Light grows tired climbing out
Picture a beam of light leaving the floor of a deep, deep well and climbing up toward you at the rim. A thrown ball would slow down as it rises, trading speed for height. Light cannot slow down — its speed is fixed for everyone — so it pays the toll a different way: it loses a little energy, which for light means its color drifts toward the red. A blue photon sent up arrives a touch redder. This reddening of light escaping gravity is called gravitational redshift.
We can guess this without any new law, using only the equivalence principle from the start of this track. Picture a windowless rocket accelerating upward. Light fired from the floor reaches the ceiling a moment later — and in that moment the ceiling has sped up, so it is rushing away from the light and sees it Doppler-shifted toward red. The equivalence principle says a rocket accelerating in space is indistinguishable from sitting still in gravity. So the same redshift must happen for light rising against real gravity. No new postulate needed; the falling-elevator idea already demands it.
Redshift and slow clocks are the same fact
Here is the beautiful twist: gravitational redshift is just gravitational time dilation wearing a different costume. Think of a light wave as a clock that ticks once per crest. A clock deep in a gravity well runs slow, so it sends up fewer crests each second. When you catch them up high — where your own clock runs faster — those crests arrive spread further apart: a longer wavelength, a redder color. Slow clocks below and red light arriving above are two descriptions of one thing.
high up (fast clock, thin gravity)
you -> ~~~~~~~~~~~~~~~~~~~~~~ crests arrive far apart = REDDER
\ /
\ light rises /
\ losing /
\ energy /
deep down -> |||||||||||| crests sent close together (slow clock)
(source) source emits blue
slow clock below <==> redshift seen above (one fact, two views)Even light has to turn
The same equivalence trick predicts something even more startling. Fire a laser straight across that accelerating rocket, wall to wall. While the light crosses, the rocket moves up, so the spot lands a hair *lower* than where it was aimed — from inside, the beam looks bent downward. By the equivalence principle, gravity must bend light the very same way. A ray of starlight skimming past a heavy mass should curve toward it. This is the bending of light, and it means light is not exempt from gravity — nothing is.
1919: the eclipse that crowned Einstein
How do you watch starlight bend past the Sun, when the Sun's glare drowns every nearby star? You wait for a total solar eclipse. In May 1919 the astronomer Arthur Eddington led expeditions to the island of Príncipe off Africa and to Brazil. As the Moon blotted out the Sun, they photographed the stars that appeared right at its edge — then compared those plates to night-time photos of the same stars taken months earlier.
- The prediction: light grazing the Sun's edge should swing by about 1.75 arcseconds — Einstein's full number, twice the old Newtonian guess.
- The look: a star really sitting *behind* the Sun's edge appears nudged slightly *outward*, away from the Sun, because its light curved on the way to us.
- The verdict: the measured shift matched Einstein, not Newton. On 6 November 1919 the result was announced, and the next morning newspapers crowned him world-famous overnight.
Two clues, one curvature
Step back and notice that redshift and bending are not two separate gravitational quirks — they are two faces of a single idea. If light slows its ticking deep in a well (redshift) and also curves its path near a mass (bending), the simplest story is that gravity is not a force pulling on light at all. Instead, the mass reshapes the very stage — curved spacetime — and light, doing nothing but going as straight as it can, follows the warp. Time runs slow where space is curved, and the beam leans toward the mass because that is the shortest path through bent geometry.