Abstract
The history of the problem of combining a set of independent observations on the same quantity is traced from antiquity to the appearance in the eighteenth century of the arithmetic mean as a statistical concept.