The systematic emergence of the test approach to index numbers took place in the early 1920s, a period marked by pluralism in American economics. Promoted by Irving Fisher, the test approach to index numbers aimed to select a universally valid, ideal formula for index numbers by employing a series of mathematical tests. However, from the first presentation of Fisher's approach, at the 1920 Annual Meeting of the American Statistical Association, to the publication of his book The Making of Index Numbers, in 1922, he faced a series of criticisms, not addressed to his ideal formula per se, but rather aimed at the very idea that a universal formula for index numbers could be singled out. The most prominent individuals involved in this debate were Wesley Mitchell, Warren Persons, Correa Walsh (Fisher's only supporter), and Allyn Young. Among them, the foremost representative of Fisher's antagonists was Mitchell. This study aims at reconstructing this antagonism, arguing that the disagreements between Fisher and Mitchell resulted from their different backgrounds and their distinct understandings of economics as a science. More specifically, this article illustrates how Fisher, as a mathematical economist, privileged a universalist conception of science, while Mitchell, as an institutionalist, understood economics as a contextual and historical discipline, and it illustrates how these preconceptions spilled over to their debates on index numbers. To illuminate such positions, this study explores their archival correspondence.