<p>The received wisdom does not question the stability of a mechanism. It has been implicitly agreed that mechanisms are a stable organization of entities that do not evolve over time. The entities do not change from one type to another in a mechanism, although new mechanists allow the entities to change in terms of quantity. This stability of mechanisms confers upon a mechanism its identity. This orthodoxy is examined in this paper. By drawing an analogy from biological and software evolution, I contend that some mechanisms do evolve. The evolution of mechanisms can result in either a distinct mechanism in kind or a “growing” (or “developing”) mechanism—the former is discrete, whereas the latter is continuous with the evolving mechanism. I argue that mechanisms can evolve in three ways. First, a mechanism is said to evolve when its type of entities or activities changes. Second, a mechanism may also evolve when its boundary changes. Third, a mechanism may be said to have evolved when it gives rise to a new behavior or phenomenon.</p>