# Earliest proof of completeness for axiomatization of Boolean Algebra

Suppose we define Boolean algebra as the system of algebraic rules (logical equivalences) obeyed by AND, OR, NOT with AND, OR, NOT defined by the usual truth tables. We also have variables, which can represent TRUE or FALSE, and we have perentheses. Possibly when a real mathematician says "Boolean algebra" they mean something more general, but I am just a poor physicist, so please bear with me.

There are infinitely many equivalences between logical propositions. However, all of these can be generated by repeated application of finitely many "local" equivalences, such as distributivity, De Morgan's laws, etc. In Chapter 11 of "A survey of Modern Algebra" by Birkhoff and MacLane, there is an implicit proof of this by showing how one can repeatedly apply a finite list of local equivalences to transform any logic proposition into a canonical form that depends only on the truth table of the proposition.

My inclination is to say that this proves the "completeness" of the given set of axioms for Boolean algebra, but please correct me if I am inadvertently misusing technical terms. My question is: if I wanted to cite the original literature for the first proof that a set of axioms for Boolean algebra is complete (in the above sense), what would I cite? My attempts to dig this up myself by library research have so far failed, so any help is appreciated!

-
When you say that a set of axioms is complete you mean to say that there is no claim which is neither provable nor disprovable from the axioms. When you say that a Boolean algebra is complete you mean that the Boolean-induced order defines a complete lattice (supremum and infimum are well-defined and exist for every set). Which one do you mean? – Asaf Karagila Nov 16 '12 at 19:41
@AsafKaragila: To me he seems to mean that a particular proof system for propositional calculus is complete, which is to claim that every tautology has a proof. And the proof system he has in mind is equational rewriting using some finite set of algebraic identities such as $(A\land B)\lor C = (A\lor C)\land(B\lor C)$. – Henning Makholm Nov 16 '12 at 19:46
@Henning: yes that is what I meant – StephenJ Nov 16 '12 at 19:56
@Henning: I had a feeling that I was forgetting another meaning of "complete"... :-) – Asaf Karagila Nov 16 '12 at 21:24