I’ve recently had to write an essay about whether or not I think Karl Popper got science ‘right’. This is a pretty open-ended question, and required what I think is an insane amount of reading. After all of it, I have decided that his deductive method provides a good foundation for science, but his ideas of falsification are flawed.

I should really establish what his ideal method of science was. Popper believed that science should be conducted in a deductive method: starting from generals and working towards particulars. He said you should start with a theory, make testable predictions, and then test them.

The predictions should also be falsifiable. This does not mean that the theory is false, but that the theory has predictions that can be tested. If the predictions do not agree with the tests, then the theory has been falsified and will be immediately rejected. If the prediction is observed then the theory is not proven true, but corroborated: it satisfies the criteria for it to stand until further evidence proves otherwise. He felt that science would not, and could not progress by verification, only by refutation.

He based his ideas partly on his hatred for Marxism, Stalinism and communism and partly on what he thought was an exemplary experiment conducted by Eddington in 1919. Communists believed that whatever theories they had about ruling the people were right, even if in the past they had been shown not to work. So if you apply Popper’s theory: their theories were falsified, but not rejected. He thought this was completely unacceptable, and politics and culture should not interfere with the scientific method.

1919 Solar Eclipse
Credit: Wikimedia Commons

But it is his basis on Eddington’s experiment that I am having a problem with. In 1919 Eddington went on an expedition to see if Einstein’s general relativity (GR) theory could be experimentally “proved”. Popper thought this was a great piece of science: Einstein had come up with a theory which had testable predictions, and Eddington went out to test them. Peter Coles (a previous lecturer of mine!) has written a great paper on the whole story.

The experiment was to test whether light would be bent by massive objects in space due to their gravitational pull. As a result, it would appear as if the source was displaced. Einstein came up with a value for the amount of displacement, so the experiment just had to see if they could come up with the same one. The issue was that to test this phenomenon you needed a solar eclipse.

So in 1919 Eddington set out to see if they could take some photographs of the displaced source during the solar eclipse. Unfortunately, for all but the last ten seconds of the eclipse, it was completely clouded over, making it impossible to get any good photographs. In the end, Eddington only managed to get two meaningful photographs.

With only two meaningful results, Eddington went to the Royal Society to present them. There was a significant amount of  ambivalence amongst those he presented to. According to Coles, “some questioned the reliability of statistical evidence from such a small number of stars.”

There was very little difference between the results collected and those predicted by both Einstein, and Newtons’ theory before him. But with popular backing from Eddington and J. J. Thomson, Einstein’s theory was accepted. Eddington was a big fan of Einstein and the GR theory, which could have made him biased, and clouded his judgement of the results.

Now, this does not sound like exemplary science to me. I was taught that I would need much more than two pieces of data in order to “prove” something. So should Popper have based his method on this piece of science? OK, so the method was what he believed in: theory, testable predictions, test predictions. But the tests didn’t show up great results, so really, Einstein’s theory was not fully “corroborated” in this instance. Eddington didn’t really produce an exemplary piece of scientific experiment. The experiment was repeated in 1922, and provided plenty of results about displaced stars, which would have corroborated Einstein’s GR theory nicely, but this was after Popper had said that this was a great piece of science.

So I think I should change the question to: did Popper only use this one example to support his claims about the nature of science? If so, then I think he was wrong to draw his conclusions as he did. If this was just one amongst many, then maybe his ideas stand up a bit better in general.

I haven’t come across any in my readings, but he probably didn’t base his ideal scientific method on just this experiment. But if people have, please let me know!

Go to the orginal article here or listen below