an even better agreement with Einstein’s prediction. Nevertheless, all up the 1922 data were still too inaccurate and variable to give unequivocal support to Einstein’s theory, and photographic expeditions continued for the next half-century.
By the 1960s, astronomers began testing Einstein’s prediction using radio waves rather than light. Radio waves have the same electromagnetic nature and speed as light, but they can be measured whenever their source, such as a quasar, passes behind the sun; there’s no need to wait for a solar eclipse. In addition, they are not so vulnerable to atmospheric effects.
Radio interferometry thus gives much more
accurate data, with current errors of just ± 0.0001 arc seconds, and so far the measured values have differed from Einstein’s prediction by as little as 0.02%. This is several hundred times closer to the GR prediction than Eddington’s result, and with far more reliable data.
In fact, the agreement is so good that astronomers are using “gravitational lensing” as a tool for exploring deeper into the cosmos – in the hunt for exoplanets and the analysis of dark matter, for example. While an ordinary lens uses refraction to bend and focus light, astronomers use the gravitational bending of electromagnetic waves by known objects to detect radiation from sources too distant or hidden to see directly. A dramatic illustration is that stunning first direct image of a black hole, in which radiation from the accretion disk is eclipsed by the massive “hole”. The halo and shadow are circular no matter which direction we’re looking from, because gravity is so strong here that light paths are bent around the black hole, so we can “see” all sides.
ALL THIS NEW IMPROVED data shows two key things. First, that the 1919 tests were far from sufficient to have overthrown Newton and launched Einstein on his celebrity trajectory. Not only did they consist of just two independent experiments, at Sobral and Principe, they lacked both the technical accuracy and a precise enough agreement with the new theory. The media razzmatazz was premature – just as today we are too often bombarded with premature announcements of various scientific and medical “breakthroughs”.
What also gets lost in this kind of media hype is that no theory has the last word on reality. Eventually, GR will no doubt be adapted to give an even better fit with physical observations, and perhaps an even better explanatory foundation.
For now, though, it has passed all tests, and so the second thing that all this data shows is that Einstein really does deserve his superstar status, and that Eddington’s faith in him was not misplaced.
GR is a truly extraordinary theory. It still offers our best explanation and most accurate accounts of known gravitational effects, and it has predicted the existence of completely new phenomena. In addition to light bending, many other predictions have been confirmed, including the recent discovery of gravity waves, and the gravitational redshift.
This latter effect is due to the slowing down of time in a gravitational field; conversely, time speeds up in weaker fields, such as those experienced away from earth on the satellites used in GPS navigation. This is a different, and much larger, effect from special relativity’s time dilation due to the relative motion of a clock and its observer, and both effects are taken into account in using GPS navigation. Without these corrections, our GPS positions would be out by a kilometre within two hours, or more than 10 kilometres a day. General relativity also opened up the whole field of cosmology, and it will continue to make headlines for some time yet. But the seeds of our confidence in Einstein’s radical tour de force were planted when Eddington and his colleagues obtained the very first broad confirmation of a new GR prediction.
Despite the limitations of its results, the 1919 expedition was an amazing scientific undertaking, and its centenary certainly deserves to be celebrated.