Pre-print, Working paper: Stock market indices are today a vital and daily tool for both economists and actors in the financial world. The multiplication and the very importance given to these indices raise the question of their accuracy and of the reliability of the methods that are used to construct them. We begin an investigation on these questions by studying the early history of these indices. We show that stock market indices appeared in the daily press in the United States at the end of the 19th century; that around World War One, they became the focus of the interest of very different groups of people, so that their construction became a more complex and specialized task. The scientific study of indices did not result initially from the stock market's importance in finance (for firms financing, for savers' portfolio choices or for investment banks' decisions), since most of the initial interest came from economists that looked at the stock market only as a measure or an index of the macroeconomic situation. The development of indices dedicated to financial studies came only in the late 1920s, and accelerated only with the birth of modern finance. This article describes the origins of stock-market indices in the interwar period, with an emphasis on France and the United States. It links this evolution with contemporary economic theories, index number theory, financial practices, and the other motivations of their authors. It examines the consequences of the methodological choices that were made and suggests that they had a surprisingly large impact on the results. In particular, we analyse in detail the motivations and technical characteristics of the most important indices that were produced during the interwar period by the French government statistical office (the Statistique générale de la France or SGF). We suggest that these indices cannot be easily compared to most usually discussed indices for other countries and that new calculations are required before international comparisons.