Water served as an emblematic locus for debates on the atomic constitution of matter. Today it is taken as common sense that water is H2O, but this was a highly disputed hypothesis for the first half-century of atomic chemistry. In Dalton’s original formulation of the atomic theory published in 1808 water was presented as HO, and consensus on the H2O formula (first proposed by Avogadro) was not reached until after the mid-century establishment of organic structural theory based on the concept of valency. The main epistemic difficulty was unobservability: molecular formulas could be ascertained only on the basis of the knowledge of atomic weights, and vice versa. There were multiple self-consistent sets of molecular formulas and atomic weights, which were employed in at least five different systems of atomic chemistry that flourished in the nineteenth century, each with its distinctive set of aims and methods and in productive mutual interaction. At the heart of the distinctive systems of atomic chemistry were different ways of operationalizing the concept of the atom (weighing, counting, and sorting atoms). It was operationalization that enabled atomic theories to become more than mere hypotheses that may or may not be consistent with observed phenomena. If we examine the crucial phase of development in which the consensus on H2O was achieved, the key was not the revival of Avogadro’s ideas by Cannizzaro, but the establishment of good atom-counting methods in substitution reactions. This, too, was a triumph of operationalization. We also need to keep in mind that the H2O consensus was not a straightforward unification of all systems of atomic chemistry; rather, it was a reconfiguration of the field which resulted in a new pluralistic phase of development.