Some light may be shed on the transformations of demand analysis in the 1930s by telling the story of one of the period’s most debated con- cepts: complementarity. What is the meaning of such sentences as “x and y are substitutes,” “y and z are complementary goods,” or “x and z are independent”? If x is a substitute for y, will y necessarily be a substi- tute for x? Is it supposed to have any empirical counterpart? What kind of meaning does it have in the rst place?1 All those questions emerged soon after the marginalist revolution and were given prominence in the 1930s. They were at the center of the reshaping of demand theory, toward the now-classic Hicks-Slutsky presentation of demand theory.