In Principles of Mathematical Analysis, Rudin specified a definition. It is very important because we often use it in convergence proofs. However, it's a definition, not an axiom. And it's not a part of the binary operator (>) defition. I think it's possible to derive it somehow.
