EN
In this paper the case of a conditional sampling design proportional to the sum of two order statistics is considered. Several strategies including the Horvitz-Thompson estimator and ratio-type estimators are discussed. The accuracy of these estimators is analyzed on the basis of computer simulation experiments.