The effect that one type of medical improvement will have on life expectancies is often computed using a life table. In classical methods, such as Greville's, the increase in life expectancy has been dealt with by assuming that deaths from a particular cause have been eradicated. Keyfitz derived a parameter that measures the increase in life expectancy by a marginal reduction in any cause of death. The parameter is additive in several causes and useful for various studies of causes of death. 

This paper is a generalization of Keyfitz's idea and deals with a case where some percent of the deaths from a particular cause are eliminated, not necessarily uniformly in all age intervals.