Skip to Main Content
Skip Nav Destination

A record-setting Eddington experiment

21 August 2018

An amateur astronomer improved on the famous 1919 experiment after collecting data during the Great American Eclipse.

Stars tracked near the eclipsed Sun
By tracking the positions of the 20 circled stars, amateur astronomer Donald Bruns was able to improve on Arthur Eddington’s century-old experiment to test general relativity. Credit: Donald Bruns

A year ago today, an estimated 216 million adults watched, either in person or on a screen, the total solar eclipse that swept across the continental US. One of them was San Diego–based amateur astronomer Donald Bruns. Executing the plan he had been perfecting for nearly two years, he observed the Great American Eclipse from Casper, Wyoming, using a $4000 telescope and a $5000 camera to carefully track the positions of stars located near the face of the eclipsed Sun.

In a solo paper published earlier this year in Classical and Quantum Gravity, Bruns reports that his efforts paid off. His measurements of starlight deflection caused by the gravity well of the Sun match the theoretical prediction and carry an uncertainty of just 3%. It’s the most accurate and precise ground-based optical version of the Eddington experiment, the century-old solar eclipse measurement that bolstered Albert Einstein’s then-controversial general theory of relativity.

Bruns’s result certainly doesn’t qualify as a scientific breakthrough. Yet from a historical perspective, there’s something satisfying about somebody finally mastering one of the most famous experiments of the 20th century. And as the experiences of Bruns and other astronomers reveal, even today’s impressive telescope and camera technology offers no easy ticket to replicating the work of Arthur Eddington.

Eddington conducted his celebrated experiment with Astronomer Royal Frank Watson Dyson. Two teams traveled to Brazil and Principe, off the coast of western Africa, to measure starlight deflection during the total solar eclipse of 29 May 1919. If Einstein’s theory was right, then a star located on the edge of the solar disk would be deflected radially away from the Sun by about 1.75 arcseconds (the deflection drops off sharply for stars farther from the Sun); according to Newton’s laws, the deflection would be half that amount. The two teams ran into weather and equipment problems on eclipse day, but in the end they reported a deflection coefficient of between 1.80″ and 2.16″, which was more consistent with Einstein’s predictions than with Newton’s. Eddington and Dyson’s result vaulted Einstein to worldwide celebrity, but it also drew accusations of bias in the data analysis (see the article by Daniel Kennefick, Physics Today, March 2009, page 37).

Since then hundreds of experiments have upheld general relativity, including Eddington-like experiments in which radio telescopes precisely measured the deflection of light from distant quasars scooting past the Sun. But despite multiple tries during 20th-century eclipses, no astronomer using an optical telescope on the ground had managed to replicate the 1919 experiment and attain both an accurate and a precise result.

Last year’s eclipse, the first total solar eclipse to sweep across the entire continental US since the year before Eddington’s expedition, proved too enticing an opportunity for a handful of astronomers, including Bruns. From November 2015 until the big day, he often worked 7 days a week for 10 hours daily. He carefully researched telescopes and cameras and then persuaded manufacturers to lend him the equipment. He ended up with a 101-mm-aperture refractor telescope from Tele Vue Optics, chosen to minimize optical distortion, and a monochrome CCD camera from Finger Lakes Instrumentation that could be cooled to reduce thermal noise.

Donald Bruns
Donald Bruns prepares to capture more calibration images following totality on 21 August 2017. Credit: Steve Lang

Bruns acquired commercial software packages to operate the telescope and camera, pinpoint the centroid of each star, and apply optical distortion corrections. He wrote scripts to perform the desired exposures during the eclipse and to analyze the data. He simulated the experiment by observing clusters of stars at night and at twilight, in brightness conditions close to those expected during totality. Three months before the eclipse, he picked out his observing spot on a mountain just south of Casper and then chose a “backup site, a backup backup site, and a backup backup backup site,” he says. Three days before totality, he bolted the telescope tripod to a concrete base to avoid any vibrations.

The weather turned out perfect on 21 August 2017, and Bruns’s equipment and software operated according to plan. He acquired his first images a minute before totality, capturing a swath of sky to the right of the Sun for calibration purposes. Just over half a minute into totality, the telescope swerved toward the Sun, and then, nearly a minute later, it moved farther to the left for another set of calibration images.

Those calibration exposures proved crucial to the analysis, Bruns says. He compared the observed positions of the stars to the far left and far right of the Sun with their accepted positions from a 2017 star catalog. (Consider that Eddington and Dyson had to take their own measurements of the reference positions of the stars, which necessitated their making observations months before and after the eclipse.) By factoring in position errors on both sides of the eclipse area, Bruns could better pinpoint the all-important stars in between. After comparing the observed and accepted locations of 20 stars between 1.5 solar radii and 5 solar radii from the eclipsed Sun, Bruns came up with a deflection coefficient of between 1.69″ and 1.81″. The final weighted average based on all the measurements was 1.7512″, almost exactly the theoretical prediction.

“It’s just an astounding result,” says Toby Dittrich, an astronomer at Portland Community College in Oregon who co-led another Eddington effort.

Not everyone was as fortunate as Bruns. Louisiana State University astronomer Bradley Schaefer, who also observed the eclipse from Wyoming, was foiled by atmospheric distortion and other factors. In the end he couldn’t filter out all the noise and pin down a final result. “To say the least, it’s disappointing and startling,” he told attendees at the American Astronomical Society’s January 2018 meeting just outside Washington, DC. “But I can’t think of what we could have done better.”

Dittrich’s team also ran into problems. Observing in the desert of eastern Oregon, he and his students hooked up their telescope to a battery powered by a solar panel. The unconventional power system introduced noise into the images. Fortunately, project co-leader Richard Berry, the former editor of Astronomy magazine, also observed the eclipse from his personal observatory on the western side of the Cascades in Oregon and collected solid data that Dittrich and his students are still analyzing. The team plans to run the data first through Bruns’s software and then through a Python program that Berry and Dittrich and his students are still coding. “We want community college students to do better than anyone in history,” Dittrich says. “Other than Don, of course.”

Close Modal

or Create an Account

Close Modal
Close Modal