On A Theorem of Wajsberg
On A Theorem of Wajsberg
Wajsberg
ADRIAN REZUS·WEDNESDAY, JULY 4, 2018
Not too long ago, I encountered (@ Facebook!) a young man from the States
who claimed he can axiomatize classical propositional logic with a single
axiom, and modus ponens plus substitution as the only rules of inference.
Well, this was well-known since about 1925. The result follows from a more
general theorem of Alfred Tarski (1925), published without proof in 1930.
(Among other things, this shows that Tarski was interested in proof theory, as
well, at an early stage of his logico-matematical carrer.)
The funny thing was that my young [FB-] friend's supposed single axiom
contained a single propositional variable (p, say)!
That was impossible, I exclaimed -- without even looking at his `argument' --,
in view of a well-known (algebraic) result, obtaind by A. H. Diamond and J. C.
C. McKinsey in 1947 -- actually a corollary of it --, saying that any axiom
system for (an arbitrary, finitely axiomatizable fragment of classical)
propositional logic must contain at least one axiom with at least _three_
distinct propositional variables! (Cf. A. H. Diamond, and J. C. C. McKinsey,
`Algebras and their subalgebras', in: Bull. Amer. Math. Soc. 53, 1947, pp. 959--
962.) The fact in itself is a bit curious (why three and not... four?), but... that’s
it!
The young man didn't believe me, and claimed Diamond and McKinsey were
wrong! (Poor McKinsey!) As I was unable to follow his `argument', I gave up
after a while...
Now, I was wrong too, but on... credits / authorship, not on the theorem itself!
Because the [three-variables] result was actually obtained, first, about
seventeen years before Diamond & McKinsey, in 1930, by Mordchaj Wajsberg
(1902--1942), a former PhD student of Jan Łukasiewicz, in his Warsaw PhD
dissertation (published partially in 1931).
[*] For an English translation, see Storrs McCall (ed.), `Polish Logic 1920--
1939', Clarendon Press, Oxford 1967, pp. 264--284. Apparenly, the editor (also
a co-translator of the paper) was not aware of the alternative algebraic proof of
Diamond and McKinsey (1948).
[20180704]