A general theory for concave gratings is presented that is based on a
recursion formula for the facet positions and that differs from previo
us theories that are based on a power-series expansion of the light pa
th function. Ln the recursion formula approach the facet positions are
determined from a numerical solution for the roots of two constraint
functions. Facet positions are determined in sequence, starting from t
he grating pole. One constraint function may be chosen to give a stigm
atic point. A variety of grating designs are discussed, including a de
sign that cannot be generated with the power-series approach. (C) 1996
Optical Society of America