Synthese 137 (2003) 211-259.
After a brief flirtation with logicism in 1917–1920, David Hilbert proposed his own program in the foundations of mathematics in 1920 and developed it, in concert with collaborators such as Paul Bernays and Wilhelm Ackermann, throughout the 1920s. The two technical pillars of the project were the development of axiomatic systems for ever stronger and more comprehensive areas of mathematics and finitistic proofs of consistency of these systems. Early advances in these areas were made by Hilbert (and Bernays) in a series of lecture courses at the University of Göttingen between 1917 and 1923, and notably in Ackermann's dissertation of 1924. The main innovation was the invention of the epsilon-calculus, on which Hilbert's axiom systems were based, and the development of the epsilon-substitution method as a basis for consistency proofs. The paper traces the development of the "simultaneous development of logic and mathematics" through the epsilon-notation and provides an analysis of Ackermann's consistency proofs for primitive recursive arithmetic and for the first comprehensive mathematical system, the latter using the substitution method. It is striking that these proofs use transfinite induction not dissimilar to that used in Gentzen's later consistency proof as well as non-primitive recursive definitions, and that these methods were accepted as finitistic at the time.
Note: his is a revised version of Chapter 3 of my dissertation.
John W. Dawson, Jr (Mathematical Reviews 2038463 (2005b:03008))
doi:10.1023/A:1026247421383 (Subscription required)