How accurate is Limber's equation?

The so-called Limber equation is widely used in the literature to relate the projected angular clustering of galaxies to the spatial clustering of galaxies in an approximate way. This paper gives estimates of where the regime of applicability of Limber's equation stops. Limber's equation is accurate for small galaxy separations but breaks down beyond a certain separation that depends mainly on the ratio sigma/R and to some degree on the power-law index, gamma, of spatial clustering xi; sigma is the one-sigma width of the galaxy distribution in comoving distance, and R the mean comoving distance. As rule-of-thumb, a 10% relative error is reached at 260 sigma/R arcmin for gamma~1.6, if the spatial clustering is a power-law. More realistic xi are discussed in the paper. Limber's equation becomes increasingly inaccurate for larger angular separations. Ignoring this effect and blindly applying Limber's equation can possibly bias results for the inferred spatial correlation. It is suggested to use in cases of doubt, or maybe even in general, the exact equation that can easily be integrated numerically in the form given in the paper.