The Twyman-Green Interferometer

Frank Twyman and A. Green patented one change to the Michelson in 1916 and turned a physics instrument into a production tool. Illuminate it with a COLLIMATED beam — a plane wave rather than a point source — and put the optic you want to test in one arm, with a mirror behind it to send the light back through. A perfect optic returns the plane wave unchanged and the fringes are dead straight. A real one returns a wavefront with its own errors printed into it, and the fringes bend by exactly that much. The pattern is a contour map of the optic's error, in steps of half a wavelength, and because the light passes through twice the error is doubled and therefore easier to see. The Fizeau wedge of blueprint 3 already maps a surface, and does it more cheaply. What it cannot do is test anything that is not a flat you can lay a reference against — a cemented doublet, a prism, a whole telescope. Moving the optic into an ARM of an interferometer removes that constraint completely, at the cost of a collimator and about two orders of magnitude in money. It also produces the one number an optical shop can put on a certificate: peak-to-valley wavefront error, against the quarter-wave criterion below which an optic is as good as its aperture allows and polishing further buys nothing at all.