The Michelson Moorley experiment to detect the presence of aether is the main experiment that gave rise to the theory of relativity. I've been trying to understand this experiment. The idea of the experiment was to split a beam of light and then align one the split beams with the aether and show that this would speed up one beam compared with the other but the interference pattern of the re-combined beams did not change, hence the aether was proved not to exist. I think this then left the unexpected result that when one of the split beams was aligned with say the path of the Earth round the sun this should alter the distance travelled by one beam compared with the other and therefore the alignment of the wavelengths and so effect the interference pattern but it did'nt. The experimental layout I have been looking at is on this webpage:- http://www.phys.unsw.edu.au/einsteinlight/jw/module3_M&M.htm In trying to understand the experiment I came up with the following. If the horizontal arm is inline with say the path of the Earth round the sun then the mirror on this arm will be travelling at around 1 x 10^4 m/s and when a photon leaves the beam splitter it will be travelling at 3 x 10^8 m/s. Hence if the mirror it bounces back off is 1m away from the splitter, in the time it takes the photon to travel 1m, 3 x 10^-8 s, the mirror will have moved 1 x 10^4 x 3 x 10^-8 = 3 x 10^-4 m or about 1000 wavelengths of visible light. So the horizontal beam will be travelling a different length compared with the vertical beam. However when the horizontal beam starts to travel back the splitter will travel the same distance towards it and hence cancel the effect out so I would expect the two beams to be back in synch! There are other things to consider such as the alignment of the beams with the rotation of the Earth on its axis. The experiment always produces the same result - no change in the fringe pattern but its confusing that the setup in the simplified way I looked at it does not appear to setup the two beams to be on different length paths. Have I mis-interpreted this experiment or does anyone have a qualitative explaintation of why this setup does alter the path lengths and hence why it was expected to produce a change in the fringe pattern?