“Eddington had needed to make significant corrections to some of the measurements, for various technical reasons, and in the end decided to leave some of the Sobral data out of the calculation entirely. Many scientists were suspicious that he had cooked the books. Although the suspicion lingered for years in some quarters, in the end the results were confirmed at eclipse after eclipse with higher and higher precision.” -Peter Coles
If ever you attempt to come out with a new scientific theory, there are three criteria you must fulfill:
- You must reproduce all the successes of the old theory, the one you’re looking to replace.
- You must explain at least one observation or measurement, successfully, that the old theory failed to explain.
- You must make a new prediction, different from the old theory’s prediction, that you can go out and test.
When Einstein’s General Relativity first came out, it met those first two criteria, but the third proved exceedingly difficult.
The only practical test that people could come up with involved measuring the position of distant stars during the day: very close to the Sun. According to Einstein, the curved space around a large mass would bend the starlight, causing its position to shift in a measurable way. While VLBI radio observations can do this today, there was no such technology a century ago. And yet, a total solar eclipse made exactly those critical observations possible.