It is well known that routine methods of computing the spectrum of a self‐adjoint operator often lead to spurious eigenvalues in a gap between two parts of the essential spectrum. We provide a geometrical explanation for a numerical method of resolving this problem, and give some examples of its use.