State machines provide a powerful way to describe dynamic behavior of systems and components. Often the suggested Pattern of Gamma et al from 1994 is used for implementing state machines. One Problem of that Pattern is the publication of the state among each other for changing to the next state by its one. An other problem in that context is the bearing of all existing states in the memory. This thesis analyses and uses a novel principle for changing states with the help of the operator Placement-new. It provides an alternative Pattern for the Implementation of state machines. That alternative pattern is simple, fast and memory sparing at the same time.