The scientific method: authoritative, impregnable. Even in the hands of humans

What gives science its authority is its ability to test ideas, in a way that can be repeated by other people, and based on prediction rather than on hindsight.

In a field like history, in which I worked for a while, one looks back on events and tries to come up with a plausible explanation for them. But plausibility isn’t proof, and there’s no clear way of distinguishing which of two or more plausible explanations is true (if any of them), which is why history is a subject of so much debate and revision.

For instance, today, on the 70th anniversary of the Dresden bombing, though we still feel horror at the number of deaths, we believe there were far fewer than Kurt Vonnegut mourned when he wrote Slaughterhouse Five. There are voices suggesting today that the bombing was even, perhaps, justified – including the voices of a number of survivors.

Dresden after the bombing of 13 February 1945

Science isn’t like that. A theory suggests that something must be a certain way, so we take a look – and it’s important that just anyone with the necessary skill and equipment can take a look – and if we find that things aren’t that way, the theory needs revising. If it is, we don’t necessarily accept the theory as true, but we feel we can perhaps keep using it as a helpful set of assumptions.

It’s true that there are some applications in the life sciences in particular, where we may be dealing with an individual’s reaction to a particular pathogen (or, to use a more technical term, a grubby little germ) and demanding that an observation be reproducible may be a tall order. However, generally, insisting on predictions that can be tested in reproducible experiments is a powerful methodology that has served us well.

One of the more famous such confirmations concerned Einstein’s General Theory of Relativity (not sure whether the capital letters are absolutely obligatory, but they somehow seem deserved). He predicted that light would be bent by gravity around really massive objects. So when light coming to us from a distant star has to travel close to the sun, it would be bent towards it, and the star would look further away from it than it should be.

In normal conditions, you can’t check that: the light from the sun completely drowns out any star shining behind it. But in a total eclipse, the sun’s light goes and stars apparently close to it can be seen. It ought therefore to be possible to see their apparent position.

Arthur Eddington
Confirmed Einstein's theory. Which he admired

In 1919, Arthur Eddington organised two expeditions to carry out the necessary observations during that year’s total eclipse, one in Brazil, the other in West Africa. Their measurement of the apparent displacement of the stars didn’t just to confirm that the phenomenon was happening, but that it was happening to the degree that Einstein’s calculations suggested.

Staggering proof of the power of Einstein’s theory, greeted with great acclaim around the world.

Except. The bending of visible light is slight. And during a total eclipse all sorts of strange currents get going in the atmosphere, creating all manner of problems seeing through it. Many people looking at Eddington’s photographs can only ask: how on Earth can you really assert anything from pictures as muddy and uncertain as these?

Eddington's picture of the 1919 total eclipse (negative and positive)
They support precise measurements, do they? Seriously? 

Interestingly, measurements carried out since by telescopes outside the Earth’s atmosphere, or on radio waves where the phenomenon is far greater, have all confirmed Eddington’s findings. So we can breathe easy (well, as easy as the weird picture of the universe that emerges from General Relativity allows). Still, it does leave a bit of a question, doesn’t it? Did Eddington really see what he claimed he’d seen? Or was it just too convenient?

The scientific method: rigorous, systematic, impartial. And authoritative as a result.

But it has to be applied by men and women. Who aren’t always quite as rigorous and systematic as one might hope. And who might, perhaps, be a little partial to one particular theory – such as Einstein’s.

Still, Eddington has ultimately been vindicated. So who cares whether his results were reliable or not? Doesn’t matter, does it? 

Or does it?