It is now some sixteen years since Sir David Cox (1972) published his epoch making paper in which he incorporated regression type arguments into life-table analysis. Central to the method was the introduction of the multiplicative hazard
with vector of covariates z, unknown regression parameters β, and so-called base-line hazard λ*(t) = λ (t, 0). Applications of the method, based on the conditional likelihood argument expounded by Cox in which the base-line hazard λ* is unknown, have proliferated in the intervening years, largely in the field of medical statistics. There have been relatively few applications in which the base-line hazard is assumed known at the outset. Specific cases include Breslow et al. (1983), Berry (1983) and Hill et al. (1985).