# Does light bend in a vacuum?

I'm familiar with gravitational lensing but still I'm wondering if there is experiments (conducted here on Earth) which show that light bends due to gravity. For example mirrors setup to hold the light or something like that.

My question is inspired by this bounty question.

• I'd love to know the answer to this. Arguably the Pound-Rebka experiment is indirect measurement since it implies gravitational deflection must happen, but it doesn't actually measure deflection. I estimate the lasers in LIGO are gravitationally deflected by about $10\mu m$ over their 300km path length, but there is no way to measure this deflection in the LIGO setup. – John Rennie Jun 29 '14 at 17:35
• I tried to run the experiment to see if light bends in a vacuum. But apparently the people over at Bed, Bath & Beyond don't like it when you experiment on the products – Jim May 8 '15 at 14:25
• @ACuriousJim You were joking, right? :-) If not, tell us more. – Kimmo Rouvari May 9 '15 at 4:28

• Just to confirm it. Even in the approximation where light freely falls, light needs about 0.1 seconds to fly around/by the Earth. During that time, it drops by $s=gt^2/2\sim 5\times 0.1^2\sim 0.05$ meters in the direction of the gravitational field. To prove that the trajectory isn't straight, one would have to measure the photon's altitude at different places of the globe with precision better than 5 centimeters which is slightly above the abilities of the GPS now, not to mention the fact that the gravitational potential surfaces of Earth aren't mapped precisely, either. – Luboš Motl Jun 29 '14 at 6:18
• Hi @WaitMeDude, the demands for precision would be even much greater if the experiment were compact. Just imagine that the size of the experiment is 3 meters. In $10^{-8}$ seconds, light travels back and forth between the mirrors, or something like that. During that time, the light only "drops" by $gt^2/2=5\times 10^{-16}$ due to the accelerating motion. To prove that it did, you must also be sure that the angles of the mirrors are accurate with the precision $10^{-16}/3$ (meters cancel). All these things are well beyond the precision we may achieve. The problem is the squaring in $gt^2/2$. – Luboš Motl Jun 29 '14 at 8:22