This year we celebrate the centenary of Albert Einstein's special theory            of relativity. Indeed, 1905 was the year in which Einstein first gave            notice of his astonishing abilities. He was but 26 and had just earned            his doctorate, but that year he published four papers on separate topics,            each of which marked a major advance in physics. The first of these,            on the photo-electric effect, would bring him the Nobel Prize, but it            was the third, on special relativity, that made him both famous and            controversial. A decade after this flurry of papers, in 1915, he unveiled            the theory of general relativity, shaking again the foundations of science.
                               
  So different was relativity from the prevailing beliefs         that most     physicists demanded proof that it could explain         phenomena that Isaac     Newton's canon could not. Satisfying         such demands was difficult,     because the difference between         the two models could only be apparent     under extreme         conditions. There seemed little hope that any     terrestrial         experiment could decide between them, but Einstein later         identified three astronomical tests. The first was the proper         calculation of the orbit of the planet Mercury—a feat that         was     beyond Newtonian physics (see "In Pursuit of         Vulcan" in     the September-October 1994 American         Scientist). The second     test required the comparison of         light emitted from atoms in the Sun     with light from similar         atoms on Earth—relativity predicted     that the Sun's         light would have a longer wavelength (an example of     the         so-called redshift). The third test posited that if relativity         was true, then rays of starlight that passed near the Sun would         be     bent compared to the same rays when the Sun was elsewhere         in the     sky. In each case, the relativistic effects are         caused by gravity     from the Sun's huge mass.
  Early         attempts to perform these tests did not silence Einstein's         critics, because some observations supported his theory and         others     did not. Thus, the general theory of relativity         yielded a much     better solution to the Mercury problem than         did Newtonian models,     but another prediction of relativity,         the redshift of the solar     spectrum, could not be verified.         (Eventually, astrophysicists     learned that several other         factors complicated the observation of     this phenomenon.) So         with one result in favor and another in doubt,     the third         test became something of a deciding vote for or against     relativity.
  Einstein first suggested how this light-bending         effect could be     measured in 1911. He predicted that those         rays of starlight that     passed closest to the Sun would be         deflected by 0.85 arcseconds     (0.00023 degree) because of the         Sun's gravitational field. However,     stars that appear next         to the Sun are only visible during a total     solar eclipse. To         test Einstein's hypothesis, one would have to take         photographs during an eclipse that showed background stars near         the     Sun's disk and compare them with photos taken months         earlier or     later, when the same stars rose in the night sky.         Did stars     appearing on opposite sides of the Sun's disk         maintain the same     spacing when the Sun was gone, or not?
         This prediction seemed easy to check. Many pictures of solar         eclipses already existed, as did photos of the night sky. Even         so,     skepticism about Einstein's theory was so prevalent that         few     astronomers rushed to their archives. And when they did         examine     previous photographs of solar eclipses, they found         that the pictures     were unsuited to proving or disproving         Einstein's claim: The     telescopes had been set to track the         Sun's motion across the sky,     not the stellar motions, and         the slight differences between these     perspectives obscured         the small, predicted shifts in star positions.     However, as         time went by and other experiments gave equivocal     results,         the solar-eclipse experiment represented the best chance to         test the truth of relativity.
                      Hoping for a Dark Noon
         As early as 1912 it seemed possible to capture the necessary         photographs with little fuss. In October of that year, a total         solar     eclipse was to run across the northern parts of South         America, and     the astronomical observatory of Córdoba         in central Argentina     was near enough to mount an expedition.         Unhappily, almost all of     South America was under clouds that     day.
  Another suitable eclipse loomed in August 1914, running         northwest to     southeast across eastern Europe. Erwin         Freundlich, a young German     astronomer, was determined to         test Einstein's theory but encountered     grave difficulty         raising money for the trip. The scientific     establishment in         Germany was uninterested in paying for it, leading     Einstein         himself to offer his own none-too-abundant finances. With     so         few options, Freundlich appealed to other countries for         collaborators that would help fund the expedition. He had only         one     taker: William Wallace Campbell and a team from the Lick         Observatory     in California. Later, the Berlin Academy         provided additional support.
  The eclipse was due August         21, but the team of Germans and Americans     established a camp         near Kiev well before that date to prepare for     the event.         Unfortunately, history intervened: On August 1, 1914,         Germany declared war on Russia, and the German astronomers were         taken prisoner. Russian forces expelled the older scientists and         held the younger ones as prisoners of war. The Russians did         allow     the Americans to stay for the eclipse, but again the         sky was totally     clouded out. Campbell later wrote "I         never knew before how     keenly an eclipse astronomer feels his         disappointment through     clouds. One wishes that he could come         home by the back door and see nobody."
    |  | 
  | Arthur Eddington | 
 
 The next         year, at the height of the First World War, Einstein         published his general theory of relativity. This timing greatly         complicated the theory's dissemination because German scientific         journals were then unavailable to the English-speaking world. It         was     an astronomer from neutral Holland who brought word of         the new     theory to Britain. Moreover, Britain was going         through a period of     almost hysterical opposition to all         things German. Ardently opposed     to this mindless, pervasive         hatred, a young British astrophysicist     named Arthur Stanley         Eddington stood almost alone. Eddington was not     only a         rising star in astronomy but a Quaker—a religious         pacifist. As such, he refused to fight in the war, although he         was     willing to risk his life providing aid to civilians         caught in the     violence. Because of his beliefs, Eddington         lived on the verge of     imprisonment during much of the war         and suffered vicious attacks for     his pacifism and efforts to         counter his peers' nationalistic     hostility toward German     science.
  Eddington learned of Einstein's general theory from         the Dutch     astronomer Willem de Sitter and was immediately         taken with it. He     was almost certainly the first (and, for a         while, the only)     English-speaker to understand the theory         and appreciate its     significance. Eddington grasped the fact         that Einstein's new work     meant that the eclipse experiment         was an even more significant test     of relativity—the         general theory predicted twice as much     deflection of light         rays passing the Sun as did the special theory.     Another         suitable eclipse would occur in 1919, and although in 1915         there was no immediate hope for peace, the British Astronomer         Royal,     Frank Dyson, began to lay plans (no doubt at         Eddington's prompting)     for an expedition to photograph the         event. Eddington, of course, was     eager to lead such an         expedition but worried that his uncertain     standing with the         authorities might cause difficulties for the     project. Then,         in a stroke of genius, Dyson wrote a carefully worded     letter         to officialdom. In response, the government notified         Eddington that he was lucky so far in having avoided prison, and         that his only hope of remaining that way was to lead Dyson's         expedition, whether Eddington liked it or not! Eddington         dutifully     bowed to the hoped-for ultimatum.
     Partly Cloudy
                 Around the         same time, an eclipse in the United States in June 1918     was         almost entirely obscured by clouds, but Campbell's team did get         some photographs. These poorly exposed plates seemed to indicate         no     relativistic effects, much to the delight of Einstein's         skeptics,     including Campbell.
                                    
  The eclipse of May 29, 1919, was to start near the border         between     Chile and Peru, then traverse South America, cross         the Atlantic     Ocean and arc down through central Africa. No         part of the path was     far from the equator, and the         desirable, longest-lasting portion was     in the Atlantic, a         few hundred miles from the coast of Liberia. The     British         planners decided that the tiny island of Principe, nestled         in the crook of Africa's Gulf of Guinea, would be best despite         the     poor astronomical viewing from low-lying tropical         regions. The     choice of Principe introduced other challenges.         One modern travel     agency advises prospective visitors to the         island that "It's     best to go between June and         September. The rest of the year is muggy     and         hot—you'll be swimming in rain and your own sweat."         Just in case Principe was cloudy at the crucial time, the         British     sent a second expedition to observe the eclipse from         Sobral, in     eastern Brazil.
  The main instruments at         both sites were existing astrographic     telescopes of         33-centimeter aperture designed specifically for         photographing star positions with high precision. Although these         telescopes were designed to automatically follow the stars,         their     temporary emplacement in the field required each         telescope to be     immobilized as a clockwork-driven flat         mirror tracked across the sky     and fed light to the main         lens. As an afterthought, the Brazil     contingent added a         small 10-centimeter telescope to its roster. In     the end, it         saved the day.
  The expeditionaries set out months ahead         of the eclipse to allow for     travel difficulties. Although         the war officially ended in November     1918, chaos continued         for months thereafter. Upon arrival, they had     to evaluate         the terrain, choose a site, and set up and test their         equipment. Eddington's group arrived at Principe in late April         and,     amid the heat and rain, found themselves under such         constant attack     by biting insects that they needed to work         under mosquito netting     most of the time. The rain grew worse         as May advanced, and the day     of the eclipse began with a         tremendous storm. The rain stopped as     the day wore on, but         the totality phase of the eclipse would start     at 2:15 p.m.         and last only five minutes. Eddington wrote:
         About 1.30 when the partial phase was well advanced,         we     began to get glimpses of the Sun, at 1.55 we could see         the crescent     (through the cloud) almost continuously and         large patches of clear     sky appearing. We had to carry out         our programme of photographs in     faith. I did not see the         eclipse, being too busy changing plates,     except for one         glance to make sure it had begun.... We took 16     photographs         ... but the cloud has interfered very much with the star         images. 
  The weather in Brazil was much         better—beautifully clear, in     fact. The observers took         19 photos with the astrograph and eight     with the small         telescope. But when the photographs were developed,     they         found that despite their precautions, the astrograph's pictures         showed, according to Dyson, "a serious change of focus, so         that, while the stars were shown, the definition was         spoiled."     Even under ideal conditions, the predicted         relativistic displacement     on the photographs was only 1/60         of a millimeter—about a     quarter of the diameter of a         star on a sharply exposed image.     Although they could measure         such a minute shift, the poor focus made     this task nearly         impossible. By contrast, the small telescope's     photographs         were clear and sharp, but on a reduced scale.
            
  |  | 
              |                    The original caption for the graphical explanation                    of the experiment read as follows: The results obtained                    by the British expeditions to observe the total eclipse of the                    sun last May verified Professor Einstein's theory that light                    is subject to gravitation. Writing in our issue of November                    15 [1919], Dr. A.C. Crommelin, one of the British observers,                    said: "The eclipse was specially favourable for the purpose,                    there being no fewer than twelve fairly bright stars near the                    limb of the sun. The process of observation consisted in taking                    photographs of these stars during totality, and comparing them                    with other plates of the same region taken when the sun was                    not in the neighbourhood. Then if the starlight is bent by the                    sun's attraction, the stars on the eclipse plates would seem                    to be pushed outward compared with those on the other plates….                    The second Sobral camera and the one used at Principe agree                    in supporting Einstein's theory…. It is of profound philosophical                    interest. Straight lines in Einstein’s space cannot exist; they                    are parts of gigantic curves." From the Illustrated London                    News of November 22, 1919. | 
 
                     Weighing the Data
                 Many         months later, back in England, Eddington pondered the         inconsistent results. Einstein's theory predicted a displacement         of     1.75 arcseconds, but none of the experiments was in         perfect     agreement with the theory. The usable photos from         Principe showed an     average difference of 1.61±0.30         arcseconds, the astrograph in     Brazil indicated a deflection         of about 0.93 arcseconds (depending on     how one weighted the         individual spoiled photos), and the little     10-centimeter         telescope gave a result of 1.98±0.12     arcseconds. The         smaller device, in addition to yielding the most     precise         data, afforded a wider field of view and supported         Einstein's theory of how the displacement should vary with         angular     distance from the edge of the Sun. But the         validation of relativity     required exact measurements,         particularly because physicists had     realized that Newtonian         theory alone could predict a stellar     displacement that was         half that of Einstein's, or about 0.83 arcseconds.
         Eventually, Eddington, after much discussion with Dyson,         suggested     an overall measurement of 1.64 arcseconds, which         he took to be in     pretty good agreement with Einstein, but he         also gave the separate     results from each telescope so others         might weight them as they saw     fit. Moreover, Dyson offered         to send exact contact copies of the     original photographic         glass plates to anyone who wished to make     their own         measurements, which should have gone far to refute the         occasional allegation that Eddington had cooked the results.
         Ironically, confirmation of Eddington's conclusion (and the         theory     of relativity) came from Campbell's team at an         eclipse in Australia     in 1922, for which they determined a         stellar displacement of     1.72±0.11 arcseconds.         Campbell had been open in his belief     that Einstein was         wrong, but when his experiment proved exactly the     opposite,         good scientist that he was, Campbell immediately admitted         his error and never opposed relativity again.
                     
                 Acknowledgment
                  I am indebted to Dr. Jeffrey Crelinsten for granting access                    to his unpublished work on this topic and for providing comments                    on an earlier version of this article.