The interferometer consists of a long plate of glass or quartz, with faces that are polished accurately flat and parallel.
Because of the steep angle of incidence, nearly all of the light is reflected, but a tiny fraction leaks out on each bounce.
As in a Fabry–Pérot interferometer, the light that leaks out has phase that depends on how many times it has bounced inside the plate.
A lens is used to overlap light that has emerged after varying numbers of bounces, producing an interference pattern.
A key difference from a Fabry–Pérot etalon is that input light that reflects from the surface of the plate does not contribute to the interference.