Some remarks on methods of approximation used for the numerical calculation of F(x, t)

Abstract

The method, evolved by F. Esscher in 1932 and described in Section IV, was for a long time the best approximation arrived at for the calculation of F(x, t). However, the error limits deduced by Esscher himself were very much wider than practical results seemed to indicate as necessary. Several papers were published to try to explain this fact (Bohman [2], Esscher [3]), but the explanations seemed somewhat inadequate.