If we imagine observing a beam of light in an accelerated elevator we will see that the light path is curved. By the equivalence principle the same must be true for light whenever gravitational forces are present. This was tested by carefully recording the position of stars near the rim of the sun during an eclipse (see Fig. 7.22) and then observing the same stars half a year later when there is no eclipse.
During the eclipse the observed starlight reaches us only after passing through a region where gravitational effects from the sun are very strong (that is why only stars near the rim are used), but the observation a year later are done at a time where the gravitational effects of the sun on starlight are negligible.
It is found that the position of the stars are displaced when photographs of both situations are compared (see Fig. 7.22). The deviations are the same as the ones predicted by General Relativity. Eddington first observed this effect in 1919 during a solar eclipse. The early 20th century telegram (see Fig. 7.23) announcing this observation for the frist time marks the change in our views about the structure of space time.