Abstract
An iterative method for solving a non-linear equation is described, in which a rational function is fitted through previously computed values. Convergence for both single and multiple roots is discussed, first for the case of fitting by linear fractions and then by a more general form.