Secrecy and authenticity properties of protocols are mutually dependent: every authentication is based on some secrets, and every secret must be authenticated. This interdependency is a significant source of complexity in reasoning about security. We describe a method to simplify it, by encapsulating the authenticity assumptions needed in the proofs of secrecy. This complements the method for encapsulating the secrecy assumptions in proofs of authenticity, presented in [1]. While logically straightforward, this idea of encapsulation in general, and the present treatment of secrecy in particular, allow formulating scalable and reusable reasoning patterns about the families of protocols of practical interest. The approach evolved as a design strategy in the Protocol Derivation Assistant (Pda), a semantically based environment and toolkit for derivational approach to security [2,3].