In order to test which theory was correct, Sir Arthur Eddington photographed the 1919 solar eclipse from the island of Principe, off the west coast of Africa. Starlight from behind the Sun should bend around the Sun's mass, and be visible inside the edges of the eclipse, as to appear that they are between the Sun (and moon, during an eclipse) and the Earth. In Eddington's photographs, not only were stars visible within the field of the eclipse, but the positions of starlight seemingly validated Einstein's prediction. Identical observations were made in Brazil, and Einstein and his theory of relativity became world famous.
 |
| The negative of Eddington's photograph of the 1919 solar eclipse. Starlight is shifted inwards, around the sun and into the eclipse. |
While the theory of general relativity has since been accepted, many physicists and historians of recent decades have cast doubt on the validity of the 1919 experiment. One claim is of inaccuracy, that Eddington was simply lucky to get reasonably close values to one of the two theories being tested. The other claim is of a confirmation bias: Eddington and his collaborators throwing out data that appeared to support Newton rather than Einstein. Thus the 1919 eclipse is now referred to as a prime example of experimenters fitting their data to an expected result. I believe that while the 1919 measurements were not sufficient to rule out Newton's gravitation by themselves, Eddington had reasonable grounds to the claim that his data supported Einstein. I do not agree, however, that his data concretely confirmed Einstein's theory, as was the public mindset at the time. Eddington's observations were simply a contribution to the acceptance of general relativity.
Sources:
http://astro.berkeley.edu/~kalas/labs/documents/kennefick_phystoday_09.pdf
http://en.wikipedia.org/wiki/Tests_of_general_relativity#Deflection_of_light_by_the_Sun
http://en.wikipedia.org/wiki/Solar_eclipse_of_May_29,_1919
Brett, this is a really interesting test if general relativity and I think you do a great job explaining the strength of this experiment. While Eddington claimed his experiment completely supported Einstein's theory, I feel this test of general relativity is a great example of how confirmation bias is a clear obstacle to scientific data and how many scientists can allow their beliefs to interfere with their experiments. This photograph visual you have provided is also quite astonishing.
ReplyDeleteI had never heard of this experiment before, so it was very intriguing to read about. This brings up an interesting question about humans and wanting to prove things at all costs. It is hard for us to accept concepts without really seeing them and I'm sure Einstein's theories were just as hard to imagine to these scientists as they are to us today. The fitting of the data to support Einstein's theory is not the most scientific way to prove the point, but at the time, were there many more options to try to get people to understand how it all worked? I'd say that in some points of view, what they did was justified just so that they could wrap their heads around such a complicated mind puzzle.
ReplyDeleteTo argue against the claims of inaccuracy and bias, perhaps it can be assumed that Eddington was able to qualitatively evaluate which experiments were accurate an which weren't. If this is the case, then clearly the ones that supported Newtonian physics would have been inaccurate, and should therefore be thrown out.
ReplyDeleteA similar story can be written about Hubble. The observations Hubble made compared to the actual velocities that have been observed today are very different, yet the constant itself and understanding has remained very similar. This is mainly attributed to Hubble "fitting" his data to his theoretical constant.
Aside: Newtonian physics would not have predicted light to bend around massive objects. This is why General Relativity is so distinct. Newtonian models state that gravity only affects objects with mass. General relativity introduced space-time and therefore allowed light to be affected by gravity, a critical difference in the two models.