With a certain graphic interpretation in mind, we say that a function whose value at every point in its domain is a nonempty set of real numbers is an Abacus. It is shown that to every collection $\mathbf{C}$ of abaci there corresponds a logic, called an abacus logic, i.e., a certain set of propositions partially ordered by generalized implication. It is also shown that to every collection $\mathbf{C}$ of abaci there corresponds a theory $\mathscr{J}_\mathbf{C}$ in a classical propositional calculus such that the abacus logic determined by $\mathbf{C}$ is isomorphic to the poset of $\mathscr{J}_\mathbf{C}$. Two examples are given. In both examples abacus logic is a lattice in which there happens to be an operation of orthocomplementation. In the first example abacus logic turns out to be the Lindenbaum algebra of $\mathscr{J}_\mathbf{C}$. In the second example abacus logic is a lattice isomorphic to the ortholattice of subspaces of a Hilbert space. Thus quantum logic can be regarded as an abacus logic. Without suggesting "hidden variables" it is finally shown that the Lindenbaum algebra of the theory in the second example is a subalgebra of the abacus logic $\mathscr{B}$ of the kind studied in example 1. It turns out that the "classical observables" associated with $\mathscr{B}$ and the "quantum observables" associated with quantum logic are not unrelated. The value of a classical observable contains, in coded form, information about the "uncertainty" of a quantum observable. This information is retrieved by decoding the value of the corresponding classical observable.