Re-reading Menger's table

Abstract

Part of the recent reassessment of Menger's thought (e.g. Alter, 1991, 1992; Klausinger, 1992) involves the question whether or not the preferences underlying his well-known table are indeed lexicographic in nature. The answer is rendered complex and the discussion confused, since various classes of lexicographic preferences are possible and have been used by various authors studying Menger. This paper shows some of these representations to be unfounded. Instead, it proposes a class of lexicographic preferences that is closer in spirit to Menger's thinking. It is shown that a partial aspect of the ordering found in Menger's table may also be represented by a real-valued utility function. But textual evidence shows that only a vector-valued representation can adequately capture all of Menger's intent.