To date, despite decades of investigations and the relative abundance of mortality data, our understanding of the phenomenon of 'mortality crossover' remains inadequate. We propose a methodology for transforming mortality data from the 'age-domain' to the 'time-domain'. We then introduce a model of selection partially offset by mobility, to simulate the dynamics of vulnerability in a population cohort that is heterogeneous in health and death. Using our model of vulnerability simulating the dynamics of mortality in the time-domain, we compare the mortality experience of the Black and White populations of the United States, identify the significance of selection and mobility as potential factors producing the crossover phenomenon, and make diagnostic use of them.