This essay argues that neither the etiological nor the dispositional account of functions conforms to the actual practice by which functions are ascribed in biology. Philip Kitcher’s account, which unifies what is common to both accounts, is assessed against what biologists are actually doing when they ascribe functions. Two problems
of Kitcher’s account are identified: it is too liberal and it tends to circularity, insofar as it presupposes teleological notions. Finally, an alternative account of functions is provided by characterizing the system of sentences that report natural history.