Among birth and death processes with discrete spaces only the simple birth, death and immigration on process (SEDI) preserves the linearity of surprisal; that is, if once the surprisal occurs to be a linear function of the observables under consideration it will remain so forever. Among onedimensional diffusion processes we show that "linear surprisal" is preserved only for those obtained as the limiting process of an SED!. The time development of the variance, however, is essentially different between an SEDI and its diffusion process limit; the variance for the latter never decreases unlike the one for the former. Some examples for violation of the linear-surprisal preservation condition are also discussed. § 1. Introduction