We discuss the rule of inference called the entailment principle which plays a significant role in the possibilistic reasoning used in the theory of approximate reasoning. We extend this principle to situations in which the knowledge is a type of combination of possibilistic and probabilistic information which we call Dempster-Shafer granules. We discuss the conjunction of these D-S granules and show that Dempster's rule of combination is a special application of conjunction followed by a particular implementation of the entailment principle.