Abstract. An approach to ordinal analysis is presented which is finitary, but highlights the semantic content of the theories under consideration, rather than the syntactic structure of their proofs. In this paper the methods are applied to the analysis of theories extending Peano arithmetic with transfinite induction and transfinite arithmetic hierarchies. §1. Introduction. As the name implies, in the field of proof theory one tends to focus on proofs. Nowhere is this emphasis more evident than in the field of ordinal analysis, where one typically designs procedures for "unwinding" derivations in appropriate deductive systems. One might wonder, however, if this emphasis is really necessary; after all, the results of an ordinal analysis describe a relationship between a system of ordinal notations and a theory, and it is natural to think of the latter as the set of semantic consequences of some axioms. From this point of view, it may seem disappointing that we have to choose a specific deductive system before we can begin the ordinal analysis.In fact, Hilbert's epsilon substitution method, historically the first attempt at finding a finitary consistency proof for arithmetic, has a more semantic character. With this method one uses so-called epsilon terms to reduce arithmetic to a quantifier-free calculus, and then one looks for a procedure that assigns numerical values to any finite set of closed terms, in a manner consistent with the axioms. The first ordinal analysis of arithmetic using epsilon terms is due to Ackermann [1]; for further developments see, for example, [20].2000 Mathematics Subject Classification. 03F15,03F05,03F30. Dedicated to Solomon Feferman on the occasion of his 70th birthday. It is an honor to be able to contribute a paper to this volume. Although I did my graduate work under Jack Silver at Berkeley, Sol served as an informal second advisor to me, and his thoughtful advice and guidance made frequent visits to Stanford both enjoyable and well worthwhile. Mathematical logic, as a discipline, is poised between philosophy and mathematics, and so has to answer to competing standards of philosophical relevance and mathematical elegance. Throughout his career, Sol has been able to strike a harmonious balance between the two, with work that is deeply satisfying on both counts. His style sets a high standard for future generations, and one that many of us will look to as a model.I would like to thank Lev Beklemishev for comments and suggestions, and the anonymous referees for their very careful readings.