Arguments are presented that the theoretical definition of an atom in a mol
ecule or of a functional grouping of atoms that derive from experimental ch
emistry must be unique. Definitions based on the orbital model or, as recen
tly proposed, in terms of domains defined by isovalued density envelopes fa
il for a number of reasons, among them being their failure to enable a quan
tum mechanical description of the atomic or group properties. Chemistry is
concerned with the observation and measurement of properties. Definitions t
hat do not predict the measurable, additive properties found for atoms in m
olecules fail to recover the essence of the atomic concept and can play no
operational or predictive role in chemistry. Atoms exist in real space and
their form determines their properties. There is but a single definition fo
r an atom, free or bound, that meets this essential requirement.