There is a unique threshold age separating early deaths from late deaths such that averting an early death decreases life disparity, but averting a late death increases inequality in lifespans.

In a stationary population, age composition and the distribution of remaining lifespans are identical. This equivalence can be used to estimate age structure if information is available on time to death.

Mortality change roils period rates. In the short term, conventional calculations of age-specific probabilities of death and life expectancy in the period immediately after the change depend on how many lives have been saved. In the long term, the probabilities and period life expectancy also...

We present and prove a formula for decomposing change in a population average into two components. One component captures the effect of direct change in the characteristic of interest, and the other captures the effect of compositional change. The decomposition is applied to time derivatives of...

Life expectancy is overestimated if mortality is declining and underestimated if mortality is increasing. This is the fundamental claim made by Bongaarts and Feeney (2002) in their article "How Long Do We Live?", where they base their claim on arguments about "tempo effects on mortality". This...

In a stationary population, the change with age in some characteristic at a point in time, summed over all the individuals in the population, equals the change in this characteristic, from the start to the end of the lifetime of each individual, averaged over all lifetimes of the individuals in...

It is well known that life expectancy can be expressed as an integral of the survival curve. The reverse - that the survival function can be expressed as an integral of life expectancy - is also true.

This reflexion is published in memory of Väinö Kannisto, who died unexpectedly on 16 February 2002.