the two measures are changing at significantly different rates, then a change in symmetry is indicated. If the mean and mode in fact rise at about the same rate, then there are two possibilities for the changing distribution of age at death. The first is complete linear shift to the right (higher age) of the mean, mode and effective omega. The second is a stretching of the distribution to the right, where the effective omega would actually be increasing faster than other measures. Of course, as seen in the United States and Japan, a combination of these effects is possible, one above and one below the mode.

# U.S. Experience and Projections

Data for the United States since 1900 provides another perspective on the analysis of deaths by age. Simple survival curves based on period life tables illustrate the degree to which the curve is being “squared” through compression of the age at death, versus a progression to higher ages for the entire distribution. Charts 11 and 12 were produced by the Office of the Actuary at the U.S. Social Security Administration using data from the National Center for Health Statistics, the Census Bureau and from Medicare and Social Security program experience.