# Returning Light

What happen if someone is just close enough to a black hole, and he emits light tangentially out, and the light got bent and circles around, what will he see, can he see the light source from his back?

             +---------->
/            \
light   /              \
^                \
observer  +   O black hole /
sees light\               /
\             /
<----------- light path

-
I read a science fiction where they send some trash (like you sent light) which comes back and hits a turbine blade to generate energy. If I remember correctly this way of harvesting energy from black holes was suggested by Roger Penrose. – Pratik Deoghare Jul 9 '11 at 10:43
@MachineCharmer - iirc, this is called the Penrose Process (which would also make a good band name) and involves extracting angular momentum from a rotating black hole. Otherwise it would appear to violate conservation of energy. – Richard Terrett Jul 9 '11 at 16:59

Such places do exist; what you're looking for is a circular null orbit. Uh, "null" means having no change in proper time, that is, the sort of path (or orbit or geodesic) that is suitable for light. The radius for this orbit is $3r_s/2$ where $r_s$ is the Schwarzschild radius. See wikipedia, Schwarzschild metric / orbital motion.
The spacecraft emitting the light would have to use thrusters (or something else, maybe a rope or what have you) to keep itself stationary, but the value is outside the event horizon at the Schwarzschild radius (where $r_s=1$), so this is possible.
I found “The mathematical theory of black holes” by S. Chandrasekhar (1992) quite useful for such a purpose. The already mentioned $3 r_s/2$ orbit for light really exists, but it is not stable, i.e. any small fluctuation or error in initial direction make impossible for light to return in the initial point and walk around black hole forever. It is because the last stable circular orbit is $3 r_s$ (see already cited wikipedia article). Yet in the mentioned book is described null geodesic (i.e. path of a light beam) asymptotically spiralling around the circle $3 r_s/2$.