Problem of estimating the population mean at current occasion in successive
sampling has been studied by various authors in design approach. Present work
is an attempt to study the similar kind of problem under a super-population
model. Optimum replacement policy and performance of the proposed chain
type di erence estimator has been discussed under the assumed super-population
model. Empirical comparison of the proposed estimator is made with respect to
sample mean estimator and suitable recommendations are made.