Free Radicals - Michael Brooks [30]
So disappointing, in fact, that Eddington decided to abandon his original notions of how to calculate the positions of the stars from the measurements on the plates. Instead, he formulated a new technique for the purpose. This new method involved a few assumptions and Einstein’s own numbers for what the displacement should be. Perhaps unsurprisingly, Eddington achieved a rather pleasing result: ‘the one good plate that I measured gave a result agreeing with Einstein and I think I have got a little confirmation from a second plate’.
The result was a displacement of 1.61 arc-seconds. Einstein’s theory – as Eddington knew – predicted 1.75 arc-seconds. Newton’s theory, the status quo, said that the displacement would be 0.8 arc-seconds. These are tiny shifts, roughly equivalent to the diameter of the smallest coin in your pocket seen from a mile away. But Eddington was happy to declare Einstein the winner. As he wrote the following year, ‘Although the material was very meagre compared with what had been hoped for, the writer (who it must be admitted was not altogether unbiassed) believed it convincing.’
But Eddington’s was not the only expedition to test general relativity at the 1919 eclipse. The Astronomer Royal had also sent a team to Sobral in the north-east of Brazil. They were equipped with an astrographic telescope, just like Eddington’s, and had enjoyed fine, clear weather, which had allowed them to come away with plenty of photographs. However, as it turned out, fine, clear weather brought its own disadvantage: the Brazilian heat had distorted a mirror used to focus starlight into the main telescope. As a result, the images obtained were slightly fuzzy. A note written on 30 May, after four of the plates had been developed, admits that ‘It seems doubtful whether much can be got from these plates.’
But a value was derived nonetheless: a deflection of 0.9 arc-seconds. This was far too low to confirm Einstein’s theory, and very close to the standard Newtonian explanation for the passage of light through the universe. Fortunately for Eddington, the Brazilian expedition had taken along another, smaller telescope. When the images obtained with this instrument were analysed, a deflection of 1.98 arc-seconds was found.
In their book on science and its methods, The Golem, Harry Collins and Trevor Pinch demonstrate that a modern analysis, one that takes into account all the experimental results, tells us that no conclusion can be drawn from the 1919 British eclipse data. The eight ‘good’ Sobral plates yield a displacement of just over 1.7 arc-seconds; the two ‘poor’ Principe plates give a value between 0.9 and 2.3 arc-seconds. The mean of the ‘poor’ Sobral plates put an upper limit of 1.6 arc-seconds on the displacement of the stars.
By November, though, Eddington had decided which of the data were the most valuable: his own two plates. They were the fuzziest of all, and the values derived from them had been obtained by using a formula that included Einstein’s own prediction for the result. Nevertheless, J.J. Thomson, now the President of the Royal Society, ruled the evidence admissible.
Perhaps Thomson was particularly sensitive to accusations over questionable data. The debate between Millikan and Ehrenhaft over Thomson’s electron was still rumbling, after all. And he would have been familiar with that scientist’s sense of just ‘knowing’ when something is right, even when a truly satisfactory proof remains elusive. So, despite the mutterings anyone might have heard, Einstein had been proved right. ‘It is difficult for the audience to weigh fully the meaning of the figures that have been put before us,’ Thomson told the assembly at the Royal Society, ‘but the Astronomer Royal and Professor Eddington have studied the material carefully, and they regard the evidence as decisively in favour of [Einstein’s] value for the displacement.’
Thomson’s announcement obviously didn’t have its intended impact. The Nobel committee excluded relativity from Einstein’s 1921 Nobel Prize in Physics (awarded,