The notion of theory split from the notion of logics in the late 1920's when Hilbert and Ackermann distinguished between deduction rules, which do not depend on the ob ject of the speech, and axioms, which depend on it. Then arose the question of how to characterize theories - defined by a set of axioms - which can be used to formalize mathematical reasoning. A first criterion is consistency : the fact that one cannot prove all propositions of a theory. However, it gradually appeared that this criterion was not a sufficient property. Witness and disjunction properties for constructive proofs, and completeness of certain methods of automated deduction are not entailed by the consistency of a theory. But those properties are all consequences of a...